{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 10623,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 9.023058116889983,
      "learning_rate": 6.269592476489028e-08,
      "loss": 0.7969,
      "step": 1
    },
    {
      "epoch": 0.0,
      "grad_norm": 16.9694817534492,
      "learning_rate": 1.2539184952978057e-07,
      "loss": 0.8867,
      "step": 2
    },
    {
      "epoch": 0.0,
      "grad_norm": 12.385859710718453,
      "learning_rate": 1.8808777429467086e-07,
      "loss": 0.8486,
      "step": 3
    },
    {
      "epoch": 0.0,
      "grad_norm": 9.50728828775519,
      "learning_rate": 2.5078369905956113e-07,
      "loss": 0.8252,
      "step": 4
    },
    {
      "epoch": 0.0,
      "grad_norm": 17.0251911072455,
      "learning_rate": 3.134796238244514e-07,
      "loss": 0.8984,
      "step": 5
    },
    {
      "epoch": 0.0,
      "grad_norm": 8.870217072319136,
      "learning_rate": 3.761755485893417e-07,
      "loss": 0.8818,
      "step": 6
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.5822399665909614,
      "learning_rate": 4.38871473354232e-07,
      "loss": 0.627,
      "step": 7
    },
    {
      "epoch": 0.0,
      "grad_norm": 11.508186314772216,
      "learning_rate": 5.015673981191223e-07,
      "loss": 0.6729,
      "step": 8
    },
    {
      "epoch": 0.0,
      "grad_norm": 10.833482875074711,
      "learning_rate": 5.642633228840126e-07,
      "loss": 0.7861,
      "step": 9
    },
    {
      "epoch": 0.0,
      "grad_norm": 8.924876847452603,
      "learning_rate": 6.269592476489028e-07,
      "loss": 0.7627,
      "step": 10
    },
    {
      "epoch": 0.0,
      "grad_norm": 9.992140582826366,
      "learning_rate": 6.896551724137931e-07,
      "loss": 0.8682,
      "step": 11
    },
    {
      "epoch": 0.0,
      "grad_norm": 9.992255972443829,
      "learning_rate": 7.523510971786834e-07,
      "loss": 0.833,
      "step": 12
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.583707363312337,
      "learning_rate": 8.150470219435737e-07,
      "loss": 0.793,
      "step": 13
    },
    {
      "epoch": 0.0,
      "grad_norm": 8.576514868016705,
      "learning_rate": 8.77742946708464e-07,
      "loss": 0.8174,
      "step": 14
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.845204108015149,
      "learning_rate": 9.404388714733543e-07,
      "loss": 0.71,
      "step": 15
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.532819953013649,
      "learning_rate": 1.0031347962382445e-06,
      "loss": 0.7734,
      "step": 16
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.2412127394832035,
      "learning_rate": 1.0658307210031348e-06,
      "loss": 0.6807,
      "step": 17
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.228747155809349,
      "learning_rate": 1.1285266457680252e-06,
      "loss": 0.7998,
      "step": 18
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.440378350444176,
      "learning_rate": 1.1912225705329155e-06,
      "loss": 0.7529,
      "step": 19
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.941928725413803,
      "learning_rate": 1.2539184952978056e-06,
      "loss": 0.6787,
      "step": 20
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.614022225661636,
      "learning_rate": 1.316614420062696e-06,
      "loss": 0.646,
      "step": 21
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.016716614514892,
      "learning_rate": 1.3793103448275862e-06,
      "loss": 0.6309,
      "step": 22
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.119791313123958,
      "learning_rate": 1.4420062695924766e-06,
      "loss": 0.6475,
      "step": 23
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.915355044013978,
      "learning_rate": 1.5047021943573669e-06,
      "loss": 0.7725,
      "step": 24
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.6164263694394307,
      "learning_rate": 1.5673981191222572e-06,
      "loss": 0.7002,
      "step": 25
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.9991189532369527,
      "learning_rate": 1.6300940438871473e-06,
      "loss": 0.7568,
      "step": 26
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.6617325537693186,
      "learning_rate": 1.6927899686520376e-06,
      "loss": 0.6895,
      "step": 27
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.403501443821879,
      "learning_rate": 1.755485893416928e-06,
      "loss": 0.7227,
      "step": 28
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.3095088515558713,
      "learning_rate": 1.8181818181818183e-06,
      "loss": 0.7119,
      "step": 29
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.225833546926646,
      "learning_rate": 1.8808777429467086e-06,
      "loss": 0.5459,
      "step": 30
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.3302988241614155,
      "learning_rate": 1.943573667711599e-06,
      "loss": 0.6611,
      "step": 31
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.510414978677273,
      "learning_rate": 2.006269592476489e-06,
      "loss": 0.751,
      "step": 32
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9974695478315863,
      "learning_rate": 2.0689655172413796e-06,
      "loss": 0.522,
      "step": 33
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.548461523964619,
      "learning_rate": 2.1316614420062697e-06,
      "loss": 0.6006,
      "step": 34
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1417592495372975,
      "learning_rate": 2.1943573667711602e-06,
      "loss": 0.6016,
      "step": 35
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.145157087258389,
      "learning_rate": 2.2570532915360503e-06,
      "loss": 0.5684,
      "step": 36
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1120724974898657,
      "learning_rate": 2.3197492163009404e-06,
      "loss": 0.5747,
      "step": 37
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.4691692949250723,
      "learning_rate": 2.382445141065831e-06,
      "loss": 0.584,
      "step": 38
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.28600267196437,
      "learning_rate": 2.445141065830721e-06,
      "loss": 0.6519,
      "step": 39
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1594967303860106,
      "learning_rate": 2.507836990595611e-06,
      "loss": 0.6714,
      "step": 40
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.174448171315612,
      "learning_rate": 2.5705329153605017e-06,
      "loss": 0.6323,
      "step": 41
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.318937624191795,
      "learning_rate": 2.633228840125392e-06,
      "loss": 0.5815,
      "step": 42
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.299798517516603,
      "learning_rate": 2.6959247648902824e-06,
      "loss": 0.5737,
      "step": 43
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0857687841936032,
      "learning_rate": 2.7586206896551725e-06,
      "loss": 0.5386,
      "step": 44
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.139264437993848,
      "learning_rate": 2.8213166144200626e-06,
      "loss": 0.5957,
      "step": 45
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0739904831671123,
      "learning_rate": 2.884012539184953e-06,
      "loss": 0.5039,
      "step": 46
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2844519816383895,
      "learning_rate": 2.9467084639498432e-06,
      "loss": 0.6562,
      "step": 47
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1953756647701352,
      "learning_rate": 3.0094043887147338e-06,
      "loss": 0.6035,
      "step": 48
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.331260439284494,
      "learning_rate": 3.072100313479624e-06,
      "loss": 0.5469,
      "step": 49
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.4303408451135318,
      "learning_rate": 3.1347962382445144e-06,
      "loss": 0.4458,
      "step": 50
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.7570529627993405,
      "learning_rate": 3.1974921630094045e-06,
      "loss": 0.5376,
      "step": 51
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8830015095654988,
      "learning_rate": 3.2601880877742946e-06,
      "loss": 0.6318,
      "step": 52
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8958162326067567,
      "learning_rate": 3.322884012539185e-06,
      "loss": 0.4512,
      "step": 53
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9571061802256204,
      "learning_rate": 3.3855799373040753e-06,
      "loss": 0.521,
      "step": 54
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0852815610096416,
      "learning_rate": 3.448275862068966e-06,
      "loss": 0.7266,
      "step": 55
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8830810778927451,
      "learning_rate": 3.510971786833856e-06,
      "loss": 0.4746,
      "step": 56
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.014033976973716,
      "learning_rate": 3.573667711598746e-06,
      "loss": 0.5044,
      "step": 57
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.7066717890252903,
      "learning_rate": 3.6363636363636366e-06,
      "loss": 0.5259,
      "step": 58
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.4109428877102963,
      "learning_rate": 3.6990595611285267e-06,
      "loss": 0.5923,
      "step": 59
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.682623861552679,
      "learning_rate": 3.7617554858934172e-06,
      "loss": 0.4351,
      "step": 60
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.826794422209083,
      "learning_rate": 3.824451410658307e-06,
      "loss": 0.4888,
      "step": 61
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8171731368058555,
      "learning_rate": 3.887147335423198e-06,
      "loss": 0.457,
      "step": 62
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.7965718462218965,
      "learning_rate": 3.949843260188088e-06,
      "loss": 0.4756,
      "step": 63
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0675771167831947,
      "learning_rate": 4.012539184952978e-06,
      "loss": 0.5625,
      "step": 64
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8961666918909523,
      "learning_rate": 4.075235109717869e-06,
      "loss": 0.5361,
      "step": 65
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.738455402002404,
      "learning_rate": 4.137931034482759e-06,
      "loss": 0.3911,
      "step": 66
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.98958206112052,
      "learning_rate": 4.20062695924765e-06,
      "loss": 0.5557,
      "step": 67
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0713991070282045,
      "learning_rate": 4.263322884012539e-06,
      "loss": 0.4795,
      "step": 68
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.980051182361179,
      "learning_rate": 4.32601880877743e-06,
      "loss": 0.5093,
      "step": 69
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9058446567411553,
      "learning_rate": 4.3887147335423205e-06,
      "loss": 0.478,
      "step": 70
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.2823661291572455,
      "learning_rate": 4.45141065830721e-06,
      "loss": 0.6318,
      "step": 71
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.001661815037,
      "learning_rate": 4.514106583072101e-06,
      "loss": 0.5352,
      "step": 72
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8609187631947401,
      "learning_rate": 4.576802507836991e-06,
      "loss": 0.4229,
      "step": 73
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9762421310026794,
      "learning_rate": 4.639498432601881e-06,
      "loss": 0.5151,
      "step": 74
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.983847987510723,
      "learning_rate": 4.7021943573667714e-06,
      "loss": 0.4624,
      "step": 75
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0783117720564364,
      "learning_rate": 4.764890282131662e-06,
      "loss": 0.5742,
      "step": 76
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.6916882106966997,
      "learning_rate": 4.8275862068965525e-06,
      "loss": 0.4355,
      "step": 77
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.122430458963372,
      "learning_rate": 4.890282131661442e-06,
      "loss": 0.5981,
      "step": 78
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9334189329545817,
      "learning_rate": 4.952978056426333e-06,
      "loss": 0.584,
      "step": 79
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.077535733913982,
      "learning_rate": 5.015673981191222e-06,
      "loss": 0.5405,
      "step": 80
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9752297953868678,
      "learning_rate": 5.078369905956113e-06,
      "loss": 0.5444,
      "step": 81
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.7391560574540463,
      "learning_rate": 5.1410658307210035e-06,
      "loss": 0.5059,
      "step": 82
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9711610185462864,
      "learning_rate": 5.203761755485894e-06,
      "loss": 0.5479,
      "step": 83
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8054867946441293,
      "learning_rate": 5.266457680250784e-06,
      "loss": 0.5063,
      "step": 84
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.1252581046564165,
      "learning_rate": 5.329153605015674e-06,
      "loss": 0.5898,
      "step": 85
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9152422036060939,
      "learning_rate": 5.391849529780565e-06,
      "loss": 0.4678,
      "step": 86
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0378165515393567,
      "learning_rate": 5.4545454545454545e-06,
      "loss": 0.6338,
      "step": 87
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9908933730221778,
      "learning_rate": 5.517241379310345e-06,
      "loss": 0.5449,
      "step": 88
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9500785916724221,
      "learning_rate": 5.5799373040752355e-06,
      "loss": 0.4937,
      "step": 89
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8050066557512077,
      "learning_rate": 5.642633228840125e-06,
      "loss": 0.5415,
      "step": 90
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9384416079780085,
      "learning_rate": 5.705329153605016e-06,
      "loss": 0.5352,
      "step": 91
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8538078548549377,
      "learning_rate": 5.768025078369906e-06,
      "loss": 0.4541,
      "step": 92
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.0177715928511475,
      "learning_rate": 5.830721003134797e-06,
      "loss": 0.5034,
      "step": 93
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7155324899554252,
      "learning_rate": 5.8934169278996865e-06,
      "loss": 0.4019,
      "step": 94
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8989209007340702,
      "learning_rate": 5.956112852664577e-06,
      "loss": 0.5352,
      "step": 95
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8021000827752889,
      "learning_rate": 6.0188087774294676e-06,
      "loss": 0.5479,
      "step": 96
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9085732255906833,
      "learning_rate": 6.081504702194357e-06,
      "loss": 0.5073,
      "step": 97
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7553997591425767,
      "learning_rate": 6.144200626959248e-06,
      "loss": 0.4297,
      "step": 98
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8142543756233065,
      "learning_rate": 6.206896551724138e-06,
      "loss": 0.4941,
      "step": 99
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7356195476297875,
      "learning_rate": 6.269592476489029e-06,
      "loss": 0.4976,
      "step": 100
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6716549250632609,
      "learning_rate": 6.3322884012539185e-06,
      "loss": 0.5132,
      "step": 101
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.0637758288954107,
      "learning_rate": 6.394984326018809e-06,
      "loss": 0.6357,
      "step": 102
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9260077620827025,
      "learning_rate": 6.4576802507837e-06,
      "loss": 0.5444,
      "step": 103
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.908176325228822,
      "learning_rate": 6.520376175548589e-06,
      "loss": 0.4453,
      "step": 104
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.037743504803647,
      "learning_rate": 6.58307210031348e-06,
      "loss": 0.5195,
      "step": 105
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.80330925912525,
      "learning_rate": 6.64576802507837e-06,
      "loss": 0.4795,
      "step": 106
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9243446862379332,
      "learning_rate": 6.70846394984326e-06,
      "loss": 0.4814,
      "step": 107
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7599796718650476,
      "learning_rate": 6.771159874608151e-06,
      "loss": 0.4707,
      "step": 108
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8029045111563946,
      "learning_rate": 6.833855799373041e-06,
      "loss": 0.5454,
      "step": 109
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8458722549599031,
      "learning_rate": 6.896551724137932e-06,
      "loss": 0.4722,
      "step": 110
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.890994107546819,
      "learning_rate": 6.959247648902821e-06,
      "loss": 0.4619,
      "step": 111
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8466762691168837,
      "learning_rate": 7.021943573667712e-06,
      "loss": 0.4155,
      "step": 112
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.807052991025989,
      "learning_rate": 7.084639498432602e-06,
      "loss": 0.4321,
      "step": 113
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.1808245312762797,
      "learning_rate": 7.147335423197492e-06,
      "loss": 0.6064,
      "step": 114
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8614425490087254,
      "learning_rate": 7.210031347962383e-06,
      "loss": 0.4468,
      "step": 115
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8122830942327706,
      "learning_rate": 7.272727272727273e-06,
      "loss": 0.5049,
      "step": 116
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9659404204242565,
      "learning_rate": 7.3354231974921645e-06,
      "loss": 0.4937,
      "step": 117
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8825639900060311,
      "learning_rate": 7.398119122257053e-06,
      "loss": 0.5498,
      "step": 118
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7511362856697943,
      "learning_rate": 7.460815047021944e-06,
      "loss": 0.374,
      "step": 119
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.038894506607495,
      "learning_rate": 7.5235109717868345e-06,
      "loss": 0.4937,
      "step": 120
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.0535214577684298,
      "learning_rate": 7.586206896551724e-06,
      "loss": 0.5317,
      "step": 121
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9369385453958636,
      "learning_rate": 7.648902821316615e-06,
      "loss": 0.5103,
      "step": 122
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8835433312105536,
      "learning_rate": 7.711598746081506e-06,
      "loss": 0.5566,
      "step": 123
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7800735188660946,
      "learning_rate": 7.774294670846396e-06,
      "loss": 0.4199,
      "step": 124
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.3358576669895803,
      "learning_rate": 7.836990595611285e-06,
      "loss": 0.4927,
      "step": 125
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1196706028382306,
      "learning_rate": 7.899686520376177e-06,
      "loss": 0.5537,
      "step": 126
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.041683173863472,
      "learning_rate": 7.962382445141067e-06,
      "loss": 0.5591,
      "step": 127
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9591571985834055,
      "learning_rate": 8.025078369905956e-06,
      "loss": 0.4888,
      "step": 128
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0570536862286066,
      "learning_rate": 8.087774294670848e-06,
      "loss": 0.5024,
      "step": 129
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1713451665101244,
      "learning_rate": 8.150470219435737e-06,
      "loss": 0.5479,
      "step": 130
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1178448137892647,
      "learning_rate": 8.213166144200627e-06,
      "loss": 0.5825,
      "step": 131
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8454615312536908,
      "learning_rate": 8.275862068965518e-06,
      "loss": 0.4536,
      "step": 132
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8074066814970438,
      "learning_rate": 8.338557993730408e-06,
      "loss": 0.4121,
      "step": 133
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0628440667324295,
      "learning_rate": 8.4012539184953e-06,
      "loss": 0.5063,
      "step": 134
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0597499243651427,
      "learning_rate": 8.463949843260189e-06,
      "loss": 0.5332,
      "step": 135
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1103955928599754,
      "learning_rate": 8.526645768025079e-06,
      "loss": 0.6128,
      "step": 136
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9076272502489735,
      "learning_rate": 8.58934169278997e-06,
      "loss": 0.3965,
      "step": 137
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.655676058749942,
      "learning_rate": 8.65203761755486e-06,
      "loss": 0.4814,
      "step": 138
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9279510724475084,
      "learning_rate": 8.71473354231975e-06,
      "loss": 0.4951,
      "step": 139
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9612736295573034,
      "learning_rate": 8.777429467084641e-06,
      "loss": 0.6255,
      "step": 140
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.964142571510337,
      "learning_rate": 8.84012539184953e-06,
      "loss": 0.4878,
      "step": 141
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8913028907754226,
      "learning_rate": 8.90282131661442e-06,
      "loss": 0.5459,
      "step": 142
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1434016580965785,
      "learning_rate": 8.965517241379312e-06,
      "loss": 0.5435,
      "step": 143
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.754327773171796,
      "learning_rate": 9.028213166144201e-06,
      "loss": 0.4072,
      "step": 144
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.906594341776699,
      "learning_rate": 9.090909090909091e-06,
      "loss": 0.4712,
      "step": 145
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8078490251228605,
      "learning_rate": 9.153605015673982e-06,
      "loss": 0.4849,
      "step": 146
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1256971155764974,
      "learning_rate": 9.216300940438872e-06,
      "loss": 0.4897,
      "step": 147
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7899915159750772,
      "learning_rate": 9.278996865203762e-06,
      "loss": 0.4155,
      "step": 148
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6234798487465936,
      "learning_rate": 9.341692789968653e-06,
      "loss": 0.4058,
      "step": 149
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8303633547315634,
      "learning_rate": 9.404388714733543e-06,
      "loss": 0.499,
      "step": 150
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7122181575673203,
      "learning_rate": 9.467084639498434e-06,
      "loss": 0.4771,
      "step": 151
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7892426547209423,
      "learning_rate": 9.529780564263324e-06,
      "loss": 0.4683,
      "step": 152
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.690564054446622,
      "learning_rate": 9.592476489028214e-06,
      "loss": 0.4907,
      "step": 153
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.728194741031277,
      "learning_rate": 9.655172413793105e-06,
      "loss": 0.4526,
      "step": 154
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7472735531705446,
      "learning_rate": 9.717868338557995e-06,
      "loss": 0.4536,
      "step": 155
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.2948941743497637,
      "learning_rate": 9.780564263322884e-06,
      "loss": 0.5615,
      "step": 156
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8400124479221855,
      "learning_rate": 9.843260188087776e-06,
      "loss": 0.4839,
      "step": 157
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1602605926295584,
      "learning_rate": 9.905956112852665e-06,
      "loss": 0.585,
      "step": 158
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9431823679655515,
      "learning_rate": 9.968652037617555e-06,
      "loss": 0.5552,
      "step": 159
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.902729270097455,
      "learning_rate": 1.0031347962382445e-05,
      "loss": 0.4556,
      "step": 160
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7637572481328014,
      "learning_rate": 1.0094043887147336e-05,
      "loss": 0.3911,
      "step": 161
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.2859024701662864,
      "learning_rate": 1.0156739811912226e-05,
      "loss": 0.4858,
      "step": 162
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8886922863280777,
      "learning_rate": 1.0219435736677117e-05,
      "loss": 0.4951,
      "step": 163
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7527159904245868,
      "learning_rate": 1.0282131661442007e-05,
      "loss": 0.4492,
      "step": 164
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9474489917621232,
      "learning_rate": 1.0344827586206898e-05,
      "loss": 0.4185,
      "step": 165
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8894545422457294,
      "learning_rate": 1.0407523510971788e-05,
      "loss": 0.4434,
      "step": 166
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.994404535729931,
      "learning_rate": 1.0470219435736678e-05,
      "loss": 0.5112,
      "step": 167
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.3295407568446937,
      "learning_rate": 1.0532915360501567e-05,
      "loss": 0.6006,
      "step": 168
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.771089521542643,
      "learning_rate": 1.0595611285266459e-05,
      "loss": 0.4688,
      "step": 169
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.932013640402231,
      "learning_rate": 1.0658307210031348e-05,
      "loss": 0.4648,
      "step": 170
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.316294127196496,
      "learning_rate": 1.072100313479624e-05,
      "loss": 0.5952,
      "step": 171
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.894492130263785,
      "learning_rate": 1.078369905956113e-05,
      "loss": 0.4775,
      "step": 172
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9598205568190752,
      "learning_rate": 1.0846394984326021e-05,
      "loss": 0.4893,
      "step": 173
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7509320544648568,
      "learning_rate": 1.0909090909090909e-05,
      "loss": 0.4844,
      "step": 174
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8058972369760768,
      "learning_rate": 1.09717868338558e-05,
      "loss": 0.4707,
      "step": 175
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7025052250557517,
      "learning_rate": 1.103448275862069e-05,
      "loss": 0.4146,
      "step": 176
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8029525232424075,
      "learning_rate": 1.1097178683385581e-05,
      "loss": 0.4556,
      "step": 177
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9707314348883833,
      "learning_rate": 1.1159874608150471e-05,
      "loss": 0.4692,
      "step": 178
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.0366039227435917,
      "learning_rate": 1.1222570532915362e-05,
      "loss": 0.499,
      "step": 179
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.1481796327832043,
      "learning_rate": 1.128526645768025e-05,
      "loss": 0.542,
      "step": 180
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8032610466950358,
      "learning_rate": 1.1347962382445142e-05,
      "loss": 0.5356,
      "step": 181
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7836907042597545,
      "learning_rate": 1.1410658307210031e-05,
      "loss": 0.5273,
      "step": 182
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.908120554462311,
      "learning_rate": 1.1473354231974923e-05,
      "loss": 0.48,
      "step": 183
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.89659344735304,
      "learning_rate": 1.1536050156739813e-05,
      "loss": 0.4878,
      "step": 184
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9772163682411097,
      "learning_rate": 1.1598746081504704e-05,
      "loss": 0.5073,
      "step": 185
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.764159473873641,
      "learning_rate": 1.1661442006269594e-05,
      "loss": 0.4419,
      "step": 186
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9006747035193419,
      "learning_rate": 1.1724137931034483e-05,
      "loss": 0.478,
      "step": 187
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.886094436817303,
      "learning_rate": 1.1786833855799373e-05,
      "loss": 0.5176,
      "step": 188
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.784458644436215,
      "learning_rate": 1.1849529780564264e-05,
      "loss": 0.458,
      "step": 189
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9949077757017175,
      "learning_rate": 1.1912225705329154e-05,
      "loss": 0.5811,
      "step": 190
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9748010592200882,
      "learning_rate": 1.1974921630094045e-05,
      "loss": 0.4678,
      "step": 191
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9498537182906828,
      "learning_rate": 1.2037617554858935e-05,
      "loss": 0.582,
      "step": 192
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5486885513684954,
      "learning_rate": 1.2100313479623827e-05,
      "loss": 0.4224,
      "step": 193
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8526386857613386,
      "learning_rate": 1.2163009404388715e-05,
      "loss": 0.5493,
      "step": 194
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7761367587425416,
      "learning_rate": 1.2225705329153606e-05,
      "loss": 0.5635,
      "step": 195
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9323413421333289,
      "learning_rate": 1.2288401253918496e-05,
      "loss": 0.4902,
      "step": 196
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.0013484940058115,
      "learning_rate": 1.2351097178683387e-05,
      "loss": 0.5063,
      "step": 197
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.0318696168440344,
      "learning_rate": 1.2413793103448277e-05,
      "loss": 0.542,
      "step": 198
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8390066570155683,
      "learning_rate": 1.2476489028213168e-05,
      "loss": 0.4678,
      "step": 199
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.0450792352727287,
      "learning_rate": 1.2539184952978058e-05,
      "loss": 0.6162,
      "step": 200
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.888356710451703,
      "learning_rate": 1.2601880877742947e-05,
      "loss": 0.4194,
      "step": 201
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8618452328945962,
      "learning_rate": 1.2664576802507837e-05,
      "loss": 0.5376,
      "step": 202
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.0911558211621672,
      "learning_rate": 1.2727272727272728e-05,
      "loss": 0.5161,
      "step": 203
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.865939928527285,
      "learning_rate": 1.2789968652037618e-05,
      "loss": 0.5415,
      "step": 204
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.958638349990764,
      "learning_rate": 1.285266457680251e-05,
      "loss": 0.5356,
      "step": 205
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.94057917615933,
      "learning_rate": 1.29153605015674e-05,
      "loss": 0.5239,
      "step": 206
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.732722906395586,
      "learning_rate": 1.297805642633229e-05,
      "loss": 0.4512,
      "step": 207
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.070791696942078,
      "learning_rate": 1.3040752351097179e-05,
      "loss": 0.5396,
      "step": 208
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.260464557665378,
      "learning_rate": 1.310344827586207e-05,
      "loss": 0.625,
      "step": 209
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.966037806602718,
      "learning_rate": 1.316614420062696e-05,
      "loss": 0.4868,
      "step": 210
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.037753952677378,
      "learning_rate": 1.3228840125391851e-05,
      "loss": 0.4761,
      "step": 211
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.871940022701378,
      "learning_rate": 1.329153605015674e-05,
      "loss": 0.4922,
      "step": 212
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6995408855743452,
      "learning_rate": 1.3354231974921632e-05,
      "loss": 0.4014,
      "step": 213
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6233094752651327,
      "learning_rate": 1.341692789968652e-05,
      "loss": 0.3706,
      "step": 214
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.157150093886172,
      "learning_rate": 1.3479623824451411e-05,
      "loss": 0.5952,
      "step": 215
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5765578257938673,
      "learning_rate": 1.3542319749216301e-05,
      "loss": 0.4189,
      "step": 216
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8183043261705307,
      "learning_rate": 1.3605015673981193e-05,
      "loss": 0.4531,
      "step": 217
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.0383653147858456,
      "learning_rate": 1.3667711598746082e-05,
      "loss": 0.5757,
      "step": 218
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.708525803988794,
      "learning_rate": 1.3730407523510974e-05,
      "loss": 0.4355,
      "step": 219
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7628572354127507,
      "learning_rate": 1.3793103448275863e-05,
      "loss": 0.499,
      "step": 220
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8770290119306023,
      "learning_rate": 1.3855799373040753e-05,
      "loss": 0.5015,
      "step": 221
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7278196700370163,
      "learning_rate": 1.3918495297805643e-05,
      "loss": 0.439,
      "step": 222
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9487352992211726,
      "learning_rate": 1.3981191222570534e-05,
      "loss": 0.5396,
      "step": 223
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8338518274842812,
      "learning_rate": 1.4043887147335424e-05,
      "loss": 0.4878,
      "step": 224
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7980856466417705,
      "learning_rate": 1.4106583072100315e-05,
      "loss": 0.3818,
      "step": 225
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9125405122165664,
      "learning_rate": 1.4169278996865205e-05,
      "loss": 0.541,
      "step": 226
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8419113609071995,
      "learning_rate": 1.4231974921630096e-05,
      "loss": 0.4883,
      "step": 227
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8221169412686136,
      "learning_rate": 1.4294670846394984e-05,
      "loss": 0.4727,
      "step": 228
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7400260498838747,
      "learning_rate": 1.4357366771159876e-05,
      "loss": 0.4722,
      "step": 229
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9564126668343087,
      "learning_rate": 1.4420062695924765e-05,
      "loss": 0.5991,
      "step": 230
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8212089119415515,
      "learning_rate": 1.4482758620689657e-05,
      "loss": 0.5054,
      "step": 231
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7311076188847947,
      "learning_rate": 1.4545454545454546e-05,
      "loss": 0.5239,
      "step": 232
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7185476679336318,
      "learning_rate": 1.4608150470219438e-05,
      "loss": 0.457,
      "step": 233
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.937639071272641,
      "learning_rate": 1.4670846394984329e-05,
      "loss": 0.5479,
      "step": 234
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6794762258600746,
      "learning_rate": 1.4733542319749217e-05,
      "loss": 0.4585,
      "step": 235
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8961853662116797,
      "learning_rate": 1.4796238244514107e-05,
      "loss": 0.4746,
      "step": 236
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.021189110538228,
      "learning_rate": 1.4858934169278998e-05,
      "loss": 0.5645,
      "step": 237
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7462957998023316,
      "learning_rate": 1.4921630094043888e-05,
      "loss": 0.5347,
      "step": 238
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.024973587228016,
      "learning_rate": 1.498432601880878e-05,
      "loss": 0.5957,
      "step": 239
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7845702682823694,
      "learning_rate": 1.5047021943573669e-05,
      "loss": 0.4624,
      "step": 240
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6878107536455798,
      "learning_rate": 1.5109717868338559e-05,
      "loss": 0.5879,
      "step": 241
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8797724233020245,
      "learning_rate": 1.5172413793103448e-05,
      "loss": 0.5898,
      "step": 242
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.889315066297107,
      "learning_rate": 1.523510971786834e-05,
      "loss": 0.5571,
      "step": 243
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.843398015270419,
      "learning_rate": 1.529780564263323e-05,
      "loss": 0.5312,
      "step": 244
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8902834777657538,
      "learning_rate": 1.536050156739812e-05,
      "loss": 0.5664,
      "step": 245
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5884417548496008,
      "learning_rate": 1.5423197492163012e-05,
      "loss": 0.394,
      "step": 246
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5556368276849346,
      "learning_rate": 1.5485893416927904e-05,
      "loss": 0.3872,
      "step": 247
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.9073941825947716,
      "learning_rate": 1.554858934169279e-05,
      "loss": 0.5781,
      "step": 248
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7556326367822597,
      "learning_rate": 1.561128526645768e-05,
      "loss": 0.4409,
      "step": 249
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.707317342671133,
      "learning_rate": 1.567398119122257e-05,
      "loss": 0.4771,
      "step": 250
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.882669714544834,
      "learning_rate": 1.5736677115987462e-05,
      "loss": 0.5283,
      "step": 251
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8237358857060046,
      "learning_rate": 1.5799373040752354e-05,
      "loss": 0.4512,
      "step": 252
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.678796378857523,
      "learning_rate": 1.586206896551724e-05,
      "loss": 0.4644,
      "step": 253
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.9239364190046764,
      "learning_rate": 1.5924764890282133e-05,
      "loss": 0.4653,
      "step": 254
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.86918103703998,
      "learning_rate": 1.598746081504702e-05,
      "loss": 0.4946,
      "step": 255
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8062998270604094,
      "learning_rate": 1.6050156739811912e-05,
      "loss": 0.4912,
      "step": 256
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.023396224626733,
      "learning_rate": 1.6112852664576804e-05,
      "loss": 0.5532,
      "step": 257
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6871832150383446,
      "learning_rate": 1.6175548589341695e-05,
      "loss": 0.4629,
      "step": 258
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.651379536225624,
      "learning_rate": 1.6238244514106583e-05,
      "loss": 0.5264,
      "step": 259
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7073176054819097,
      "learning_rate": 1.6300940438871475e-05,
      "loss": 0.5396,
      "step": 260
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6443638815102917,
      "learning_rate": 1.6363636363636366e-05,
      "loss": 0.4487,
      "step": 261
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7707814014753998,
      "learning_rate": 1.6426332288401254e-05,
      "loss": 0.5879,
      "step": 262
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.846753976545353,
      "learning_rate": 1.6489028213166145e-05,
      "loss": 0.4531,
      "step": 263
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8220677018370501,
      "learning_rate": 1.6551724137931037e-05,
      "loss": 0.4561,
      "step": 264
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6938900933374959,
      "learning_rate": 1.6614420062695925e-05,
      "loss": 0.4766,
      "step": 265
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8169800616514784,
      "learning_rate": 1.6677115987460816e-05,
      "loss": 0.5029,
      "step": 266
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7999663694437504,
      "learning_rate": 1.6739811912225707e-05,
      "loss": 0.4292,
      "step": 267
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.81235959847101,
      "learning_rate": 1.68025078369906e-05,
      "loss": 0.5439,
      "step": 268
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.0241159345772077,
      "learning_rate": 1.6865203761755487e-05,
      "loss": 0.5181,
      "step": 269
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.650097631279815,
      "learning_rate": 1.6927899686520378e-05,
      "loss": 0.4321,
      "step": 270
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8616340973051335,
      "learning_rate": 1.6990595611285266e-05,
      "loss": 0.5479,
      "step": 271
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6746358614336778,
      "learning_rate": 1.7053291536050158e-05,
      "loss": 0.5312,
      "step": 272
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6972787636358178,
      "learning_rate": 1.711598746081505e-05,
      "loss": 0.4614,
      "step": 273
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6616790637004555,
      "learning_rate": 1.717868338557994e-05,
      "loss": 0.4868,
      "step": 274
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8428964852160914,
      "learning_rate": 1.7241379310344828e-05,
      "loss": 0.5967,
      "step": 275
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.9784622264796836,
      "learning_rate": 1.730407523510972e-05,
      "loss": 0.5483,
      "step": 276
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.0235665563119087,
      "learning_rate": 1.7366771159874608e-05,
      "loss": 0.6221,
      "step": 277
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8384180083508583,
      "learning_rate": 1.74294670846395e-05,
      "loss": 0.5034,
      "step": 278
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.9060028659299673,
      "learning_rate": 1.749216300940439e-05,
      "loss": 0.5767,
      "step": 279
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.901641712489501,
      "learning_rate": 1.7554858934169282e-05,
      "loss": 0.6411,
      "step": 280
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8666321937162536,
      "learning_rate": 1.7617554858934173e-05,
      "loss": 0.5264,
      "step": 281
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.73999514834233,
      "learning_rate": 1.768025078369906e-05,
      "loss": 0.5029,
      "step": 282
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6304336899777285,
      "learning_rate": 1.774294670846395e-05,
      "loss": 0.4204,
      "step": 283
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6642578119421991,
      "learning_rate": 1.780564263322884e-05,
      "loss": 0.5381,
      "step": 284
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.780132662996848,
      "learning_rate": 1.7868338557993732e-05,
      "loss": 0.6426,
      "step": 285
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5240620265172975,
      "learning_rate": 1.7931034482758623e-05,
      "loss": 0.4326,
      "step": 286
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.772344429967332,
      "learning_rate": 1.7993730407523515e-05,
      "loss": 0.5049,
      "step": 287
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.805685834544162,
      "learning_rate": 1.8056426332288403e-05,
      "loss": 0.5444,
      "step": 288
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6450394506581576,
      "learning_rate": 1.811912225705329e-05,
      "loss": 0.5029,
      "step": 289
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.9340712964783056,
      "learning_rate": 1.8181818181818182e-05,
      "loss": 0.5557,
      "step": 290
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8045345752243789,
      "learning_rate": 1.8244514106583073e-05,
      "loss": 0.406,
      "step": 291
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8749494693984174,
      "learning_rate": 1.8307210031347965e-05,
      "loss": 0.5371,
      "step": 292
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8300544752875916,
      "learning_rate": 1.8369905956112856e-05,
      "loss": 0.5874,
      "step": 293
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5875656678236558,
      "learning_rate": 1.8432601880877744e-05,
      "loss": 0.4463,
      "step": 294
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6522542190601486,
      "learning_rate": 1.8495297805642636e-05,
      "loss": 0.5283,
      "step": 295
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.680468873544775,
      "learning_rate": 1.8557993730407524e-05,
      "loss": 0.478,
      "step": 296
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.9105612498703688,
      "learning_rate": 1.8620689655172415e-05,
      "loss": 0.5249,
      "step": 297
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6182232605503126,
      "learning_rate": 1.8683385579937306e-05,
      "loss": 0.4961,
      "step": 298
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5853677700983764,
      "learning_rate": 1.8746081504702194e-05,
      "loss": 0.4878,
      "step": 299
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8279366365325727,
      "learning_rate": 1.8808777429467086e-05,
      "loss": 0.4995,
      "step": 300
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6176240112739193,
      "learning_rate": 1.8871473354231977e-05,
      "loss": 0.4023,
      "step": 301
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.663183913335847,
      "learning_rate": 1.893416927899687e-05,
      "loss": 0.4609,
      "step": 302
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.945927560252233,
      "learning_rate": 1.8996865203761756e-05,
      "loss": 0.5337,
      "step": 303
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7616140825249242,
      "learning_rate": 1.9059561128526648e-05,
      "loss": 0.5195,
      "step": 304
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.647566957724041,
      "learning_rate": 1.9122257053291536e-05,
      "loss": 0.4839,
      "step": 305
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6324150359935408,
      "learning_rate": 1.9184952978056427e-05,
      "loss": 0.5225,
      "step": 306
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7066596984581384,
      "learning_rate": 1.924764890282132e-05,
      "loss": 0.6318,
      "step": 307
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6602039166996816,
      "learning_rate": 1.931034482758621e-05,
      "loss": 0.4619,
      "step": 308
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8174823824368722,
      "learning_rate": 1.9373040752351098e-05,
      "loss": 0.5791,
      "step": 309
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.506237008708466,
      "learning_rate": 1.943573667711599e-05,
      "loss": 0.5103,
      "step": 310
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.429272749671718,
      "learning_rate": 1.9498432601880877e-05,
      "loss": 0.3833,
      "step": 311
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5426583943406786,
      "learning_rate": 1.956112852664577e-05,
      "loss": 0.3882,
      "step": 312
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.689135080346264,
      "learning_rate": 1.962382445141066e-05,
      "loss": 0.418,
      "step": 313
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6483684254539837,
      "learning_rate": 1.968652037617555e-05,
      "loss": 0.5083,
      "step": 314
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.9968698720887912,
      "learning_rate": 1.9749216300940443e-05,
      "loss": 0.5356,
      "step": 315
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.903941761841389,
      "learning_rate": 1.981191222570533e-05,
      "loss": 0.5332,
      "step": 316
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7729431748310827,
      "learning_rate": 1.987460815047022e-05,
      "loss": 0.5254,
      "step": 317
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8331332335499708,
      "learning_rate": 1.993730407523511e-05,
      "loss": 0.5513,
      "step": 318
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8711277163313713,
      "learning_rate": 2e-05,
      "loss": 0.5811,
      "step": 319
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4964762285065774,
      "learning_rate": 1.999999953520864e-05,
      "loss": 0.4561,
      "step": 320
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.9112025950286362,
      "learning_rate": 1.9999998140834606e-05,
      "loss": 0.4521,
      "step": 321
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6879635842982812,
      "learning_rate": 1.9999995816878023e-05,
      "loss": 0.5664,
      "step": 322
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6900071470365563,
      "learning_rate": 1.9999992563339106e-05,
      "loss": 0.5205,
      "step": 323
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6482111576774154,
      "learning_rate": 1.9999988380218164e-05,
      "loss": 0.4922,
      "step": 324
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.9414874108250397,
      "learning_rate": 1.9999983267515578e-05,
      "loss": 0.5703,
      "step": 325
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.900405128570437,
      "learning_rate": 1.9999977225231833e-05,
      "loss": 0.625,
      "step": 326
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8437482551775326,
      "learning_rate": 1.999997025336748e-05,
      "loss": 0.563,
      "step": 327
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5838463053527982,
      "learning_rate": 1.9999962351923176e-05,
      "loss": 0.4937,
      "step": 328
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7300327756588703,
      "learning_rate": 1.999995352089965e-05,
      "loss": 0.6172,
      "step": 329
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7963153911790948,
      "learning_rate": 1.9999943760297725e-05,
      "loss": 0.564,
      "step": 330
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.686174081429555,
      "learning_rate": 1.999993307011831e-05,
      "loss": 0.5537,
      "step": 331
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7881195046139318,
      "learning_rate": 1.9999921450362392e-05,
      "loss": 0.5518,
      "step": 332
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5712529300516382,
      "learning_rate": 1.999990890103106e-05,
      "loss": 0.4443,
      "step": 333
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.712095103726284,
      "learning_rate": 1.9999895422125477e-05,
      "loss": 0.5669,
      "step": 334
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6514029275759141,
      "learning_rate": 1.9999881013646893e-05,
      "loss": 0.4497,
      "step": 335
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4532041932347768,
      "learning_rate": 1.999986567559665e-05,
      "loss": 0.4199,
      "step": 336
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7069859016148823,
      "learning_rate": 1.9999849407976176e-05,
      "loss": 0.5137,
      "step": 337
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6373183474397914,
      "learning_rate": 1.9999832210786977e-05,
      "loss": 0.4077,
      "step": 338
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.520890598681407,
      "learning_rate": 1.999981408403066e-05,
      "loss": 0.4238,
      "step": 339
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.706437967266896,
      "learning_rate": 1.99997950277089e-05,
      "loss": 0.4731,
      "step": 340
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7181551536196362,
      "learning_rate": 1.9999775041823478e-05,
      "loss": 0.5415,
      "step": 341
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7901976061019116,
      "learning_rate": 1.9999754126376247e-05,
      "loss": 0.5483,
      "step": 342
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7139877762095423,
      "learning_rate": 1.999973228136915e-05,
      "loss": 0.5781,
      "step": 343
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.066630382444348,
      "learning_rate": 1.9999709506804223e-05,
      "loss": 0.6338,
      "step": 344
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6919034873297285,
      "learning_rate": 1.999968580268358e-05,
      "loss": 0.4492,
      "step": 345
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.493813394409096,
      "learning_rate": 1.999966116900942e-05,
      "loss": 0.4785,
      "step": 346
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6401713298780496,
      "learning_rate": 1.9999635605784042e-05,
      "loss": 0.5713,
      "step": 347
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6095336409699277,
      "learning_rate": 1.9999609113009817e-05,
      "loss": 0.4673,
      "step": 348
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7162022317777648,
      "learning_rate": 1.9999581690689204e-05,
      "loss": 0.5073,
      "step": 349
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3482125585579654,
      "learning_rate": 1.999955333882476e-05,
      "loss": 0.3892,
      "step": 350
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5279144611817468,
      "learning_rate": 1.9999524057419116e-05,
      "loss": 0.501,
      "step": 351
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5421242356811362,
      "learning_rate": 1.9999493846474995e-05,
      "loss": 0.4438,
      "step": 352
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5937891045257986,
      "learning_rate": 1.9999462705995206e-05,
      "loss": 0.4043,
      "step": 353
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7622929120454616,
      "learning_rate": 1.9999430635982643e-05,
      "loss": 0.5708,
      "step": 354
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6106184493182458,
      "learning_rate": 1.9999397636440284e-05,
      "loss": 0.4663,
      "step": 355
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6204923672478087,
      "learning_rate": 1.9999363707371203e-05,
      "loss": 0.4897,
      "step": 356
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6497604453038568,
      "learning_rate": 1.999932884877855e-05,
      "loss": 0.437,
      "step": 357
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4596281940374551,
      "learning_rate": 1.9999293060665565e-05,
      "loss": 0.4153,
      "step": 358
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.773715297812362,
      "learning_rate": 1.9999256343035577e-05,
      "loss": 0.5659,
      "step": 359
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6748162999818443,
      "learning_rate": 1.9999218695892e-05,
      "loss": 0.4927,
      "step": 360
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6005869451330348,
      "learning_rate": 1.9999180119238327e-05,
      "loss": 0.4507,
      "step": 361
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8270664143817201,
      "learning_rate": 1.9999140613078154e-05,
      "loss": 0.6138,
      "step": 362
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6869496916717341,
      "learning_rate": 1.9999100177415143e-05,
      "loss": 0.5327,
      "step": 363
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6259552544131068,
      "learning_rate": 1.999905881225306e-05,
      "loss": 0.5405,
      "step": 364
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.411342454443797,
      "learning_rate": 1.9999016517595752e-05,
      "loss": 0.4697,
      "step": 365
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.402442881222897,
      "learning_rate": 1.9998973293447143e-05,
      "loss": 0.4512,
      "step": 366
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4227052830957307,
      "learning_rate": 1.9998929139811257e-05,
      "loss": 0.4409,
      "step": 367
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5923244717876555,
      "learning_rate": 1.9998884056692195e-05,
      "loss": 0.4658,
      "step": 368
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6678174889240536,
      "learning_rate": 1.999883804409415e-05,
      "loss": 0.5635,
      "step": 369
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6384338843461632,
      "learning_rate": 1.9998791102021396e-05,
      "loss": 0.5278,
      "step": 370
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7293141323080579,
      "learning_rate": 1.99987432304783e-05,
      "loss": 0.5356,
      "step": 371
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3286317325186037,
      "learning_rate": 1.9998694429469315e-05,
      "loss": 0.4199,
      "step": 372
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5451725003417762,
      "learning_rate": 1.999864469899897e-05,
      "loss": 0.5303,
      "step": 373
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5692283596798156,
      "learning_rate": 1.9998594039071894e-05,
      "loss": 0.5361,
      "step": 374
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5560459399354958,
      "learning_rate": 1.9998542449692794e-05,
      "loss": 0.5259,
      "step": 375
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5921644960150632,
      "learning_rate": 1.999848993086646e-05,
      "loss": 0.5063,
      "step": 376
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.580527340660811,
      "learning_rate": 1.9998436482597784e-05,
      "loss": 0.4414,
      "step": 377
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6157167960986585,
      "learning_rate": 1.9998382104891728e-05,
      "loss": 0.5317,
      "step": 378
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4033825280019678,
      "learning_rate": 1.9998326797753352e-05,
      "loss": 0.4614,
      "step": 379
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5996971444699404,
      "learning_rate": 1.999827056118779e-05,
      "loss": 0.4849,
      "step": 380
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.466557767996865,
      "learning_rate": 1.9998213395200277e-05,
      "loss": 0.4785,
      "step": 381
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4578185343633587,
      "learning_rate": 1.9998155299796122e-05,
      "loss": 0.3945,
      "step": 382
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4183366740563634,
      "learning_rate": 1.999809627498073e-05,
      "loss": 0.5005,
      "step": 383
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6070860058448246,
      "learning_rate": 1.9998036320759582e-05,
      "loss": 0.5327,
      "step": 384
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5020012197098862,
      "learning_rate": 1.9997975437138256e-05,
      "loss": 0.4521,
      "step": 385
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6277416452937208,
      "learning_rate": 1.999791362412241e-05,
      "loss": 0.5806,
      "step": 386
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5378755487570124,
      "learning_rate": 1.9997850881717788e-05,
      "loss": 0.4448,
      "step": 387
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4518092214672391,
      "learning_rate": 1.9997787209930222e-05,
      "loss": 0.3911,
      "step": 388
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0436898973039965,
      "learning_rate": 1.999772260876564e-05,
      "loss": 0.5698,
      "step": 389
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4765194987848635,
      "learning_rate": 1.9997657078230034e-05,
      "loss": 0.4561,
      "step": 390
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6384032281579854,
      "learning_rate": 1.9997590618329507e-05,
      "loss": 0.5518,
      "step": 391
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.665611154918358,
      "learning_rate": 1.999752322907023e-05,
      "loss": 0.4746,
      "step": 392
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.623847241962191,
      "learning_rate": 1.999745491045847e-05,
      "loss": 0.5542,
      "step": 393
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.611112592679862,
      "learning_rate": 1.9997385662500577e-05,
      "loss": 0.5122,
      "step": 394
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5198912174812218,
      "learning_rate": 1.999731548520299e-05,
      "loss": 0.4712,
      "step": 395
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4033316289055529,
      "learning_rate": 1.9997244378572227e-05,
      "loss": 0.4258,
      "step": 396
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.42076489742244,
      "learning_rate": 1.9997172342614902e-05,
      "loss": 0.4038,
      "step": 397
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6643206846988874,
      "learning_rate": 1.9997099377337714e-05,
      "loss": 0.5713,
      "step": 398
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7231929125099428,
      "learning_rate": 1.999702548274744e-05,
      "loss": 0.4761,
      "step": 399
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6897154033276929,
      "learning_rate": 1.9996950658850956e-05,
      "loss": 0.5483,
      "step": 400
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8160872956477558,
      "learning_rate": 1.9996874905655214e-05,
      "loss": 0.5103,
      "step": 401
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6040040155633168,
      "learning_rate": 1.999679822316725e-05,
      "loss": 0.5479,
      "step": 402
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5369713388061474,
      "learning_rate": 1.99967206113942e-05,
      "loss": 0.5244,
      "step": 403
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4535288919257459,
      "learning_rate": 1.999664207034328e-05,
      "loss": 0.4419,
      "step": 404
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6206848966531935,
      "learning_rate": 1.9996562600021786e-05,
      "loss": 0.4688,
      "step": 405
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5102181013294833,
      "learning_rate": 1.9996482200437104e-05,
      "loss": 0.4229,
      "step": 406
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6436552936115683,
      "learning_rate": 1.999640087159671e-05,
      "loss": 0.4795,
      "step": 407
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3094400642370503,
      "learning_rate": 1.999631861350817e-05,
      "loss": 0.4038,
      "step": 408
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3979984745625296,
      "learning_rate": 1.999623542617912e-05,
      "loss": 0.4302,
      "step": 409
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6036797480809681,
      "learning_rate": 1.99961513096173e-05,
      "loss": 0.4761,
      "step": 410
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4764256395269906,
      "learning_rate": 1.9996066263830533e-05,
      "loss": 0.4595,
      "step": 411
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5240639319563671,
      "learning_rate": 1.9995980288826714e-05,
      "loss": 0.4956,
      "step": 412
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3679044120060413,
      "learning_rate": 1.999589338461384e-05,
      "loss": 0.4424,
      "step": 413
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7101720144453696,
      "learning_rate": 1.9995805551199988e-05,
      "loss": 0.5083,
      "step": 414
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6388058723271897,
      "learning_rate": 1.999571678859333e-05,
      "loss": 0.4692,
      "step": 415
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5306525934932445,
      "learning_rate": 1.999562709680211e-05,
      "loss": 0.4756,
      "step": 416
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4994285577094324,
      "learning_rate": 1.9995536475834667e-05,
      "loss": 0.4683,
      "step": 417
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5236103669749108,
      "learning_rate": 1.9995444925699427e-05,
      "loss": 0.458,
      "step": 418
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7589427107138509,
      "learning_rate": 1.99953524464049e-05,
      "loss": 0.6133,
      "step": 419
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6797234659963485,
      "learning_rate": 1.999525903795968e-05,
      "loss": 0.5332,
      "step": 420
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6007937849121396,
      "learning_rate": 1.9995164700372448e-05,
      "loss": 0.5073,
      "step": 421
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6217028262188353,
      "learning_rate": 1.9995069433651985e-05,
      "loss": 0.4951,
      "step": 422
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.926440145288664,
      "learning_rate": 1.9994973237807133e-05,
      "loss": 0.583,
      "step": 423
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6460950977520106,
      "learning_rate": 1.999487611284684e-05,
      "loss": 0.626,
      "step": 424
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4969215524142199,
      "learning_rate": 1.9994778058780136e-05,
      "loss": 0.5649,
      "step": 425
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4628682816514367,
      "learning_rate": 1.999467907561614e-05,
      "loss": 0.4199,
      "step": 426
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4281189204162543,
      "learning_rate": 1.999457916336404e-05,
      "loss": 0.4443,
      "step": 427
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5134476182839527,
      "learning_rate": 1.9994478322033134e-05,
      "loss": 0.519,
      "step": 428
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.43788440283071,
      "learning_rate": 1.9994376551632796e-05,
      "loss": 0.46,
      "step": 429
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4752709602470842,
      "learning_rate": 1.9994273852172484e-05,
      "loss": 0.3965,
      "step": 430
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5330347624488943,
      "learning_rate": 1.999417022366174e-05,
      "loss": 0.5195,
      "step": 431
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.402354718087985,
      "learning_rate": 1.9994065666110206e-05,
      "loss": 0.4619,
      "step": 432
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.559871723909221,
      "learning_rate": 1.99939601795276e-05,
      "loss": 0.604,
      "step": 433
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7904627134860422,
      "learning_rate": 1.9993853763923724e-05,
      "loss": 0.519,
      "step": 434
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5607252603311672,
      "learning_rate": 1.999374641930847e-05,
      "loss": 0.4736,
      "step": 435
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5453245370338478,
      "learning_rate": 1.9993638145691817e-05,
      "loss": 0.5347,
      "step": 436
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7539362617154786,
      "learning_rate": 1.999352894308383e-05,
      "loss": 0.5249,
      "step": 437
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5113467125178164,
      "learning_rate": 1.9993418811494663e-05,
      "loss": 0.5161,
      "step": 438
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.782014465318087,
      "learning_rate": 1.9993307750934555e-05,
      "loss": 0.48,
      "step": 439
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5690232513212532,
      "learning_rate": 1.9993195761413823e-05,
      "loss": 0.5161,
      "step": 440
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4728582185364028,
      "learning_rate": 1.9993082842942883e-05,
      "loss": 0.3979,
      "step": 441
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3579363701024922,
      "learning_rate": 1.9992968995532228e-05,
      "loss": 0.3677,
      "step": 442
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5491064017062564,
      "learning_rate": 1.999285421919245e-05,
      "loss": 0.5161,
      "step": 443
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4960721160414894,
      "learning_rate": 1.9992738513934205e-05,
      "loss": 0.4717,
      "step": 444
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.548686181230436,
      "learning_rate": 1.9992621879768256e-05,
      "loss": 0.4507,
      "step": 445
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7483871625424023,
      "learning_rate": 1.9992504316705446e-05,
      "loss": 0.5234,
      "step": 446
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5330824278223798,
      "learning_rate": 1.99923858247567e-05,
      "loss": 0.479,
      "step": 447
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.616236815559328,
      "learning_rate": 1.9992266403933037e-05,
      "loss": 0.481,
      "step": 448
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5633305324080298,
      "learning_rate": 1.9992146054245552e-05,
      "loss": 0.4829,
      "step": 449
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4563630632083757,
      "learning_rate": 1.999202477570544e-05,
      "loss": 0.4399,
      "step": 450
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8237700698918915,
      "learning_rate": 1.9991902568323972e-05,
      "loss": 0.6499,
      "step": 451
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7337354186728475,
      "learning_rate": 1.9991779432112503e-05,
      "loss": 0.5112,
      "step": 452
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4600819149791386,
      "learning_rate": 1.9991655367082487e-05,
      "loss": 0.5815,
      "step": 453
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3477881098500668,
      "learning_rate": 1.999153037324545e-05,
      "loss": 0.4272,
      "step": 454
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7423237156515956,
      "learning_rate": 1.999140445061302e-05,
      "loss": 0.5806,
      "step": 455
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6981022209106569,
      "learning_rate": 1.9991277599196897e-05,
      "loss": 0.5239,
      "step": 456
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3437857736310783,
      "learning_rate": 1.999114981900887e-05,
      "loss": 0.4326,
      "step": 457
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6777912005611682,
      "learning_rate": 1.9991021110060825e-05,
      "loss": 0.5688,
      "step": 458
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6526974350487134,
      "learning_rate": 1.999089147236472e-05,
      "loss": 0.5361,
      "step": 459
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.605546911247771,
      "learning_rate": 1.9990760905932605e-05,
      "loss": 0.5474,
      "step": 460
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5420090009531073,
      "learning_rate": 1.9990629410776624e-05,
      "loss": 0.5527,
      "step": 461
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6374380764726255,
      "learning_rate": 1.9990496986908994e-05,
      "loss": 0.5693,
      "step": 462
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.551331270822273,
      "learning_rate": 1.9990363634342032e-05,
      "loss": 0.5239,
      "step": 463
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6651813973013292,
      "learning_rate": 1.9990229353088123e-05,
      "loss": 0.665,
      "step": 464
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6673893623538176,
      "learning_rate": 1.9990094143159763e-05,
      "loss": 0.6582,
      "step": 465
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.583210150131019,
      "learning_rate": 1.9989958004569514e-05,
      "loss": 0.5122,
      "step": 466
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5448177327734416,
      "learning_rate": 1.9989820937330028e-05,
      "loss": 0.502,
      "step": 467
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3352274155454866,
      "learning_rate": 1.998968294145405e-05,
      "loss": 0.459,
      "step": 468
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.845802444035111,
      "learning_rate": 1.9989544016954414e-05,
      "loss": 0.5898,
      "step": 469
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6228410429963986,
      "learning_rate": 1.9989404163844026e-05,
      "loss": 0.5903,
      "step": 470
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3016334519925992,
      "learning_rate": 1.9989263382135882e-05,
      "loss": 0.4321,
      "step": 471
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.48643291003588,
      "learning_rate": 1.998912167184308e-05,
      "loss": 0.5435,
      "step": 472
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4812628005619872,
      "learning_rate": 1.998897903297879e-05,
      "loss": 0.4634,
      "step": 473
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5480288221340428,
      "learning_rate": 1.998883546555627e-05,
      "loss": 0.5532,
      "step": 474
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4509373196410293,
      "learning_rate": 1.9988690969588867e-05,
      "loss": 0.5073,
      "step": 475
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.365595283101366,
      "learning_rate": 1.998854554509001e-05,
      "loss": 0.457,
      "step": 476
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5251633017986141,
      "learning_rate": 1.998839919207322e-05,
      "loss": 0.541,
      "step": 477
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4135375643204238,
      "learning_rate": 1.9988251910552103e-05,
      "loss": 0.3994,
      "step": 478
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4734286358671957,
      "learning_rate": 1.9988103700540345e-05,
      "loss": 0.4766,
      "step": 479
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5325665881259969,
      "learning_rate": 1.9987954562051724e-05,
      "loss": 0.563,
      "step": 480
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4846158670347254,
      "learning_rate": 1.9987804495100112e-05,
      "loss": 0.5396,
      "step": 481
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7647791898562728,
      "learning_rate": 1.998765349969945e-05,
      "loss": 0.6587,
      "step": 482
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7310798798036746,
      "learning_rate": 1.9987501575863776e-05,
      "loss": 0.623,
      "step": 483
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.500385970029389,
      "learning_rate": 1.9987348723607218e-05,
      "loss": 0.4048,
      "step": 484
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7362199048564129,
      "learning_rate": 1.998719494294398e-05,
      "loss": 0.5347,
      "step": 485
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.829011071176407,
      "learning_rate": 1.9987040233888355e-05,
      "loss": 0.5039,
      "step": 486
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6370845731881438,
      "learning_rate": 1.998688459645473e-05,
      "loss": 0.5654,
      "step": 487
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6073664809980979,
      "learning_rate": 1.998672803065757e-05,
      "loss": 0.4741,
      "step": 488
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8235768142711108,
      "learning_rate": 1.998657053651143e-05,
      "loss": 0.6426,
      "step": 489
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7361910169173773,
      "learning_rate": 1.998641211403095e-05,
      "loss": 0.5513,
      "step": 490
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.667979258190254,
      "learning_rate": 1.9986252763230856e-05,
      "loss": 0.5835,
      "step": 491
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.745884708739697,
      "learning_rate": 1.998609248412596e-05,
      "loss": 0.6104,
      "step": 492
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2945292031526974,
      "learning_rate": 1.9985931276731162e-05,
      "loss": 0.4409,
      "step": 493
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5059912108021898,
      "learning_rate": 1.998576914106145e-05,
      "loss": 0.5894,
      "step": 494
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4792152282410982,
      "learning_rate": 1.9985606077131895e-05,
      "loss": 0.4902,
      "step": 495
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4462601299679776,
      "learning_rate": 1.9985442084957657e-05,
      "loss": 0.4336,
      "step": 496
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5631022390479647,
      "learning_rate": 1.9985277164553972e-05,
      "loss": 0.5093,
      "step": 497
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6398939322305879,
      "learning_rate": 1.9985111315936177e-05,
      "loss": 0.6284,
      "step": 498
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2570845470323695,
      "learning_rate": 1.998494453911969e-05,
      "loss": 0.3994,
      "step": 499
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4025277137709862,
      "learning_rate": 1.9984776834120015e-05,
      "loss": 0.4399,
      "step": 500
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4883219602094895,
      "learning_rate": 1.9984608200952736e-05,
      "loss": 0.5562,
      "step": 501
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3931630348985937,
      "learning_rate": 1.9984438639633534e-05,
      "loss": 0.4243,
      "step": 502
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3986931902152848,
      "learning_rate": 1.998426815017817e-05,
      "loss": 0.4512,
      "step": 503
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.560477306858998,
      "learning_rate": 1.9984096732602485e-05,
      "loss": 0.5444,
      "step": 504
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5093680883291258,
      "learning_rate": 1.9983924386922427e-05,
      "loss": 0.4624,
      "step": 505
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4689926526733654,
      "learning_rate": 1.998375111315401e-05,
      "loss": 0.4478,
      "step": 506
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.484419913297081,
      "learning_rate": 1.998357691131334e-05,
      "loss": 0.4712,
      "step": 507
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5366732107885184,
      "learning_rate": 1.998340178141661e-05,
      "loss": 0.5249,
      "step": 508
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5899178165404961,
      "learning_rate": 1.9983225723480104e-05,
      "loss": 0.5815,
      "step": 509
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.472362119251607,
      "learning_rate": 1.9983048737520186e-05,
      "loss": 0.457,
      "step": 510
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5324287683777682,
      "learning_rate": 1.998287082355331e-05,
      "loss": 0.481,
      "step": 511
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.437652337397896,
      "learning_rate": 1.998269198159601e-05,
      "loss": 0.4536,
      "step": 512
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7171094674168979,
      "learning_rate": 1.9982512211664914e-05,
      "loss": 0.5317,
      "step": 513
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.42815441603342,
      "learning_rate": 1.9982331513776733e-05,
      "loss": 0.4565,
      "step": 514
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5763188918391096,
      "learning_rate": 1.9982149887948264e-05,
      "loss": 0.4214,
      "step": 515
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5115698346489592,
      "learning_rate": 1.998196733419639e-05,
      "loss": 0.5493,
      "step": 516
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.426445739140189,
      "learning_rate": 1.998178385253808e-05,
      "loss": 0.4641,
      "step": 517
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2785706282194595,
      "learning_rate": 1.9981599442990397e-05,
      "loss": 0.3975,
      "step": 518
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3574331049465373,
      "learning_rate": 1.9981414105570473e-05,
      "loss": 0.418,
      "step": 519
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4970784118068157,
      "learning_rate": 1.9981227840295544e-05,
      "loss": 0.4619,
      "step": 520
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6049830634681108,
      "learning_rate": 1.9981040647182923e-05,
      "loss": 0.5234,
      "step": 521
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4172629094362656,
      "learning_rate": 1.998085252625001e-05,
      "loss": 0.4536,
      "step": 522
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5680635516337798,
      "learning_rate": 1.9980663477514294e-05,
      "loss": 0.5225,
      "step": 523
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.470018750853335,
      "learning_rate": 1.9980473500993346e-05,
      "loss": 0.4946,
      "step": 524
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4486680363527842,
      "learning_rate": 1.9980282596704828e-05,
      "loss": 0.4141,
      "step": 525
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.713238525919704,
      "learning_rate": 1.9980090764666486e-05,
      "loss": 0.5732,
      "step": 526
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5646469524824356,
      "learning_rate": 1.997989800489615e-05,
      "loss": 0.5396,
      "step": 527
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3765052455744407,
      "learning_rate": 1.9979704317411745e-05,
      "loss": 0.4351,
      "step": 528
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5933988880554875,
      "learning_rate": 1.997950970223127e-05,
      "loss": 0.5132,
      "step": 529
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3316994371493966,
      "learning_rate": 1.9979314159372815e-05,
      "loss": 0.4697,
      "step": 530
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4632348751859219,
      "learning_rate": 1.9979117688854565e-05,
      "loss": 0.4893,
      "step": 531
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5582672853432127,
      "learning_rate": 1.9978920290694776e-05,
      "loss": 0.5474,
      "step": 532
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5030135142222265,
      "learning_rate": 1.99787219649118e-05,
      "loss": 0.5684,
      "step": 533
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4230181810706906,
      "learning_rate": 1.9978522711524076e-05,
      "loss": 0.4824,
      "step": 534
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7026311057599557,
      "learning_rate": 1.997832253055012e-05,
      "loss": 0.6182,
      "step": 535
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6060295002563518,
      "learning_rate": 1.9978121422008547e-05,
      "loss": 0.4756,
      "step": 536
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4260012297608853,
      "learning_rate": 1.997791938591805e-05,
      "loss": 0.5166,
      "step": 537
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.406690627068418,
      "learning_rate": 1.9977716422297404e-05,
      "loss": 0.4775,
      "step": 538
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5434045136123276,
      "learning_rate": 1.9977512531165484e-05,
      "loss": 0.4941,
      "step": 539
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3542768143552273,
      "learning_rate": 1.997730771254124e-05,
      "loss": 0.481,
      "step": 540
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3657794297547423,
      "learning_rate": 1.9977101966443713e-05,
      "loss": 0.4473,
      "step": 541
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1513455344640724,
      "learning_rate": 1.9976895292892028e-05,
      "loss": 0.3418,
      "step": 542
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3423030612921667,
      "learning_rate": 1.9976687691905394e-05,
      "loss": 0.4365,
      "step": 543
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.443262143583192,
      "learning_rate": 1.9976479163503112e-05,
      "loss": 0.4585,
      "step": 544
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3440451656011425,
      "learning_rate": 1.997626970770457e-05,
      "loss": 0.4097,
      "step": 545
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3225120451733088,
      "learning_rate": 1.997605932452923e-05,
      "loss": 0.396,
      "step": 546
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5261068717402149,
      "learning_rate": 1.9975848013996658e-05,
      "loss": 0.4707,
      "step": 547
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6472474059949334,
      "learning_rate": 1.9975635776126492e-05,
      "loss": 0.6118,
      "step": 548
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4490998418473824,
      "learning_rate": 1.9975422610938463e-05,
      "loss": 0.4619,
      "step": 549
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4886577850196165,
      "learning_rate": 1.9975208518452384e-05,
      "loss": 0.4492,
      "step": 550
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5794114969063617,
      "learning_rate": 1.997499349868816e-05,
      "loss": 0.5376,
      "step": 551
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.556200064953959,
      "learning_rate": 1.9974777551665773e-05,
      "loss": 0.4819,
      "step": 552
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4772334555189368,
      "learning_rate": 1.9974560677405307e-05,
      "loss": 0.4624,
      "step": 553
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5489066068150599,
      "learning_rate": 1.9974342875926913e-05,
      "loss": 0.4546,
      "step": 554
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5647190560689617,
      "learning_rate": 1.997412414725084e-05,
      "loss": 0.4546,
      "step": 555
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3357352078169606,
      "learning_rate": 1.9973904491397426e-05,
      "loss": 0.4414,
      "step": 556
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5515603421735404,
      "learning_rate": 1.997368390838708e-05,
      "loss": 0.4917,
      "step": 557
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3255068643329826,
      "learning_rate": 1.9973462398240316e-05,
      "loss": 0.4482,
      "step": 558
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5722686803203894,
      "learning_rate": 1.997323996097772e-05,
      "loss": 0.5112,
      "step": 559
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6149951293790745,
      "learning_rate": 1.9973016596619973e-05,
      "loss": 0.585,
      "step": 560
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6985763439128798,
      "learning_rate": 1.997279230518784e-05,
      "loss": 0.5322,
      "step": 561
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5608586947307654,
      "learning_rate": 1.9972567086702163e-05,
      "loss": 0.5254,
      "step": 562
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.49814927000226,
      "learning_rate": 1.997234094118388e-05,
      "loss": 0.4326,
      "step": 563
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.480160181147905,
      "learning_rate": 1.9972113868654016e-05,
      "loss": 0.5332,
      "step": 564
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6995247536227285,
      "learning_rate": 1.9971885869133683e-05,
      "loss": 0.6504,
      "step": 565
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3334147942483932,
      "learning_rate": 1.9971656942644068e-05,
      "loss": 0.4253,
      "step": 566
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5370628341221133,
      "learning_rate": 1.9971427089206458e-05,
      "loss": 0.5771,
      "step": 567
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5861293404066574,
      "learning_rate": 1.997119630884221e-05,
      "loss": 0.5449,
      "step": 568
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2942356462731295,
      "learning_rate": 1.9970964601572788e-05,
      "loss": 0.4546,
      "step": 569
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4849108038483698,
      "learning_rate": 1.9970731967419724e-05,
      "loss": 0.5459,
      "step": 570
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3255477841033008,
      "learning_rate": 1.997049840640465e-05,
      "loss": 0.4204,
      "step": 571
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.330247327814002,
      "learning_rate": 1.9970263918549274e-05,
      "loss": 0.4478,
      "step": 572
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3336611010366823,
      "learning_rate": 1.9970028503875387e-05,
      "loss": 0.4331,
      "step": 573
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5027150515066197,
      "learning_rate": 1.996979216240488e-05,
      "loss": 0.5625,
      "step": 574
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3805236583993028,
      "learning_rate": 1.9969554894159723e-05,
      "loss": 0.4741,
      "step": 575
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3586600072514714,
      "learning_rate": 1.996931669916197e-05,
      "loss": 0.4951,
      "step": 576
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5223106353611267,
      "learning_rate": 1.9969077577433763e-05,
      "loss": 0.5264,
      "step": 577
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5385942426385915,
      "learning_rate": 1.9968837528997333e-05,
      "loss": 0.4937,
      "step": 578
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3920032267551488,
      "learning_rate": 1.9968596553874993e-05,
      "loss": 0.563,
      "step": 579
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6086325922290647,
      "learning_rate": 1.9968354652089142e-05,
      "loss": 0.6299,
      "step": 580
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.291804715467673,
      "learning_rate": 1.9968111823662268e-05,
      "loss": 0.4067,
      "step": 581
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4966623846995006,
      "learning_rate": 1.996786806861694e-05,
      "loss": 0.481,
      "step": 582
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5491577428009684,
      "learning_rate": 1.9967623386975826e-05,
      "loss": 0.5347,
      "step": 583
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5949084600271346,
      "learning_rate": 1.9967377778761667e-05,
      "loss": 0.625,
      "step": 584
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3774379606464244,
      "learning_rate": 1.996713124399729e-05,
      "loss": 0.4736,
      "step": 585
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4203695781493644,
      "learning_rate": 1.9966883782705617e-05,
      "loss": 0.4043,
      "step": 586
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3605940688723812,
      "learning_rate": 1.996663539490965e-05,
      "loss": 0.4019,
      "step": 587
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.483544103222794,
      "learning_rate": 1.996638608063248e-05,
      "loss": 0.4473,
      "step": 588
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3162666838373345,
      "learning_rate": 1.996613583989728e-05,
      "loss": 0.4453,
      "step": 589
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6291795331292993,
      "learning_rate": 1.9965884672727313e-05,
      "loss": 0.5527,
      "step": 590
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5120996810322944,
      "learning_rate": 1.996563257914593e-05,
      "loss": 0.5005,
      "step": 591
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6147610115765059,
      "learning_rate": 1.9965379559176562e-05,
      "loss": 0.5542,
      "step": 592
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8227699255086989,
      "learning_rate": 1.996512561284273e-05,
      "loss": 0.5791,
      "step": 593
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5115139393502182,
      "learning_rate": 1.9964870740168046e-05,
      "loss": 0.5127,
      "step": 594
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.434228822840142,
      "learning_rate": 1.9964614941176194e-05,
      "loss": 0.52,
      "step": 595
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.414477307910178,
      "learning_rate": 1.9964358215890955e-05,
      "loss": 0.4517,
      "step": 596
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6808391221398626,
      "learning_rate": 1.9964100564336197e-05,
      "loss": 0.5957,
      "step": 597
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3225161912744736,
      "learning_rate": 1.9963841986535864e-05,
      "loss": 0.3862,
      "step": 598
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4631239224873172,
      "learning_rate": 1.9963582482514003e-05,
      "loss": 0.5181,
      "step": 599
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3215255389066791,
      "learning_rate": 1.996332205229473e-05,
      "loss": 0.4702,
      "step": 600
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.307840079983866,
      "learning_rate": 1.996306069590225e-05,
      "loss": 0.4248,
      "step": 601
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.565398072468361,
      "learning_rate": 1.996279841336087e-05,
      "loss": 0.4658,
      "step": 602
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4477255710904193,
      "learning_rate": 1.9962535204694964e-05,
      "loss": 0.4595,
      "step": 603
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4093696514723524,
      "learning_rate": 1.9962271069929e-05,
      "loss": 0.4956,
      "step": 604
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6789904771775233,
      "learning_rate": 1.9962006009087535e-05,
      "loss": 0.5737,
      "step": 605
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5863459993810665,
      "learning_rate": 1.9961740022195202e-05,
      "loss": 0.584,
      "step": 606
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.327528653391696,
      "learning_rate": 1.9961473109276735e-05,
      "loss": 0.4272,
      "step": 607
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.655461486404094,
      "learning_rate": 1.9961205270356937e-05,
      "loss": 0.52,
      "step": 608
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3664589910464306,
      "learning_rate": 1.996093650546071e-05,
      "loss": 0.4248,
      "step": 609
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4064874304036996,
      "learning_rate": 1.9960666814613043e-05,
      "loss": 0.5005,
      "step": 610
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5051040856497053,
      "learning_rate": 1.9960396197838997e-05,
      "loss": 0.5669,
      "step": 611
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5363514282119046,
      "learning_rate": 1.996012465516373e-05,
      "loss": 0.6357,
      "step": 612
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3561817467112773,
      "learning_rate": 1.9959852186612492e-05,
      "loss": 0.4253,
      "step": 613
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3603939131219724,
      "learning_rate": 1.9959578792210604e-05,
      "loss": 0.4189,
      "step": 614
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.656626689424805,
      "learning_rate": 1.9959304471983477e-05,
      "loss": 0.5889,
      "step": 615
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3774273208525318,
      "learning_rate": 1.995902922595662e-05,
      "loss": 0.4751,
      "step": 616
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3494501416677702,
      "learning_rate": 1.995875305415561e-05,
      "loss": 0.4512,
      "step": 617
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4073869565957418,
      "learning_rate": 1.9958475956606133e-05,
      "loss": 0.459,
      "step": 618
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3872128380806057,
      "learning_rate": 1.995819793333394e-05,
      "loss": 0.396,
      "step": 619
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5171025933756503,
      "learning_rate": 1.995791898436487e-05,
      "loss": 0.5991,
      "step": 620
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.921396098392842,
      "learning_rate": 1.9957639109724858e-05,
      "loss": 0.5132,
      "step": 621
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3376249112818386,
      "learning_rate": 1.995735830943992e-05,
      "loss": 0.4663,
      "step": 622
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5176629998517275,
      "learning_rate": 1.9957076583536166e-05,
      "loss": 0.5171,
      "step": 623
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3681994782519846,
      "learning_rate": 1.9956793932039774e-05,
      "loss": 0.4707,
      "step": 624
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3636138681441032,
      "learning_rate": 1.9956510354977022e-05,
      "loss": 0.4746,
      "step": 625
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4807370554449848,
      "learning_rate": 1.9956225852374275e-05,
      "loss": 0.5073,
      "step": 626
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5366832229615366,
      "learning_rate": 1.995594042425798e-05,
      "loss": 0.4463,
      "step": 627
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3694168910928592,
      "learning_rate": 1.9955654070654664e-05,
      "loss": 0.4302,
      "step": 628
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6664837971462987,
      "learning_rate": 1.995536679159095e-05,
      "loss": 0.4844,
      "step": 629
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6169138202904674,
      "learning_rate": 1.995507858709354e-05,
      "loss": 0.5308,
      "step": 630
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.288549312467579,
      "learning_rate": 1.995478945718923e-05,
      "loss": 0.4209,
      "step": 631
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4757049857783961,
      "learning_rate": 1.9954499401904893e-05,
      "loss": 0.4829,
      "step": 632
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.573257457918332,
      "learning_rate": 1.995420842126749e-05,
      "loss": 0.4961,
      "step": 633
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.26523037966077,
      "learning_rate": 1.9953916515304077e-05,
      "loss": 0.4092,
      "step": 634
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5771731397895892,
      "learning_rate": 1.9953623684041786e-05,
      "loss": 0.5684,
      "step": 635
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4132960323140658,
      "learning_rate": 1.9953329927507833e-05,
      "loss": 0.478,
      "step": 636
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.551105861115843,
      "learning_rate": 1.995303524572953e-05,
      "loss": 0.5171,
      "step": 637
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5277994706024751,
      "learning_rate": 1.995273963873427e-05,
      "loss": 0.4795,
      "step": 638
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4525836983516638,
      "learning_rate": 1.9952443106549535e-05,
      "loss": 0.542,
      "step": 639
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5291163124229241,
      "learning_rate": 1.9952145649202885e-05,
      "loss": 0.4878,
      "step": 640
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4513059607618193,
      "learning_rate": 1.995184726672197e-05,
      "loss": 0.4888,
      "step": 641
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6778355861652696,
      "learning_rate": 1.9951547959134535e-05,
      "loss": 0.6128,
      "step": 642
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.436110637200094,
      "learning_rate": 1.9951247726468393e-05,
      "loss": 0.4351,
      "step": 643
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4570073930092748,
      "learning_rate": 1.9950946568751458e-05,
      "loss": 0.5083,
      "step": 644
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.477686815965051,
      "learning_rate": 1.995064448601173e-05,
      "loss": 0.4844,
      "step": 645
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6437746824557906,
      "learning_rate": 1.9950341478277282e-05,
      "loss": 0.6035,
      "step": 646
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3958021660546818,
      "learning_rate": 1.9950037545576288e-05,
      "loss": 0.4648,
      "step": 647
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4043339283334764,
      "learning_rate": 1.9949732687936992e-05,
      "loss": 0.4683,
      "step": 648
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.523450581989599,
      "learning_rate": 1.994942690538774e-05,
      "loss": 0.5352,
      "step": 649
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3421044906887702,
      "learning_rate": 1.9949120197956956e-05,
      "loss": 0.4863,
      "step": 650
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2936245224781133,
      "learning_rate": 1.9948812565673153e-05,
      "loss": 0.4307,
      "step": 651
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2540781724424328,
      "learning_rate": 1.9948504008564928e-05,
      "loss": 0.4072,
      "step": 652
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4639637895972029,
      "learning_rate": 1.9948194526660956e-05,
      "loss": 0.5322,
      "step": 653
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4496361562247393,
      "learning_rate": 1.9947884119990012e-05,
      "loss": 0.4771,
      "step": 654
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.620506909552827,
      "learning_rate": 1.994757278858095e-05,
      "loss": 0.5981,
      "step": 655
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3307110621949598,
      "learning_rate": 1.994726053246271e-05,
      "loss": 0.4956,
      "step": 656
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.503566797518634,
      "learning_rate": 1.9946947351664324e-05,
      "loss": 0.4941,
      "step": 657
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.361857488667517,
      "learning_rate": 1.99466332462149e-05,
      "loss": 0.4604,
      "step": 658
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3639828549297999,
      "learning_rate": 1.9946318216143633e-05,
      "loss": 0.4717,
      "step": 659
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3393606374048481,
      "learning_rate": 1.9946002261479817e-05,
      "loss": 0.4199,
      "step": 660
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6050249828563097,
      "learning_rate": 1.9945685382252816e-05,
      "loss": 0.584,
      "step": 661
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3240489296369595,
      "learning_rate": 1.9945367578492085e-05,
      "loss": 0.4438,
      "step": 662
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4647427399944368,
      "learning_rate": 1.994504885022717e-05,
      "loss": 0.5356,
      "step": 663
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8272959603944448,
      "learning_rate": 1.9944729197487702e-05,
      "loss": 0.6196,
      "step": 664
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4888157561582651,
      "learning_rate": 1.9944408620303393e-05,
      "loss": 0.4395,
      "step": 665
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5295154521660042,
      "learning_rate": 1.994408711870404e-05,
      "loss": 0.5522,
      "step": 666
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7065033020170686,
      "learning_rate": 1.9943764692719535e-05,
      "loss": 0.5566,
      "step": 667
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5031526333508516,
      "learning_rate": 1.9943441342379843e-05,
      "loss": 0.4653,
      "step": 668
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6764544238761847,
      "learning_rate": 1.994311706771503e-05,
      "loss": 0.5532,
      "step": 669
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4193473508417653,
      "learning_rate": 1.9942791868755233e-05,
      "loss": 0.5142,
      "step": 670
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4347538913834945,
      "learning_rate": 1.9942465745530687e-05,
      "loss": 0.4541,
      "step": 671
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3093525362244116,
      "learning_rate": 1.99421386980717e-05,
      "loss": 0.4185,
      "step": 672
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5588506005035447,
      "learning_rate": 1.994181072640869e-05,
      "loss": 0.6611,
      "step": 673
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4006257024480118,
      "learning_rate": 1.9941481830572126e-05,
      "loss": 0.4585,
      "step": 674
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3831331566197211,
      "learning_rate": 1.9941152010592594e-05,
      "loss": 0.4985,
      "step": 675
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5039031079150404,
      "learning_rate": 1.994082126650075e-05,
      "loss": 0.5288,
      "step": 676
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3881616036263524,
      "learning_rate": 1.9940489598327336e-05,
      "loss": 0.4702,
      "step": 677
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3874714674598454,
      "learning_rate": 1.994015700610319e-05,
      "loss": 0.5034,
      "step": 678
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.428389687527317,
      "learning_rate": 1.9939823489859226e-05,
      "loss": 0.4624,
      "step": 679
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4001787025925405,
      "learning_rate": 1.993948904962644e-05,
      "loss": 0.4829,
      "step": 680
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6500050638688402,
      "learning_rate": 1.9939153685435932e-05,
      "loss": 0.5454,
      "step": 681
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6276263958169501,
      "learning_rate": 1.9938817397318872e-05,
      "loss": 0.5327,
      "step": 682
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4835102353945226,
      "learning_rate": 1.993848018530652e-05,
      "loss": 0.5591,
      "step": 683
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4969909704314155,
      "learning_rate": 1.9938142049430223e-05,
      "loss": 0.4785,
      "step": 684
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4027866695071185,
      "learning_rate": 1.9937802989721417e-05,
      "loss": 0.4766,
      "step": 685
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3465115996995531,
      "learning_rate": 1.9937463006211617e-05,
      "loss": 0.4507,
      "step": 686
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6551752991025934,
      "learning_rate": 1.9937122098932428e-05,
      "loss": 0.6338,
      "step": 687
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3803731236903725,
      "learning_rate": 1.993678026791554e-05,
      "loss": 0.4805,
      "step": 688
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5171490874929938,
      "learning_rate": 1.9936437513192728e-05,
      "loss": 0.4878,
      "step": 689
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4308358498508913,
      "learning_rate": 1.9936093834795853e-05,
      "loss": 0.4585,
      "step": 690
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.364733752517677,
      "learning_rate": 1.9935749232756868e-05,
      "loss": 0.4639,
      "step": 691
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4464481116638688,
      "learning_rate": 1.9935403707107804e-05,
      "loss": 0.4614,
      "step": 692
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3818825264811792,
      "learning_rate": 1.9935057257880777e-05,
      "loss": 0.5142,
      "step": 693
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6981752148082516,
      "learning_rate": 1.9934709885108e-05,
      "loss": 0.7178,
      "step": 694
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3713499939324802,
      "learning_rate": 1.9934361588821757e-05,
      "loss": 0.5195,
      "step": 695
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4965008525514543,
      "learning_rate": 1.9934012369054422e-05,
      "loss": 0.5176,
      "step": 696
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4267110462910018,
      "learning_rate": 1.993366222583847e-05,
      "loss": 0.5552,
      "step": 697
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.387045586039698,
      "learning_rate": 1.993331115920644e-05,
      "loss": 0.5249,
      "step": 698
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4132492436734525,
      "learning_rate": 1.993295916919097e-05,
      "loss": 0.52,
      "step": 699
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5125344568799406,
      "learning_rate": 1.993260625582478e-05,
      "loss": 0.4839,
      "step": 700
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.463385135608391,
      "learning_rate": 1.9932252419140675e-05,
      "loss": 0.4844,
      "step": 701
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3229178645383404,
      "learning_rate": 1.993189765917155e-05,
      "loss": 0.4121,
      "step": 702
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.579825843621078,
      "learning_rate": 1.993154197595038e-05,
      "loss": 0.478,
      "step": 703
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4061295064095387,
      "learning_rate": 1.9931185369510228e-05,
      "loss": 0.5171,
      "step": 704
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4245782980609212,
      "learning_rate": 1.993082783988425e-05,
      "loss": 0.4902,
      "step": 705
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3987117601171128,
      "learning_rate": 1.993046938710567e-05,
      "loss": 0.4878,
      "step": 706
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3540057231524971,
      "learning_rate": 1.993011001120782e-05,
      "loss": 0.4683,
      "step": 707
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.625875273558657,
      "learning_rate": 1.9929749712224104e-05,
      "loss": 0.5342,
      "step": 708
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3533245979555162,
      "learning_rate": 1.992938849018801e-05,
      "loss": 0.4941,
      "step": 709
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6568637250364056,
      "learning_rate": 1.992902634513312e-05,
      "loss": 0.5449,
      "step": 710
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4003882883693233,
      "learning_rate": 1.99286632770931e-05,
      "loss": 0.458,
      "step": 711
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3614699068221654,
      "learning_rate": 1.9928299286101696e-05,
      "loss": 0.541,
      "step": 712
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5274174822531656,
      "learning_rate": 1.992793437219275e-05,
      "loss": 0.5381,
      "step": 713
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.690944559007341,
      "learning_rate": 1.992756853540018e-05,
      "loss": 0.5103,
      "step": 714
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6757934345377818,
      "learning_rate": 1.9927201775757995e-05,
      "loss": 0.5806,
      "step": 715
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3561642284208326,
      "learning_rate": 1.9926834093300283e-05,
      "loss": 0.4409,
      "step": 716
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3498455210091262,
      "learning_rate": 1.9926465488061232e-05,
      "loss": 0.4082,
      "step": 717
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3171470024462135,
      "learning_rate": 1.9926095960075097e-05,
      "loss": 0.4014,
      "step": 718
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4324165981146808,
      "learning_rate": 1.9925725509376236e-05,
      "loss": 0.5293,
      "step": 719
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4642505556541625,
      "learning_rate": 1.9925354135999083e-05,
      "loss": 0.4868,
      "step": 720
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6295046113609803,
      "learning_rate": 1.9924981839978167e-05,
      "loss": 0.6372,
      "step": 721
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6068443164852861,
      "learning_rate": 1.992460862134808e-05,
      "loss": 0.5186,
      "step": 722
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4878662662753734,
      "learning_rate": 1.992423448014353e-05,
      "loss": 0.5352,
      "step": 723
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3234754187378048,
      "learning_rate": 1.9923859416399295e-05,
      "loss": 0.4131,
      "step": 724
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4152738677595955,
      "learning_rate": 1.992348343015023e-05,
      "loss": 0.4668,
      "step": 725
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5942990729615092,
      "learning_rate": 1.9923106521431296e-05,
      "loss": 0.5771,
      "step": 726
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4609628327922837,
      "learning_rate": 1.9922728690277528e-05,
      "loss": 0.4692,
      "step": 727
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2240431139029484,
      "learning_rate": 1.9922349936724048e-05,
      "loss": 0.375,
      "step": 728
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.443746663930552,
      "learning_rate": 1.9921970260806064e-05,
      "loss": 0.5298,
      "step": 729
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5637356845317356,
      "learning_rate": 1.992158966255887e-05,
      "loss": 0.5059,
      "step": 730
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4002798249445148,
      "learning_rate": 1.9921208142017844e-05,
      "loss": 0.4663,
      "step": 731
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3502349353132679,
      "learning_rate": 1.9920825699218453e-05,
      "loss": 0.4561,
      "step": 732
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5477893097150155,
      "learning_rate": 1.9920442334196248e-05,
      "loss": 0.5811,
      "step": 733
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7261275456839555,
      "learning_rate": 1.9920058046986867e-05,
      "loss": 0.6011,
      "step": 734
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4107782936063442,
      "learning_rate": 1.991967283762603e-05,
      "loss": 0.438,
      "step": 735
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2491021711822623,
      "learning_rate": 1.991928670614955e-05,
      "loss": 0.3965,
      "step": 736
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3998682164531064,
      "learning_rate": 1.9918899652593316e-05,
      "loss": 0.4233,
      "step": 737
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3895510054212274,
      "learning_rate": 1.9918511676993315e-05,
      "loss": 0.4316,
      "step": 738
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4861915127607581,
      "learning_rate": 1.99181227793856e-05,
      "loss": 0.4829,
      "step": 739
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4023092647410027,
      "learning_rate": 1.9917732959806336e-05,
      "loss": 0.4355,
      "step": 740
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4013816240082366,
      "learning_rate": 1.9917342218291752e-05,
      "loss": 0.4292,
      "step": 741
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3256482134636904,
      "learning_rate": 1.9916950554878178e-05,
      "loss": 0.4751,
      "step": 742
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2605235251195641,
      "learning_rate": 1.9916557969602007e-05,
      "loss": 0.417,
      "step": 743
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4190630036034266,
      "learning_rate": 1.991616446249975e-05,
      "loss": 0.4785,
      "step": 744
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4148807401066745,
      "learning_rate": 1.9915770033607978e-05,
      "loss": 0.4795,
      "step": 745
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4757503299792147,
      "learning_rate": 1.9915374682963358e-05,
      "loss": 0.5073,
      "step": 746
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4931413260278041,
      "learning_rate": 1.991497841060264e-05,
      "loss": 0.5078,
      "step": 747
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5809145320567446,
      "learning_rate": 1.9914581216562664e-05,
      "loss": 0.54,
      "step": 748
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.280053930331413,
      "learning_rate": 1.991418310088035e-05,
      "loss": 0.436,
      "step": 749
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2872468145338951,
      "learning_rate": 1.9913784063592708e-05,
      "loss": 0.3965,
      "step": 750
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4593003361422499,
      "learning_rate": 1.991338410473683e-05,
      "loss": 0.5254,
      "step": 751
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4731736730150486,
      "learning_rate": 1.9912983224349894e-05,
      "loss": 0.4717,
      "step": 752
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5016792853270995,
      "learning_rate": 1.991258142246917e-05,
      "loss": 0.4468,
      "step": 753
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3403959976187045,
      "learning_rate": 1.9912178699132005e-05,
      "loss": 0.5054,
      "step": 754
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4035522141254781,
      "learning_rate": 1.9911775054375835e-05,
      "loss": 0.439,
      "step": 755
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6534652283539084,
      "learning_rate": 1.9911370488238185e-05,
      "loss": 0.4946,
      "step": 756
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4893842594747118,
      "learning_rate": 1.9910965000756662e-05,
      "loss": 0.5283,
      "step": 757
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3578166482241028,
      "learning_rate": 1.9910558591968956e-05,
      "loss": 0.4614,
      "step": 758
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2992118849092897,
      "learning_rate": 1.991015126191285e-05,
      "loss": 0.4492,
      "step": 759
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4696046386257664,
      "learning_rate": 1.990974301062621e-05,
      "loss": 0.5518,
      "step": 760
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3786380106739593,
      "learning_rate": 1.990933383814698e-05,
      "loss": 0.4932,
      "step": 761
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.382821090587405,
      "learning_rate": 1.99089237445132e-05,
      "loss": 0.4658,
      "step": 762
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3726153161626071,
      "learning_rate": 1.9908512729762995e-05,
      "loss": 0.5156,
      "step": 763
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2828312821853352,
      "learning_rate": 1.9908100793934566e-05,
      "loss": 0.4766,
      "step": 764
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5452933764100583,
      "learning_rate": 1.990768793706621e-05,
      "loss": 0.4951,
      "step": 765
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.423319706665499,
      "learning_rate": 1.9907274159196302e-05,
      "loss": 0.6055,
      "step": 766
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6992317038296305,
      "learning_rate": 1.9906859460363307e-05,
      "loss": 0.6592,
      "step": 767
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2145247407952926,
      "learning_rate": 1.990644384060578e-05,
      "loss": 0.4038,
      "step": 768
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.580401221439093,
      "learning_rate": 1.990602729996235e-05,
      "loss": 0.5801,
      "step": 769
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5303646020827033,
      "learning_rate": 1.990560983847174e-05,
      "loss": 0.5801,
      "step": 770
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4147764352043266,
      "learning_rate": 1.9905191456172757e-05,
      "loss": 0.4985,
      "step": 771
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4855976044700083,
      "learning_rate": 1.990477215310429e-05,
      "loss": 0.4775,
      "step": 772
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.358089717313229,
      "learning_rate": 1.990435192930532e-05,
      "loss": 0.415,
      "step": 773
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3952452262678412,
      "learning_rate": 1.9903930784814908e-05,
      "loss": 0.5161,
      "step": 774
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4940603161490598,
      "learning_rate": 1.9903508719672208e-05,
      "loss": 0.5952,
      "step": 775
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.60890680342218,
      "learning_rate": 1.990308573391645e-05,
      "loss": 0.6011,
      "step": 776
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.338854666140026,
      "learning_rate": 1.9902661827586953e-05,
      "loss": 0.4941,
      "step": 777
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3907253575126928,
      "learning_rate": 1.9902237000723127e-05,
      "loss": 0.5576,
      "step": 778
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4189730930311113,
      "learning_rate": 1.9901811253364458e-05,
      "loss": 0.542,
      "step": 779
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6330334904985881,
      "learning_rate": 1.9901384585550527e-05,
      "loss": 0.5503,
      "step": 780
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6876361500326877,
      "learning_rate": 1.990095699732099e-05,
      "loss": 0.5889,
      "step": 781
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.434779315911078,
      "learning_rate": 1.990052848871561e-05,
      "loss": 0.5024,
      "step": 782
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3725529036117499,
      "learning_rate": 1.9900099059774197e-05,
      "loss": 0.4888,
      "step": 783
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.330566603493168,
      "learning_rate": 1.989966871053669e-05,
      "loss": 0.4443,
      "step": 784
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.511036958212691,
      "learning_rate": 1.989923744104309e-05,
      "loss": 0.5352,
      "step": 785
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.336143181120464,
      "learning_rate": 1.9898805251333477e-05,
      "loss": 0.4443,
      "step": 786
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3774556896867642,
      "learning_rate": 1.9898372141448033e-05,
      "loss": 0.4424,
      "step": 787
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4247682949898,
      "learning_rate": 1.989793811142702e-05,
      "loss": 0.4653,
      "step": 788
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.64947512899262,
      "learning_rate": 1.9897503161310786e-05,
      "loss": 0.6333,
      "step": 789
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6642489777312126,
      "learning_rate": 1.989706729113976e-05,
      "loss": 0.6592,
      "step": 790
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4678835143463094,
      "learning_rate": 1.989663050095446e-05,
      "loss": 0.3945,
      "step": 791
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.459277866942333,
      "learning_rate": 1.989619279079549e-05,
      "loss": 0.4453,
      "step": 792
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4481499643177103,
      "learning_rate": 1.989575416070354e-05,
      "loss": 0.4839,
      "step": 793
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5182746655856338,
      "learning_rate": 1.9895314610719382e-05,
      "loss": 0.4912,
      "step": 794
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5668926166110424,
      "learning_rate": 1.9894874140883877e-05,
      "loss": 0.5366,
      "step": 795
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4462446168744485,
      "learning_rate": 1.9894432751237974e-05,
      "loss": 0.4795,
      "step": 796
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4937186780133718,
      "learning_rate": 1.9893990441822698e-05,
      "loss": 0.5557,
      "step": 797
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4048331200520277,
      "learning_rate": 1.9893547212679162e-05,
      "loss": 0.4692,
      "step": 798
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.429393755617368,
      "learning_rate": 1.989310306384858e-05,
      "loss": 0.5015,
      "step": 799
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6235070878946678,
      "learning_rate": 1.9892657995372227e-05,
      "loss": 0.6016,
      "step": 800
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3107328879412803,
      "learning_rate": 1.989221200729149e-05,
      "loss": 0.4536,
      "step": 801
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3671800533534217,
      "learning_rate": 1.989176509964781e-05,
      "loss": 0.5186,
      "step": 802
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.406391834169605,
      "learning_rate": 1.9891317272482744e-05,
      "loss": 0.5063,
      "step": 803
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6697336389383366,
      "learning_rate": 1.9890868525837917e-05,
      "loss": 0.5547,
      "step": 804
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6290030573990912,
      "learning_rate": 1.989041885975504e-05,
      "loss": 0.5254,
      "step": 805
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6501955348048283,
      "learning_rate": 1.9889968274275916e-05,
      "loss": 0.4912,
      "step": 806
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4477912471888867,
      "learning_rate": 1.9889516769442436e-05,
      "loss": 0.4365,
      "step": 807
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.481521569259303,
      "learning_rate": 1.9889064345296563e-05,
      "loss": 0.4653,
      "step": 808
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.647947270843963,
      "learning_rate": 1.9888611001880357e-05,
      "loss": 0.4507,
      "step": 809
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4205928875926819,
      "learning_rate": 1.988815673923596e-05,
      "loss": 0.4556,
      "step": 810
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4238972968141317,
      "learning_rate": 1.9887701557405598e-05,
      "loss": 0.4404,
      "step": 811
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.48025079141048,
      "learning_rate": 1.988724545643159e-05,
      "loss": 0.481,
      "step": 812
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3426516438363516,
      "learning_rate": 1.9886788436356325e-05,
      "loss": 0.4214,
      "step": 813
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4384098531835068,
      "learning_rate": 1.9886330497222294e-05,
      "loss": 0.5039,
      "step": 814
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5579369300911863,
      "learning_rate": 1.988587163907206e-05,
      "loss": 0.5435,
      "step": 815
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3748110186550906,
      "learning_rate": 1.9885411861948287e-05,
      "loss": 0.4541,
      "step": 816
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3527659263862963,
      "learning_rate": 1.9884951165893706e-05,
      "loss": 0.4634,
      "step": 817
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4746105707676895,
      "learning_rate": 1.9884489550951146e-05,
      "loss": 0.5308,
      "step": 818
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5636471887095342,
      "learning_rate": 1.9884027017163515e-05,
      "loss": 0.5386,
      "step": 819
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5064882472265493,
      "learning_rate": 1.9883563564573815e-05,
      "loss": 0.478,
      "step": 820
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2817401009043339,
      "learning_rate": 1.9883099193225125e-05,
      "loss": 0.3921,
      "step": 821
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4341017913069745,
      "learning_rate": 1.9882633903160612e-05,
      "loss": 0.4971,
      "step": 822
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6088480911301521,
      "learning_rate": 1.988216769442353e-05,
      "loss": 0.522,
      "step": 823
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.297656343014956,
      "learning_rate": 1.9881700567057214e-05,
      "loss": 0.4688,
      "step": 824
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5331797801139821,
      "learning_rate": 1.988123252110509e-05,
      "loss": 0.5728,
      "step": 825
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4300010049910392,
      "learning_rate": 1.9880763556610666e-05,
      "loss": 0.4995,
      "step": 826
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3979250521520108,
      "learning_rate": 1.988029367361753e-05,
      "loss": 0.4805,
      "step": 827
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4671943825135032,
      "learning_rate": 1.9879822872169378e-05,
      "loss": 0.5527,
      "step": 828
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3931499685893414,
      "learning_rate": 1.9879351152309955e-05,
      "loss": 0.5215,
      "step": 829
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.493354375247885,
      "learning_rate": 1.9878878514083124e-05,
      "loss": 0.4878,
      "step": 830
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3697678707661307,
      "learning_rate": 1.9878404957532817e-05,
      "loss": 0.5093,
      "step": 831
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3199038618646766,
      "learning_rate": 1.9877930482703057e-05,
      "loss": 0.5947,
      "step": 832
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3788062594045494,
      "learning_rate": 1.9877455089637944e-05,
      "loss": 0.5454,
      "step": 833
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.512765385756017,
      "learning_rate": 1.9876978778381675e-05,
      "loss": 0.5659,
      "step": 834
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3437627459094428,
      "learning_rate": 1.9876501548978527e-05,
      "loss": 0.4287,
      "step": 835
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3045602359517694,
      "learning_rate": 1.9876023401472865e-05,
      "loss": 0.4775,
      "step": 836
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.476193057265912,
      "learning_rate": 1.987554433590913e-05,
      "loss": 0.5361,
      "step": 837
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.324705327880853,
      "learning_rate": 1.987506435233186e-05,
      "loss": 0.4795,
      "step": 838
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2433254122915762,
      "learning_rate": 1.987458345078567e-05,
      "loss": 0.4062,
      "step": 839
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2916601461709605,
      "learning_rate": 1.9874101631315268e-05,
      "loss": 0.4253,
      "step": 840
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.467958688031402,
      "learning_rate": 1.9873618893965442e-05,
      "loss": 0.4868,
      "step": 841
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5570506410611524,
      "learning_rate": 1.987313523878106e-05,
      "loss": 0.6318,
      "step": 842
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.482599763605817,
      "learning_rate": 1.987265066580709e-05,
      "loss": 0.542,
      "step": 843
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.459642891944371,
      "learning_rate": 1.9872165175088578e-05,
      "loss": 0.4678,
      "step": 844
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4571875661993285,
      "learning_rate": 1.9871678766670647e-05,
      "loss": 0.5024,
      "step": 845
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.462008039103072,
      "learning_rate": 1.9871191440598515e-05,
      "loss": 0.5552,
      "step": 846
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.29590220206523,
      "learning_rate": 1.9870703196917485e-05,
      "loss": 0.4868,
      "step": 847
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.264670254901286,
      "learning_rate": 1.9870214035672945e-05,
      "loss": 0.458,
      "step": 848
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7107744961516387,
      "learning_rate": 1.986972395691036e-05,
      "loss": 0.5737,
      "step": 849
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4373678431225398,
      "learning_rate": 1.9869232960675292e-05,
      "loss": 0.5107,
      "step": 850
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4943672774651149,
      "learning_rate": 1.9868741047013382e-05,
      "loss": 0.5469,
      "step": 851
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.548875656271839,
      "learning_rate": 1.986824821597036e-05,
      "loss": 0.627,
      "step": 852
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.255182911649015,
      "learning_rate": 1.9867754467592037e-05,
      "loss": 0.4961,
      "step": 853
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.481658789866771,
      "learning_rate": 1.9867259801924306e-05,
      "loss": 0.4902,
      "step": 854
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3635998659684887,
      "learning_rate": 1.9866764219013154e-05,
      "loss": 0.5283,
      "step": 855
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.523324916139995,
      "learning_rate": 1.9866267718904655e-05,
      "loss": 0.5288,
      "step": 856
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2589743592539957,
      "learning_rate": 1.9865770301644956e-05,
      "loss": 0.4277,
      "step": 857
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.608721642161341,
      "learning_rate": 1.9865271967280297e-05,
      "loss": 0.5781,
      "step": 858
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5377274695456995,
      "learning_rate": 1.9864772715857e-05,
      "loss": 0.5688,
      "step": 859
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3094038221584756,
      "learning_rate": 1.9864272547421482e-05,
      "loss": 0.4209,
      "step": 860
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.347644170391953,
      "learning_rate": 1.9863771462020235e-05,
      "loss": 0.4951,
      "step": 861
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5287586525083605,
      "learning_rate": 1.9863269459699836e-05,
      "loss": 0.5371,
      "step": 862
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.558028469771246,
      "learning_rate": 1.986276654050695e-05,
      "loss": 0.5073,
      "step": 863
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5416709455626412,
      "learning_rate": 1.986226270448833e-05,
      "loss": 0.5068,
      "step": 864
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4258062022325213,
      "learning_rate": 1.9861757951690813e-05,
      "loss": 0.5508,
      "step": 865
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.623362668392341,
      "learning_rate": 1.9861252282161313e-05,
      "loss": 0.5098,
      "step": 866
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.491606116205011,
      "learning_rate": 1.9860745695946848e-05,
      "loss": 0.4585,
      "step": 867
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4178953543549644,
      "learning_rate": 1.9860238193094497e-05,
      "loss": 0.5649,
      "step": 868
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4047373713251796,
      "learning_rate": 1.9859729773651446e-05,
      "loss": 0.4121,
      "step": 869
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3817042929149035,
      "learning_rate": 1.985922043766495e-05,
      "loss": 0.4478,
      "step": 870
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4530480720069954,
      "learning_rate": 1.985871018518236e-05,
      "loss": 0.5786,
      "step": 871
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4976802562597697,
      "learning_rate": 1.9858199016251106e-05,
      "loss": 0.5049,
      "step": 872
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4949894875065948,
      "learning_rate": 1.985768693091871e-05,
      "loss": 0.5317,
      "step": 873
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3907562519490881,
      "learning_rate": 1.9857173929232768e-05,
      "loss": 0.4321,
      "step": 874
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4429402700648206,
      "learning_rate": 1.985666001124097e-05,
      "loss": 0.4653,
      "step": 875
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5142437966267708,
      "learning_rate": 1.9856145176991093e-05,
      "loss": 0.5498,
      "step": 876
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.502468737394059,
      "learning_rate": 1.9855629426530992e-05,
      "loss": 0.5166,
      "step": 877
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4238835759864816,
      "learning_rate": 1.9855112759908607e-05,
      "loss": 0.3955,
      "step": 878
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3992484868419746,
      "learning_rate": 1.9854595177171968e-05,
      "loss": 0.4575,
      "step": 879
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3109301671058282,
      "learning_rate": 1.9854076678369197e-05,
      "loss": 0.5015,
      "step": 880
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.469292683218551,
      "learning_rate": 1.985355726354848e-05,
      "loss": 0.5781,
      "step": 881
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4080684014485552,
      "learning_rate": 1.985303693275811e-05,
      "loss": 0.4658,
      "step": 882
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.710855438578213,
      "learning_rate": 1.9852515686046453e-05,
      "loss": 0.666,
      "step": 883
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.533984538003839,
      "learning_rate": 1.985199352346196e-05,
      "loss": 0.5293,
      "step": 884
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3298939018574873,
      "learning_rate": 1.9851470445053173e-05,
      "loss": 0.4434,
      "step": 885
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5207488181041502,
      "learning_rate": 1.985094645086872e-05,
      "loss": 0.5293,
      "step": 886
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4378822527280573,
      "learning_rate": 1.9850421540957307e-05,
      "loss": 0.5112,
      "step": 887
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5369297860222126,
      "learning_rate": 1.9849895715367728e-05,
      "loss": 0.5317,
      "step": 888
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.678926375682574,
      "learning_rate": 1.9849368974148865e-05,
      "loss": 0.5728,
      "step": 889
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5362936975446289,
      "learning_rate": 1.984884131734968e-05,
      "loss": 0.5884,
      "step": 890
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3841005780169016,
      "learning_rate": 1.9848312745019224e-05,
      "loss": 0.5728,
      "step": 891
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2304790612506336,
      "learning_rate": 1.984778325720664e-05,
      "loss": 0.4585,
      "step": 892
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3660536516368273,
      "learning_rate": 1.9847252853961136e-05,
      "loss": 0.4863,
      "step": 893
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3811771953933039,
      "learning_rate": 1.984672153533202e-05,
      "loss": 0.5142,
      "step": 894
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4106753887507333,
      "learning_rate": 1.984618930136869e-05,
      "loss": 0.5273,
      "step": 895
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3205778330378921,
      "learning_rate": 1.9845656152120617e-05,
      "loss": 0.4951,
      "step": 896
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4723027952208814,
      "learning_rate": 1.984512208763736e-05,
      "loss": 0.4932,
      "step": 897
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4651541495764147,
      "learning_rate": 1.9844587107968567e-05,
      "loss": 0.5596,
      "step": 898
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.55392493502501,
      "learning_rate": 1.9844051213163967e-05,
      "loss": 0.583,
      "step": 899
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.420822058055589,
      "learning_rate": 1.9843514403273378e-05,
      "loss": 0.5718,
      "step": 900
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.593591091359301,
      "learning_rate": 1.98429766783467e-05,
      "loss": 0.5537,
      "step": 901
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2699116281949072,
      "learning_rate": 1.984243803843392e-05,
      "loss": 0.4287,
      "step": 902
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5958736923202048,
      "learning_rate": 1.98418984835851e-05,
      "loss": 0.5303,
      "step": 903
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4895930333183813,
      "learning_rate": 1.9841358013850413e-05,
      "loss": 0.5449,
      "step": 904
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4313602443342202,
      "learning_rate": 1.9840816629280087e-05,
      "loss": 0.5444,
      "step": 905
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.28629464236138,
      "learning_rate": 1.9840274329924452e-05,
      "loss": 0.459,
      "step": 906
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4701045252721656,
      "learning_rate": 1.983973111583392e-05,
      "loss": 0.4868,
      "step": 907
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.296369084500699,
      "learning_rate": 1.9839186987058986e-05,
      "loss": 0.4106,
      "step": 908
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.660069318602158,
      "learning_rate": 1.9838641943650234e-05,
      "loss": 0.6641,
      "step": 909
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.243334916686844,
      "learning_rate": 1.9838095985658324e-05,
      "loss": 0.4463,
      "step": 910
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.408795887313064,
      "learning_rate": 1.9837549113134015e-05,
      "loss": 0.5034,
      "step": 911
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5561005200132643,
      "learning_rate": 1.983700132612814e-05,
      "loss": 0.5786,
      "step": 912
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4622173108431848,
      "learning_rate": 1.9836452624691617e-05,
      "loss": 0.5283,
      "step": 913
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.541791171060721,
      "learning_rate": 1.9835903008875458e-05,
      "loss": 0.5811,
      "step": 914
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4922247846243972,
      "learning_rate": 1.983535247873075e-05,
      "loss": 0.5518,
      "step": 915
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4725703392573863,
      "learning_rate": 1.9834801034308674e-05,
      "loss": 0.4795,
      "step": 916
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.326910933693184,
      "learning_rate": 1.9834248675660484e-05,
      "loss": 0.478,
      "step": 917
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4243890367875014,
      "learning_rate": 1.9833695402837536e-05,
      "loss": 0.436,
      "step": 918
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.267302681949012,
      "learning_rate": 1.9833141215891253e-05,
      "loss": 0.4028,
      "step": 919
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2764546645033694,
      "learning_rate": 1.9832586114873154e-05,
      "loss": 0.4287,
      "step": 920
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4332432856244937,
      "learning_rate": 1.983203009983484e-05,
      "loss": 0.48,
      "step": 921
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2901822036948623,
      "learning_rate": 1.9831473170828003e-05,
      "loss": 0.5112,
      "step": 922
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5418789913396223,
      "learning_rate": 1.9830915327904402e-05,
      "loss": 0.5156,
      "step": 923
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5146759075953287,
      "learning_rate": 1.9830356571115904e-05,
      "loss": 0.585,
      "step": 924
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5116154850755572,
      "learning_rate": 1.9829796900514445e-05,
      "loss": 0.4907,
      "step": 925
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2909828947745143,
      "learning_rate": 1.982923631615205e-05,
      "loss": 0.4468,
      "step": 926
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.601317945493633,
      "learning_rate": 1.9828674818080837e-05,
      "loss": 0.5986,
      "step": 927
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2981257967911897,
      "learning_rate": 1.9828112406352994e-05,
      "loss": 0.4424,
      "step": 928
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4984009490809516,
      "learning_rate": 1.9827549081020806e-05,
      "loss": 0.5649,
      "step": 929
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4656795698636051,
      "learning_rate": 1.9826984842136637e-05,
      "loss": 0.4712,
      "step": 930
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4641307675722581,
      "learning_rate": 1.9826419689752935e-05,
      "loss": 0.4531,
      "step": 931
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3432141420285735,
      "learning_rate": 1.9825853623922244e-05,
      "loss": 0.4263,
      "step": 932
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3545365258393816,
      "learning_rate": 1.9825286644697176e-05,
      "loss": 0.4946,
      "step": 933
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3265958409055116,
      "learning_rate": 1.982471875213044e-05,
      "loss": 0.4395,
      "step": 934
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.17271374317102,
      "learning_rate": 1.9824149946274827e-05,
      "loss": 0.3594,
      "step": 935
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4574107463078858,
      "learning_rate": 1.982358022718321e-05,
      "loss": 0.5151,
      "step": 936
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.287628807367388,
      "learning_rate": 1.9823009594908553e-05,
      "loss": 0.4468,
      "step": 937
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2814547243978813,
      "learning_rate": 1.9822438049503894e-05,
      "loss": 0.4526,
      "step": 938
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3512945530948144,
      "learning_rate": 1.982186559102237e-05,
      "loss": 0.4868,
      "step": 939
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3011951504747994,
      "learning_rate": 1.982129221951719e-05,
      "loss": 0.4526,
      "step": 940
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4207175761288797,
      "learning_rate": 1.982071793504166e-05,
      "loss": 0.4512,
      "step": 941
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.277948692725099,
      "learning_rate": 1.982014273764916e-05,
      "loss": 0.4307,
      "step": 942
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4925562194983075,
      "learning_rate": 1.981956662739316e-05,
      "loss": 0.5488,
      "step": 943
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3984817139303765,
      "learning_rate": 1.9818989604327218e-05,
      "loss": 0.5542,
      "step": 944
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4307507405445445,
      "learning_rate": 1.9818411668504965e-05,
      "loss": 0.4912,
      "step": 945
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.406895983789908,
      "learning_rate": 1.981783281998013e-05,
      "loss": 0.4321,
      "step": 946
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5067307812751443,
      "learning_rate": 1.9817253058806525e-05,
      "loss": 0.6035,
      "step": 947
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6195855582912122,
      "learning_rate": 1.9816672385038033e-05,
      "loss": 0.6123,
      "step": 948
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3531817485372206,
      "learning_rate": 1.9816090798728648e-05,
      "loss": 0.4365,
      "step": 949
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4116171857516902,
      "learning_rate": 1.9815508299932417e-05,
      "loss": 0.4741,
      "step": 950
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5866130552111473,
      "learning_rate": 1.98149248887035e-05,
      "loss": 0.5967,
      "step": 951
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2889905822209926,
      "learning_rate": 1.9814340565096124e-05,
      "loss": 0.4102,
      "step": 952
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4029154300243303,
      "learning_rate": 1.981375532916461e-05,
      "loss": 0.4756,
      "step": 953
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6038383023996923,
      "learning_rate": 1.981316918096336e-05,
      "loss": 0.4976,
      "step": 954
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4466012677615843,
      "learning_rate": 1.9812582120546854e-05,
      "loss": 0.5195,
      "step": 955
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4327366065837883,
      "learning_rate": 1.9811994147969676e-05,
      "loss": 0.5405,
      "step": 956
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3606133360440416,
      "learning_rate": 1.9811405263286475e-05,
      "loss": 0.5005,
      "step": 957
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3898457263865347,
      "learning_rate": 1.9810815466551996e-05,
      "loss": 0.5142,
      "step": 958
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4670094532834037,
      "learning_rate": 1.9810224757821063e-05,
      "loss": 0.4692,
      "step": 959
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4397020370889504,
      "learning_rate": 1.980963313714859e-05,
      "loss": 0.4688,
      "step": 960
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3372545015357489,
      "learning_rate": 1.9809040604589572e-05,
      "loss": 0.4941,
      "step": 961
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4191289638028337,
      "learning_rate": 1.9808447160199087e-05,
      "loss": 0.5474,
      "step": 962
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4712482852497524,
      "learning_rate": 1.9807852804032306e-05,
      "loss": 0.5176,
      "step": 963
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3393767526049003,
      "learning_rate": 1.9807257536144474e-05,
      "loss": 0.4404,
      "step": 964
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5915101662854148,
      "learning_rate": 1.980666135659093e-05,
      "loss": 0.624,
      "step": 965
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3783501295998515,
      "learning_rate": 1.9806064265427093e-05,
      "loss": 0.5981,
      "step": 966
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3388704732848007,
      "learning_rate": 1.9805466262708464e-05,
      "loss": 0.4712,
      "step": 967
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.383889986531494,
      "learning_rate": 1.980486734849064e-05,
      "loss": 0.5259,
      "step": 968
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4260855947109286,
      "learning_rate": 1.9804267522829287e-05,
      "loss": 0.5098,
      "step": 969
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4355731584477662,
      "learning_rate": 1.9803666785780165e-05,
      "loss": 0.5674,
      "step": 970
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3344133852906463,
      "learning_rate": 1.9803065137399123e-05,
      "loss": 0.4414,
      "step": 971
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.451280136889525,
      "learning_rate": 1.980246257774208e-05,
      "loss": 0.5547,
      "step": 972
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5331401754718794,
      "learning_rate": 1.980185910686506e-05,
      "loss": 0.5767,
      "step": 973
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4243827700318499,
      "learning_rate": 1.9801254724824152e-05,
      "loss": 0.4678,
      "step": 974
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.366982875147301,
      "learning_rate": 1.9800649431675544e-05,
      "loss": 0.5063,
      "step": 975
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3359182617876415,
      "learning_rate": 1.98000432274755e-05,
      "loss": 0.4624,
      "step": 976
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4654827407893802,
      "learning_rate": 1.9799436112280374e-05,
      "loss": 0.5278,
      "step": 977
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5866159099808494,
      "learning_rate": 1.9798828086146596e-05,
      "loss": 0.5498,
      "step": 978
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6149949579184906,
      "learning_rate": 1.9798219149130692e-05,
      "loss": 0.5938,
      "step": 979
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5049279022878352,
      "learning_rate": 1.979760930128927e-05,
      "loss": 0.4434,
      "step": 980
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4177211282003719,
      "learning_rate": 1.9796998542679015e-05,
      "loss": 0.4902,
      "step": 981
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5833049821563199,
      "learning_rate": 1.979638687335671e-05,
      "loss": 0.542,
      "step": 982
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3730829945937553,
      "learning_rate": 1.9795774293379206e-05,
      "loss": 0.4414,
      "step": 983
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.314762452323158,
      "learning_rate": 1.979516080280345e-05,
      "loss": 0.416,
      "step": 984
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4082685026086879,
      "learning_rate": 1.979454640168647e-05,
      "loss": 0.4878,
      "step": 985
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5770719387026348,
      "learning_rate": 1.9793931090085385e-05,
      "loss": 0.6484,
      "step": 986
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.450861179316513,
      "learning_rate": 1.9793314868057387e-05,
      "loss": 0.5352,
      "step": 987
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4777478351236237,
      "learning_rate": 1.9792697735659766e-05,
      "loss": 0.5034,
      "step": 988
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2236396578344049,
      "learning_rate": 1.979207969294988e-05,
      "loss": 0.4409,
      "step": 989
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5139017973595386,
      "learning_rate": 1.979146073998519e-05,
      "loss": 0.4868,
      "step": 990
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2099745210339607,
      "learning_rate": 1.979084087682323e-05,
      "loss": 0.4624,
      "step": 991
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.225515378126224,
      "learning_rate": 1.979022010352162e-05,
      "loss": 0.437,
      "step": 992
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5151059199803072,
      "learning_rate": 1.9789598420138065e-05,
      "loss": 0.5273,
      "step": 993
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2887277309981273,
      "learning_rate": 1.9788975826730362e-05,
      "loss": 0.4395,
      "step": 994
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3355830041674372,
      "learning_rate": 1.9788352323356376e-05,
      "loss": 0.436,
      "step": 995
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3191176495185708,
      "learning_rate": 1.9787727910074077e-05,
      "loss": 0.4683,
      "step": 996
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4758162149160374,
      "learning_rate": 1.97871025869415e-05,
      "loss": 0.5063,
      "step": 997
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4436968420030611,
      "learning_rate": 1.9786476354016782e-05,
      "loss": 0.4531,
      "step": 998
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.466753085899181,
      "learning_rate": 1.9785849211358133e-05,
      "loss": 0.4023,
      "step": 999
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3580632917364361,
      "learning_rate": 1.9785221159023852e-05,
      "loss": 0.4429,
      "step": 1000
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3432980151382157,
      "learning_rate": 1.9784592197072317e-05,
      "loss": 0.4766,
      "step": 1001
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.365145403477454,
      "learning_rate": 1.9783962325562004e-05,
      "loss": 0.4478,
      "step": 1002
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4392132533295519,
      "learning_rate": 1.9783331544551457e-05,
      "loss": 0.48,
      "step": 1003
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3076323305430062,
      "learning_rate": 1.9782699854099316e-05,
      "loss": 0.4839,
      "step": 1004
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3363752951852934,
      "learning_rate": 1.97820672542643e-05,
      "loss": 0.52,
      "step": 1005
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3632764596789735,
      "learning_rate": 1.9781433745105218e-05,
      "loss": 0.4941,
      "step": 1006
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6300343154517176,
      "learning_rate": 1.9780799326680956e-05,
      "loss": 0.5034,
      "step": 1007
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4374287071518081,
      "learning_rate": 1.978016399905049e-05,
      "loss": 0.5122,
      "step": 1008
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5193125355014512,
      "learning_rate": 1.9779527762272877e-05,
      "loss": 0.5508,
      "step": 1009
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3392337069545506,
      "learning_rate": 1.9778890616407266e-05,
      "loss": 0.3979,
      "step": 1010
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2862783216483686,
      "learning_rate": 1.977825256151288e-05,
      "loss": 0.4365,
      "step": 1011
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2772281787829693,
      "learning_rate": 1.9777613597649033e-05,
      "loss": 0.3779,
      "step": 1012
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3893528779899904,
      "learning_rate": 1.9776973724875123e-05,
      "loss": 0.4741,
      "step": 1013
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6080709109939548,
      "learning_rate": 1.9776332943250628e-05,
      "loss": 0.5542,
      "step": 1014
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3251067343097829,
      "learning_rate": 1.9775691252835113e-05,
      "loss": 0.4683,
      "step": 1015
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4459064516925437,
      "learning_rate": 1.9775048653688235e-05,
      "loss": 0.4785,
      "step": 1016
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6355154557670404,
      "learning_rate": 1.9774405145869728e-05,
      "loss": 0.6875,
      "step": 1017
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5104220317230586,
      "learning_rate": 1.9773760729439405e-05,
      "loss": 0.4272,
      "step": 1018
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5687370953178155,
      "learning_rate": 1.9773115404457175e-05,
      "loss": 0.5542,
      "step": 1019
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.504182672200429,
      "learning_rate": 1.9772469170983026e-05,
      "loss": 0.5884,
      "step": 1020
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.402652209511977,
      "learning_rate": 1.9771822029077028e-05,
      "loss": 0.4883,
      "step": 1021
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4509019029585228,
      "learning_rate": 1.977117397879934e-05,
      "loss": 0.5176,
      "step": 1022
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5219578354339445,
      "learning_rate": 1.9770525020210204e-05,
      "loss": 0.5195,
      "step": 1023
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2421775556437544,
      "learning_rate": 1.976987515336995e-05,
      "loss": 0.4551,
      "step": 1024
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4474492999566317,
      "learning_rate": 1.9769224378338978e-05,
      "loss": 0.5347,
      "step": 1025
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.433230017089214,
      "learning_rate": 1.976857269517779e-05,
      "loss": 0.4492,
      "step": 1026
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3595530644432205,
      "learning_rate": 1.976792010394697e-05,
      "loss": 0.4858,
      "step": 1027
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4158203020996243,
      "learning_rate": 1.9767266604707172e-05,
      "loss": 0.4653,
      "step": 1028
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.438425830899571,
      "learning_rate": 1.9766612197519146e-05,
      "loss": 0.5264,
      "step": 1029
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6503422499402378,
      "learning_rate": 1.976595688244373e-05,
      "loss": 0.6045,
      "step": 1030
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2679924023611728,
      "learning_rate": 1.9765300659541837e-05,
      "loss": 0.374,
      "step": 1031
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.429591746688921,
      "learning_rate": 1.976464352887447e-05,
      "loss": 0.4883,
      "step": 1032
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2809155759208672,
      "learning_rate": 1.9763985490502714e-05,
      "loss": 0.4546,
      "step": 1033
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5761851867069898,
      "learning_rate": 1.976332654448774e-05,
      "loss": 0.542,
      "step": 1034
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.446279305090369,
      "learning_rate": 1.9762666690890796e-05,
      "loss": 0.4907,
      "step": 1035
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4174097892992075,
      "learning_rate": 1.976200592977323e-05,
      "loss": 0.481,
      "step": 1036
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3897235955325116,
      "learning_rate": 1.9761344261196464e-05,
      "loss": 0.521,
      "step": 1037
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4858771822027177,
      "learning_rate": 1.9760681685222e-05,
      "loss": 0.4688,
      "step": 1038
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3977780430060287,
      "learning_rate": 1.976001820191143e-05,
      "loss": 0.4409,
      "step": 1039
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5121738733615409,
      "learning_rate": 1.975935381132644e-05,
      "loss": 0.6343,
      "step": 1040
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3385217766481359,
      "learning_rate": 1.9758688513528783e-05,
      "loss": 0.5073,
      "step": 1041
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4148838755188622,
      "learning_rate": 1.9758022308580306e-05,
      "loss": 0.5513,
      "step": 1042
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2816669692801324,
      "learning_rate": 1.975735519654293e-05,
      "loss": 0.4458,
      "step": 1043
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2017122423043753,
      "learning_rate": 1.9756687177478683e-05,
      "loss": 0.4321,
      "step": 1044
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4716747031910962,
      "learning_rate": 1.975601825144965e-05,
      "loss": 0.5435,
      "step": 1045
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.407904224923888,
      "learning_rate": 1.9755348418518022e-05,
      "loss": 0.4336,
      "step": 1046
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4858245903991456,
      "learning_rate": 1.9754677678746064e-05,
      "loss": 0.4912,
      "step": 1047
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.438775809067378,
      "learning_rate": 1.9754006032196123e-05,
      "loss": 0.4565,
      "step": 1048
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3031832793905382,
      "learning_rate": 1.9753333478930632e-05,
      "loss": 0.4517,
      "step": 1049
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5194450811196583,
      "learning_rate": 1.975266001901212e-05,
      "loss": 0.4985,
      "step": 1050
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4658293746774724,
      "learning_rate": 1.9751985652503187e-05,
      "loss": 0.4614,
      "step": 1051
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3679297105311918,
      "learning_rate": 1.9751310379466514e-05,
      "loss": 0.5244,
      "step": 1052
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3059248585035073,
      "learning_rate": 1.9750634199964882e-05,
      "loss": 0.4033,
      "step": 1053
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.399948466935977,
      "learning_rate": 1.9749957114061143e-05,
      "loss": 0.4634,
      "step": 1054
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6189883469421626,
      "learning_rate": 1.9749279121818235e-05,
      "loss": 0.4971,
      "step": 1055
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4182176584122856,
      "learning_rate": 1.9748600223299192e-05,
      "loss": 0.5195,
      "step": 1056
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.378600345221092,
      "learning_rate": 1.9747920418567117e-05,
      "loss": 0.4731,
      "step": 1057
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4360656784406123,
      "learning_rate": 1.9747239707685205e-05,
      "loss": 0.4937,
      "step": 1058
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4522582696017565,
      "learning_rate": 1.974655809071673e-05,
      "loss": 0.5977,
      "step": 1059
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.440208129476182,
      "learning_rate": 1.9745875567725058e-05,
      "loss": 0.5132,
      "step": 1060
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4185651196796003,
      "learning_rate": 1.9745192138773633e-05,
      "loss": 0.3921,
      "step": 1061
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3263172800662013,
      "learning_rate": 1.974450780392599e-05,
      "loss": 0.4893,
      "step": 1062
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.264710539878131,
      "learning_rate": 1.9743822563245738e-05,
      "loss": 0.4502,
      "step": 1063
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2579451393527896,
      "learning_rate": 1.9743136416796578e-05,
      "loss": 0.3926,
      "step": 1064
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4375420419494163,
      "learning_rate": 1.9742449364642293e-05,
      "loss": 0.5503,
      "step": 1065
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5367320326411922,
      "learning_rate": 1.974176140684675e-05,
      "loss": 0.5015,
      "step": 1066
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.359721490146022,
      "learning_rate": 1.97410725434739e-05,
      "loss": 0.4941,
      "step": 1067
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4671422703602892,
      "learning_rate": 1.974038277458778e-05,
      "loss": 0.4946,
      "step": 1068
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3925559175114959,
      "learning_rate": 1.973969210025251e-05,
      "loss": 0.54,
      "step": 1069
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2390327035968698,
      "learning_rate": 1.973900052053229e-05,
      "loss": 0.4048,
      "step": 1070
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4176360058463882,
      "learning_rate": 1.973830803549141e-05,
      "loss": 0.4722,
      "step": 1071
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.469834431010734,
      "learning_rate": 1.9737614645194246e-05,
      "loss": 0.5752,
      "step": 1072
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.422292484318057,
      "learning_rate": 1.9736920349705252e-05,
      "loss": 0.5938,
      "step": 1073
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3622063916795104,
      "learning_rate": 1.9736225149088964e-05,
      "loss": 0.5063,
      "step": 1074
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4647637162912845,
      "learning_rate": 1.9735529043410012e-05,
      "loss": 0.5073,
      "step": 1075
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3449715905400403,
      "learning_rate": 1.97348320327331e-05,
      "loss": 0.4971,
      "step": 1076
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.303122283495748,
      "learning_rate": 1.9734134117123028e-05,
      "loss": 0.4155,
      "step": 1077
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.320539685434242,
      "learning_rate": 1.973343529664467e-05,
      "loss": 0.479,
      "step": 1078
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.346393772662831,
      "learning_rate": 1.9732735571362985e-05,
      "loss": 0.4956,
      "step": 1079
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2398158989007608,
      "learning_rate": 1.973203494134302e-05,
      "loss": 0.3677,
      "step": 1080
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3983922289340591,
      "learning_rate": 1.9731333406649905e-05,
      "loss": 0.4902,
      "step": 1081
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4552560110862272,
      "learning_rate": 1.973063096734885e-05,
      "loss": 0.4795,
      "step": 1082
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3425295269140536,
      "learning_rate": 1.9729927623505153e-05,
      "loss": 0.4717,
      "step": 1083
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3414221155823884,
      "learning_rate": 1.97292233751842e-05,
      "loss": 0.4663,
      "step": 1084
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2520867083556844,
      "learning_rate": 1.9728518222451454e-05,
      "loss": 0.4585,
      "step": 1085
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4058600699227664,
      "learning_rate": 1.9727812165372463e-05,
      "loss": 0.5161,
      "step": 1086
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.1907036176627308,
      "learning_rate": 1.972710520401287e-05,
      "loss": 0.3931,
      "step": 1087
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6136758796611501,
      "learning_rate": 1.972639733843838e-05,
      "loss": 0.5571,
      "step": 1088
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4110705590716281,
      "learning_rate": 1.97256885687148e-05,
      "loss": 0.479,
      "step": 1089
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3073552902001018,
      "learning_rate": 1.9724978894908017e-05,
      "loss": 0.4766,
      "step": 1090
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5914828444650972,
      "learning_rate": 1.9724268317084e-05,
      "loss": 0.562,
      "step": 1091
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4632829897327995,
      "learning_rate": 1.9723556835308807e-05,
      "loss": 0.4893,
      "step": 1092
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2512986866857425,
      "learning_rate": 1.9722844449648572e-05,
      "loss": 0.3965,
      "step": 1093
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2974532827047969,
      "learning_rate": 1.972213116016952e-05,
      "loss": 0.5317,
      "step": 1094
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4541929554526547,
      "learning_rate": 1.972141696693795e-05,
      "loss": 0.4321,
      "step": 1095
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2248535086763503,
      "learning_rate": 1.972070187002026e-05,
      "loss": 0.4287,
      "step": 1096
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5376991058034462,
      "learning_rate": 1.9719985869482925e-05,
      "loss": 0.5576,
      "step": 1097
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.408475981008393,
      "learning_rate": 1.9719268965392495e-05,
      "loss": 0.4663,
      "step": 1098
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3350048094188893,
      "learning_rate": 1.971855115781562e-05,
      "loss": 0.4722,
      "step": 1099
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.585775563299218,
      "learning_rate": 1.971783244681902e-05,
      "loss": 0.6289,
      "step": 1100
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4396703036147784,
      "learning_rate": 1.971711283246951e-05,
      "loss": 0.6143,
      "step": 1101
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6406534705810207,
      "learning_rate": 1.9716392314833982e-05,
      "loss": 0.5723,
      "step": 1102
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6063625024431978,
      "learning_rate": 1.9715670893979416e-05,
      "loss": 0.501,
      "step": 1103
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3899192999238985,
      "learning_rate": 1.9714948569972873e-05,
      "loss": 0.5117,
      "step": 1104
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.424732370386904,
      "learning_rate": 1.9714225342881495e-05,
      "loss": 0.4858,
      "step": 1105
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.1903957677757222,
      "learning_rate": 1.9713501212772518e-05,
      "loss": 0.4175,
      "step": 1106
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2578257261932844,
      "learning_rate": 1.9712776179713252e-05,
      "loss": 0.564,
      "step": 1107
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5208713952235482,
      "learning_rate": 1.9712050243771095e-05,
      "loss": 0.5908,
      "step": 1108
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.1510993489619998,
      "learning_rate": 1.9711323405013535e-05,
      "loss": 0.3716,
      "step": 1109
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.273319526922376,
      "learning_rate": 1.9710595663508125e-05,
      "loss": 0.4673,
      "step": 1110
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5314767338229565,
      "learning_rate": 1.9709867019322528e-05,
      "loss": 0.6763,
      "step": 1111
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4579257859542363,
      "learning_rate": 1.970913747252447e-05,
      "loss": 0.4883,
      "step": 1112
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5232016310096248,
      "learning_rate": 1.970840702318177e-05,
      "loss": 0.5811,
      "step": 1113
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3468360287936312,
      "learning_rate": 1.9707675671362328e-05,
      "loss": 0.4624,
      "step": 1114
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.434252221142746,
      "learning_rate": 1.970694341713413e-05,
      "loss": 0.519,
      "step": 1115
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.508092913488423,
      "learning_rate": 1.9706210260565246e-05,
      "loss": 0.5195,
      "step": 1116
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.45493989956685,
      "learning_rate": 1.970547620172383e-05,
      "loss": 0.5195,
      "step": 1117
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2405153788320662,
      "learning_rate": 1.9704741240678117e-05,
      "loss": 0.3896,
      "step": 1118
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2962761246657495,
      "learning_rate": 1.9704005377496428e-05,
      "loss": 0.4312,
      "step": 1119
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4996398194705138,
      "learning_rate": 1.9703268612247172e-05,
      "loss": 0.5049,
      "step": 1120
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1166673223579475,
      "learning_rate": 1.9702530944998825e-05,
      "loss": 0.4253,
      "step": 1121
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5267437905829886,
      "learning_rate": 1.9701792375819974e-05,
      "loss": 0.5142,
      "step": 1122
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3541298058882774,
      "learning_rate": 1.9701052904779265e-05,
      "loss": 0.4795,
      "step": 1123
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4115400883725617,
      "learning_rate": 1.9700312531945444e-05,
      "loss": 0.481,
      "step": 1124
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4455743201320277,
      "learning_rate": 1.969957125738733e-05,
      "loss": 0.5566,
      "step": 1125
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5467210001555358,
      "learning_rate": 1.969882908117383e-05,
      "loss": 0.5996,
      "step": 1126
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5075589531534888,
      "learning_rate": 1.969808600337394e-05,
      "loss": 0.5649,
      "step": 1127
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2782215926077898,
      "learning_rate": 1.969734202405674e-05,
      "loss": 0.4722,
      "step": 1128
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5855153609795938,
      "learning_rate": 1.9696597143291377e-05,
      "loss": 0.5771,
      "step": 1129
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.455772169611351,
      "learning_rate": 1.9695851361147098e-05,
      "loss": 0.5503,
      "step": 1130
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6422803927672336,
      "learning_rate": 1.9695104677693234e-05,
      "loss": 0.666,
      "step": 1131
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4814963271767199,
      "learning_rate": 1.969435709299919e-05,
      "loss": 0.5684,
      "step": 1132
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3608521236453879,
      "learning_rate": 1.9693608607134466e-05,
      "loss": 0.4609,
      "step": 1133
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4728912287369476,
      "learning_rate": 1.9692859220168634e-05,
      "loss": 0.5645,
      "step": 1134
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.30499055508731,
      "learning_rate": 1.969210893217136e-05,
      "loss": 0.3657,
      "step": 1135
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.340371205972667,
      "learning_rate": 1.9691357743212385e-05,
      "loss": 0.4604,
      "step": 1136
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3144512724099189,
      "learning_rate": 1.969060565336154e-05,
      "loss": 0.4678,
      "step": 1137
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.506460853868738,
      "learning_rate": 1.9689852662688743e-05,
      "loss": 0.5625,
      "step": 1138
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3231383484944559,
      "learning_rate": 1.9689098771263982e-05,
      "loss": 0.3965,
      "step": 1139
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4072254085339928,
      "learning_rate": 1.9688343979157348e-05,
      "loss": 0.4673,
      "step": 1140
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.388610856346805,
      "learning_rate": 1.9687588286438995e-05,
      "loss": 0.459,
      "step": 1141
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.426649734223562,
      "learning_rate": 1.9686831693179178e-05,
      "loss": 0.5684,
      "step": 1142
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4420482743913523,
      "learning_rate": 1.9686074199448222e-05,
      "loss": 0.4277,
      "step": 1143
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.335847727031508,
      "learning_rate": 1.9685315805316548e-05,
      "loss": 0.4512,
      "step": 1144
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5887997089617394,
      "learning_rate": 1.9684556510854655e-05,
      "loss": 0.6094,
      "step": 1145
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4715017077979302,
      "learning_rate": 1.968379631613312e-05,
      "loss": 0.5098,
      "step": 1146
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3381830587584356,
      "learning_rate": 1.9683035221222617e-05,
      "loss": 0.5479,
      "step": 1147
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3884558234725104,
      "learning_rate": 1.968227322619389e-05,
      "loss": 0.4927,
      "step": 1148
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5478407444897602,
      "learning_rate": 1.968151033111777e-05,
      "loss": 0.5444,
      "step": 1149
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3658726158731074,
      "learning_rate": 1.968074653606519e-05,
      "loss": 0.4307,
      "step": 1150
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3279531465681376,
      "learning_rate": 1.967998184110713e-05,
      "loss": 0.4663,
      "step": 1151
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.405195188114155,
      "learning_rate": 1.9679216246314694e-05,
      "loss": 0.4995,
      "step": 1152
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5725554122859284,
      "learning_rate": 1.9678449751759036e-05,
      "loss": 0.4922,
      "step": 1153
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6876391720875625,
      "learning_rate": 1.9677682357511415e-05,
      "loss": 0.6279,
      "step": 1154
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3020857763747091,
      "learning_rate": 1.9676914063643165e-05,
      "loss": 0.4795,
      "step": 1155
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3449904596669406,
      "learning_rate": 1.9676144870225705e-05,
      "loss": 0.4756,
      "step": 1156
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3082746901295739,
      "learning_rate": 1.967537477733054e-05,
      "loss": 0.4224,
      "step": 1157
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5139235361420502,
      "learning_rate": 1.9674603785029252e-05,
      "loss": 0.5195,
      "step": 1158
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4451900645460256,
      "learning_rate": 1.967383189339352e-05,
      "loss": 0.5977,
      "step": 1159
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4553362378902055,
      "learning_rate": 1.9673059102495084e-05,
      "loss": 0.5366,
      "step": 1160
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.64532240467246,
      "learning_rate": 1.967228541240579e-05,
      "loss": 0.5942,
      "step": 1161
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4276191918058296,
      "learning_rate": 1.9671510823197562e-05,
      "loss": 0.5908,
      "step": 1162
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3588426235949198,
      "learning_rate": 1.9670735334942398e-05,
      "loss": 0.4229,
      "step": 1163
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3881379364158486,
      "learning_rate": 1.966995894771239e-05,
      "loss": 0.5288,
      "step": 1164
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4754276200856564,
      "learning_rate": 1.9669181661579708e-05,
      "loss": 0.4829,
      "step": 1165
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6438899269731078,
      "learning_rate": 1.9668403476616604e-05,
      "loss": 0.6396,
      "step": 1166
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2394928161458503,
      "learning_rate": 1.9667624392895423e-05,
      "loss": 0.4888,
      "step": 1167
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3408532475426407,
      "learning_rate": 1.966684441048858e-05,
      "loss": 0.4883,
      "step": 1168
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4163354327094917,
      "learning_rate": 1.966606352946859e-05,
      "loss": 0.4736,
      "step": 1169
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2367678908383766,
      "learning_rate": 1.9665281749908034e-05,
      "loss": 0.4746,
      "step": 1170
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2545480963147269,
      "learning_rate": 1.966449907187959e-05,
      "loss": 0.4824,
      "step": 1171
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.511661768134115,
      "learning_rate": 1.9663715495456012e-05,
      "loss": 0.5361,
      "step": 1172
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3193711980779252,
      "learning_rate": 1.9662931020710138e-05,
      "loss": 0.4453,
      "step": 1173
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4740000980868542,
      "learning_rate": 1.9662145647714896e-05,
      "loss": 0.5054,
      "step": 1174
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4512936511447094,
      "learning_rate": 1.9661359376543295e-05,
      "loss": 0.5332,
      "step": 1175
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.274057664394431,
      "learning_rate": 1.9660572207268416e-05,
      "loss": 0.4951,
      "step": 1176
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2617455106701447,
      "learning_rate": 1.9659784139963437e-05,
      "loss": 0.4531,
      "step": 1177
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5272106997211543,
      "learning_rate": 1.9658995174701622e-05,
      "loss": 0.5332,
      "step": 1178
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3679882572142552,
      "learning_rate": 1.9658205311556304e-05,
      "loss": 0.4922,
      "step": 1179
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4149156243178438,
      "learning_rate": 1.9657414550600907e-05,
      "loss": 0.4775,
      "step": 1180
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3505868433473762,
      "learning_rate": 1.9656622891908944e-05,
      "loss": 0.457,
      "step": 1181
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3218704799045364,
      "learning_rate": 1.9655830335554004e-05,
      "loss": 0.4741,
      "step": 1182
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3358853462228215,
      "learning_rate": 1.9655036881609763e-05,
      "loss": 0.5254,
      "step": 1183
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3212760451984895,
      "learning_rate": 1.9654242530149974e-05,
      "loss": 0.4055,
      "step": 1184
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4181706060342694,
      "learning_rate": 1.9653447281248484e-05,
      "loss": 0.4912,
      "step": 1185
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5815877562722658,
      "learning_rate": 1.9652651134979215e-05,
      "loss": 0.4688,
      "step": 1186
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3708063392080219,
      "learning_rate": 1.9651854091416175e-05,
      "loss": 0.5112,
      "step": 1187
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.556247754367963,
      "learning_rate": 1.965105615063346e-05,
      "loss": 0.5396,
      "step": 1188
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4221623471604905,
      "learning_rate": 1.965025731270524e-05,
      "loss": 0.543,
      "step": 1189
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.472376130640175,
      "learning_rate": 1.964945757770578e-05,
      "loss": 0.5654,
      "step": 1190
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3200802127839302,
      "learning_rate": 1.9648656945709413e-05,
      "loss": 0.4966,
      "step": 1191
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3399486974445807,
      "learning_rate": 1.964785541679057e-05,
      "loss": 0.4224,
      "step": 1192
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5701943783222678,
      "learning_rate": 1.964705299102376e-05,
      "loss": 0.5278,
      "step": 1193
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4123610668840818,
      "learning_rate": 1.9646249668483575e-05,
      "loss": 0.5308,
      "step": 1194
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.601353861664167,
      "learning_rate": 1.964544544924469e-05,
      "loss": 0.519,
      "step": 1195
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2807095387212584,
      "learning_rate": 1.9644640333381862e-05,
      "loss": 0.4473,
      "step": 1196
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.267385806074116,
      "learning_rate": 1.964383432096993e-05,
      "loss": 0.4546,
      "step": 1197
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3450662771081714,
      "learning_rate": 1.964302741208383e-05,
      "loss": 0.5161,
      "step": 1198
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3104075767601928,
      "learning_rate": 1.9642219606798566e-05,
      "loss": 0.4448,
      "step": 1199
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3988933396354533,
      "learning_rate": 1.964141090518923e-05,
      "loss": 0.5215,
      "step": 1200
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.453933063456526,
      "learning_rate": 1.964060130733099e-05,
      "loss": 0.5703,
      "step": 1201
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4331912762307188,
      "learning_rate": 1.963979081329912e-05,
      "loss": 0.4824,
      "step": 1202
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5532658899898204,
      "learning_rate": 1.9638979423168948e-05,
      "loss": 0.4224,
      "step": 1203
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.1966899805586897,
      "learning_rate": 1.9638167137015905e-05,
      "loss": 0.4033,
      "step": 1204
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.424424750555133,
      "learning_rate": 1.9637353954915503e-05,
      "loss": 0.4697,
      "step": 1205
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.395931177769199,
      "learning_rate": 1.9636539876943325e-05,
      "loss": 0.4829,
      "step": 1206
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4931923672864458,
      "learning_rate": 1.9635724903175055e-05,
      "loss": 0.6035,
      "step": 1207
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.371090919365124,
      "learning_rate": 1.963490903368645e-05,
      "loss": 0.5073,
      "step": 1208
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2626443416444144,
      "learning_rate": 1.963409226855335e-05,
      "loss": 0.4756,
      "step": 1209
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3349878140337879,
      "learning_rate": 1.963327460785168e-05,
      "loss": 0.502,
      "step": 1210
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3863278025089711,
      "learning_rate": 1.9632456051657448e-05,
      "loss": 0.4082,
      "step": 1211
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4286846393607633,
      "learning_rate": 1.9631636600046748e-05,
      "loss": 0.4707,
      "step": 1212
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2486430452604214,
      "learning_rate": 1.9630816253095754e-05,
      "loss": 0.4878,
      "step": 1213
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3777727540636955,
      "learning_rate": 1.9629995010880724e-05,
      "loss": 0.4238,
      "step": 1214
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4138428951055082,
      "learning_rate": 1.9629172873477995e-05,
      "loss": 0.5454,
      "step": 1215
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.291322804327372,
      "learning_rate": 1.9628349840963997e-05,
      "loss": 0.4121,
      "step": 1216
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4187696000283827,
      "learning_rate": 1.9627525913415234e-05,
      "loss": 0.4673,
      "step": 1217
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.672076875428312,
      "learning_rate": 1.96267010909083e-05,
      "loss": 0.6128,
      "step": 1218
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4611697924494988,
      "learning_rate": 1.9625875373519866e-05,
      "loss": 0.54,
      "step": 1219
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6524261988863052,
      "learning_rate": 1.962504876132669e-05,
      "loss": 0.582,
      "step": 1220
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2638821212622047,
      "learning_rate": 1.962422125440562e-05,
      "loss": 0.4438,
      "step": 1221
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2403396424298683,
      "learning_rate": 1.962339285283357e-05,
      "loss": 0.4951,
      "step": 1222
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5267651460358547,
      "learning_rate": 1.9622563556687545e-05,
      "loss": 0.5527,
      "step": 1223
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4395873509530808,
      "learning_rate": 1.9621733366044644e-05,
      "loss": 0.4131,
      "step": 1224
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3987245823830523,
      "learning_rate": 1.962090228098203e-05,
      "loss": 0.5229,
      "step": 1225
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3064527400729613,
      "learning_rate": 1.9620070301576974e-05,
      "loss": 0.3599,
      "step": 1226
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3619871277061613,
      "learning_rate": 1.96192374279068e-05,
      "loss": 0.4922,
      "step": 1227
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4774875882296814,
      "learning_rate": 1.9618403660048937e-05,
      "loss": 0.5703,
      "step": 1228
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.412873084338238,
      "learning_rate": 1.9617568998080893e-05,
      "loss": 0.5493,
      "step": 1229
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.327586348734748,
      "learning_rate": 1.9616733442080253e-05,
      "loss": 0.5229,
      "step": 1230
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6168712527945073,
      "learning_rate": 1.961589699212469e-05,
      "loss": 0.5908,
      "step": 1231
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4227568100878825,
      "learning_rate": 1.9615059648291956e-05,
      "loss": 0.5527,
      "step": 1232
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2738091077541955,
      "learning_rate": 1.9614221410659896e-05,
      "loss": 0.4795,
      "step": 1233
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3308117896606046,
      "learning_rate": 1.9613382279306426e-05,
      "loss": 0.4658,
      "step": 1234
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3018344795094647,
      "learning_rate": 1.961254225430955e-05,
      "loss": 0.5107,
      "step": 1235
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5847231406581617,
      "learning_rate": 1.961170133574736e-05,
      "loss": 0.4751,
      "step": 1236
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.341891089200775,
      "learning_rate": 1.961085952369802e-05,
      "loss": 0.4258,
      "step": 1237
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.564184858079409,
      "learning_rate": 1.9610016818239788e-05,
      "loss": 0.5464,
      "step": 1238
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.306306452966194,
      "learning_rate": 1.9609173219450998e-05,
      "loss": 0.4009,
      "step": 1239
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4059528507700727,
      "learning_rate": 1.960832872741007e-05,
      "loss": 0.5444,
      "step": 1240
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4044869931346122,
      "learning_rate": 1.9607483342195505e-05,
      "loss": 0.4956,
      "step": 1241
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4006320102105527,
      "learning_rate": 1.9606637063885892e-05,
      "loss": 0.5449,
      "step": 1242
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2047896850184867,
      "learning_rate": 1.9605789892559902e-05,
      "loss": 0.3916,
      "step": 1243
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2889404423956505,
      "learning_rate": 1.9604941828296275e-05,
      "loss": 0.4741,
      "step": 1244
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3444355401650079,
      "learning_rate": 1.960409287117386e-05,
      "loss": 0.4619,
      "step": 1245
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5235459011430672,
      "learning_rate": 1.9603243021271562e-05,
      "loss": 0.5107,
      "step": 1246
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6290166966096817,
      "learning_rate": 1.960239227866839e-05,
      "loss": 0.5747,
      "step": 1247
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2942940391641622,
      "learning_rate": 1.9601540643443423e-05,
      "loss": 0.4731,
      "step": 1248
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3547859268396565,
      "learning_rate": 1.960068811567583e-05,
      "loss": 0.4639,
      "step": 1249
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3371734490984444,
      "learning_rate": 1.9599834695444863e-05,
      "loss": 0.4546,
      "step": 1250
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3052894648817224,
      "learning_rate": 1.9598980382829848e-05,
      "loss": 0.4062,
      "step": 1251
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5255586787458542,
      "learning_rate": 1.9598125177910204e-05,
      "loss": 0.626,
      "step": 1252
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.400601856493461,
      "learning_rate": 1.959726908076543e-05,
      "loss": 0.4761,
      "step": 1253
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2308968071162,
      "learning_rate": 1.9596412091475107e-05,
      "loss": 0.4077,
      "step": 1254
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3472227835697808,
      "learning_rate": 1.9595554210118896e-05,
      "loss": 0.4189,
      "step": 1255
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4779657785955773,
      "learning_rate": 1.9594695436776548e-05,
      "loss": 0.4507,
      "step": 1256
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4182946990021843,
      "learning_rate": 1.9593835771527893e-05,
      "loss": 0.4805,
      "step": 1257
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3465464324897523,
      "learning_rate": 1.9592975214452844e-05,
      "loss": 0.4692,
      "step": 1258
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5071386788722667,
      "learning_rate": 1.959211376563139e-05,
      "loss": 0.4585,
      "step": 1259
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4072205336510821,
      "learning_rate": 1.959125142514362e-05,
      "loss": 0.5171,
      "step": 1260
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2256615785152092,
      "learning_rate": 1.959038819306969e-05,
      "loss": 0.4434,
      "step": 1261
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.325441009573899,
      "learning_rate": 1.958952406948985e-05,
      "loss": 0.459,
      "step": 1262
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2777279816440135,
      "learning_rate": 1.9588659054484417e-05,
      "loss": 0.3413,
      "step": 1263
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.324702782987114,
      "learning_rate": 1.958779314813381e-05,
      "loss": 0.4604,
      "step": 1264
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3905065544997919,
      "learning_rate": 1.9586926350518522e-05,
      "loss": 0.4971,
      "step": 1265
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3220898771166374,
      "learning_rate": 1.9586058661719124e-05,
      "loss": 0.521,
      "step": 1266
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3053165017335613,
      "learning_rate": 1.9585190081816277e-05,
      "loss": 0.4575,
      "step": 1267
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4453859947823227,
      "learning_rate": 1.958432061089073e-05,
      "loss": 0.4526,
      "step": 1268
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.435565056014725,
      "learning_rate": 1.9583450249023292e-05,
      "loss": 0.5796,
      "step": 1269
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3014068137687589,
      "learning_rate": 1.9582578996294882e-05,
      "loss": 0.4707,
      "step": 1270
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3023701536323553,
      "learning_rate": 1.9581706852786492e-05,
      "loss": 0.4751,
      "step": 1271
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3315704700004212,
      "learning_rate": 1.9580833818579184e-05,
      "loss": 0.4946,
      "step": 1272
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.221064152802467,
      "learning_rate": 1.9579959893754123e-05,
      "loss": 0.4316,
      "step": 1273
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3203477827153376,
      "learning_rate": 1.9579085078392543e-05,
      "loss": 0.4834,
      "step": 1274
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3758881462253347,
      "learning_rate": 1.9578209372575766e-05,
      "loss": 0.457,
      "step": 1275
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4920551804620035,
      "learning_rate": 1.95773327763852e-05,
      "loss": 0.4956,
      "step": 1276
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.265897283913735,
      "learning_rate": 1.9576455289902327e-05,
      "loss": 0.4219,
      "step": 1277
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.533308902979511,
      "learning_rate": 1.9575576913208718e-05,
      "loss": 0.5195,
      "step": 1278
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4946922700443588,
      "learning_rate": 1.9574697646386027e-05,
      "loss": 0.5977,
      "step": 1279
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3739202509208983,
      "learning_rate": 1.957381748951599e-05,
      "loss": 0.4185,
      "step": 1280
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5855342536803647,
      "learning_rate": 1.9572936442680417e-05,
      "loss": 0.5591,
      "step": 1281
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4848176003841025,
      "learning_rate": 1.957205450596122e-05,
      "loss": 0.4336,
      "step": 1282
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7276584137944204,
      "learning_rate": 1.9571171679440374e-05,
      "loss": 0.6992,
      "step": 1283
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5834089267399407,
      "learning_rate": 1.9570287963199947e-05,
      "loss": 0.5957,
      "step": 1284
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3705397612965156,
      "learning_rate": 1.956940335732209e-05,
      "loss": 0.4766,
      "step": 1285
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4304531381535566,
      "learning_rate": 1.9568517861889035e-05,
      "loss": 0.5063,
      "step": 1286
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4133958119607888,
      "learning_rate": 1.9567631476983088e-05,
      "loss": 0.5605,
      "step": 1287
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4167321023947343,
      "learning_rate": 1.9566744202686657e-05,
      "loss": 0.5728,
      "step": 1288
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2500936970446348,
      "learning_rate": 1.9565856039082213e-05,
      "loss": 0.4136,
      "step": 1289
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2792278767305165,
      "learning_rate": 1.9564966986252326e-05,
      "loss": 0.501,
      "step": 1290
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4177134783982768,
      "learning_rate": 1.956407704427963e-05,
      "loss": 0.5146,
      "step": 1291
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2810625907325173,
      "learning_rate": 1.9563186213246864e-05,
      "loss": 0.4863,
      "step": 1292
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4837782923848104,
      "learning_rate": 1.9562294493236834e-05,
      "loss": 0.501,
      "step": 1293
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5499673176012718,
      "learning_rate": 1.9561401884332424e-05,
      "loss": 0.5947,
      "step": 1294
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5445619608116261,
      "learning_rate": 1.9560508386616624e-05,
      "loss": 0.4614,
      "step": 1295
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2964939792544314,
      "learning_rate": 1.9559614000172483e-05,
      "loss": 0.5552,
      "step": 1296
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4018173881632394,
      "learning_rate": 1.9558718725083143e-05,
      "loss": 0.5474,
      "step": 1297
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4048125632157245,
      "learning_rate": 1.955782256143183e-05,
      "loss": 0.5278,
      "step": 1298
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2909425310478881,
      "learning_rate": 1.9556925509301844e-05,
      "loss": 0.4927,
      "step": 1299
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.1462327583424399,
      "learning_rate": 1.9556027568776577e-05,
      "loss": 0.3667,
      "step": 1300
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3504026418460136,
      "learning_rate": 1.9555128739939504e-05,
      "loss": 0.5107,
      "step": 1301
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4679770768813878,
      "learning_rate": 1.9554229022874175e-05,
      "loss": 0.5527,
      "step": 1302
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.1521562301150758,
      "learning_rate": 1.9553328417664223e-05,
      "loss": 0.4019,
      "step": 1303
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3709235705819942,
      "learning_rate": 1.9552426924393368e-05,
      "loss": 0.4644,
      "step": 1304
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3451341170394822,
      "learning_rate": 1.9551524543145417e-05,
      "loss": 0.4326,
      "step": 1305
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3116492908796922,
      "learning_rate": 1.9550621274004248e-05,
      "loss": 0.4849,
      "step": 1306
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3944584004494738,
      "learning_rate": 1.9549717117053828e-05,
      "loss": 0.5273,
      "step": 1307
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3089740491056612,
      "learning_rate": 1.9548812072378208e-05,
      "loss": 0.4473,
      "step": 1308
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2466562495308007,
      "learning_rate": 1.9547906140061515e-05,
      "loss": 0.4707,
      "step": 1309
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4853407400996277,
      "learning_rate": 1.9546999320187966e-05,
      "loss": 0.625,
      "step": 1310
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5460016507925818,
      "learning_rate": 1.954609161284186e-05,
      "loss": 0.5176,
      "step": 1311
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5888228879393962,
      "learning_rate": 1.9545183018107576e-05,
      "loss": 0.563,
      "step": 1312
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4033557653242719,
      "learning_rate": 1.9544273536069573e-05,
      "loss": 0.5151,
      "step": 1313
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4216738433923612,
      "learning_rate": 1.9543363166812387e-05,
      "loss": 0.5732,
      "step": 1314
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3172510362698884,
      "learning_rate": 1.9542451910420655e-05,
      "loss": 0.5474,
      "step": 1315
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2748691669735104,
      "learning_rate": 1.9541539766979087e-05,
      "loss": 0.4365,
      "step": 1316
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4962287909974805,
      "learning_rate": 1.954062673657247e-05,
      "loss": 0.5493,
      "step": 1317
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2305387649880188,
      "learning_rate": 1.9539712819285674e-05,
      "loss": 0.4478,
      "step": 1318
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.626697857013644,
      "learning_rate": 1.953879801520366e-05,
      "loss": 0.6113,
      "step": 1319
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4168879117774467,
      "learning_rate": 1.953788232441147e-05,
      "loss": 0.5095,
      "step": 1320
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.419559167760876,
      "learning_rate": 1.9536965746994213e-05,
      "loss": 0.4756,
      "step": 1321
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4703241945464274,
      "learning_rate": 1.9536048283037105e-05,
      "loss": 0.5728,
      "step": 1322
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2061352203259517,
      "learning_rate": 1.9535129932625425e-05,
      "loss": 0.3892,
      "step": 1323
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4352665592024356,
      "learning_rate": 1.9534210695844543e-05,
      "loss": 0.4551,
      "step": 1324
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2420675510030188,
      "learning_rate": 1.9533290572779912e-05,
      "loss": 0.4487,
      "step": 1325
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.340815265703987,
      "learning_rate": 1.9532369563517066e-05,
      "loss": 0.439,
      "step": 1326
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.377550147668576,
      "learning_rate": 1.953144766814161e-05,
      "loss": 0.4385,
      "step": 1327
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3711529305306656,
      "learning_rate": 1.9530524886739254e-05,
      "loss": 0.4731,
      "step": 1328
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3028873154492406,
      "learning_rate": 1.952960121939577e-05,
      "loss": 0.415,
      "step": 1329
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2046486879423945,
      "learning_rate": 1.9528676666197026e-05,
      "loss": 0.3745,
      "step": 1330
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.662697005208789,
      "learning_rate": 1.9527751227228964e-05,
      "loss": 0.6504,
      "step": 1331
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4584695460391852,
      "learning_rate": 1.9526824902577614e-05,
      "loss": 0.5029,
      "step": 1332
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3283835653631506,
      "learning_rate": 1.9525897692329082e-05,
      "loss": 0.4854,
      "step": 1333
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.481189103436298,
      "learning_rate": 1.952496959656956e-05,
      "loss": 0.5181,
      "step": 1334
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5447946473309349,
      "learning_rate": 1.9524040615385324e-05,
      "loss": 0.522,
      "step": 1335
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3798930658729276,
      "learning_rate": 1.9523110748862733e-05,
      "loss": 0.5835,
      "step": 1336
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3796987182349507,
      "learning_rate": 1.952217999708822e-05,
      "loss": 0.4917,
      "step": 1337
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4293849324160142,
      "learning_rate": 1.952124836014831e-05,
      "loss": 0.5439,
      "step": 1338
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.338259274450012,
      "learning_rate": 1.9520315838129602e-05,
      "loss": 0.4419,
      "step": 1339
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3093208867878379,
      "learning_rate": 1.951938243111879e-05,
      "loss": 0.4741,
      "step": 1340
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6267521676212038,
      "learning_rate": 1.9518448139202632e-05,
      "loss": 0.6543,
      "step": 1341
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3601654334807223,
      "learning_rate": 1.9517512962467987e-05,
      "loss": 0.4624,
      "step": 1342
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5639055859556679,
      "learning_rate": 1.951657690100178e-05,
      "loss": 0.5991,
      "step": 1343
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2983188223283655,
      "learning_rate": 1.951563995489103e-05,
      "loss": 0.4355,
      "step": 1344
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4043531897690515,
      "learning_rate": 1.9514702124222837e-05,
      "loss": 0.4355,
      "step": 1345
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.345538260147685,
      "learning_rate": 1.951376340908437e-05,
      "loss": 0.4751,
      "step": 1346
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5347221552905914,
      "learning_rate": 1.95128238095629e-05,
      "loss": 0.6055,
      "step": 1347
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4976954211050493,
      "learning_rate": 1.9511883325745767e-05,
      "loss": 0.5322,
      "step": 1348
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4755915902280938,
      "learning_rate": 1.9510941957720396e-05,
      "loss": 0.5034,
      "step": 1349
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4819031941903391,
      "learning_rate": 1.9509999705574293e-05,
      "loss": 0.5454,
      "step": 1350
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3463949355069629,
      "learning_rate": 1.950905656939505e-05,
      "loss": 0.4922,
      "step": 1351
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3619789011998646,
      "learning_rate": 1.9508112549270346e-05,
      "loss": 0.5684,
      "step": 1352
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.433352300909516,
      "learning_rate": 1.9507167645287926e-05,
      "loss": 0.4912,
      "step": 1353
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.416047173315535,
      "learning_rate": 1.950622185753563e-05,
      "loss": 0.5225,
      "step": 1354
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5804570579102557,
      "learning_rate": 1.9505275186101378e-05,
      "loss": 0.5762,
      "step": 1355
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4860709518105044,
      "learning_rate": 1.950432763107317e-05,
      "loss": 0.5601,
      "step": 1356
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3317119201163696,
      "learning_rate": 1.9503379192539086e-05,
      "loss": 0.5039,
      "step": 1357
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5547291554275517,
      "learning_rate": 1.9502429870587295e-05,
      "loss": 0.5415,
      "step": 1358
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2618560814487103,
      "learning_rate": 1.9501479665306046e-05,
      "loss": 0.4141,
      "step": 1359
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2314513703326107,
      "learning_rate": 1.9500528576783667e-05,
      "loss": 0.4761,
      "step": 1360
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.534833999112915,
      "learning_rate": 1.9499576605108564e-05,
      "loss": 0.563,
      "step": 1361
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2975395545941855,
      "learning_rate": 1.949862375036924e-05,
      "loss": 0.5493,
      "step": 1362
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3930480429425698,
      "learning_rate": 1.949767001265426e-05,
      "loss": 0.4907,
      "step": 1363
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3332595913401106,
      "learning_rate": 1.949671539205229e-05,
      "loss": 0.4302,
      "step": 1364
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3079322391515713,
      "learning_rate": 1.9495759888652072e-05,
      "loss": 0.4634,
      "step": 1365
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3640244944777513,
      "learning_rate": 1.9494803502542415e-05,
      "loss": 0.4805,
      "step": 1366
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3644166715365083,
      "learning_rate": 1.949384623381224e-05,
      "loss": 0.4756,
      "step": 1367
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4473502386042352,
      "learning_rate": 1.949288808255052e-05,
      "loss": 0.4868,
      "step": 1368
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.561785679249979,
      "learning_rate": 1.9491929048846328e-05,
      "loss": 0.5601,
      "step": 1369
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3737803254594716,
      "learning_rate": 1.9490969132788815e-05,
      "loss": 0.4517,
      "step": 1370
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3225624111214813,
      "learning_rate": 1.9490008334467212e-05,
      "loss": 0.4116,
      "step": 1371
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4675005363974052,
      "learning_rate": 1.948904665397083e-05,
      "loss": 0.4521,
      "step": 1372
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3919736249448442,
      "learning_rate": 1.948808409138907e-05,
      "loss": 0.4805,
      "step": 1373
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.378449837464275,
      "learning_rate": 1.948712064681141e-05,
      "loss": 0.4668,
      "step": 1374
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3934936837887357,
      "learning_rate": 1.9486156320327406e-05,
      "loss": 0.48,
      "step": 1375
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2312444374901514,
      "learning_rate": 1.9485191112026707e-05,
      "loss": 0.3984,
      "step": 1376
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4248834654590088,
      "learning_rate": 1.9484225021999032e-05,
      "loss": 0.4331,
      "step": 1377
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4882717207009393,
      "learning_rate": 1.9483258050334183e-05,
      "loss": 0.6094,
      "step": 1378
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3583784219512254,
      "learning_rate": 1.9482290197122054e-05,
      "loss": 0.5098,
      "step": 1379
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.50214505364165,
      "learning_rate": 1.9481321462452617e-05,
      "loss": 0.5049,
      "step": 1380
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.426196615341957,
      "learning_rate": 1.9480351846415918e-05,
      "loss": 0.458,
      "step": 1381
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3038516515170553,
      "learning_rate": 1.9479381349102095e-05,
      "loss": 0.4937,
      "step": 1382
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.417958482712624,
      "learning_rate": 1.947840997060136e-05,
      "loss": 0.5117,
      "step": 1383
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4499673856168422,
      "learning_rate": 1.9477437711004015e-05,
      "loss": 0.54,
      "step": 1384
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5790171620892945,
      "learning_rate": 1.9476464570400434e-05,
      "loss": 0.5283,
      "step": 1385
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4050344583131003,
      "learning_rate": 1.9475490548881083e-05,
      "loss": 0.4907,
      "step": 1386
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4115834179862894,
      "learning_rate": 1.9474515646536507e-05,
      "loss": 0.5488,
      "step": 1387
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5040225850485571,
      "learning_rate": 1.947353986345733e-05,
      "loss": 0.5381,
      "step": 1388
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3059847272585463,
      "learning_rate": 1.9472563199734254e-05,
      "loss": 0.4854,
      "step": 1389
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4479654782092204,
      "learning_rate": 1.9471585655458073e-05,
      "loss": 0.5308,
      "step": 1390
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.415450539374477,
      "learning_rate": 1.9470607230719654e-05,
      "loss": 0.4878,
      "step": 1391
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5206227151382226,
      "learning_rate": 1.9469627925609956e-05,
      "loss": 0.5801,
      "step": 1392
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3527874719554465,
      "learning_rate": 1.946864774022001e-05,
      "loss": 0.4814,
      "step": 1393
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.377448746339702,
      "learning_rate": 1.946766667464093e-05,
      "loss": 0.5879,
      "step": 1394
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3533096794200343,
      "learning_rate": 1.9466684728963914e-05,
      "loss": 0.521,
      "step": 1395
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2820476947873525,
      "learning_rate": 1.9465701903280246e-05,
      "loss": 0.4419,
      "step": 1396
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4088050229723903,
      "learning_rate": 1.9464718197681284e-05,
      "loss": 0.4595,
      "step": 1397
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5062151947463285,
      "learning_rate": 1.9463733612258476e-05,
      "loss": 0.5327,
      "step": 1398
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4424084587699268,
      "learning_rate": 1.9462748147103342e-05,
      "loss": 0.4683,
      "step": 1399
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1380140351069692,
      "learning_rate": 1.9461761802307494e-05,
      "loss": 0.3989,
      "step": 1400
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4167914114109375,
      "learning_rate": 1.9460774577962622e-05,
      "loss": 0.4658,
      "step": 1401
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.388376941961101,
      "learning_rate": 1.945978647416049e-05,
      "loss": 0.5039,
      "step": 1402
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4559340622550017,
      "learning_rate": 1.9458797490992954e-05,
      "loss": 0.5488,
      "step": 1403
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4657121444694556,
      "learning_rate": 1.9457807628551947e-05,
      "loss": 0.4961,
      "step": 1404
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3567708105727652,
      "learning_rate": 1.945681688692949e-05,
      "loss": 0.4326,
      "step": 1405
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5686238431522173,
      "learning_rate": 1.9455825266217674e-05,
      "loss": 0.5649,
      "step": 1406
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3886700511703083,
      "learning_rate": 1.945483276650868e-05,
      "loss": 0.4756,
      "step": 1407
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3334462919529313,
      "learning_rate": 1.945383938789477e-05,
      "loss": 0.5205,
      "step": 1408
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2623364331937654,
      "learning_rate": 1.945284513046829e-05,
      "loss": 0.4487,
      "step": 1409
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.224613865675036,
      "learning_rate": 1.945184999432166e-05,
      "loss": 0.4873,
      "step": 1410
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.288573352678353,
      "learning_rate": 1.9450853979547384e-05,
      "loss": 0.4673,
      "step": 1411
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.331118091549076,
      "learning_rate": 1.9449857086238058e-05,
      "loss": 0.4336,
      "step": 1412
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3445470830590738,
      "learning_rate": 1.9448859314486342e-05,
      "loss": 0.4878,
      "step": 1413
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4711681696503103,
      "learning_rate": 1.9447860664384998e-05,
      "loss": 0.4849,
      "step": 1414
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3858681189400182,
      "learning_rate": 1.9446861136026846e-05,
      "loss": 0.5044,
      "step": 1415
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.322042390837048,
      "learning_rate": 1.9445860729504812e-05,
      "loss": 0.5425,
      "step": 1416
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3031813599061213,
      "learning_rate": 1.9444859444911884e-05,
      "loss": 0.5195,
      "step": 1417
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3417499083101903,
      "learning_rate": 1.9443857282341144e-05,
      "loss": 0.5059,
      "step": 1418
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4401369604889214,
      "learning_rate": 1.944285424188575e-05,
      "loss": 0.585,
      "step": 1419
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3127650643640183,
      "learning_rate": 1.9441850323638944e-05,
      "loss": 0.5312,
      "step": 1420
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4070531576750571,
      "learning_rate": 1.9440845527694047e-05,
      "loss": 0.4253,
      "step": 1421
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3323472519303963,
      "learning_rate": 1.9439839854144463e-05,
      "loss": 0.3843,
      "step": 1422
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3265006261203267,
      "learning_rate": 1.9438833303083677e-05,
      "loss": 0.5176,
      "step": 1423
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3141120160111324,
      "learning_rate": 1.943782587460526e-05,
      "loss": 0.4419,
      "step": 1424
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5647010090809466,
      "learning_rate": 1.9436817568802854e-05,
      "loss": 0.6094,
      "step": 1425
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3405533216379137,
      "learning_rate": 1.94358083857702e-05,
      "loss": 0.5005,
      "step": 1426
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2607219823446694,
      "learning_rate": 1.9434798325601098e-05,
      "loss": 0.4321,
      "step": 1427
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4437159467108913,
      "learning_rate": 1.9433787388389453e-05,
      "loss": 0.4995,
      "step": 1428
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4315248593860213,
      "learning_rate": 1.943277557422923e-05,
      "loss": 0.4648,
      "step": 1429
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5346410200516805,
      "learning_rate": 1.943176288321449e-05,
      "loss": 0.5151,
      "step": 1430
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.8825010864117366,
      "learning_rate": 1.943074931543937e-05,
      "loss": 0.6069,
      "step": 1431
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3571946997682727,
      "learning_rate": 1.942973487099809e-05,
      "loss": 0.4604,
      "step": 1432
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.488005045934584,
      "learning_rate": 1.9428719549984955e-05,
      "loss": 0.5366,
      "step": 1433
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6390189753149327,
      "learning_rate": 1.9427703352494335e-05,
      "loss": 0.583,
      "step": 1434
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6172085183306266,
      "learning_rate": 1.942668627862071e-05,
      "loss": 0.6104,
      "step": 1435
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.565282517618452,
      "learning_rate": 1.9425668328458616e-05,
      "loss": 0.5854,
      "step": 1436
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3220963457597608,
      "learning_rate": 1.942464950210268e-05,
      "loss": 0.4976,
      "step": 1437
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4480326782043156,
      "learning_rate": 1.9423629799647618e-05,
      "loss": 0.5361,
      "step": 1438
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6342763042592743,
      "learning_rate": 1.9422609221188208e-05,
      "loss": 0.5347,
      "step": 1439
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5074541050137635,
      "learning_rate": 1.942158776681933e-05,
      "loss": 0.5264,
      "step": 1440
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2088565362710477,
      "learning_rate": 1.942056543663593e-05,
      "loss": 0.3706,
      "step": 1441
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.1848326434901424,
      "learning_rate": 1.941954223073305e-05,
      "loss": 0.3979,
      "step": 1442
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2729136637466234,
      "learning_rate": 1.94185181492058e-05,
      "loss": 0.4458,
      "step": 1443
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.701004557940086,
      "learning_rate": 1.9417493192149376e-05,
      "loss": 0.5991,
      "step": 1444
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3314225755313136,
      "learning_rate": 1.941646735965906e-05,
      "loss": 0.4126,
      "step": 1445
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5211220071880724,
      "learning_rate": 1.941544065183021e-05,
      "loss": 0.5596,
      "step": 1446
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3080274316773102,
      "learning_rate": 1.9414413068758266e-05,
      "loss": 0.4287,
      "step": 1447
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5021761328296055,
      "learning_rate": 1.9413384610538752e-05,
      "loss": 0.5967,
      "step": 1448
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5050961064593789,
      "learning_rate": 1.941235527726727e-05,
      "loss": 0.5405,
      "step": 1449
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5019372320968516,
      "learning_rate": 1.941132506903951e-05,
      "loss": 0.5732,
      "step": 1450
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2658021990536998,
      "learning_rate": 1.9410293985951233e-05,
      "loss": 0.4355,
      "step": 1451
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3735954261418017,
      "learning_rate": 1.9409262028098285e-05,
      "loss": 0.5181,
      "step": 1452
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4715082563908313,
      "learning_rate": 1.94082291955766e-05,
      "loss": 0.5029,
      "step": 1453
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4817092487323225,
      "learning_rate": 1.9407195488482185e-05,
      "loss": 0.5278,
      "step": 1454
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3300822729632205,
      "learning_rate": 1.9406160906911137e-05,
      "loss": 0.4897,
      "step": 1455
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.346122741370758,
      "learning_rate": 1.9405125450959623e-05,
      "loss": 0.4526,
      "step": 1456
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.228696919094752,
      "learning_rate": 1.94040891207239e-05,
      "loss": 0.4863,
      "step": 1457
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.598585500666822,
      "learning_rate": 1.9403051916300296e-05,
      "loss": 0.6162,
      "step": 1458
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2746112826154703,
      "learning_rate": 1.9402013837785242e-05,
      "loss": 0.4844,
      "step": 1459
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4386287030265987,
      "learning_rate": 1.9400974885275226e-05,
      "loss": 0.5869,
      "step": 1460
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3475752850135974,
      "learning_rate": 1.939993505886683e-05,
      "loss": 0.4741,
      "step": 1461
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6392108925567226,
      "learning_rate": 1.9398894358656713e-05,
      "loss": 0.6133,
      "step": 1462
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3635830820362334,
      "learning_rate": 1.939785278474162e-05,
      "loss": 0.4624,
      "step": 1463
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4079972599055215,
      "learning_rate": 1.9396810337218373e-05,
      "loss": 0.6182,
      "step": 1464
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.43918278043144,
      "learning_rate": 1.939576701618387e-05,
      "loss": 0.6118,
      "step": 1465
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3752282022436453,
      "learning_rate": 1.9394722821735105e-05,
      "loss": 0.5591,
      "step": 1466
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.268798527088114,
      "learning_rate": 1.9393677753969137e-05,
      "loss": 0.5322,
      "step": 1467
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3887879884207404,
      "learning_rate": 1.939263181298312e-05,
      "loss": 0.459,
      "step": 1468
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2402700829919415,
      "learning_rate": 1.939158499887428e-05,
      "loss": 0.5151,
      "step": 1469
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2912546915868668,
      "learning_rate": 1.9390537311739927e-05,
      "loss": 0.5225,
      "step": 1470
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5364241185863337,
      "learning_rate": 1.938948875167745e-05,
      "loss": 0.665,
      "step": 1471
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1649791590981946,
      "learning_rate": 1.9388439318784328e-05,
      "loss": 0.4893,
      "step": 1472
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3368249651182127,
      "learning_rate": 1.9387389013158108e-05,
      "loss": 0.4717,
      "step": 1473
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1658682566136114,
      "learning_rate": 1.9386337834896428e-05,
      "loss": 0.3745,
      "step": 1474
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.286111375643965,
      "learning_rate": 1.9385285784097e-05,
      "loss": 0.5581,
      "step": 1475
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2760597593432768,
      "learning_rate": 1.9384232860857627e-05,
      "loss": 0.4185,
      "step": 1476
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2798675048465196,
      "learning_rate": 1.938317906527618e-05,
      "loss": 0.4351,
      "step": 1477
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3694420345435225,
      "learning_rate": 1.938212439745062e-05,
      "loss": 0.5366,
      "step": 1478
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4534107203583015,
      "learning_rate": 1.9381068857478994e-05,
      "loss": 0.5107,
      "step": 1479
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6049376635751578,
      "learning_rate": 1.938001244545941e-05,
      "loss": 0.5244,
      "step": 1480
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.431682188133012,
      "learning_rate": 1.9378955161490086e-05,
      "loss": 0.5273,
      "step": 1481
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5228603645027499,
      "learning_rate": 1.937789700566929e-05,
      "loss": 0.519,
      "step": 1482
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.289505893400877,
      "learning_rate": 1.93768379780954e-05,
      "loss": 0.4487,
      "step": 1483
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.335504934062169,
      "learning_rate": 1.937577807886685e-05,
      "loss": 0.4932,
      "step": 1484
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.680182632814276,
      "learning_rate": 1.9374717308082172e-05,
      "loss": 0.6123,
      "step": 1485
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4172758598635837,
      "learning_rate": 1.9373655665839973e-05,
      "loss": 0.4932,
      "step": 1486
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3925556345811374,
      "learning_rate": 1.937259315223894e-05,
      "loss": 0.4858,
      "step": 1487
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4914108444852507,
      "learning_rate": 1.937152976737785e-05,
      "loss": 0.4668,
      "step": 1488
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4632714576666066,
      "learning_rate": 1.937046551135554e-05,
      "loss": 0.436,
      "step": 1489
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4936982336018694,
      "learning_rate": 1.9369400384270948e-05,
      "loss": 0.5376,
      "step": 1490
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5616381498630005,
      "learning_rate": 1.9368334386223092e-05,
      "loss": 0.5195,
      "step": 1491
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4378627774802553,
      "learning_rate": 1.9367267517311057e-05,
      "loss": 0.481,
      "step": 1492
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2751865428892204,
      "learning_rate": 1.9366199777634018e-05,
      "loss": 0.3877,
      "step": 1493
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4259718551667826,
      "learning_rate": 1.9365131167291237e-05,
      "loss": 0.5005,
      "step": 1494
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.579249438912697,
      "learning_rate": 1.9364061686382042e-05,
      "loss": 0.5024,
      "step": 1495
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4585526685951027,
      "learning_rate": 1.9362991335005853e-05,
      "loss": 0.4751,
      "step": 1496
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4402871584129624,
      "learning_rate": 1.9361920113262172e-05,
      "loss": 0.5166,
      "step": 1497
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.298347937997259,
      "learning_rate": 1.9360848021250573e-05,
      "loss": 0.4556,
      "step": 1498
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.436289178915474,
      "learning_rate": 1.935977505907072e-05,
      "loss": 0.5146,
      "step": 1499
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5756417922212014,
      "learning_rate": 1.935870122682235e-05,
      "loss": 0.5923,
      "step": 1500
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3263007661178101,
      "learning_rate": 1.9357626524605286e-05,
      "loss": 0.4463,
      "step": 1501
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5219757506207505,
      "learning_rate": 1.935655095251943e-05,
      "loss": 0.4771,
      "step": 1502
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3851398869262774,
      "learning_rate": 1.9355474510664763e-05,
      "loss": 0.4614,
      "step": 1503
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3277043686469598,
      "learning_rate": 1.9354397199141356e-05,
      "loss": 0.4551,
      "step": 1504
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2636976839798384,
      "learning_rate": 1.9353319018049346e-05,
      "loss": 0.3784,
      "step": 1505
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.395855431688192,
      "learning_rate": 1.9352239967488965e-05,
      "loss": 0.5205,
      "step": 1506
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.409517277937351,
      "learning_rate": 1.9351160047560516e-05,
      "loss": 0.5098,
      "step": 1507
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2703949804084946,
      "learning_rate": 1.935007925836439e-05,
      "loss": 0.4507,
      "step": 1508
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4317491155698112,
      "learning_rate": 1.9348997600001052e-05,
      "loss": 0.5239,
      "step": 1509
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3713593189797215,
      "learning_rate": 1.934791507257105e-05,
      "loss": 0.4443,
      "step": 1510
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3933023184213866,
      "learning_rate": 1.934683167617502e-05,
      "loss": 0.436,
      "step": 1511
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4102466454882172,
      "learning_rate": 1.9345747410913666e-05,
      "loss": 0.5371,
      "step": 1512
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.531483930891419,
      "learning_rate": 1.9344662276887787e-05,
      "loss": 0.626,
      "step": 1513
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4453201888812905,
      "learning_rate": 1.9343576274198246e-05,
      "loss": 0.5205,
      "step": 1514
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3789058457808554,
      "learning_rate": 1.9342489402945997e-05,
      "loss": 0.5117,
      "step": 1515
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.291445743367134,
      "learning_rate": 1.9341401663232083e-05,
      "loss": 0.4487,
      "step": 1516
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3175739872633216,
      "learning_rate": 1.934031305515761e-05,
      "loss": 0.4683,
      "step": 1517
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4614421578183223,
      "learning_rate": 1.933922357882378e-05,
      "loss": 0.5063,
      "step": 1518
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5344873330908058,
      "learning_rate": 1.933813323433186e-05,
      "loss": 0.5601,
      "step": 1519
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3772181757449145,
      "learning_rate": 1.933704202178321e-05,
      "loss": 0.4756,
      "step": 1520
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.343693195027829,
      "learning_rate": 1.9335949941279267e-05,
      "loss": 0.4956,
      "step": 1521
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.311704894948002,
      "learning_rate": 1.9334856992921555e-05,
      "loss": 0.4805,
      "step": 1522
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3557550710774522,
      "learning_rate": 1.9333763176811663e-05,
      "loss": 0.4863,
      "step": 1523
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4939977599903318,
      "learning_rate": 1.933266849305128e-05,
      "loss": 0.5293,
      "step": 1524
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.209435434862085,
      "learning_rate": 1.9331572941742157e-05,
      "loss": 0.4067,
      "step": 1525
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.549312397625571,
      "learning_rate": 1.9330476522986136e-05,
      "loss": 0.6211,
      "step": 1526
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.398128739248394,
      "learning_rate": 1.9329379236885145e-05,
      "loss": 0.5259,
      "step": 1527
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3387186393113941,
      "learning_rate": 1.932828108354118e-05,
      "loss": 0.5254,
      "step": 1528
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4974529432429289,
      "learning_rate": 1.9327182063056325e-05,
      "loss": 0.5527,
      "step": 1529
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3904366736758427,
      "learning_rate": 1.9326082175532744e-05,
      "loss": 0.5356,
      "step": 1530
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4845002669506921,
      "learning_rate": 1.932498142107268e-05,
      "loss": 0.5444,
      "step": 1531
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2922469213234555,
      "learning_rate": 1.9323879799778452e-05,
      "loss": 0.4341,
      "step": 1532
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.395447646462076,
      "learning_rate": 1.9322777311752473e-05,
      "loss": 0.5522,
      "step": 1533
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4374015400966205,
      "learning_rate": 1.9321673957097226e-05,
      "loss": 0.5752,
      "step": 1534
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3990246193842566,
      "learning_rate": 1.9320569735915273e-05,
      "loss": 0.5127,
      "step": 1535
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2678703121795167,
      "learning_rate": 1.9319464648309265e-05,
      "loss": 0.4243,
      "step": 1536
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.288578625693972,
      "learning_rate": 1.9318358694381926e-05,
      "loss": 0.4424,
      "step": 1537
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2667549090131986,
      "learning_rate": 1.9317251874236066e-05,
      "loss": 0.3921,
      "step": 1538
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3477454032906435,
      "learning_rate": 1.931614418797457e-05,
      "loss": 0.4492,
      "step": 1539
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4385612429503434,
      "learning_rate": 1.9315035635700412e-05,
      "loss": 0.5176,
      "step": 1540
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.600101600921362,
      "learning_rate": 1.9313926217516637e-05,
      "loss": 0.5332,
      "step": 1541
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5187242773089265,
      "learning_rate": 1.9312815933526375e-05,
      "loss": 0.4922,
      "step": 1542
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.42837798309172,
      "learning_rate": 1.9311704783832835e-05,
      "loss": 0.519,
      "step": 1543
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4011375425550259,
      "learning_rate": 1.9310592768539315e-05,
      "loss": 0.5078,
      "step": 1544
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.521977107312,
      "learning_rate": 1.9309479887749175e-05,
      "loss": 0.5747,
      "step": 1545
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2545933558596956,
      "learning_rate": 1.930836614156587e-05,
      "loss": 0.4131,
      "step": 1546
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.410494477202916,
      "learning_rate": 1.9307251530092937e-05,
      "loss": 0.5366,
      "step": 1547
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3025965909798534,
      "learning_rate": 1.930613605343398e-05,
      "loss": 0.4834,
      "step": 1548
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5532111623585725,
      "learning_rate": 1.93050197116927e-05,
      "loss": 0.4775,
      "step": 1549
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2695967970244835,
      "learning_rate": 1.9303902504972866e-05,
      "loss": 0.5093,
      "step": 1550
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3892791611327686,
      "learning_rate": 1.9302784433378333e-05,
      "loss": 0.4907,
      "step": 1551
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.425534899354286,
      "learning_rate": 1.9301665497013034e-05,
      "loss": 0.5083,
      "step": 1552
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.316208958047847,
      "learning_rate": 1.9300545695980985e-05,
      "loss": 0.4839,
      "step": 1553
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.482307816254752,
      "learning_rate": 1.929942503038628e-05,
      "loss": 0.4849,
      "step": 1554
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.461596448486347,
      "learning_rate": 1.9298303500333088e-05,
      "loss": 0.4951,
      "step": 1555
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.475009123960136,
      "learning_rate": 1.9297181105925675e-05,
      "loss": 0.4932,
      "step": 1556
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5005680298685045,
      "learning_rate": 1.929605784726837e-05,
      "loss": 0.5117,
      "step": 1557
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3050225486906364,
      "learning_rate": 1.9294933724465593e-05,
      "loss": 0.4707,
      "step": 1558
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3994684285759689,
      "learning_rate": 1.9293808737621837e-05,
      "loss": 0.4404,
      "step": 1559
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4605119250553165,
      "learning_rate": 1.9292682886841683e-05,
      "loss": 0.5586,
      "step": 1560
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3531718340613823,
      "learning_rate": 1.9291556172229784e-05,
      "loss": 0.5298,
      "step": 1561
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.223337046322433,
      "learning_rate": 1.929042859389088e-05,
      "loss": 0.4072,
      "step": 1562
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4749999216538503,
      "learning_rate": 1.9289300151929784e-05,
      "loss": 0.5049,
      "step": 1563
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4991801844437909,
      "learning_rate": 1.9288170846451402e-05,
      "loss": 0.5239,
      "step": 1564
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4184154420925157,
      "learning_rate": 1.928704067756071e-05,
      "loss": 0.5498,
      "step": 1565
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.331978284560686,
      "learning_rate": 1.928590964536276e-05,
      "loss": 0.5054,
      "step": 1566
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2954868177913135,
      "learning_rate": 1.9284777749962696e-05,
      "loss": 0.4888,
      "step": 1567
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.326340166111839,
      "learning_rate": 1.928364499146574e-05,
      "loss": 0.4678,
      "step": 1568
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2708992738912093,
      "learning_rate": 1.9282511369977185e-05,
      "loss": 0.4497,
      "step": 1569
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3776623906151362,
      "learning_rate": 1.9281376885602412e-05,
      "loss": 0.5605,
      "step": 1570
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4720846062478692,
      "learning_rate": 1.9280241538446885e-05,
      "loss": 0.5498,
      "step": 1571
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3433654524882146,
      "learning_rate": 1.927910532861614e-05,
      "loss": 0.4106,
      "step": 1572
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3884944906757517,
      "learning_rate": 1.9277968256215794e-05,
      "loss": 0.5513,
      "step": 1573
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3513216501681815,
      "learning_rate": 1.9276830321351558e-05,
      "loss": 0.5039,
      "step": 1574
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4378456574315905,
      "learning_rate": 1.9275691524129203e-05,
      "loss": 0.4771,
      "step": 1575
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4143107423934906,
      "learning_rate": 1.9274551864654593e-05,
      "loss": 0.4512,
      "step": 1576
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3076257964685267,
      "learning_rate": 1.9273411343033667e-05,
      "loss": 0.499,
      "step": 1577
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2659584047792705,
      "learning_rate": 1.9272269959372444e-05,
      "loss": 0.4512,
      "step": 1578
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.1988271619448188,
      "learning_rate": 1.9271127713777033e-05,
      "loss": 0.4619,
      "step": 1579
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5218959811420465,
      "learning_rate": 1.926998460635361e-05,
      "loss": 0.5806,
      "step": 1580
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2491949465703451,
      "learning_rate": 1.9268840637208436e-05,
      "loss": 0.4585,
      "step": 1581
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3315699163103158,
      "learning_rate": 1.926769580644785e-05,
      "loss": 0.4204,
      "step": 1582
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3095984298991012,
      "learning_rate": 1.926655011417828e-05,
      "loss": 0.4561,
      "step": 1583
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.338947246593069,
      "learning_rate": 1.9265403560506223e-05,
      "loss": 0.4644,
      "step": 1584
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.1099063080038118,
      "learning_rate": 1.9264256145538262e-05,
      "loss": 0.3657,
      "step": 1585
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.426834158037802,
      "learning_rate": 1.926310786938106e-05,
      "loss": 0.499,
      "step": 1586
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2509420208367592,
      "learning_rate": 1.9261958732141352e-05,
      "loss": 0.4263,
      "step": 1587
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4377316085504073,
      "learning_rate": 1.926080873392597e-05,
      "loss": 0.5396,
      "step": 1588
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2916599544290488,
      "learning_rate": 1.925965787484181e-05,
      "loss": 0.4775,
      "step": 1589
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.16776539841199,
      "learning_rate": 1.9258506154995854e-05,
      "loss": 0.343,
      "step": 1590
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2991467620241395,
      "learning_rate": 1.9257353574495164e-05,
      "loss": 0.4863,
      "step": 1591
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4087488984525423,
      "learning_rate": 1.925620013344688e-05,
      "loss": 0.5508,
      "step": 1592
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.450687540282814,
      "learning_rate": 1.925504583195823e-05,
      "loss": 0.5312,
      "step": 1593
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2553286535013224,
      "learning_rate": 1.925389067013651e-05,
      "loss": 0.4077,
      "step": 1594
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2601256810642758,
      "learning_rate": 1.925273464808911e-05,
      "loss": 0.4185,
      "step": 1595
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3609911626690527,
      "learning_rate": 1.925157776592348e-05,
      "loss": 0.4688,
      "step": 1596
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2501701422040956,
      "learning_rate": 1.925042002374717e-05,
      "loss": 0.3589,
      "step": 1597
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2844904731716575,
      "learning_rate": 1.9249261421667796e-05,
      "loss": 0.4648,
      "step": 1598
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.313899943720688,
      "learning_rate": 1.9248101959793066e-05,
      "loss": 0.4456,
      "step": 1599
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.383780296981636,
      "learning_rate": 1.924694163823076e-05,
      "loss": 0.437,
      "step": 1600
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2160407288719073,
      "learning_rate": 1.9245780457088736e-05,
      "loss": 0.4087,
      "step": 1601
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3008630067437985,
      "learning_rate": 1.9244618416474938e-05,
      "loss": 0.4302,
      "step": 1602
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4111891977860216,
      "learning_rate": 1.9243455516497388e-05,
      "loss": 0.481,
      "step": 1603
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3699226586789695,
      "learning_rate": 1.924229175726418e-05,
      "loss": 0.4688,
      "step": 1604
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4503184992869769,
      "learning_rate": 1.9241127138883508e-05,
      "loss": 0.5215,
      "step": 1605
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5007827145877224,
      "learning_rate": 1.9239961661463623e-05,
      "loss": 0.5474,
      "step": 1606
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4040055869952697,
      "learning_rate": 1.9238795325112867e-05,
      "loss": 0.5327,
      "step": 1607
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3940753187211865,
      "learning_rate": 1.9237628129939665e-05,
      "loss": 0.4307,
      "step": 1608
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6319870455929395,
      "learning_rate": 1.9236460076052515e-05,
      "loss": 0.5894,
      "step": 1609
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2286443264646345,
      "learning_rate": 1.9235291163559996e-05,
      "loss": 0.4238,
      "step": 1610
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6410407763944672,
      "learning_rate": 1.923412139257077e-05,
      "loss": 0.6543,
      "step": 1611
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4671615863778469,
      "learning_rate": 1.9232950763193576e-05,
      "loss": 0.6182,
      "step": 1612
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3204390295190824,
      "learning_rate": 1.9231779275537233e-05,
      "loss": 0.4282,
      "step": 1613
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.464176608427387,
      "learning_rate": 1.923060692971064e-05,
      "loss": 0.5381,
      "step": 1614
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3143298704662114,
      "learning_rate": 1.9229433725822776e-05,
      "loss": 0.4561,
      "step": 1615
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5631419553927788,
      "learning_rate": 1.9228259663982705e-05,
      "loss": 0.5684,
      "step": 1616
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2309330043555384,
      "learning_rate": 1.922708474429956e-05,
      "loss": 0.4697,
      "step": 1617
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4345813711721422,
      "learning_rate": 1.9225908966882563e-05,
      "loss": 0.5649,
      "step": 1618
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.1921838186674425,
      "learning_rate": 1.922473233184101e-05,
      "loss": 0.3999,
      "step": 1619
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.380999031003653,
      "learning_rate": 1.922355483928428e-05,
      "loss": 0.5732,
      "step": 1620
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.259045521526527,
      "learning_rate": 1.922237648932183e-05,
      "loss": 0.4746,
      "step": 1621
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2893393087182108,
      "learning_rate": 1.92211972820632e-05,
      "loss": 0.4053,
      "step": 1622
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.290075898562747,
      "learning_rate": 1.9220017217618006e-05,
      "loss": 0.478,
      "step": 1623
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3272193615444934,
      "learning_rate": 1.921883629609594e-05,
      "loss": 0.5894,
      "step": 1624
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4012306886829706,
      "learning_rate": 1.9217654517606786e-05,
      "loss": 0.478,
      "step": 1625
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5085767916750623,
      "learning_rate": 1.921647188226039e-05,
      "loss": 0.5537,
      "step": 1626
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5041016445842972,
      "learning_rate": 1.92152883901667e-05,
      "loss": 0.5039,
      "step": 1627
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3411840203666847,
      "learning_rate": 1.921410404143572e-05,
      "loss": 0.4907,
      "step": 1628
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5612923086900239,
      "learning_rate": 1.9212918836177555e-05,
      "loss": 0.6216,
      "step": 1629
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3342125084878977,
      "learning_rate": 1.9211732774502372e-05,
      "loss": 0.4346,
      "step": 1630
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4546717938617213,
      "learning_rate": 1.921054585652043e-05,
      "loss": 0.4795,
      "step": 1631
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3647348698282926,
      "learning_rate": 1.920935808234206e-05,
      "loss": 0.564,
      "step": 1632
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2351576644624442,
      "learning_rate": 1.9208169452077678e-05,
      "loss": 0.4272,
      "step": 1633
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3780956722477973,
      "learning_rate": 1.9206979965837775e-05,
      "loss": 0.4561,
      "step": 1634
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.479410962542304,
      "learning_rate": 1.9205789623732923e-05,
      "loss": 0.5767,
      "step": 1635
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3435917610314811,
      "learning_rate": 1.9204598425873773e-05,
      "loss": 0.4971,
      "step": 1636
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4071125158973306,
      "learning_rate": 1.9203406372371065e-05,
      "loss": 0.5171,
      "step": 1637
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4250391158429547,
      "learning_rate": 1.92022134633356e-05,
      "loss": 0.5225,
      "step": 1638
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.295651426180104,
      "learning_rate": 1.9201019698878272e-05,
      "loss": 0.3955,
      "step": 1639
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.474334388517874,
      "learning_rate": 1.919982507911006e-05,
      "loss": 0.5132,
      "step": 1640
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2976758793225156,
      "learning_rate": 1.9198629604141996e-05,
      "loss": 0.4976,
      "step": 1641
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5146521743643582,
      "learning_rate": 1.9197433274085225e-05,
      "loss": 0.4917,
      "step": 1642
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.241672157709371,
      "learning_rate": 1.919623608905095e-05,
      "loss": 0.4521,
      "step": 1643
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.1894724112337551,
      "learning_rate": 1.9195038049150455e-05,
      "loss": 0.4595,
      "step": 1644
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4129785409000983,
      "learning_rate": 1.919383915449512e-05,
      "loss": 0.5898,
      "step": 1645
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2011897861521441,
      "learning_rate": 1.9192639405196377e-05,
      "loss": 0.4565,
      "step": 1646
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3996320603915107,
      "learning_rate": 1.9191438801365763e-05,
      "loss": 0.5449,
      "step": 1647
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3283854383449878,
      "learning_rate": 1.919023734311488e-05,
      "loss": 0.5161,
      "step": 1648
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4123787046733667,
      "learning_rate": 1.918903503055541e-05,
      "loss": 0.5269,
      "step": 1649
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3653428473776241,
      "learning_rate": 1.918783186379913e-05,
      "loss": 0.4927,
      "step": 1650
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3766320931590617,
      "learning_rate": 1.9186627842957873e-05,
      "loss": 0.4644,
      "step": 1651
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6623228596703783,
      "learning_rate": 1.9185422968143566e-05,
      "loss": 0.6289,
      "step": 1652
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3643896329460552,
      "learning_rate": 1.9184217239468213e-05,
      "loss": 0.4575,
      "step": 1653
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2309536977952862,
      "learning_rate": 1.9183010657043894e-05,
      "loss": 0.4058,
      "step": 1654
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.371711827240924,
      "learning_rate": 1.9181803220982776e-05,
      "loss": 0.4985,
      "step": 1655
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5917257288072661,
      "learning_rate": 1.9180594931397094e-05,
      "loss": 0.584,
      "step": 1656
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3210070929812157,
      "learning_rate": 1.9179385788399176e-05,
      "loss": 0.478,
      "step": 1657
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3450855940782993,
      "learning_rate": 1.917817579210141e-05,
      "loss": 0.5171,
      "step": 1658
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4502840570562603,
      "learning_rate": 1.9176964942616286e-05,
      "loss": 0.5967,
      "step": 1659
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5281917221763675,
      "learning_rate": 1.917575324005636e-05,
      "loss": 0.5942,
      "step": 1660
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5311135437363903,
      "learning_rate": 1.9174540684534267e-05,
      "loss": 0.5996,
      "step": 1661
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2824012117542651,
      "learning_rate": 1.9173327276162724e-05,
      "loss": 0.437,
      "step": 1662
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3183076644775358,
      "learning_rate": 1.917211301505453e-05,
      "loss": 0.5166,
      "step": 1663
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2688298722699158,
      "learning_rate": 1.9170897901322563e-05,
      "loss": 0.4634,
      "step": 1664
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3606454229944422,
      "learning_rate": 1.916968193507977e-05,
      "loss": 0.4185,
      "step": 1665
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3433620154388475,
      "learning_rate": 1.9168465116439196e-05,
      "loss": 0.4766,
      "step": 1666
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.117566295757557,
      "learning_rate": 1.916724744551394e-05,
      "loss": 0.3838,
      "step": 1667
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2027825313117038,
      "learning_rate": 1.9166028922417208e-05,
      "loss": 0.4229,
      "step": 1668
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3789490518992984,
      "learning_rate": 1.9164809547262262e-05,
      "loss": 0.6006,
      "step": 1669
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3457145867635452,
      "learning_rate": 1.916358932016246e-05,
      "loss": 0.4858,
      "step": 1670
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3528788198141088,
      "learning_rate": 1.916236824123123e-05,
      "loss": 0.4912,
      "step": 1671
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3373879508367132,
      "learning_rate": 1.916114631058208e-05,
      "loss": 0.4414,
      "step": 1672
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4473379522172565,
      "learning_rate": 1.91599235283286e-05,
      "loss": 0.543,
      "step": 1673
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6064135328482532,
      "learning_rate": 1.9158699894584456e-05,
      "loss": 0.5381,
      "step": 1674
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.238057041358882,
      "learning_rate": 1.9157475409463396e-05,
      "loss": 0.4519,
      "step": 1675
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2230038216894874,
      "learning_rate": 1.915625007307925e-05,
      "loss": 0.4316,
      "step": 1676
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2527880091710049,
      "learning_rate": 1.9155023885545914e-05,
      "loss": 0.4785,
      "step": 1677
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1511875581888071,
      "learning_rate": 1.915379684697738e-05,
      "loss": 0.4692,
      "step": 1678
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2858549760419637,
      "learning_rate": 1.915256895748771e-05,
      "loss": 0.4546,
      "step": 1679
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5186524351224586,
      "learning_rate": 1.9151340217191042e-05,
      "loss": 0.6631,
      "step": 1680
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3183732776329582,
      "learning_rate": 1.9150110626201604e-05,
      "loss": 0.4233,
      "step": 1681
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4546657085889103,
      "learning_rate": 1.9148880184633695e-05,
      "loss": 0.542,
      "step": 1682
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.366064151110794,
      "learning_rate": 1.914764889260169e-05,
      "loss": 0.4814,
      "step": 1683
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4657495896396773,
      "learning_rate": 1.914641675022005e-05,
      "loss": 0.5039,
      "step": 1684
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3893783933861021,
      "learning_rate": 1.914518375760332e-05,
      "loss": 0.4561,
      "step": 1685
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4244259859127197,
      "learning_rate": 1.9143949914866106e-05,
      "loss": 0.4463,
      "step": 1686
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.347439486418999,
      "learning_rate": 1.914271522212311e-05,
      "loss": 0.4961,
      "step": 1687
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2870314210300506,
      "learning_rate": 1.9141479679489107e-05,
      "loss": 0.4849,
      "step": 1688
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3010930711166941,
      "learning_rate": 1.914024328707895e-05,
      "loss": 0.4751,
      "step": 1689
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3433206535268085,
      "learning_rate": 1.9139006045007567e-05,
      "loss": 0.4873,
      "step": 1690
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.30018537013275,
      "learning_rate": 1.913776795338998e-05,
      "loss": 0.4146,
      "step": 1691
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2345092111152667,
      "learning_rate": 1.9136529012341268e-05,
      "loss": 0.3877,
      "step": 1692
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.534659059008195,
      "learning_rate": 1.9135289221976608e-05,
      "loss": 0.5679,
      "step": 1693
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.497030461295887,
      "learning_rate": 1.9134048582411253e-05,
      "loss": 0.5405,
      "step": 1694
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4113577038431948,
      "learning_rate": 1.9132807093760523e-05,
      "loss": 0.5054,
      "step": 1695
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.389876670849188,
      "learning_rate": 1.9131564756139824e-05,
      "loss": 0.4922,
      "step": 1696
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.452092219869638,
      "learning_rate": 1.9130321569664646e-05,
      "loss": 0.4795,
      "step": 1697
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3846796144696758,
      "learning_rate": 1.9129077534450556e-05,
      "loss": 0.4556,
      "step": 1698
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4360348335080464,
      "learning_rate": 1.912783265061319e-05,
      "loss": 0.5498,
      "step": 1699
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1677086101182694,
      "learning_rate": 1.9126586918268275e-05,
      "loss": 0.4492,
      "step": 1700
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2694967467328062,
      "learning_rate": 1.9125340337531612e-05,
      "loss": 0.4854,
      "step": 1701
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2428692372419785,
      "learning_rate": 1.912409290851908e-05,
      "loss": 0.4268,
      "step": 1702
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3647094515901101,
      "learning_rate": 1.9122844631346632e-05,
      "loss": 0.5034,
      "step": 1703
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.321716249636648,
      "learning_rate": 1.9121595506130317e-05,
      "loss": 0.4224,
      "step": 1704
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2829942093758973,
      "learning_rate": 1.9120345532986243e-05,
      "loss": 0.5181,
      "step": 1705
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3616061863610014,
      "learning_rate": 1.911909471203061e-05,
      "loss": 0.5303,
      "step": 1706
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3605761076285985,
      "learning_rate": 1.911784304337969e-05,
      "loss": 0.5015,
      "step": 1707
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5028296578145492,
      "learning_rate": 1.911659052714984e-05,
      "loss": 0.5752,
      "step": 1708
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3256322336964317,
      "learning_rate": 1.911533716345748e-05,
      "loss": 0.4614,
      "step": 1709
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4742658766009122,
      "learning_rate": 1.9114082952419134e-05,
      "loss": 0.5469,
      "step": 1710
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.452751674811923,
      "learning_rate": 1.9112827894151386e-05,
      "loss": 0.4897,
      "step": 1711
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3015372076545233,
      "learning_rate": 1.9111571988770903e-05,
      "loss": 0.4087,
      "step": 1712
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.344164951271242,
      "learning_rate": 1.9110315236394434e-05,
      "loss": 0.439,
      "step": 1713
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2566634855619665,
      "learning_rate": 1.9109057637138805e-05,
      "loss": 0.4512,
      "step": 1714
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3850814056055665,
      "learning_rate": 1.9107799191120913e-05,
      "loss": 0.5444,
      "step": 1715
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3297930505970632,
      "learning_rate": 1.910653989845775e-05,
      "loss": 0.4663,
      "step": 1716
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.449573472066402,
      "learning_rate": 1.9105279759266376e-05,
      "loss": 0.5601,
      "step": 1717
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4420283562474898,
      "learning_rate": 1.9104018773663924e-05,
      "loss": 0.6133,
      "step": 1718
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3192384909220056,
      "learning_rate": 1.9102756941767625e-05,
      "loss": 0.4595,
      "step": 1719
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6543203562967128,
      "learning_rate": 1.9101494263694764e-05,
      "loss": 0.5405,
      "step": 1720
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.316396587877852,
      "learning_rate": 1.9100230739562728e-05,
      "loss": 0.4204,
      "step": 1721
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.571817249700429,
      "learning_rate": 1.9098966369488967e-05,
      "loss": 0.5322,
      "step": 1722
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3609476415664803,
      "learning_rate": 1.9097701153591015e-05,
      "loss": 0.5039,
      "step": 1723
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2259005451976086,
      "learning_rate": 1.9096435091986485e-05,
      "loss": 0.4429,
      "step": 1724
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3407346173915504,
      "learning_rate": 1.9095168184793067e-05,
      "loss": 0.5288,
      "step": 1725
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3443169789949942,
      "learning_rate": 1.9093900432128532e-05,
      "loss": 0.4834,
      "step": 1726
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3550388617359101,
      "learning_rate": 1.9092631834110723e-05,
      "loss": 0.5078,
      "step": 1727
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4042067498911706,
      "learning_rate": 1.9091362390857578e-05,
      "loss": 0.4844,
      "step": 1728
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3020897931635647,
      "learning_rate": 1.909009210248709e-05,
      "loss": 0.5229,
      "step": 1729
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3663538152975119,
      "learning_rate": 1.9088820969117348e-05,
      "loss": 0.4448,
      "step": 1730
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2048814380216197,
      "learning_rate": 1.9087548990866515e-05,
      "loss": 0.4082,
      "step": 1731
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3699340316730577,
      "learning_rate": 1.9086276167852834e-05,
      "loss": 0.5366,
      "step": 1732
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5581843306440477,
      "learning_rate": 1.908500250019462e-05,
      "loss": 0.5776,
      "step": 1733
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3430321049346448,
      "learning_rate": 1.908372798801027e-05,
      "loss": 0.5254,
      "step": 1734
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2718413033829328,
      "learning_rate": 1.9082452631418265e-05,
      "loss": 0.4937,
      "step": 1735
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4091610555936633,
      "learning_rate": 1.9081176430537158e-05,
      "loss": 0.5254,
      "step": 1736
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3664807553534362,
      "learning_rate": 1.9079899385485583e-05,
      "loss": 0.5718,
      "step": 1737
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2206027892992777,
      "learning_rate": 1.907862149638225e-05,
      "loss": 0.4619,
      "step": 1738
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3921034179791556,
      "learning_rate": 1.907734276334595e-05,
      "loss": 0.4707,
      "step": 1739
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3875741707535443,
      "learning_rate": 1.907606318649555e-05,
      "loss": 0.5107,
      "step": 1740
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3778494530738254,
      "learning_rate": 1.9074782765950007e-05,
      "loss": 0.4365,
      "step": 1741
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5393681251163105,
      "learning_rate": 1.9073501501828335e-05,
      "loss": 0.5273,
      "step": 1742
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.400026232704665,
      "learning_rate": 1.9072219394249644e-05,
      "loss": 0.5,
      "step": 1743
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4855419083883028,
      "learning_rate": 1.9070936443333113e-05,
      "loss": 0.4956,
      "step": 1744
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3869523890052753,
      "learning_rate": 1.9069652649198004e-05,
      "loss": 0.4932,
      "step": 1745
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1592408155289877,
      "learning_rate": 1.906836801196366e-05,
      "loss": 0.4541,
      "step": 1746
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6056067561455853,
      "learning_rate": 1.9067082531749496e-05,
      "loss": 0.5757,
      "step": 1747
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.341751599109826,
      "learning_rate": 1.9065796208675005e-05,
      "loss": 0.4814,
      "step": 1748
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5872669978577292,
      "learning_rate": 1.9064509042859767e-05,
      "loss": 0.5303,
      "step": 1749
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4971932961210355,
      "learning_rate": 1.906322103442343e-05,
      "loss": 0.5605,
      "step": 1750
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3851212404178812,
      "learning_rate": 1.9061932183485726e-05,
      "loss": 0.5039,
      "step": 1751
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2711147304238455,
      "learning_rate": 1.906064249016647e-05,
      "loss": 0.459,
      "step": 1752
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3188086065117302,
      "learning_rate": 1.905935195458554e-05,
      "loss": 0.4976,
      "step": 1753
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2379764646538622,
      "learning_rate": 1.9058060576862912e-05,
      "loss": 0.4927,
      "step": 1754
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.419342412103798,
      "learning_rate": 1.905676835711862e-05,
      "loss": 0.5132,
      "step": 1755
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4576360430688815,
      "learning_rate": 1.9055475295472792e-05,
      "loss": 0.5981,
      "step": 1756
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.471440964986542,
      "learning_rate": 1.905418139204563e-05,
      "loss": 0.4883,
      "step": 1757
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1190019305514216,
      "learning_rate": 1.9052886646957413e-05,
      "loss": 0.377,
      "step": 1758
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.384532992585479,
      "learning_rate": 1.9051591060328496e-05,
      "loss": 0.4907,
      "step": 1759
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.327872715093799,
      "learning_rate": 1.9050294632279317e-05,
      "loss": 0.5044,
      "step": 1760
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3120691063540797,
      "learning_rate": 1.9048997362930384e-05,
      "loss": 0.4736,
      "step": 1761
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2983638813687473,
      "learning_rate": 1.9047699252402294e-05,
      "loss": 0.4092,
      "step": 1762
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4236106051966777,
      "learning_rate": 1.9046400300815716e-05,
      "loss": 0.4326,
      "step": 1763
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2835565218978733,
      "learning_rate": 1.90451005082914e-05,
      "loss": 0.4165,
      "step": 1764
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.223309099444229,
      "learning_rate": 1.904379987495017e-05,
      "loss": 0.5093,
      "step": 1765
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3740710252372172,
      "learning_rate": 1.904249840091293e-05,
      "loss": 0.5742,
      "step": 1766
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1860245098292197,
      "learning_rate": 1.9041196086300666e-05,
      "loss": 0.457,
      "step": 1767
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3937307588212662,
      "learning_rate": 1.9039892931234434e-05,
      "loss": 0.4863,
      "step": 1768
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5279812109386235,
      "learning_rate": 1.903858893583538e-05,
      "loss": 0.4629,
      "step": 1769
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5553220178777023,
      "learning_rate": 1.9037284100224714e-05,
      "loss": 0.5864,
      "step": 1770
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2933537055442985,
      "learning_rate": 1.9035978424523737e-05,
      "loss": 0.4878,
      "step": 1771
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3635342186390704,
      "learning_rate": 1.903467190885382e-05,
      "loss": 0.4624,
      "step": 1772
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3825462149377117,
      "learning_rate": 1.903336455333641e-05,
      "loss": 0.4312,
      "step": 1773
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.384144127304629,
      "learning_rate": 1.9032056358093048e-05,
      "loss": 0.4121,
      "step": 1774
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3591983069987252,
      "learning_rate": 1.903074732324533e-05,
      "loss": 0.478,
      "step": 1775
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3924363200859637,
      "learning_rate": 1.9029437448914945e-05,
      "loss": 0.5015,
      "step": 1776
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.548787998936849,
      "learning_rate": 1.902812673522366e-05,
      "loss": 0.5825,
      "step": 1777
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4096939025952473,
      "learning_rate": 1.9026815182293315e-05,
      "loss": 0.5024,
      "step": 1778
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3323329925221892,
      "learning_rate": 1.9025502790245824e-05,
      "loss": 0.4878,
      "step": 1779
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3120514581191334,
      "learning_rate": 1.9024189559203193e-05,
      "loss": 0.4546,
      "step": 1780
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3231473673228933,
      "learning_rate": 1.9022875489287496e-05,
      "loss": 0.457,
      "step": 1781
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4304031790138878,
      "learning_rate": 1.9021560580620883e-05,
      "loss": 0.5264,
      "step": 1782
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.295090516387524,
      "learning_rate": 1.902024483332559e-05,
      "loss": 0.4468,
      "step": 1783
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4640055393045466,
      "learning_rate": 1.9018928247523924e-05,
      "loss": 0.5151,
      "step": 1784
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3381707182404003,
      "learning_rate": 1.901761082333827e-05,
      "loss": 0.4658,
      "step": 1785
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3922469942231612,
      "learning_rate": 1.90162925608911e-05,
      "loss": 0.4897,
      "step": 1786
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4291252887698753,
      "learning_rate": 1.901497346030495e-05,
      "loss": 0.5278,
      "step": 1787
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4017878700876496,
      "learning_rate": 1.9013653521702448e-05,
      "loss": 0.4482,
      "step": 1788
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2668765012418657,
      "learning_rate": 1.901233274520629e-05,
      "loss": 0.4487,
      "step": 1789
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6104563436736237,
      "learning_rate": 1.9011011130939254e-05,
      "loss": 0.5249,
      "step": 1790
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.589917684808301,
      "learning_rate": 1.900968867902419e-05,
      "loss": 0.4858,
      "step": 1791
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3353851286108847,
      "learning_rate": 1.9008365389584042e-05,
      "loss": 0.4863,
      "step": 1792
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.211412737212925,
      "learning_rate": 1.900704126274181e-05,
      "loss": 0.3574,
      "step": 1793
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3414740514901287,
      "learning_rate": 1.9005716298620585e-05,
      "loss": 0.4702,
      "step": 1794
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.591438892263213,
      "learning_rate": 1.9004390497343536e-05,
      "loss": 0.5469,
      "step": 1795
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4321737678885746,
      "learning_rate": 1.9003063859033906e-05,
      "loss": 0.4282,
      "step": 1796
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5292904314935651,
      "learning_rate": 1.9001736383815023e-05,
      "loss": 0.5938,
      "step": 1797
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.448753920200164,
      "learning_rate": 1.900040807181027e-05,
      "loss": 0.542,
      "step": 1798
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1179128039282553,
      "learning_rate": 1.8999078923143142e-05,
      "loss": 0.3721,
      "step": 1799
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2660774189635953,
      "learning_rate": 1.8997748937937188e-05,
      "loss": 0.4214,
      "step": 1800
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.27862608304779,
      "learning_rate": 1.899641811631604e-05,
      "loss": 0.4385,
      "step": 1801
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3090400091297718,
      "learning_rate": 1.8995086458403408e-05,
      "loss": 0.458,
      "step": 1802
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4544945371791764,
      "learning_rate": 1.8993753964323086e-05,
      "loss": 0.5938,
      "step": 1803
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4268311078985674,
      "learning_rate": 1.8992420634198934e-05,
      "loss": 0.5791,
      "step": 1804
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4130722562338502,
      "learning_rate": 1.8991086468154897e-05,
      "loss": 0.5098,
      "step": 1805
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4439078873166302,
      "learning_rate": 1.8989751466315004e-05,
      "loss": 0.5298,
      "step": 1806
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4594381028736478,
      "learning_rate": 1.8988415628803345e-05,
      "loss": 0.4946,
      "step": 1807
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3358768550263984,
      "learning_rate": 1.8987078955744103e-05,
      "loss": 0.4458,
      "step": 1808
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5150622372257514,
      "learning_rate": 1.898574144726153e-05,
      "loss": 0.5415,
      "step": 1809
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2268521108030426,
      "learning_rate": 1.8984403103479957e-05,
      "loss": 0.4121,
      "step": 1810
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.293003666230569,
      "learning_rate": 1.89830639245238e-05,
      "loss": 0.4453,
      "step": 1811
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2847057068600816,
      "learning_rate": 1.8981723910517546e-05,
      "loss": 0.4868,
      "step": 1812
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3245987116152724,
      "learning_rate": 1.8980383061585752e-05,
      "loss": 0.5176,
      "step": 1813
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.372274978375831,
      "learning_rate": 1.8979041377853068e-05,
      "loss": 0.5908,
      "step": 1814
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4278083501797936,
      "learning_rate": 1.8977698859444217e-05,
      "loss": 0.4922,
      "step": 1815
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2403701740634436,
      "learning_rate": 1.8976355506483988e-05,
      "loss": 0.457,
      "step": 1816
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4815304210030804,
      "learning_rate": 1.8975011319097264e-05,
      "loss": 0.5713,
      "step": 1817
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4997819900577614,
      "learning_rate": 1.8973666297409e-05,
      "loss": 0.4863,
      "step": 1818
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.252663092178203,
      "learning_rate": 1.8972320441544224e-05,
      "loss": 0.3887,
      "step": 1819
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2667220844684757,
      "learning_rate": 1.897097375162804e-05,
      "loss": 0.4463,
      "step": 1820
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3513148040311687,
      "learning_rate": 1.896962622778564e-05,
      "loss": 0.52,
      "step": 1821
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5327672916373871,
      "learning_rate": 1.8968277870142283e-05,
      "loss": 0.5273,
      "step": 1822
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.285094857465307,
      "learning_rate": 1.8966928678823317e-05,
      "loss": 0.4224,
      "step": 1823
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4843674991159124,
      "learning_rate": 1.8965578653954152e-05,
      "loss": 0.6094,
      "step": 1824
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2388176227866996,
      "learning_rate": 1.896422779566029e-05,
      "loss": 0.4141,
      "step": 1825
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.49959563487157,
      "learning_rate": 1.8962876104067303e-05,
      "loss": 0.4922,
      "step": 1826
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2891590636171075,
      "learning_rate": 1.8961523579300836e-05,
      "loss": 0.4644,
      "step": 1827
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2996998674130613,
      "learning_rate": 1.896017022148663e-05,
      "loss": 0.4595,
      "step": 1828
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4632837414471014,
      "learning_rate": 1.895881603075048e-05,
      "loss": 0.5229,
      "step": 1829
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.560639405983952,
      "learning_rate": 1.8957461007218272e-05,
      "loss": 0.5815,
      "step": 1830
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.349019945879509,
      "learning_rate": 1.8956105151015966e-05,
      "loss": 0.4023,
      "step": 1831
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2996719594809036,
      "learning_rate": 1.8954748462269604e-05,
      "loss": 0.415,
      "step": 1832
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.276247099698208,
      "learning_rate": 1.89533909411053e-05,
      "loss": 0.4834,
      "step": 1833
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.447830321825076,
      "learning_rate": 1.895203258764924e-05,
      "loss": 0.4795,
      "step": 1834
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2637619222757357,
      "learning_rate": 1.8950673402027706e-05,
      "loss": 0.3882,
      "step": 1835
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6323499663977261,
      "learning_rate": 1.8949313384367038e-05,
      "loss": 0.5835,
      "step": 1836
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5209160160782305,
      "learning_rate": 1.8947952534793663e-05,
      "loss": 0.519,
      "step": 1837
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.0403579152075246,
      "learning_rate": 1.894659085343408e-05,
      "loss": 0.8242,
      "step": 1838
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3236733613621143,
      "learning_rate": 1.894522834041487e-05,
      "loss": 0.481,
      "step": 1839
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3674137754391713,
      "learning_rate": 1.8943864995862692e-05,
      "loss": 0.5249,
      "step": 1840
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4106182232547877,
      "learning_rate": 1.8942500819904285e-05,
      "loss": 0.5571,
      "step": 1841
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4381476591987619,
      "learning_rate": 1.8941135812666445e-05,
      "loss": 0.5605,
      "step": 1842
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5017574517981254,
      "learning_rate": 1.8939769974276076e-05,
      "loss": 0.4263,
      "step": 1843
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3456152086959932,
      "learning_rate": 1.8938403304860137e-05,
      "loss": 0.4634,
      "step": 1844
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.410884663511288,
      "learning_rate": 1.893703580454567e-05,
      "loss": 0.5107,
      "step": 1845
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2804124580898186,
      "learning_rate": 1.89356674734598e-05,
      "loss": 0.4238,
      "step": 1846
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5521598103716714,
      "learning_rate": 1.8934298311729728e-05,
      "loss": 0.5381,
      "step": 1847
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3951878818779306,
      "learning_rate": 1.8932928319482714e-05,
      "loss": 0.5234,
      "step": 1848
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4325220950832864,
      "learning_rate": 1.8931557496846124e-05,
      "loss": 0.4766,
      "step": 1849
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4650194199448525,
      "learning_rate": 1.8930185843947382e-05,
      "loss": 0.4893,
      "step": 1850
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3365164603819417,
      "learning_rate": 1.8928813360914e-05,
      "loss": 0.5342,
      "step": 1851
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3924882465640656,
      "learning_rate": 1.892744004787355e-05,
      "loss": 0.5645,
      "step": 1852
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3077526607913268,
      "learning_rate": 1.8926065904953703e-05,
      "loss": 0.5146,
      "step": 1853
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2072562936369702,
      "learning_rate": 1.8924690932282193e-05,
      "loss": 0.4473,
      "step": 1854
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3230414822664653,
      "learning_rate": 1.8923315129986838e-05,
      "loss": 0.5288,
      "step": 1855
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5071856996904043,
      "learning_rate": 1.8921938498195523e-05,
      "loss": 0.4683,
      "step": 1856
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2636968828202724,
      "learning_rate": 1.892056103703623e-05,
      "loss": 0.4795,
      "step": 1857
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3251537379174965,
      "learning_rate": 1.891918274663699e-05,
      "loss": 0.4551,
      "step": 1858
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3229651787020629,
      "learning_rate": 1.891780362712594e-05,
      "loss": 0.4951,
      "step": 1859
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4328108001825304,
      "learning_rate": 1.891642367863127e-05,
      "loss": 0.5366,
      "step": 1860
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4763010121329145,
      "learning_rate": 1.891504290128127e-05,
      "loss": 0.6299,
      "step": 1861
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2969206429633262,
      "learning_rate": 1.891366129520428e-05,
      "loss": 0.4966,
      "step": 1862
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3334001860344356,
      "learning_rate": 1.8912278860528742e-05,
      "loss": 0.4985,
      "step": 1863
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4221326992347272,
      "learning_rate": 1.891089559738316e-05,
      "loss": 0.5913,
      "step": 1864
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3428253031575235,
      "learning_rate": 1.8909511505896122e-05,
      "loss": 0.4854,
      "step": 1865
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2198607435411277,
      "learning_rate": 1.890812658619629e-05,
      "loss": 0.4204,
      "step": 1866
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3386590205074966,
      "learning_rate": 1.8906740838412404e-05,
      "loss": 0.4355,
      "step": 1867
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3325732562467194,
      "learning_rate": 1.890535426267328e-05,
      "loss": 0.562,
      "step": 1868
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4559805981859006,
      "learning_rate": 1.8903966859107816e-05,
      "loss": 0.5776,
      "step": 1869
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.281444251437626,
      "learning_rate": 1.8902578627844975e-05,
      "loss": 0.3564,
      "step": 1870
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4695316422803655,
      "learning_rate": 1.890118956901381e-05,
      "loss": 0.5698,
      "step": 1871
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3901661580979736,
      "learning_rate": 1.8899799682743442e-05,
      "loss": 0.4255,
      "step": 1872
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3899419459986238,
      "learning_rate": 1.8898408969163078e-05,
      "loss": 0.603,
      "step": 1873
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5915362002020685,
      "learning_rate": 1.889701742840199e-05,
      "loss": 0.5869,
      "step": 1874
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1836858712956415,
      "learning_rate": 1.8895625060589538e-05,
      "loss": 0.3984,
      "step": 1875
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6550415550247273,
      "learning_rate": 1.8894231865855152e-05,
      "loss": 0.6514,
      "step": 1876
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5644084889136898,
      "learning_rate": 1.8892837844328338e-05,
      "loss": 0.5498,
      "step": 1877
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6208936616334684,
      "learning_rate": 1.8891442996138686e-05,
      "loss": 0.5498,
      "step": 1878
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.58945358690966,
      "learning_rate": 1.8890047321415856e-05,
      "loss": 0.5781,
      "step": 1879
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2351883677218591,
      "learning_rate": 1.8888650820289594e-05,
      "loss": 0.4365,
      "step": 1880
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.311097311377065,
      "learning_rate": 1.8887253492889708e-05,
      "loss": 0.4448,
      "step": 1881
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4800838059708674,
      "learning_rate": 1.8885855339346097e-05,
      "loss": 0.582,
      "step": 1882
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.21054946430668,
      "learning_rate": 1.8884456359788725e-05,
      "loss": 0.4683,
      "step": 1883
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2369867877479095,
      "learning_rate": 1.8883056554347643e-05,
      "loss": 0.4126,
      "step": 1884
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4546608542915707,
      "learning_rate": 1.8881655923152975e-05,
      "loss": 0.5112,
      "step": 1885
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2219698431982604,
      "learning_rate": 1.888025446633492e-05,
      "loss": 0.4077,
      "step": 1886
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3161728979783731,
      "learning_rate": 1.8878852184023754e-05,
      "loss": 0.4614,
      "step": 1887
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1365685523673625,
      "learning_rate": 1.8877449076349833e-05,
      "loss": 0.4116,
      "step": 1888
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1520829235422292,
      "learning_rate": 1.8876045143443583e-05,
      "loss": 0.4521,
      "step": 1889
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.292756853785611,
      "learning_rate": 1.8874640385435515e-05,
      "loss": 0.4849,
      "step": 1890
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3241923859615223,
      "learning_rate": 1.8873234802456216e-05,
      "loss": 0.5063,
      "step": 1891
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2859009137439545,
      "learning_rate": 1.887182839463634e-05,
      "loss": 0.4038,
      "step": 1892
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3333537407947886,
      "learning_rate": 1.8870421162106628e-05,
      "loss": 0.4927,
      "step": 1893
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2960513684409156,
      "learning_rate": 1.8869013104997896e-05,
      "loss": 0.415,
      "step": 1894
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2982769279580884,
      "learning_rate": 1.8867604223441027e-05,
      "loss": 0.4419,
      "step": 1895
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4340768993616895,
      "learning_rate": 1.8866194517566993e-05,
      "loss": 0.542,
      "step": 1896
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3788024770150666,
      "learning_rate": 1.886478398750684e-05,
      "loss": 0.4595,
      "step": 1897
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4288407883989303,
      "learning_rate": 1.886337263339168e-05,
      "loss": 0.502,
      "step": 1898
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.298308123721369,
      "learning_rate": 1.8861960455352723e-05,
      "loss": 0.5137,
      "step": 1899
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4698340153294234,
      "learning_rate": 1.8860547453521232e-05,
      "loss": 0.5396,
      "step": 1900
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2692642861493124,
      "learning_rate": 1.8859133628028564e-05,
      "loss": 0.417,
      "step": 1901
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1967718995253538,
      "learning_rate": 1.8857718979006135e-05,
      "loss": 0.4453,
      "step": 1902
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4070697695189065,
      "learning_rate": 1.885630350658546e-05,
      "loss": 0.5317,
      "step": 1903
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.354394968518872,
      "learning_rate": 1.885488721089812e-05,
      "loss": 0.4946,
      "step": 1904
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4200304231033405,
      "learning_rate": 1.885347009207576e-05,
      "loss": 0.4932,
      "step": 1905
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4154706308525962,
      "learning_rate": 1.8852052150250123e-05,
      "loss": 0.4741,
      "step": 1906
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.333181876374129,
      "learning_rate": 1.885063338555301e-05,
      "loss": 0.4707,
      "step": 1907
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3981450676920255,
      "learning_rate": 1.8849213798116318e-05,
      "loss": 0.5347,
      "step": 1908
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3346031230237965,
      "learning_rate": 1.8847793388071997e-05,
      "loss": 0.415,
      "step": 1909
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3860622477599085,
      "learning_rate": 1.8846372155552095e-05,
      "loss": 0.5234,
      "step": 1910
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4359006639583898,
      "learning_rate": 1.884495010068872e-05,
      "loss": 0.563,
      "step": 1911
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.406947669772113,
      "learning_rate": 1.884352722361407e-05,
      "loss": 0.5464,
      "step": 1912
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.519606654605986,
      "learning_rate": 1.8842103524460414e-05,
      "loss": 0.6226,
      "step": 1913
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2632537515400695,
      "learning_rate": 1.8840679003360088e-05,
      "loss": 0.4185,
      "step": 1914
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3370858847640676,
      "learning_rate": 1.8839253660445523e-05,
      "loss": 0.4673,
      "step": 1915
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.353579573060149,
      "learning_rate": 1.883782749584921e-05,
      "loss": 0.4326,
      "step": 1916
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3908437628704888,
      "learning_rate": 1.8836400509703727e-05,
      "loss": 0.5088,
      "step": 1917
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2963433608790271,
      "learning_rate": 1.883497270214172e-05,
      "loss": 0.4155,
      "step": 1918
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.395932387653434,
      "learning_rate": 1.8833544073295918e-05,
      "loss": 0.5552,
      "step": 1919
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.207716917141677,
      "learning_rate": 1.8832114623299125e-05,
      "loss": 0.4072,
      "step": 1920
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.373721861517437,
      "learning_rate": 1.8830684352284217e-05,
      "loss": 0.4756,
      "step": 1921
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4726853373662716,
      "learning_rate": 1.882925326038415e-05,
      "loss": 0.6064,
      "step": 1922
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4558783954055512,
      "learning_rate": 1.8827821347731955e-05,
      "loss": 0.5542,
      "step": 1923
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.8166639510683014,
      "learning_rate": 1.8826388614460746e-05,
      "loss": 0.6631,
      "step": 1924
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.150357404584835,
      "learning_rate": 1.88249550607037e-05,
      "loss": 0.4492,
      "step": 1925
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4061668168984032,
      "learning_rate": 1.8823520686594087e-05,
      "loss": 0.457,
      "step": 1926
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3139278621104586,
      "learning_rate": 1.8822085492265235e-05,
      "loss": 0.478,
      "step": 1927
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.428009607056949,
      "learning_rate": 1.8820649477850562e-05,
      "loss": 0.5273,
      "step": 1928
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.396465438707096,
      "learning_rate": 1.881921264348355e-05,
      "loss": 0.4756,
      "step": 1929
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6071742030999694,
      "learning_rate": 1.8817774989297776e-05,
      "loss": 0.5161,
      "step": 1930
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.297023506818782,
      "learning_rate": 1.8816336515426873e-05,
      "loss": 0.4775,
      "step": 1931
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3376940558702854,
      "learning_rate": 1.8814897222004564e-05,
      "loss": 0.4316,
      "step": 1932
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5206692603326624,
      "learning_rate": 1.8813457109164642e-05,
      "loss": 0.6206,
      "step": 1933
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3200809888474931,
      "learning_rate": 1.8812016177040975e-05,
      "loss": 0.4434,
      "step": 1934
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5096040506233552,
      "learning_rate": 1.8810574425767512e-05,
      "loss": 0.6123,
      "step": 1935
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1818154610824168,
      "learning_rate": 1.8809131855478276e-05,
      "loss": 0.4214,
      "step": 1936
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3387220525634618,
      "learning_rate": 1.8807688466307362e-05,
      "loss": 0.5024,
      "step": 1937
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4823439589103107,
      "learning_rate": 1.880624425838895e-05,
      "loss": 0.4375,
      "step": 1938
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3972070865348007,
      "learning_rate": 1.8804799231857292e-05,
      "loss": 0.5044,
      "step": 1939
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3379615240598481,
      "learning_rate": 1.8803353386846708e-05,
      "loss": 0.5244,
      "step": 1940
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.333879918979691,
      "learning_rate": 1.8801906723491606e-05,
      "loss": 0.4902,
      "step": 1941
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.355802844962232,
      "learning_rate": 1.8800459241926466e-05,
      "loss": 0.4482,
      "step": 1942
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3103488099384693,
      "learning_rate": 1.879901094228584e-05,
      "loss": 0.4766,
      "step": 1943
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2570429225073403,
      "learning_rate": 1.8797561824704364e-05,
      "loss": 0.4419,
      "step": 1944
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3692312009602732,
      "learning_rate": 1.8796111889316742e-05,
      "loss": 0.5679,
      "step": 1945
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2616510776614638,
      "learning_rate": 1.879466113625776e-05,
      "loss": 0.4033,
      "step": 1946
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3536213257881051,
      "learning_rate": 1.8793209565662273e-05,
      "loss": 0.3789,
      "step": 1947
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3526534377910528,
      "learning_rate": 1.8791757177665223e-05,
      "loss": 0.5098,
      "step": 1948
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2537418713689632,
      "learning_rate": 1.8790303972401616e-05,
      "loss": 0.4248,
      "step": 1949
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.388883216280266,
      "learning_rate": 1.878884995000654e-05,
      "loss": 0.5059,
      "step": 1950
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2925015100642325,
      "learning_rate": 1.8787395110615163e-05,
      "loss": 0.4434,
      "step": 1951
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3418605329271427,
      "learning_rate": 1.878593945436272e-05,
      "loss": 0.5225,
      "step": 1952
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4660142863924668,
      "learning_rate": 1.8784482981384523e-05,
      "loss": 0.5986,
      "step": 1953
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3757464339886867,
      "learning_rate": 1.8783025691815974e-05,
      "loss": 0.4766,
      "step": 1954
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2388355871594263,
      "learning_rate": 1.878156758579253e-05,
      "loss": 0.4731,
      "step": 1955
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4730613280241214,
      "learning_rate": 1.8780108663449742e-05,
      "loss": 0.54,
      "step": 1956
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3916297258356503,
      "learning_rate": 1.8778648924923222e-05,
      "loss": 0.457,
      "step": 1957
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5558365706563324,
      "learning_rate": 1.8777188370348667e-05,
      "loss": 0.6372,
      "step": 1958
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.313574639508453,
      "learning_rate": 1.877572699986185e-05,
      "loss": 0.5239,
      "step": 1959
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5159265474019428,
      "learning_rate": 1.8774264813598614e-05,
      "loss": 0.4932,
      "step": 1960
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2273838085669613,
      "learning_rate": 1.8772801811694882e-05,
      "loss": 0.521,
      "step": 1961
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2877106674018008,
      "learning_rate": 1.8771337994286656e-05,
      "loss": 0.498,
      "step": 1962
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1895101223862088,
      "learning_rate": 1.8769873361510004e-05,
      "loss": 0.417,
      "step": 1963
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3537137600844955,
      "learning_rate": 1.876840791350108e-05,
      "loss": 0.5205,
      "step": 1964
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4512311542551282,
      "learning_rate": 1.8766941650396112e-05,
      "loss": 0.4521,
      "step": 1965
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3383370657173952,
      "learning_rate": 1.876547457233139e-05,
      "loss": 0.5005,
      "step": 1966
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.292566266344416,
      "learning_rate": 1.8764006679443306e-05,
      "loss": 0.4609,
      "step": 1967
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3159524324851595,
      "learning_rate": 1.87625379718683e-05,
      "loss": 0.4658,
      "step": 1968
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.23954770696975,
      "learning_rate": 1.876106844974291e-05,
      "loss": 0.4473,
      "step": 1969
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3313154204975677,
      "learning_rate": 1.875959811320373e-05,
      "loss": 0.5483,
      "step": 1970
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3471233783452705,
      "learning_rate": 1.875812696238745e-05,
      "loss": 0.5342,
      "step": 1971
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4220629786653263,
      "learning_rate": 1.8756654997430823e-05,
      "loss": 0.5117,
      "step": 1972
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4471423408731328,
      "learning_rate": 1.8755182218470675e-05,
      "loss": 0.4907,
      "step": 1973
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.357344896464073,
      "learning_rate": 1.8753708625643924e-05,
      "loss": 0.5259,
      "step": 1974
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3665396020568312,
      "learning_rate": 1.8752234219087538e-05,
      "loss": 0.4819,
      "step": 1975
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2450020533905972,
      "learning_rate": 1.8750758998938584e-05,
      "loss": 0.4272,
      "step": 1976
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2883174540460447,
      "learning_rate": 1.8749282965334198e-05,
      "loss": 0.4858,
      "step": 1977
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2078286472041029,
      "learning_rate": 1.8747806118411588e-05,
      "loss": 0.4497,
      "step": 1978
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2994521489251767,
      "learning_rate": 1.8746328458308034e-05,
      "loss": 0.4541,
      "step": 1979
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3707763999112146,
      "learning_rate": 1.87448499851609e-05,
      "loss": 0.4766,
      "step": 1980
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.23686552759167,
      "learning_rate": 1.8743370699107624e-05,
      "loss": 0.4395,
      "step": 1981
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.176084604555685,
      "learning_rate": 1.8741890600285714e-05,
      "loss": 0.4873,
      "step": 1982
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4200701193018375,
      "learning_rate": 1.8740409688832762e-05,
      "loss": 0.4688,
      "step": 1983
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5077869374100952,
      "learning_rate": 1.873892796488643e-05,
      "loss": 0.5762,
      "step": 1984
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3193606520957177,
      "learning_rate": 1.8737445428584456e-05,
      "loss": 0.4863,
      "step": 1985
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4282282102422703,
      "learning_rate": 1.8735962080064652e-05,
      "loss": 0.4971,
      "step": 1986
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5915003750297603,
      "learning_rate": 1.8734477919464905e-05,
      "loss": 0.5488,
      "step": 1987
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2958307034984744,
      "learning_rate": 1.8732992946923187e-05,
      "loss": 0.4785,
      "step": 1988
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.446538667588698,
      "learning_rate": 1.8731507162577536e-05,
      "loss": 0.5078,
      "step": 1989
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2498775909651432,
      "learning_rate": 1.8730020566566068e-05,
      "loss": 0.4021,
      "step": 1990
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3857741925783573,
      "learning_rate": 1.8728533159026972e-05,
      "loss": 0.5327,
      "step": 1991
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3780447513469136,
      "learning_rate": 1.8727044940098516e-05,
      "loss": 0.5005,
      "step": 1992
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4064983640034463,
      "learning_rate": 1.872555590991904e-05,
      "loss": 0.4229,
      "step": 1993
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3584853292912071,
      "learning_rate": 1.872406606862697e-05,
      "loss": 0.4438,
      "step": 1994
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.417229799186225,
      "learning_rate": 1.8722575416360794e-05,
      "loss": 0.4497,
      "step": 1995
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3187130373566283,
      "learning_rate": 1.8721083953259078e-05,
      "loss": 0.479,
      "step": 1996
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5209376750354844,
      "learning_rate": 1.8719591679460464e-05,
      "loss": 0.5684,
      "step": 1997
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4774630164826157,
      "learning_rate": 1.871809859510368e-05,
      "loss": 0.4756,
      "step": 1998
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3545761796704539,
      "learning_rate": 1.8716604700327516e-05,
      "loss": 0.4551,
      "step": 1999
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.625580117506669,
      "learning_rate": 1.8715109995270836e-05,
      "loss": 0.6313,
      "step": 2000
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2964424810651338,
      "learning_rate": 1.8713614480072594e-05,
      "loss": 0.417,
      "step": 2001
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2754299386685477,
      "learning_rate": 1.8712118154871808e-05,
      "loss": 0.46,
      "step": 2002
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.43703330029361,
      "learning_rate": 1.871062101980757e-05,
      "loss": 0.501,
      "step": 2003
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.470970397551701,
      "learning_rate": 1.870912307501906e-05,
      "loss": 0.5645,
      "step": 2004
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4246282287318215,
      "learning_rate": 1.870762432064551e-05,
      "loss": 0.5581,
      "step": 2005
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3124900681584324,
      "learning_rate": 1.8706124756826255e-05,
      "loss": 0.5127,
      "step": 2006
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2795338264382672,
      "learning_rate": 1.8704624383700686e-05,
      "loss": 0.439,
      "step": 2007
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4039752917150505,
      "learning_rate": 1.8703123201408277e-05,
      "loss": 0.5405,
      "step": 2008
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4069876817547957,
      "learning_rate": 1.8701621210088574e-05,
      "loss": 0.5947,
      "step": 2009
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3277043695851425,
      "learning_rate": 1.8700118409881198e-05,
      "loss": 0.4038,
      "step": 2010
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6188803603335655,
      "learning_rate": 1.8698614800925853e-05,
      "loss": 0.584,
      "step": 2011
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.380331262016451,
      "learning_rate": 1.86971103833623e-05,
      "loss": 0.5679,
      "step": 2012
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3526967250547794,
      "learning_rate": 1.8695605157330398e-05,
      "loss": 0.522,
      "step": 2013
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.30998521526328,
      "learning_rate": 1.869409912297007e-05,
      "loss": 0.5166,
      "step": 2014
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2960796656744995,
      "learning_rate": 1.8692592280421305e-05,
      "loss": 0.439,
      "step": 2015
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3954475372502635,
      "learning_rate": 1.8691084629824186e-05,
      "loss": 0.5811,
      "step": 2016
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.265866653055348,
      "learning_rate": 1.868957617131886e-05,
      "loss": 0.375,
      "step": 2017
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3555451657670892,
      "learning_rate": 1.8688066905045545e-05,
      "loss": 0.5698,
      "step": 2018
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4067987369788852,
      "learning_rate": 1.8686556831144545e-05,
      "loss": 0.5596,
      "step": 2019
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1647330453636298,
      "learning_rate": 1.8685045949756232e-05,
      "loss": 0.4014,
      "step": 2020
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4637736424350762,
      "learning_rate": 1.8683534261021058e-05,
      "loss": 0.5112,
      "step": 2021
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2849218128895623,
      "learning_rate": 1.8682021765079537e-05,
      "loss": 0.4771,
      "step": 2022
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4545108267486586,
      "learning_rate": 1.8680508462072282e-05,
      "loss": 0.4595,
      "step": 2023
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.312516018807833,
      "learning_rate": 1.867899435213996e-05,
      "loss": 0.4585,
      "step": 2024
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4024564392083394,
      "learning_rate": 1.867747943542332e-05,
      "loss": 0.4282,
      "step": 2025
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2172817450280968,
      "learning_rate": 1.8675963712063184e-05,
      "loss": 0.3999,
      "step": 2026
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.21969336784812,
      "learning_rate": 1.8674447182200457e-05,
      "loss": 0.4277,
      "step": 2027
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4177351822504418,
      "learning_rate": 1.8672929845976106e-05,
      "loss": 0.4512,
      "step": 2028
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5154700923695699,
      "learning_rate": 1.8671411703531186e-05,
      "loss": 0.5806,
      "step": 2029
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2090716795407401,
      "learning_rate": 1.8669892755006816e-05,
      "loss": 0.3625,
      "step": 2030
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3852793158011525,
      "learning_rate": 1.8668373000544197e-05,
      "loss": 0.4946,
      "step": 2031
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4407569462003686,
      "learning_rate": 1.8666852440284607e-05,
      "loss": 0.5015,
      "step": 2032
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6704198124606995,
      "learning_rate": 1.8665331074369385e-05,
      "loss": 0.6133,
      "step": 2033
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4611872079897876,
      "learning_rate": 1.8663808902939965e-05,
      "loss": 0.5083,
      "step": 2034
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3670834620095176,
      "learning_rate": 1.8662285926137837e-05,
      "loss": 0.5449,
      "step": 2035
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3643790334386274,
      "learning_rate": 1.866076214410458e-05,
      "loss": 0.4424,
      "step": 2036
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4178056891771276,
      "learning_rate": 1.8659237556981836e-05,
      "loss": 0.4844,
      "step": 2037
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.326002172699595,
      "learning_rate": 1.8657712164911337e-05,
      "loss": 0.4927,
      "step": 2038
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1892124213459303,
      "learning_rate": 1.865618596803487e-05,
      "loss": 0.3735,
      "step": 2039
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3904735263909274,
      "learning_rate": 1.865465896649432e-05,
      "loss": 0.501,
      "step": 2040
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2277468625963626,
      "learning_rate": 1.8653131160431622e-05,
      "loss": 0.4146,
      "step": 2041
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.422365634535188,
      "learning_rate": 1.865160254998881e-05,
      "loss": 0.6094,
      "step": 2042
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5638968855686621,
      "learning_rate": 1.8650073135307973e-05,
      "loss": 0.4463,
      "step": 2043
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4919550997225395,
      "learning_rate": 1.8648542916531283e-05,
      "loss": 0.5801,
      "step": 2044
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4623795235253483,
      "learning_rate": 1.8647011893800987e-05,
      "loss": 0.5391,
      "step": 2045
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2866752042827665,
      "learning_rate": 1.864548006725941e-05,
      "loss": 0.46,
      "step": 2046
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4291067454935829,
      "learning_rate": 1.8643947437048944e-05,
      "loss": 0.5356,
      "step": 2047
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3081278632730096,
      "learning_rate": 1.8642414003312063e-05,
      "loss": 0.4868,
      "step": 2048
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.477470844542444,
      "learning_rate": 1.8640879766191305e-05,
      "loss": 0.542,
      "step": 2049
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.393766979640637,
      "learning_rate": 1.8639344725829302e-05,
      "loss": 0.5088,
      "step": 2050
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.584520695747727,
      "learning_rate": 1.863780888236874e-05,
      "loss": 0.6006,
      "step": 2051
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4192744990941009,
      "learning_rate": 1.863627223595239e-05,
      "loss": 0.4839,
      "step": 2052
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3211032554338367,
      "learning_rate": 1.8634734786723096e-05,
      "loss": 0.4111,
      "step": 2053
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.223183917349571,
      "learning_rate": 1.8633196534823778e-05,
      "loss": 0.4297,
      "step": 2054
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4706066654717196,
      "learning_rate": 1.863165748039743e-05,
      "loss": 0.5742,
      "step": 2055
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3071827658981217,
      "learning_rate": 1.8630117623587116e-05,
      "loss": 0.436,
      "step": 2056
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5444431557611547,
      "learning_rate": 1.862857696453598e-05,
      "loss": 0.5225,
      "step": 2057
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3613305924332957,
      "learning_rate": 1.862703550338724e-05,
      "loss": 0.5005,
      "step": 2058
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.389596954306525,
      "learning_rate": 1.862549324028419e-05,
      "loss": 0.4907,
      "step": 2059
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4280578735565819,
      "learning_rate": 1.862395017537019e-05,
      "loss": 0.5298,
      "step": 2060
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.293472920954391,
      "learning_rate": 1.8622406308788683e-05,
      "loss": 0.4507,
      "step": 2061
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5093355493883265,
      "learning_rate": 1.862086164068319e-05,
      "loss": 0.5938,
      "step": 2062
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2155042585408184,
      "learning_rate": 1.8619316171197292e-05,
      "loss": 0.4502,
      "step": 2063
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4109012683148343,
      "learning_rate": 1.861776990047466e-05,
      "loss": 0.5029,
      "step": 2064
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.217867508457121,
      "learning_rate": 1.8616222828659024e-05,
      "loss": 0.4028,
      "step": 2065
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2279165047639837,
      "learning_rate": 1.8614674955894206e-05,
      "loss": 0.3911,
      "step": 2066
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2607429188381256,
      "learning_rate": 1.8613126282324092e-05,
      "loss": 0.4648,
      "step": 2067
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.401451280288754,
      "learning_rate": 1.861157680809264e-05,
      "loss": 0.5054,
      "step": 2068
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1691678201064182,
      "learning_rate": 1.861002653334389e-05,
      "loss": 0.4648,
      "step": 2069
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.300196314889133,
      "learning_rate": 1.860847545822195e-05,
      "loss": 0.5093,
      "step": 2070
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3917979447594173,
      "learning_rate": 1.8606923582871007e-05,
      "loss": 0.4814,
      "step": 2071
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3416153781303524,
      "learning_rate": 1.8605370907435322e-05,
      "loss": 0.4873,
      "step": 2072
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.299679932795415,
      "learning_rate": 1.860381743205923e-05,
      "loss": 0.4399,
      "step": 2073
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4419291444001348,
      "learning_rate": 1.8602263156887136e-05,
      "loss": 0.5244,
      "step": 2074
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3179273642006848,
      "learning_rate": 1.8600708082063518e-05,
      "loss": 0.5449,
      "step": 2075
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.32122522488447,
      "learning_rate": 1.8599152207732945e-05,
      "loss": 0.4961,
      "step": 2076
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4691869021909598,
      "learning_rate": 1.859759553404004e-05,
      "loss": 0.5859,
      "step": 2077
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1860161615316411,
      "learning_rate": 1.859603806112951e-05,
      "loss": 0.4048,
      "step": 2078
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.300482651625352,
      "learning_rate": 1.859447978914614e-05,
      "loss": 0.479,
      "step": 2079
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.200003042465027,
      "learning_rate": 1.8592920718234775e-05,
      "loss": 0.3906,
      "step": 2080
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3532012831744298,
      "learning_rate": 1.859136084854035e-05,
      "loss": 0.4556,
      "step": 2081
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3722527427252276,
      "learning_rate": 1.8589800180207865e-05,
      "loss": 0.4678,
      "step": 2082
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3926412298694624,
      "learning_rate": 1.85882387133824e-05,
      "loss": 0.5034,
      "step": 2083
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5409311059266633,
      "learning_rate": 1.8586676448209107e-05,
      "loss": 0.5283,
      "step": 2084
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4446539037528883,
      "learning_rate": 1.8585113384833206e-05,
      "loss": 0.5034,
      "step": 2085
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2321214044683828,
      "learning_rate": 1.85835495234e-05,
      "loss": 0.4385,
      "step": 2086
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3299640147890952,
      "learning_rate": 1.8581984864054866e-05,
      "loss": 0.5049,
      "step": 2087
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.330132770422781,
      "learning_rate": 1.858041940694325e-05,
      "loss": 0.4805,
      "step": 2088
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6266913065725812,
      "learning_rate": 1.8578853152210666e-05,
      "loss": 0.5771,
      "step": 2089
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3132143300157957,
      "learning_rate": 1.8577286100002723e-05,
      "loss": 0.4463,
      "step": 2090
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.339590218693674,
      "learning_rate": 1.8575718250465083e-05,
      "loss": 0.4766,
      "step": 2091
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3507544364990753,
      "learning_rate": 1.8574149603743495e-05,
      "loss": 0.4434,
      "step": 2092
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3777618208180062,
      "learning_rate": 1.8572580159983772e-05,
      "loss": 0.4961,
      "step": 2093
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4053087400770559,
      "learning_rate": 1.8571009919331817e-05,
      "loss": 0.5723,
      "step": 2094
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3245063135654072,
      "learning_rate": 1.8569438881933587e-05,
      "loss": 0.4995,
      "step": 2095
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2922289333292327,
      "learning_rate": 1.8567867047935128e-05,
      "loss": 0.5049,
      "step": 2096
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2295862274498903,
      "learning_rate": 1.8566294417482552e-05,
      "loss": 0.4316,
      "step": 2097
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.592225176478738,
      "learning_rate": 1.8564720990722048e-05,
      "loss": 0.6426,
      "step": 2098
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3400530419941463,
      "learning_rate": 1.8563146767799884e-05,
      "loss": 0.4097,
      "step": 2099
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3756897447808882,
      "learning_rate": 1.8561571748862394e-05,
      "loss": 0.4678,
      "step": 2100
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6128304919358414,
      "learning_rate": 1.855999593405599e-05,
      "loss": 0.5464,
      "step": 2101
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3078323457013912,
      "learning_rate": 1.8558419323527152e-05,
      "loss": 0.522,
      "step": 2102
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3519985142947497,
      "learning_rate": 1.8556841917422443e-05,
      "loss": 0.5679,
      "step": 2103
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.307351895802653,
      "learning_rate": 1.8555263715888493e-05,
      "loss": 0.4492,
      "step": 2104
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.568412915827262,
      "learning_rate": 1.8553684719072017e-05,
      "loss": 0.5942,
      "step": 2105
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.361918524128438,
      "learning_rate": 1.855210492711979e-05,
      "loss": 0.4956,
      "step": 2106
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4525772654714246,
      "learning_rate": 1.8550524340178664e-05,
      "loss": 0.5303,
      "step": 2107
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3651905047189836,
      "learning_rate": 1.8548942958395573e-05,
      "loss": 0.4849,
      "step": 2108
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1266632505234493,
      "learning_rate": 1.8547360781917515e-05,
      "loss": 0.4062,
      "step": 2109
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2386688613619241,
      "learning_rate": 1.854577781089157e-05,
      "loss": 0.4014,
      "step": 2110
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4263248095245806,
      "learning_rate": 1.8544194045464888e-05,
      "loss": 0.5176,
      "step": 2111
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2856744354426883,
      "learning_rate": 1.854260948578469e-05,
      "loss": 0.4336,
      "step": 2112
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4828843418221176,
      "learning_rate": 1.8541024131998277e-05,
      "loss": 0.5674,
      "step": 2113
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6706976551931223,
      "learning_rate": 1.853943798425302e-05,
      "loss": 0.626,
      "step": 2114
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4497152807836013,
      "learning_rate": 1.8537851042696363e-05,
      "loss": 0.5757,
      "step": 2115
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4497515249153907,
      "learning_rate": 1.8536263307475826e-05,
      "loss": 0.5947,
      "step": 2116
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.360012485929054,
      "learning_rate": 1.8534674778739004e-05,
      "loss": 0.4517,
      "step": 2117
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5632842663635773,
      "learning_rate": 1.853308545663356e-05,
      "loss": 0.5522,
      "step": 2118
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5092924046427318,
      "learning_rate": 1.853149534130724e-05,
      "loss": 0.5615,
      "step": 2119
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4589920649537562,
      "learning_rate": 1.8529904432907858e-05,
      "loss": 0.5605,
      "step": 2120
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.219010525827927,
      "learning_rate": 1.8528312731583295e-05,
      "loss": 0.3955,
      "step": 2121
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2874104968536297,
      "learning_rate": 1.8526720237481518e-05,
      "loss": 0.457,
      "step": 2122
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.378577103323036,
      "learning_rate": 1.8525126950750564e-05,
      "loss": 0.5283,
      "step": 2123
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3842290982884777,
      "learning_rate": 1.852353287153854e-05,
      "loss": 0.5298,
      "step": 2124
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1924825943624795,
      "learning_rate": 1.8521937999993627e-05,
      "loss": 0.418,
      "step": 2125
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2045658494988607,
      "learning_rate": 1.8520342336264082e-05,
      "loss": 0.4092,
      "step": 2126
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.230659675461521,
      "learning_rate": 1.8518745880498242e-05,
      "loss": 0.4531,
      "step": 2127
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5016544617052048,
      "learning_rate": 1.8517148632844502e-05,
      "loss": 0.5303,
      "step": 2128
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4734729335256358,
      "learning_rate": 1.8515550593451347e-05,
      "loss": 0.5396,
      "step": 2129
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4224418413652369,
      "learning_rate": 1.8513951762467318e-05,
      "loss": 0.4844,
      "step": 2130
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2503463883706734,
      "learning_rate": 1.851235214004105e-05,
      "loss": 0.5015,
      "step": 2131
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3360149507321541,
      "learning_rate": 1.8510751726321233e-05,
      "loss": 0.5679,
      "step": 2132
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5217177291780926,
      "learning_rate": 1.850915052145664e-05,
      "loss": 0.4824,
      "step": 2133
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4649367577332817,
      "learning_rate": 1.850754852559612e-05,
      "loss": 0.5146,
      "step": 2134
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4150051266520092,
      "learning_rate": 1.8505945738888593e-05,
      "loss": 0.5269,
      "step": 2135
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.388358420386928,
      "learning_rate": 1.850434216148305e-05,
      "loss": 0.6309,
      "step": 2136
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4697714319374666,
      "learning_rate": 1.8502737793528552e-05,
      "loss": 0.4639,
      "step": 2137
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2672048065015207,
      "learning_rate": 1.850113263517424e-05,
      "loss": 0.3882,
      "step": 2138
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.366313743159405,
      "learning_rate": 1.849952668656933e-05,
      "loss": 0.4463,
      "step": 2139
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2285956912981,
      "learning_rate": 1.849791994786311e-05,
      "loss": 0.4229,
      "step": 2140
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2970382009999852,
      "learning_rate": 1.849631241920493e-05,
      "loss": 0.3955,
      "step": 2141
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3968559284598985,
      "learning_rate": 1.849470410074423e-05,
      "loss": 0.4441,
      "step": 2142
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3878734236712533,
      "learning_rate": 1.849309499263052e-05,
      "loss": 0.4287,
      "step": 2143
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.462503086129237,
      "learning_rate": 1.8491485095013373e-05,
      "loss": 0.4648,
      "step": 2144
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5388759121913271,
      "learning_rate": 1.8489874408042445e-05,
      "loss": 0.6069,
      "step": 2145
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3427105562838897,
      "learning_rate": 1.8488262931867464e-05,
      "loss": 0.3818,
      "step": 2146
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4785954507414918,
      "learning_rate": 1.848665066663823e-05,
      "loss": 0.5493,
      "step": 2147
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3931525636681819,
      "learning_rate": 1.848503761250461e-05,
      "loss": 0.4517,
      "step": 2148
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3650879010807944,
      "learning_rate": 1.848342376961656e-05,
      "loss": 0.4951,
      "step": 2149
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.518270059614588,
      "learning_rate": 1.8481809138124092e-05,
      "loss": 0.5747,
      "step": 2150
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.485006930707545,
      "learning_rate": 1.8480193718177305e-05,
      "loss": 0.5356,
      "step": 2151
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5868288850491956,
      "learning_rate": 1.8478577509926367e-05,
      "loss": 0.5732,
      "step": 2152
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2546542214711067,
      "learning_rate": 1.847696051352151e-05,
      "loss": 0.4106,
      "step": 2153
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3590907211274679,
      "learning_rate": 1.8475342729113053e-05,
      "loss": 0.5396,
      "step": 2154
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4007735860455535,
      "learning_rate": 1.8473724156851384e-05,
      "loss": 0.5288,
      "step": 2155
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3346095489550673,
      "learning_rate": 1.8472104796886957e-05,
      "loss": 0.5698,
      "step": 2156
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.279370836136493,
      "learning_rate": 1.847048464937031e-05,
      "loss": 0.5205,
      "step": 2157
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2944281884204218,
      "learning_rate": 1.8468863714452044e-05,
      "loss": 0.4668,
      "step": 2158
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3058626227701757,
      "learning_rate": 1.8467241992282842e-05,
      "loss": 0.4585,
      "step": 2159
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3338769563380908,
      "learning_rate": 1.846561948301346e-05,
      "loss": 0.5161,
      "step": 2160
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3275663024282167,
      "learning_rate": 1.8463996186794716e-05,
      "loss": 0.4702,
      "step": 2161
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.396588643733537,
      "learning_rate": 1.8462372103777512e-05,
      "loss": 0.4995,
      "step": 2162
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.197065403412127,
      "learning_rate": 1.846074723411282e-05,
      "loss": 0.5547,
      "step": 2163
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2855836889990786,
      "learning_rate": 1.845912157795169e-05,
      "loss": 0.562,
      "step": 2164
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.394680471863229,
      "learning_rate": 1.845749513544523e-05,
      "loss": 0.5884,
      "step": 2165
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3451813465060671,
      "learning_rate": 1.845586790674464e-05,
      "loss": 0.5415,
      "step": 2166
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3014445168299544,
      "learning_rate": 1.845423989200118e-05,
      "loss": 0.5034,
      "step": 2167
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2614405880258557,
      "learning_rate": 1.8452611091366188e-05,
      "loss": 0.4399,
      "step": 2168
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2927119644835716,
      "learning_rate": 1.8450981504991077e-05,
      "loss": 0.4229,
      "step": 2169
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2428751188328766,
      "learning_rate": 1.8449351133027327e-05,
      "loss": 0.4771,
      "step": 2170
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3008525787301388,
      "learning_rate": 1.8447719975626496e-05,
      "loss": 0.4282,
      "step": 2171
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4951525191850168,
      "learning_rate": 1.8446088032940217e-05,
      "loss": 0.5654,
      "step": 2172
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.577591330054329,
      "learning_rate": 1.8444455305120187e-05,
      "loss": 0.6323,
      "step": 2173
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2047148673505594,
      "learning_rate": 1.8442821792318183e-05,
      "loss": 0.4053,
      "step": 2174
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.414840760891781,
      "learning_rate": 1.8441187494686055e-05,
      "loss": 0.5405,
      "step": 2175
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.325347369221581,
      "learning_rate": 1.8439552412375725e-05,
      "loss": 0.437,
      "step": 2176
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.467105342342801,
      "learning_rate": 1.8437916545539185e-05,
      "loss": 0.5713,
      "step": 2177
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4173845980994546,
      "learning_rate": 1.8436279894328507e-05,
      "loss": 0.4805,
      "step": 2178
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5371392500998329,
      "learning_rate": 1.8434642458895823e-05,
      "loss": 0.4712,
      "step": 2179
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2487308987683128,
      "learning_rate": 1.8433004239393353e-05,
      "loss": 0.4858,
      "step": 2180
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4428255802821606,
      "learning_rate": 1.8431365235973383e-05,
      "loss": 0.4663,
      "step": 2181
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5244043631870299,
      "learning_rate": 1.8429725448788267e-05,
      "loss": 0.5229,
      "step": 2182
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.304075003388159,
      "learning_rate": 1.8428084877990443e-05,
      "loss": 0.4717,
      "step": 2183
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5545724622696833,
      "learning_rate": 1.8426443523732412e-05,
      "loss": 0.6572,
      "step": 2184
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.446866026176347,
      "learning_rate": 1.8424801386166752e-05,
      "loss": 0.564,
      "step": 2185
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5735553224083219,
      "learning_rate": 1.842315846544611e-05,
      "loss": 0.5557,
      "step": 2186
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3011662015602496,
      "learning_rate": 1.8421514761723216e-05,
      "loss": 0.46,
      "step": 2187
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3890724369115734,
      "learning_rate": 1.841987027515086e-05,
      "loss": 0.4819,
      "step": 2188
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.448134918113577,
      "learning_rate": 1.8418225005881916e-05,
      "loss": 0.5796,
      "step": 2189
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.500846833816944,
      "learning_rate": 1.8416578954069318e-05,
      "loss": 0.5654,
      "step": 2190
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.431884372858895,
      "learning_rate": 1.841493211986609e-05,
      "loss": 0.5195,
      "step": 2191
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3632490704793372,
      "learning_rate": 1.841328450342531e-05,
      "loss": 0.5239,
      "step": 2192
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4068075203600663,
      "learning_rate": 1.8411636104900138e-05,
      "loss": 0.667,
      "step": 2193
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2371690556105543,
      "learning_rate": 1.840998692444381e-05,
      "loss": 0.5356,
      "step": 2194
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3376905775869656,
      "learning_rate": 1.840833696220963e-05,
      "loss": 0.5239,
      "step": 2195
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2183104046196653,
      "learning_rate": 1.840668621835098e-05,
      "loss": 0.4497,
      "step": 2196
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3919694662790318,
      "learning_rate": 1.84050346930213e-05,
      "loss": 0.5933,
      "step": 2197
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3236218909783108,
      "learning_rate": 1.8403382386374124e-05,
      "loss": 0.5718,
      "step": 2198
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2197712605882407,
      "learning_rate": 1.840172929856304e-05,
      "loss": 0.395,
      "step": 2199
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1624033108848213,
      "learning_rate": 1.840007542974172e-05,
      "loss": 0.5107,
      "step": 2200
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1977286629530295,
      "learning_rate": 1.8398420780063905e-05,
      "loss": 0.479,
      "step": 2201
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.483759291101431,
      "learning_rate": 1.8396765349683404e-05,
      "loss": 0.561,
      "step": 2202
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3957917641409265,
      "learning_rate": 1.8395109138754104e-05,
      "loss": 0.5029,
      "step": 2203
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3358288282129966,
      "learning_rate": 1.8393452147429965e-05,
      "loss": 0.519,
      "step": 2204
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3308083128996973,
      "learning_rate": 1.8391794375865025e-05,
      "loss": 0.4409,
      "step": 2205
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3180711158558858,
      "learning_rate": 1.8390135824213376e-05,
      "loss": 0.5146,
      "step": 2206
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.372020458236938,
      "learning_rate": 1.8388476492629198e-05,
      "loss": 0.5234,
      "step": 2207
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4925934859337984,
      "learning_rate": 1.838681638126674e-05,
      "loss": 0.542,
      "step": 2208
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4460539924396567,
      "learning_rate": 1.8385155490280327e-05,
      "loss": 0.5898,
      "step": 2209
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4358517797724153,
      "learning_rate": 1.838349381982435e-05,
      "loss": 0.5703,
      "step": 2210
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5313053606766662,
      "learning_rate": 1.838183137005327e-05,
      "loss": 0.5732,
      "step": 2211
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4678784548654296,
      "learning_rate": 1.8380168141121635e-05,
      "loss": 0.5889,
      "step": 2212
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1971872545228122,
      "learning_rate": 1.8378504133184047e-05,
      "loss": 0.4233,
      "step": 2213
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5147748867067956,
      "learning_rate": 1.8376839346395194e-05,
      "loss": 0.4409,
      "step": 2214
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3014887917809617,
      "learning_rate": 1.837517378090983e-05,
      "loss": 0.4277,
      "step": 2215
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3935176465408428,
      "learning_rate": 1.8373507436882784e-05,
      "loss": 0.4712,
      "step": 2216
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2179813565518496,
      "learning_rate": 1.8371840314468956e-05,
      "loss": 0.4048,
      "step": 2217
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2231099119413649,
      "learning_rate": 1.837017241382332e-05,
      "loss": 0.4556,
      "step": 2218
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3012137937382224,
      "learning_rate": 1.836850373510092e-05,
      "loss": 0.4907,
      "step": 2219
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2226419434119697,
      "learning_rate": 1.8366834278456872e-05,
      "loss": 0.4731,
      "step": 2220
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.442872222059224,
      "learning_rate": 1.8365164044046367e-05,
      "loss": 0.5547,
      "step": 2221
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4096228531784485,
      "learning_rate": 1.836349303202467e-05,
      "loss": 0.5078,
      "step": 2222
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4362560110045615,
      "learning_rate": 1.836182124254711e-05,
      "loss": 0.4819,
      "step": 2223
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.320448245327657,
      "learning_rate": 1.83601486757691e-05,
      "loss": 0.5415,
      "step": 2224
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4137103978050445,
      "learning_rate": 1.835847533184611e-05,
      "loss": 0.5801,
      "step": 2225
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4047844137373988,
      "learning_rate": 1.83568012109337e-05,
      "loss": 0.5205,
      "step": 2226
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3582776465841373,
      "learning_rate": 1.8355126313187496e-05,
      "loss": 0.4668,
      "step": 2227
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4033718341288306,
      "learning_rate": 1.8353450638763178e-05,
      "loss": 0.6338,
      "step": 2228
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.414745395467186,
      "learning_rate": 1.835177418781653e-05,
      "loss": 0.5547,
      "step": 2229
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3319925024268449,
      "learning_rate": 1.8350096960503383e-05,
      "loss": 0.4907,
      "step": 2230
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.293925962451803,
      "learning_rate": 1.834841895697965e-05,
      "loss": 0.6011,
      "step": 2231
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2684149779553282,
      "learning_rate": 1.8346740177401318e-05,
      "loss": 0.4858,
      "step": 2232
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4334433174835264,
      "learning_rate": 1.8345060621924443e-05,
      "loss": 0.5596,
      "step": 2233
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3481934602248513,
      "learning_rate": 1.8343380290705153e-05,
      "loss": 0.5151,
      "step": 2234
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4133482890462776,
      "learning_rate": 1.834169918389965e-05,
      "loss": 0.542,
      "step": 2235
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3744831326977651,
      "learning_rate": 1.83400173016642e-05,
      "loss": 0.5391,
      "step": 2236
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.39465541307052,
      "learning_rate": 1.833833464415516e-05,
      "loss": 0.4922,
      "step": 2237
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5421182353583816,
      "learning_rate": 1.8336651211528934e-05,
      "loss": 0.6348,
      "step": 2238
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4045698747631776,
      "learning_rate": 1.833496700394202e-05,
      "loss": 0.5508,
      "step": 2239
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3176224156234553,
      "learning_rate": 1.833328202155098e-05,
      "loss": 0.4546,
      "step": 2240
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3459080523291138,
      "learning_rate": 1.833159626451244e-05,
      "loss": 0.4248,
      "step": 2241
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.328231932891867,
      "learning_rate": 1.8329909732983105e-05,
      "loss": 0.5337,
      "step": 2242
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4920319962986128,
      "learning_rate": 1.832822242711976e-05,
      "loss": 0.5879,
      "step": 2243
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2355785008008169,
      "learning_rate": 1.832653434707925e-05,
      "loss": 0.4663,
      "step": 2244
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3501220723718068,
      "learning_rate": 1.832484549301849e-05,
      "loss": 0.4883,
      "step": 2245
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5558368970074936,
      "learning_rate": 1.8323155865094483e-05,
      "loss": 0.6431,
      "step": 2246
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4444065642136457,
      "learning_rate": 1.8321465463464287e-05,
      "loss": 0.5508,
      "step": 2247
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5087243161585913,
      "learning_rate": 1.831977428828504e-05,
      "loss": 0.5918,
      "step": 2248
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.385836795950072,
      "learning_rate": 1.8318082339713955e-05,
      "loss": 0.4517,
      "step": 2249
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.612465011743367,
      "learning_rate": 1.831638961790831e-05,
      "loss": 0.6289,
      "step": 2250
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2464585975396858,
      "learning_rate": 1.8314696123025456e-05,
      "loss": 0.4932,
      "step": 2251
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.212945963973869,
      "learning_rate": 1.8313001855222812e-05,
      "loss": 0.4199,
      "step": 2252
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3349946138319002,
      "learning_rate": 1.8311306814657886e-05,
      "loss": 0.4551,
      "step": 2253
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5779838881150217,
      "learning_rate": 1.8309611001488242e-05,
      "loss": 0.5874,
      "step": 2254
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4743325235886182,
      "learning_rate": 1.8307914415871516e-05,
      "loss": 0.4863,
      "step": 2255
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5162322791754335,
      "learning_rate": 1.8306217057965427e-05,
      "loss": 0.5352,
      "step": 2256
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3198477164303086,
      "learning_rate": 1.8304518927927745e-05,
      "loss": 0.4429,
      "step": 2257
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.416271634891278,
      "learning_rate": 1.830282002591634e-05,
      "loss": 0.4927,
      "step": 2258
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2498193651202352,
      "learning_rate": 1.830112035208913e-05,
      "loss": 0.3682,
      "step": 2259
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2325237873262798,
      "learning_rate": 1.8299419906604115e-05,
      "loss": 0.5386,
      "step": 2260
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3453272270972658,
      "learning_rate": 1.8297718689619374e-05,
      "loss": 0.4487,
      "step": 2261
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.234334503899102,
      "learning_rate": 1.8296016701293037e-05,
      "loss": 0.4619,
      "step": 2262
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.270052805588769,
      "learning_rate": 1.829431394178332e-05,
      "loss": 0.3804,
      "step": 2263
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.480633295795306,
      "learning_rate": 1.8292610411248515e-05,
      "loss": 0.5498,
      "step": 2264
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4201502773917376,
      "learning_rate": 1.8290906109846974e-05,
      "loss": 0.5176,
      "step": 2265
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4230584112221913,
      "learning_rate": 1.828920103773713e-05,
      "loss": 0.4375,
      "step": 2266
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.330096336224516,
      "learning_rate": 1.8287495195077482e-05,
      "loss": 0.4268,
      "step": 2267
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3356798293722631,
      "learning_rate": 1.8285788582026602e-05,
      "loss": 0.5225,
      "step": 2268
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3217488696777788,
      "learning_rate": 1.828408119874313e-05,
      "loss": 0.5229,
      "step": 2269
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.436129967394272,
      "learning_rate": 1.8282373045385786e-05,
      "loss": 0.5518,
      "step": 2270
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.391599666127308,
      "learning_rate": 1.8280664122113356e-05,
      "loss": 0.522,
      "step": 2271
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2813857734746184,
      "learning_rate": 1.82789544290847e-05,
      "loss": 0.4839,
      "step": 2272
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.227259175944006,
      "learning_rate": 1.8277243966458744e-05,
      "loss": 0.4609,
      "step": 2273
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2669969169183428,
      "learning_rate": 1.827553273439449e-05,
      "loss": 0.4746,
      "step": 2274
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5993475500986958,
      "learning_rate": 1.8273820733051016e-05,
      "loss": 0.499,
      "step": 2275
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3604483790405926,
      "learning_rate": 1.8272107962587465e-05,
      "loss": 0.5068,
      "step": 2276
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4462427879361277,
      "learning_rate": 1.827039442316305e-05,
      "loss": 0.5376,
      "step": 2277
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4043192203042436,
      "learning_rate": 1.8268680114937064e-05,
      "loss": 0.4106,
      "step": 2278
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4927400959100188,
      "learning_rate": 1.8266965038068856e-05,
      "loss": 0.4917,
      "step": 2279
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4455690146065603,
      "learning_rate": 1.8265249192717868e-05,
      "loss": 0.5186,
      "step": 2280
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5434594917622473,
      "learning_rate": 1.8263532579043598e-05,
      "loss": 0.5537,
      "step": 2281
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3955643065131598,
      "learning_rate": 1.8261815197205617e-05,
      "loss": 0.4502,
      "step": 2282
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4204123947995992,
      "learning_rate": 1.8260097047363575e-05,
      "loss": 0.5015,
      "step": 2283
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4095612256776542,
      "learning_rate": 1.8258378129677184e-05,
      "loss": 0.502,
      "step": 2284
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1705257273535914,
      "learning_rate": 1.8256658444306233e-05,
      "loss": 0.3989,
      "step": 2285
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4987105139751518,
      "learning_rate": 1.825493799141058e-05,
      "loss": 0.5312,
      "step": 2286
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4940551610781954,
      "learning_rate": 1.8253216771150153e-05,
      "loss": 0.5635,
      "step": 2287
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4221134024800306,
      "learning_rate": 1.825149478368496e-05,
      "loss": 0.457,
      "step": 2288
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.504953715490201,
      "learning_rate": 1.8249772029175067e-05,
      "loss": 0.583,
      "step": 2289
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2421525702182588,
      "learning_rate": 1.8248048507780626e-05,
      "loss": 0.4155,
      "step": 2290
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3448168644790213,
      "learning_rate": 1.8246324219661848e-05,
      "loss": 0.481,
      "step": 2291
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4143450654981553,
      "learning_rate": 1.8244599164979015e-05,
      "loss": 0.4536,
      "step": 2292
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.330179031390935,
      "learning_rate": 1.8242873343892494e-05,
      "loss": 0.4873,
      "step": 2293
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2476865343105044,
      "learning_rate": 1.824114675656271e-05,
      "loss": 0.4404,
      "step": 2294
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5728589643596456,
      "learning_rate": 1.823941940315017e-05,
      "loss": 0.4692,
      "step": 2295
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.22251941048911,
      "learning_rate": 1.823769128381543e-05,
      "loss": 0.4663,
      "step": 2296
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4179770283116426,
      "learning_rate": 1.823596239871915e-05,
      "loss": 0.4507,
      "step": 2297
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4102214655358014,
      "learning_rate": 1.8234232748022033e-05,
      "loss": 0.5166,
      "step": 2298
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.288761182005454,
      "learning_rate": 1.823250233188487e-05,
      "loss": 0.5547,
      "step": 2299
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3860360523381534,
      "learning_rate": 1.8230771150468517e-05,
      "loss": 0.4531,
      "step": 2300
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3879595340330015,
      "learning_rate": 1.8229039203933903e-05,
      "loss": 0.4609,
      "step": 2301
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3876056951397915,
      "learning_rate": 1.8227306492442022e-05,
      "loss": 0.502,
      "step": 2302
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4291226151733054,
      "learning_rate": 1.8225573016153945e-05,
      "loss": 0.5723,
      "step": 2303
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3580347328151114,
      "learning_rate": 1.822383877523082e-05,
      "loss": 0.4897,
      "step": 2304
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5368344538914944,
      "learning_rate": 1.8222103769833844e-05,
      "loss": 0.5625,
      "step": 2305
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5933182304795996,
      "learning_rate": 1.8220368000124316e-05,
      "loss": 0.6221,
      "step": 2306
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2402679574152324,
      "learning_rate": 1.8218631466263584e-05,
      "loss": 0.481,
      "step": 2307
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2857837771924543,
      "learning_rate": 1.821689416841307e-05,
      "loss": 0.4956,
      "step": 2308
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4755191645059453,
      "learning_rate": 1.8215156106734274e-05,
      "loss": 0.605,
      "step": 2309
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3723848186852579,
      "learning_rate": 1.821341728138876e-05,
      "loss": 0.4619,
      "step": 2310
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1996780225265211,
      "learning_rate": 1.821167769253817e-05,
      "loss": 0.5586,
      "step": 2311
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3110449979077115,
      "learning_rate": 1.820993734034421e-05,
      "loss": 0.5093,
      "step": 2312
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3208815171648607,
      "learning_rate": 1.8208196224968663e-05,
      "loss": 0.5781,
      "step": 2313
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4510892206184898,
      "learning_rate": 1.820645434657338e-05,
      "loss": 0.4961,
      "step": 2314
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3685870405585692,
      "learning_rate": 1.8204711705320282e-05,
      "loss": 0.4897,
      "step": 2315
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5073690555501087,
      "learning_rate": 1.820296830137136e-05,
      "loss": 0.501,
      "step": 2316
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.234523150422742,
      "learning_rate": 1.820122413488868e-05,
      "loss": 0.416,
      "step": 2317
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3835067913037864,
      "learning_rate": 1.8199479206034374e-05,
      "loss": 0.5229,
      "step": 2318
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2979645160315747,
      "learning_rate": 1.8197733514970655e-05,
      "loss": 0.4932,
      "step": 2319
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3643480883865313,
      "learning_rate": 1.819598706185979e-05,
      "loss": 0.417,
      "step": 2320
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3297457033482383,
      "learning_rate": 1.8194239846864133e-05,
      "loss": 0.5117,
      "step": 2321
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4261858170244097,
      "learning_rate": 1.81924918701461e-05,
      "loss": 0.4863,
      "step": 2322
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2304846216350156,
      "learning_rate": 1.8190743131868176e-05,
      "loss": 0.4106,
      "step": 2323
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.406520519870464,
      "learning_rate": 1.8188993632192926e-05,
      "loss": 0.5518,
      "step": 2324
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1641794646039911,
      "learning_rate": 1.8187243371282976e-05,
      "loss": 0.4331,
      "step": 2325
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1820887873459056,
      "learning_rate": 1.8185492349301035e-05,
      "loss": 0.4062,
      "step": 2326
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3301870808385348,
      "learning_rate": 1.8183740566409867e-05,
      "loss": 0.5151,
      "step": 2327
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3111863080047887,
      "learning_rate": 1.8181988022772315e-05,
      "loss": 0.4146,
      "step": 2328
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.317345639948561,
      "learning_rate": 1.81802347185513e-05,
      "loss": 0.4502,
      "step": 2329
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2292734370012321,
      "learning_rate": 1.8178480653909795e-05,
      "loss": 0.4077,
      "step": 2330
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2584508412005744,
      "learning_rate": 1.8176725829010868e-05,
      "loss": 0.4468,
      "step": 2331
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2249775682912405,
      "learning_rate": 1.817497024401763e-05,
      "loss": 0.416,
      "step": 2332
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1902220001661177,
      "learning_rate": 1.817321389909329e-05,
      "loss": 0.4155,
      "step": 2333
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3258931396583509,
      "learning_rate": 1.817145679440111e-05,
      "loss": 0.5088,
      "step": 2334
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.471298376283359,
      "learning_rate": 1.816969893010442e-05,
      "loss": 0.5303,
      "step": 2335
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4682533106637052,
      "learning_rate": 1.8167940306366642e-05,
      "loss": 0.5562,
      "step": 2336
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3152298296975262,
      "learning_rate": 1.8166180923351244e-05,
      "loss": 0.4751,
      "step": 2337
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2798546959482195,
      "learning_rate": 1.8164420781221777e-05,
      "loss": 0.5386,
      "step": 2338
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2518351224845186,
      "learning_rate": 1.8162659880141865e-05,
      "loss": 0.4277,
      "step": 2339
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.537227091049253,
      "learning_rate": 1.8160898220275196e-05,
      "loss": 0.5156,
      "step": 2340
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6679874155514764,
      "learning_rate": 1.815913580178553e-05,
      "loss": 0.6318,
      "step": 2341
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3586869876087921,
      "learning_rate": 1.81573726248367e-05,
      "loss": 0.5361,
      "step": 2342
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3005304145005048,
      "learning_rate": 1.8155608689592604e-05,
      "loss": 0.52,
      "step": 2343
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2391790051212435,
      "learning_rate": 1.815384399621722e-05,
      "loss": 0.4727,
      "step": 2344
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.377714533851604,
      "learning_rate": 1.8152078544874582e-05,
      "loss": 0.5015,
      "step": 2345
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.356276327965675,
      "learning_rate": 1.8150312335728818e-05,
      "loss": 0.4995,
      "step": 2346
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3321147338461545,
      "learning_rate": 1.8148545368944096e-05,
      "loss": 0.5283,
      "step": 2347
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2546977653891136,
      "learning_rate": 1.814677764468468e-05,
      "loss": 0.4902,
      "step": 2348
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4630277857234826,
      "learning_rate": 1.8145009163114894e-05,
      "loss": 0.5244,
      "step": 2349
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.377833617994189,
      "learning_rate": 1.8143239924399127e-05,
      "loss": 0.5801,
      "step": 2350
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3256327030304818,
      "learning_rate": 1.8141469928701852e-05,
      "loss": 0.5503,
      "step": 2351
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4644790822469342,
      "learning_rate": 1.81396991761876e-05,
      "loss": 0.5259,
      "step": 2352
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6350142486177255,
      "learning_rate": 1.8137927667020975e-05,
      "loss": 0.4707,
      "step": 2353
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2857473602115153,
      "learning_rate": 1.8136155401366658e-05,
      "loss": 0.4912,
      "step": 2354
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3558810624322,
      "learning_rate": 1.8134382379389396e-05,
      "loss": 0.4917,
      "step": 2355
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2210411844479943,
      "learning_rate": 1.8132608601254003e-05,
      "loss": 0.3721,
      "step": 2356
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4921368322910147,
      "learning_rate": 1.8130834067125368e-05,
      "loss": 0.4785,
      "step": 2357
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2015292831217153,
      "learning_rate": 1.8129058777168447e-05,
      "loss": 0.4014,
      "step": 2358
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.292066395334125,
      "learning_rate": 1.812728273154827e-05,
      "loss": 0.4717,
      "step": 2359
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3546671766063194,
      "learning_rate": 1.8125505930429936e-05,
      "loss": 0.4346,
      "step": 2360
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3310965800956696,
      "learning_rate": 1.812372837397861e-05,
      "loss": 0.4805,
      "step": 2361
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3504312891978165,
      "learning_rate": 1.812195006235953e-05,
      "loss": 0.4375,
      "step": 2362
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.419554121215576,
      "learning_rate": 1.812017099573801e-05,
      "loss": 0.5688,
      "step": 2363
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4610903555105992,
      "learning_rate": 1.8118391174279426e-05,
      "loss": 0.4819,
      "step": 2364
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4026226836209217,
      "learning_rate": 1.8116610598149225e-05,
      "loss": 0.4619,
      "step": 2365
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5385514920919752,
      "learning_rate": 1.811482926751293e-05,
      "loss": 0.5654,
      "step": 2366
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3496246974373094,
      "learning_rate": 1.8113047182536128e-05,
      "loss": 0.4575,
      "step": 2367
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4661488762881667,
      "learning_rate": 1.811126434338448e-05,
      "loss": 0.5488,
      "step": 2368
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.431649204454558,
      "learning_rate": 1.8109480750223714e-05,
      "loss": 0.5732,
      "step": 2369
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4601318429461436,
      "learning_rate": 1.810769640321963e-05,
      "loss": 0.5269,
      "step": 2370
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4030794083409102,
      "learning_rate": 1.81059113025381e-05,
      "loss": 0.4795,
      "step": 2371
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4800978875607946,
      "learning_rate": 1.8104125448345066e-05,
      "loss": 0.5352,
      "step": 2372
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2417657550660375,
      "learning_rate": 1.810233884080653e-05,
      "loss": 0.4233,
      "step": 2373
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3210848260365218,
      "learning_rate": 1.8100551480088574e-05,
      "loss": 0.5063,
      "step": 2374
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3063706699641529,
      "learning_rate": 1.8098763366357354e-05,
      "loss": 0.5107,
      "step": 2375
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3248285660457346,
      "learning_rate": 1.8096974499779084e-05,
      "loss": 0.4961,
      "step": 2376
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.374104800910922,
      "learning_rate": 1.8095184880520058e-05,
      "loss": 0.5654,
      "step": 2377
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2035703106157347,
      "learning_rate": 1.8093394508746635e-05,
      "loss": 0.478,
      "step": 2378
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2846657805229285,
      "learning_rate": 1.8091603384625243e-05,
      "loss": 0.4336,
      "step": 2379
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3204829817854582,
      "learning_rate": 1.8089811508322382e-05,
      "loss": 0.3999,
      "step": 2380
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5068079156668206,
      "learning_rate": 1.8088018880004622e-05,
      "loss": 0.5474,
      "step": 2381
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3290718237112409,
      "learning_rate": 1.8086225499838602e-05,
      "loss": 0.519,
      "step": 2382
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4590196949475944,
      "learning_rate": 1.8084431367991032e-05,
      "loss": 0.4868,
      "step": 2383
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.498307425808981,
      "learning_rate": 1.80826364846287e-05,
      "loss": 0.5376,
      "step": 2384
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2844309165441408,
      "learning_rate": 1.808084084991844e-05,
      "loss": 0.5049,
      "step": 2385
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4172015903319761,
      "learning_rate": 1.8079044464027177e-05,
      "loss": 0.4961,
      "step": 2386
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4745520252375217,
      "learning_rate": 1.8077247327121904e-05,
      "loss": 0.5537,
      "step": 2387
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5763409503943817,
      "learning_rate": 1.8075449439369678e-05,
      "loss": 0.6602,
      "step": 2388
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.255680745542088,
      "learning_rate": 1.8073650800937627e-05,
      "loss": 0.5049,
      "step": 2389
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.339399837370153,
      "learning_rate": 1.8071851411992948e-05,
      "loss": 0.4912,
      "step": 2390
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4771286769166854,
      "learning_rate": 1.8070051272702905e-05,
      "loss": 0.5732,
      "step": 2391
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3261430169302106,
      "learning_rate": 1.806825038323485e-05,
      "loss": 0.5024,
      "step": 2392
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.266752269953783,
      "learning_rate": 1.8066448743756173e-05,
      "loss": 0.4268,
      "step": 2393
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5157398662012722,
      "learning_rate": 1.8064646354434363e-05,
      "loss": 0.584,
      "step": 2394
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4155340409434034,
      "learning_rate": 1.8062843215436965e-05,
      "loss": 0.5024,
      "step": 2395
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4007216542485565,
      "learning_rate": 1.8061039326931592e-05,
      "loss": 0.5166,
      "step": 2396
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5504190833251694,
      "learning_rate": 1.8059234689085935e-05,
      "loss": 0.4995,
      "step": 2397
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2625977410658926,
      "learning_rate": 1.8057429302067748e-05,
      "loss": 0.4214,
      "step": 2398
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4743574961086505,
      "learning_rate": 1.8055623166044855e-05,
      "loss": 0.499,
      "step": 2399
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.402489499517517,
      "learning_rate": 1.8053816281185154e-05,
      "loss": 0.4912,
      "step": 2400
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.450372525766138,
      "learning_rate": 1.8052008647656605e-05,
      "loss": 0.541,
      "step": 2401
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.510284283220455,
      "learning_rate": 1.8050200265627247e-05,
      "loss": 0.5342,
      "step": 2402
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.429962732684969,
      "learning_rate": 1.8048391135265185e-05,
      "loss": 0.5386,
      "step": 2403
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3055599121215684,
      "learning_rate": 1.8046581256738592e-05,
      "loss": 0.4766,
      "step": 2404
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3583477908594401,
      "learning_rate": 1.8044770630215706e-05,
      "loss": 0.5044,
      "step": 2405
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3341129372161518,
      "learning_rate": 1.8042959255864846e-05,
      "loss": 0.5068,
      "step": 2406
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3799496553086217,
      "learning_rate": 1.804114713385439e-05,
      "loss": 0.6045,
      "step": 2407
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.252920441324624,
      "learning_rate": 1.8039334264352794e-05,
      "loss": 0.4844,
      "step": 2408
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6519676706739719,
      "learning_rate": 1.8037520647528576e-05,
      "loss": 0.6758,
      "step": 2409
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5483256381772599,
      "learning_rate": 1.803570628355033e-05,
      "loss": 0.5503,
      "step": 2410
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2856588757121121,
      "learning_rate": 1.803389117258671e-05,
      "loss": 0.502,
      "step": 2411
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6197083536571295,
      "learning_rate": 1.803207531480645e-05,
      "loss": 0.52,
      "step": 2412
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2742878605480854,
      "learning_rate": 1.8030258710378348e-05,
      "loss": 0.4131,
      "step": 2413
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3487125798196513,
      "learning_rate": 1.8028441359471273e-05,
      "loss": 0.4546,
      "step": 2414
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.288683392397607,
      "learning_rate": 1.8026623262254164e-05,
      "loss": 0.4722,
      "step": 2415
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2460520903242929,
      "learning_rate": 1.8024804418896026e-05,
      "loss": 0.4624,
      "step": 2416
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.325954176535644,
      "learning_rate": 1.8022984829565935e-05,
      "loss": 0.4297,
      "step": 2417
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3845459308077255,
      "learning_rate": 1.8021164494433038e-05,
      "loss": 0.4922,
      "step": 2418
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2291077850900152,
      "learning_rate": 1.801934341366655e-05,
      "loss": 0.4507,
      "step": 2419
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3950691229462975,
      "learning_rate": 1.801752158743576e-05,
      "loss": 0.5015,
      "step": 2420
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.378351807994783,
      "learning_rate": 1.801569901591001e-05,
      "loss": 0.5156,
      "step": 2421
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5564577230660366,
      "learning_rate": 1.8013875699258738e-05,
      "loss": 0.5479,
      "step": 2422
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3184275688718883,
      "learning_rate": 1.8012051637651423e-05,
      "loss": 0.5029,
      "step": 2423
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4153603008088804,
      "learning_rate": 1.8010226831257636e-05,
      "loss": 0.5156,
      "step": 2424
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5620908654478451,
      "learning_rate": 1.8008401280247005e-05,
      "loss": 0.542,
      "step": 2425
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.532563287653426,
      "learning_rate": 1.8006574984789226e-05,
      "loss": 0.6006,
      "step": 2426
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3120968837509583,
      "learning_rate": 1.8004747945054076e-05,
      "loss": 0.4155,
      "step": 2427
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4447010957794681,
      "learning_rate": 1.8002920161211384e-05,
      "loss": 0.4897,
      "step": 2428
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2996839936343034,
      "learning_rate": 1.800109163343107e-05,
      "loss": 0.5044,
      "step": 2429
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3787121416267127,
      "learning_rate": 1.7999262361883102e-05,
      "loss": 0.4683,
      "step": 2430
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3303478063609673,
      "learning_rate": 1.7997432346737524e-05,
      "loss": 0.4912,
      "step": 2431
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4173535088494995,
      "learning_rate": 1.7995601588164456e-05,
      "loss": 0.499,
      "step": 2432
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3708297223451467,
      "learning_rate": 1.7993770086334082e-05,
      "loss": 0.4961,
      "step": 2433
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2512420794926489,
      "learning_rate": 1.7991937841416652e-05,
      "loss": 0.4531,
      "step": 2434
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2957579461298365,
      "learning_rate": 1.7990104853582494e-05,
      "loss": 0.4771,
      "step": 2435
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5649496215307808,
      "learning_rate": 1.7988271123001994e-05,
      "loss": 0.5269,
      "step": 2436
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6995398219176634,
      "learning_rate": 1.798643664984561e-05,
      "loss": 0.5835,
      "step": 2437
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3572769027102496,
      "learning_rate": 1.798460143428388e-05,
      "loss": 0.5103,
      "step": 2438
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3348480485223682,
      "learning_rate": 1.7982765476487398e-05,
      "loss": 0.4644,
      "step": 2439
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3328852686415573,
      "learning_rate": 1.7980928776626833e-05,
      "loss": 0.4824,
      "step": 2440
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2128079767615878,
      "learning_rate": 1.7979091334872915e-05,
      "loss": 0.4556,
      "step": 2441
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2074377432045016,
      "learning_rate": 1.797725315139646e-05,
      "loss": 0.4292,
      "step": 2442
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3314275820542953,
      "learning_rate": 1.7975414226368334e-05,
      "loss": 0.498,
      "step": 2443
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5028308252992963,
      "learning_rate": 1.797357455995948e-05,
      "loss": 0.5312,
      "step": 2444
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3618000454676107,
      "learning_rate": 1.7971734152340918e-05,
      "loss": 0.4932,
      "step": 2445
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2277892365794505,
      "learning_rate": 1.7969893003683726e-05,
      "loss": 0.4702,
      "step": 2446
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3790645190602109,
      "learning_rate": 1.7968051114159046e-05,
      "loss": 0.5337,
      "step": 2447
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3254216167761153,
      "learning_rate": 1.7966208483938108e-05,
      "loss": 0.5586,
      "step": 2448
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.40559210099027,
      "learning_rate": 1.7964365113192195e-05,
      "loss": 0.4751,
      "step": 2449
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4302037555977902,
      "learning_rate": 1.796252100209266e-05,
      "loss": 0.5029,
      "step": 2450
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.400090635199245,
      "learning_rate": 1.7960676150810937e-05,
      "loss": 0.5679,
      "step": 2451
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2535999905412265,
      "learning_rate": 1.7958830559518513e-05,
      "loss": 0.4238,
      "step": 2452
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3160801662352417,
      "learning_rate": 1.795698422838695e-05,
      "loss": 0.4941,
      "step": 2453
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.303076798541286,
      "learning_rate": 1.7955137157587886e-05,
      "loss": 0.4985,
      "step": 2454
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3243850734864093,
      "learning_rate": 1.795328934729302e-05,
      "loss": 0.5068,
      "step": 2455
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3801003770049687,
      "learning_rate": 1.7951440797674117e-05,
      "loss": 0.5967,
      "step": 2456
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.344044555008663,
      "learning_rate": 1.7949591508903016e-05,
      "loss": 0.459,
      "step": 2457
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3825298008220244,
      "learning_rate": 1.7947741481151628e-05,
      "loss": 0.5063,
      "step": 2458
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4744457204014263,
      "learning_rate": 1.7945890714591926e-05,
      "loss": 0.5239,
      "step": 2459
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.248801927909229,
      "learning_rate": 1.794403920939595e-05,
      "loss": 0.5259,
      "step": 2460
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3851613930414626,
      "learning_rate": 1.794218696573582e-05,
      "loss": 0.4546,
      "step": 2461
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3782589162982068,
      "learning_rate": 1.7940333983783715e-05,
      "loss": 0.54,
      "step": 2462
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3971098235934318,
      "learning_rate": 1.793848026371188e-05,
      "loss": 0.4409,
      "step": 2463
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5052205015199442,
      "learning_rate": 1.793662580569264e-05,
      "loss": 0.5674,
      "step": 2464
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5054232378467587,
      "learning_rate": 1.7934770609898377e-05,
      "loss": 0.6138,
      "step": 2465
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2117954661959187,
      "learning_rate": 1.7932914676501553e-05,
      "loss": 0.4111,
      "step": 2466
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2861911599872387,
      "learning_rate": 1.7931058005674687e-05,
      "loss": 0.4517,
      "step": 2467
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3678216858112275,
      "learning_rate": 1.7929200597590375e-05,
      "loss": 0.5,
      "step": 2468
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3525057890061791,
      "learning_rate": 1.7927342452421275e-05,
      "loss": 0.5127,
      "step": 2469
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4058715605994445,
      "learning_rate": 1.7925483570340118e-05,
      "loss": 0.4922,
      "step": 2470
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2412398887949352,
      "learning_rate": 1.7923623951519708e-05,
      "loss": 0.4658,
      "step": 2471
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3959676958748684,
      "learning_rate": 1.7921763596132905e-05,
      "loss": 0.4941,
      "step": 2472
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2460862730488393,
      "learning_rate": 1.7919902504352646e-05,
      "loss": 0.5547,
      "step": 2473
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3815733246818538,
      "learning_rate": 1.791804067635194e-05,
      "loss": 0.4819,
      "step": 2474
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2996038542597772,
      "learning_rate": 1.791617811230385e-05,
      "loss": 0.4575,
      "step": 2475
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2717636599288198,
      "learning_rate": 1.7914314812381524e-05,
      "loss": 0.4966,
      "step": 2476
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.387045734337128,
      "learning_rate": 1.791245077675817e-05,
      "loss": 0.606,
      "step": 2477
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5233598440429628,
      "learning_rate": 1.7910586005607063e-05,
      "loss": 0.6206,
      "step": 2478
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3699115403300575,
      "learning_rate": 1.7908720499101552e-05,
      "loss": 0.5918,
      "step": 2479
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.281274293658059,
      "learning_rate": 1.7906854257415048e-05,
      "loss": 0.4414,
      "step": 2480
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4832043025433883,
      "learning_rate": 1.7904987280721037e-05,
      "loss": 0.5176,
      "step": 2481
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2605450463554237,
      "learning_rate": 1.7903119569193066e-05,
      "loss": 0.4067,
      "step": 2482
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.15203987665172,
      "learning_rate": 1.790125112300476e-05,
      "loss": 0.4307,
      "step": 2483
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4584730649738942,
      "learning_rate": 1.78993819423298e-05,
      "loss": 0.541,
      "step": 2484
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3078622239955062,
      "learning_rate": 1.7897512027341945e-05,
      "loss": 0.4399,
      "step": 2485
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4293866231387005,
      "learning_rate": 1.789564137821502e-05,
      "loss": 0.5781,
      "step": 2486
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3695041224534932,
      "learning_rate": 1.7893769995122916e-05,
      "loss": 0.5259,
      "step": 2487
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3882411137450008,
      "learning_rate": 1.7891897878239595e-05,
      "loss": 0.5518,
      "step": 2488
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.337107948469726,
      "learning_rate": 1.7890025027739084e-05,
      "loss": 0.5483,
      "step": 2489
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.44406375603782,
      "learning_rate": 1.7888151443795478e-05,
      "loss": 0.5107,
      "step": 2490
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1166180954352594,
      "learning_rate": 1.7886277126582947e-05,
      "loss": 0.3774,
      "step": 2491
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3391448012618543,
      "learning_rate": 1.7884402076275723e-05,
      "loss": 0.543,
      "step": 2492
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2683462045522738,
      "learning_rate": 1.7882526293048102e-05,
      "loss": 0.4639,
      "step": 2493
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4442272023010518,
      "learning_rate": 1.788064977707446e-05,
      "loss": 0.4575,
      "step": 2494
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4101782133021363,
      "learning_rate": 1.7878772528529232e-05,
      "loss": 0.5425,
      "step": 2495
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3639589728543347,
      "learning_rate": 1.7876894547586924e-05,
      "loss": 0.4976,
      "step": 2496
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4156042524587915,
      "learning_rate": 1.7875015834422113e-05,
      "loss": 0.5049,
      "step": 2497
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.7345648747949725,
      "learning_rate": 1.7873136389209435e-05,
      "loss": 0.4604,
      "step": 2498
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2956965169169161,
      "learning_rate": 1.7871256212123605e-05,
      "loss": 0.4878,
      "step": 2499
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.385269070835936,
      "learning_rate": 1.78693753033394e-05,
      "loss": 0.4912,
      "step": 2500
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.314909566652341,
      "learning_rate": 1.7867493663031664e-05,
      "loss": 0.5283,
      "step": 2501
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4201858801286902,
      "learning_rate": 1.7865611291375313e-05,
      "loss": 0.5381,
      "step": 2502
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.379103412394212,
      "learning_rate": 1.7863728188545326e-05,
      "loss": 0.5254,
      "step": 2503
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3795570359391895,
      "learning_rate": 1.7861844354716757e-05,
      "loss": 0.5605,
      "step": 2504
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.521871524814506,
      "learning_rate": 1.7859959790064723e-05,
      "loss": 0.5459,
      "step": 2505
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3199563583292755,
      "learning_rate": 1.7858074494764406e-05,
      "loss": 0.5342,
      "step": 2506
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2047558508575942,
      "learning_rate": 1.7856188468991064e-05,
      "loss": 0.4512,
      "step": 2507
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5306751864741508,
      "learning_rate": 1.785430171292002e-05,
      "loss": 0.5728,
      "step": 2508
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4166123444552978,
      "learning_rate": 1.7852414226726654e-05,
      "loss": 0.5767,
      "step": 2509
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2267557344847018,
      "learning_rate": 1.7850526010586437e-05,
      "loss": 0.4644,
      "step": 2510
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3617313955975312,
      "learning_rate": 1.7848637064674887e-05,
      "loss": 0.4893,
      "step": 2511
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1934820007798521,
      "learning_rate": 1.78467473891676e-05,
      "loss": 0.4395,
      "step": 2512
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.429312743698147,
      "learning_rate": 1.784485698424023e-05,
      "loss": 0.479,
      "step": 2513
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2077402049777413,
      "learning_rate": 1.784296585006851e-05,
      "loss": 0.4458,
      "step": 2514
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3013529729013096,
      "learning_rate": 1.784107398682824e-05,
      "loss": 0.479,
      "step": 2515
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1411175988276245,
      "learning_rate": 1.7839181394695285e-05,
      "loss": 0.4082,
      "step": 2516
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.288361843613262,
      "learning_rate": 1.7837288073845566e-05,
      "loss": 0.5376,
      "step": 2517
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.401997821234701,
      "learning_rate": 1.7835394024455097e-05,
      "loss": 0.4272,
      "step": 2518
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.534287846293908,
      "learning_rate": 1.783349924669994e-05,
      "loss": 0.6572,
      "step": 2519
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3733756211945043,
      "learning_rate": 1.7831603740756223e-05,
      "loss": 0.4917,
      "step": 2520
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3657301176302705,
      "learning_rate": 1.7829707506800157e-05,
      "loss": 0.4541,
      "step": 2521
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2576784927418545,
      "learning_rate": 1.782781054500801e-05,
      "loss": 0.376,
      "step": 2522
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.314471686003398,
      "learning_rate": 1.782591285555613e-05,
      "loss": 0.4438,
      "step": 2523
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4393587534153203,
      "learning_rate": 1.7824014438620906e-05,
      "loss": 0.5488,
      "step": 2524
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5152481933740591,
      "learning_rate": 1.7822115294378824e-05,
      "loss": 0.5869,
      "step": 2525
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5205545507687819,
      "learning_rate": 1.7820215423006418e-05,
      "loss": 0.5898,
      "step": 2526
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.35963990250712,
      "learning_rate": 1.78183148246803e-05,
      "loss": 0.5298,
      "step": 2527
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.47658855395952,
      "learning_rate": 1.7816413499577146e-05,
      "loss": 0.5259,
      "step": 2528
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4334517420546746,
      "learning_rate": 1.78145114478737e-05,
      "loss": 0.4824,
      "step": 2529
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5779640932588652,
      "learning_rate": 1.7812608669746774e-05,
      "loss": 0.5601,
      "step": 2530
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3441625603382095,
      "learning_rate": 1.7810705165373245e-05,
      "loss": 0.4883,
      "step": 2531
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1712538281930958,
      "learning_rate": 1.7808800934930062e-05,
      "loss": 0.377,
      "step": 2532
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2898436363836985,
      "learning_rate": 1.7806895978594237e-05,
      "loss": 0.4663,
      "step": 2533
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3557661159730081,
      "learning_rate": 1.7804990296542856e-05,
      "loss": 0.5093,
      "step": 2534
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.395764920841167,
      "learning_rate": 1.7803083888953058e-05,
      "loss": 0.4858,
      "step": 2535
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3940915332122288,
      "learning_rate": 1.780117675600207e-05,
      "loss": 0.5254,
      "step": 2536
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3190278594526834,
      "learning_rate": 1.779926889786717e-05,
      "loss": 0.4741,
      "step": 2537
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3252425256155205,
      "learning_rate": 1.7797360314725707e-05,
      "loss": 0.4521,
      "step": 2538
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.233005417154304,
      "learning_rate": 1.779545100675511e-05,
      "loss": 0.4946,
      "step": 2539
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2601225188707432,
      "learning_rate": 1.779354097413285e-05,
      "loss": 0.4263,
      "step": 2540
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4087914780699726,
      "learning_rate": 1.7791630217036492e-05,
      "loss": 0.4937,
      "step": 2541
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4085316857390204,
      "learning_rate": 1.7789718735643655e-05,
      "loss": 0.5723,
      "step": 2542
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4233563027439906,
      "learning_rate": 1.7787806530132022e-05,
      "loss": 0.5928,
      "step": 2543
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4921842250845496,
      "learning_rate": 1.7785893600679353e-05,
      "loss": 0.542,
      "step": 2544
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.405465912595566,
      "learning_rate": 1.778397994746347e-05,
      "loss": 0.4912,
      "step": 2545
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3391770952885036,
      "learning_rate": 1.7782065570662263e-05,
      "loss": 0.5415,
      "step": 2546
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.754922132445484,
      "learning_rate": 1.7780150470453682e-05,
      "loss": 0.5825,
      "step": 2547
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2329320973075881,
      "learning_rate": 1.777823464701576e-05,
      "loss": 0.4795,
      "step": 2548
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.403457737189202,
      "learning_rate": 1.777631810052659e-05,
      "loss": 0.5728,
      "step": 2549
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3409946055315183,
      "learning_rate": 1.777440083116432e-05,
      "loss": 0.4297,
      "step": 2550
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4124773159050308,
      "learning_rate": 1.777248283910719e-05,
      "loss": 0.5459,
      "step": 2551
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5088232637772803,
      "learning_rate": 1.777056412453348e-05,
      "loss": 0.6455,
      "step": 2552
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5119377569113475,
      "learning_rate": 1.776864468762156e-05,
      "loss": 0.4526,
      "step": 2553
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.464347738427798,
      "learning_rate": 1.7766724528549856e-05,
      "loss": 0.5791,
      "step": 2554
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4113899418427234,
      "learning_rate": 1.776480364749686e-05,
      "loss": 0.5068,
      "step": 2555
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.448962350881202,
      "learning_rate": 1.7762882044641133e-05,
      "loss": 0.604,
      "step": 2556
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3539489168677181,
      "learning_rate": 1.7760959720161306e-05,
      "loss": 0.4766,
      "step": 2557
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1449057440529997,
      "learning_rate": 1.7759036674236074e-05,
      "loss": 0.4517,
      "step": 2558
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1201753029986117,
      "learning_rate": 1.77571129070442e-05,
      "loss": 0.4194,
      "step": 2559
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.264477014539888,
      "learning_rate": 1.7755188418764517e-05,
      "loss": 0.499,
      "step": 2560
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3035443992155984,
      "learning_rate": 1.7753263209575914e-05,
      "loss": 0.4805,
      "step": 2561
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4006337774833741,
      "learning_rate": 1.7751337279657365e-05,
      "loss": 0.4604,
      "step": 2562
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1532300942978808,
      "learning_rate": 1.77494106291879e-05,
      "loss": 0.3975,
      "step": 2563
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3763199115218319,
      "learning_rate": 1.7747483258346607e-05,
      "loss": 0.4888,
      "step": 2564
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2785014565041328,
      "learning_rate": 1.774555516731266e-05,
      "loss": 0.4902,
      "step": 2565
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4364221688422025,
      "learning_rate": 1.7743626356265292e-05,
      "loss": 0.5083,
      "step": 2566
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.363965955750412,
      "learning_rate": 1.7741696825383797e-05,
      "loss": 0.5186,
      "step": 2567
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3922948653733327,
      "learning_rate": 1.7739766574847542e-05,
      "loss": 0.4873,
      "step": 2568
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4431081535403065,
      "learning_rate": 1.7737835604835962e-05,
      "loss": 0.5659,
      "step": 2569
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5693328985880792,
      "learning_rate": 1.7735903915528553e-05,
      "loss": 0.5415,
      "step": 2570
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2076771066597518,
      "learning_rate": 1.7733971507104887e-05,
      "loss": 0.4106,
      "step": 2571
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3950457484458954,
      "learning_rate": 1.7732038379744592e-05,
      "loss": 0.5635,
      "step": 2572
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4600885067739426,
      "learning_rate": 1.773010453362737e-05,
      "loss": 0.6128,
      "step": 2573
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3495700373479298,
      "learning_rate": 1.772816996893299e-05,
      "loss": 0.5034,
      "step": 2574
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4686426506789245,
      "learning_rate": 1.7726234685841283e-05,
      "loss": 0.521,
      "step": 2575
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5148040983967144,
      "learning_rate": 1.772429868453215e-05,
      "loss": 0.5288,
      "step": 2576
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.247150927992013,
      "learning_rate": 1.772236196518556e-05,
      "loss": 0.4365,
      "step": 2577
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.385143300822547,
      "learning_rate": 1.7720424527981545e-05,
      "loss": 0.5386,
      "step": 2578
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4022837226932807,
      "learning_rate": 1.7718486373100207e-05,
      "loss": 0.4727,
      "step": 2579
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4277831610671945,
      "learning_rate": 1.7716547500721715e-05,
      "loss": 0.4771,
      "step": 2580
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4414358439197579,
      "learning_rate": 1.77146079110263e-05,
      "loss": 0.4888,
      "step": 2581
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2842541217127,
      "learning_rate": 1.7712667604194263e-05,
      "loss": 0.4521,
      "step": 2582
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.364029183691624,
      "learning_rate": 1.7710726580405977e-05,
      "loss": 0.5073,
      "step": 2583
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.377222552728353,
      "learning_rate": 1.770878483984187e-05,
      "loss": 0.501,
      "step": 2584
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3566398360288194,
      "learning_rate": 1.7706842382682445e-05,
      "loss": 0.584,
      "step": 2585
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2665928849739898,
      "learning_rate": 1.770489920910827e-05,
      "loss": 0.4556,
      "step": 2586
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2543242548474256,
      "learning_rate": 1.770295531929998e-05,
      "loss": 0.397,
      "step": 2587
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.446613024290769,
      "learning_rate": 1.770101071343827e-05,
      "loss": 0.4756,
      "step": 2588
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.40899999448991,
      "learning_rate": 1.7699065391703914e-05,
      "loss": 0.5645,
      "step": 2589
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2446130137784344,
      "learning_rate": 1.7697119354277746e-05,
      "loss": 0.4321,
      "step": 2590
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4549418107777907,
      "learning_rate": 1.769517260134066e-05,
      "loss": 0.5176,
      "step": 2591
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3421854989966506,
      "learning_rate": 1.769322513307363e-05,
      "loss": 0.4619,
      "step": 2592
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4405716839476284,
      "learning_rate": 1.769127694965768e-05,
      "loss": 0.6592,
      "step": 2593
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4976228794013822,
      "learning_rate": 1.768932805127392e-05,
      "loss": 0.5376,
      "step": 2594
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.270089860906637,
      "learning_rate": 1.768737843810351e-05,
      "loss": 0.4712,
      "step": 2595
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4535645110046067,
      "learning_rate": 1.7685428110327683e-05,
      "loss": 0.5874,
      "step": 2596
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1960101835022006,
      "learning_rate": 1.7683477068127742e-05,
      "loss": 0.4028,
      "step": 2597
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3940923773532643,
      "learning_rate": 1.7681525311685046e-05,
      "loss": 0.5234,
      "step": 2598
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.293528539720967,
      "learning_rate": 1.7679572841181033e-05,
      "loss": 0.4395,
      "step": 2599
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4347285200051867,
      "learning_rate": 1.76776196567972e-05,
      "loss": 0.4502,
      "step": 2600
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2790734747952364,
      "learning_rate": 1.767566575871511e-05,
      "loss": 0.397,
      "step": 2601
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.550503840179546,
      "learning_rate": 1.7673711147116392e-05,
      "loss": 0.5947,
      "step": 2602
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5808941271845272,
      "learning_rate": 1.767175582218275e-05,
      "loss": 0.5981,
      "step": 2603
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.270259029146733,
      "learning_rate": 1.766979978409594e-05,
      "loss": 0.5225,
      "step": 2604
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3230399529734342,
      "learning_rate": 1.76678430330378e-05,
      "loss": 0.4893,
      "step": 2605
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.360048536689945,
      "learning_rate": 1.766588556919022e-05,
      "loss": 0.4336,
      "step": 2606
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4602492486238345,
      "learning_rate": 1.766392739273516e-05,
      "loss": 0.563,
      "step": 2607
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5175471148074848,
      "learning_rate": 1.766196850385466e-05,
      "loss": 0.4463,
      "step": 2608
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3840827644585911,
      "learning_rate": 1.7660008902730804e-05,
      "loss": 0.4336,
      "step": 2609
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3907229263185141,
      "learning_rate": 1.7658048589545757e-05,
      "loss": 0.48,
      "step": 2610
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2899610297905344,
      "learning_rate": 1.7656087564481746e-05,
      "loss": 0.4087,
      "step": 2611
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4569011254159865,
      "learning_rate": 1.7654125827721066e-05,
      "loss": 0.5415,
      "step": 2612
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.456308549863808,
      "learning_rate": 1.7652163379446073e-05,
      "loss": 0.4658,
      "step": 2613
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1938707599461904,
      "learning_rate": 1.7650200219839198e-05,
      "loss": 0.3862,
      "step": 2614
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.537950201569925,
      "learning_rate": 1.7648236349082928e-05,
      "loss": 0.5347,
      "step": 2615
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3428683321670147,
      "learning_rate": 1.7646271767359824e-05,
      "loss": 0.4866,
      "step": 2616
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3945126630164402,
      "learning_rate": 1.764430647485251e-05,
      "loss": 0.5,
      "step": 2617
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.36819889425275,
      "learning_rate": 1.7642340471743675e-05,
      "loss": 0.5132,
      "step": 2618
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2253019947481842,
      "learning_rate": 1.7640373758216075e-05,
      "loss": 0.4722,
      "step": 2619
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1530338598644019,
      "learning_rate": 1.7638406334452535e-05,
      "loss": 0.4292,
      "step": 2620
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.470539769559348,
      "learning_rate": 1.7636438200635942e-05,
      "loss": 0.4902,
      "step": 2621
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3978424332388768,
      "learning_rate": 1.7634469356949246e-05,
      "loss": 0.4629,
      "step": 2622
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3666639752273184,
      "learning_rate": 1.7632499803575473e-05,
      "loss": 0.4888,
      "step": 2623
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3565828405314537,
      "learning_rate": 1.7630529540697708e-05,
      "loss": 0.479,
      "step": 2624
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3491795728112603,
      "learning_rate": 1.7628558568499103e-05,
      "loss": 0.4028,
      "step": 2625
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4393601961515667,
      "learning_rate": 1.7626586887162875e-05,
      "loss": 0.499,
      "step": 2626
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3932860911457912,
      "learning_rate": 1.7624614496872304e-05,
      "loss": 0.5122,
      "step": 2627
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.459650495629095,
      "learning_rate": 1.762264139781075e-05,
      "loss": 0.5186,
      "step": 2628
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.346560939666723,
      "learning_rate": 1.7620667590161626e-05,
      "loss": 0.4497,
      "step": 2629
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5653646734521218,
      "learning_rate": 1.7618693074108405e-05,
      "loss": 0.5732,
      "step": 2630
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.392413207990118,
      "learning_rate": 1.7616717849834644e-05,
      "loss": 0.5439,
      "step": 2631
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.343654622442227,
      "learning_rate": 1.7614741917523956e-05,
      "loss": 0.459,
      "step": 2632
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4412666575965691,
      "learning_rate": 1.7612765277360013e-05,
      "loss": 0.5293,
      "step": 2633
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4392057145573283,
      "learning_rate": 1.761078792952657e-05,
      "loss": 0.6733,
      "step": 2634
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1259696440572249,
      "learning_rate": 1.7608809874207426e-05,
      "loss": 0.397,
      "step": 2635
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3182133371370022,
      "learning_rate": 1.7606831111586467e-05,
      "loss": 0.4858,
      "step": 2636
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.341635943707837,
      "learning_rate": 1.7604851641847633e-05,
      "loss": 0.5376,
      "step": 2637
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.358331684187602,
      "learning_rate": 1.7602871465174934e-05,
      "loss": 0.4775,
      "step": 2638
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4514820547999747,
      "learning_rate": 1.7600890581752435e-05,
      "loss": 0.519,
      "step": 2639
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5225318313990204,
      "learning_rate": 1.7598908991764288e-05,
      "loss": 0.5527,
      "step": 2640
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1589258619146954,
      "learning_rate": 1.7596926695394692e-05,
      "loss": 0.4072,
      "step": 2641
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.339349991551602,
      "learning_rate": 1.7594943692827913e-05,
      "loss": 0.4282,
      "step": 2642
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2598056106360316,
      "learning_rate": 1.75929599842483e-05,
      "loss": 0.5273,
      "step": 2643
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3997597963851938,
      "learning_rate": 1.759097556984024e-05,
      "loss": 0.5532,
      "step": 2644
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2154357493514292,
      "learning_rate": 1.7588990449788213e-05,
      "loss": 0.4888,
      "step": 2645
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.329984319062216,
      "learning_rate": 1.7587004624276747e-05,
      "loss": 0.4795,
      "step": 2646
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4612668072914727,
      "learning_rate": 1.758501809349044e-05,
      "loss": 0.5957,
      "step": 2647
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3941618598672503,
      "learning_rate": 1.7583030857613957e-05,
      "loss": 0.5337,
      "step": 2648
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2861005755924715,
      "learning_rate": 1.7581042916832028e-05,
      "loss": 0.4585,
      "step": 2649
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3093949834605734,
      "learning_rate": 1.7579054271329457e-05,
      "loss": 0.4438,
      "step": 2650
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5457844787318669,
      "learning_rate": 1.7577064921291092e-05,
      "loss": 0.5083,
      "step": 2651
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3344248148951017,
      "learning_rate": 1.7575074866901863e-05,
      "loss": 0.4956,
      "step": 2652
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.232913939880477,
      "learning_rate": 1.757308410834677e-05,
      "loss": 0.5073,
      "step": 2653
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3901580994392446,
      "learning_rate": 1.757109264581086e-05,
      "loss": 0.5537,
      "step": 2654
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3504867782253551,
      "learning_rate": 1.756910047947926e-05,
      "loss": 0.4912,
      "step": 2655
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3416737118713347,
      "learning_rate": 1.7567107609537163e-05,
      "loss": 0.5566,
      "step": 2656
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2089263296500756,
      "learning_rate": 1.756511403616982e-05,
      "loss": 0.4043,
      "step": 2657
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2241991215381218,
      "learning_rate": 1.7563119759562545e-05,
      "loss": 0.4487,
      "step": 2658
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4202599380061636,
      "learning_rate": 1.7561124779900723e-05,
      "loss": 0.5024,
      "step": 2659
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3310290415599517,
      "learning_rate": 1.755912909736981e-05,
      "loss": 0.5068,
      "step": 2660
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3356085044420232,
      "learning_rate": 1.755713271215532e-05,
      "loss": 0.6162,
      "step": 2661
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2741175569423273,
      "learning_rate": 1.7555135624442833e-05,
      "loss": 0.4766,
      "step": 2662
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3029885947174902,
      "learning_rate": 1.755313783441799e-05,
      "loss": 0.5112,
      "step": 2663
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5945705359136388,
      "learning_rate": 1.755113934226651e-05,
      "loss": 0.6211,
      "step": 2664
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4621046371765094,
      "learning_rate": 1.754914014817416e-05,
      "loss": 0.5049,
      "step": 2665
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3756433287026821,
      "learning_rate": 1.754714025232679e-05,
      "loss": 0.5474,
      "step": 2666
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4149761198164585,
      "learning_rate": 1.7545139654910302e-05,
      "loss": 0.5298,
      "step": 2667
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2886487590096432,
      "learning_rate": 1.7543138356110675e-05,
      "loss": 0.4595,
      "step": 2668
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2666328446888495,
      "learning_rate": 1.7541136356113934e-05,
      "loss": 0.4077,
      "step": 2669
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.438512248915248,
      "learning_rate": 1.753913365510619e-05,
      "loss": 0.5259,
      "step": 2670
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2726112494355486,
      "learning_rate": 1.7537130253273613e-05,
      "loss": 0.3955,
      "step": 2671
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.604692057773581,
      "learning_rate": 1.7535126150802428e-05,
      "loss": 0.6104,
      "step": 2672
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5304152486633158,
      "learning_rate": 1.753312134787894e-05,
      "loss": 0.5635,
      "step": 2673
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4044963433990558,
      "learning_rate": 1.7531115844689505e-05,
      "loss": 0.5483,
      "step": 2674
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4039604725332,
      "learning_rate": 1.7529109641420557e-05,
      "loss": 0.5576,
      "step": 2675
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4416245612278484,
      "learning_rate": 1.7527102738258588e-05,
      "loss": 0.522,
      "step": 2676
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3110856989735586,
      "learning_rate": 1.7525095135390152e-05,
      "loss": 0.5278,
      "step": 2677
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3925905478855232,
      "learning_rate": 1.752308683300188e-05,
      "loss": 0.5845,
      "step": 2678
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.608276820863452,
      "learning_rate": 1.7521077831280453e-05,
      "loss": 0.5879,
      "step": 2679
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4739423182612783,
      "learning_rate": 1.751906813041263e-05,
      "loss": 0.5298,
      "step": 2680
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.285165031072395,
      "learning_rate": 1.7517057730585224e-05,
      "loss": 0.4888,
      "step": 2681
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2686189444981888,
      "learning_rate": 1.751504663198512e-05,
      "loss": 0.5698,
      "step": 2682
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2044010030890555,
      "learning_rate": 1.751303483479927e-05,
      "loss": 0.4775,
      "step": 2683
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.294450942361991,
      "learning_rate": 1.7511022339214682e-05,
      "loss": 0.4985,
      "step": 2684
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.276808951654449,
      "learning_rate": 1.750900914541844e-05,
      "loss": 0.5254,
      "step": 2685
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.173461024915496,
      "learning_rate": 1.750699525359768e-05,
      "loss": 0.4341,
      "step": 2686
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2259798534791915,
      "learning_rate": 1.7504980663939614e-05,
      "loss": 0.4341,
      "step": 2687
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.530725685834471,
      "learning_rate": 1.7502965376631515e-05,
      "loss": 0.4961,
      "step": 2688
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3521653475176858,
      "learning_rate": 1.750094939186072e-05,
      "loss": 0.5088,
      "step": 2689
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3770965225355605,
      "learning_rate": 1.749893270981463e-05,
      "loss": 0.4331,
      "step": 2690
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2940948415840585,
      "learning_rate": 1.7496915330680713e-05,
      "loss": 0.5146,
      "step": 2691
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2360541199060266,
      "learning_rate": 1.7494897254646503e-05,
      "loss": 0.4663,
      "step": 2692
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3590007546532537,
      "learning_rate": 1.7492878481899595e-05,
      "loss": 0.4722,
      "step": 2693
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.417961130052328,
      "learning_rate": 1.7490859012627652e-05,
      "loss": 0.6021,
      "step": 2694
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4303262552732479,
      "learning_rate": 1.7488838847018397e-05,
      "loss": 0.5903,
      "step": 2695
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3451804320951883,
      "learning_rate": 1.7486817985259627e-05,
      "loss": 0.4854,
      "step": 2696
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4311692819634816,
      "learning_rate": 1.7484796427539194e-05,
      "loss": 0.543,
      "step": 2697
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2091776285185676,
      "learning_rate": 1.7482774174045017e-05,
      "loss": 0.3809,
      "step": 2698
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5051550035260757,
      "learning_rate": 1.7480751224965083e-05,
      "loss": 0.4868,
      "step": 2699
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3230251469527186,
      "learning_rate": 1.747872758048744e-05,
      "loss": 0.4478,
      "step": 2700
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3524060957390949,
      "learning_rate": 1.747670324080021e-05,
      "loss": 0.522,
      "step": 2701
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.411700525774363,
      "learning_rate": 1.7474678206091563e-05,
      "loss": 0.52,
      "step": 2702
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2612144148504045,
      "learning_rate": 1.7472652476549747e-05,
      "loss": 0.501,
      "step": 2703
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2728714240723422,
      "learning_rate": 1.7470626052363068e-05,
      "loss": 0.5293,
      "step": 2704
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3491735794572561,
      "learning_rate": 1.74685989337199e-05,
      "loss": 0.5249,
      "step": 2705
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4340569317853717,
      "learning_rate": 1.7466571120808684e-05,
      "loss": 0.5312,
      "step": 2706
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2781834994913854,
      "learning_rate": 1.746454261381792e-05,
      "loss": 0.4326,
      "step": 2707
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.31125149501473,
      "learning_rate": 1.7462513412936168e-05,
      "loss": 0.5791,
      "step": 2708
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3096985158567274,
      "learning_rate": 1.7460483518352068e-05,
      "loss": 0.4878,
      "step": 2709
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.426342470303575,
      "learning_rate": 1.745845293025431e-05,
      "loss": 0.5083,
      "step": 2710
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2877885135632692,
      "learning_rate": 1.7456421648831658e-05,
      "loss": 0.4321,
      "step": 2711
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2889931216274675,
      "learning_rate": 1.745438967427293e-05,
      "loss": 0.4399,
      "step": 2712
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2986748574756133,
      "learning_rate": 1.7452357006767026e-05,
      "loss": 0.519,
      "step": 2713
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.360640939711408,
      "learning_rate": 1.745032364650289e-05,
      "loss": 0.5044,
      "step": 2714
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4008846342304992,
      "learning_rate": 1.744828959366954e-05,
      "loss": 0.5757,
      "step": 2715
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4138146694752973,
      "learning_rate": 1.744625484845606e-05,
      "loss": 0.4683,
      "step": 2716
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3346370984215556,
      "learning_rate": 1.7444219411051598e-05,
      "loss": 0.5278,
      "step": 2717
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.372368749355295,
      "learning_rate": 1.744218328164536e-05,
      "loss": 0.4946,
      "step": 2718
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.502692708454557,
      "learning_rate": 1.744014646042663e-05,
      "loss": 0.5947,
      "step": 2719
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3058008544225046,
      "learning_rate": 1.7438108947584737e-05,
      "loss": 0.561,
      "step": 2720
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3240760055796272,
      "learning_rate": 1.7436070743309093e-05,
      "loss": 0.498,
      "step": 2721
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3406828952432002,
      "learning_rate": 1.743403184778916e-05,
      "loss": 0.5605,
      "step": 2722
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.442680713345824,
      "learning_rate": 1.743199226121447e-05,
      "loss": 0.4712,
      "step": 2723
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4697002153251617,
      "learning_rate": 1.7429951983774626e-05,
      "loss": 0.5596,
      "step": 2724
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3742425206510946,
      "learning_rate": 1.742791101565928e-05,
      "loss": 0.4907,
      "step": 2725
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3140030328133008,
      "learning_rate": 1.7425869357058167e-05,
      "loss": 0.5293,
      "step": 2726
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4520284137267863,
      "learning_rate": 1.742382700816107e-05,
      "loss": 0.5127,
      "step": 2727
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.380464535732484,
      "learning_rate": 1.742178396915784e-05,
      "loss": 0.4966,
      "step": 2728
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3911426900313824,
      "learning_rate": 1.74197402402384e-05,
      "loss": 0.48,
      "step": 2729
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3587535298491942,
      "learning_rate": 1.7417695821592727e-05,
      "loss": 0.417,
      "step": 2730
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3569753180125206,
      "learning_rate": 1.7415650713410867e-05,
      "loss": 0.5303,
      "step": 2731
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2992484381251794,
      "learning_rate": 1.7413604915882932e-05,
      "loss": 0.4409,
      "step": 2732
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4625916403919887,
      "learning_rate": 1.7411558429199095e-05,
      "loss": 0.5337,
      "step": 2733
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1978695437828895,
      "learning_rate": 1.7409511253549592e-05,
      "loss": 0.3892,
      "step": 2734
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2643816410700146,
      "learning_rate": 1.7407463389124728e-05,
      "loss": 0.459,
      "step": 2735
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3358656640354656,
      "learning_rate": 1.7405414836114868e-05,
      "loss": 0.4888,
      "step": 2736
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.521242288561055,
      "learning_rate": 1.740336559471044e-05,
      "loss": 0.5283,
      "step": 2737
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2166739167335696,
      "learning_rate": 1.7401315665101942e-05,
      "loss": 0.3726,
      "step": 2738
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5790204132804273,
      "learning_rate": 1.7399265047479926e-05,
      "loss": 0.6045,
      "step": 2739
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4525835210519642,
      "learning_rate": 1.739721374203502e-05,
      "loss": 0.5293,
      "step": 2740
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3154268506144426,
      "learning_rate": 1.7395161748957905e-05,
      "loss": 0.4497,
      "step": 2741
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2163304217354713,
      "learning_rate": 1.7393109068439336e-05,
      "loss": 0.4277,
      "step": 2742
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3264207382394997,
      "learning_rate": 1.739105570067012e-05,
      "loss": 0.4849,
      "step": 2743
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4513558545029286,
      "learning_rate": 1.7389001645841137e-05,
      "loss": 0.5044,
      "step": 2744
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4013525474192088,
      "learning_rate": 1.738694690414333e-05,
      "loss": 0.5142,
      "step": 2745
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.36129945225726,
      "learning_rate": 1.7384891475767706e-05,
      "loss": 0.5078,
      "step": 2746
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3284894160219591,
      "learning_rate": 1.738283536090533e-05,
      "loss": 0.48,
      "step": 2747
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.380228572565522,
      "learning_rate": 1.7380778559747335e-05,
      "loss": 0.5405,
      "step": 2748
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3531953892281945,
      "learning_rate": 1.7378721072484923e-05,
      "loss": 0.5947,
      "step": 2749
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.202869080065457,
      "learning_rate": 1.7376662899309346e-05,
      "loss": 0.4702,
      "step": 2750
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3791929747325966,
      "learning_rate": 1.7374604040411934e-05,
      "loss": 0.5806,
      "step": 2751
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.318544239400494,
      "learning_rate": 1.7372544495984076e-05,
      "loss": 0.6016,
      "step": 2752
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4654052672885225,
      "learning_rate": 1.7370484266217223e-05,
      "loss": 0.5913,
      "step": 2753
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1886370884434283,
      "learning_rate": 1.7368423351302884e-05,
      "loss": 0.5049,
      "step": 2754
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4352419995632095,
      "learning_rate": 1.7366361751432645e-05,
      "loss": 0.5464,
      "step": 2755
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5065152698943822,
      "learning_rate": 1.7364299466798146e-05,
      "loss": 0.5967,
      "step": 2756
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3062399611644036,
      "learning_rate": 1.7362236497591097e-05,
      "loss": 0.4937,
      "step": 2757
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2227271405738018,
      "learning_rate": 1.7360172844003263e-05,
      "loss": 0.4312,
      "step": 2758
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3447785261233591,
      "learning_rate": 1.7358108506226477e-05,
      "loss": 0.585,
      "step": 2759
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1377180703919518,
      "learning_rate": 1.7356043484452643e-05,
      "loss": 0.418,
      "step": 2760
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2526482161926438,
      "learning_rate": 1.7353977778873718e-05,
      "loss": 0.4282,
      "step": 2761
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3299634149932205,
      "learning_rate": 1.7351911389681725e-05,
      "loss": 0.5352,
      "step": 2762
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2149169641776583,
      "learning_rate": 1.7349844317068754e-05,
      "loss": 0.4463,
      "step": 2763
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3692548977567658,
      "learning_rate": 1.7347776561226956e-05,
      "loss": 0.4966,
      "step": 2764
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3341925778155734,
      "learning_rate": 1.7345708122348543e-05,
      "loss": 0.5107,
      "step": 2765
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2618309201343767,
      "learning_rate": 1.73436390006258e-05,
      "loss": 0.4927,
      "step": 2766
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1615728519139457,
      "learning_rate": 1.7341569196251065e-05,
      "loss": 0.4658,
      "step": 2767
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2050973790390356,
      "learning_rate": 1.7339498709416744e-05,
      "loss": 0.4941,
      "step": 2768
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4258698484193462,
      "learning_rate": 1.7337427540315305e-05,
      "loss": 0.5908,
      "step": 2769
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4735008162644831,
      "learning_rate": 1.733535568913928e-05,
      "loss": 0.5239,
      "step": 2770
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3241725756775524,
      "learning_rate": 1.7333283156081266e-05,
      "loss": 0.4819,
      "step": 2771
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2977237411736042,
      "learning_rate": 1.733120994133392e-05,
      "loss": 0.4883,
      "step": 2772
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3400142280475562,
      "learning_rate": 1.732913604508997e-05,
      "loss": 0.4385,
      "step": 2773
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4434603997658457,
      "learning_rate": 1.73270614675422e-05,
      "loss": 0.4517,
      "step": 2774
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2532051426277258,
      "learning_rate": 1.732498620888345e-05,
      "loss": 0.4702,
      "step": 2775
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2834814579726976,
      "learning_rate": 1.7322910269306645e-05,
      "loss": 0.3672,
      "step": 2776
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3141253741438437,
      "learning_rate": 1.7320833649004754e-05,
      "loss": 0.4946,
      "step": 2777
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4206357059436392,
      "learning_rate": 1.7318756348170817e-05,
      "loss": 0.5176,
      "step": 2778
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3597416278140178,
      "learning_rate": 1.7316678366997935e-05,
      "loss": 0.4409,
      "step": 2779
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3336882055085002,
      "learning_rate": 1.731459970567928e-05,
      "loss": 0.459,
      "step": 2780
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3947572278638933,
      "learning_rate": 1.731252036440807e-05,
      "loss": 0.5273,
      "step": 2781
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.414280194175203,
      "learning_rate": 1.7310440343377608e-05,
      "loss": 0.5566,
      "step": 2782
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5706112982218268,
      "learning_rate": 1.730835964278124e-05,
      "loss": 0.7002,
      "step": 2783
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3909414772036175,
      "learning_rate": 1.7306278262812393e-05,
      "loss": 0.4785,
      "step": 2784
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3238325602793894,
      "learning_rate": 1.7304196203664544e-05,
      "loss": 0.5063,
      "step": 2785
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3579453480749328,
      "learning_rate": 1.7302113465531233e-05,
      "loss": 0.5054,
      "step": 2786
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4586090899718036,
      "learning_rate": 1.7300030048606077e-05,
      "loss": 0.6357,
      "step": 2787
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1825884047192128,
      "learning_rate": 1.729794595308274e-05,
      "loss": 0.4731,
      "step": 2788
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3074083587742178,
      "learning_rate": 1.7295861179154954e-05,
      "loss": 0.4307,
      "step": 2789
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3757833378189643,
      "learning_rate": 1.729377572701653e-05,
      "loss": 0.4966,
      "step": 2790
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2546149728472258,
      "learning_rate": 1.729168959686131e-05,
      "loss": 0.4595,
      "step": 2791
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4443206090249574,
      "learning_rate": 1.7289602788883227e-05,
      "loss": 0.5073,
      "step": 2792
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3175362295777644,
      "learning_rate": 1.728751530327627e-05,
      "loss": 0.4751,
      "step": 2793
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4185006882009816,
      "learning_rate": 1.7285427140234476e-05,
      "loss": 0.4624,
      "step": 2794
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.335047508453984,
      "learning_rate": 1.728333829995197e-05,
      "loss": 0.4814,
      "step": 2795
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3848517594187313,
      "learning_rate": 1.7281248782622916e-05,
      "loss": 0.5044,
      "step": 2796
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4261869009477737,
      "learning_rate": 1.7279158588441558e-05,
      "loss": 0.5542,
      "step": 2797
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.182093648166414,
      "learning_rate": 1.7277067717602197e-05,
      "loss": 0.395,
      "step": 2798
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2952106992549,
      "learning_rate": 1.7274976170299197e-05,
      "loss": 0.4536,
      "step": 2799
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.194362398739174,
      "learning_rate": 1.7272883946726986e-05,
      "loss": 0.4038,
      "step": 2800
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.243987990268697,
      "learning_rate": 1.727079104708005e-05,
      "loss": 0.4526,
      "step": 2801
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.381296340151661,
      "learning_rate": 1.7268697471552937e-05,
      "loss": 0.4473,
      "step": 2802
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2545752903917318,
      "learning_rate": 1.7266603220340273e-05,
      "loss": 0.3975,
      "step": 2803
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4779358297948606,
      "learning_rate": 1.7264508293636726e-05,
      "loss": 0.5586,
      "step": 2804
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3866128509484994,
      "learning_rate": 1.7262412691637044e-05,
      "loss": 0.4688,
      "step": 2805
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6003543118115264,
      "learning_rate": 1.7260316414536026e-05,
      "loss": 0.5176,
      "step": 2806
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4986762603717503,
      "learning_rate": 1.7258219462528543e-05,
      "loss": 0.5718,
      "step": 2807
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3895111470147035,
      "learning_rate": 1.725612183580952e-05,
      "loss": 0.4717,
      "step": 2808
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2612345705831773,
      "learning_rate": 1.7254023534573946e-05,
      "loss": 0.4666,
      "step": 2809
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.7160187603280241,
      "learning_rate": 1.7251924559016885e-05,
      "loss": 0.5977,
      "step": 2810
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3535730763972205,
      "learning_rate": 1.7249824909333445e-05,
      "loss": 0.5073,
      "step": 2811
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.259176132148292,
      "learning_rate": 1.7247724585718807e-05,
      "loss": 0.479,
      "step": 2812
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2151947075019096,
      "learning_rate": 1.7245623588368217e-05,
      "loss": 0.3667,
      "step": 2813
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3596255802171393,
      "learning_rate": 1.7243521917476984e-05,
      "loss": 0.4854,
      "step": 2814
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.391882147095078,
      "learning_rate": 1.7241419573240463e-05,
      "loss": 0.4722,
      "step": 2815
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3667482366795995,
      "learning_rate": 1.7239316555854096e-05,
      "loss": 0.4736,
      "step": 2816
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3228826311593018,
      "learning_rate": 1.723721286551337e-05,
      "loss": 0.4517,
      "step": 2817
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2776707250194137,
      "learning_rate": 1.7235108502413844e-05,
      "loss": 0.4512,
      "step": 2818
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2844938988068046,
      "learning_rate": 1.7233003466751133e-05,
      "loss": 0.4248,
      "step": 2819
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4213484897254252,
      "learning_rate": 1.7230897758720916e-05,
      "loss": 0.665,
      "step": 2820
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2953348735698662,
      "learning_rate": 1.722879137851894e-05,
      "loss": 0.5352,
      "step": 2821
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4682905023536843,
      "learning_rate": 1.7226684326341014e-05,
      "loss": 0.5464,
      "step": 2822
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4913183528719567,
      "learning_rate": 1.7224576602382993e-05,
      "loss": 0.6191,
      "step": 2823
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3508199145854263,
      "learning_rate": 1.722246820684082e-05,
      "loss": 0.4331,
      "step": 2824
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2484087065517,
      "learning_rate": 1.722035913991048e-05,
      "loss": 0.415,
      "step": 2825
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2113007236078779,
      "learning_rate": 1.7218249401788033e-05,
      "loss": 0.4771,
      "step": 2826
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2874048456535585,
      "learning_rate": 1.7216138992669593e-05,
      "loss": 0.4966,
      "step": 2827
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.290594087497585,
      "learning_rate": 1.7214027912751342e-05,
      "loss": 0.439,
      "step": 2828
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2652005739836842,
      "learning_rate": 1.7211916162229524e-05,
      "loss": 0.438,
      "step": 2829
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4304074177291861,
      "learning_rate": 1.720980374130044e-05,
      "loss": 0.5459,
      "step": 2830
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.314490204371776,
      "learning_rate": 1.720769065016046e-05,
      "loss": 0.5063,
      "step": 2831
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3401947045403626,
      "learning_rate": 1.720557688900601e-05,
      "loss": 0.4683,
      "step": 2832
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.360370107942328,
      "learning_rate": 1.7203462458033586e-05,
      "loss": 0.4604,
      "step": 2833
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2066219183993516,
      "learning_rate": 1.720134735743974e-05,
      "loss": 0.4336,
      "step": 2834
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.307119313955858,
      "learning_rate": 1.7199231587421087e-05,
      "loss": 0.4604,
      "step": 2835
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3526967028839236,
      "learning_rate": 1.7197115148174305e-05,
      "loss": 0.5112,
      "step": 2836
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.197479326642742,
      "learning_rate": 1.7194998039896134e-05,
      "loss": 0.4512,
      "step": 2837
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4371577746442938,
      "learning_rate": 1.719288026278338e-05,
      "loss": 0.4849,
      "step": 2838
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4031374186097727,
      "learning_rate": 1.719076181703291e-05,
      "loss": 0.4858,
      "step": 2839
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.425785264477991,
      "learning_rate": 1.7188642702841643e-05,
      "loss": 0.4897,
      "step": 2840
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1325465113295239,
      "learning_rate": 1.7186522920406572e-05,
      "loss": 0.3901,
      "step": 2841
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1936813594894953,
      "learning_rate": 1.7184402469924748e-05,
      "loss": 0.4985,
      "step": 2842
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6041542889419518,
      "learning_rate": 1.7182281351593288e-05,
      "loss": 0.5825,
      "step": 2843
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3416137664041912,
      "learning_rate": 1.718015956560936e-05,
      "loss": 0.4678,
      "step": 2844
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2004308589071013,
      "learning_rate": 1.7178037112170213e-05,
      "loss": 0.3931,
      "step": 2845
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3200049137551455,
      "learning_rate": 1.7175913991473137e-05,
      "loss": 0.4434,
      "step": 2846
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1702430632408791,
      "learning_rate": 1.7173790203715494e-05,
      "loss": 0.4092,
      "step": 2847
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3767310022008439,
      "learning_rate": 1.7171665749094713e-05,
      "loss": 0.5483,
      "step": 2848
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3480596523380344,
      "learning_rate": 1.7169540627808276e-05,
      "loss": 0.4919,
      "step": 2849
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.313402615932262,
      "learning_rate": 1.716741484005373e-05,
      "loss": 0.4453,
      "step": 2850
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3173835384628874,
      "learning_rate": 1.7165288386028683e-05,
      "loss": 0.4604,
      "step": 2851
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2815740120549644,
      "learning_rate": 1.7163161265930814e-05,
      "loss": 0.354,
      "step": 2852
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2425339387626761,
      "learning_rate": 1.716103347995785e-05,
      "loss": 0.3911,
      "step": 2853
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2374436572788605,
      "learning_rate": 1.715890502830759e-05,
      "loss": 0.4668,
      "step": 2854
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2490677450843353,
      "learning_rate": 1.7156775911177888e-05,
      "loss": 0.3975,
      "step": 2855
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4571561711843983,
      "learning_rate": 1.7154646128766663e-05,
      "loss": 0.5557,
      "step": 2856
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3790040336501896,
      "learning_rate": 1.7152515681271896e-05,
      "loss": 0.4312,
      "step": 2857
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.47197255005257,
      "learning_rate": 1.7150384568891634e-05,
      "loss": 0.5083,
      "step": 2858
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4544554582120743,
      "learning_rate": 1.714825279182398e-05,
      "loss": 0.5317,
      "step": 2859
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6482829579784322,
      "learning_rate": 1.7146120350267094e-05,
      "loss": 0.5542,
      "step": 2860
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3596016368563577,
      "learning_rate": 1.7143987244419212e-05,
      "loss": 0.4619,
      "step": 2861
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3178646398611515,
      "learning_rate": 1.7141853474478618e-05,
      "loss": 0.5142,
      "step": 2862
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.379286576014486,
      "learning_rate": 1.713971904064367e-05,
      "loss": 0.4946,
      "step": 2863
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4679442506314213,
      "learning_rate": 1.7137583943112776e-05,
      "loss": 0.5376,
      "step": 2864
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4943615155515881,
      "learning_rate": 1.713544818208441e-05,
      "loss": 0.5151,
      "step": 2865
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.342319091144411,
      "learning_rate": 1.7133311757757112e-05,
      "loss": 0.4341,
      "step": 2866
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.352877133489478,
      "learning_rate": 1.713117467032948e-05,
      "loss": 0.5464,
      "step": 2867
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.305355577854755,
      "learning_rate": 1.7129036920000174e-05,
      "loss": 0.5527,
      "step": 2868
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3778627223539854,
      "learning_rate": 1.7126898506967913e-05,
      "loss": 0.5103,
      "step": 2869
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.44201178641325,
      "learning_rate": 1.7124759431431485e-05,
      "loss": 0.4692,
      "step": 2870
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1873641438154818,
      "learning_rate": 1.712261969358973e-05,
      "loss": 0.4663,
      "step": 2871
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4398560043419977,
      "learning_rate": 1.7120479293641558e-05,
      "loss": 0.4863,
      "step": 2872
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2488046274145619,
      "learning_rate": 1.711833823178593e-05,
      "loss": 0.3716,
      "step": 2873
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3465293625065737,
      "learning_rate": 1.7116196508221886e-05,
      "loss": 0.4775,
      "step": 2874
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3080735790543203,
      "learning_rate": 1.7114054123148508e-05,
      "loss": 0.543,
      "step": 2875
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3546863662099247,
      "learning_rate": 1.7111911076764954e-05,
      "loss": 0.5181,
      "step": 2876
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3162171851923719,
      "learning_rate": 1.7109767369270437e-05,
      "loss": 0.4561,
      "step": 2877
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3813196590011156,
      "learning_rate": 1.710762300086423e-05,
      "loss": 0.4761,
      "step": 2878
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2329317532064261,
      "learning_rate": 1.7105477971745668e-05,
      "loss": 0.4238,
      "step": 2879
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3255919511927923,
      "learning_rate": 1.7103332282114156e-05,
      "loss": 0.5669,
      "step": 2880
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1479262885376875,
      "learning_rate": 1.7101185932169147e-05,
      "loss": 0.4111,
      "step": 2881
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.167281713040837,
      "learning_rate": 1.7099038922110164e-05,
      "loss": 0.4282,
      "step": 2882
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3009453471268981,
      "learning_rate": 1.709689125213679e-05,
      "loss": 0.4341,
      "step": 2883
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4249548822628484,
      "learning_rate": 1.7094742922448674e-05,
      "loss": 0.4868,
      "step": 2884
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4874529917427957,
      "learning_rate": 1.7092593933245513e-05,
      "loss": 0.541,
      "step": 2885
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.379764009764511,
      "learning_rate": 1.709044428472707e-05,
      "loss": 0.5859,
      "step": 2886
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4623037655220423,
      "learning_rate": 1.7088293977093187e-05,
      "loss": 0.4658,
      "step": 2887
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2451299119611006,
      "learning_rate": 1.7086143010543737e-05,
      "loss": 0.3843,
      "step": 2888
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3128720472801856,
      "learning_rate": 1.7083991385278686e-05,
      "loss": 0.5029,
      "step": 2889
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2345640802484878,
      "learning_rate": 1.7081839101498033e-05,
      "loss": 0.4893,
      "step": 2890
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5020418028219975,
      "learning_rate": 1.7079686159401853e-05,
      "loss": 0.542,
      "step": 2891
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5133112345708852,
      "learning_rate": 1.7077532559190282e-05,
      "loss": 0.5527,
      "step": 2892
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.276316221140839,
      "learning_rate": 1.7075378301063518e-05,
      "loss": 0.4995,
      "step": 2893
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2608516516440536,
      "learning_rate": 1.7073223385221806e-05,
      "loss": 0.4648,
      "step": 2894
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3055295406328757,
      "learning_rate": 1.7071067811865477e-05,
      "loss": 0.5093,
      "step": 2895
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4340422355755489,
      "learning_rate": 1.70689115811949e-05,
      "loss": 0.54,
      "step": 2896
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.229882456211707,
      "learning_rate": 1.706675469341052e-05,
      "loss": 0.479,
      "step": 2897
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1969804779550521,
      "learning_rate": 1.7064597148712834e-05,
      "loss": 0.4097,
      "step": 2898
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3267922387656592,
      "learning_rate": 1.7062438947302405e-05,
      "loss": 0.4482,
      "step": 2899
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4898288185975495,
      "learning_rate": 1.7060280089379854e-05,
      "loss": 0.5654,
      "step": 2900
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4137703926795706,
      "learning_rate": 1.7058120575145868e-05,
      "loss": 0.5249,
      "step": 2901
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.262791151313917,
      "learning_rate": 1.7055960404801187e-05,
      "loss": 0.4653,
      "step": 2902
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2534588970417158,
      "learning_rate": 1.7053799578546623e-05,
      "loss": 0.4609,
      "step": 2903
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4881567133051357,
      "learning_rate": 1.705163809658304e-05,
      "loss": 0.5371,
      "step": 2904
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.248552358572823,
      "learning_rate": 1.704947595911136e-05,
      "loss": 0.4229,
      "step": 2905
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3047607670350703,
      "learning_rate": 1.704731316633258e-05,
      "loss": 0.4941,
      "step": 2906
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4175642707603724,
      "learning_rate": 1.7045149718447743e-05,
      "loss": 0.52,
      "step": 2907
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3651222829695195,
      "learning_rate": 1.7042985615657964e-05,
      "loss": 0.542,
      "step": 2908
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2044140992711407,
      "learning_rate": 1.7040820858164413e-05,
      "loss": 0.4136,
      "step": 2909
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3452058953371404,
      "learning_rate": 1.703865544616832e-05,
      "loss": 0.4131,
      "step": 2910
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2465317106382714,
      "learning_rate": 1.7036489379870982e-05,
      "loss": 0.4805,
      "step": 2911
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2876158760218628,
      "learning_rate": 1.7034322659473748e-05,
      "loss": 0.5049,
      "step": 2912
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.216079219814481,
      "learning_rate": 1.7032155285178036e-05,
      "loss": 0.4585,
      "step": 2913
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3905296824970068,
      "learning_rate": 1.702998725718532e-05,
      "loss": 0.4453,
      "step": 2914
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.390738180585651,
      "learning_rate": 1.7027818575697136e-05,
      "loss": 0.5005,
      "step": 2915
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.538102532948229,
      "learning_rate": 1.7025649240915085e-05,
      "loss": 0.6089,
      "step": 2916
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4480543796194807,
      "learning_rate": 1.7023479253040817e-05,
      "loss": 0.5171,
      "step": 2917
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.293318100060326,
      "learning_rate": 1.7021308612276056e-05,
      "loss": 0.4736,
      "step": 2918
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4879315589248068,
      "learning_rate": 1.7019137318822577e-05,
      "loss": 0.5586,
      "step": 2919
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2872840517600688,
      "learning_rate": 1.7016965372882227e-05,
      "loss": 0.4937,
      "step": 2920
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4111669867227854,
      "learning_rate": 1.7014792774656896e-05,
      "loss": 0.5503,
      "step": 2921
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3405650968007532,
      "learning_rate": 1.7012619524348552e-05,
      "loss": 0.54,
      "step": 2922
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4527555439505393,
      "learning_rate": 1.7010445622159214e-05,
      "loss": 0.4854,
      "step": 2923
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4712387609728002,
      "learning_rate": 1.7008271068290966e-05,
      "loss": 0.5044,
      "step": 2924
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5759647931529404,
      "learning_rate": 1.7006095862945948e-05,
      "loss": 0.6182,
      "step": 2925
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4539046469992531,
      "learning_rate": 1.700392000632637e-05,
      "loss": 0.4805,
      "step": 2926
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.513705191075327,
      "learning_rate": 1.7001743498634487e-05,
      "loss": 0.5288,
      "step": 2927
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2349182605454254,
      "learning_rate": 1.6999566340072628e-05,
      "loss": 0.4375,
      "step": 2928
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3504440900241932,
      "learning_rate": 1.6997388530843176e-05,
      "loss": 0.5332,
      "step": 2929
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5444458978689304,
      "learning_rate": 1.6995210071148582e-05,
      "loss": 0.5,
      "step": 2930
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.408942892740854,
      "learning_rate": 1.6993030961191345e-05,
      "loss": 0.5049,
      "step": 2931
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.316012783295793,
      "learning_rate": 1.699085120117403e-05,
      "loss": 0.4653,
      "step": 2932
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.331547037339535,
      "learning_rate": 1.6988670791299274e-05,
      "loss": 0.5269,
      "step": 2933
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1525722170554802,
      "learning_rate": 1.6986489731769755e-05,
      "loss": 0.4043,
      "step": 2934
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3380771345526736,
      "learning_rate": 1.6984308022788227e-05,
      "loss": 0.4873,
      "step": 2935
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3922442233830785,
      "learning_rate": 1.698212566455749e-05,
      "loss": 0.501,
      "step": 2936
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3229800970223664,
      "learning_rate": 1.6979942657280414e-05,
      "loss": 0.5391,
      "step": 2937
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.389742316335388,
      "learning_rate": 1.6977759001159934e-05,
      "loss": 0.4863,
      "step": 2938
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2766316901301868,
      "learning_rate": 1.6975574696399033e-05,
      "loss": 0.498,
      "step": 2939
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.339984626246454,
      "learning_rate": 1.6973389743200764e-05,
      "loss": 0.5132,
      "step": 2940
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4196026398243609,
      "learning_rate": 1.6971204141768235e-05,
      "loss": 0.5571,
      "step": 2941
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3809163952535246,
      "learning_rate": 1.696901789230461e-05,
      "loss": 0.5415,
      "step": 2942
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.401687249005795,
      "learning_rate": 1.6966830995013134e-05,
      "loss": 0.499,
      "step": 2943
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.440917173134199,
      "learning_rate": 1.6964643450097077e-05,
      "loss": 0.5308,
      "step": 2944
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4415955460000824,
      "learning_rate": 1.6962455257759806e-05,
      "loss": 0.499,
      "step": 2945
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3974407376297067,
      "learning_rate": 1.696026641820472e-05,
      "loss": 0.4634,
      "step": 2946
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4635412622234152,
      "learning_rate": 1.69580769316353e-05,
      "loss": 0.519,
      "step": 2947
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3767398064299676,
      "learning_rate": 1.695588679825507e-05,
      "loss": 0.4878,
      "step": 2948
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3069822606979613,
      "learning_rate": 1.6953696018267627e-05,
      "loss": 0.4399,
      "step": 2949
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3781883314341763,
      "learning_rate": 1.6951504591876614e-05,
      "loss": 0.4985,
      "step": 2950
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2117702025129782,
      "learning_rate": 1.694931251928575e-05,
      "loss": 0.4341,
      "step": 2951
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3703026788818908,
      "learning_rate": 1.69471198006988e-05,
      "loss": 0.4727,
      "step": 2952
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4394596196010898,
      "learning_rate": 1.6944926436319596e-05,
      "loss": 0.4888,
      "step": 2953
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4263244877916328,
      "learning_rate": 1.6942732426352032e-05,
      "loss": 0.4932,
      "step": 2954
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5140602872148399,
      "learning_rate": 1.694053777100006e-05,
      "loss": 0.5532,
      "step": 2955
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5686176435220065,
      "learning_rate": 1.693834247046769e-05,
      "loss": 0.6333,
      "step": 2956
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.342913375788647,
      "learning_rate": 1.6936146524958994e-05,
      "loss": 0.4326,
      "step": 2957
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5512731708319023,
      "learning_rate": 1.6933949934678104e-05,
      "loss": 0.5464,
      "step": 2958
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.279998489081924,
      "learning_rate": 1.693175269982921e-05,
      "loss": 0.4912,
      "step": 2959
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2773647605887795,
      "learning_rate": 1.692955482061656e-05,
      "loss": 0.4941,
      "step": 2960
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3843888253706083,
      "learning_rate": 1.6927356297244474e-05,
      "loss": 0.4614,
      "step": 2961
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.303060518750089,
      "learning_rate": 1.6925157129917308e-05,
      "loss": 0.5083,
      "step": 2962
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4335587470932234,
      "learning_rate": 1.692295731883951e-05,
      "loss": 0.4653,
      "step": 2963
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4003086102691173,
      "learning_rate": 1.6920756864215558e-05,
      "loss": 0.4941,
      "step": 2964
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3329519493820852,
      "learning_rate": 1.691855576625001e-05,
      "loss": 0.5684,
      "step": 2965
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3931860855573477,
      "learning_rate": 1.6916354025147473e-05,
      "loss": 0.5396,
      "step": 2966
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2883115517286714,
      "learning_rate": 1.691415164111261e-05,
      "loss": 0.4136,
      "step": 2967
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3167177191220423,
      "learning_rate": 1.6911948614350165e-05,
      "loss": 0.6162,
      "step": 2968
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.30594974100752,
      "learning_rate": 1.6909744945064924e-05,
      "loss": 0.5122,
      "step": 2969
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2773248553896908,
      "learning_rate": 1.6907540633461728e-05,
      "loss": 0.4238,
      "step": 2970
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1823060026303112,
      "learning_rate": 1.690533567974549e-05,
      "loss": 0.4302,
      "step": 2971
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2693435055952549,
      "learning_rate": 1.6903130084121183e-05,
      "loss": 0.4326,
      "step": 2972
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3664702364816081,
      "learning_rate": 1.6900923846793832e-05,
      "loss": 0.4556,
      "step": 2973
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2183934613988598,
      "learning_rate": 1.6898716967968522e-05,
      "loss": 0.439,
      "step": 2974
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2862544886441525,
      "learning_rate": 1.689650944785041e-05,
      "loss": 0.4395,
      "step": 2975
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3549377056354144,
      "learning_rate": 1.689430128664469e-05,
      "loss": 0.4512,
      "step": 2976
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3528983940676063,
      "learning_rate": 1.6892092484556638e-05,
      "loss": 0.4692,
      "step": 2977
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3260687442737384,
      "learning_rate": 1.6889883041791578e-05,
      "loss": 0.4722,
      "step": 2978
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4082456669243857,
      "learning_rate": 1.68876729585549e-05,
      "loss": 0.6245,
      "step": 2979
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.444181379084105,
      "learning_rate": 1.6885462235052038e-05,
      "loss": 0.5381,
      "step": 2980
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1901889983030878,
      "learning_rate": 1.688325087148851e-05,
      "loss": 0.4224,
      "step": 2981
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3392915577675253,
      "learning_rate": 1.6881038868069875e-05,
      "loss": 0.4546,
      "step": 2982
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.529412139815621,
      "learning_rate": 1.6878826225001756e-05,
      "loss": 0.4785,
      "step": 2983
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2751637401359206,
      "learning_rate": 1.687661294248984e-05,
      "loss": 0.417,
      "step": 2984
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4384238645883147,
      "learning_rate": 1.6874399020739865e-05,
      "loss": 0.5186,
      "step": 2985
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3117064195508559,
      "learning_rate": 1.6872184459957637e-05,
      "loss": 0.3916,
      "step": 2986
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3057984853500884,
      "learning_rate": 1.686996926034902e-05,
      "loss": 0.4653,
      "step": 2987
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4131188882287156,
      "learning_rate": 1.6867753422119926e-05,
      "loss": 0.5366,
      "step": 2988
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.255353191766856,
      "learning_rate": 1.6865536945476348e-05,
      "loss": 0.4863,
      "step": 2989
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3293944115092722,
      "learning_rate": 1.6863319830624313e-05,
      "loss": 0.4775,
      "step": 2990
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3541665403805203,
      "learning_rate": 1.686110207776993e-05,
      "loss": 0.542,
      "step": 2991
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5380677111941714,
      "learning_rate": 1.6858883687119353e-05,
      "loss": 0.5229,
      "step": 2992
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4732058351638977,
      "learning_rate": 1.6856664658878797e-05,
      "loss": 0.5215,
      "step": 2993
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.463165898052546,
      "learning_rate": 1.6854444993254547e-05,
      "loss": 0.5151,
      "step": 2994
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4312987917180577,
      "learning_rate": 1.6852224690452937e-05,
      "loss": 0.4824,
      "step": 2995
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.376383247031167,
      "learning_rate": 1.685000375068036e-05,
      "loss": 0.5024,
      "step": 2996
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.266946014949892,
      "learning_rate": 1.6847782174143264e-05,
      "loss": 0.4399,
      "step": 2997
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.460901211780402,
      "learning_rate": 1.6845559961048178e-05,
      "loss": 0.5991,
      "step": 2998
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4138544839118707,
      "learning_rate": 1.6843337111601663e-05,
      "loss": 0.5547,
      "step": 2999
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4832382304421874,
      "learning_rate": 1.6841113626010358e-05,
      "loss": 0.5312,
      "step": 3000
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.325056678056675,
      "learning_rate": 1.683888950448095e-05,
      "loss": 0.5073,
      "step": 3001
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3266054988665168,
      "learning_rate": 1.6836664747220193e-05,
      "loss": 0.4883,
      "step": 3002
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1179471716998959,
      "learning_rate": 1.683443935443489e-05,
      "loss": 0.4067,
      "step": 3003
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3259116211804358,
      "learning_rate": 1.6832213326331918e-05,
      "loss": 0.5151,
      "step": 3004
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3747616776430744,
      "learning_rate": 1.6829986663118203e-05,
      "loss": 0.4785,
      "step": 3005
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4572144991202676,
      "learning_rate": 1.682775936500073e-05,
      "loss": 0.6152,
      "step": 3006
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4024241075243917,
      "learning_rate": 1.6825531432186545e-05,
      "loss": 0.4673,
      "step": 3007
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3107381543598853,
      "learning_rate": 1.6823302864882748e-05,
      "loss": 0.4053,
      "step": 3008
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2817584979397554,
      "learning_rate": 1.682107366329651e-05,
      "loss": 0.4795,
      "step": 3009
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.465606795744995,
      "learning_rate": 1.6818843827635052e-05,
      "loss": 0.5674,
      "step": 3010
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3542422459311987,
      "learning_rate": 1.6816613358105655e-05,
      "loss": 0.5591,
      "step": 3011
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4093942654351543,
      "learning_rate": 1.681438225491566e-05,
      "loss": 0.5249,
      "step": 3012
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2328959671112008,
      "learning_rate": 1.6812150518272465e-05,
      "loss": 0.4316,
      "step": 3013
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3704975807307929,
      "learning_rate": 1.6809918148383525e-05,
      "loss": 0.439,
      "step": 3014
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3293986959813575,
      "learning_rate": 1.680768514545637e-05,
      "loss": 0.4243,
      "step": 3015
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3653303314019087,
      "learning_rate": 1.6805451509698562e-05,
      "loss": 0.5513,
      "step": 3016
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3101111092609143,
      "learning_rate": 1.680321724131774e-05,
      "loss": 0.4707,
      "step": 3017
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3536088773730397,
      "learning_rate": 1.6800982340521605e-05,
      "loss": 0.4585,
      "step": 3018
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5657882484784045,
      "learning_rate": 1.67987468075179e-05,
      "loss": 0.6304,
      "step": 3019
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4422829413814284,
      "learning_rate": 1.679651064251444e-05,
      "loss": 0.5439,
      "step": 3020
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3897327176850738,
      "learning_rate": 1.6794273845719096e-05,
      "loss": 0.5249,
      "step": 3021
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3959531940611332,
      "learning_rate": 1.6792036417339797e-05,
      "loss": 0.4829,
      "step": 3022
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3423683403174145,
      "learning_rate": 1.6789798357584524e-05,
      "loss": 0.4253,
      "step": 3023
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.406717192712657,
      "learning_rate": 1.6787559666661336e-05,
      "loss": 0.5889,
      "step": 3024
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4710629142958027,
      "learning_rate": 1.6785320344778325e-05,
      "loss": 0.5454,
      "step": 3025
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.352356538110007,
      "learning_rate": 1.678308039214366e-05,
      "loss": 0.4414,
      "step": 3026
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4029533880053262,
      "learning_rate": 1.6780839808965566e-05,
      "loss": 0.564,
      "step": 3027
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6626209602902238,
      "learning_rate": 1.6778598595452324e-05,
      "loss": 0.5532,
      "step": 3028
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1983173400069533,
      "learning_rate": 1.6776356751812262e-05,
      "loss": 0.416,
      "step": 3029
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.459452723598835,
      "learning_rate": 1.677411427825379e-05,
      "loss": 0.5352,
      "step": 3030
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.252168418289274,
      "learning_rate": 1.677187117498536e-05,
      "loss": 0.4253,
      "step": 3031
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4235704723032256,
      "learning_rate": 1.6769627442215485e-05,
      "loss": 0.5649,
      "step": 3032
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3999706177206797,
      "learning_rate": 1.6767383080152744e-05,
      "loss": 0.4043,
      "step": 3033
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.328894421942908,
      "learning_rate": 1.6765138089005765e-05,
      "loss": 0.457,
      "step": 3034
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2875750811192082,
      "learning_rate": 1.6762892468983237e-05,
      "loss": 0.4517,
      "step": 3035
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4349347075260326,
      "learning_rate": 1.6760646220293916e-05,
      "loss": 0.5034,
      "step": 3036
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4023347600382672,
      "learning_rate": 1.6758399343146602e-05,
      "loss": 0.4258,
      "step": 3037
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2697344522677116,
      "learning_rate": 1.6756151837750167e-05,
      "loss": 0.5132,
      "step": 3038
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4080334664267558,
      "learning_rate": 1.6753903704313527e-05,
      "loss": 0.4248,
      "step": 3039
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.240661731401082,
      "learning_rate": 1.6751654943045672e-05,
      "loss": 0.4375,
      "step": 3040
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.382653187801813,
      "learning_rate": 1.674940555415564e-05,
      "loss": 0.4819,
      "step": 3041
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3293252095526886,
      "learning_rate": 1.674715553785253e-05,
      "loss": 0.5605,
      "step": 3042
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2577072651611934,
      "learning_rate": 1.6744904894345504e-05,
      "loss": 0.4932,
      "step": 3043
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.319342877234591,
      "learning_rate": 1.674265362384377e-05,
      "loss": 0.5254,
      "step": 3044
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.323232098216517,
      "learning_rate": 1.6740401726556608e-05,
      "loss": 0.4917,
      "step": 3045
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.559913459554824,
      "learning_rate": 1.6738149202693347e-05,
      "loss": 0.6196,
      "step": 3046
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.357218091424362,
      "learning_rate": 1.6735896052463384e-05,
      "loss": 0.5986,
      "step": 3047
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3758643986900052,
      "learning_rate": 1.673364227607616e-05,
      "loss": 0.4868,
      "step": 3048
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2868707994007056,
      "learning_rate": 1.673138787374119e-05,
      "loss": 0.4595,
      "step": 3049
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4004799439711086,
      "learning_rate": 1.672913284566803e-05,
      "loss": 0.5903,
      "step": 3050
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.332820838862316,
      "learning_rate": 1.672687719206631e-05,
      "loss": 0.4878,
      "step": 3051
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5239121922215526,
      "learning_rate": 1.672462091314571e-05,
      "loss": 0.5093,
      "step": 3052
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3998684131029804,
      "learning_rate": 1.672236400911597e-05,
      "loss": 0.498,
      "step": 3053
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2713335517579007,
      "learning_rate": 1.672010648018689e-05,
      "loss": 0.4043,
      "step": 3054
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3921191913667688,
      "learning_rate": 1.6717848326568327e-05,
      "loss": 0.4502,
      "step": 3055
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.238559117942201,
      "learning_rate": 1.6715589548470187e-05,
      "loss": 0.4399,
      "step": 3056
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4007102862941805,
      "learning_rate": 1.6713330146102447e-05,
      "loss": 0.5518,
      "step": 3057
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2469092143188207,
      "learning_rate": 1.6711070119675138e-05,
      "loss": 0.4453,
      "step": 3058
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.409457215392977,
      "learning_rate": 1.6708809469398347e-05,
      "loss": 0.4976,
      "step": 3059
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.284010072555825,
      "learning_rate": 1.6706548195482222e-05,
      "loss": 0.4521,
      "step": 3060
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.309582764551307,
      "learning_rate": 1.6704286298136966e-05,
      "loss": 0.5103,
      "step": 3061
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4068638203410628,
      "learning_rate": 1.670202377757284e-05,
      "loss": 0.6382,
      "step": 3062
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2533074435810034,
      "learning_rate": 1.6699760634000166e-05,
      "loss": 0.4355,
      "step": 3063
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4088065312612688,
      "learning_rate": 1.669749686762932e-05,
      "loss": 0.5884,
      "step": 3064
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.305415732336322,
      "learning_rate": 1.669523247867074e-05,
      "loss": 0.5029,
      "step": 3065
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.273969479262472,
      "learning_rate": 1.6692967467334915e-05,
      "loss": 0.4507,
      "step": 3066
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3090493143568795,
      "learning_rate": 1.6690701833832398e-05,
      "loss": 0.4551,
      "step": 3067
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4787024256420225,
      "learning_rate": 1.668843557837381e-05,
      "loss": 0.5366,
      "step": 3068
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.389759715856025,
      "learning_rate": 1.6686168701169797e-05,
      "loss": 0.5264,
      "step": 3069
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3202656210271875,
      "learning_rate": 1.66839012024311e-05,
      "loss": 0.4849,
      "step": 3070
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3124177167225144,
      "learning_rate": 1.66816330823685e-05,
      "loss": 0.4961,
      "step": 3071
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3083329347007864,
      "learning_rate": 1.667936434119283e-05,
      "loss": 0.4204,
      "step": 3072
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3297324300915916,
      "learning_rate": 1.6677094979114993e-05,
      "loss": 0.6143,
      "step": 3073
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1764702334198307,
      "learning_rate": 1.6674824996345947e-05,
      "loss": 0.4116,
      "step": 3074
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4405697744830677,
      "learning_rate": 1.66725543930967e-05,
      "loss": 0.5493,
      "step": 3075
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2491751073561066,
      "learning_rate": 1.6670283169578333e-05,
      "loss": 0.3574,
      "step": 3076
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3449596821236662,
      "learning_rate": 1.6668011326001962e-05,
      "loss": 0.4951,
      "step": 3077
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3759062822513326,
      "learning_rate": 1.6665738862578783e-05,
      "loss": 0.4346,
      "step": 3078
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2556765959631526,
      "learning_rate": 1.6663465779520042e-05,
      "loss": 0.4639,
      "step": 3079
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2599460801034064,
      "learning_rate": 1.666119207703703e-05,
      "loss": 0.4927,
      "step": 3080
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3884869735154874,
      "learning_rate": 1.665891775534112e-05,
      "loss": 0.459,
      "step": 3081
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3490360610365597,
      "learning_rate": 1.6656642814643716e-05,
      "loss": 0.457,
      "step": 3082
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.429613073260145,
      "learning_rate": 1.6654367255156303e-05,
      "loss": 0.6016,
      "step": 3083
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4676910287341232,
      "learning_rate": 1.6652091077090405e-05,
      "loss": 0.6313,
      "step": 3084
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3529173145274198,
      "learning_rate": 1.664981428065762e-05,
      "loss": 0.4624,
      "step": 3085
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2951976763983875,
      "learning_rate": 1.6647536866069587e-05,
      "loss": 0.4688,
      "step": 3086
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.622699294231002,
      "learning_rate": 1.6645258833538015e-05,
      "loss": 0.5557,
      "step": 3087
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4340251732869291,
      "learning_rate": 1.6642980183274665e-05,
      "loss": 0.543,
      "step": 3088
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3298851144304737,
      "learning_rate": 1.6640700915491354e-05,
      "loss": 0.459,
      "step": 3089
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.41147195098515,
      "learning_rate": 1.6638421030399962e-05,
      "loss": 0.4673,
      "step": 3090
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7542291073814402,
      "learning_rate": 1.6636140528212427e-05,
      "loss": 0.6069,
      "step": 3091
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.63699437642561,
      "learning_rate": 1.663385940914073e-05,
      "loss": 0.5054,
      "step": 3092
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.40399010151968,
      "learning_rate": 1.6631577673396925e-05,
      "loss": 0.5874,
      "step": 3093
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.196675920466446,
      "learning_rate": 1.6629295321193125e-05,
      "loss": 0.3813,
      "step": 3094
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.23265755184038,
      "learning_rate": 1.6627012352741482e-05,
      "loss": 0.4482,
      "step": 3095
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.372248857569726,
      "learning_rate": 1.6624728768254225e-05,
      "loss": 0.5059,
      "step": 3096
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2880841771278253,
      "learning_rate": 1.6622444567943627e-05,
      "loss": 0.4458,
      "step": 3097
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3860233327657003,
      "learning_rate": 1.662015975202203e-05,
      "loss": 0.4683,
      "step": 3098
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2825429399846855,
      "learning_rate": 1.6617874320701813e-05,
      "loss": 0.4028,
      "step": 3099
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2717059841056564,
      "learning_rate": 1.6615588274195445e-05,
      "loss": 0.4824,
      "step": 3100
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3540098322954195,
      "learning_rate": 1.661330161271542e-05,
      "loss": 0.4609,
      "step": 3101
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.230599532267572,
      "learning_rate": 1.6611014336474303e-05,
      "loss": 0.458,
      "step": 3102
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4113349327535463,
      "learning_rate": 1.6608726445684715e-05,
      "loss": 0.4795,
      "step": 3103
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2496738110069312,
      "learning_rate": 1.6606437940559342e-05,
      "loss": 0.4248,
      "step": 3104
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3451059203605857,
      "learning_rate": 1.6604148821310912e-05,
      "loss": 0.5146,
      "step": 3105
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3405469874908145,
      "learning_rate": 1.6601859088152215e-05,
      "loss": 0.5132,
      "step": 3106
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3645543228899595,
      "learning_rate": 1.6599568741296112e-05,
      "loss": 0.5781,
      "step": 3107
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5405407298285227,
      "learning_rate": 1.6597277780955502e-05,
      "loss": 0.5859,
      "step": 3108
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1601181441363282,
      "learning_rate": 1.6594986207343343e-05,
      "loss": 0.4038,
      "step": 3109
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2090584498811818,
      "learning_rate": 1.6592694020672667e-05,
      "loss": 0.4399,
      "step": 3110
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2359650556973096,
      "learning_rate": 1.659040122115655e-05,
      "loss": 0.4932,
      "step": 3111
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3551299184759567,
      "learning_rate": 1.658810780900812e-05,
      "loss": 0.5229,
      "step": 3112
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.407730116492141,
      "learning_rate": 1.6585813784440575e-05,
      "loss": 0.5781,
      "step": 3113
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5121670330217905,
      "learning_rate": 1.6583519147667157e-05,
      "loss": 0.5273,
      "step": 3114
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3318201934994693,
      "learning_rate": 1.658122389890118e-05,
      "loss": 0.4858,
      "step": 3115
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6552478009544278,
      "learning_rate": 1.6578928038355998e-05,
      "loss": 0.6006,
      "step": 3116
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2678013909459604,
      "learning_rate": 1.6576631566245037e-05,
      "loss": 0.4663,
      "step": 3117
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4852040451479438,
      "learning_rate": 1.6574334482781768e-05,
      "loss": 0.5918,
      "step": 3118
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.371431893803585,
      "learning_rate": 1.6572036788179728e-05,
      "loss": 0.5806,
      "step": 3119
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3415319327358082,
      "learning_rate": 1.65697384826525e-05,
      "loss": 0.5435,
      "step": 3120
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3850222900931433,
      "learning_rate": 1.6567439566413737e-05,
      "loss": 0.4766,
      "step": 3121
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3884856366969902,
      "learning_rate": 1.6565140039677142e-05,
      "loss": 0.5278,
      "step": 3122
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.410664643400656,
      "learning_rate": 1.6562839902656476e-05,
      "loss": 0.5586,
      "step": 3123
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.391954240700714,
      "learning_rate": 1.656053915556555e-05,
      "loss": 0.501,
      "step": 3124
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3501785721872315,
      "learning_rate": 1.6558237798618243e-05,
      "loss": 0.5835,
      "step": 3125
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2603390261729603,
      "learning_rate": 1.655593583202848e-05,
      "loss": 0.5156,
      "step": 3126
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.279375220942681,
      "learning_rate": 1.6553633256010254e-05,
      "loss": 0.4282,
      "step": 3127
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.335178312618041,
      "learning_rate": 1.6551330070777603e-05,
      "loss": 0.4863,
      "step": 3128
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3760004889319735,
      "learning_rate": 1.6549026276544627e-05,
      "loss": 0.4839,
      "step": 3129
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2769803105017576,
      "learning_rate": 1.6546721873525488e-05,
      "loss": 0.4536,
      "step": 3130
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3861515115617438,
      "learning_rate": 1.65444168619344e-05,
      "loss": 0.4507,
      "step": 3131
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3410883204418642,
      "learning_rate": 1.6542111241985623e-05,
      "loss": 0.4717,
      "step": 3132
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2895879886651669,
      "learning_rate": 1.6539805013893493e-05,
      "loss": 0.5273,
      "step": 3133
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.375970490563541,
      "learning_rate": 1.653749817787239e-05,
      "loss": 0.4595,
      "step": 3134
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2290888902144517,
      "learning_rate": 1.653519073413675e-05,
      "loss": 0.4512,
      "step": 3135
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.296190783373663,
      "learning_rate": 1.6532882682901076e-05,
      "loss": 0.3931,
      "step": 3136
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3387707401028666,
      "learning_rate": 1.6530574024379915e-05,
      "loss": 0.5171,
      "step": 3137
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6203405661097,
      "learning_rate": 1.6528264758787876e-05,
      "loss": 0.541,
      "step": 3138
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6312723300466776,
      "learning_rate": 1.652595488633963e-05,
      "loss": 0.5435,
      "step": 3139
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4119597724198214,
      "learning_rate": 1.6523644407249893e-05,
      "loss": 0.4443,
      "step": 3140
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1619581164059771,
      "learning_rate": 1.652133332173344e-05,
      "loss": 0.3843,
      "step": 3141
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3666846604152492,
      "learning_rate": 1.6519021630005115e-05,
      "loss": 0.5137,
      "step": 3142
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3367628671805474,
      "learning_rate": 1.6516709332279806e-05,
      "loss": 0.5171,
      "step": 3143
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2259190549958403,
      "learning_rate": 1.6514396428772457e-05,
      "loss": 0.5029,
      "step": 3144
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2651166690834321,
      "learning_rate": 1.6512082919698072e-05,
      "loss": 0.4658,
      "step": 3145
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5462459910391642,
      "learning_rate": 1.650976880527171e-05,
      "loss": 0.5913,
      "step": 3146
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5629950761525535,
      "learning_rate": 1.650745408570849e-05,
      "loss": 0.4473,
      "step": 3147
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.323887027262089,
      "learning_rate": 1.6505138761223586e-05,
      "loss": 0.4956,
      "step": 3148
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4574011007412744,
      "learning_rate": 1.650282283203222e-05,
      "loss": 0.5234,
      "step": 3149
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.428669305392533,
      "learning_rate": 1.6500506298349682e-05,
      "loss": 0.5254,
      "step": 3150
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3736390024619347,
      "learning_rate": 1.649818916039131e-05,
      "loss": 0.5742,
      "step": 3151
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3830393929314355,
      "learning_rate": 1.6495871418372503e-05,
      "loss": 0.457,
      "step": 3152
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.128954891480001,
      "learning_rate": 1.6493553072508716e-05,
      "loss": 0.4507,
      "step": 3153
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1427011063286479,
      "learning_rate": 1.6491234123015454e-05,
      "loss": 0.4175,
      "step": 3154
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3094027011273899,
      "learning_rate": 1.6488914570108287e-05,
      "loss": 0.4658,
      "step": 3155
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3502263608130267,
      "learning_rate": 1.6486594414002836e-05,
      "loss": 0.4736,
      "step": 3156
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3419046918233197,
      "learning_rate": 1.6484273654914772e-05,
      "loss": 0.5693,
      "step": 3157
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3277109554694686,
      "learning_rate": 1.6481952293059835e-05,
      "loss": 0.4219,
      "step": 3158
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3206122933653914,
      "learning_rate": 1.6479630328653814e-05,
      "loss": 0.4688,
      "step": 3159
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.234438374976514,
      "learning_rate": 1.6477307761912555e-05,
      "loss": 0.4634,
      "step": 3160
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.427570266704197,
      "learning_rate": 1.6474984593051965e-05,
      "loss": 0.5527,
      "step": 3161
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.441855810849257,
      "learning_rate": 1.6472660822287987e-05,
      "loss": 0.5796,
      "step": 3162
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3399644483836708,
      "learning_rate": 1.647033644983665e-05,
      "loss": 0.5205,
      "step": 3163
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3681991508255693,
      "learning_rate": 1.6468011475914015e-05,
      "loss": 0.5195,
      "step": 3164
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2792261961699285,
      "learning_rate": 1.6465685900736204e-05,
      "loss": 0.4448,
      "step": 3165
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3996794696400934,
      "learning_rate": 1.6463359724519413e-05,
      "loss": 0.4844,
      "step": 3166
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.394605048827088,
      "learning_rate": 1.646103294747987e-05,
      "loss": 0.4741,
      "step": 3167
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2428574314884075,
      "learning_rate": 1.6458705569833866e-05,
      "loss": 0.4375,
      "step": 3168
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.378219366405179,
      "learning_rate": 1.6456377591797754e-05,
      "loss": 0.5249,
      "step": 3169
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2899100273258741,
      "learning_rate": 1.645404901358794e-05,
      "loss": 0.5146,
      "step": 3170
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2806389786009775,
      "learning_rate": 1.645171983542088e-05,
      "loss": 0.4946,
      "step": 3171
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1980607520165285,
      "learning_rate": 1.644939005751309e-05,
      "loss": 0.4478,
      "step": 3172
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2442977743430574,
      "learning_rate": 1.644705968008115e-05,
      "loss": 0.4468,
      "step": 3173
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2969926260558238,
      "learning_rate": 1.644472870334168e-05,
      "loss": 0.3984,
      "step": 3174
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4067829769952451,
      "learning_rate": 1.6442397127511366e-05,
      "loss": 0.5244,
      "step": 3175
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5480420399510983,
      "learning_rate": 1.644006495280695e-05,
      "loss": 0.5513,
      "step": 3176
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1624446947679636,
      "learning_rate": 1.6437732179445222e-05,
      "loss": 0.438,
      "step": 3177
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3607933347347896,
      "learning_rate": 1.6435398807643035e-05,
      "loss": 0.5615,
      "step": 3178
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3232977254355742,
      "learning_rate": 1.6433064837617294e-05,
      "loss": 0.4878,
      "step": 3179
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3033693176585446,
      "learning_rate": 1.6430730269584963e-05,
      "loss": 0.4736,
      "step": 3180
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3433880014222186,
      "learning_rate": 1.642839510376306e-05,
      "loss": 0.4292,
      "step": 3181
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3218718290526084,
      "learning_rate": 1.6426059340368653e-05,
      "loss": 0.4751,
      "step": 3182
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3808538538566877,
      "learning_rate": 1.6423722979618883e-05,
      "loss": 0.4956,
      "step": 3183
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3567914628662179,
      "learning_rate": 1.6421386021730915e-05,
      "loss": 0.521,
      "step": 3184
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.588096316782127,
      "learning_rate": 1.6419048466922004e-05,
      "loss": 0.6953,
      "step": 3185
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4273777694144598,
      "learning_rate": 1.6416710315409437e-05,
      "loss": 0.4355,
      "step": 3186
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.395145072190652,
      "learning_rate": 1.641437156741057e-05,
      "loss": 0.5674,
      "step": 3187
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.319680920160828,
      "learning_rate": 1.6412032223142806e-05,
      "loss": 0.4883,
      "step": 3188
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3276768682239641,
      "learning_rate": 1.6409692282823604e-05,
      "loss": 0.4541,
      "step": 3189
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2012043360292608,
      "learning_rate": 1.6407351746670484e-05,
      "loss": 0.4336,
      "step": 3190
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2807039266653317,
      "learning_rate": 1.6405010614901017e-05,
      "loss": 0.4297,
      "step": 3191
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3416694689021336,
      "learning_rate": 1.6402668887732833e-05,
      "loss": 0.4912,
      "step": 3192
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4407170349219054,
      "learning_rate": 1.6400326565383614e-05,
      "loss": 0.4639,
      "step": 3193
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3066409126920908,
      "learning_rate": 1.6397983648071093e-05,
      "loss": 0.4883,
      "step": 3194
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.38811309080013,
      "learning_rate": 1.6395640136013073e-05,
      "loss": 0.4854,
      "step": 3195
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3723529107810635,
      "learning_rate": 1.6393296029427395e-05,
      "loss": 0.459,
      "step": 3196
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3959209891667372,
      "learning_rate": 1.6390951328531966e-05,
      "loss": 0.6025,
      "step": 3197
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2570200616317353,
      "learning_rate": 1.6388606033544745e-05,
      "loss": 0.439,
      "step": 3198
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3246713545529,
      "learning_rate": 1.6386260144683744e-05,
      "loss": 0.5322,
      "step": 3199
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2263764090951328,
      "learning_rate": 1.638391366216704e-05,
      "loss": 0.3545,
      "step": 3200
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4326310030444132,
      "learning_rate": 1.6381566586212752e-05,
      "loss": 0.4521,
      "step": 3201
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2666521785942098,
      "learning_rate": 1.637921891703906e-05,
      "loss": 0.4648,
      "step": 3202
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.642032257154202,
      "learning_rate": 1.63768706548642e-05,
      "loss": 0.6084,
      "step": 3203
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3515669344209629,
      "learning_rate": 1.6374521799906468e-05,
      "loss": 0.4185,
      "step": 3204
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4761839691066845,
      "learning_rate": 1.63721723523842e-05,
      "loss": 0.6279,
      "step": 3205
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3374697965104885,
      "learning_rate": 1.6369822312515805e-05,
      "loss": 0.4927,
      "step": 3206
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.327773937730122,
      "learning_rate": 1.6367471680519734e-05,
      "loss": 0.4243,
      "step": 3207
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4324756873678322,
      "learning_rate": 1.6365120456614498e-05,
      "loss": 0.6104,
      "step": 3208
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3062405942256923,
      "learning_rate": 1.6362768641018662e-05,
      "loss": 0.4429,
      "step": 3209
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.419403464196248,
      "learning_rate": 1.636041623395085e-05,
      "loss": 0.4751,
      "step": 3210
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2793256213861752,
      "learning_rate": 1.6358063235629733e-05,
      "loss": 0.4648,
      "step": 3211
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.239989360014822,
      "learning_rate": 1.6355709646274048e-05,
      "loss": 0.4619,
      "step": 3212
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3590530531547897,
      "learning_rate": 1.6353355466102575e-05,
      "loss": 0.4795,
      "step": 3213
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.340022825813162,
      "learning_rate": 1.6351000695334157e-05,
      "loss": 0.4355,
      "step": 3214
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.205241901873085,
      "learning_rate": 1.6348645334187686e-05,
      "loss": 0.4004,
      "step": 3215
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2815713571170364,
      "learning_rate": 1.6346289382882117e-05,
      "loss": 0.3799,
      "step": 3216
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3214654310897207,
      "learning_rate": 1.6343932841636455e-05,
      "loss": 0.4644,
      "step": 3217
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3158208601092805,
      "learning_rate": 1.6341575710669758e-05,
      "loss": 0.4531,
      "step": 3218
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3850760684512617,
      "learning_rate": 1.633921799020114e-05,
      "loss": 0.5186,
      "step": 3219
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5179287314895118,
      "learning_rate": 1.6336859680449773e-05,
      "loss": 0.6523,
      "step": 3220
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3058469815452836,
      "learning_rate": 1.633450078163488e-05,
      "loss": 0.5352,
      "step": 3221
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5464335431793546,
      "learning_rate": 1.6332141293975742e-05,
      "loss": 0.5483,
      "step": 3222
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4166721488493383,
      "learning_rate": 1.632978121769169e-05,
      "loss": 0.519,
      "step": 3223
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3105083926021388,
      "learning_rate": 1.6327420553002113e-05,
      "loss": 0.5552,
      "step": 3224
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3867245916473057,
      "learning_rate": 1.632505930012646e-05,
      "loss": 0.4663,
      "step": 3225
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5102093746348741,
      "learning_rate": 1.632269745928422e-05,
      "loss": 0.4736,
      "step": 3226
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3035399582398566,
      "learning_rate": 1.632033503069495e-05,
      "loss": 0.4434,
      "step": 3227
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3544896336351513,
      "learning_rate": 1.6317972014578252e-05,
      "loss": 0.4941,
      "step": 3228
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3504149697918713,
      "learning_rate": 1.63156084111538e-05,
      "loss": 0.5088,
      "step": 3229
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3560624456135284,
      "learning_rate": 1.6313244220641304e-05,
      "loss": 0.4902,
      "step": 3230
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4571076442777227,
      "learning_rate": 1.631087944326053e-05,
      "loss": 0.5454,
      "step": 3231
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3473469332793142,
      "learning_rate": 1.630851407923131e-05,
      "loss": 0.5991,
      "step": 3232
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.312071204886614,
      "learning_rate": 1.6306148128773522e-05,
      "loss": 0.5225,
      "step": 3233
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2965345472411698,
      "learning_rate": 1.6303781592107102e-05,
      "loss": 0.4648,
      "step": 3234
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4156137403061935,
      "learning_rate": 1.6301414469452037e-05,
      "loss": 0.521,
      "step": 3235
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.260898261596813,
      "learning_rate": 1.6299046761028373e-05,
      "loss": 0.4062,
      "step": 3236
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.410108445776272,
      "learning_rate": 1.629667846705621e-05,
      "loss": 0.5674,
      "step": 3237
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2484706127918686,
      "learning_rate": 1.6294309587755693e-05,
      "loss": 0.5093,
      "step": 3238
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3042891294937087,
      "learning_rate": 1.6291940123347033e-05,
      "loss": 0.4722,
      "step": 3239
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3815172784921794,
      "learning_rate": 1.6289570074050492e-05,
      "loss": 0.5176,
      "step": 3240
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.445142184709925,
      "learning_rate": 1.628719944008639e-05,
      "loss": 0.5166,
      "step": 3241
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.460734161962173,
      "learning_rate": 1.6284828221675085e-05,
      "loss": 0.5498,
      "step": 3242
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.120494836991512,
      "learning_rate": 1.6282456419037013e-05,
      "loss": 0.4712,
      "step": 3243
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.317633837799979,
      "learning_rate": 1.628008403239265e-05,
      "loss": 0.5024,
      "step": 3244
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3594739806785916,
      "learning_rate": 1.6277711061962525e-05,
      "loss": 0.4521,
      "step": 3245
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2465064509243597,
      "learning_rate": 1.6275337507967228e-05,
      "loss": 0.4551,
      "step": 3246
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2990880492356613,
      "learning_rate": 1.6272963370627398e-05,
      "loss": 0.4224,
      "step": 3247
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3755726325727815,
      "learning_rate": 1.6270588650163737e-05,
      "loss": 0.5576,
      "step": 3248
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3171691764896238,
      "learning_rate": 1.626821334679699e-05,
      "loss": 0.4775,
      "step": 3249
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.42579142616148,
      "learning_rate": 1.626583746074796e-05,
      "loss": 0.542,
      "step": 3250
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3238390853687563,
      "learning_rate": 1.6263460992237507e-05,
      "loss": 0.4995,
      "step": 3251
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2620982548206425,
      "learning_rate": 1.6261083941486543e-05,
      "loss": 0.4146,
      "step": 3252
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5375085386025726,
      "learning_rate": 1.6258706308716035e-05,
      "loss": 0.5034,
      "step": 3253
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4908647521306502,
      "learning_rate": 1.6256328094147003e-05,
      "loss": 0.583,
      "step": 3254
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3845979224278475,
      "learning_rate": 1.6253949298000527e-05,
      "loss": 0.5215,
      "step": 3255
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4822001455899485,
      "learning_rate": 1.6251569920497725e-05,
      "loss": 0.5303,
      "step": 3256
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5001365608432464,
      "learning_rate": 1.624918996185979e-05,
      "loss": 0.5981,
      "step": 3257
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2089882996929024,
      "learning_rate": 1.6246809422307954e-05,
      "loss": 0.4087,
      "step": 3258
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.465847728976822,
      "learning_rate": 1.6244428302063506e-05,
      "loss": 0.5225,
      "step": 3259
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4643745566454507,
      "learning_rate": 1.6242046601347796e-05,
      "loss": 0.5938,
      "step": 3260
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.384919322780448,
      "learning_rate": 1.623966432038222e-05,
      "loss": 0.4526,
      "step": 3261
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4092071218463407,
      "learning_rate": 1.6237281459388233e-05,
      "loss": 0.5173,
      "step": 3262
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3488971756399828,
      "learning_rate": 1.6234898018587336e-05,
      "loss": 0.4854,
      "step": 3263
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3619468931799914,
      "learning_rate": 1.6232513998201094e-05,
      "loss": 0.4194,
      "step": 3264
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3271393910601326,
      "learning_rate": 1.6230129398451124e-05,
      "loss": 0.4331,
      "step": 3265
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.321403582048089,
      "learning_rate": 1.6227744219559086e-05,
      "loss": 0.4629,
      "step": 3266
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3280483022665392,
      "learning_rate": 1.622535846174671e-05,
      "loss": 0.502,
      "step": 3267
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.334480351306957,
      "learning_rate": 1.6222972125235766e-05,
      "loss": 0.479,
      "step": 3268
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.121624460693256,
      "learning_rate": 1.6220585210248093e-05,
      "loss": 0.4106,
      "step": 3269
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0820798497748891,
      "learning_rate": 1.6218197717005562e-05,
      "loss": 0.3396,
      "step": 3270
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3999223878045244,
      "learning_rate": 1.6215809645730115e-05,
      "loss": 0.522,
      "step": 3271
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.42320264532421,
      "learning_rate": 1.621342099664375e-05,
      "loss": 0.5835,
      "step": 3272
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4913276875044534,
      "learning_rate": 1.6211031769968503e-05,
      "loss": 0.5244,
      "step": 3273
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2491415290922028,
      "learning_rate": 1.6208641965926474e-05,
      "loss": 0.4453,
      "step": 3274
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2936570045339784,
      "learning_rate": 1.6206251584739817e-05,
      "loss": 0.4229,
      "step": 3275
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4038168346309712,
      "learning_rate": 1.620386062663074e-05,
      "loss": 0.564,
      "step": 3276
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3222787058575187,
      "learning_rate": 1.6201469091821498e-05,
      "loss": 0.4731,
      "step": 3277
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4716745537694393,
      "learning_rate": 1.6199076980534406e-05,
      "loss": 0.5947,
      "step": 3278
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3893472870448456,
      "learning_rate": 1.6196684292991827e-05,
      "loss": 0.5454,
      "step": 3279
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3917817164044874,
      "learning_rate": 1.6194291029416188e-05,
      "loss": 0.4873,
      "step": 3280
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3041659697012284,
      "learning_rate": 1.6191897190029956e-05,
      "loss": 0.4546,
      "step": 3281
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2742414526689236,
      "learning_rate": 1.6189502775055662e-05,
      "loss": 0.4692,
      "step": 3282
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.335914842963182,
      "learning_rate": 1.618710778471588e-05,
      "loss": 0.479,
      "step": 3283
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5182013500346356,
      "learning_rate": 1.618471221923326e-05,
      "loss": 0.5928,
      "step": 3284
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.528589686196238,
      "learning_rate": 1.6182316078830473e-05,
      "loss": 0.5146,
      "step": 3285
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4993226989028419,
      "learning_rate": 1.617991936373027e-05,
      "loss": 0.5054,
      "step": 3286
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2193186271571361,
      "learning_rate": 1.6177522074155436e-05,
      "loss": 0.4438,
      "step": 3287
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.194353195784632,
      "learning_rate": 1.617512421032883e-05,
      "loss": 0.4277,
      "step": 3288
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4170283930452356,
      "learning_rate": 1.6172725772473343e-05,
      "loss": 0.5015,
      "step": 3289
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4689646428272307,
      "learning_rate": 1.617032676081194e-05,
      "loss": 0.5371,
      "step": 3290
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.436744896921848,
      "learning_rate": 1.616792717556762e-05,
      "loss": 0.502,
      "step": 3291
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4458586927685293,
      "learning_rate": 1.616552701696345e-05,
      "loss": 0.4946,
      "step": 3292
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4054702121295273,
      "learning_rate": 1.616312628522254e-05,
      "loss": 0.5347,
      "step": 3293
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6476905979254846,
      "learning_rate": 1.6160724980568066e-05,
      "loss": 0.625,
      "step": 3294
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5515755019846946,
      "learning_rate": 1.615832310322324e-05,
      "loss": 0.6396,
      "step": 3295
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2449753980647642,
      "learning_rate": 1.615592065341134e-05,
      "loss": 0.4233,
      "step": 3296
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3315010685289095,
      "learning_rate": 1.6153517631355696e-05,
      "loss": 0.46,
      "step": 3297
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.297041347964648,
      "learning_rate": 1.6151114037279682e-05,
      "loss": 0.4751,
      "step": 3298
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2970048698815004,
      "learning_rate": 1.614870987140674e-05,
      "loss": 0.4912,
      "step": 3299
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2234669435775798,
      "learning_rate": 1.614630513396035e-05,
      "loss": 0.501,
      "step": 3300
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.385493830046165,
      "learning_rate": 1.6143899825164058e-05,
      "loss": 0.5649,
      "step": 3301
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3114451099598707,
      "learning_rate": 1.6141493945241453e-05,
      "loss": 0.5044,
      "step": 3302
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4962453663785023,
      "learning_rate": 1.6139087494416184e-05,
      "loss": 0.6196,
      "step": 3303
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0595843045119344,
      "learning_rate": 1.613668047291195e-05,
      "loss": 0.3972,
      "step": 3304
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3981199773246762,
      "learning_rate": 1.6134272880952506e-05,
      "loss": 0.606,
      "step": 3305
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2489017481726874,
      "learning_rate": 1.613186471876165e-05,
      "loss": 0.4448,
      "step": 3306
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.41289454920694,
      "learning_rate": 1.612945598656325e-05,
      "loss": 0.5562,
      "step": 3307
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2963284416219283,
      "learning_rate": 1.6127046684581212e-05,
      "loss": 0.5269,
      "step": 3308
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.385610960583107,
      "learning_rate": 1.6124636813039502e-05,
      "loss": 0.457,
      "step": 3309
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2639494689269555,
      "learning_rate": 1.6122226372162137e-05,
      "loss": 0.4761,
      "step": 3310
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4813218996350261,
      "learning_rate": 1.6119815362173188e-05,
      "loss": 0.5522,
      "step": 3311
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2370370996051867,
      "learning_rate": 1.6117403783296778e-05,
      "loss": 0.4619,
      "step": 3312
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2621435881041116,
      "learning_rate": 1.6114991635757085e-05,
      "loss": 0.4771,
      "step": 3313
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4140313901504307,
      "learning_rate": 1.611257891977833e-05,
      "loss": 0.4819,
      "step": 3314
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3071740569435957,
      "learning_rate": 1.6110165635584807e-05,
      "loss": 0.5801,
      "step": 3315
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.311299008207973,
      "learning_rate": 1.6107751783400845e-05,
      "loss": 0.4282,
      "step": 3316
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2473232006686528,
      "learning_rate": 1.610533736345083e-05,
      "loss": 0.4541,
      "step": 3317
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1739506610164847,
      "learning_rate": 1.6102922375959204e-05,
      "loss": 0.4717,
      "step": 3318
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2105787697281332,
      "learning_rate": 1.6100506821150455e-05,
      "loss": 0.4634,
      "step": 3319
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3659795105334367,
      "learning_rate": 1.6098090699249144e-05,
      "loss": 0.4956,
      "step": 3320
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5189189556321967,
      "learning_rate": 1.609567401047985e-05,
      "loss": 0.5522,
      "step": 3321
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3149473283351947,
      "learning_rate": 1.6093256755067236e-05,
      "loss": 0.4922,
      "step": 3322
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2562857303521722,
      "learning_rate": 1.6090838933236004e-05,
      "loss": 0.4961,
      "step": 3323
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2718527331655562,
      "learning_rate": 1.608842054521091e-05,
      "loss": 0.4556,
      "step": 3324
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4572007910338651,
      "learning_rate": 1.6086001591216764e-05,
      "loss": 0.563,
      "step": 3325
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3096370118122116,
      "learning_rate": 1.6083582071478424e-05,
      "loss": 0.4775,
      "step": 3326
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.357241916944019,
      "learning_rate": 1.6081161986220807e-05,
      "loss": 0.5815,
      "step": 3327
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4018580268727006,
      "learning_rate": 1.6078741335668882e-05,
      "loss": 0.4966,
      "step": 3328
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.229330204620163,
      "learning_rate": 1.6076320120047667e-05,
      "loss": 0.5195,
      "step": 3329
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2968601189706812,
      "learning_rate": 1.607389833958223e-05,
      "loss": 0.4297,
      "step": 3330
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3022609649911592,
      "learning_rate": 1.6071475994497702e-05,
      "loss": 0.4751,
      "step": 3331
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.347568903619191,
      "learning_rate": 1.6069053085019258e-05,
      "loss": 0.4072,
      "step": 3332
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.225537120183506,
      "learning_rate": 1.6066629611372127e-05,
      "loss": 0.4233,
      "step": 3333
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4716826519038102,
      "learning_rate": 1.6064205573781587e-05,
      "loss": 0.5078,
      "step": 3334
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3428523160017114,
      "learning_rate": 1.6061780972472978e-05,
      "loss": 0.5894,
      "step": 3335
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3448020932673206,
      "learning_rate": 1.6059355807671683e-05,
      "loss": 0.4419,
      "step": 3336
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5356818531400356,
      "learning_rate": 1.6056930079603144e-05,
      "loss": 0.5239,
      "step": 3337
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3915153336361188,
      "learning_rate": 1.6054503788492852e-05,
      "loss": 0.4971,
      "step": 3338
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.291773445074028,
      "learning_rate": 1.605207693456635e-05,
      "loss": 0.5703,
      "step": 3339
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2756786465710235,
      "learning_rate": 1.6049649518049234e-05,
      "loss": 0.4146,
      "step": 3340
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3095593762896791,
      "learning_rate": 1.6047221539167152e-05,
      "loss": 0.4858,
      "step": 3341
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2183753067516903,
      "learning_rate": 1.6044792998145804e-05,
      "loss": 0.4731,
      "step": 3342
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3468100457310583,
      "learning_rate": 1.6042363895210948e-05,
      "loss": 0.4517,
      "step": 3343
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1885087766860594,
      "learning_rate": 1.6039934230588384e-05,
      "loss": 0.5181,
      "step": 3344
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.421274441051791,
      "learning_rate": 1.6037504004503967e-05,
      "loss": 0.4819,
      "step": 3345
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.172794665953639,
      "learning_rate": 1.6035073217183613e-05,
      "loss": 0.4058,
      "step": 3346
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2747896162583592,
      "learning_rate": 1.6032641868853283e-05,
      "loss": 0.5244,
      "step": 3347
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3656773353903704,
      "learning_rate": 1.6030209959738988e-05,
      "loss": 0.4893,
      "step": 3348
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3173689252888638,
      "learning_rate": 1.6027777490066798e-05,
      "loss": 0.4756,
      "step": 3349
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3901927363182454,
      "learning_rate": 1.6025344460062826e-05,
      "loss": 0.4761,
      "step": 3350
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2555397749682988,
      "learning_rate": 1.6022910869953245e-05,
      "loss": 0.417,
      "step": 3351
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.220009085989529,
      "learning_rate": 1.602047671996428e-05,
      "loss": 0.4595,
      "step": 3352
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3727107547857669,
      "learning_rate": 1.6018042010322197e-05,
      "loss": 0.4219,
      "step": 3353
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2606892880492186,
      "learning_rate": 1.6015606741253334e-05,
      "loss": 0.4619,
      "step": 3354
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3205776036545274,
      "learning_rate": 1.601317091298406e-05,
      "loss": 0.4624,
      "step": 3355
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4244025601635162,
      "learning_rate": 1.601073452574081e-05,
      "loss": 0.4492,
      "step": 3356
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.311849641243136,
      "learning_rate": 1.6008297579750063e-05,
      "loss": 0.416,
      "step": 3357
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3858950373796672,
      "learning_rate": 1.6005860075238358e-05,
      "loss": 0.4556,
      "step": 3358
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2562838825016804,
      "learning_rate": 1.6003422012432275e-05,
      "loss": 0.4536,
      "step": 3359
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4566290971444347,
      "learning_rate": 1.6000983391558457e-05,
      "loss": 0.4932,
      "step": 3360
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4473051671975836,
      "learning_rate": 1.5998544212843597e-05,
      "loss": 0.4971,
      "step": 3361
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4058177805015581,
      "learning_rate": 1.5996104476514426e-05,
      "loss": 0.4556,
      "step": 3362
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.33745699581217,
      "learning_rate": 1.5993664182797747e-05,
      "loss": 0.4268,
      "step": 3363
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1840501865206203,
      "learning_rate": 1.59912233319204e-05,
      "loss": 0.4971,
      "step": 3364
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5505613545130785,
      "learning_rate": 1.5988781924109293e-05,
      "loss": 0.5361,
      "step": 3365
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3082256609422636,
      "learning_rate": 1.598633995959136e-05,
      "loss": 0.4136,
      "step": 3366
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3567357876938013,
      "learning_rate": 1.5983897438593612e-05,
      "loss": 0.5049,
      "step": 3367
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.14537455585065,
      "learning_rate": 1.5981454361343097e-05,
      "loss": 0.4019,
      "step": 3368
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.425390926833432,
      "learning_rate": 1.597901072806692e-05,
      "loss": 0.5352,
      "step": 3369
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5068676978204434,
      "learning_rate": 1.5976566538992237e-05,
      "loss": 0.5708,
      "step": 3370
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4198595830006944,
      "learning_rate": 1.597412179434626e-05,
      "loss": 0.4761,
      "step": 3371
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3386332913409973,
      "learning_rate": 1.5971676494356237e-05,
      "loss": 0.5122,
      "step": 3372
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3514968667985796,
      "learning_rate": 1.5969230639249492e-05,
      "loss": 0.4136,
      "step": 3373
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.323165653058559,
      "learning_rate": 1.596678422925338e-05,
      "loss": 0.5557,
      "step": 3374
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.353978496742663,
      "learning_rate": 1.596433726459532e-05,
      "loss": 0.4761,
      "step": 3375
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3692854722782335,
      "learning_rate": 1.5961889745502767e-05,
      "loss": 0.5459,
      "step": 3376
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3720466287323423,
      "learning_rate": 1.5959441672203254e-05,
      "loss": 0.5166,
      "step": 3377
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.196551812601061,
      "learning_rate": 1.5956993044924334e-05,
      "loss": 0.4414,
      "step": 3378
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3970520052206217,
      "learning_rate": 1.5954543863893638e-05,
      "loss": 0.584,
      "step": 3379
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3970180501119096,
      "learning_rate": 1.5952094129338834e-05,
      "loss": 0.4526,
      "step": 3380
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3988117706778938,
      "learning_rate": 1.594964384148764e-05,
      "loss": 0.5625,
      "step": 3381
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4228420831393556,
      "learning_rate": 1.5947193000567844e-05,
      "loss": 0.563,
      "step": 3382
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4173534861445403,
      "learning_rate": 1.5944741606807257e-05,
      "loss": 0.5244,
      "step": 3383
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2939946318893394,
      "learning_rate": 1.5942289660433766e-05,
      "loss": 0.4756,
      "step": 3384
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2432838781781463,
      "learning_rate": 1.5939837161675297e-05,
      "loss": 0.4058,
      "step": 3385
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2183118118730156,
      "learning_rate": 1.5937384110759824e-05,
      "loss": 0.4014,
      "step": 3386
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2813850173950119,
      "learning_rate": 1.5934930507915386e-05,
      "loss": 0.5308,
      "step": 3387
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3723659681374976,
      "learning_rate": 1.5932476353370068e-05,
      "loss": 0.4556,
      "step": 3388
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2807419781642386,
      "learning_rate": 1.5930021647351997e-05,
      "loss": 0.4844,
      "step": 3389
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2434724511109452,
      "learning_rate": 1.5927566390089362e-05,
      "loss": 0.4639,
      "step": 3390
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.30486894484889,
      "learning_rate": 1.5925110581810396e-05,
      "loss": 0.5264,
      "step": 3391
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.366212871490849,
      "learning_rate": 1.592265422274339e-05,
      "loss": 0.5068,
      "step": 3392
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2816630491231245,
      "learning_rate": 1.5920197313116682e-05,
      "loss": 0.4893,
      "step": 3393
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2839262628506865,
      "learning_rate": 1.591773985315866e-05,
      "loss": 0.4922,
      "step": 3394
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5093234725453697,
      "learning_rate": 1.591528184309777e-05,
      "loss": 0.5894,
      "step": 3395
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3041088733524993,
      "learning_rate": 1.59128232831625e-05,
      "loss": 0.5171,
      "step": 3396
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2629720220197695,
      "learning_rate": 1.5910364173581395e-05,
      "loss": 0.4355,
      "step": 3397
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.428460656020844,
      "learning_rate": 1.5907904514583047e-05,
      "loss": 0.562,
      "step": 3398
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.348239446745653,
      "learning_rate": 1.590544430639611e-05,
      "loss": 0.5435,
      "step": 3399
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3812477202873268,
      "learning_rate": 1.5902983549249272e-05,
      "loss": 0.5498,
      "step": 3400
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2500967910519811,
      "learning_rate": 1.5900522243371283e-05,
      "loss": 0.5239,
      "step": 3401
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4220880389348665,
      "learning_rate": 1.589806038899094e-05,
      "loss": 0.4839,
      "step": 3402
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2574668117266823,
      "learning_rate": 1.58955979863371e-05,
      "loss": 0.4648,
      "step": 3403
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.317534640239552,
      "learning_rate": 1.5893135035638658e-05,
      "loss": 0.458,
      "step": 3404
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.352315915774499,
      "learning_rate": 1.5890671537124565e-05,
      "loss": 0.5737,
      "step": 3405
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2910932380798614,
      "learning_rate": 1.5888207491023824e-05,
      "loss": 0.4209,
      "step": 3406
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3189501279689777,
      "learning_rate": 1.5885742897565494e-05,
      "loss": 0.4551,
      "step": 3407
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3318481971127194,
      "learning_rate": 1.588327775697867e-05,
      "loss": 0.5508,
      "step": 3408
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.400435187861884,
      "learning_rate": 1.5880812069492516e-05,
      "loss": 0.585,
      "step": 3409
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2441626932069492,
      "learning_rate": 1.5878345835336232e-05,
      "loss": 0.4795,
      "step": 3410
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5042234805549282,
      "learning_rate": 1.5875879054739075e-05,
      "loss": 0.5586,
      "step": 3411
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.399562288659144,
      "learning_rate": 1.587341172793036e-05,
      "loss": 0.5391,
      "step": 3412
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2048428476980417,
      "learning_rate": 1.5870943855139437e-05,
      "loss": 0.4263,
      "step": 3413
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5437741298711076,
      "learning_rate": 1.5868475436595713e-05,
      "loss": 0.6304,
      "step": 3414
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.181961163032277,
      "learning_rate": 1.586600647252866e-05,
      "loss": 0.4619,
      "step": 3415
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4101409350267844,
      "learning_rate": 1.586353696316778e-05,
      "loss": 0.4927,
      "step": 3416
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3988313799324972,
      "learning_rate": 1.5861066908742638e-05,
      "loss": 0.5171,
      "step": 3417
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3527012266992633,
      "learning_rate": 1.585859630948284e-05,
      "loss": 0.4683,
      "step": 3418
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4407542478715394,
      "learning_rate": 1.5856125165618056e-05,
      "loss": 0.4824,
      "step": 3419
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4441773062870664,
      "learning_rate": 1.5853653477377996e-05,
      "loss": 0.4624,
      "step": 3420
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4237591802141998,
      "learning_rate": 1.585118124499242e-05,
      "loss": 0.5508,
      "step": 3421
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4245659715245766,
      "learning_rate": 1.584870846869115e-05,
      "loss": 0.4956,
      "step": 3422
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.328748690289947,
      "learning_rate": 1.5846235148704047e-05,
      "loss": 0.416,
      "step": 3423
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2158526570339871,
      "learning_rate": 1.5843761285261027e-05,
      "loss": 0.3936,
      "step": 3424
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3215621822351784,
      "learning_rate": 1.5841286878592055e-05,
      "loss": 0.4302,
      "step": 3425
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4128752177846557,
      "learning_rate": 1.583881192892715e-05,
      "loss": 0.4741,
      "step": 3426
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2994342016149971,
      "learning_rate": 1.5836336436496377e-05,
      "loss": 0.4126,
      "step": 3427
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1808788858464292,
      "learning_rate": 1.5833860401529855e-05,
      "loss": 0.5044,
      "step": 3428
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2843215295715804,
      "learning_rate": 1.5831383824257748e-05,
      "loss": 0.4229,
      "step": 3429
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2945249218590575,
      "learning_rate": 1.582890670491028e-05,
      "loss": 0.3809,
      "step": 3430
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3665963633743718,
      "learning_rate": 1.5826429043717716e-05,
      "loss": 0.5098,
      "step": 3431
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.258103048607019,
      "learning_rate": 1.5823950840910376e-05,
      "loss": 0.4321,
      "step": 3432
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2804149227497543,
      "learning_rate": 1.582147209671863e-05,
      "loss": 0.4536,
      "step": 3433
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.142586070464346,
      "learning_rate": 1.5818992811372898e-05,
      "loss": 0.4185,
      "step": 3434
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2499716020551757,
      "learning_rate": 1.581651298510365e-05,
      "loss": 0.5117,
      "step": 3435
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.342237523527866,
      "learning_rate": 1.58140326181414e-05,
      "loss": 0.4556,
      "step": 3436
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3519842230708532,
      "learning_rate": 1.5811551710716732e-05,
      "loss": 0.5381,
      "step": 3437
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3576690829811005,
      "learning_rate": 1.5809070263060256e-05,
      "loss": 0.4902,
      "step": 3438
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.240055594773596,
      "learning_rate": 1.580658827540265e-05,
      "loss": 0.4458,
      "step": 3439
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2919562123794281,
      "learning_rate": 1.5804105747974626e-05,
      "loss": 0.4976,
      "step": 3440
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6091636367268247,
      "learning_rate": 1.5801622681006966e-05,
      "loss": 0.6348,
      "step": 3441
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5847957166826898,
      "learning_rate": 1.5799139074730487e-05,
      "loss": 0.5742,
      "step": 3442
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2760933995740855,
      "learning_rate": 1.5796654929376057e-05,
      "loss": 0.4829,
      "step": 3443
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4332807069000808,
      "learning_rate": 1.5794170245174605e-05,
      "loss": 0.4585,
      "step": 3444
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6836476231616373,
      "learning_rate": 1.5791685022357098e-05,
      "loss": 0.623,
      "step": 3445
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2819934517894394,
      "learning_rate": 1.5789199261154557e-05,
      "loss": 0.4707,
      "step": 3446
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6334795660826193,
      "learning_rate": 1.578671296179806e-05,
      "loss": 0.6138,
      "step": 3447
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2457388080556606,
      "learning_rate": 1.5784226124518724e-05,
      "loss": 0.4941,
      "step": 3448
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.414440436721024,
      "learning_rate": 1.5781738749547724e-05,
      "loss": 0.4976,
      "step": 3449
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.469773705003075,
      "learning_rate": 1.5779250837116275e-05,
      "loss": 0.4917,
      "step": 3450
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3235805479978697,
      "learning_rate": 1.577676238745566e-05,
      "loss": 0.54,
      "step": 3451
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4730715280390814,
      "learning_rate": 1.5774273400797195e-05,
      "loss": 0.5527,
      "step": 3452
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3850333885913322,
      "learning_rate": 1.5771783877372252e-05,
      "loss": 0.5483,
      "step": 3453
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2877340140202211,
      "learning_rate": 1.576929381741226e-05,
      "loss": 0.4282,
      "step": 3454
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2457798130591384,
      "learning_rate": 1.5766803221148676e-05,
      "loss": 0.4707,
      "step": 3455
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.317648990208334,
      "learning_rate": 1.576431208881303e-05,
      "loss": 0.4756,
      "step": 3456
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2850682788832597,
      "learning_rate": 1.5761820420636894e-05,
      "loss": 0.479,
      "step": 3457
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6365201283580728,
      "learning_rate": 1.5759328216851886e-05,
      "loss": 0.5596,
      "step": 3458
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.539474996080632,
      "learning_rate": 1.5756835477689683e-05,
      "loss": 0.6113,
      "step": 3459
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4808107072046512,
      "learning_rate": 1.5754342203382003e-05,
      "loss": 0.5938,
      "step": 3460
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.438992873286106,
      "learning_rate": 1.575184839416061e-05,
      "loss": 0.561,
      "step": 3461
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3182319751989797,
      "learning_rate": 1.5749354050257334e-05,
      "loss": 0.4648,
      "step": 3462
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.213426932555141,
      "learning_rate": 1.574685917190404e-05,
      "loss": 0.4585,
      "step": 3463
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4671020918543414,
      "learning_rate": 1.5744363759332647e-05,
      "loss": 0.5908,
      "step": 3464
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3514373410462628,
      "learning_rate": 1.5741867812775125e-05,
      "loss": 0.4546,
      "step": 3465
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3436881648717143,
      "learning_rate": 1.5739371332463496e-05,
      "loss": 0.4683,
      "step": 3466
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.456086335356034,
      "learning_rate": 1.573687431862982e-05,
      "loss": 0.5566,
      "step": 3467
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3191289794062184,
      "learning_rate": 1.5734376771506222e-05,
      "loss": 0.4868,
      "step": 3468
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4911534022080988,
      "learning_rate": 1.5731878691324874e-05,
      "loss": 0.5513,
      "step": 3469
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4691091199346529,
      "learning_rate": 1.5729380078317982e-05,
      "loss": 0.5601,
      "step": 3470
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3747890959267801,
      "learning_rate": 1.572688093271782e-05,
      "loss": 0.4824,
      "step": 3471
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5804207569973647,
      "learning_rate": 1.57243812547567e-05,
      "loss": 0.6187,
      "step": 3472
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.230542408764944,
      "learning_rate": 1.5721881044666996e-05,
      "loss": 0.5068,
      "step": 3473
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2665879305735286,
      "learning_rate": 1.5719380302681114e-05,
      "loss": 0.4844,
      "step": 3474
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1467402020186295,
      "learning_rate": 1.5716879029031523e-05,
      "loss": 0.4082,
      "step": 3475
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2941275139756734,
      "learning_rate": 1.5714377223950734e-05,
      "loss": 0.5249,
      "step": 3476
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3492863451523898,
      "learning_rate": 1.5711874887671318e-05,
      "loss": 0.5581,
      "step": 3477
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2509611212957625,
      "learning_rate": 1.5709372020425877e-05,
      "loss": 0.4199,
      "step": 3478
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4722941458072276,
      "learning_rate": 1.5706868622447084e-05,
      "loss": 0.5117,
      "step": 3479
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3896514247976255,
      "learning_rate": 1.570436469396764e-05,
      "loss": 0.5298,
      "step": 3480
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.344701506901641,
      "learning_rate": 1.5701860235220314e-05,
      "loss": 0.4424,
      "step": 3481
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2485132218196777,
      "learning_rate": 1.569935524643791e-05,
      "loss": 0.5039,
      "step": 3482
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2803555309121053,
      "learning_rate": 1.5696849727853297e-05,
      "loss": 0.4312,
      "step": 3483
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3133244742457175,
      "learning_rate": 1.5694343679699377e-05,
      "loss": 0.4644,
      "step": 3484
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4704151544153903,
      "learning_rate": 1.5691837102209103e-05,
      "loss": 0.5693,
      "step": 3485
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.44804193310135,
      "learning_rate": 1.568932999561549e-05,
      "loss": 0.5767,
      "step": 3486
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3316066385583467,
      "learning_rate": 1.568682236015159e-05,
      "loss": 0.459,
      "step": 3487
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.372264965096391,
      "learning_rate": 1.5684314196050516e-05,
      "loss": 0.478,
      "step": 3488
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2421202342795596,
      "learning_rate": 1.568180550354541e-05,
      "loss": 0.4087,
      "step": 3489
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3462215781106004,
      "learning_rate": 1.567929628286949e-05,
      "loss": 0.5044,
      "step": 3490
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.433885706023974,
      "learning_rate": 1.5676786534255993e-05,
      "loss": 0.5381,
      "step": 3491
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3168060457420172,
      "learning_rate": 1.5674276257938234e-05,
      "loss": 0.4644,
      "step": 3492
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3972586520005263,
      "learning_rate": 1.5671765454149558e-05,
      "loss": 0.6104,
      "step": 3493
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3370055650416863,
      "learning_rate": 1.566925412312337e-05,
      "loss": 0.5376,
      "step": 3494
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2219586531548692,
      "learning_rate": 1.566674226509311e-05,
      "loss": 0.4785,
      "step": 3495
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4019109527039944,
      "learning_rate": 1.5664229880292286e-05,
      "loss": 0.6436,
      "step": 3496
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3936678153451592,
      "learning_rate": 1.5661716968954436e-05,
      "loss": 0.5493,
      "step": 3497
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.308502858312181,
      "learning_rate": 1.5659203531313162e-05,
      "loss": 0.4707,
      "step": 3498
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3043196062041333,
      "learning_rate": 1.565668956760211e-05,
      "loss": 0.4785,
      "step": 3499
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2177983282255926,
      "learning_rate": 1.5654175078054965e-05,
      "loss": 0.4702,
      "step": 3500
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2651119783438836,
      "learning_rate": 1.5651660062905476e-05,
      "loss": 0.4165,
      "step": 3501
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4172828722022872,
      "learning_rate": 1.5649144522387437e-05,
      "loss": 0.5796,
      "step": 3502
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3439679077776334,
      "learning_rate": 1.564662845673468e-05,
      "loss": 0.4404,
      "step": 3503
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3063097946499533,
      "learning_rate": 1.56441118661811e-05,
      "loss": 0.5259,
      "step": 3504
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2847356951863893,
      "learning_rate": 1.564159475096064e-05,
      "loss": 0.4287,
      "step": 3505
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3730767548084137,
      "learning_rate": 1.5639077111307276e-05,
      "loss": 0.4907,
      "step": 3506
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1167305315665201,
      "learning_rate": 1.563655894745505e-05,
      "loss": 0.374,
      "step": 3507
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.504440068894076,
      "learning_rate": 1.5634040259638044e-05,
      "loss": 0.584,
      "step": 3508
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3487103324776581,
      "learning_rate": 1.563152104809039e-05,
      "loss": 0.4834,
      "step": 3509
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4352961350734168,
      "learning_rate": 1.562900131304627e-05,
      "loss": 0.5234,
      "step": 3510
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.254565018326531,
      "learning_rate": 1.5626481054739916e-05,
      "loss": 0.4385,
      "step": 3511
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3074397082906155,
      "learning_rate": 1.5623960273405605e-05,
      "loss": 0.4497,
      "step": 3512
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3858303793589999,
      "learning_rate": 1.5621438969277667e-05,
      "loss": 0.543,
      "step": 3513
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3938791575017626,
      "learning_rate": 1.5618917142590472e-05,
      "loss": 0.5,
      "step": 3514
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.148406433659883,
      "learning_rate": 1.5616394793578455e-05,
      "loss": 0.438,
      "step": 3515
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.474315743397625,
      "learning_rate": 1.5613871922476082e-05,
      "loss": 0.5918,
      "step": 3516
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3019075740889432,
      "learning_rate": 1.5611348529517872e-05,
      "loss": 0.5327,
      "step": 3517
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4331740541083156,
      "learning_rate": 1.56088246149384e-05,
      "loss": 0.5249,
      "step": 3518
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2141174874683922,
      "learning_rate": 1.560630017897229e-05,
      "loss": 0.3506,
      "step": 3519
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2021962322663355,
      "learning_rate": 1.5603775221854195e-05,
      "loss": 0.4434,
      "step": 3520
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2417825385982273,
      "learning_rate": 1.560124974381884e-05,
      "loss": 0.5259,
      "step": 3521
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4493187366288793,
      "learning_rate": 1.559872374510099e-05,
      "loss": 0.5781,
      "step": 3522
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5314145723909574,
      "learning_rate": 1.5596197225935458e-05,
      "loss": 0.5396,
      "step": 3523
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4051070778100858,
      "learning_rate": 1.5593670186557096e-05,
      "loss": 0.5918,
      "step": 3524
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.530885345644857,
      "learning_rate": 1.5591142627200825e-05,
      "loss": 0.4858,
      "step": 3525
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.523910992392368,
      "learning_rate": 1.5588614548101593e-05,
      "loss": 0.6318,
      "step": 3526
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3351611604575337,
      "learning_rate": 1.558608594949441e-05,
      "loss": 0.5449,
      "step": 3527
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3656071510326033,
      "learning_rate": 1.5583556831614333e-05,
      "loss": 0.5088,
      "step": 3528
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2343447591802204,
      "learning_rate": 1.5581027194696458e-05,
      "loss": 0.4971,
      "step": 3529
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2471244723310353,
      "learning_rate": 1.557849703897594e-05,
      "loss": 0.439,
      "step": 3530
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.279345435231848,
      "learning_rate": 1.557596636468798e-05,
      "loss": 0.5439,
      "step": 3531
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3810751896812414,
      "learning_rate": 1.557343517206782e-05,
      "loss": 0.5576,
      "step": 3532
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2551374768028685,
      "learning_rate": 1.5570903461350754e-05,
      "loss": 0.4663,
      "step": 3533
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3213963720031345,
      "learning_rate": 1.5568371232772134e-05,
      "loss": 0.5537,
      "step": 3534
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.293576961839571,
      "learning_rate": 1.5565838486567343e-05,
      "loss": 0.5205,
      "step": 3535
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4072676164683218,
      "learning_rate": 1.5563305222971826e-05,
      "loss": 0.4976,
      "step": 3536
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.1757461783588063,
      "learning_rate": 1.556077144222107e-05,
      "loss": 0.3545,
      "step": 3537
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2280331247225205,
      "learning_rate": 1.5558237144550608e-05,
      "loss": 0.4595,
      "step": 3538
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2465966090240976,
      "learning_rate": 1.5555702330196024e-05,
      "loss": 0.4639,
      "step": 3539
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2358201122755634,
      "learning_rate": 1.5553166999392954e-05,
      "loss": 0.4951,
      "step": 3540
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.338458093154734,
      "learning_rate": 1.5550631152377075e-05,
      "loss": 0.478,
      "step": 3541
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9684010634969923,
      "learning_rate": 1.5548094789384113e-05,
      "loss": 0.3442,
      "step": 3542
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2607930432384393,
      "learning_rate": 1.554555791064985e-05,
      "loss": 0.4556,
      "step": 3543
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.1478507843129124,
      "learning_rate": 1.5543020516410104e-05,
      "loss": 0.373,
      "step": 3544
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.1815185140653188,
      "learning_rate": 1.5540482606900752e-05,
      "loss": 0.3711,
      "step": 3545
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0691212357541922,
      "learning_rate": 1.553794418235771e-05,
      "loss": 0.3096,
      "step": 3546
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.267349401126991,
      "learning_rate": 1.5535405243016945e-05,
      "loss": 0.4551,
      "step": 3547
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2056749093773311,
      "learning_rate": 1.5532865789114477e-05,
      "loss": 0.3887,
      "step": 3548
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3303339365079503,
      "learning_rate": 1.5530325820886366e-05,
      "loss": 0.4204,
      "step": 3549
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.1653432139533353,
      "learning_rate": 1.5527785338568718e-05,
      "loss": 0.3389,
      "step": 3550
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5632514563105926,
      "learning_rate": 1.55252443423977e-05,
      "loss": 0.3916,
      "step": 3551
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4800532327289446,
      "learning_rate": 1.5522702832609516e-05,
      "loss": 0.3943,
      "step": 3552
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3607794457565412,
      "learning_rate": 1.552016080944042e-05,
      "loss": 0.3599,
      "step": 3553
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3719236765426337,
      "learning_rate": 1.5517618273126714e-05,
      "loss": 0.3188,
      "step": 3554
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4507358687662242,
      "learning_rate": 1.5515075223904745e-05,
      "loss": 0.3511,
      "step": 3555
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.701256396325289,
      "learning_rate": 1.5512531662010918e-05,
      "loss": 0.417,
      "step": 3556
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5081113389261729,
      "learning_rate": 1.550998758768167e-05,
      "loss": 0.3071,
      "step": 3557
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4276093077406713,
      "learning_rate": 1.5507443001153497e-05,
      "loss": 0.2556,
      "step": 3558
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.556605829206475,
      "learning_rate": 1.550489790266294e-05,
      "loss": 0.3306,
      "step": 3559
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.5186854017628661,
      "learning_rate": 1.550235229244659e-05,
      "loss": 0.323,
      "step": 3560
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.4279138483536427,
      "learning_rate": 1.5499806170741073e-05,
      "loss": 0.3203,
      "step": 3561
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.4811620404749868,
      "learning_rate": 1.5497259537783084e-05,
      "loss": 0.3257,
      "step": 3562
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.4738460655487196,
      "learning_rate": 1.5494712393809343e-05,
      "loss": 0.2871,
      "step": 3563
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.4053540946121286,
      "learning_rate": 1.5492164739056635e-05,
      "loss": 0.3169,
      "step": 3564
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.4225132197649668,
      "learning_rate": 1.5489616573761784e-05,
      "loss": 0.3623,
      "step": 3565
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.4831429128999856,
      "learning_rate": 1.548706789816166e-05,
      "loss": 0.3267,
      "step": 3566
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.4069981700801724,
      "learning_rate": 1.5484518712493188e-05,
      "loss": 0.334,
      "step": 3567
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.4189319610564288,
      "learning_rate": 1.5481969016993335e-05,
      "loss": 0.2952,
      "step": 3568
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.45381832132131,
      "learning_rate": 1.547941881189911e-05,
      "loss": 0.3196,
      "step": 3569
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.470497344840741,
      "learning_rate": 1.5476868097447586e-05,
      "loss": 0.4111,
      "step": 3570
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.3797480383566056,
      "learning_rate": 1.5474316873875864e-05,
      "loss": 0.3643,
      "step": 3571
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.4813253513307676,
      "learning_rate": 1.547176514142111e-05,
      "loss": 0.321,
      "step": 3572
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.5340788232623674,
      "learning_rate": 1.546921290032052e-05,
      "loss": 0.3457,
      "step": 3573
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.4898818241964076,
      "learning_rate": 1.546666015081135e-05,
      "loss": 0.2776,
      "step": 3574
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.7510824880901297,
      "learning_rate": 1.5464106893130896e-05,
      "loss": 0.4517,
      "step": 3575
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.854102959769189,
      "learning_rate": 1.546155312751651e-05,
      "loss": 0.4658,
      "step": 3576
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.541408128955297,
      "learning_rate": 1.5458998854205585e-05,
      "loss": 0.3789,
      "step": 3577
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.3854703995228934,
      "learning_rate": 1.5456444073435558e-05,
      "loss": 0.2812,
      "step": 3578
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.4959222679004367,
      "learning_rate": 1.5453888785443916e-05,
      "loss": 0.3076,
      "step": 3579
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.4317423719783118,
      "learning_rate": 1.5451332990468202e-05,
      "loss": 0.2839,
      "step": 3580
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.3558311397679406,
      "learning_rate": 1.544877668874599e-05,
      "loss": 0.2949,
      "step": 3581
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.6223487790623166,
      "learning_rate": 1.5446219880514913e-05,
      "loss": 0.2737,
      "step": 3582
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.822078358526578,
      "learning_rate": 1.5443662566012645e-05,
      "loss": 0.4058,
      "step": 3583
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.5343897115696987,
      "learning_rate": 1.5441104745476913e-05,
      "loss": 0.3481,
      "step": 3584
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.7358014900804546,
      "learning_rate": 1.543854641914549e-05,
      "loss": 0.4067,
      "step": 3585
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.6362259903947847,
      "learning_rate": 1.5435987587256183e-05,
      "loss": 0.3203,
      "step": 3586
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.6971310164510125,
      "learning_rate": 1.543342825004687e-05,
      "loss": 0.3848,
      "step": 3587
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.8163270502996036,
      "learning_rate": 1.543086840775545e-05,
      "loss": 0.3613,
      "step": 3588
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.400210908021434,
      "learning_rate": 1.5428308060619893e-05,
      "loss": 0.3064,
      "step": 3589
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.4860755759805304,
      "learning_rate": 1.5425747208878195e-05,
      "loss": 0.3677,
      "step": 3590
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.507534199507966,
      "learning_rate": 1.542318585276841e-05,
      "loss": 0.3594,
      "step": 3591
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.3893352642461245,
      "learning_rate": 1.5420623992528643e-05,
      "loss": 0.3164,
      "step": 3592
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.541381778710299,
      "learning_rate": 1.5418061628397037e-05,
      "loss": 0.3853,
      "step": 3593
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.5158497382977771,
      "learning_rate": 1.541549876061178e-05,
      "loss": 0.3955,
      "step": 3594
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.4846020940730422,
      "learning_rate": 1.5412935389411124e-05,
      "loss": 0.3521,
      "step": 3595
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.6709208853440303,
      "learning_rate": 1.5410371515033343e-05,
      "loss": 0.395,
      "step": 3596
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.4873183886288837,
      "learning_rate": 1.5407807137716774e-05,
      "loss": 0.2939,
      "step": 3597
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.4830681128085466,
      "learning_rate": 1.54052422576998e-05,
      "loss": 0.3296,
      "step": 3598
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.4699618285703753,
      "learning_rate": 1.5402676875220847e-05,
      "loss": 0.3181,
      "step": 3599
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.5063447193324462,
      "learning_rate": 1.5400110990518386e-05,
      "loss": 0.3589,
      "step": 3600
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.7465077115134413,
      "learning_rate": 1.539754460383094e-05,
      "loss": 0.3569,
      "step": 3601
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.5742581500580013,
      "learning_rate": 1.5394977715397073e-05,
      "loss": 0.3635,
      "step": 3602
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.4852161436370117,
      "learning_rate": 1.5392410325455397e-05,
      "loss": 0.3291,
      "step": 3603
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.4494157729584143,
      "learning_rate": 1.538984243424458e-05,
      "loss": 0.2471,
      "step": 3604
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.6142577981200255,
      "learning_rate": 1.5387274042003327e-05,
      "loss": 0.3391,
      "step": 3605
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.7011205405025542,
      "learning_rate": 1.5384705148970384e-05,
      "loss": 0.333,
      "step": 3606
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.6960605916289762,
      "learning_rate": 1.5382135755384554e-05,
      "loss": 0.3652,
      "step": 3607
    },
    {
      "epoch": 1.02,
      "grad_norm": 2.8059824597946887,
      "learning_rate": 1.537956586148469e-05,
      "loss": 0.3411,
      "step": 3608
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.5401882794858641,
      "learning_rate": 1.5376995467509673e-05,
      "loss": 0.3115,
      "step": 3609
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.401706610522498,
      "learning_rate": 1.5374424573698453e-05,
      "loss": 0.2805,
      "step": 3610
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.7248547075699865,
      "learning_rate": 1.537185318029001e-05,
      "loss": 0.3662,
      "step": 3611
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.518824398514535,
      "learning_rate": 1.5369281287523376e-05,
      "loss": 0.3208,
      "step": 3612
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.7080642612685168,
      "learning_rate": 1.536670889563763e-05,
      "loss": 0.2791,
      "step": 3613
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.3715376618446395,
      "learning_rate": 1.5364136004871906e-05,
      "loss": 0.2756,
      "step": 3614
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.5118911960962578,
      "learning_rate": 1.5361562615465366e-05,
      "loss": 0.3296,
      "step": 3615
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.6329428047377381,
      "learning_rate": 1.5358988727657227e-05,
      "loss": 0.4211,
      "step": 3616
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.511750333456279,
      "learning_rate": 1.5356414341686758e-05,
      "loss": 0.3071,
      "step": 3617
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.5365138483281502,
      "learning_rate": 1.535383945779327e-05,
      "loss": 0.3474,
      "step": 3618
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.5037275966594943,
      "learning_rate": 1.5351264076216114e-05,
      "loss": 0.3213,
      "step": 3619
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.595658943149936,
      "learning_rate": 1.5348688197194696e-05,
      "loss": 0.3838,
      "step": 3620
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.4459994619680332,
      "learning_rate": 1.534611182096847e-05,
      "loss": 0.3335,
      "step": 3621
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.5323595505514478,
      "learning_rate": 1.5343534947776924e-05,
      "loss": 0.3979,
      "step": 3622
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.3887126441781588,
      "learning_rate": 1.5340957577859605e-05,
      "loss": 0.314,
      "step": 3623
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.4170067435813996,
      "learning_rate": 1.5338379711456096e-05,
      "loss": 0.3145,
      "step": 3624
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.4443070670591518,
      "learning_rate": 1.5335801348806036e-05,
      "loss": 0.3201,
      "step": 3625
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.3939784889888347,
      "learning_rate": 1.53332224901491e-05,
      "loss": 0.2939,
      "step": 3626
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.4639468547204189,
      "learning_rate": 1.5330643135725022e-05,
      "loss": 0.3027,
      "step": 3627
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.6398931803022263,
      "learning_rate": 1.5328063285773567e-05,
      "loss": 0.376,
      "step": 3628
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.4810000310506806,
      "learning_rate": 1.5325482940534554e-05,
      "loss": 0.3,
      "step": 3629
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.5409889722835188,
      "learning_rate": 1.532290210024785e-05,
      "loss": 0.3606,
      "step": 3630
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.6637851770196659,
      "learning_rate": 1.5320320765153367e-05,
      "loss": 0.3887,
      "step": 3631
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.5117054056367765,
      "learning_rate": 1.531773893549106e-05,
      "loss": 0.3188,
      "step": 3632
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.5927411282563662,
      "learning_rate": 1.5315156611500927e-05,
      "loss": 0.3228,
      "step": 3633
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.7900085606876877,
      "learning_rate": 1.531257379342302e-05,
      "loss": 0.4492,
      "step": 3634
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.4742799022820807,
      "learning_rate": 1.5309990481497438e-05,
      "loss": 0.3252,
      "step": 3635
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.4402101627066803,
      "learning_rate": 1.5307406675964315e-05,
      "loss": 0.3171,
      "step": 3636
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.8354080697792037,
      "learning_rate": 1.530482237706383e-05,
      "loss": 0.3696,
      "step": 3637
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.6847780760765625,
      "learning_rate": 1.5302237585036236e-05,
      "loss": 0.3843,
      "step": 3638
    },
    {
      "epoch": 1.03,
      "grad_norm": 2.0289057350311044,
      "learning_rate": 1.5299652300121792e-05,
      "loss": 0.3875,
      "step": 3639
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.6613417124402041,
      "learning_rate": 1.529706652256083e-05,
      "loss": 0.3289,
      "step": 3640
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.6105142089088034,
      "learning_rate": 1.5294480252593718e-05,
      "loss": 0.375,
      "step": 3641
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.8210834419097757,
      "learning_rate": 1.5291893490460868e-05,
      "loss": 0.3896,
      "step": 3642
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.925535179838433,
      "learning_rate": 1.5289306236402744e-05,
      "loss": 0.4927,
      "step": 3643
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.4926622676705736,
      "learning_rate": 1.5286718490659854e-05,
      "loss": 0.3359,
      "step": 3644
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.5502253148401146,
      "learning_rate": 1.5284130253472746e-05,
      "loss": 0.3274,
      "step": 3645
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.721584276200717,
      "learning_rate": 1.5281541525082024e-05,
      "loss": 0.3701,
      "step": 3646
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.4669115706585625,
      "learning_rate": 1.5278952305728325e-05,
      "loss": 0.3059,
      "step": 3647
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.6184943241704015,
      "learning_rate": 1.5276362595652347e-05,
      "loss": 0.394,
      "step": 3648
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.484726400386461,
      "learning_rate": 1.5273772395094814e-05,
      "loss": 0.3511,
      "step": 3649
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.3894670993315918,
      "learning_rate": 1.5271181704296513e-05,
      "loss": 0.3169,
      "step": 3650
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.5821883938611925,
      "learning_rate": 1.526859052349827e-05,
      "loss": 0.4219,
      "step": 3651
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.687249600899217,
      "learning_rate": 1.526599885294096e-05,
      "loss": 0.3867,
      "step": 3652
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.6885564943032338,
      "learning_rate": 1.5263406692865494e-05,
      "loss": 0.3755,
      "step": 3653
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.7370836749605112,
      "learning_rate": 1.5260814043512838e-05,
      "loss": 0.324,
      "step": 3654
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.655435782023133,
      "learning_rate": 1.5258220905123997e-05,
      "loss": 0.3711,
      "step": 3655
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.6060219121499657,
      "learning_rate": 1.5255627277940023e-05,
      "loss": 0.3799,
      "step": 3656
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.4793924542161976,
      "learning_rate": 1.5253033162202027e-05,
      "loss": 0.3066,
      "step": 3657
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.4409300080465355,
      "learning_rate": 1.5250438558151142e-05,
      "loss": 0.2891,
      "step": 3658
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.6018097263004611,
      "learning_rate": 1.524784346602856e-05,
      "loss": 0.3325,
      "step": 3659
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.5380799526968103,
      "learning_rate": 1.5245247886075518e-05,
      "loss": 0.4072,
      "step": 3660
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.7550942688604494,
      "learning_rate": 1.5242651818533299e-05,
      "loss": 0.3511,
      "step": 3661
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.4985024521801573,
      "learning_rate": 1.5240055263643223e-05,
      "loss": 0.2695,
      "step": 3662
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.5565713862763706,
      "learning_rate": 1.5237458221646668e-05,
      "loss": 0.3245,
      "step": 3663
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.6264536145176478,
      "learning_rate": 1.5234860692785045e-05,
      "loss": 0.325,
      "step": 3664
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.6325792828781653,
      "learning_rate": 1.5232262677299816e-05,
      "loss": 0.3818,
      "step": 3665
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.5487000246681335,
      "learning_rate": 1.5229664175432494e-05,
      "loss": 0.395,
      "step": 3666
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.5139870092540229,
      "learning_rate": 1.5227065187424623e-05,
      "loss": 0.3352,
      "step": 3667
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.577591761985644,
      "learning_rate": 1.5224465713517811e-05,
      "loss": 0.2969,
      "step": 3668
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.4176261445671359,
      "learning_rate": 1.522186575395369e-05,
      "loss": 0.3193,
      "step": 3669
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.5586242613598074,
      "learning_rate": 1.5219265308973952e-05,
      "loss": 0.324,
      "step": 3670
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.3369607897530629,
      "learning_rate": 1.5216664378820327e-05,
      "loss": 0.2832,
      "step": 3671
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.5557326347003313,
      "learning_rate": 1.5214062963734599e-05,
      "loss": 0.283,
      "step": 3672
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.5364970934540052,
      "learning_rate": 1.5211461063958589e-05,
      "loss": 0.3435,
      "step": 3673
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.7894926693846525,
      "learning_rate": 1.5208858679734161e-05,
      "loss": 0.3774,
      "step": 3674
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.8265776716484519,
      "learning_rate": 1.5206255811303235e-05,
      "loss": 0.4316,
      "step": 3675
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.4009939721555256,
      "learning_rate": 1.5203652458907763e-05,
      "loss": 0.2551,
      "step": 3676
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.6496762557520113,
      "learning_rate": 1.5201048622789747e-05,
      "loss": 0.3979,
      "step": 3677
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.4170743517826594,
      "learning_rate": 1.5198444303191244e-05,
      "loss": 0.2407,
      "step": 3678
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.4454971878773377,
      "learning_rate": 1.5195839500354337e-05,
      "loss": 0.333,
      "step": 3679
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.6871577621663036,
      "learning_rate": 1.519323421452117e-05,
      "loss": 0.3618,
      "step": 3680
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.4916674264586223,
      "learning_rate": 1.5190628445933925e-05,
      "loss": 0.3433,
      "step": 3681
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.541403607287981,
      "learning_rate": 1.5188022194834831e-05,
      "loss": 0.3643,
      "step": 3682
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.485449064670457,
      "learning_rate": 1.5185415461466155e-05,
      "loss": 0.2744,
      "step": 3683
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.4822578999917,
      "learning_rate": 1.5182808246070222e-05,
      "loss": 0.301,
      "step": 3684
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.7404232320291966,
      "learning_rate": 1.5180200548889387e-05,
      "loss": 0.418,
      "step": 3685
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.456969982938576,
      "learning_rate": 1.517759237016606e-05,
      "loss": 0.2781,
      "step": 3686
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.4534236483264331,
      "learning_rate": 1.5174983710142694e-05,
      "loss": 0.2976,
      "step": 3687
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.599283006440858,
      "learning_rate": 1.5172374569061787e-05,
      "loss": 0.3823,
      "step": 3688
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.5172458145803003,
      "learning_rate": 1.5169764947165879e-05,
      "loss": 0.3623,
      "step": 3689
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.4551824077883115,
      "learning_rate": 1.5167154844697549e-05,
      "loss": 0.343,
      "step": 3690
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.8777773556472674,
      "learning_rate": 1.5164544261899439e-05,
      "loss": 0.4893,
      "step": 3691
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.3078506174677793,
      "learning_rate": 1.5161933199014216e-05,
      "loss": 0.2588,
      "step": 3692
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.4977815482549186,
      "learning_rate": 1.5159321656284602e-05,
      "loss": 0.3013,
      "step": 3693
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.4883717201069901,
      "learning_rate": 1.5156709633953364e-05,
      "loss": 0.3184,
      "step": 3694
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.510902981569387,
      "learning_rate": 1.515409713226331e-05,
      "loss": 0.2869,
      "step": 3695
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.4691729033923477,
      "learning_rate": 1.5151484151457292e-05,
      "loss": 0.3323,
      "step": 3696
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.574652381301715,
      "learning_rate": 1.5148870691778208e-05,
      "loss": 0.3794,
      "step": 3697
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.5806168145985826,
      "learning_rate": 1.5146256753469004e-05,
      "loss": 0.3301,
      "step": 3698
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.5548362922741303,
      "learning_rate": 1.5143642336772663e-05,
      "loss": 0.3145,
      "step": 3699
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.4120372318643062,
      "learning_rate": 1.5141027441932217e-05,
      "loss": 0.3208,
      "step": 3700
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.6775944747604845,
      "learning_rate": 1.5138412069190747e-05,
      "loss": 0.3433,
      "step": 3701
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.5145952863956298,
      "learning_rate": 1.513579621879137e-05,
      "loss": 0.3364,
      "step": 3702
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.6470859555836679,
      "learning_rate": 1.513317989097725e-05,
      "loss": 0.3398,
      "step": 3703
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.6310189494507088,
      "learning_rate": 1.5130563085991599e-05,
      "loss": 0.2979,
      "step": 3704
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.7370738171571436,
      "learning_rate": 1.5127945804077668e-05,
      "loss": 0.3462,
      "step": 3705
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.867443953759633,
      "learning_rate": 1.5125328045478755e-05,
      "loss": 0.4321,
      "step": 3706
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.59872006713413,
      "learning_rate": 1.5122709810438205e-05,
      "loss": 0.2954,
      "step": 3707
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.553115744692842,
      "learning_rate": 1.5120091099199403e-05,
      "loss": 0.283,
      "step": 3708
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.8990277155862176,
      "learning_rate": 1.511747191200578e-05,
      "loss": 0.3599,
      "step": 3709
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.5829882196402232,
      "learning_rate": 1.5114852249100811e-05,
      "loss": 0.3401,
      "step": 3710
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.9483365684405685,
      "learning_rate": 1.5112232110728016e-05,
      "loss": 0.3872,
      "step": 3711
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.5555542037061838,
      "learning_rate": 1.5109611497130959e-05,
      "loss": 0.2974,
      "step": 3712
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.5574443227165546,
      "learning_rate": 1.510699040855324e-05,
      "loss": 0.3237,
      "step": 3713
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.5605250508615411,
      "learning_rate": 1.5104368845238525e-05,
      "loss": 0.3628,
      "step": 3714
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.6152135442794437,
      "learning_rate": 1.5101746807430502e-05,
      "loss": 0.3418,
      "step": 3715
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.8875513934301675,
      "learning_rate": 1.509912429537291e-05,
      "loss": 0.4482,
      "step": 3716
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.538274797663273,
      "learning_rate": 1.509650130930953e-05,
      "loss": 0.2837,
      "step": 3717
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.6185118977291406,
      "learning_rate": 1.5093877849484201e-05,
      "loss": 0.3481,
      "step": 3718
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.5237138731407918,
      "learning_rate": 1.5091253916140789e-05,
      "loss": 0.3408,
      "step": 3719
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.6089488712029743,
      "learning_rate": 1.5088629509523207e-05,
      "loss": 0.3574,
      "step": 3720
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.9084120255604686,
      "learning_rate": 1.5086004629875426e-05,
      "loss": 0.4277,
      "step": 3721
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.6701911039602704,
      "learning_rate": 1.5083379277441437e-05,
      "loss": 0.4004,
      "step": 3722
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.5401958281798074,
      "learning_rate": 1.5080753452465296e-05,
      "loss": 0.3223,
      "step": 3723
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.6605017690073947,
      "learning_rate": 1.5078127155191094e-05,
      "loss": 0.3704,
      "step": 3724
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.6414129199753638,
      "learning_rate": 1.5075500385862967e-05,
      "loss": 0.3359,
      "step": 3725
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.6937974587006297,
      "learning_rate": 1.5072873144725093e-05,
      "loss": 0.3564,
      "step": 3726
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.38800562527389,
      "learning_rate": 1.5070245432021699e-05,
      "loss": 0.3203,
      "step": 3727
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.53963986097933,
      "learning_rate": 1.5067617247997053e-05,
      "loss": 0.3755,
      "step": 3728
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.5569164928268933,
      "learning_rate": 1.5064988592895463e-05,
      "loss": 0.3831,
      "step": 3729
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.6195719430089155,
      "learning_rate": 1.5062359466961283e-05,
      "loss": 0.3828,
      "step": 3730
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.5363103181909195,
      "learning_rate": 1.5059729870438917e-05,
      "loss": 0.3662,
      "step": 3731
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.8070233791456392,
      "learning_rate": 1.5057099803572806e-05,
      "loss": 0.4092,
      "step": 3732
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.7664782357173465,
      "learning_rate": 1.5054469266607435e-05,
      "loss": 0.376,
      "step": 3733
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.6565298637858947,
      "learning_rate": 1.5051838259787332e-05,
      "loss": 0.3691,
      "step": 3734
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.6898168207247022,
      "learning_rate": 1.5049206783357082e-05,
      "loss": 0.3442,
      "step": 3735
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.6447481559369412,
      "learning_rate": 1.5046574837561289e-05,
      "loss": 0.3687,
      "step": 3736
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.4488570343862868,
      "learning_rate": 1.504394242264462e-05,
      "loss": 0.3364,
      "step": 3737
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.6141171583617133,
      "learning_rate": 1.504130953885178e-05,
      "loss": 0.3711,
      "step": 3738
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.4465407270195014,
      "learning_rate": 1.5038676186427515e-05,
      "loss": 0.3054,
      "step": 3739
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.3846677379722523,
      "learning_rate": 1.5036042365616621e-05,
      "loss": 0.2378,
      "step": 3740
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.7678307078383626,
      "learning_rate": 1.5033408076663932e-05,
      "loss": 0.3979,
      "step": 3741
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.5795822473963832,
      "learning_rate": 1.5030773319814324e-05,
      "loss": 0.3027,
      "step": 3742
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.5957027423164587,
      "learning_rate": 1.502813809531272e-05,
      "loss": 0.312,
      "step": 3743
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.7035679992020627,
      "learning_rate": 1.5025502403404089e-05,
      "loss": 0.3389,
      "step": 3744
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.8061406180656556,
      "learning_rate": 1.5022866244333438e-05,
      "loss": 0.4351,
      "step": 3745
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.6147023678837928,
      "learning_rate": 1.502022961834582e-05,
      "loss": 0.3906,
      "step": 3746
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.6559265294088459,
      "learning_rate": 1.5017592525686333e-05,
      "loss": 0.4092,
      "step": 3747
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.7261332477301974,
      "learning_rate": 1.5014954966600117e-05,
      "loss": 0.3574,
      "step": 3748
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.5810908231308887,
      "learning_rate": 1.501231694133235e-05,
      "loss": 0.3174,
      "step": 3749
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.7565147668760623,
      "learning_rate": 1.5009678450128263e-05,
      "loss": 0.3608,
      "step": 3750
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.4775456932666768,
      "learning_rate": 1.5007039493233123e-05,
      "loss": 0.3403,
      "step": 3751
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.65098898234367,
      "learning_rate": 1.5004400070892246e-05,
      "loss": 0.353,
      "step": 3752
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.4792765919037965,
      "learning_rate": 1.5001760183350981e-05,
      "loss": 0.3022,
      "step": 3753
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.5405058109206757,
      "learning_rate": 1.4999119830854739e-05,
      "loss": 0.3374,
      "step": 3754
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.4235026909526287,
      "learning_rate": 1.4996479013648952e-05,
      "loss": 0.3525,
      "step": 3755
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.5701189735170358,
      "learning_rate": 1.499383773197911e-05,
      "loss": 0.3638,
      "step": 3756
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.468580333663465,
      "learning_rate": 1.4991195986090744e-05,
      "loss": 0.2996,
      "step": 3757
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.4792638070528505,
      "learning_rate": 1.4988553776229421e-05,
      "loss": 0.3213,
      "step": 3758
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.383263191952761,
      "learning_rate": 1.4985911102640762e-05,
      "loss": 0.333,
      "step": 3759
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.715127495095224,
      "learning_rate": 1.498326796557042e-05,
      "loss": 0.3674,
      "step": 3760
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.8693287008833397,
      "learning_rate": 1.4980624365264103e-05,
      "loss": 0.4111,
      "step": 3761
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.6774819988425125,
      "learning_rate": 1.4977980301967549e-05,
      "loss": 0.3408,
      "step": 3762
    },
    {
      "epoch": 1.06,
      "grad_norm": 2.0086435344578377,
      "learning_rate": 1.4975335775926547e-05,
      "loss": 0.46,
      "step": 3763
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.6549989123444349,
      "learning_rate": 1.497269078738693e-05,
      "loss": 0.3647,
      "step": 3764
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.4732900706615988,
      "learning_rate": 1.4970045336594571e-05,
      "loss": 0.3066,
      "step": 3765
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.5670800916919128,
      "learning_rate": 1.4967399423795384e-05,
      "loss": 0.3408,
      "step": 3766
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.558206746400973,
      "learning_rate": 1.4964753049235333e-05,
      "loss": 0.3574,
      "step": 3767
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.5543452783705907,
      "learning_rate": 1.4962106213160415e-05,
      "loss": 0.3071,
      "step": 3768
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.4646573413872135,
      "learning_rate": 1.4959458915816681e-05,
      "loss": 0.2849,
      "step": 3769
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.694315080915635,
      "learning_rate": 1.495681115745021e-05,
      "loss": 0.3069,
      "step": 3770
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.9285544510941233,
      "learning_rate": 1.4954162938307143e-05,
      "loss": 0.4355,
      "step": 3771
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.5633022525748397,
      "learning_rate": 1.4951514258633652e-05,
      "loss": 0.291,
      "step": 3772
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.5687395246654112,
      "learning_rate": 1.4948865118675948e-05,
      "loss": 0.3726,
      "step": 3773
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.7045054501795565,
      "learning_rate": 1.4946215518680299e-05,
      "loss": 0.3662,
      "step": 3774
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.6908380324092216,
      "learning_rate": 1.4943565458892999e-05,
      "loss": 0.3706,
      "step": 3775
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.8431008070972752,
      "learning_rate": 1.4940914939560398e-05,
      "loss": 0.4448,
      "step": 3776
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.5665995695602648,
      "learning_rate": 1.4938263960928878e-05,
      "loss": 0.3584,
      "step": 3777
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.5896537005118103,
      "learning_rate": 1.493561252324488e-05,
      "loss": 0.2966,
      "step": 3778
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.7953247319256378,
      "learning_rate": 1.4932960626754867e-05,
      "loss": 0.3457,
      "step": 3779
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.5378234007747746,
      "learning_rate": 1.4930308271705357e-05,
      "loss": 0.3472,
      "step": 3780
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.8050492504807598,
      "learning_rate": 1.4927655458342914e-05,
      "loss": 0.4302,
      "step": 3781
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.6864170707650814,
      "learning_rate": 1.4925002186914133e-05,
      "loss": 0.3218,
      "step": 3782
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.7119771842234601,
      "learning_rate": 1.4922348457665656e-05,
      "loss": 0.3955,
      "step": 3783
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.4667475174183726,
      "learning_rate": 1.4919694270844176e-05,
      "loss": 0.3137,
      "step": 3784
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.6329518833559318,
      "learning_rate": 1.4917039626696416e-05,
      "loss": 0.3936,
      "step": 3785
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.476507907466845,
      "learning_rate": 1.491438452546915e-05,
      "loss": 0.2842,
      "step": 3786
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.5172038409792252,
      "learning_rate": 1.4911728967409189e-05,
      "loss": 0.3062,
      "step": 3787
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.7443050567339031,
      "learning_rate": 1.4909072952763395e-05,
      "loss": 0.3381,
      "step": 3788
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.5667841592013974,
      "learning_rate": 1.4906416481778656e-05,
      "loss": 0.3267,
      "step": 3789
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.7385298007860561,
      "learning_rate": 1.4903759554701922e-05,
      "loss": 0.3223,
      "step": 3790
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.6304794829859495,
      "learning_rate": 1.4901102171780175e-05,
      "loss": 0.3369,
      "step": 3791
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.5264769137374012,
      "learning_rate": 1.4898444333260436e-05,
      "loss": 0.3186,
      "step": 3792
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.6484918828737325,
      "learning_rate": 1.4895786039389779e-05,
      "loss": 0.3909,
      "step": 3793
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.8199343262614274,
      "learning_rate": 1.4893127290415312e-05,
      "loss": 0.4429,
      "step": 3794
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.510222662852172,
      "learning_rate": 1.4890468086584187e-05,
      "loss": 0.3604,
      "step": 3795
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.434186224887021,
      "learning_rate": 1.4887808428143595e-05,
      "loss": 0.2981,
      "step": 3796
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.5695798796464935,
      "learning_rate": 1.4885148315340783e-05,
      "loss": 0.3105,
      "step": 3797
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.6391043138068853,
      "learning_rate": 1.4882487748423025e-05,
      "loss": 0.3062,
      "step": 3798
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.9694328278126227,
      "learning_rate": 1.487982672763764e-05,
      "loss": 0.416,
      "step": 3799
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.6710791818441428,
      "learning_rate": 1.4877165253231995e-05,
      "loss": 0.4014,
      "step": 3800
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.6361918076276956,
      "learning_rate": 1.48745033254535e-05,
      "loss": 0.3271,
      "step": 3801
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.7861125463167236,
      "learning_rate": 1.4871840944549596e-05,
      "loss": 0.4097,
      "step": 3802
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.5277839203710883,
      "learning_rate": 1.4869178110767777e-05,
      "loss": 0.3291,
      "step": 3803
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.5526239060920835,
      "learning_rate": 1.4866514824355572e-05,
      "loss": 0.3806,
      "step": 3804
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.53132160764535,
      "learning_rate": 1.4863851085560563e-05,
      "loss": 0.3008,
      "step": 3805
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.750744777051714,
      "learning_rate": 1.4861186894630359e-05,
      "loss": 0.3823,
      "step": 3806
    },
    {
      "epoch": 1.08,
      "grad_norm": 2.240084183708366,
      "learning_rate": 1.4858522251812621e-05,
      "loss": 0.5298,
      "step": 3807
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.6065590186550764,
      "learning_rate": 1.485585715735505e-05,
      "loss": 0.3506,
      "step": 3808
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.5071336258257855,
      "learning_rate": 1.4853191611505391e-05,
      "loss": 0.2974,
      "step": 3809
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.5082351399037102,
      "learning_rate": 1.4850525614511427e-05,
      "loss": 0.3262,
      "step": 3810
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.5387848560196522,
      "learning_rate": 1.484785916662098e-05,
      "loss": 0.3574,
      "step": 3811
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.5892107340990813,
      "learning_rate": 1.4845192268081924e-05,
      "loss": 0.3428,
      "step": 3812
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.6182644239002615,
      "learning_rate": 1.4842524919142164e-05,
      "loss": 0.4238,
      "step": 3813
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.5913385935169047,
      "learning_rate": 1.4839857120049658e-05,
      "loss": 0.3691,
      "step": 3814
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.7272722248553924,
      "learning_rate": 1.4837188871052399e-05,
      "loss": 0.3657,
      "step": 3815
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.7895294675319324,
      "learning_rate": 1.4834520172398421e-05,
      "loss": 0.4355,
      "step": 3816
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.6107789736691744,
      "learning_rate": 1.4831851024335799e-05,
      "loss": 0.4031,
      "step": 3817
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.803160827837147,
      "learning_rate": 1.482918142711266e-05,
      "loss": 0.3464,
      "step": 3818
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.4310936002636991,
      "learning_rate": 1.4826511380977155e-05,
      "loss": 0.2861,
      "step": 3819
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.6803278144384919,
      "learning_rate": 1.4823840886177494e-05,
      "loss": 0.3423,
      "step": 3820
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.6687774080461648,
      "learning_rate": 1.4821169942961922e-05,
      "loss": 0.3721,
      "step": 3821
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.615007649282075,
      "learning_rate": 1.4818498551578721e-05,
      "loss": 0.3706,
      "step": 3822
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.629652189717176,
      "learning_rate": 1.481582671227622e-05,
      "loss": 0.3713,
      "step": 3823
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.6482548976972111,
      "learning_rate": 1.4813154425302791e-05,
      "loss": 0.3384,
      "step": 3824
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.9698550472619143,
      "learning_rate": 1.4810481690906846e-05,
      "loss": 0.4067,
      "step": 3825
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.7315985019828277,
      "learning_rate": 1.4807808509336831e-05,
      "loss": 0.3552,
      "step": 3826
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.7642997228613853,
      "learning_rate": 1.480513488084125e-05,
      "loss": 0.355,
      "step": 3827
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.3862448916390764,
      "learning_rate": 1.480246080566863e-05,
      "loss": 0.3181,
      "step": 3828
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.5730378096207727,
      "learning_rate": 1.4799786284067554e-05,
      "loss": 0.3633,
      "step": 3829
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.3733216828887953,
      "learning_rate": 1.4797111316286639e-05,
      "loss": 0.2947,
      "step": 3830
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.6085240734468007,
      "learning_rate": 1.4794435902574543e-05,
      "loss": 0.3555,
      "step": 3831
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.5612167306247238,
      "learning_rate": 1.4791760043179975e-05,
      "loss": 0.3887,
      "step": 3832
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.556765240548713,
      "learning_rate": 1.478908373835167e-05,
      "loss": 0.4038,
      "step": 3833
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.6108499818561042,
      "learning_rate": 1.4786406988338415e-05,
      "loss": 0.3096,
      "step": 3834
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.601629732605131,
      "learning_rate": 1.4783729793389043e-05,
      "loss": 0.3115,
      "step": 3835
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.9504894810972064,
      "learning_rate": 1.4781052153752411e-05,
      "loss": 0.4473,
      "step": 3836
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.964927536014361,
      "learning_rate": 1.4778374069677435e-05,
      "loss": 0.3955,
      "step": 3837
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.5005293895210394,
      "learning_rate": 1.4775695541413063e-05,
      "loss": 0.3745,
      "step": 3838
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.6091638986457337,
      "learning_rate": 1.4773016569208283e-05,
      "loss": 0.3555,
      "step": 3839
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.6427779421417585,
      "learning_rate": 1.4770337153312131e-05,
      "loss": 0.3359,
      "step": 3840
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.587840021506337,
      "learning_rate": 1.476765729397368e-05,
      "loss": 0.3901,
      "step": 3841
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.5655466501826356,
      "learning_rate": 1.4764976991442045e-05,
      "loss": 0.3708,
      "step": 3842
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.7249325041181662,
      "learning_rate": 1.4762296245966387e-05,
      "loss": 0.3726,
      "step": 3843
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.4741909847338701,
      "learning_rate": 1.4759615057795895e-05,
      "loss": 0.2527,
      "step": 3844
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.564602903635674,
      "learning_rate": 1.4756933427179814e-05,
      "loss": 0.3167,
      "step": 3845
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.4913570141402313,
      "learning_rate": 1.475425135436742e-05,
      "loss": 0.3262,
      "step": 3846
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.305686651053948,
      "learning_rate": 1.4751568839608036e-05,
      "loss": 0.2427,
      "step": 3847
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.637362927053281,
      "learning_rate": 1.4748885883151028e-05,
      "loss": 0.3076,
      "step": 3848
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.6963543701310972,
      "learning_rate": 1.4746202485245789e-05,
      "loss": 0.334,
      "step": 3849
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.9364329015173343,
      "learning_rate": 1.474351864614177e-05,
      "loss": 0.4141,
      "step": 3850
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.8269309585069815,
      "learning_rate": 1.4740834366088454e-05,
      "loss": 0.4014,
      "step": 3851
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.8581888050572524,
      "learning_rate": 1.4738149645335369e-05,
      "loss": 0.3862,
      "step": 3852
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.692562331732015,
      "learning_rate": 1.4735464484132079e-05,
      "loss": 0.4316,
      "step": 3853
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.792115812029989,
      "learning_rate": 1.4732778882728193e-05,
      "loss": 0.355,
      "step": 3854
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.4280637159971863,
      "learning_rate": 1.4730092841373362e-05,
      "loss": 0.2817,
      "step": 3855
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.5640663631225884,
      "learning_rate": 1.4727406360317274e-05,
      "loss": 0.3335,
      "step": 3856
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.7359565343329029,
      "learning_rate": 1.4724719439809659e-05,
      "loss": 0.3657,
      "step": 3857
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.687069966090469,
      "learning_rate": 1.472203208010029e-05,
      "loss": 0.4106,
      "step": 3858
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.7373017458282232,
      "learning_rate": 1.4719344281438977e-05,
      "loss": 0.3804,
      "step": 3859
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.8232885620955417,
      "learning_rate": 1.4716656044075577e-05,
      "loss": 0.3335,
      "step": 3860
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.6690150803661115,
      "learning_rate": 1.4713967368259981e-05,
      "loss": 0.3315,
      "step": 3861
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.5954030597010582,
      "learning_rate": 1.471127825424212e-05,
      "loss": 0.3643,
      "step": 3862
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.5338358490492356,
      "learning_rate": 1.4708588702271978e-05,
      "loss": 0.3853,
      "step": 3863
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.5273483762506026,
      "learning_rate": 1.4705898712599563e-05,
      "loss": 0.2842,
      "step": 3864
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.6504788564704702,
      "learning_rate": 1.4703208285474942e-05,
      "loss": 0.3643,
      "step": 3865
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.819355959868359,
      "learning_rate": 1.4700517421148199e-05,
      "loss": 0.3574,
      "step": 3866
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.6929474031612466,
      "learning_rate": 1.4697826119869483e-05,
      "loss": 0.3657,
      "step": 3867
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.6229566722439859,
      "learning_rate": 1.4695134381888969e-05,
      "loss": 0.322,
      "step": 3868
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.606018086446139,
      "learning_rate": 1.4692442207456875e-05,
      "loss": 0.2922,
      "step": 3869
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.7833377816089147,
      "learning_rate": 1.468974959682346e-05,
      "loss": 0.4746,
      "step": 3870
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.659295389937837,
      "learning_rate": 1.468705655023903e-05,
      "loss": 0.3115,
      "step": 3871
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.5493813184574905,
      "learning_rate": 1.468436306795392e-05,
      "loss": 0.3296,
      "step": 3872
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.479189803413764,
      "learning_rate": 1.4681669150218516e-05,
      "loss": 0.2983,
      "step": 3873
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.797906312999725,
      "learning_rate": 1.4678974797283235e-05,
      "loss": 0.3613,
      "step": 3874
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.527678119890287,
      "learning_rate": 1.4676280009398544e-05,
      "loss": 0.3711,
      "step": 3875
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.5193729814573802,
      "learning_rate": 1.4673584786814943e-05,
      "loss": 0.2717,
      "step": 3876
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.6538557931449045,
      "learning_rate": 1.4670889129782976e-05,
      "loss": 0.3618,
      "step": 3877
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.4348797646457236,
      "learning_rate": 1.4668193038553229e-05,
      "loss": 0.3176,
      "step": 3878
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.6980093167058228,
      "learning_rate": 1.466549651337632e-05,
      "loss": 0.3499,
      "step": 3879
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.477292876230161,
      "learning_rate": 1.466279955450292e-05,
      "loss": 0.3105,
      "step": 3880
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.757819149133433,
      "learning_rate": 1.4660102162183732e-05,
      "loss": 0.3916,
      "step": 3881
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.6093240019649697,
      "learning_rate": 1.4657404336669498e-05,
      "loss": 0.3367,
      "step": 3882
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.8703687579396622,
      "learning_rate": 1.4654706078211003e-05,
      "loss": 0.3711,
      "step": 3883
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.678383110503484,
      "learning_rate": 1.4652007387059077e-05,
      "loss": 0.3633,
      "step": 3884
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.5687935949289813,
      "learning_rate": 1.4649308263464583e-05,
      "loss": 0.3281,
      "step": 3885
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.7426838006966132,
      "learning_rate": 1.4646608707678428e-05,
      "loss": 0.3867,
      "step": 3886
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.7704944150785176,
      "learning_rate": 1.4643908719951551e-05,
      "loss": 0.4004,
      "step": 3887
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.6183382472160597,
      "learning_rate": 1.464120830053495e-05,
      "loss": 0.3862,
      "step": 3888
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.7823778833047752,
      "learning_rate": 1.4638507449679642e-05,
      "loss": 0.376,
      "step": 3889
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.5940025755061036,
      "learning_rate": 1.4635806167636698e-05,
      "loss": 0.301,
      "step": 3890
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.637550607487515,
      "learning_rate": 1.4633104454657225e-05,
      "loss": 0.3491,
      "step": 3891
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.6905120978872277,
      "learning_rate": 1.4630402310992367e-05,
      "loss": 0.3936,
      "step": 3892
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.7098887309722592,
      "learning_rate": 1.4627699736893309e-05,
      "loss": 0.3433,
      "step": 3893
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.4423974071775254,
      "learning_rate": 1.4624996732611284e-05,
      "loss": 0.2949,
      "step": 3894
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.6452326187634865,
      "learning_rate": 1.4622293298397554e-05,
      "loss": 0.3594,
      "step": 3895
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.481731626026696,
      "learning_rate": 1.4619589434503426e-05,
      "loss": 0.3291,
      "step": 3896
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.6269644593408583,
      "learning_rate": 1.4616885141180244e-05,
      "loss": 0.2939,
      "step": 3897
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.6744051665251087,
      "learning_rate": 1.46141804186794e-05,
      "loss": 0.3435,
      "step": 3898
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.504277569225843,
      "learning_rate": 1.4611475267252318e-05,
      "loss": 0.3154,
      "step": 3899
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.69773055566041,
      "learning_rate": 1.4608769687150459e-05,
      "loss": 0.3359,
      "step": 3900
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.8385616544459384,
      "learning_rate": 1.460606367862534e-05,
      "loss": 0.4395,
      "step": 3901
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.3478879147075022,
      "learning_rate": 1.4603357241928499e-05,
      "loss": 0.272,
      "step": 3902
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.4969562875536746,
      "learning_rate": 1.4600650377311523e-05,
      "loss": 0.3276,
      "step": 3903
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.7318196583190018,
      "learning_rate": 1.4597943085026037e-05,
      "loss": 0.3618,
      "step": 3904
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.8733717197231408,
      "learning_rate": 1.459523536532371e-05,
      "loss": 0.4258,
      "step": 3905
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.7376792519544766,
      "learning_rate": 1.459252721845624e-05,
      "loss": 0.3867,
      "step": 3906
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.4352277360422847,
      "learning_rate": 1.4589818644675378e-05,
      "loss": 0.3179,
      "step": 3907
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.7197603967886468,
      "learning_rate": 1.4587109644232906e-05,
      "loss": 0.4233,
      "step": 3908
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.4583301018343358,
      "learning_rate": 1.4584400217380647e-05,
      "loss": 0.3047,
      "step": 3909
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.6896282991065632,
      "learning_rate": 1.4581690364370466e-05,
      "loss": 0.3574,
      "step": 3910
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.5228159605709277,
      "learning_rate": 1.4578980085454268e-05,
      "loss": 0.3105,
      "step": 3911
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.715181063861323,
      "learning_rate": 1.4576269380883992e-05,
      "loss": 0.3945,
      "step": 3912
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.7562867357172363,
      "learning_rate": 1.4573558250911624e-05,
      "loss": 0.3306,
      "step": 3913
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.9582183673812603,
      "learning_rate": 1.457084669578918e-05,
      "loss": 0.4487,
      "step": 3914
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.482862578189193,
      "learning_rate": 1.4568134715768727e-05,
      "loss": 0.3237,
      "step": 3915
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.699540406657032,
      "learning_rate": 1.4565422311102367e-05,
      "loss": 0.3159,
      "step": 3916
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.6723637152258548,
      "learning_rate": 1.4562709482042237e-05,
      "loss": 0.3677,
      "step": 3917
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.6296766729402452,
      "learning_rate": 1.4559996228840518e-05,
      "loss": 0.3652,
      "step": 3918
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.6325266088564305,
      "learning_rate": 1.4557282551749428e-05,
      "loss": 0.353,
      "step": 3919
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.5933137706687357,
      "learning_rate": 1.455456845102123e-05,
      "loss": 0.3655,
      "step": 3920
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.4828935775883048,
      "learning_rate": 1.4551853926908215e-05,
      "loss": 0.304,
      "step": 3921
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.5760868337981802,
      "learning_rate": 1.454913897966273e-05,
      "loss": 0.3406,
      "step": 3922
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.5977317248434888,
      "learning_rate": 1.454642360953714e-05,
      "loss": 0.3118,
      "step": 3923
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.8047797690181202,
      "learning_rate": 1.454370781678387e-05,
      "loss": 0.3079,
      "step": 3924
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.8566816570409495,
      "learning_rate": 1.4540991601655374e-05,
      "loss": 0.397,
      "step": 3925
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.9405495678178146,
      "learning_rate": 1.4538274964404146e-05,
      "loss": 0.3723,
      "step": 3926
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.8447016850049227,
      "learning_rate": 1.4535557905282716e-05,
      "loss": 0.3628,
      "step": 3927
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.737749095898118,
      "learning_rate": 1.4532840424543664e-05,
      "loss": 0.3403,
      "step": 3928
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.7614455933361683,
      "learning_rate": 1.4530122522439598e-05,
      "loss": 0.3792,
      "step": 3929
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.6120064718146716,
      "learning_rate": 1.4527404199223173e-05,
      "loss": 0.2681,
      "step": 3930
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.5174513522930206,
      "learning_rate": 1.4524685455147071e-05,
      "loss": 0.2764,
      "step": 3931
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.7901431979681957,
      "learning_rate": 1.4521966290464033e-05,
      "loss": 0.417,
      "step": 3932
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.6948911654412129,
      "learning_rate": 1.4519246705426822e-05,
      "loss": 0.3606,
      "step": 3933
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.8479271080441462,
      "learning_rate": 1.4516526700288243e-05,
      "loss": 0.4053,
      "step": 3934
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.5945682861271369,
      "learning_rate": 1.451380627530115e-05,
      "loss": 0.2974,
      "step": 3935
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.618861684364816,
      "learning_rate": 1.4511085430718423e-05,
      "loss": 0.3569,
      "step": 3936
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.731184524452198,
      "learning_rate": 1.4508364166792993e-05,
      "loss": 0.373,
      "step": 3937
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.630560733196001,
      "learning_rate": 1.450564248377782e-05,
      "loss": 0.3584,
      "step": 3938
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.6053927190652655,
      "learning_rate": 1.4502920381925905e-05,
      "loss": 0.3369,
      "step": 3939
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.610987847649454,
      "learning_rate": 1.4500197861490293e-05,
      "loss": 0.293,
      "step": 3940
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.5590580796270346,
      "learning_rate": 1.4497474922724064e-05,
      "loss": 0.3174,
      "step": 3941
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.6708342916752486,
      "learning_rate": 1.4494751565880338e-05,
      "loss": 0.396,
      "step": 3942
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.4581649611298855,
      "learning_rate": 1.4492027791212275e-05,
      "loss": 0.2944,
      "step": 3943
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.8579966406111537,
      "learning_rate": 1.4489303598973067e-05,
      "loss": 0.4116,
      "step": 3944
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.4409428989462787,
      "learning_rate": 1.448657898941596e-05,
      "loss": 0.2808,
      "step": 3945
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.622583969009655,
      "learning_rate": 1.4483853962794218e-05,
      "loss": 0.4214,
      "step": 3946
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.7023334012364377,
      "learning_rate": 1.4481128519361163e-05,
      "loss": 0.4219,
      "step": 3947
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.6450941265201995,
      "learning_rate": 1.4478402659370144e-05,
      "loss": 0.3154,
      "step": 3948
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.351041383085924,
      "learning_rate": 1.447567638307455e-05,
      "loss": 0.3105,
      "step": 3949
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.656082230810699,
      "learning_rate": 1.4472949690727813e-05,
      "loss": 0.3188,
      "step": 3950
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.647610583950766,
      "learning_rate": 1.4470222582583404e-05,
      "loss": 0.3667,
      "step": 3951
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.676463191827808,
      "learning_rate": 1.4467495058894829e-05,
      "loss": 0.3506,
      "step": 3952
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.7324081727242424,
      "learning_rate": 1.446476711991563e-05,
      "loss": 0.3564,
      "step": 3953
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.7126244620378708,
      "learning_rate": 1.4462038765899397e-05,
      "loss": 0.3188,
      "step": 3954
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.6133572826904448,
      "learning_rate": 1.445930999709975e-05,
      "loss": 0.3262,
      "step": 3955
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.9139446678827947,
      "learning_rate": 1.4456580813770353e-05,
      "loss": 0.3916,
      "step": 3956
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.6382832115296615,
      "learning_rate": 1.4453851216164902e-05,
      "loss": 0.3672,
      "step": 3957
    },
    {
      "epoch": 1.12,
      "grad_norm": 2.0797515926079675,
      "learning_rate": 1.4451121204537144e-05,
      "loss": 0.4497,
      "step": 3958
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.4346443678417315,
      "learning_rate": 1.4448390779140844e-05,
      "loss": 0.2854,
      "step": 3959
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.6061675621654836,
      "learning_rate": 1.4445659940229827e-05,
      "loss": 0.3647,
      "step": 3960
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.4469723221404585,
      "learning_rate": 1.4442928688057945e-05,
      "loss": 0.3008,
      "step": 3961
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.7825914393653854,
      "learning_rate": 1.4440197022879088e-05,
      "loss": 0.314,
      "step": 3962
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.69515257081763,
      "learning_rate": 1.4437464944947186e-05,
      "loss": 0.3721,
      "step": 3963
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.6319889638185487,
      "learning_rate": 1.4434732454516214e-05,
      "loss": 0.3369,
      "step": 3964
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.721570843889218,
      "learning_rate": 1.4431999551840175e-05,
      "loss": 0.3818,
      "step": 3965
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.6517907805203043,
      "learning_rate": 1.4429266237173116e-05,
      "loss": 0.3052,
      "step": 3966
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.6349933992771135,
      "learning_rate": 1.442653251076912e-05,
      "loss": 0.3208,
      "step": 3967
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.6316649792045554,
      "learning_rate": 1.4423798372882315e-05,
      "loss": 0.377,
      "step": 3968
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.5136183286702412,
      "learning_rate": 1.4421063823766855e-05,
      "loss": 0.2896,
      "step": 3969
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.6570125400984987,
      "learning_rate": 1.441832886367694e-05,
      "loss": 0.3564,
      "step": 3970
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.9554980815305372,
      "learning_rate": 1.441559349286681e-05,
      "loss": 0.4126,
      "step": 3971
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.5641359652875682,
      "learning_rate": 1.4412857711590742e-05,
      "loss": 0.3042,
      "step": 3972
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.626712552689843,
      "learning_rate": 1.4410121520103045e-05,
      "loss": 0.3735,
      "step": 3973
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.5055493764686394,
      "learning_rate": 1.440738491865807e-05,
      "loss": 0.3447,
      "step": 3974
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.849567868346941,
      "learning_rate": 1.4404647907510212e-05,
      "loss": 0.4072,
      "step": 3975
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.578825928071026,
      "learning_rate": 1.4401910486913892e-05,
      "loss": 0.2732,
      "step": 3976
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.6198401985276634,
      "learning_rate": 1.4399172657123584e-05,
      "loss": 0.3579,
      "step": 3977
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.8607581238266666,
      "learning_rate": 1.4396434418393786e-05,
      "loss": 0.4199,
      "step": 3978
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.8622798930186057,
      "learning_rate": 1.4393695770979038e-05,
      "loss": 0.4146,
      "step": 3979
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.434545684442707,
      "learning_rate": 1.4390956715133928e-05,
      "loss": 0.2668,
      "step": 3980
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.6202656954243893,
      "learning_rate": 1.4388217251113068e-05,
      "loss": 0.3633,
      "step": 3981
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.6750964072102912,
      "learning_rate": 1.4385477379171116e-05,
      "loss": 0.3325,
      "step": 3982
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.5555037027154448,
      "learning_rate": 1.4382737099562765e-05,
      "loss": 0.3608,
      "step": 3983
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.5773099916080826,
      "learning_rate": 1.4379996412542742e-05,
      "loss": 0.3242,
      "step": 3984
    },
    {
      "epoch": 1.13,
      "grad_norm": 2.029974682897359,
      "learning_rate": 1.4377255318365828e-05,
      "loss": 0.4414,
      "step": 3985
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.5658326000100582,
      "learning_rate": 1.4374513817286819e-05,
      "loss": 0.3357,
      "step": 3986
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.5516419474582372,
      "learning_rate": 1.4371771909560566e-05,
      "loss": 0.355,
      "step": 3987
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.7494442651294018,
      "learning_rate": 1.4369029595441953e-05,
      "loss": 0.4111,
      "step": 3988
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.4149525341563163,
      "learning_rate": 1.4366286875185895e-05,
      "loss": 0.3179,
      "step": 3989
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.7849612808331072,
      "learning_rate": 1.4363543749047354e-05,
      "loss": 0.4751,
      "step": 3990
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.637013798339755,
      "learning_rate": 1.436080021728133e-05,
      "loss": 0.3418,
      "step": 3991
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.4086017479702184,
      "learning_rate": 1.4358056280142851e-05,
      "loss": 0.3013,
      "step": 3992
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.6163518526507366,
      "learning_rate": 1.4355311937886988e-05,
      "loss": 0.3237,
      "step": 3993
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.5213817010849588,
      "learning_rate": 1.4352567190768859e-05,
      "loss": 0.3604,
      "step": 3994
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.5010562703258663,
      "learning_rate": 1.4349822039043602e-05,
      "loss": 0.3052,
      "step": 3995
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.627405416171989,
      "learning_rate": 1.4347076482966404e-05,
      "loss": 0.3218,
      "step": 3996
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.4814469002998314,
      "learning_rate": 1.4344330522792489e-05,
      "loss": 0.2964,
      "step": 3997
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.4805905067267153,
      "learning_rate": 1.4341584158777115e-05,
      "loss": 0.3162,
      "step": 3998
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.7710872559001831,
      "learning_rate": 1.4338837391175582e-05,
      "loss": 0.3711,
      "step": 3999
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.5880435164795006,
      "learning_rate": 1.4336090220243222e-05,
      "loss": 0.2988,
      "step": 4000
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.7708513399889318,
      "learning_rate": 1.4333342646235407e-05,
      "loss": 0.3589,
      "step": 4001
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.735097146107318,
      "learning_rate": 1.433059466940755e-05,
      "loss": 0.3516,
      "step": 4002
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.6358657540393386,
      "learning_rate": 1.4327846290015094e-05,
      "loss": 0.3081,
      "step": 4003
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.6474009055214687,
      "learning_rate": 1.4325097508313524e-05,
      "loss": 0.3447,
      "step": 4004
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.724720220671931,
      "learning_rate": 1.4322348324558366e-05,
      "loss": 0.396,
      "step": 4005
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.868754224429411,
      "learning_rate": 1.4319598739005174e-05,
      "loss": 0.3916,
      "step": 4006
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.5488734194483735,
      "learning_rate": 1.431684875190955e-05,
      "loss": 0.3479,
      "step": 4007
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.5215314062681207,
      "learning_rate": 1.4314098363527122e-05,
      "loss": 0.3042,
      "step": 4008
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.6566889266006213,
      "learning_rate": 1.431134757411357e-05,
      "loss": 0.3281,
      "step": 4009
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.485370493864094,
      "learning_rate": 1.4308596383924593e-05,
      "loss": 0.3254,
      "step": 4010
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.5593954051419314,
      "learning_rate": 1.4305844793215943e-05,
      "loss": 0.3408,
      "step": 4011
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.2794177959077988,
      "learning_rate": 1.4303092802243404e-05,
      "loss": 0.2568,
      "step": 4012
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.5312134710525749,
      "learning_rate": 1.4300340411262792e-05,
      "loss": 0.4062,
      "step": 4013
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.722503697968406,
      "learning_rate": 1.4297587620529965e-05,
      "loss": 0.3477,
      "step": 4014
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.479281597008075,
      "learning_rate": 1.4294834430300822e-05,
      "loss": 0.272,
      "step": 4015
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.6806864165023803,
      "learning_rate": 1.4292080840831289e-05,
      "loss": 0.3042,
      "step": 4016
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.7012567302306372,
      "learning_rate": 1.4289326852377339e-05,
      "loss": 0.3376,
      "step": 4017
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.609101694359922,
      "learning_rate": 1.4286572465194977e-05,
      "loss": 0.3853,
      "step": 4018
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.680913994449485,
      "learning_rate": 1.4283817679540246e-05,
      "loss": 0.3357,
      "step": 4019
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.630238737875649,
      "learning_rate": 1.4281062495669224e-05,
      "loss": 0.3433,
      "step": 4020
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.7857923483594165,
      "learning_rate": 1.4278306913838034e-05,
      "loss": 0.3413,
      "step": 4021
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.6949066503294576,
      "learning_rate": 1.4275550934302822e-05,
      "loss": 0.3164,
      "step": 4022
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.6258622730140413,
      "learning_rate": 1.4272794557319785e-05,
      "loss": 0.3579,
      "step": 4023
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.357339479884339,
      "learning_rate": 1.427003778314515e-05,
      "loss": 0.2603,
      "step": 4024
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.8577590952767942,
      "learning_rate": 1.426728061203518e-05,
      "loss": 0.4043,
      "step": 4025
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.4629551690298368,
      "learning_rate": 1.426452304424618e-05,
      "loss": 0.3069,
      "step": 4026
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.3629628105310612,
      "learning_rate": 1.4261765080034487e-05,
      "loss": 0.2659,
      "step": 4027
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.9245772139717843,
      "learning_rate": 1.4259006719656476e-05,
      "loss": 0.3716,
      "step": 4028
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.8140398463114957,
      "learning_rate": 1.425624796336856e-05,
      "loss": 0.3408,
      "step": 4029
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.4254047938837613,
      "learning_rate": 1.4253488811427188e-05,
      "loss": 0.2656,
      "step": 4030
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.670039365304275,
      "learning_rate": 1.4250729264088845e-05,
      "loss": 0.3486,
      "step": 4031
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.6058104545214291,
      "learning_rate": 1.4247969321610055e-05,
      "loss": 0.3457,
      "step": 4032
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.7280729144031555,
      "learning_rate": 1.4245208984247378e-05,
      "loss": 0.3184,
      "step": 4033
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.591270775976831,
      "learning_rate": 1.424244825225741e-05,
      "loss": 0.3369,
      "step": 4034
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.7735489674780218,
      "learning_rate": 1.4239687125896783e-05,
      "loss": 0.3611,
      "step": 4035
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.7755888185889843,
      "learning_rate": 1.423692560542217e-05,
      "loss": 0.3477,
      "step": 4036
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.5493350790692657,
      "learning_rate": 1.4234163691090269e-05,
      "loss": 0.2998,
      "step": 4037
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.570707952500652,
      "learning_rate": 1.423140138315783e-05,
      "loss": 0.302,
      "step": 4038
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.7676912017458228,
      "learning_rate": 1.4228638681881633e-05,
      "loss": 0.418,
      "step": 4039
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.7651726306716402,
      "learning_rate": 1.4225875587518485e-05,
      "loss": 0.3921,
      "step": 4040
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.667520079274115,
      "learning_rate": 1.422311210032525e-05,
      "loss": 0.3206,
      "step": 4041
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.497201160913538,
      "learning_rate": 1.422034822055881e-05,
      "loss": 0.3196,
      "step": 4042
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.670013248859228,
      "learning_rate": 1.4217583948476094e-05,
      "loss": 0.3252,
      "step": 4043
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.611427983907567,
      "learning_rate": 1.4214819284334058e-05,
      "loss": 0.395,
      "step": 4044
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.6449261910564716,
      "learning_rate": 1.4212054228389712e-05,
      "loss": 0.2949,
      "step": 4045
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.4785461629723833,
      "learning_rate": 1.4209288780900074e-05,
      "loss": 0.2961,
      "step": 4046
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.5586901105456201,
      "learning_rate": 1.4206522942122232e-05,
      "loss": 0.2966,
      "step": 4047
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.6422859281790603,
      "learning_rate": 1.4203756712313284e-05,
      "loss": 0.3677,
      "step": 4048
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.83228264157126,
      "learning_rate": 1.4200990091730376e-05,
      "loss": 0.438,
      "step": 4049
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.5494686177360901,
      "learning_rate": 1.4198223080630686e-05,
      "loss": 0.2898,
      "step": 4050
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.7585732412305266,
      "learning_rate": 1.4195455679271435e-05,
      "loss": 0.3572,
      "step": 4051
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.5435927470915072,
      "learning_rate": 1.4192687887909875e-05,
      "loss": 0.3384,
      "step": 4052
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.3472551881387038,
      "learning_rate": 1.4189919706803292e-05,
      "loss": 0.2566,
      "step": 4053
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.6223865142846745,
      "learning_rate": 1.4187151136209015e-05,
      "loss": 0.3789,
      "step": 4054
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.6249289332467596,
      "learning_rate": 1.4184382176384403e-05,
      "loss": 0.3137,
      "step": 4055
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.7131742492550466,
      "learning_rate": 1.4181612827586854e-05,
      "loss": 0.312,
      "step": 4056
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.7850386261659474,
      "learning_rate": 1.4178843090073802e-05,
      "loss": 0.356,
      "step": 4057
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.6155756646539199,
      "learning_rate": 1.4176072964102719e-05,
      "loss": 0.3215,
      "step": 4058
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.6803336603684456,
      "learning_rate": 1.4173302449931107e-05,
      "loss": 0.3369,
      "step": 4059
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.6312633349364154,
      "learning_rate": 1.4170531547816513e-05,
      "loss": 0.3335,
      "step": 4060
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.60485419670689,
      "learning_rate": 1.4167760258016514e-05,
      "loss": 0.3477,
      "step": 4061
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.647674596512531,
      "learning_rate": 1.4164988580788723e-05,
      "loss": 0.3293,
      "step": 4062
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.4157536864243194,
      "learning_rate": 1.4162216516390787e-05,
      "loss": 0.2495,
      "step": 4063
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.8091520965836432,
      "learning_rate": 1.4159444065080398e-05,
      "loss": 0.3594,
      "step": 4064
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.5407982847830168,
      "learning_rate": 1.4156671227115278e-05,
      "loss": 0.3293,
      "step": 4065
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.7812329578557449,
      "learning_rate": 1.4153898002753184e-05,
      "loss": 0.4121,
      "step": 4066
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.7436351214063122,
      "learning_rate": 1.4151124392251905e-05,
      "loss": 0.3447,
      "step": 4067
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.8080385959206628,
      "learning_rate": 1.4148350395869279e-05,
      "loss": 0.4155,
      "step": 4068
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.6686117354737862,
      "learning_rate": 1.414557601386317e-05,
      "loss": 0.314,
      "step": 4069
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.65978623455919,
      "learning_rate": 1.4142801246491476e-05,
      "loss": 0.3242,
      "step": 4070
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.736126485067494,
      "learning_rate": 1.4140026094012136e-05,
      "loss": 0.3516,
      "step": 4071
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.5046916514295634,
      "learning_rate": 1.4137250556683129e-05,
      "loss": 0.2681,
      "step": 4072
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.5459813942247358,
      "learning_rate": 1.4134474634762454e-05,
      "loss": 0.2842,
      "step": 4073
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.6597146417865645,
      "learning_rate": 1.4131698328508163e-05,
      "loss": 0.3652,
      "step": 4074
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.6059104078878566,
      "learning_rate": 1.4128921638178336e-05,
      "loss": 0.2898,
      "step": 4075
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.5677258274791306,
      "learning_rate": 1.4126144564031085e-05,
      "loss": 0.3357,
      "step": 4076
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.7631750857573538,
      "learning_rate": 1.412336710632457e-05,
      "loss": 0.3892,
      "step": 4077
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.5668561678147601,
      "learning_rate": 1.4120589265316974e-05,
      "loss": 0.2852,
      "step": 4078
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.582495609469782,
      "learning_rate": 1.4117811041266518e-05,
      "loss": 0.3315,
      "step": 4079
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.5698292313543891,
      "learning_rate": 1.4115032434431461e-05,
      "loss": 0.3096,
      "step": 4080
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.7181545825701792,
      "learning_rate": 1.4112253445070103e-05,
      "loss": 0.3359,
      "step": 4081
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.534184776934453,
      "learning_rate": 1.410947407344077e-05,
      "loss": 0.3088,
      "step": 4082
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.7264852120080179,
      "learning_rate": 1.410669431980183e-05,
      "loss": 0.2979,
      "step": 4083
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.6338507977571617,
      "learning_rate": 1.410391418441168e-05,
      "loss": 0.2832,
      "step": 4084
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.4432382701988637,
      "learning_rate": 1.4101133667528761e-05,
      "loss": 0.3198,
      "step": 4085
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.4891327321610295,
      "learning_rate": 1.409835276941154e-05,
      "loss": 0.27,
      "step": 4086
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.5731110925284197,
      "learning_rate": 1.4095571490318532e-05,
      "loss": 0.3682,
      "step": 4087
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.5927946307481753,
      "learning_rate": 1.4092789830508271e-05,
      "loss": 0.3091,
      "step": 4088
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.67522576937605,
      "learning_rate": 1.4090007790239344e-05,
      "loss": 0.4033,
      "step": 4089
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.4718086630358047,
      "learning_rate": 1.4087225369770356e-05,
      "loss": 0.2959,
      "step": 4090
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.8880831192527485,
      "learning_rate": 1.4084442569359964e-05,
      "loss": 0.4282,
      "step": 4091
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.544709649665869,
      "learning_rate": 1.4081659389266846e-05,
      "loss": 0.2888,
      "step": 4092
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.8288708676058794,
      "learning_rate": 1.4078875829749724e-05,
      "loss": 0.3418,
      "step": 4093
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.7510912895175692,
      "learning_rate": 1.4076091891067355e-05,
      "loss": 0.3511,
      "step": 4094
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.363116684191321,
      "learning_rate": 1.4073307573478528e-05,
      "loss": 0.3047,
      "step": 4095
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.6706833079547552,
      "learning_rate": 1.4070522877242068e-05,
      "loss": 0.3853,
      "step": 4096
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.6092224465826863,
      "learning_rate": 1.4067737802616832e-05,
      "loss": 0.3315,
      "step": 4097
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.5624672847819572,
      "learning_rate": 1.406495234986172e-05,
      "loss": 0.417,
      "step": 4098
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.6561173400840388,
      "learning_rate": 1.4062166519235665e-05,
      "loss": 0.3818,
      "step": 4099
    },
    {
      "epoch": 1.16,
      "grad_norm": 2.0266638397982506,
      "learning_rate": 1.4059380310997626e-05,
      "loss": 0.4746,
      "step": 4100
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.6013567916421636,
      "learning_rate": 1.4056593725406608e-05,
      "loss": 0.3438,
      "step": 4101
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.606342310200096,
      "learning_rate": 1.4053806762721652e-05,
      "loss": 0.3467,
      "step": 4102
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.5604916099635402,
      "learning_rate": 1.405101942320182e-05,
      "loss": 0.3413,
      "step": 4103
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.7260911218793171,
      "learning_rate": 1.4048231707106222e-05,
      "loss": 0.373,
      "step": 4104
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.9722096526814443,
      "learning_rate": 1.4045443614694004e-05,
      "loss": 0.4561,
      "step": 4105
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.5868292921940697,
      "learning_rate": 1.4042655146224333e-05,
      "loss": 0.3086,
      "step": 4106
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.8449136187234958,
      "learning_rate": 1.4039866301956427e-05,
      "loss": 0.3979,
      "step": 4107
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.6154908867983502,
      "learning_rate": 1.4037077082149533e-05,
      "loss": 0.395,
      "step": 4108
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.7255070517381634,
      "learning_rate": 1.4034287487062928e-05,
      "loss": 0.373,
      "step": 4109
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.4870102562136684,
      "learning_rate": 1.403149751695593e-05,
      "loss": 0.3662,
      "step": 4110
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.5776592603532573,
      "learning_rate": 1.4028707172087885e-05,
      "loss": 0.3159,
      "step": 4111
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.5990324685218542,
      "learning_rate": 1.402591645271819e-05,
      "loss": 0.3848,
      "step": 4112
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.4203600398537417,
      "learning_rate": 1.4023125359106253e-05,
      "loss": 0.3149,
      "step": 4113
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.7461568265360043,
      "learning_rate": 1.4020333891511536e-05,
      "loss": 0.3325,
      "step": 4114
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.8011119388841323,
      "learning_rate": 1.4017542050193532e-05,
      "loss": 0.416,
      "step": 4115
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.6669276474623858,
      "learning_rate": 1.4014749835411754e-05,
      "loss": 0.3467,
      "step": 4116
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.3766454623076605,
      "learning_rate": 1.401195724742577e-05,
      "loss": 0.2966,
      "step": 4117
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.6093131705931263,
      "learning_rate": 1.4009164286495179e-05,
      "loss": 0.3608,
      "step": 4118
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.7025869341533233,
      "learning_rate": 1.40063709528796e-05,
      "loss": 0.4282,
      "step": 4119
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.497947087404865,
      "learning_rate": 1.40035772468387e-05,
      "loss": 0.2957,
      "step": 4120
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.7051701050567616,
      "learning_rate": 1.4000783168632178e-05,
      "loss": 0.3723,
      "step": 4121
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.6060323001701062,
      "learning_rate": 1.3997988718519766e-05,
      "loss": 0.3413,
      "step": 4122
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.7691982212920598,
      "learning_rate": 1.399519389676123e-05,
      "loss": 0.3782,
      "step": 4123
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.7124574119170926,
      "learning_rate": 1.3992398703616372e-05,
      "loss": 0.3423,
      "step": 4124
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.6350397197879551,
      "learning_rate": 1.3989603139345037e-05,
      "loss": 0.3499,
      "step": 4125
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.6028988004455782,
      "learning_rate": 1.3986807204207077e-05,
      "loss": 0.3423,
      "step": 4126
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.7618160172049047,
      "learning_rate": 1.3984010898462417e-05,
      "loss": 0.3391,
      "step": 4127
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.9993508021871995,
      "learning_rate": 1.3981214222370985e-05,
      "loss": 0.4248,
      "step": 4128
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.6084566660218824,
      "learning_rate": 1.397841717619276e-05,
      "loss": 0.3643,
      "step": 4129
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.6595606009016217,
      "learning_rate": 1.3975619760187746e-05,
      "loss": 0.3359,
      "step": 4130
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.6888590300039485,
      "learning_rate": 1.3972821974615991e-05,
      "loss": 0.3613,
      "step": 4131
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.4669001511187651,
      "learning_rate": 1.3970023819737573e-05,
      "loss": 0.3125,
      "step": 4132
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.5598227778789195,
      "learning_rate": 1.3967225295812595e-05,
      "loss": 0.356,
      "step": 4133
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.579268741678119,
      "learning_rate": 1.3964426403101212e-05,
      "loss": 0.2866,
      "step": 4134
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.7412635589557104,
      "learning_rate": 1.3961627141863603e-05,
      "loss": 0.4121,
      "step": 4135
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.8636331699517728,
      "learning_rate": 1.3958827512359979e-05,
      "loss": 0.3369,
      "step": 4136
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.7317073785265547,
      "learning_rate": 1.395602751485059e-05,
      "loss": 0.3628,
      "step": 4137
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.4323325401248537,
      "learning_rate": 1.3953227149595718e-05,
      "loss": 0.2698,
      "step": 4138
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.5725616736816763,
      "learning_rate": 1.3950426416855685e-05,
      "loss": 0.3164,
      "step": 4139
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.8300547513130403,
      "learning_rate": 1.3947625316890836e-05,
      "loss": 0.3926,
      "step": 4140
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.660681016850317,
      "learning_rate": 1.3944823849961557e-05,
      "loss": 0.3696,
      "step": 4141
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.3600755931712523,
      "learning_rate": 1.3942022016328276e-05,
      "loss": 0.2432,
      "step": 4142
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.704269477986042,
      "learning_rate": 1.3939219816251435e-05,
      "loss": 0.3271,
      "step": 4143
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.7896796352132442,
      "learning_rate": 1.3936417249991528e-05,
      "loss": 0.3813,
      "step": 4144
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.4664440531684715,
      "learning_rate": 1.3933614317809076e-05,
      "loss": 0.2568,
      "step": 4145
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.601537095197704,
      "learning_rate": 1.3930811019964633e-05,
      "loss": 0.3457,
      "step": 4146
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.5677700319580283,
      "learning_rate": 1.3928007356718792e-05,
      "loss": 0.2966,
      "step": 4147
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.6053979654600001,
      "learning_rate": 1.3925203328332173e-05,
      "loss": 0.3792,
      "step": 4148
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.739888161440692,
      "learning_rate": 1.3922398935065437e-05,
      "loss": 0.3408,
      "step": 4149
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.776161898912191,
      "learning_rate": 1.3919594177179272e-05,
      "loss": 0.3608,
      "step": 4150
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.5705830126173876,
      "learning_rate": 1.3916789054934408e-05,
      "loss": 0.2905,
      "step": 4151
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.676161795401396,
      "learning_rate": 1.39139835685916e-05,
      "loss": 0.3057,
      "step": 4152
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.8652711839779101,
      "learning_rate": 1.3911177718411643e-05,
      "loss": 0.3364,
      "step": 4153
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.6963370010541343,
      "learning_rate": 1.3908371504655365e-05,
      "loss": 0.3159,
      "step": 4154
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.663819264271806,
      "learning_rate": 1.3905564927583625e-05,
      "loss": 0.3306,
      "step": 4155
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.7351619392408295,
      "learning_rate": 1.3902757987457318e-05,
      "loss": 0.3315,
      "step": 4156
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.5776321352364286,
      "learning_rate": 1.3899950684537371e-05,
      "loss": 0.3435,
      "step": 4157
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.533334888274493,
      "learning_rate": 1.3897143019084749e-05,
      "loss": 0.3472,
      "step": 4158
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.694385188525624,
      "learning_rate": 1.3894334991360448e-05,
      "loss": 0.3042,
      "step": 4159
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.5232227182571076,
      "learning_rate": 1.3891526601625492e-05,
      "loss": 0.3145,
      "step": 4160
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.6169575013912691,
      "learning_rate": 1.388871785014095e-05,
      "loss": 0.293,
      "step": 4161
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.5664911732642246,
      "learning_rate": 1.3885908737167918e-05,
      "loss": 0.3025,
      "step": 4162
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.7598610331884288,
      "learning_rate": 1.3883099262967521e-05,
      "loss": 0.3926,
      "step": 4163
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.2463932927739603,
      "learning_rate": 1.3880289427800925e-05,
      "loss": 0.2219,
      "step": 4164
    },
    {
      "epoch": 1.18,
      "grad_norm": 2.124774698363275,
      "learning_rate": 1.3877479231929333e-05,
      "loss": 0.3691,
      "step": 4165
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.5975055214946188,
      "learning_rate": 1.3874668675613972e-05,
      "loss": 0.3257,
      "step": 4166
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.706088862279809,
      "learning_rate": 1.3871857759116104e-05,
      "loss": 0.376,
      "step": 4167
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.9706665357416815,
      "learning_rate": 1.386904648269703e-05,
      "loss": 0.5156,
      "step": 4168
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.6403138526455823,
      "learning_rate": 1.3866234846618083e-05,
      "loss": 0.3774,
      "step": 4169
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.580274216309871,
      "learning_rate": 1.3863422851140624e-05,
      "loss": 0.2896,
      "step": 4170
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.5681057081761385,
      "learning_rate": 1.386061049652605e-05,
      "loss": 0.3394,
      "step": 4171
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.6855351799749079,
      "learning_rate": 1.3857797783035803e-05,
      "loss": 0.3259,
      "step": 4172
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.6569875307395379,
      "learning_rate": 1.3854984710931332e-05,
      "loss": 0.335,
      "step": 4173
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.6572369362339907,
      "learning_rate": 1.3852171280474147e-05,
      "loss": 0.3389,
      "step": 4174
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.6667830093959641,
      "learning_rate": 1.3849357491925779e-05,
      "loss": 0.3833,
      "step": 4175
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.8346821455010982,
      "learning_rate": 1.3846543345547787e-05,
      "loss": 0.3445,
      "step": 4176
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.6485845985580259,
      "learning_rate": 1.3843728841601771e-05,
      "loss": 0.3491,
      "step": 4177
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.712408215625194,
      "learning_rate": 1.3840913980349366e-05,
      "loss": 0.333,
      "step": 4178
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.4847163786333348,
      "learning_rate": 1.3838098762052237e-05,
      "loss": 0.2717,
      "step": 4179
    },
    {
      "epoch": 1.18,
      "grad_norm": 2.117481197745979,
      "learning_rate": 1.3835283186972077e-05,
      "loss": 0.4771,
      "step": 4180
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.7590078322112566,
      "learning_rate": 1.3832467255370618e-05,
      "loss": 0.3945,
      "step": 4181
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.778890251604596,
      "learning_rate": 1.3829650967509629e-05,
      "loss": 0.3818,
      "step": 4182
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.853497864026999,
      "learning_rate": 1.3826834323650899e-05,
      "loss": 0.3557,
      "step": 4183
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.8522139696172621,
      "learning_rate": 1.3824017324056264e-05,
      "loss": 0.3838,
      "step": 4184
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.6459347362779355,
      "learning_rate": 1.382119996898759e-05,
      "loss": 0.313,
      "step": 4185
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.665174003420284,
      "learning_rate": 1.3818382258706766e-05,
      "loss": 0.3093,
      "step": 4186
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.7070290836855935,
      "learning_rate": 1.381556419347573e-05,
      "loss": 0.3433,
      "step": 4187
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.5595816793142254,
      "learning_rate": 1.3812745773556438e-05,
      "loss": 0.3113,
      "step": 4188
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.706156958168763,
      "learning_rate": 1.3809926999210886e-05,
      "loss": 0.3398,
      "step": 4189
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.994306650635538,
      "learning_rate": 1.3807107870701102e-05,
      "loss": 0.4077,
      "step": 4190
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.7787269519311735,
      "learning_rate": 1.3804288388289152e-05,
      "loss": 0.4067,
      "step": 4191
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.6029665661479293,
      "learning_rate": 1.3801468552237127e-05,
      "loss": 0.3379,
      "step": 4192
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.5569050367307546,
      "learning_rate": 1.3798648362807154e-05,
      "loss": 0.281,
      "step": 4193
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.5549734545248022,
      "learning_rate": 1.3795827820261392e-05,
      "loss": 0.335,
      "step": 4194
    },
    {
      "epoch": 1.18,
      "grad_norm": 2.227311499212996,
      "learning_rate": 1.379300692486204e-05,
      "loss": 0.4502,
      "step": 4195
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.7399060899708911,
      "learning_rate": 1.3790185676871312e-05,
      "loss": 0.3257,
      "step": 4196
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.6288480196031763,
      "learning_rate": 1.3787364076551478e-05,
      "loss": 0.3076,
      "step": 4197
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.8211214659417467,
      "learning_rate": 1.3784542124164821e-05,
      "loss": 0.3228,
      "step": 4198
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.717150684516093,
      "learning_rate": 1.378171981997367e-05,
      "loss": 0.3438,
      "step": 4199
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.5540220588182063,
      "learning_rate": 1.3778897164240378e-05,
      "loss": 0.3003,
      "step": 4200
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.4939753143090142,
      "learning_rate": 1.3776074157227335e-05,
      "loss": 0.3516,
      "step": 4201
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.6964649149772422,
      "learning_rate": 1.3773250799196964e-05,
      "loss": 0.3225,
      "step": 4202
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.6318703838043302,
      "learning_rate": 1.3770427090411717e-05,
      "loss": 0.3284,
      "step": 4203
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.5062009875757698,
      "learning_rate": 1.3767603031134087e-05,
      "loss": 0.3279,
      "step": 4204
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.5105113956604999,
      "learning_rate": 1.3764778621626586e-05,
      "loss": 0.24,
      "step": 4205
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.7453134826234098,
      "learning_rate": 1.3761953862151773e-05,
      "loss": 0.3877,
      "step": 4206
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.7116540420755686,
      "learning_rate": 1.3759128752972229e-05,
      "loss": 0.3735,
      "step": 4207
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.5577572998700258,
      "learning_rate": 1.375630329435057e-05,
      "loss": 0.3523,
      "step": 4208
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.6676419425969569,
      "learning_rate": 1.3753477486549449e-05,
      "loss": 0.3477,
      "step": 4209
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.6195057795881258,
      "learning_rate": 1.3750651329831548e-05,
      "loss": 0.4233,
      "step": 4210
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.92256510834093,
      "learning_rate": 1.3747824824459577e-05,
      "loss": 0.3999,
      "step": 4211
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.3390179139518106,
      "learning_rate": 1.3744997970696291e-05,
      "loss": 0.2468,
      "step": 4212
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.6322367653865417,
      "learning_rate": 1.3742170768804464e-05,
      "loss": 0.397,
      "step": 4213
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.5664672068694026,
      "learning_rate": 1.3739343219046908e-05,
      "loss": 0.3389,
      "step": 4214
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.6636060795119598,
      "learning_rate": 1.3736515321686468e-05,
      "loss": 0.3838,
      "step": 4215
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.7266739954385624,
      "learning_rate": 1.373368707698602e-05,
      "loss": 0.3181,
      "step": 4216
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.72595706223026,
      "learning_rate": 1.3730858485208471e-05,
      "loss": 0.3706,
      "step": 4217
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.723209947381202,
      "learning_rate": 1.3728029546616769e-05,
      "loss": 0.356,
      "step": 4218
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.4795525501478612,
      "learning_rate": 1.3725200261473879e-05,
      "loss": 0.3105,
      "step": 4219
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.8559024900689773,
      "learning_rate": 1.3722370630042809e-05,
      "loss": 0.3169,
      "step": 4220
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.5895615192135018,
      "learning_rate": 1.3719540652586601e-05,
      "loss": 0.344,
      "step": 4221
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.7214671103698287,
      "learning_rate": 1.3716710329368319e-05,
      "loss": 0.3706,
      "step": 4222
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.9009453820547255,
      "learning_rate": 1.3713879660651069e-05,
      "loss": 0.4609,
      "step": 4223
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.641414768214502,
      "learning_rate": 1.3711048646697979e-05,
      "loss": 0.3086,
      "step": 4224
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.707598381080663,
      "learning_rate": 1.3708217287772227e-05,
      "loss": 0.3438,
      "step": 4225
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.6392130683129817,
      "learning_rate": 1.3705385584136997e-05,
      "loss": 0.3655,
      "step": 4226
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.620659133391431,
      "learning_rate": 1.3702553536055529e-05,
      "loss": 0.3608,
      "step": 4227
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.9338668685359262,
      "learning_rate": 1.3699721143791083e-05,
      "loss": 0.3833,
      "step": 4228
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.7581748769540013,
      "learning_rate": 1.3696888407606952e-05,
      "loss": 0.397,
      "step": 4229
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.5993043644480787,
      "learning_rate": 1.369405532776646e-05,
      "loss": 0.2595,
      "step": 4230
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.6362163487180437,
      "learning_rate": 1.3691221904532972e-05,
      "loss": 0.3247,
      "step": 4231
    },
    {
      "epoch": 1.2,
      "grad_norm": 2.290765471378604,
      "learning_rate": 1.3688388138169873e-05,
      "loss": 0.4629,
      "step": 4232
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.7106122331852414,
      "learning_rate": 1.3685554028940586e-05,
      "loss": 0.4492,
      "step": 4233
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.7146846037313208,
      "learning_rate": 1.3682719577108566e-05,
      "loss": 0.356,
      "step": 4234
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.4696909735668118,
      "learning_rate": 1.3679884782937295e-05,
      "loss": 0.3271,
      "step": 4235
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.7327552711879546,
      "learning_rate": 1.3677049646690298e-05,
      "loss": 0.4048,
      "step": 4236
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.5977647201212795,
      "learning_rate": 1.3674214168631114e-05,
      "loss": 0.3113,
      "step": 4237
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.463313987051893,
      "learning_rate": 1.3671378349023333e-05,
      "loss": 0.2988,
      "step": 4238
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.4366127102801667,
      "learning_rate": 1.3668542188130567e-05,
      "loss": 0.2979,
      "step": 4239
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.7608438949626544,
      "learning_rate": 1.3665705686216457e-05,
      "loss": 0.3799,
      "step": 4240
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.4434170014338752,
      "learning_rate": 1.3662868843544676e-05,
      "loss": 0.3164,
      "step": 4241
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.4956520341367219,
      "learning_rate": 1.3660031660378945e-05,
      "loss": 0.344,
      "step": 4242
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.8953790160346908,
      "learning_rate": 1.365719413698299e-05,
      "loss": 0.4272,
      "step": 4243
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.4738878911499096,
      "learning_rate": 1.365435627362059e-05,
      "loss": 0.3301,
      "step": 4244
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.4764825083690611,
      "learning_rate": 1.3651518070555545e-05,
      "loss": 0.3535,
      "step": 4245
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.6683468654965308,
      "learning_rate": 1.3648679528051689e-05,
      "loss": 0.3103,
      "step": 4246
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.4321691739267048,
      "learning_rate": 1.364584064637289e-05,
      "loss": 0.2571,
      "step": 4247
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.513620288720645,
      "learning_rate": 1.3643001425783045e-05,
      "loss": 0.3203,
      "step": 4248
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.5470880348852842,
      "learning_rate": 1.3640161866546083e-05,
      "loss": 0.3115,
      "step": 4249
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.4337507253371886,
      "learning_rate": 1.3637321968925964e-05,
      "loss": 0.3496,
      "step": 4250
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.6987792968758013,
      "learning_rate": 1.3634481733186675e-05,
      "loss": 0.321,
      "step": 4251
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.6985797960294395,
      "learning_rate": 1.3631641159592253e-05,
      "loss": 0.4141,
      "step": 4252
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.633260314659198,
      "learning_rate": 1.3628800248406738e-05,
      "loss": 0.3398,
      "step": 4253
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.9153810841946015,
      "learning_rate": 1.3625958999894225e-05,
      "loss": 0.4414,
      "step": 4254
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.6248762235300436,
      "learning_rate": 1.3623117414318827e-05,
      "loss": 0.3086,
      "step": 4255
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.6100502438512023,
      "learning_rate": 1.3620275491944695e-05,
      "loss": 0.2778,
      "step": 4256
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.9133561739667877,
      "learning_rate": 1.361743323303601e-05,
      "loss": 0.394,
      "step": 4257
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.5854456423247225,
      "learning_rate": 1.3614590637856986e-05,
      "loss": 0.3337,
      "step": 4258
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.5393275753145068,
      "learning_rate": 1.3611747706671859e-05,
      "loss": 0.3298,
      "step": 4259
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.5347510515905467,
      "learning_rate": 1.3608904439744905e-05,
      "loss": 0.3735,
      "step": 4260
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.8884425793249606,
      "learning_rate": 1.3606060837340431e-05,
      "loss": 0.3574,
      "step": 4261
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.6960507870803028,
      "learning_rate": 1.3603216899722775e-05,
      "loss": 0.3208,
      "step": 4262
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.672492543100102,
      "learning_rate": 1.3600372627156304e-05,
      "loss": 0.386,
      "step": 4263
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.476255633930814,
      "learning_rate": 1.3597528019905411e-05,
      "loss": 0.313,
      "step": 4264
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.3593877066831517,
      "learning_rate": 1.3594683078234532e-05,
      "loss": 0.2646,
      "step": 4265
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.6283383378877292,
      "learning_rate": 1.3591837802408123e-05,
      "loss": 0.4185,
      "step": 4266
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.629345485704324,
      "learning_rate": 1.3588992192690683e-05,
      "loss": 0.2932,
      "step": 4267
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.636898750413856,
      "learning_rate": 1.3586146249346728e-05,
      "loss": 0.4033,
      "step": 4268
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.4423252225466463,
      "learning_rate": 1.3583299972640815e-05,
      "loss": 0.2915,
      "step": 4269
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.7448401558306799,
      "learning_rate": 1.3580453362837527e-05,
      "loss": 0.436,
      "step": 4270
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.5745245744613872,
      "learning_rate": 1.3577606420201483e-05,
      "loss": 0.3354,
      "step": 4271
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.9088587678133564,
      "learning_rate": 1.357475914499733e-05,
      "loss": 0.4153,
      "step": 4272
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.7714407688749434,
      "learning_rate": 1.3571911537489739e-05,
      "loss": 0.3062,
      "step": 4273
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.3431592864110553,
      "learning_rate": 1.3569063597943428e-05,
      "loss": 0.2456,
      "step": 4274
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.6226525276311705,
      "learning_rate": 1.3566215326623131e-05,
      "loss": 0.3022,
      "step": 4275
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.652358004073436,
      "learning_rate": 1.3563366723793621e-05,
      "loss": 0.356,
      "step": 4276
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.7051955223871074,
      "learning_rate": 1.3560517789719696e-05,
      "loss": 0.3652,
      "step": 4277
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.8392514732107166,
      "learning_rate": 1.355766852466619e-05,
      "loss": 0.3901,
      "step": 4278
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.5665151903058259,
      "learning_rate": 1.3554818928897965e-05,
      "loss": 0.3145,
      "step": 4279
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.7058283285792677,
      "learning_rate": 1.355196900267992e-05,
      "loss": 0.3389,
      "step": 4280
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.563291252294466,
      "learning_rate": 1.3549118746276968e-05,
      "loss": 0.3518,
      "step": 4281
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.576459117816438,
      "learning_rate": 1.3546268159954075e-05,
      "loss": 0.3599,
      "step": 4282
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.7646371941576533,
      "learning_rate": 1.3543417243976217e-05,
      "loss": 0.3535,
      "step": 4283
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.6934865078478047,
      "learning_rate": 1.3540565998608419e-05,
      "loss": 0.3232,
      "step": 4284
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.7716725341477162,
      "learning_rate": 1.3537714424115725e-05,
      "loss": 0.3501,
      "step": 4285
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.5604639394480124,
      "learning_rate": 1.353486252076321e-05,
      "loss": 0.324,
      "step": 4286
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.7517500774692403,
      "learning_rate": 1.353201028881598e-05,
      "loss": 0.3657,
      "step": 4287
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.5002806826089499,
      "learning_rate": 1.3529157728539179e-05,
      "loss": 0.3066,
      "step": 4288
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.4674873427289954,
      "learning_rate": 1.3526304840197976e-05,
      "loss": 0.3252,
      "step": 4289
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.5179575980032733,
      "learning_rate": 1.3523451624057566e-05,
      "loss": 0.3196,
      "step": 4290
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.5303340586034164,
      "learning_rate": 1.3520598080383183e-05,
      "loss": 0.2949,
      "step": 4291
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.815677576539279,
      "learning_rate": 1.3517744209440085e-05,
      "loss": 0.3528,
      "step": 4292
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.9004219322883664,
      "learning_rate": 1.3514890011493564e-05,
      "loss": 0.4229,
      "step": 4293
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.4968806919889472,
      "learning_rate": 1.3512035486808942e-05,
      "loss": 0.2759,
      "step": 4294
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.7040764082569295,
      "learning_rate": 1.350918063565157e-05,
      "loss": 0.3999,
      "step": 4295
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.6799714954470997,
      "learning_rate": 1.3506325458286829e-05,
      "loss": 0.3774,
      "step": 4296
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.6546430519698965,
      "learning_rate": 1.3503469954980133e-05,
      "loss": 0.3293,
      "step": 4297
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.7756285438814474,
      "learning_rate": 1.3500614125996924e-05,
      "loss": 0.3301,
      "step": 4298
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.764306197730283,
      "learning_rate": 1.3497757971602677e-05,
      "loss": 0.4155,
      "step": 4299
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.6225728438181375,
      "learning_rate": 1.3494901492062889e-05,
      "loss": 0.3291,
      "step": 4300
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.653711222878537,
      "learning_rate": 1.34920446876431e-05,
      "loss": 0.365,
      "step": 4301
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.7255448629136565,
      "learning_rate": 1.3489187558608871e-05,
      "loss": 0.3064,
      "step": 4302
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.6482048390241466,
      "learning_rate": 1.3486330105225797e-05,
      "loss": 0.3574,
      "step": 4303
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.7772774020817927,
      "learning_rate": 1.3483472327759496e-05,
      "loss": 0.4116,
      "step": 4304
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.5067085227384147,
      "learning_rate": 1.3480614226475632e-05,
      "loss": 0.2988,
      "step": 4305
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.7729919916722812,
      "learning_rate": 1.3477755801639883e-05,
      "loss": 0.4097,
      "step": 4306
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.5753776858953512,
      "learning_rate": 1.3474897053517963e-05,
      "loss": 0.3354,
      "step": 4307
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.9610804677876976,
      "learning_rate": 1.347203798237562e-05,
      "loss": 0.4087,
      "step": 4308
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.712635181595457,
      "learning_rate": 1.3469178588478621e-05,
      "loss": 0.3801,
      "step": 4309
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.8632455578434823,
      "learning_rate": 1.346631887209278e-05,
      "loss": 0.4111,
      "step": 4310
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.4873742896716557,
      "learning_rate": 1.3463458833483923e-05,
      "loss": 0.2827,
      "step": 4311
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.5517343785132298,
      "learning_rate": 1.3460598472917923e-05,
      "loss": 0.2927,
      "step": 4312
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.3820258807120493,
      "learning_rate": 1.3457737790660663e-05,
      "loss": 0.2566,
      "step": 4313
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.412869102587919,
      "learning_rate": 1.3454876786978076e-05,
      "loss": 0.29,
      "step": 4314
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.8547610217553894,
      "learning_rate": 1.3452015462136113e-05,
      "loss": 0.3469,
      "step": 4315
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.653893295889704,
      "learning_rate": 1.3449153816400758e-05,
      "loss": 0.3408,
      "step": 4316
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.8604311772765536,
      "learning_rate": 1.3446291850038021e-05,
      "loss": 0.3716,
      "step": 4317
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.8388194707432786,
      "learning_rate": 1.3443429563313952e-05,
      "loss": 0.4114,
      "step": 4318
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.700649534997186,
      "learning_rate": 1.344056695649462e-05,
      "loss": 0.2998,
      "step": 4319
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.7538640882273115,
      "learning_rate": 1.343770402984613e-05,
      "loss": 0.3584,
      "step": 4320
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.7704020705649688,
      "learning_rate": 1.3434840783634611e-05,
      "loss": 0.394,
      "step": 4321
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.9955062990575674,
      "learning_rate": 1.3431977218126234e-05,
      "loss": 0.355,
      "step": 4322
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.7727278901128574,
      "learning_rate": 1.3429113333587181e-05,
      "loss": 0.3506,
      "step": 4323
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.5113365121319258,
      "learning_rate": 1.3426249130283677e-05,
      "loss": 0.3228,
      "step": 4324
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.70405306434983,
      "learning_rate": 1.3423384608481978e-05,
      "loss": 0.3813,
      "step": 4325
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.7335208974569023,
      "learning_rate": 1.342051976844836e-05,
      "loss": 0.3376,
      "step": 4326
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.6524803422231988,
      "learning_rate": 1.3417654610449131e-05,
      "loss": 0.3401,
      "step": 4327
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.6519400385718341,
      "learning_rate": 1.341478913475064e-05,
      "loss": 0.3379,
      "step": 4328
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.446936414130687,
      "learning_rate": 1.341192334161925e-05,
      "loss": 0.2915,
      "step": 4329
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.7204059803946201,
      "learning_rate": 1.3409057231321363e-05,
      "loss": 0.3499,
      "step": 4330
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.6555681350788483,
      "learning_rate": 1.3406190804123405e-05,
      "loss": 0.293,
      "step": 4331
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.560132386643744,
      "learning_rate": 1.3403324060291837e-05,
      "loss": 0.3501,
      "step": 4332
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.6237801598140533,
      "learning_rate": 1.3400457000093145e-05,
      "loss": 0.3198,
      "step": 4333
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.5934900696501144,
      "learning_rate": 1.3397589623793845e-05,
      "loss": 0.3516,
      "step": 4334
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.5503566238237132,
      "learning_rate": 1.3394721931660488e-05,
      "loss": 0.3328,
      "step": 4335
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.500528236437697,
      "learning_rate": 1.3391853923959644e-05,
      "loss": 0.27,
      "step": 4336
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.5351806187592152,
      "learning_rate": 1.3388985600957922e-05,
      "loss": 0.3379,
      "step": 4337
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.6590706693635167,
      "learning_rate": 1.3386116962921951e-05,
      "loss": 0.2866,
      "step": 4338
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.6725334539492056,
      "learning_rate": 1.3383248010118404e-05,
      "loss": 0.2961,
      "step": 4339
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.8460893386356445,
      "learning_rate": 1.3380378742813964e-05,
      "loss": 0.2961,
      "step": 4340
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.4305591051854352,
      "learning_rate": 1.337750916127536e-05,
      "loss": 0.2739,
      "step": 4341
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.4344317325819533,
      "learning_rate": 1.3374639265769337e-05,
      "loss": 0.2351,
      "step": 4342
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.5585541740160826,
      "learning_rate": 1.3371769056562683e-05,
      "loss": 0.2874,
      "step": 4343
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.3942936016794698,
      "learning_rate": 1.3368898533922202e-05,
      "loss": 0.2246,
      "step": 4344
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.6214675058025674,
      "learning_rate": 1.3366027698114734e-05,
      "loss": 0.3281,
      "step": 4345
    },
    {
      "epoch": 1.23,
      "grad_norm": 2.1298271170722503,
      "learning_rate": 1.3363156549407151e-05,
      "loss": 0.4429,
      "step": 4346
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.9866601525510046,
      "learning_rate": 1.3360285088066343e-05,
      "loss": 0.3936,
      "step": 4347
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.7847220465603375,
      "learning_rate": 1.3357413314359242e-05,
      "loss": 0.3457,
      "step": 4348
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.5024899976013277,
      "learning_rate": 1.3354541228552802e-05,
      "loss": 0.2935,
      "step": 4349
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.5121781459927148,
      "learning_rate": 1.3351668830914004e-05,
      "loss": 0.2798,
      "step": 4350
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.7057566789417917,
      "learning_rate": 1.3348796121709862e-05,
      "loss": 0.4009,
      "step": 4351
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.669662561126739,
      "learning_rate": 1.3345923101207423e-05,
      "loss": 0.3149,
      "step": 4352
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.6018290283784806,
      "learning_rate": 1.334304976967375e-05,
      "loss": 0.3716,
      "step": 4353
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.6528194871054886,
      "learning_rate": 1.334017612737595e-05,
      "loss": 0.3584,
      "step": 4354
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.632252474152988,
      "learning_rate": 1.333730217458115e-05,
      "loss": 0.3486,
      "step": 4355
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.7199402309212053,
      "learning_rate": 1.3334427911556506e-05,
      "loss": 0.3799,
      "step": 4356
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.731243793119649,
      "learning_rate": 1.3331553338569204e-05,
      "loss": 0.2891,
      "step": 4357
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.8020756229210517,
      "learning_rate": 1.3328678455886461e-05,
      "loss": 0.3823,
      "step": 4358
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.6717880132108323,
      "learning_rate": 1.3325803263775521e-05,
      "loss": 0.3066,
      "step": 4359
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.8155006155409694,
      "learning_rate": 1.3322927762503656e-05,
      "loss": 0.3813,
      "step": 4360
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.5924814394809286,
      "learning_rate": 1.3320051952338166e-05,
      "loss": 0.3257,
      "step": 4361
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.6476271337793515,
      "learning_rate": 1.3317175833546387e-05,
      "loss": 0.2837,
      "step": 4362
    },
    {
      "epoch": 1.23,
      "grad_norm": 2.145408773710861,
      "learning_rate": 1.3314299406395677e-05,
      "loss": 0.481,
      "step": 4363
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.776504615308031,
      "learning_rate": 1.3311422671153416e-05,
      "loss": 0.3342,
      "step": 4364
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.5591477445951303,
      "learning_rate": 1.3308545628087029e-05,
      "loss": 0.3298,
      "step": 4365
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.7132328752153803,
      "learning_rate": 1.3305668277463954e-05,
      "loss": 0.3511,
      "step": 4366
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.6855138212218714,
      "learning_rate": 1.3302790619551673e-05,
      "loss": 0.3179,
      "step": 4367
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.55454063617089,
      "learning_rate": 1.329991265461768e-05,
      "loss": 0.3496,
      "step": 4368
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.5966825269853913,
      "learning_rate": 1.3297034382929512e-05,
      "loss": 0.3284,
      "step": 4369
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.6311697550356312,
      "learning_rate": 1.329415580475472e-05,
      "loss": 0.3691,
      "step": 4370
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.8174602537814317,
      "learning_rate": 1.32912769203609e-05,
      "loss": 0.4331,
      "step": 4371
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.8802800236742194,
      "learning_rate": 1.3288397730015666e-05,
      "loss": 0.4312,
      "step": 4372
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.6379444573559503,
      "learning_rate": 1.328551823398666e-05,
      "loss": 0.3354,
      "step": 4373
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.5640851935127276,
      "learning_rate": 1.3282638432541553e-05,
      "loss": 0.2861,
      "step": 4374
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.515716845572626,
      "learning_rate": 1.3279758325948054e-05,
      "loss": 0.2981,
      "step": 4375
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.890927785428697,
      "learning_rate": 1.3276877914473887e-05,
      "loss": 0.436,
      "step": 4376
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.655976411892525,
      "learning_rate": 1.3273997198386814e-05,
      "loss": 0.3179,
      "step": 4377
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.6434560234003304,
      "learning_rate": 1.3271116177954615e-05,
      "loss": 0.2954,
      "step": 4378
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.6990555458944674,
      "learning_rate": 1.3268234853445113e-05,
      "loss": 0.3496,
      "step": 4379
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.7750194808892992,
      "learning_rate": 1.3265353225126143e-05,
      "loss": 0.3252,
      "step": 4380
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.7288398971887515,
      "learning_rate": 1.3262471293265577e-05,
      "loss": 0.3333,
      "step": 4381
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.8463652883681996,
      "learning_rate": 1.3259589058131323e-05,
      "loss": 0.3887,
      "step": 4382
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.6819787547438148,
      "learning_rate": 1.32567065199913e-05,
      "loss": 0.3508,
      "step": 4383
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.6371757402783995,
      "learning_rate": 1.3253823679113466e-05,
      "loss": 0.3599,
      "step": 4384
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.6031607828106595,
      "learning_rate": 1.3250940535765808e-05,
      "loss": 0.2717,
      "step": 4385
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.5990038084627258,
      "learning_rate": 1.3248057090216336e-05,
      "loss": 0.335,
      "step": 4386
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.661033484014287,
      "learning_rate": 1.3245173342733084e-05,
      "loss": 0.3721,
      "step": 4387
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.8365089836562127,
      "learning_rate": 1.3242289293584132e-05,
      "loss": 0.3711,
      "step": 4388
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.6574244693004396,
      "learning_rate": 1.3239404943037566e-05,
      "loss": 0.3628,
      "step": 4389
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.7139288085059627,
      "learning_rate": 1.3236520291361516e-05,
      "loss": 0.3496,
      "step": 4390
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.8722199224994076,
      "learning_rate": 1.3233635338824132e-05,
      "loss": 0.4058,
      "step": 4391
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.4763736483131407,
      "learning_rate": 1.3230750085693599e-05,
      "loss": 0.3093,
      "step": 4392
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.8797062867353087,
      "learning_rate": 1.3227864532238113e-05,
      "loss": 0.3687,
      "step": 4393
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.775120633697327,
      "learning_rate": 1.3224978678725921e-05,
      "loss": 0.3413,
      "step": 4394
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.7340565376962362,
      "learning_rate": 1.3222092525425286e-05,
      "loss": 0.3809,
      "step": 4395
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.5880284716002862,
      "learning_rate": 1.3219206072604496e-05,
      "loss": 0.3052,
      "step": 4396
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.6943776605168746,
      "learning_rate": 1.321631932053187e-05,
      "loss": 0.3606,
      "step": 4397
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.5731738209968045,
      "learning_rate": 1.3213432269475759e-05,
      "loss": 0.3232,
      "step": 4398
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.5081803601894281,
      "learning_rate": 1.3210544919704539e-05,
      "loss": 0.2998,
      "step": 4399
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.503856408259567,
      "learning_rate": 1.3207657271486607e-05,
      "loss": 0.2793,
      "step": 4400
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.7693048557589963,
      "learning_rate": 1.3204769325090403e-05,
      "loss": 0.3845,
      "step": 4401
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.554651068495546,
      "learning_rate": 1.3201881080784378e-05,
      "loss": 0.3335,
      "step": 4402
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.7039598189010485,
      "learning_rate": 1.3198992538837021e-05,
      "loss": 0.2571,
      "step": 4403
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.7093980093347039,
      "learning_rate": 1.3196103699516844e-05,
      "loss": 0.2776,
      "step": 4404
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.7720458555833694,
      "learning_rate": 1.3193214563092392e-05,
      "loss": 0.3574,
      "step": 4405
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.7574597110100398,
      "learning_rate": 1.319032512983223e-05,
      "loss": 0.3433,
      "step": 4406
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.6460245044406605,
      "learning_rate": 1.318743540000496e-05,
      "loss": 0.3623,
      "step": 4407
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.5788895124981317,
      "learning_rate": 1.3184545373879201e-05,
      "loss": 0.2791,
      "step": 4408
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.5896333609531403,
      "learning_rate": 1.318165505172361e-05,
      "loss": 0.3069,
      "step": 4409
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.9276917905997373,
      "learning_rate": 1.3178764433806858e-05,
      "loss": 0.3389,
      "step": 4410
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.6568198216114647,
      "learning_rate": 1.3175873520397659e-05,
      "loss": 0.3184,
      "step": 4411
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.6096411919500517,
      "learning_rate": 1.3172982311764749e-05,
      "loss": 0.3296,
      "step": 4412
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.7290030910130723,
      "learning_rate": 1.3170090808176883e-05,
      "loss": 0.3755,
      "step": 4413
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.573969631314692,
      "learning_rate": 1.316719900990285e-05,
      "loss": 0.3311,
      "step": 4414
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.6641894406076876,
      "learning_rate": 1.3164306917211475e-05,
      "loss": 0.3518,
      "step": 4415
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.7446466068874222,
      "learning_rate": 1.3161414530371598e-05,
      "loss": 0.3386,
      "step": 4416
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.5681686681801543,
      "learning_rate": 1.3158521849652087e-05,
      "loss": 0.3142,
      "step": 4417
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.6250018458032542,
      "learning_rate": 1.3155628875321842e-05,
      "loss": 0.2766,
      "step": 4418
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.7196669294402378,
      "learning_rate": 1.315273560764979e-05,
      "loss": 0.3647,
      "step": 4419
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.613693525625962,
      "learning_rate": 1.3149842046904885e-05,
      "loss": 0.3037,
      "step": 4420
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.5153479266502876,
      "learning_rate": 1.3146948193356105e-05,
      "loss": 0.3076,
      "step": 4421
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.5700040086598914,
      "learning_rate": 1.3144054047272462e-05,
      "loss": 0.2986,
      "step": 4422
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.91571459532067,
      "learning_rate": 1.3141159608922984e-05,
      "loss": 0.3232,
      "step": 4423
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.6109614959262484,
      "learning_rate": 1.3138264878576738e-05,
      "loss": 0.3374,
      "step": 4424
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.8751128274695947,
      "learning_rate": 1.3135369856502813e-05,
      "loss": 0.356,
      "step": 4425
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.6185533989574907,
      "learning_rate": 1.3132474542970324e-05,
      "loss": 0.3403,
      "step": 4426
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.7564810917512996,
      "learning_rate": 1.3129578938248411e-05,
      "loss": 0.3657,
      "step": 4427
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.6302746096843415,
      "learning_rate": 1.3126683042606251e-05,
      "loss": 0.3257,
      "step": 4428
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.6374614842767783,
      "learning_rate": 1.3123786856313036e-05,
      "loss": 0.3474,
      "step": 4429
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.7860718410930871,
      "learning_rate": 1.3120890379637996e-05,
      "loss": 0.3667,
      "step": 4430
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.6096527854502245,
      "learning_rate": 1.3117993612850377e-05,
      "loss": 0.3262,
      "step": 4431
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.5483897183202324,
      "learning_rate": 1.3115096556219462e-05,
      "loss": 0.3228,
      "step": 4432
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.612081142074054,
      "learning_rate": 1.3112199210014552e-05,
      "loss": 0.3203,
      "step": 4433
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.6870202637638003,
      "learning_rate": 1.3109301574504978e-05,
      "loss": 0.3711,
      "step": 4434
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.7371421158476237,
      "learning_rate": 1.3106403649960109e-05,
      "loss": 0.3589,
      "step": 4435
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.5892122066885688,
      "learning_rate": 1.310350543664932e-05,
      "loss": 0.394,
      "step": 4436
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.5801794528348334,
      "learning_rate": 1.310060693484203e-05,
      "loss": 0.2969,
      "step": 4437
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.5829841210419267,
      "learning_rate": 1.3097708144807679e-05,
      "loss": 0.3965,
      "step": 4438
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.56016109331299,
      "learning_rate": 1.3094809066815731e-05,
      "loss": 0.3008,
      "step": 4439
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.701529495947141,
      "learning_rate": 1.3091909701135676e-05,
      "loss": 0.3745,
      "step": 4440
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.560525994731936,
      "learning_rate": 1.3089010048037043e-05,
      "loss": 0.3379,
      "step": 4441
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.6247063239004225,
      "learning_rate": 1.3086110107789371e-05,
      "loss": 0.3259,
      "step": 4442
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.4624095583000858,
      "learning_rate": 1.3083209880662237e-05,
      "loss": 0.2915,
      "step": 4443
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.8569409315605465,
      "learning_rate": 1.3080309366925239e-05,
      "loss": 0.3936,
      "step": 4444
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.5094096106233308,
      "learning_rate": 1.3077408566848009e-05,
      "loss": 0.3062,
      "step": 4445
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.5126109877662004,
      "learning_rate": 1.3074507480700192e-05,
      "loss": 0.2979,
      "step": 4446
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.6594187271798284,
      "learning_rate": 1.3071606108751475e-05,
      "loss": 0.3774,
      "step": 4447
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.5017262776272688,
      "learning_rate": 1.3068704451271561e-05,
      "loss": 0.2502,
      "step": 4448
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.9909886443015001,
      "learning_rate": 1.3065802508530186e-05,
      "loss": 0.3916,
      "step": 4449
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.8328310558775678,
      "learning_rate": 1.3062900280797104e-05,
      "loss": 0.3633,
      "step": 4450
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.7455791572550765,
      "learning_rate": 1.3059997768342108e-05,
      "loss": 0.3589,
      "step": 4451
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.6771816109807562,
      "learning_rate": 1.3057094971435005e-05,
      "loss": 0.3545,
      "step": 4452
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.617600123762209,
      "learning_rate": 1.3054191890345633e-05,
      "loss": 0.332,
      "step": 4453
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.891111963417076,
      "learning_rate": 1.3051288525343866e-05,
      "loss": 0.3335,
      "step": 4454
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.7488480496287706,
      "learning_rate": 1.3048384876699588e-05,
      "loss": 0.376,
      "step": 4455
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.7617530114539892,
      "learning_rate": 1.304548094468272e-05,
      "loss": 0.2871,
      "step": 4456
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.6184508823329544,
      "learning_rate": 1.3042576729563203e-05,
      "loss": 0.3057,
      "step": 4457
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.6742268957305149,
      "learning_rate": 1.3039672231611015e-05,
      "loss": 0.3369,
      "step": 4458
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.543685582877137,
      "learning_rate": 1.3036767451096148e-05,
      "loss": 0.3354,
      "step": 4459
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.5872677202312926,
      "learning_rate": 1.3033862388288628e-05,
      "loss": 0.3228,
      "step": 4460
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.779457596022529,
      "learning_rate": 1.30309570434585e-05,
      "loss": 0.3569,
      "step": 4461
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.6905508293905913,
      "learning_rate": 1.3028051416875845e-05,
      "loss": 0.3848,
      "step": 4462
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.8336977642231127,
      "learning_rate": 1.302514550881076e-05,
      "loss": 0.377,
      "step": 4463
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.9840467979648557,
      "learning_rate": 1.3022239319533376e-05,
      "loss": 0.3818,
      "step": 4464
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.590891034402144,
      "learning_rate": 1.3019332849313851e-05,
      "loss": 0.3599,
      "step": 4465
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.5647240153227864,
      "learning_rate": 1.301642609842236e-05,
      "loss": 0.2688,
      "step": 4466
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.5279641092800542,
      "learning_rate": 1.3013519067129108e-05,
      "loss": 0.261,
      "step": 4467
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.7352080878931255,
      "learning_rate": 1.3010611755704334e-05,
      "loss": 0.3696,
      "step": 4468
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.8130402103184518,
      "learning_rate": 1.3007704164418294e-05,
      "loss": 0.3723,
      "step": 4469
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.8201833466449346,
      "learning_rate": 1.3004796293541269e-05,
      "loss": 0.311,
      "step": 4470
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.6838560052807687,
      "learning_rate": 1.3001888143343578e-05,
      "loss": 0.2656,
      "step": 4471
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.8587119803311039,
      "learning_rate": 1.2998979714095547e-05,
      "loss": 0.3301,
      "step": 4472
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.6965525140635003,
      "learning_rate": 1.2996071006067548e-05,
      "loss": 0.3457,
      "step": 4473
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.5688444562621602,
      "learning_rate": 1.2993162019529963e-05,
      "loss": 0.3452,
      "step": 4474
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.7173859396252817,
      "learning_rate": 1.2990252754753212e-05,
      "loss": 0.3228,
      "step": 4475
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.8313233208331445,
      "learning_rate": 1.2987343212007728e-05,
      "loss": 0.3718,
      "step": 4476
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.606001635127521,
      "learning_rate": 1.2984433391563984e-05,
      "loss": 0.3247,
      "step": 4477
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.8984243046935205,
      "learning_rate": 1.2981523293692468e-05,
      "loss": 0.3799,
      "step": 4478
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.7093740562183226,
      "learning_rate": 1.2978612918663702e-05,
      "loss": 0.3425,
      "step": 4479
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.7737818737665327,
      "learning_rate": 1.297570226674822e-05,
      "loss": 0.4126,
      "step": 4480
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.6584145757518887,
      "learning_rate": 1.2972791338216601e-05,
      "loss": 0.373,
      "step": 4481
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.6349503159916288,
      "learning_rate": 1.2969880133339437e-05,
      "loss": 0.3774,
      "step": 4482
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.6474194733400975,
      "learning_rate": 1.2966968652387348e-05,
      "loss": 0.3667,
      "step": 4483
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.747498449772196,
      "learning_rate": 1.2964056895630976e-05,
      "loss": 0.3848,
      "step": 4484
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.5751093197423915,
      "learning_rate": 1.2961144863341e-05,
      "loss": 0.2888,
      "step": 4485
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.6520441289796115,
      "learning_rate": 1.2958232555788115e-05,
      "loss": 0.4038,
      "step": 4486
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.6484436713803379,
      "learning_rate": 1.2955319973243043e-05,
      "loss": 0.2986,
      "step": 4487
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.604735859498446,
      "learning_rate": 1.2952407115976531e-05,
      "loss": 0.3252,
      "step": 4488
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.5838085496260168,
      "learning_rate": 1.2949493984259355e-05,
      "loss": 0.2786,
      "step": 4489
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.6531069184961946,
      "learning_rate": 1.294658057836232e-05,
      "loss": 0.3516,
      "step": 4490
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.610002004413302,
      "learning_rate": 1.294366689855624e-05,
      "loss": 0.323,
      "step": 4491
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.7725143776982641,
      "learning_rate": 1.2940752945111977e-05,
      "loss": 0.3931,
      "step": 4492
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.8210123551275168,
      "learning_rate": 1.2937838718300398e-05,
      "loss": 0.363,
      "step": 4493
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.8317230680855279,
      "learning_rate": 1.293492421839241e-05,
      "loss": 0.3364,
      "step": 4494
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.6780462453987597,
      "learning_rate": 1.293200944565894e-05,
      "loss": 0.3784,
      "step": 4495
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.5446541483170406,
      "learning_rate": 1.2929094400370936e-05,
      "loss": 0.2788,
      "step": 4496
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.7317119097024671,
      "learning_rate": 1.2926179082799377e-05,
      "loss": 0.3828,
      "step": 4497
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.706843450019343,
      "learning_rate": 1.292326349321527e-05,
      "loss": 0.321,
      "step": 4498
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.5648749942257172,
      "learning_rate": 1.2920347631889637e-05,
      "loss": 0.3105,
      "step": 4499
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.95486041590381,
      "learning_rate": 1.2917431499093538e-05,
      "loss": 0.3965,
      "step": 4500
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.701668813842967,
      "learning_rate": 1.2914515095098043e-05,
      "loss": 0.3328,
      "step": 4501
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.9391474521409375,
      "learning_rate": 1.291159842017427e-05,
      "loss": 0.3984,
      "step": 4502
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.769642262155629,
      "learning_rate": 1.290868147459333e-05,
      "loss": 0.3535,
      "step": 4503
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.593032511334354,
      "learning_rate": 1.290576425862639e-05,
      "loss": 0.3098,
      "step": 4504
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.7916541995988355,
      "learning_rate": 1.2902846772544625e-05,
      "loss": 0.3447,
      "step": 4505
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.9127241367886083,
      "learning_rate": 1.2899929016619241e-05,
      "loss": 0.373,
      "step": 4506
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.6657383361238467,
      "learning_rate": 1.2897010991121465e-05,
      "loss": 0.3403,
      "step": 4507
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.5920976073187252,
      "learning_rate": 1.2894092696322554e-05,
      "loss": 0.2712,
      "step": 4508
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.5583463431112654,
      "learning_rate": 1.2891174132493792e-05,
      "loss": 0.3081,
      "step": 4509
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.7906595052744991,
      "learning_rate": 1.288825529990647e-05,
      "loss": 0.4248,
      "step": 4510
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.8860529177323833,
      "learning_rate": 1.2885336198831933e-05,
      "loss": 0.3647,
      "step": 4511
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.5912129874126513,
      "learning_rate": 1.2882416829541526e-05,
      "loss": 0.2922,
      "step": 4512
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.8485953377330666,
      "learning_rate": 1.2879497192306634e-05,
      "loss": 0.376,
      "step": 4513
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.6362911420593629,
      "learning_rate": 1.2876577287398653e-05,
      "loss": 0.312,
      "step": 4514
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.994687085113273,
      "learning_rate": 1.2873657115089022e-05,
      "loss": 0.3755,
      "step": 4515
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.529999741106312,
      "learning_rate": 1.2870736675649192e-05,
      "loss": 0.3245,
      "step": 4516
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.6175899201453048,
      "learning_rate": 1.286781596935064e-05,
      "loss": 0.3398,
      "step": 4517
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.7231507800252661,
      "learning_rate": 1.286489499646487e-05,
      "loss": 0.3726,
      "step": 4518
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.6688761083127375,
      "learning_rate": 1.2861973757263416e-05,
      "loss": 0.3628,
      "step": 4519
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.9275611692953312,
      "learning_rate": 1.2859052252017824e-05,
      "loss": 0.397,
      "step": 4520
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.6369842228216813,
      "learning_rate": 1.2856130480999673e-05,
      "loss": 0.3555,
      "step": 4521
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.5544946718524273,
      "learning_rate": 1.2853208444480572e-05,
      "loss": 0.3438,
      "step": 4522
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.6559656332378037,
      "learning_rate": 1.2850286142732141e-05,
      "loss": 0.363,
      "step": 4523
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.6935926123984193,
      "learning_rate": 1.2847363576026037e-05,
      "loss": 0.3799,
      "step": 4524
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.66188770918031,
      "learning_rate": 1.2844440744633934e-05,
      "loss": 0.356,
      "step": 4525
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.7257872993063395,
      "learning_rate": 1.2841517648827538e-05,
      "loss": 0.415,
      "step": 4526
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.7499673610365512,
      "learning_rate": 1.2838594288878567e-05,
      "loss": 0.3955,
      "step": 4527
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.4264138157047788,
      "learning_rate": 1.2835670665058779e-05,
      "loss": 0.2854,
      "step": 4528
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.6537708625389769,
      "learning_rate": 1.2832746777639947e-05,
      "loss": 0.3594,
      "step": 4529
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.4397115872152197,
      "learning_rate": 1.2829822626893867e-05,
      "loss": 0.2646,
      "step": 4530
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.741091598587738,
      "learning_rate": 1.2826898213092365e-05,
      "loss": 0.3486,
      "step": 4531
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.9666912614730223,
      "learning_rate": 1.2823973536507295e-05,
      "loss": 0.4331,
      "step": 4532
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.693323413604661,
      "learning_rate": 1.282104859741052e-05,
      "loss": 0.3608,
      "step": 4533
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.6264578878697835,
      "learning_rate": 1.2818123396073942e-05,
      "loss": 0.3696,
      "step": 4534
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.5513120938297509,
      "learning_rate": 1.2815197932769486e-05,
      "loss": 0.2905,
      "step": 4535
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.5489927313756573,
      "learning_rate": 1.2812272207769092e-05,
      "loss": 0.3206,
      "step": 4536
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.639052834305715,
      "learning_rate": 1.2809346221344734e-05,
      "loss": 0.3096,
      "step": 4537
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.7437887279620445,
      "learning_rate": 1.2806419973768407e-05,
      "loss": 0.4214,
      "step": 4538
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.6531308253980825,
      "learning_rate": 1.280349346531213e-05,
      "loss": 0.3804,
      "step": 4539
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.7263611373852528,
      "learning_rate": 1.2800566696247943e-05,
      "loss": 0.377,
      "step": 4540
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.4670811296219826,
      "learning_rate": 1.2797639666847912e-05,
      "loss": 0.2915,
      "step": 4541
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.840968797906768,
      "learning_rate": 1.2794712377384135e-05,
      "loss": 0.3257,
      "step": 4542
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.833256176829522,
      "learning_rate": 1.2791784828128727e-05,
      "loss": 0.3599,
      "step": 4543
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.7124104740742703,
      "learning_rate": 1.2788857019353821e-05,
      "loss": 0.3645,
      "step": 4544
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.7385337350606462,
      "learning_rate": 1.278592895133159e-05,
      "loss": 0.3647,
      "step": 4545
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.6467561445622598,
      "learning_rate": 1.2783000624334214e-05,
      "loss": 0.3398,
      "step": 4546
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.6455835722215268,
      "learning_rate": 1.2780072038633913e-05,
      "loss": 0.3157,
      "step": 4547
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.5747766352342412,
      "learning_rate": 1.2777143194502915e-05,
      "loss": 0.2717,
      "step": 4548
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.6007472334955324,
      "learning_rate": 1.2774214092213489e-05,
      "loss": 0.3066,
      "step": 4549
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.571410720138621,
      "learning_rate": 1.2771284732037912e-05,
      "loss": 0.3101,
      "step": 4550
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.6623380257406997,
      "learning_rate": 1.2768355114248493e-05,
      "loss": 0.3398,
      "step": 4551
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.7229047403619318,
      "learning_rate": 1.2765425239117572e-05,
      "loss": 0.3394,
      "step": 4552
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.9033977293758697,
      "learning_rate": 1.2762495106917496e-05,
      "loss": 0.438,
      "step": 4553
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.788707189709312,
      "learning_rate": 1.2759564717920649e-05,
      "loss": 0.3374,
      "step": 4554
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.7355191887460462,
      "learning_rate": 1.2756634072399434e-05,
      "loss": 0.3618,
      "step": 4555
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.7213963860371162,
      "learning_rate": 1.2753703170626279e-05,
      "loss": 0.3394,
      "step": 4556
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.6865664908990747,
      "learning_rate": 1.2750772012873637e-05,
      "loss": 0.2932,
      "step": 4557
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.7054257193364892,
      "learning_rate": 1.2747840599413981e-05,
      "loss": 0.2993,
      "step": 4558
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.636956980289227,
      "learning_rate": 1.274490893051981e-05,
      "loss": 0.3203,
      "step": 4559
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.5676366855154549,
      "learning_rate": 1.274197700646365e-05,
      "loss": 0.335,
      "step": 4560
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.6255121118836142,
      "learning_rate": 1.2739044827518043e-05,
      "loss": 0.3452,
      "step": 4561
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.482240788441769,
      "learning_rate": 1.2736112393955567e-05,
      "loss": 0.313,
      "step": 4562
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.8394076483290858,
      "learning_rate": 1.2733179706048805e-05,
      "loss": 0.3354,
      "step": 4563
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.7543771765119571,
      "learning_rate": 1.2730246764070385e-05,
      "loss": 0.3765,
      "step": 4564
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.6542366464423763,
      "learning_rate": 1.2727313568292942e-05,
      "loss": 0.3174,
      "step": 4565
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.6718058592302316,
      "learning_rate": 1.2724380118989144e-05,
      "loss": 0.3484,
      "step": 4566
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.5697041774106322,
      "learning_rate": 1.2721446416431676e-05,
      "loss": 0.3633,
      "step": 4567
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.841926291770861,
      "learning_rate": 1.271851246089325e-05,
      "loss": 0.3398,
      "step": 4568
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.7491386664637214,
      "learning_rate": 1.2715578252646608e-05,
      "loss": 0.4021,
      "step": 4569
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.655469199750724,
      "learning_rate": 1.2712643791964501e-05,
      "loss": 0.3179,
      "step": 4570
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.4760286232715227,
      "learning_rate": 1.2709709079119712e-05,
      "loss": 0.2759,
      "step": 4571
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.9587397901997206,
      "learning_rate": 1.2706774114385054e-05,
      "loss": 0.4277,
      "step": 4572
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.536895771319605,
      "learning_rate": 1.270383889803335e-05,
      "loss": 0.3374,
      "step": 4573
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.8338887558243726,
      "learning_rate": 1.2700903430337456e-05,
      "loss": 0.3594,
      "step": 4574
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.517221834084261,
      "learning_rate": 1.2697967711570243e-05,
      "loss": 0.2837,
      "step": 4575
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.6029535345487154,
      "learning_rate": 1.2695031742004617e-05,
      "loss": 0.2925,
      "step": 4576
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.6848652261431267,
      "learning_rate": 1.2692095521913494e-05,
      "loss": 0.3354,
      "step": 4577
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.5776643151067156,
      "learning_rate": 1.2689159051569824e-05,
      "loss": 0.2932,
      "step": 4578
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.970570901820663,
      "learning_rate": 1.268622233124658e-05,
      "loss": 0.4575,
      "step": 4579
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.6806543618812027,
      "learning_rate": 1.2683285361216745e-05,
      "loss": 0.3994,
      "step": 4580
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.6797055346192082,
      "learning_rate": 1.268034814175334e-05,
      "loss": 0.321,
      "step": 4581
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.750507603282102,
      "learning_rate": 1.2677410673129406e-05,
      "loss": 0.3721,
      "step": 4582
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.6989614785681186,
      "learning_rate": 1.2674472955618001e-05,
      "loss": 0.3235,
      "step": 4583
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.7006677013484721,
      "learning_rate": 1.2671534989492209e-05,
      "loss": 0.2834,
      "step": 4584
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.790010471279744,
      "learning_rate": 1.2668596775025143e-05,
      "loss": 0.3657,
      "step": 4585
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.7168427041624408,
      "learning_rate": 1.2665658312489932e-05,
      "loss": 0.3721,
      "step": 4586
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.777042177855204,
      "learning_rate": 1.2662719602159732e-05,
      "loss": 0.4231,
      "step": 4587
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.7592805225806039,
      "learning_rate": 1.2659780644307718e-05,
      "loss": 0.3618,
      "step": 4588
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.5140424634652145,
      "learning_rate": 1.2656841439207093e-05,
      "loss": 0.3525,
      "step": 4589
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.5820589174673925,
      "learning_rate": 1.2653901987131074e-05,
      "loss": 0.2815,
      "step": 4590
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.3080778229913583,
      "learning_rate": 1.2650962288352916e-05,
      "loss": 0.2559,
      "step": 4591
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.9815303569253928,
      "learning_rate": 1.2648022343145885e-05,
      "loss": 0.4155,
      "step": 4592
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.5562438277679074,
      "learning_rate": 1.2645082151783272e-05,
      "loss": 0.3486,
      "step": 4593
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.6914906604746451,
      "learning_rate": 1.2642141714538391e-05,
      "loss": 0.3706,
      "step": 4594
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.7296284361532792,
      "learning_rate": 1.2639201031684584e-05,
      "loss": 0.3625,
      "step": 4595
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.6054178065298481,
      "learning_rate": 1.2636260103495209e-05,
      "loss": 0.3337,
      "step": 4596
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.7084497539600987,
      "learning_rate": 1.2633318930243647e-05,
      "loss": 0.3911,
      "step": 4597
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.6016662220035636,
      "learning_rate": 1.2630377512203314e-05,
      "loss": 0.3289,
      "step": 4598
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.7150181061179122,
      "learning_rate": 1.2627435849647629e-05,
      "loss": 0.3604,
      "step": 4599
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.767462765462553,
      "learning_rate": 1.262449394285005e-05,
      "loss": 0.3945,
      "step": 4600
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.7505973901489895,
      "learning_rate": 1.2621551792084045e-05,
      "loss": 0.3159,
      "step": 4601
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.5029828510711156,
      "learning_rate": 1.2618609397623121e-05,
      "loss": 0.2974,
      "step": 4602
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.4984604767705212,
      "learning_rate": 1.2615666759740788e-05,
      "loss": 0.2749,
      "step": 4603
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.6973484067682179,
      "learning_rate": 1.2612723878710594e-05,
      "loss": 0.334,
      "step": 4604
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.4080406349460697,
      "learning_rate": 1.2609780754806105e-05,
      "loss": 0.2302,
      "step": 4605
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.7055344163830708,
      "learning_rate": 1.2606837388300905e-05,
      "loss": 0.3262,
      "step": 4606
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.9762143133253172,
      "learning_rate": 1.2603893779468604e-05,
      "loss": 0.4336,
      "step": 4607
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.8764935706783656,
      "learning_rate": 1.2600949928582839e-05,
      "loss": 0.4849,
      "step": 4608
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.7079046838946064,
      "learning_rate": 1.2598005835917263e-05,
      "loss": 0.3311,
      "step": 4609
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.7408018809613945,
      "learning_rate": 1.2595061501745556e-05,
      "loss": 0.3442,
      "step": 4610
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.9595758804698435,
      "learning_rate": 1.2592116926341411e-05,
      "loss": 0.4741,
      "step": 4611
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.8989458044184206,
      "learning_rate": 1.258917210997856e-05,
      "loss": 0.3599,
      "step": 4612
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.5005106507479178,
      "learning_rate": 1.2586227052930742e-05,
      "loss": 0.2844,
      "step": 4613
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.6488079054073232,
      "learning_rate": 1.2583281755471726e-05,
      "loss": 0.3628,
      "step": 4614
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.4467625116530698,
      "learning_rate": 1.2580336217875303e-05,
      "loss": 0.2527,
      "step": 4615
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.5675923703694017,
      "learning_rate": 1.2577390440415284e-05,
      "loss": 0.2842,
      "step": 4616
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.7047821421914997,
      "learning_rate": 1.2574444423365503e-05,
      "loss": 0.3389,
      "step": 4617
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.651334797333183,
      "learning_rate": 1.2571498166999816e-05,
      "loss": 0.3,
      "step": 4618
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.6754119566352306,
      "learning_rate": 1.2568551671592106e-05,
      "loss": 0.3025,
      "step": 4619
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.7469015711269238,
      "learning_rate": 1.2565604937416267e-05,
      "loss": 0.3843,
      "step": 4620
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.9309176741740264,
      "learning_rate": 1.256265796474623e-05,
      "loss": 0.4619,
      "step": 4621
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.5272637743829354,
      "learning_rate": 1.2559710753855937e-05,
      "loss": 0.353,
      "step": 4622
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.7417463058720057,
      "learning_rate": 1.2556763305019353e-05,
      "loss": 0.4414,
      "step": 4623
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.5905896578844092,
      "learning_rate": 1.255381561851047e-05,
      "loss": 0.3159,
      "step": 4624
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.7541825740737895,
      "learning_rate": 1.2550867694603302e-05,
      "loss": 0.3169,
      "step": 4625
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.8602263792851315,
      "learning_rate": 1.2547919533571879e-05,
      "loss": 0.3823,
      "step": 4626
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.3596438004967315,
      "learning_rate": 1.2544971135690263e-05,
      "loss": 0.2209,
      "step": 4627
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.8721296717772624,
      "learning_rate": 1.2542022501232522e-05,
      "loss": 0.4067,
      "step": 4628
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.6602831666462705,
      "learning_rate": 1.2539073630472768e-05,
      "loss": 0.3108,
      "step": 4629
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.6994369261391387,
      "learning_rate": 1.2536124523685114e-05,
      "loss": 0.2727,
      "step": 4630
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.8836795022979826,
      "learning_rate": 1.2533175181143704e-05,
      "loss": 0.3389,
      "step": 4631
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.9068618211614181,
      "learning_rate": 1.2530225603122713e-05,
      "loss": 0.3643,
      "step": 4632
    },
    {
      "epoch": 1.31,
      "grad_norm": 2.1162960774869264,
      "learning_rate": 1.2527275789896315e-05,
      "loss": 0.4341,
      "step": 4633
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.6648053182630194,
      "learning_rate": 1.2524325741738732e-05,
      "loss": 0.374,
      "step": 4634
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.7264797288484155,
      "learning_rate": 1.2521375458924187e-05,
      "loss": 0.3784,
      "step": 4635
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.829473846746986,
      "learning_rate": 1.2518424941726939e-05,
      "loss": 0.3433,
      "step": 4636
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.809846482476772,
      "learning_rate": 1.2515474190421258e-05,
      "loss": 0.3809,
      "step": 4637
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.789023305262927,
      "learning_rate": 1.2512523205281444e-05,
      "loss": 0.3137,
      "step": 4638
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.6721812317630147,
      "learning_rate": 1.2509571986581814e-05,
      "loss": 0.3359,
      "step": 4639
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.6666246239193316,
      "learning_rate": 1.2506620534596711e-05,
      "loss": 0.3462,
      "step": 4640
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.6794336092890385,
      "learning_rate": 1.250366884960049e-05,
      "loss": 0.3467,
      "step": 4641
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.716090104538274,
      "learning_rate": 1.2500716931867543e-05,
      "loss": 0.3154,
      "step": 4642
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.8392540589464013,
      "learning_rate": 1.249776478167227e-05,
      "loss": 0.356,
      "step": 4643
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.6752596623253246,
      "learning_rate": 1.24948123992891e-05,
      "loss": 0.3208,
      "step": 4644
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.8283889197515715,
      "learning_rate": 1.2491859784992477e-05,
      "loss": 0.3945,
      "step": 4645
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.7045750380626028,
      "learning_rate": 1.248890693905688e-05,
      "loss": 0.3015,
      "step": 4646
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.5622287622813078,
      "learning_rate": 1.248595386175679e-05,
      "loss": 0.3228,
      "step": 4647
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.6904664315120712,
      "learning_rate": 1.2483000553366727e-05,
      "loss": 0.4038,
      "step": 4648
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.7784318731184303,
      "learning_rate": 1.2480047014161223e-05,
      "loss": 0.3911,
      "step": 4649
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.6608144610562632,
      "learning_rate": 1.247709324441483e-05,
      "loss": 0.3398,
      "step": 4650
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.6505562553452917,
      "learning_rate": 1.2474139244402134e-05,
      "loss": 0.3022,
      "step": 4651
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.772329061610606,
      "learning_rate": 1.2471185014397728e-05,
      "loss": 0.3506,
      "step": 4652
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.726386092959217,
      "learning_rate": 1.2468230554676235e-05,
      "loss": 0.3525,
      "step": 4653
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.8015378875899222,
      "learning_rate": 1.2465275865512291e-05,
      "loss": 0.384,
      "step": 4654
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.8800519915975427,
      "learning_rate": 1.2462320947180565e-05,
      "loss": 0.3442,
      "step": 4655
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.8086692079400846,
      "learning_rate": 1.2459365799955741e-05,
      "loss": 0.4946,
      "step": 4656
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.7645181473715132,
      "learning_rate": 1.245641042411252e-05,
      "loss": 0.3372,
      "step": 4657
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.4554893765696777,
      "learning_rate": 1.2453454819925627e-05,
      "loss": 0.3013,
      "step": 4658
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.7822071525541143,
      "learning_rate": 1.245049898766982e-05,
      "loss": 0.4272,
      "step": 4659
    },
    {
      "epoch": 1.32,
      "grad_norm": 2.2135495246166235,
      "learning_rate": 1.2447542927619857e-05,
      "loss": 0.3926,
      "step": 4660
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.6736103269127196,
      "learning_rate": 1.2444586640050536e-05,
      "loss": 0.3159,
      "step": 4661
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.7596653590130846,
      "learning_rate": 1.2441630125236664e-05,
      "loss": 0.311,
      "step": 4662
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.8296567269118633,
      "learning_rate": 1.2438673383453073e-05,
      "loss": 0.3403,
      "step": 4663
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.6769758234667826,
      "learning_rate": 1.243571641497462e-05,
      "loss": 0.396,
      "step": 4664
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.648090741739503,
      "learning_rate": 1.2432759220076177e-05,
      "loss": 0.3735,
      "step": 4665
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.7501841293761073,
      "learning_rate": 1.242980179903264e-05,
      "loss": 0.3501,
      "step": 4666
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.6777694433377468,
      "learning_rate": 1.2426844152118926e-05,
      "loss": 0.374,
      "step": 4667
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.8864485162641331,
      "learning_rate": 1.2423886279609975e-05,
      "loss": 0.3721,
      "step": 4668
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.6828965972634766,
      "learning_rate": 1.2420928181780745e-05,
      "loss": 0.3352,
      "step": 4669
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.8248143133933372,
      "learning_rate": 1.2417969858906214e-05,
      "loss": 0.3901,
      "step": 4670
    },
    {
      "epoch": 1.32,
      "grad_norm": 2.0356223650052416,
      "learning_rate": 1.241501131126138e-05,
      "loss": 0.4062,
      "step": 4671
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.607470399218247,
      "learning_rate": 1.2412052539121273e-05,
      "loss": 0.3467,
      "step": 4672
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.7888256593260798,
      "learning_rate": 1.2409093542760925e-05,
      "loss": 0.3389,
      "step": 4673
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.6679352094167827,
      "learning_rate": 1.240613432245541e-05,
      "loss": 0.3218,
      "step": 4674
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.7015759113377547,
      "learning_rate": 1.2403174878479802e-05,
      "loss": 0.3711,
      "step": 4675
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.7620300733652037,
      "learning_rate": 1.240021521110921e-05,
      "loss": 0.4048,
      "step": 4676
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.5921196509795852,
      "learning_rate": 1.239725532061876e-05,
      "loss": 0.3281,
      "step": 4677
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.9682961353046895,
      "learning_rate": 1.2394295207283598e-05,
      "loss": 0.4258,
      "step": 4678
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.637831240157806,
      "learning_rate": 1.239133487137889e-05,
      "loss": 0.3267,
      "step": 4679
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.5701844465166832,
      "learning_rate": 1.2388374313179828e-05,
      "loss": 0.344,
      "step": 4680
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.7269496958991126,
      "learning_rate": 1.2385413532961612e-05,
      "loss": 0.3711,
      "step": 4681
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.7631908118813528,
      "learning_rate": 1.2382452530999479e-05,
      "loss": 0.3306,
      "step": 4682
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.7508991927082835,
      "learning_rate": 1.2379491307568676e-05,
      "loss": 0.3655,
      "step": 4683
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.626131194202727,
      "learning_rate": 1.2376529862944472e-05,
      "loss": 0.3364,
      "step": 4684
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.9619342729101197,
      "learning_rate": 1.2373568197402157e-05,
      "loss": 0.4424,
      "step": 4685
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.9821658057516551,
      "learning_rate": 1.2370606311217047e-05,
      "loss": 0.4194,
      "step": 4686
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.5827975457284955,
      "learning_rate": 1.2367644204664468e-05,
      "loss": 0.3394,
      "step": 4687
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.8073751188053275,
      "learning_rate": 1.2364681878019776e-05,
      "loss": 0.2979,
      "step": 4688
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.7884919635470555,
      "learning_rate": 1.2361719331558346e-05,
      "loss": 0.2974,
      "step": 4689
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.510467461629062,
      "learning_rate": 1.2358756565555563e-05,
      "loss": 0.2783,
      "step": 4690
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.685256695761019,
      "learning_rate": 1.2355793580286848e-05,
      "loss": 0.3599,
      "step": 4691
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.9802195400301865,
      "learning_rate": 1.2352830376027634e-05,
      "loss": 0.4355,
      "step": 4692
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.6347822770374192,
      "learning_rate": 1.2349866953053372e-05,
      "loss": 0.3564,
      "step": 4693
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.8583053067726927,
      "learning_rate": 1.2346903311639537e-05,
      "loss": 0.3325,
      "step": 4694
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.5568140078618873,
      "learning_rate": 1.2343939452061628e-05,
      "loss": 0.2781,
      "step": 4695
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.2974572765564245,
      "learning_rate": 1.2340975374595157e-05,
      "loss": 0.2825,
      "step": 4696
    },
    {
      "epoch": 1.33,
      "grad_norm": 2.3031088885870137,
      "learning_rate": 1.2338011079515661e-05,
      "loss": 0.4937,
      "step": 4697
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.6958835455085401,
      "learning_rate": 1.2335046567098694e-05,
      "loss": 0.3599,
      "step": 4698
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.7058098376571658,
      "learning_rate": 1.2332081837619836e-05,
      "loss": 0.3357,
      "step": 4699
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.8413795552010177,
      "learning_rate": 1.2329116891354677e-05,
      "loss": 0.3438,
      "step": 4700
    },
    {
      "epoch": 1.33,
      "grad_norm": 2.0764554586609694,
      "learning_rate": 1.2326151728578839e-05,
      "loss": 0.4385,
      "step": 4701
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.6976868139028305,
      "learning_rate": 1.2323186349567955e-05,
      "loss": 0.3311,
      "step": 4702
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.6908132054912934,
      "learning_rate": 1.232022075459768e-05,
      "loss": 0.3059,
      "step": 4703
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.551335439554422,
      "learning_rate": 1.2317254943943695e-05,
      "loss": 0.3306,
      "step": 4704
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.498964352101603,
      "learning_rate": 1.2314288917881696e-05,
      "loss": 0.3264,
      "step": 4705
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.5400113475747554,
      "learning_rate": 1.2311322676687398e-05,
      "loss": 0.3169,
      "step": 4706
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.5917795711331806,
      "learning_rate": 1.2308356220636535e-05,
      "loss": 0.2979,
      "step": 4707
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.5609657225184326,
      "learning_rate": 1.230538955000487e-05,
      "loss": 0.29,
      "step": 4708
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.7303811261115936,
      "learning_rate": 1.2302422665068174e-05,
      "loss": 0.3452,
      "step": 4709
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.672297055926268,
      "learning_rate": 1.2299455566102248e-05,
      "loss": 0.3467,
      "step": 4710
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.6401144100916927,
      "learning_rate": 1.2296488253382902e-05,
      "loss": 0.334,
      "step": 4711
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.7429336627657004,
      "learning_rate": 1.229352072718598e-05,
      "loss": 0.3721,
      "step": 4712
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.7661911702024566,
      "learning_rate": 1.2290552987787332e-05,
      "loss": 0.3589,
      "step": 4713
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.7668872167448513,
      "learning_rate": 1.2287585035462838e-05,
      "loss": 0.3105,
      "step": 4714
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.6539891446352941,
      "learning_rate": 1.228461687048839e-05,
      "loss": 0.2983,
      "step": 4715
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.8304874058572185,
      "learning_rate": 1.2281648493139911e-05,
      "loss": 0.3313,
      "step": 4716
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.7887787463750633,
      "learning_rate": 1.2278679903693325e-05,
      "loss": 0.3789,
      "step": 4717
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.653859198097434,
      "learning_rate": 1.2275711102424595e-05,
      "loss": 0.3438,
      "step": 4718
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.8116784745537555,
      "learning_rate": 1.2272742089609694e-05,
      "loss": 0.3789,
      "step": 4719
    },
    {
      "epoch": 1.33,
      "grad_norm": 2.1055203941014,
      "learning_rate": 1.2269772865524612e-05,
      "loss": 0.4668,
      "step": 4720
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.8244247807199467,
      "learning_rate": 1.2266803430445372e-05,
      "loss": 0.3022,
      "step": 4721
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.975440201706423,
      "learning_rate": 1.2263833784647998e-05,
      "loss": 0.4126,
      "step": 4722
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.7557938732893819,
      "learning_rate": 1.2260863928408551e-05,
      "loss": 0.3438,
      "step": 4723
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.9534186785074945,
      "learning_rate": 1.2257893862003093e-05,
      "loss": 0.3687,
      "step": 4724
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.6457272855644576,
      "learning_rate": 1.225492358570773e-05,
      "loss": 0.3384,
      "step": 4725
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.6289021977566196,
      "learning_rate": 1.2251953099798566e-05,
      "loss": 0.3191,
      "step": 4726
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.6288810363194637,
      "learning_rate": 1.2248982404551733e-05,
      "loss": 0.3071,
      "step": 4727
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.8600828372624327,
      "learning_rate": 1.224601150024338e-05,
      "loss": 0.4404,
      "step": 4728
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.8876593228313279,
      "learning_rate": 1.2243040387149682e-05,
      "loss": 0.3584,
      "step": 4729
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.9237905383626541,
      "learning_rate": 1.2240069065546823e-05,
      "loss": 0.4497,
      "step": 4730
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.823047705767659,
      "learning_rate": 1.2237097535711016e-05,
      "loss": 0.3774,
      "step": 4731
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.8793772934920572,
      "learning_rate": 1.2234125797918488e-05,
      "loss": 0.374,
      "step": 4732
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.7696823593006847,
      "learning_rate": 1.2231153852445485e-05,
      "loss": 0.3721,
      "step": 4733
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.638990634620649,
      "learning_rate": 1.2228181699568276e-05,
      "loss": 0.3462,
      "step": 4734
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.762739009322588,
      "learning_rate": 1.2225209339563144e-05,
      "loss": 0.3279,
      "step": 4735
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.6507213578978446,
      "learning_rate": 1.2222236772706402e-05,
      "loss": 0.3867,
      "step": 4736
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.7693050604084883,
      "learning_rate": 1.2219263999274367e-05,
      "loss": 0.3357,
      "step": 4737
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.4783831762446826,
      "learning_rate": 1.2216291019543385e-05,
      "loss": 0.3418,
      "step": 4738
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.7036094020262598,
      "learning_rate": 1.221331783378982e-05,
      "loss": 0.3857,
      "step": 4739
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.6631647160026093,
      "learning_rate": 1.2210344442290054e-05,
      "loss": 0.3555,
      "step": 4740
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.7345228123289045,
      "learning_rate": 1.2207370845320488e-05,
      "loss": 0.3445,
      "step": 4741
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.668051291063268,
      "learning_rate": 1.2204397043157541e-05,
      "loss": 0.3877,
      "step": 4742
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.624458221908496,
      "learning_rate": 1.2201423036077657e-05,
      "loss": 0.3589,
      "step": 4743
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.7087307701591414,
      "learning_rate": 1.2198448824357292e-05,
      "loss": 0.3027,
      "step": 4744
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.6095932095002772,
      "learning_rate": 1.2195474408272919e-05,
      "loss": 0.3665,
      "step": 4745
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.784080461073997,
      "learning_rate": 1.2192499788101044e-05,
      "loss": 0.3965,
      "step": 4746
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.7736620343489595,
      "learning_rate": 1.2189524964118174e-05,
      "loss": 0.3208,
      "step": 4747
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.6496401197862929,
      "learning_rate": 1.2186549936600847e-05,
      "loss": 0.2812,
      "step": 4748
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.6726524178038849,
      "learning_rate": 1.2183574705825618e-05,
      "loss": 0.3013,
      "step": 4749
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.5105390331347917,
      "learning_rate": 1.2180599272069058e-05,
      "loss": 0.2798,
      "step": 4750
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.5424629113065988,
      "learning_rate": 1.2177623635607753e-05,
      "loss": 0.2852,
      "step": 4751
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.7633373419597802,
      "learning_rate": 1.2174647796718322e-05,
      "loss": 0.3398,
      "step": 4752
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.7293537369154643,
      "learning_rate": 1.217167175567739e-05,
      "loss": 0.3252,
      "step": 4753
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.5735631495623699,
      "learning_rate": 1.2168695512761604e-05,
      "loss": 0.3291,
      "step": 4754
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.8373929557430868,
      "learning_rate": 1.2165719068247626e-05,
      "loss": 0.3447,
      "step": 4755
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.7174057298421266,
      "learning_rate": 1.216274242241215e-05,
      "loss": 0.2729,
      "step": 4756
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.8465819994691137,
      "learning_rate": 1.2159765575531877e-05,
      "loss": 0.4072,
      "step": 4757
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.9393725167046114,
      "learning_rate": 1.2156788527883524e-05,
      "loss": 0.3325,
      "step": 4758
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.820469668308305,
      "learning_rate": 1.2153811279743841e-05,
      "loss": 0.3342,
      "step": 4759
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.8347428669514536,
      "learning_rate": 1.215083383138958e-05,
      "loss": 0.3882,
      "step": 4760
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.767559080987566,
      "learning_rate": 1.2147856183097524e-05,
      "loss": 0.3413,
      "step": 4761
    },
    {
      "epoch": 1.34,
      "grad_norm": 2.0971358856584636,
      "learning_rate": 1.2144878335144469e-05,
      "loss": 0.4175,
      "step": 4762
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.5109159826746703,
      "learning_rate": 1.214190028780723e-05,
      "loss": 0.312,
      "step": 4763
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.7451281128853664,
      "learning_rate": 1.213892204136264e-05,
      "loss": 0.3203,
      "step": 4764
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.671660757230755,
      "learning_rate": 1.2135943596087554e-05,
      "loss": 0.3223,
      "step": 4765
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.7206567927551566,
      "learning_rate": 1.2132964952258841e-05,
      "loss": 0.3516,
      "step": 4766
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.7228403901703,
      "learning_rate": 1.2129986110153395e-05,
      "loss": 0.3647,
      "step": 4767
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.7173111492959447,
      "learning_rate": 1.2127007070048117e-05,
      "loss": 0.394,
      "step": 4768
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.6842977665480772,
      "learning_rate": 1.2124027832219942e-05,
      "loss": 0.3438,
      "step": 4769
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.7780281534661158,
      "learning_rate": 1.2121048396945807e-05,
      "loss": 0.3428,
      "step": 4770
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.8036424662197883,
      "learning_rate": 1.2118068764502677e-05,
      "loss": 0.3545,
      "step": 4771
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.6308734213787721,
      "learning_rate": 1.2115088935167538e-05,
      "loss": 0.3584,
      "step": 4772
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.485052745911109,
      "learning_rate": 1.2112108909217386e-05,
      "loss": 0.2854,
      "step": 4773
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.8705621048076044,
      "learning_rate": 1.2109128686929235e-05,
      "loss": 0.3647,
      "step": 4774
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.755775255945816,
      "learning_rate": 1.210614826858013e-05,
      "loss": 0.3906,
      "step": 4775
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.8361949458886273,
      "learning_rate": 1.2103167654447121e-05,
      "loss": 0.3481,
      "step": 4776
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.620687291201517,
      "learning_rate": 1.210018684480728e-05,
      "loss": 0.3589,
      "step": 4777
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.6744771157622131,
      "learning_rate": 1.20972058399377e-05,
      "loss": 0.3049,
      "step": 4778
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.7464936154494894,
      "learning_rate": 1.2094224640115488e-05,
      "loss": 0.3489,
      "step": 4779
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.623478964812256,
      "learning_rate": 1.2091243245617774e-05,
      "loss": 0.2847,
      "step": 4780
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.6426906533767083,
      "learning_rate": 1.20882616567217e-05,
      "loss": 0.3096,
      "step": 4781
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.760129547088209,
      "learning_rate": 1.2085279873704433e-05,
      "loss": 0.3716,
      "step": 4782
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.9846434197712834,
      "learning_rate": 1.208229789684315e-05,
      "loss": 0.4556,
      "step": 4783
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.8515486466038389,
      "learning_rate": 1.2079315726415053e-05,
      "loss": 0.3623,
      "step": 4784
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.5952766440259336,
      "learning_rate": 1.2076333362697358e-05,
      "loss": 0.2805,
      "step": 4785
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.4549773760851576,
      "learning_rate": 1.2073350805967308e-05,
      "loss": 0.2566,
      "step": 4786
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.7817847860667542,
      "learning_rate": 1.2070368056502142e-05,
      "loss": 0.3628,
      "step": 4787
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.9021802755283768,
      "learning_rate": 1.2067385114579144e-05,
      "loss": 0.3945,
      "step": 4788
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.4887866243724737,
      "learning_rate": 1.2064401980475595e-05,
      "loss": 0.2461,
      "step": 4789
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.9192562815568797,
      "learning_rate": 1.2061418654468808e-05,
      "loss": 0.3718,
      "step": 4790
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.9236784669654394,
      "learning_rate": 1.2058435136836101e-05,
      "loss": 0.4243,
      "step": 4791
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.859439073481088,
      "learning_rate": 1.2055451427854825e-05,
      "loss": 0.3494,
      "step": 4792
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.5285569667482066,
      "learning_rate": 1.2052467527802337e-05,
      "loss": 0.2976,
      "step": 4793
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.8495624158395516,
      "learning_rate": 1.2049483436956009e-05,
      "loss": 0.4224,
      "step": 4794
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.6755489486732653,
      "learning_rate": 1.2046499155593245e-05,
      "loss": 0.3184,
      "step": 4795
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.6710182393487787,
      "learning_rate": 1.2043514683991455e-05,
      "loss": 0.2947,
      "step": 4796
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.7611696092897178,
      "learning_rate": 1.2040530022428074e-05,
      "loss": 0.3525,
      "step": 4797
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.6022974407295207,
      "learning_rate": 1.2037545171180545e-05,
      "loss": 0.3145,
      "step": 4798
    },
    {
      "epoch": 1.36,
      "grad_norm": 2.0529927220538915,
      "learning_rate": 1.2034560130526341e-05,
      "loss": 0.3916,
      "step": 4799
    },
    {
      "epoch": 1.36,
      "grad_norm": 2.0886933411152797,
      "learning_rate": 1.203157490074294e-05,
      "loss": 0.3904,
      "step": 4800
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.8674963911436302,
      "learning_rate": 1.2028589482107847e-05,
      "loss": 0.4541,
      "step": 4801
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.6588527273658042,
      "learning_rate": 1.2025603874898582e-05,
      "loss": 0.3105,
      "step": 4802
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.4869034660270948,
      "learning_rate": 1.2022618079392683e-05,
      "loss": 0.283,
      "step": 4803
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.8037533356491045,
      "learning_rate": 1.2019632095867697e-05,
      "loss": 0.4009,
      "step": 4804
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.7481760552853958,
      "learning_rate": 1.2016645924601207e-05,
      "loss": 0.332,
      "step": 4805
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.818125911266759,
      "learning_rate": 1.2013659565870795e-05,
      "loss": 0.3628,
      "step": 4806
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.844612941086835,
      "learning_rate": 1.201067301995407e-05,
      "loss": 0.4404,
      "step": 4807
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.978808978523622,
      "learning_rate": 1.200768628712865e-05,
      "loss": 0.3491,
      "step": 4808
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.774285067039147,
      "learning_rate": 1.2004699367672187e-05,
      "loss": 0.3115,
      "step": 4809
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.6740229387628522,
      "learning_rate": 1.2001712261862335e-05,
      "loss": 0.2878,
      "step": 4810
    },
    {
      "epoch": 1.36,
      "grad_norm": 2.0142354001873994,
      "learning_rate": 1.1998724969976767e-05,
      "loss": 0.4785,
      "step": 4811
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.7177765713844477,
      "learning_rate": 1.1995737492293183e-05,
      "loss": 0.4092,
      "step": 4812
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.6190218046422789,
      "learning_rate": 1.199274982908929e-05,
      "loss": 0.3101,
      "step": 4813
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.5799269045175266,
      "learning_rate": 1.1989761980642816e-05,
      "loss": 0.3286,
      "step": 4814
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.7579516712303975,
      "learning_rate": 1.1986773947231505e-05,
      "loss": 0.3828,
      "step": 4815
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.688707265044341,
      "learning_rate": 1.1983785729133125e-05,
      "loss": 0.3652,
      "step": 4816
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.8220744349723499,
      "learning_rate": 1.1980797326625446e-05,
      "loss": 0.325,
      "step": 4817
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.5557607069613184,
      "learning_rate": 1.1977808739986275e-05,
      "loss": 0.3164,
      "step": 4818
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.7035732932623262,
      "learning_rate": 1.1974819969493421e-05,
      "loss": 0.3901,
      "step": 4819
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.75346496950868,
      "learning_rate": 1.1971831015424713e-05,
      "loss": 0.3271,
      "step": 4820
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.4194524843631848,
      "learning_rate": 1.1968841878057999e-05,
      "loss": 0.2766,
      "step": 4821
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.5897208642379885,
      "learning_rate": 1.196585255767115e-05,
      "loss": 0.2893,
      "step": 4822
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.6240247840093427,
      "learning_rate": 1.1962863054542045e-05,
      "loss": 0.3677,
      "step": 4823
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.637563567194311,
      "learning_rate": 1.195987336894858e-05,
      "loss": 0.3877,
      "step": 4824
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.922812754448883,
      "learning_rate": 1.1956883501168672e-05,
      "loss": 0.3667,
      "step": 4825
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.809732561662456,
      "learning_rate": 1.195389345148026e-05,
      "loss": 0.4053,
      "step": 4826
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.5078687750386095,
      "learning_rate": 1.1950903220161286e-05,
      "loss": 0.3447,
      "step": 4827
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.531383211235644,
      "learning_rate": 1.1947912807489716e-05,
      "loss": 0.2695,
      "step": 4828
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.6540223598597739,
      "learning_rate": 1.1944922213743543e-05,
      "loss": 0.2974,
      "step": 4829
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.7416031524381528,
      "learning_rate": 1.194193143920076e-05,
      "loss": 0.3118,
      "step": 4830
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.8285662437436476,
      "learning_rate": 1.1938940484139387e-05,
      "loss": 0.3491,
      "step": 4831
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.6637005231346742,
      "learning_rate": 1.1935949348837457e-05,
      "loss": 0.2966,
      "step": 4832
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.6242285308793354,
      "learning_rate": 1.193295803357302e-05,
      "loss": 0.353,
      "step": 4833
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.760234415185392,
      "learning_rate": 1.1929966538624143e-05,
      "loss": 0.3452,
      "step": 4834
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.475499568067183,
      "learning_rate": 1.1926974864268914e-05,
      "loss": 0.3054,
      "step": 4835
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.7532849965574515,
      "learning_rate": 1.1923983010785431e-05,
      "loss": 0.3784,
      "step": 4836
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.7423984856035288,
      "learning_rate": 1.1920990978451812e-05,
      "loss": 0.3481,
      "step": 4837
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.7641642880281212,
      "learning_rate": 1.1917998767546188e-05,
      "loss": 0.3252,
      "step": 4838
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.9525567478378183,
      "learning_rate": 1.1915006378346719e-05,
      "loss": 0.3533,
      "step": 4839
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.6615048483928325,
      "learning_rate": 1.1912013811131562e-05,
      "loss": 0.3677,
      "step": 4840
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.6346913524735487,
      "learning_rate": 1.1909021066178906e-05,
      "loss": 0.3916,
      "step": 4841
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.6366192101063435,
      "learning_rate": 1.1906028143766952e-05,
      "loss": 0.332,
      "step": 4842
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.7124141813278813,
      "learning_rate": 1.1903035044173914e-05,
      "loss": 0.3418,
      "step": 4843
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.7242285032665228,
      "learning_rate": 1.1900041767678024e-05,
      "loss": 0.3291,
      "step": 4844
    },
    {
      "epoch": 1.37,
      "grad_norm": 2.1929649775822555,
      "learning_rate": 1.1897048314557538e-05,
      "loss": 0.4829,
      "step": 4845
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.6153089776790435,
      "learning_rate": 1.189405468509072e-05,
      "loss": 0.3337,
      "step": 4846
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.7150189930567685,
      "learning_rate": 1.1891060879555847e-05,
      "loss": 0.3379,
      "step": 4847
    },
    {
      "epoch": 1.37,
      "grad_norm": 2.0223913116972416,
      "learning_rate": 1.1888066898231223e-05,
      "loss": 0.438,
      "step": 4848
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.7178078326116493,
      "learning_rate": 1.1885072741395165e-05,
      "loss": 0.3276,
      "step": 4849
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.7931902333280716,
      "learning_rate": 1.1882078409326003e-05,
      "loss": 0.397,
      "step": 4850
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.566691038861583,
      "learning_rate": 1.1879083902302082e-05,
      "loss": 0.3442,
      "step": 4851
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.6275145153424535,
      "learning_rate": 1.187608922060177e-05,
      "loss": 0.3027,
      "step": 4852
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.7174390267567405,
      "learning_rate": 1.1873094364503446e-05,
      "loss": 0.4277,
      "step": 4853
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.6801008707324352,
      "learning_rate": 1.1870099334285507e-05,
      "loss": 0.3877,
      "step": 4854
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.5886899256302578,
      "learning_rate": 1.1867104130226363e-05,
      "loss": 0.3433,
      "step": 4855
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.7703686584740501,
      "learning_rate": 1.1864108752604451e-05,
      "loss": 0.3926,
      "step": 4856
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.8112161051792726,
      "learning_rate": 1.1861113201698204e-05,
      "loss": 0.374,
      "step": 4857
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.835130951507719,
      "learning_rate": 1.1858117477786094e-05,
      "loss": 0.3677,
      "step": 4858
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.5998127831655935,
      "learning_rate": 1.1855121581146591e-05,
      "loss": 0.3423,
      "step": 4859
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.5291570406195085,
      "learning_rate": 1.1852125512058194e-05,
      "loss": 0.2817,
      "step": 4860
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.6177353635862446,
      "learning_rate": 1.1849129270799407e-05,
      "loss": 0.2893,
      "step": 4861
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.7191094317637687,
      "learning_rate": 1.184613285764876e-05,
      "loss": 0.3896,
      "step": 4862
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.8735819908781142,
      "learning_rate": 1.1843136272884795e-05,
      "loss": 0.3669,
      "step": 4863
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.7517511988161765,
      "learning_rate": 1.1840139516786062e-05,
      "loss": 0.3494,
      "step": 4864
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.5822558903959516,
      "learning_rate": 1.1837142589631143e-05,
      "loss": 0.2903,
      "step": 4865
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.5067564090526457,
      "learning_rate": 1.1834145491698623e-05,
      "loss": 0.2891,
      "step": 4866
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.6200242105940454,
      "learning_rate": 1.1831148223267108e-05,
      "loss": 0.2847,
      "step": 4867
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.9304665045063392,
      "learning_rate": 1.1828150784615216e-05,
      "loss": 0.4116,
      "step": 4868
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.590336796398658,
      "learning_rate": 1.1825153176021591e-05,
      "loss": 0.2832,
      "step": 4869
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.7886233670001348,
      "learning_rate": 1.1822155397764873e-05,
      "loss": 0.4199,
      "step": 4870
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.9787199294433209,
      "learning_rate": 1.1819157450123745e-05,
      "loss": 0.3906,
      "step": 4871
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.686786216883196,
      "learning_rate": 1.1816159333376882e-05,
      "loss": 0.2991,
      "step": 4872
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.7179040474810368,
      "learning_rate": 1.1813161047802986e-05,
      "loss": 0.3394,
      "step": 4873
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.8934567402165978,
      "learning_rate": 1.1810162593680768e-05,
      "loss": 0.3318,
      "step": 4874
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.7914818343957333,
      "learning_rate": 1.1807163971288967e-05,
      "loss": 0.394,
      "step": 4875
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.883971309767671,
      "learning_rate": 1.1804165180906326e-05,
      "loss": 0.3672,
      "step": 4876
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.6960532934668267,
      "learning_rate": 1.1801166222811607e-05,
      "loss": 0.2571,
      "step": 4877
    },
    {
      "epoch": 1.38,
      "grad_norm": 2.007131316896604,
      "learning_rate": 1.1798167097283585e-05,
      "loss": 0.3862,
      "step": 4878
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.7069965598607775,
      "learning_rate": 1.1795167804601062e-05,
      "loss": 0.3108,
      "step": 4879
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.5967456167620628,
      "learning_rate": 1.179216834504284e-05,
      "loss": 0.3452,
      "step": 4880
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.6116218322319753,
      "learning_rate": 1.1789168718887745e-05,
      "loss": 0.3201,
      "step": 4881
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.6237918513511425,
      "learning_rate": 1.1786168926414617e-05,
      "loss": 0.3696,
      "step": 4882
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.7100913737442869,
      "learning_rate": 1.1783168967902314e-05,
      "loss": 0.3208,
      "step": 4883
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.746943859455795,
      "learning_rate": 1.1780168843629705e-05,
      "loss": 0.364,
      "step": 4884
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.7229010322328782,
      "learning_rate": 1.1777168553875673e-05,
      "loss": 0.3452,
      "step": 4885
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.6631326058548084,
      "learning_rate": 1.177416809891913e-05,
      "loss": 0.3193,
      "step": 4886
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.522258706331626,
      "learning_rate": 1.1771167479038978e-05,
      "loss": 0.3,
      "step": 4887
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.6273179207686796,
      "learning_rate": 1.1768166694514164e-05,
      "loss": 0.3794,
      "step": 4888
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.7046242610650084,
      "learning_rate": 1.1765165745623628e-05,
      "loss": 0.397,
      "step": 4889
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.6654882666881956,
      "learning_rate": 1.1762164632646334e-05,
      "loss": 0.3364,
      "step": 4890
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.5653271413277483,
      "learning_rate": 1.1759163355861258e-05,
      "loss": 0.3032,
      "step": 4891
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.7523050802957971,
      "learning_rate": 1.1756161915547399e-05,
      "loss": 0.3884,
      "step": 4892
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.6606237500217882,
      "learning_rate": 1.1753160311983764e-05,
      "loss": 0.3442,
      "step": 4893
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.4498749200329193,
      "learning_rate": 1.1750158545449377e-05,
      "loss": 0.2747,
      "step": 4894
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.785671156191868,
      "learning_rate": 1.1747156616223272e-05,
      "loss": 0.4238,
      "step": 4895
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.755539426797289,
      "learning_rate": 1.174415452458451e-05,
      "loss": 0.4131,
      "step": 4896
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.765125892696864,
      "learning_rate": 1.1741152270812155e-05,
      "loss": 0.3433,
      "step": 4897
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.6806311618447527,
      "learning_rate": 1.1738149855185295e-05,
      "loss": 0.2874,
      "step": 4898
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.8127684362557022,
      "learning_rate": 1.1735147277983027e-05,
      "loss": 0.3037,
      "step": 4899
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.6233530190539922,
      "learning_rate": 1.1732144539484467e-05,
      "loss": 0.2817,
      "step": 4900
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.8129346124444923,
      "learning_rate": 1.1729141639968745e-05,
      "loss": 0.4146,
      "step": 4901
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.7886733560398804,
      "learning_rate": 1.1726138579715002e-05,
      "loss": 0.439,
      "step": 4902
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.6764640195096567,
      "learning_rate": 1.1723135359002403e-05,
      "loss": 0.3125,
      "step": 4903
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.815378653343011,
      "learning_rate": 1.1720131978110115e-05,
      "loss": 0.4282,
      "step": 4904
    },
    {
      "epoch": 1.39,
      "grad_norm": 2.011993500845039,
      "learning_rate": 1.171712843731733e-05,
      "loss": 0.3623,
      "step": 4905
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.8060821372639955,
      "learning_rate": 1.1714124736903254e-05,
      "loss": 0.3618,
      "step": 4906
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.7345613773391324,
      "learning_rate": 1.1711120877147107e-05,
      "loss": 0.3325,
      "step": 4907
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.5252816799239102,
      "learning_rate": 1.1708116858328116e-05,
      "loss": 0.3232,
      "step": 4908
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.9122296928925242,
      "learning_rate": 1.1705112680725538e-05,
      "loss": 0.3838,
      "step": 4909
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.840099313542174,
      "learning_rate": 1.1702108344618627e-05,
      "loss": 0.3389,
      "step": 4910
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.9085243366659919,
      "learning_rate": 1.1699103850286668e-05,
      "loss": 0.4009,
      "step": 4911
    },
    {
      "epoch": 1.39,
      "grad_norm": 2.093591935263173,
      "learning_rate": 1.1696099198008953e-05,
      "loss": 0.5132,
      "step": 4912
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.7347464729052846,
      "learning_rate": 1.1693094388064786e-05,
      "loss": 0.3521,
      "step": 4913
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.5534648844750614,
      "learning_rate": 1.1690089420733486e-05,
      "loss": 0.2812,
      "step": 4914
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.7981180385128552,
      "learning_rate": 1.1687084296294398e-05,
      "loss": 0.3838,
      "step": 4915
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.7277191437570432,
      "learning_rate": 1.168407901502687e-05,
      "loss": 0.2969,
      "step": 4916
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.7894940344492527,
      "learning_rate": 1.1681073577210262e-05,
      "loss": 0.4016,
      "step": 4917
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.570690275819502,
      "learning_rate": 1.1678067983123965e-05,
      "loss": 0.333,
      "step": 4918
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.7915096231097263,
      "learning_rate": 1.1675062233047365e-05,
      "loss": 0.3677,
      "step": 4919
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.7255135652993128,
      "learning_rate": 1.1672056327259876e-05,
      "loss": 0.3572,
      "step": 4920
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.7286082004146486,
      "learning_rate": 1.1669050266040917e-05,
      "loss": 0.325,
      "step": 4921
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.9895216582253994,
      "learning_rate": 1.1666044049669934e-05,
      "loss": 0.438,
      "step": 4922
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.7313388600304314,
      "learning_rate": 1.1663037678426371e-05,
      "loss": 0.3872,
      "step": 4923
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.7610739773067838,
      "learning_rate": 1.16600311525897e-05,
      "loss": 0.3491,
      "step": 4924
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.6607463523013208,
      "learning_rate": 1.1657024472439402e-05,
      "loss": 0.3472,
      "step": 4925
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.6384134025134565,
      "learning_rate": 1.1654017638254976e-05,
      "loss": 0.2935,
      "step": 4926
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.938698161705344,
      "learning_rate": 1.1651010650315923e-05,
      "loss": 0.4268,
      "step": 4927
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.7135124155978887,
      "learning_rate": 1.1648003508901775e-05,
      "loss": 0.3613,
      "step": 4928
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.5346800197663402,
      "learning_rate": 1.164499621429207e-05,
      "loss": 0.3433,
      "step": 4929
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.7032217989416962,
      "learning_rate": 1.1641988766766359e-05,
      "loss": 0.3804,
      "step": 4930
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.8549787526860155,
      "learning_rate": 1.1638981166604206e-05,
      "loss": 0.3862,
      "step": 4931
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.6944731098613928,
      "learning_rate": 1.16359734140852e-05,
      "loss": 0.356,
      "step": 4932
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.7731963479316069,
      "learning_rate": 1.1632965509488932e-05,
      "loss": 0.3545,
      "step": 4933
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.6007349962938215,
      "learning_rate": 1.1629957453095013e-05,
      "loss": 0.3352,
      "step": 4934
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.6868965032924448,
      "learning_rate": 1.1626949245183061e-05,
      "loss": 0.3069,
      "step": 4935
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.8451591310447484,
      "learning_rate": 1.1623940886032723e-05,
      "loss": 0.3691,
      "step": 4936
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.7745546633964395,
      "learning_rate": 1.1620932375923644e-05,
      "loss": 0.394,
      "step": 4937
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.6998771310971268,
      "learning_rate": 1.1617923715135493e-05,
      "loss": 0.3159,
      "step": 4938
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.6856859089957776,
      "learning_rate": 1.1614914903947952e-05,
      "loss": 0.3569,
      "step": 4939
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.5232004171435072,
      "learning_rate": 1.1611905942640707e-05,
      "loss": 0.2859,
      "step": 4940
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.7262162311526252,
      "learning_rate": 1.1608896831493475e-05,
      "loss": 0.3516,
      "step": 4941
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.7928204673461863,
      "learning_rate": 1.1605887570785972e-05,
      "loss": 0.3447,
      "step": 4942
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.8976700910956659,
      "learning_rate": 1.1602878160797936e-05,
      "loss": 0.4404,
      "step": 4943
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.7134028225707911,
      "learning_rate": 1.1599868601809114e-05,
      "loss": 0.3391,
      "step": 4944
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.740395733290271,
      "learning_rate": 1.1596858894099272e-05,
      "loss": 0.3071,
      "step": 4945
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.4944365760692435,
      "learning_rate": 1.1593849037948189e-05,
      "loss": 0.2947,
      "step": 4946
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.7435093724779995,
      "learning_rate": 1.1590839033635652e-05,
      "loss": 0.3081,
      "step": 4947
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.7537307812517386,
      "learning_rate": 1.1587828881441468e-05,
      "loss": 0.3147,
      "step": 4948
    },
    {
      "epoch": 1.4,
      "grad_norm": 2.00867622367392,
      "learning_rate": 1.1584818581645453e-05,
      "loss": 0.3887,
      "step": 4949
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.9740282432602547,
      "learning_rate": 1.1581808134527443e-05,
      "loss": 0.3647,
      "step": 4950
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.6946295212892657,
      "learning_rate": 1.1578797540367284e-05,
      "loss": 0.3435,
      "step": 4951
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.9630171120918007,
      "learning_rate": 1.157578679944483e-05,
      "loss": 0.3687,
      "step": 4952
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.7975234619451714,
      "learning_rate": 1.157277591203996e-05,
      "loss": 0.3037,
      "step": 4953
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.7193016075383696,
      "learning_rate": 1.1569764878432559e-05,
      "loss": 0.3381,
      "step": 4954
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.7351122946594855,
      "learning_rate": 1.1566753698902527e-05,
      "loss": 0.3403,
      "step": 4955
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.9771542076696358,
      "learning_rate": 1.156374237372978e-05,
      "loss": 0.3999,
      "step": 4956
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.6771321153573002,
      "learning_rate": 1.1560730903194242e-05,
      "loss": 0.3315,
      "step": 4957
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.6743558669526528,
      "learning_rate": 1.1557719287575858e-05,
      "loss": 0.3684,
      "step": 4958
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.542451645205218,
      "learning_rate": 1.155470752715458e-05,
      "loss": 0.281,
      "step": 4959
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.8093992074855947,
      "learning_rate": 1.1551695622210377e-05,
      "loss": 0.3579,
      "step": 4960
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.6947028801870627,
      "learning_rate": 1.1548683573023229e-05,
      "loss": 0.314,
      "step": 4961
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.789805952796858,
      "learning_rate": 1.1545671379873134e-05,
      "loss": 0.3708,
      "step": 4962
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.4978378346936319,
      "learning_rate": 1.1542659043040097e-05,
      "loss": 0.2986,
      "step": 4963
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.6281760230358373,
      "learning_rate": 1.1539646562804143e-05,
      "loss": 0.3267,
      "step": 4964
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.6010127253818536,
      "learning_rate": 1.1536633939445302e-05,
      "loss": 0.3455,
      "step": 4965
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.6608415958300098,
      "learning_rate": 1.153362117324363e-05,
      "loss": 0.3564,
      "step": 4966
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.5845350412788457,
      "learning_rate": 1.153060826447918e-05,
      "loss": 0.2996,
      "step": 4967
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.5687043868765416,
      "learning_rate": 1.152759521343203e-05,
      "loss": 0.2627,
      "step": 4968
    },
    {
      "epoch": 1.4,
      "grad_norm": 2.051103099652399,
      "learning_rate": 1.1524582020382271e-05,
      "loss": 0.3579,
      "step": 4969
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.7507223910830514,
      "learning_rate": 1.1521568685610003e-05,
      "loss": 0.3467,
      "step": 4970
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.94893834382892,
      "learning_rate": 1.1518555209395334e-05,
      "loss": 0.342,
      "step": 4971
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.974292760938375,
      "learning_rate": 1.1515541592018402e-05,
      "loss": 0.3901,
      "step": 4972
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.7172160326906831,
      "learning_rate": 1.1512527833759339e-05,
      "loss": 0.3171,
      "step": 4973
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.703988403623384,
      "learning_rate": 1.1509513934898303e-05,
      "loss": 0.3452,
      "step": 4974
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.4259493737005924,
      "learning_rate": 1.1506499895715462e-05,
      "loss": 0.2791,
      "step": 4975
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.6278913514093658,
      "learning_rate": 1.1503485716490994e-05,
      "loss": 0.3154,
      "step": 4976
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.7930179981812315,
      "learning_rate": 1.1500471397505091e-05,
      "loss": 0.3901,
      "step": 4977
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.6400490704993025,
      "learning_rate": 1.1497456939037957e-05,
      "loss": 0.314,
      "step": 4978
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.972494161166659,
      "learning_rate": 1.1494442341369819e-05,
      "loss": 0.4443,
      "step": 4979
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.573737103651978,
      "learning_rate": 1.1491427604780898e-05,
      "loss": 0.2983,
      "step": 4980
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.914414519025899,
      "learning_rate": 1.1488412729551449e-05,
      "loss": 0.3564,
      "step": 4981
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.8353900359745021,
      "learning_rate": 1.1485397715961719e-05,
      "loss": 0.4194,
      "step": 4982
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.7148610677976082,
      "learning_rate": 1.148238256429199e-05,
      "loss": 0.3599,
      "step": 4983
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.6403939163896957,
      "learning_rate": 1.1479367274822535e-05,
      "loss": 0.3254,
      "step": 4984
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.6613107989975489,
      "learning_rate": 1.1476351847833656e-05,
      "loss": 0.3286,
      "step": 4985
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.5480948003950685,
      "learning_rate": 1.1473336283605661e-05,
      "loss": 0.2803,
      "step": 4986
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.8388765720442197,
      "learning_rate": 1.1470320582418873e-05,
      "loss": 0.3943,
      "step": 4987
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.670466085261108,
      "learning_rate": 1.1467304744553618e-05,
      "loss": 0.3691,
      "step": 4988
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.850793517600746,
      "learning_rate": 1.1464288770290255e-05,
      "loss": 0.3853,
      "step": 4989
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.82431071496689,
      "learning_rate": 1.1461272659909137e-05,
      "loss": 0.3799,
      "step": 4990
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.7843376380379468,
      "learning_rate": 1.1458256413690634e-05,
      "loss": 0.3916,
      "step": 4991
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.6500280534768044,
      "learning_rate": 1.1455240031915139e-05,
      "loss": 0.3555,
      "step": 4992
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.7044640853390465,
      "learning_rate": 1.1452223514863046e-05,
      "loss": 0.3477,
      "step": 4993
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.6441298221436988,
      "learning_rate": 1.1449206862814762e-05,
      "loss": 0.312,
      "step": 4994
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.7347975087555638,
      "learning_rate": 1.144619007605071e-05,
      "loss": 0.3716,
      "step": 4995
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.7166486787240045,
      "learning_rate": 1.1443173154851335e-05,
      "loss": 0.3022,
      "step": 4996
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.8896281445443592,
      "learning_rate": 1.1440156099497071e-05,
      "loss": 0.3586,
      "step": 4997
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.4984482821546385,
      "learning_rate": 1.1437138910268387e-05,
      "loss": 0.2742,
      "step": 4998
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.607561712225394,
      "learning_rate": 1.1434121587445752e-05,
      "loss": 0.3018,
      "step": 4999
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.8220463574707249,
      "learning_rate": 1.1431104131309654e-05,
      "loss": 0.4287,
      "step": 5000
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.78092810954384,
      "learning_rate": 1.1428086542140587e-05,
      "loss": 0.3857,
      "step": 5001
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.5422572582898528,
      "learning_rate": 1.1425068820219063e-05,
      "loss": 0.303,
      "step": 5002
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.4904133448945047,
      "learning_rate": 1.1422050965825603e-05,
      "loss": 0.2692,
      "step": 5003
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.93536567096739,
      "learning_rate": 1.1419032979240748e-05,
      "loss": 0.4272,
      "step": 5004
    },
    {
      "epoch": 1.41,
      "grad_norm": 2.0630711898729985,
      "learning_rate": 1.1416014860745032e-05,
      "loss": 0.4443,
      "step": 5005
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.6373375937770864,
      "learning_rate": 1.1412996610619028e-05,
      "loss": 0.3335,
      "step": 5006
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.5734061581120657,
      "learning_rate": 1.1409978229143297e-05,
      "loss": 0.2771,
      "step": 5007
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.8298036303056373,
      "learning_rate": 1.1406959716598424e-05,
      "loss": 0.3584,
      "step": 5008
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.6399853916912206,
      "learning_rate": 1.1403941073265014e-05,
      "loss": 0.3413,
      "step": 5009
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.63561739987647,
      "learning_rate": 1.1400922299423663e-05,
      "loss": 0.3408,
      "step": 5010
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.5275906413616323,
      "learning_rate": 1.1397903395354996e-05,
      "loss": 0.2368,
      "step": 5011
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.7505332099329547,
      "learning_rate": 1.1394884361339647e-05,
      "loss": 0.3098,
      "step": 5012
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.6014400504967479,
      "learning_rate": 1.1391865197658256e-05,
      "loss": 0.3291,
      "step": 5013
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.8191541217172245,
      "learning_rate": 1.1388845904591482e-05,
      "loss": 0.3069,
      "step": 5014
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.596626453168252,
      "learning_rate": 1.1385826482419993e-05,
      "loss": 0.3206,
      "step": 5015
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.7442683584544785,
      "learning_rate": 1.1382806931424468e-05,
      "loss": 0.3228,
      "step": 5016
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.3676232004970121,
      "learning_rate": 1.1379787251885603e-05,
      "loss": 0.2727,
      "step": 5017
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.8060103882830303,
      "learning_rate": 1.1376767444084096e-05,
      "loss": 0.416,
      "step": 5018
    },
    {
      "epoch": 1.42,
      "grad_norm": 2.040932826409291,
      "learning_rate": 1.1373747508300668e-05,
      "loss": 0.4551,
      "step": 5019
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.7353399229932849,
      "learning_rate": 1.1370727444816045e-05,
      "loss": 0.332,
      "step": 5020
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.9872133235615659,
      "learning_rate": 1.1367707253910969e-05,
      "loss": 0.4546,
      "step": 5021
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.6579704077597521,
      "learning_rate": 1.1364686935866186e-05,
      "loss": 0.3096,
      "step": 5022
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.8985159876527231,
      "learning_rate": 1.1361666490962468e-05,
      "loss": 0.366,
      "step": 5023
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.6332382691505114,
      "learning_rate": 1.1358645919480585e-05,
      "loss": 0.3125,
      "step": 5024
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.8185532709240697,
      "learning_rate": 1.1355625221701321e-05,
      "loss": 0.3474,
      "step": 5025
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.6310230371376984,
      "learning_rate": 1.1352604397905485e-05,
      "loss": 0.3508,
      "step": 5026
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.667270991711228,
      "learning_rate": 1.1349583448373878e-05,
      "loss": 0.3259,
      "step": 5027
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.9859710654779346,
      "learning_rate": 1.1346562373387326e-05,
      "loss": 0.4297,
      "step": 5028
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.7584702050558267,
      "learning_rate": 1.1343541173226664e-05,
      "loss": 0.3113,
      "step": 5029
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.7051849089512037,
      "learning_rate": 1.1340519848172735e-05,
      "loss": 0.3721,
      "step": 5030
    },
    {
      "epoch": 1.42,
      "grad_norm": 2.0129897340019203,
      "learning_rate": 1.1337498398506397e-05,
      "loss": 0.4062,
      "step": 5031
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.9445226010237395,
      "learning_rate": 1.133447682450852e-05,
      "loss": 0.3638,
      "step": 5032
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.830371012309395,
      "learning_rate": 1.1331455126459983e-05,
      "loss": 0.3618,
      "step": 5033
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.563108149969493,
      "learning_rate": 1.132843330464168e-05,
      "loss": 0.2849,
      "step": 5034
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.7792100280971657,
      "learning_rate": 1.1325411359334509e-05,
      "loss": 0.3721,
      "step": 5035
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.7032894574309183,
      "learning_rate": 1.1322389290819391e-05,
      "loss": 0.3438,
      "step": 5036
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.7377370402036887,
      "learning_rate": 1.1319367099377248e-05,
      "loss": 0.4116,
      "step": 5037
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.825593100047972,
      "learning_rate": 1.1316344785289022e-05,
      "loss": 0.3423,
      "step": 5038
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.6594627411312397,
      "learning_rate": 1.1313322348835658e-05,
      "loss": 0.3511,
      "step": 5039
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.7552857239915152,
      "learning_rate": 1.1310299790298118e-05,
      "loss": 0.3359,
      "step": 5040
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.6934425950442362,
      "learning_rate": 1.130727710995737e-05,
      "loss": 0.3311,
      "step": 5041
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.5992831721210308,
      "learning_rate": 1.1304254308094405e-05,
      "loss": 0.2993,
      "step": 5042
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.73535522467751,
      "learning_rate": 1.1301231384990213e-05,
      "loss": 0.4048,
      "step": 5043
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.5564479492016516,
      "learning_rate": 1.1298208340925798e-05,
      "loss": 0.2959,
      "step": 5044
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.7313277526134474,
      "learning_rate": 1.129518517618218e-05,
      "loss": 0.3687,
      "step": 5045
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.6103233821952625,
      "learning_rate": 1.1292161891040388e-05,
      "loss": 0.3433,
      "step": 5046
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.4820971163799475,
      "learning_rate": 1.1289138485781456e-05,
      "loss": 0.2778,
      "step": 5047
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.9328671832685358,
      "learning_rate": 1.128611496068644e-05,
      "loss": 0.3721,
      "step": 5048
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.7603114382161158,
      "learning_rate": 1.12830913160364e-05,
      "loss": 0.3296,
      "step": 5049
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.596229052514812,
      "learning_rate": 1.1280067552112408e-05,
      "loss": 0.3132,
      "step": 5050
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.868262434788021,
      "learning_rate": 1.1277043669195549e-05,
      "loss": 0.3564,
      "step": 5051
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.9124942352357353,
      "learning_rate": 1.1274019667566913e-05,
      "loss": 0.4258,
      "step": 5052
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.8094151226365243,
      "learning_rate": 1.1270995547507617e-05,
      "loss": 0.3789,
      "step": 5053
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.9466850409527359,
      "learning_rate": 1.1267971309298767e-05,
      "loss": 0.3652,
      "step": 5054
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.8224300608413904,
      "learning_rate": 1.1264946953221496e-05,
      "loss": 0.4185,
      "step": 5055
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.9232642481548956,
      "learning_rate": 1.1261922479556944e-05,
      "loss": 0.4121,
      "step": 5056
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.816582685292471,
      "learning_rate": 1.1258897888586256e-05,
      "loss": 0.3447,
      "step": 5057
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.8311362360933203,
      "learning_rate": 1.1255873180590595e-05,
      "loss": 0.436,
      "step": 5058
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.7892858337724435,
      "learning_rate": 1.1252848355851136e-05,
      "loss": 0.3237,
      "step": 5059
    },
    {
      "epoch": 1.43,
      "grad_norm": 2.0841567088748927,
      "learning_rate": 1.124982341464906e-05,
      "loss": 0.4175,
      "step": 5060
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.710363138365904,
      "learning_rate": 1.1246798357265554e-05,
      "loss": 0.3313,
      "step": 5061
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.8091976994110068,
      "learning_rate": 1.124377318398183e-05,
      "loss": 0.356,
      "step": 5062
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.602938928325915,
      "learning_rate": 1.12407478950791e-05,
      "loss": 0.3271,
      "step": 5063
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.836712040515779,
      "learning_rate": 1.1237722490838592e-05,
      "loss": 0.3723,
      "step": 5064
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.7136110682238372,
      "learning_rate": 1.1234696971541534e-05,
      "loss": 0.3582,
      "step": 5065
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.9338521699059885,
      "learning_rate": 1.1231671337469185e-05,
      "loss": 0.3818,
      "step": 5066
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.8411818106598041,
      "learning_rate": 1.1228645588902793e-05,
      "loss": 0.3896,
      "step": 5067
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.8484232718541294,
      "learning_rate": 1.1225619726123632e-05,
      "loss": 0.4053,
      "step": 5068
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.780673965432986,
      "learning_rate": 1.1222593749412982e-05,
      "loss": 0.3433,
      "step": 5069
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.8162853526760163,
      "learning_rate": 1.1219567659052126e-05,
      "loss": 0.376,
      "step": 5070
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.8451258723725357,
      "learning_rate": 1.1216541455322367e-05,
      "loss": 0.3733,
      "step": 5071
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.852825472754412,
      "learning_rate": 1.121351513850502e-05,
      "loss": 0.3789,
      "step": 5072
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.8305086689974273,
      "learning_rate": 1.12104887088814e-05,
      "loss": 0.3926,
      "step": 5073
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.757594108842674,
      "learning_rate": 1.1207462166732844e-05,
      "loss": 0.3765,
      "step": 5074
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.8925693614804768,
      "learning_rate": 1.1204435512340688e-05,
      "loss": 0.3516,
      "step": 5075
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.6606283379635538,
      "learning_rate": 1.120140874598629e-05,
      "loss": 0.2625,
      "step": 5076
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.4505478488205135,
      "learning_rate": 1.119838186795101e-05,
      "loss": 0.2651,
      "step": 5077
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.717754934867722,
      "learning_rate": 1.1195354878516222e-05,
      "loss": 0.334,
      "step": 5078
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.9205919268501832,
      "learning_rate": 1.1192327777963313e-05,
      "loss": 0.4727,
      "step": 5079
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.4664547417494493,
      "learning_rate": 1.118930056657367e-05,
      "loss": 0.2993,
      "step": 5080
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.5602350366766429,
      "learning_rate": 1.1186273244628705e-05,
      "loss": 0.3376,
      "step": 5081
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.7318561899800489,
      "learning_rate": 1.1183245812409828e-05,
      "loss": 0.3721,
      "step": 5082
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.6554640268985548,
      "learning_rate": 1.1180218270198466e-05,
      "loss": 0.3276,
      "step": 5083
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.5764270076871385,
      "learning_rate": 1.117719061827605e-05,
      "loss": 0.2935,
      "step": 5084
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.8423126968215593,
      "learning_rate": 1.1174162856924029e-05,
      "loss": 0.334,
      "step": 5085
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.9064890087242503,
      "learning_rate": 1.1171134986423859e-05,
      "loss": 0.3892,
      "step": 5086
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.6227672878740733,
      "learning_rate": 1.1168107007057006e-05,
      "loss": 0.3315,
      "step": 5087
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.6639598898644354,
      "learning_rate": 1.1165078919104942e-05,
      "loss": 0.3279,
      "step": 5088
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.5593106221384838,
      "learning_rate": 1.1162050722849153e-05,
      "loss": 0.3413,
      "step": 5089
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.8072473382207999,
      "learning_rate": 1.115902241857114e-05,
      "loss": 0.4106,
      "step": 5090
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.9436649690152572,
      "learning_rate": 1.1155994006552406e-05,
      "loss": 0.3926,
      "step": 5091
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.75309930748677,
      "learning_rate": 1.1152965487074466e-05,
      "loss": 0.3672,
      "step": 5092
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.765053489782449,
      "learning_rate": 1.1149936860418846e-05,
      "loss": 0.3638,
      "step": 5093
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.6321890755467419,
      "learning_rate": 1.1146908126867082e-05,
      "loss": 0.3428,
      "step": 5094
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.647454010525124,
      "learning_rate": 1.1143879286700723e-05,
      "loss": 0.3264,
      "step": 5095
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.8801819050515143,
      "learning_rate": 1.1140850340201319e-05,
      "loss": 0.3623,
      "step": 5096
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.7376784599565458,
      "learning_rate": 1.1137821287650438e-05,
      "loss": 0.3467,
      "step": 5097
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.765202779264421,
      "learning_rate": 1.113479212932966e-05,
      "loss": 0.3599,
      "step": 5098
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.6085002896450329,
      "learning_rate": 1.1131762865520566e-05,
      "loss": 0.3047,
      "step": 5099
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.8641405149932626,
      "learning_rate": 1.1128733496504751e-05,
      "loss": 0.3677,
      "step": 5100
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.6219006262976459,
      "learning_rate": 1.112570402256382e-05,
      "loss": 0.3281,
      "step": 5101
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.6754527089283031,
      "learning_rate": 1.1122674443979387e-05,
      "loss": 0.3535,
      "step": 5102
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.6568958427428246,
      "learning_rate": 1.1119644761033079e-05,
      "loss": 0.3174,
      "step": 5103
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.7270576760447702,
      "learning_rate": 1.111661497400653e-05,
      "loss": 0.2917,
      "step": 5104
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.7728914964683562,
      "learning_rate": 1.111358508318138e-05,
      "loss": 0.3418,
      "step": 5105
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.9680972674049368,
      "learning_rate": 1.1110555088839289e-05,
      "loss": 0.355,
      "step": 5106
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.659742018836075,
      "learning_rate": 1.1107524991261913e-05,
      "loss": 0.3005,
      "step": 5107
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.5729448019618606,
      "learning_rate": 1.1104494790730929e-05,
      "loss": 0.3008,
      "step": 5108
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.9519750103382036,
      "learning_rate": 1.1101464487528017e-05,
      "loss": 0.4048,
      "step": 5109
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.7727604148190967,
      "learning_rate": 1.1098434081934871e-05,
      "loss": 0.3967,
      "step": 5110
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.4865126620475917,
      "learning_rate": 1.1095403574233185e-05,
      "loss": 0.3037,
      "step": 5111
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.8869775344109276,
      "learning_rate": 1.1092372964704681e-05,
      "loss": 0.334,
      "step": 5112
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.6457560422045368,
      "learning_rate": 1.1089342253631074e-05,
      "loss": 0.2566,
      "step": 5113
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.6756837789041177,
      "learning_rate": 1.108631144129409e-05,
      "loss": 0.2991,
      "step": 5114
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.7781464186651623,
      "learning_rate": 1.1083280527975475e-05,
      "loss": 0.3484,
      "step": 5115
    },
    {
      "epoch": 1.44,
      "grad_norm": 2.057720543245449,
      "learning_rate": 1.1080249513956973e-05,
      "loss": 0.4348,
      "step": 5116
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.5515680482675274,
      "learning_rate": 1.1077218399520344e-05,
      "loss": 0.3149,
      "step": 5117
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.891156447721561,
      "learning_rate": 1.1074187184947351e-05,
      "loss": 0.3984,
      "step": 5118
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.7979588532503357,
      "learning_rate": 1.1071155870519777e-05,
      "loss": 0.3623,
      "step": 5119
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.7808391647126085,
      "learning_rate": 1.1068124456519402e-05,
      "loss": 0.3647,
      "step": 5120
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.7806851710942788,
      "learning_rate": 1.1065092943228024e-05,
      "loss": 0.2939,
      "step": 5121
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.6630070213074903,
      "learning_rate": 1.1062061330927445e-05,
      "loss": 0.3604,
      "step": 5122
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.835715306010616,
      "learning_rate": 1.1059029619899483e-05,
      "loss": 0.3701,
      "step": 5123
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.9813429014473802,
      "learning_rate": 1.1055997810425954e-05,
      "loss": 0.3657,
      "step": 5124
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.691363838580002,
      "learning_rate": 1.1052965902788694e-05,
      "loss": 0.3027,
      "step": 5125
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.8860247612580057,
      "learning_rate": 1.1049933897269547e-05,
      "loss": 0.3696,
      "step": 5126
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.7827496859563277,
      "learning_rate": 1.1046901794150358e-05,
      "loss": 0.3196,
      "step": 5127
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.7557666957232299,
      "learning_rate": 1.1043869593712984e-05,
      "loss": 0.3235,
      "step": 5128
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.7208268467156254,
      "learning_rate": 1.1040837296239298e-05,
      "loss": 0.3486,
      "step": 5129
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.748197358464241,
      "learning_rate": 1.1037804902011175e-05,
      "loss": 0.3442,
      "step": 5130
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.600640369207113,
      "learning_rate": 1.1034772411310503e-05,
      "loss": 0.3142,
      "step": 5131
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.8044247607901143,
      "learning_rate": 1.1031739824419175e-05,
      "loss": 0.3384,
      "step": 5132
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.5674939966914008,
      "learning_rate": 1.1028707141619095e-05,
      "loss": 0.2952,
      "step": 5133
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.813490506299369,
      "learning_rate": 1.1025674363192179e-05,
      "loss": 0.3794,
      "step": 5134
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.7353770741616548,
      "learning_rate": 1.1022641489420342e-05,
      "loss": 0.3369,
      "step": 5135
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.850956585516933,
      "learning_rate": 1.1019608520585525e-05,
      "loss": 0.3862,
      "step": 5136
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.7981278116734989,
      "learning_rate": 1.1016575456969658e-05,
      "loss": 0.3252,
      "step": 5137
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.5828483131786037,
      "learning_rate": 1.1013542298854696e-05,
      "loss": 0.3325,
      "step": 5138
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.7066931277295916,
      "learning_rate": 1.101050904652259e-05,
      "loss": 0.3694,
      "step": 5139
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.8387040271822475,
      "learning_rate": 1.1007475700255313e-05,
      "loss": 0.3262,
      "step": 5140
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.7824895380443693,
      "learning_rate": 1.1004442260334832e-05,
      "loss": 0.3765,
      "step": 5141
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.4503747058924326,
      "learning_rate": 1.1001408727043135e-05,
      "loss": 0.2441,
      "step": 5142
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.487988486213241,
      "learning_rate": 1.0998375100662215e-05,
      "loss": 0.2615,
      "step": 5143
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.9861332099243751,
      "learning_rate": 1.099534138147407e-05,
      "loss": 0.3857,
      "step": 5144
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.7229143434502965,
      "learning_rate": 1.0992307569760709e-05,
      "loss": 0.3608,
      "step": 5145
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.7122574287774264,
      "learning_rate": 1.0989273665804154e-05,
      "loss": 0.2773,
      "step": 5146
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.6004560680886697,
      "learning_rate": 1.0986239669886425e-05,
      "loss": 0.3059,
      "step": 5147
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.6647735831385184,
      "learning_rate": 1.0983205582289563e-05,
      "loss": 0.4009,
      "step": 5148
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.6724001172715228,
      "learning_rate": 1.098017140329561e-05,
      "loss": 0.3418,
      "step": 5149
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.6184740061692282,
      "learning_rate": 1.0977137133186613e-05,
      "loss": 0.3203,
      "step": 5150
    },
    {
      "epoch": 1.45,
      "grad_norm": 2.057158956620459,
      "learning_rate": 1.0974102772244638e-05,
      "loss": 0.3398,
      "step": 5151
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.91780790540262,
      "learning_rate": 1.0971068320751753e-05,
      "loss": 0.458,
      "step": 5152
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.4981268464806088,
      "learning_rate": 1.0968033778990038e-05,
      "loss": 0.2561,
      "step": 5153
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.6147202922843096,
      "learning_rate": 1.096499914724157e-05,
      "loss": 0.3218,
      "step": 5154
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.811428978128112,
      "learning_rate": 1.0961964425788452e-05,
      "loss": 0.3579,
      "step": 5155
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.7551213057782513,
      "learning_rate": 1.0958929614912782e-05,
      "loss": 0.3479,
      "step": 5156
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.7380756814830058,
      "learning_rate": 1.0955894714896675e-05,
      "loss": 0.27,
      "step": 5157
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.7535874826311741,
      "learning_rate": 1.0952859726022245e-05,
      "loss": 0.3481,
      "step": 5158
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.716866020334158,
      "learning_rate": 1.094982464857162e-05,
      "loss": 0.27,
      "step": 5159
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.6487559128269218,
      "learning_rate": 1.094678948282694e-05,
      "loss": 0.323,
      "step": 5160
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.7049763264616036,
      "learning_rate": 1.0943754229070344e-05,
      "loss": 0.3081,
      "step": 5161
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.728041370867689,
      "learning_rate": 1.0940718887583985e-05,
      "loss": 0.3394,
      "step": 5162
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.768697137730591,
      "learning_rate": 1.0937683458650029e-05,
      "loss": 0.3408,
      "step": 5163
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.8283327486081113,
      "learning_rate": 1.093464794255063e-05,
      "loss": 0.4106,
      "step": 5164
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.6825348171757168,
      "learning_rate": 1.093161233956798e-05,
      "loss": 0.334,
      "step": 5165
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.82525340957076,
      "learning_rate": 1.0928576649984254e-05,
      "loss": 0.3506,
      "step": 5166
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.8498210527589363,
      "learning_rate": 1.0925540874081649e-05,
      "loss": 0.3293,
      "step": 5167
    },
    {
      "epoch": 1.46,
      "grad_norm": 2.0073019553895723,
      "learning_rate": 1.092250501214236e-05,
      "loss": 0.4336,
      "step": 5168
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.9474752552492163,
      "learning_rate": 1.0919469064448604e-05,
      "loss": 0.4023,
      "step": 5169
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.6125148855131228,
      "learning_rate": 1.0916433031282592e-05,
      "loss": 0.3647,
      "step": 5170
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.8205900834259139,
      "learning_rate": 1.0913396912926546e-05,
      "loss": 0.312,
      "step": 5171
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.6382155757464363,
      "learning_rate": 1.0910360709662701e-05,
      "loss": 0.3179,
      "step": 5172
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.5923164812691317,
      "learning_rate": 1.0907324421773302e-05,
      "loss": 0.3057,
      "step": 5173
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.7295968549968714,
      "learning_rate": 1.090428804954059e-05,
      "loss": 0.3506,
      "step": 5174
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.6525465452515693,
      "learning_rate": 1.0901251593246822e-05,
      "loss": 0.332,
      "step": 5175
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.5076069864949302,
      "learning_rate": 1.0898215053174268e-05,
      "loss": 0.2913,
      "step": 5176
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.6603080373644958,
      "learning_rate": 1.0895178429605189e-05,
      "loss": 0.2598,
      "step": 5177
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.8681395045452889,
      "learning_rate": 1.0892141722821873e-05,
      "loss": 0.3813,
      "step": 5178
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.7391875489672834,
      "learning_rate": 1.0889104933106604e-05,
      "loss": 0.3423,
      "step": 5179
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.7329731312862173,
      "learning_rate": 1.0886068060741676e-05,
      "loss": 0.3711,
      "step": 5180
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.8677439196805612,
      "learning_rate": 1.0883031106009393e-05,
      "loss": 0.3564,
      "step": 5181
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.6262192002706395,
      "learning_rate": 1.0879994069192064e-05,
      "loss": 0.2754,
      "step": 5182
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.7315569402970195,
      "learning_rate": 1.0876956950572006e-05,
      "loss": 0.3062,
      "step": 5183
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.7606664030970671,
      "learning_rate": 1.0873919750431548e-05,
      "loss": 0.3623,
      "step": 5184
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.8357316397597674,
      "learning_rate": 1.0870882469053016e-05,
      "loss": 0.3423,
      "step": 5185
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.49282123647918,
      "learning_rate": 1.0867845106718758e-05,
      "loss": 0.2483,
      "step": 5186
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.7092850571399698,
      "learning_rate": 1.0864807663711118e-05,
      "loss": 0.3872,
      "step": 5187
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.8544442368315057,
      "learning_rate": 1.0861770140312449e-05,
      "loss": 0.3672,
      "step": 5188
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.849880535080518,
      "learning_rate": 1.085873253680512e-05,
      "loss": 0.3711,
      "step": 5189
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.9502169806894623,
      "learning_rate": 1.0855694853471499e-05,
      "loss": 0.4131,
      "step": 5190
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.8132976465163815,
      "learning_rate": 1.0852657090593961e-05,
      "loss": 0.3352,
      "step": 5191
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.851132687705419,
      "learning_rate": 1.0849619248454893e-05,
      "loss": 0.3384,
      "step": 5192
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.96394884640311,
      "learning_rate": 1.0846581327336692e-05,
      "loss": 0.4775,
      "step": 5193
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.9550425413874881,
      "learning_rate": 1.0843543327521748e-05,
      "loss": 0.3716,
      "step": 5194
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.7662897114552227,
      "learning_rate": 1.0840505249292477e-05,
      "loss": 0.3457,
      "step": 5195
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.7167484403392361,
      "learning_rate": 1.083746709293129e-05,
      "loss": 0.2871,
      "step": 5196
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.9334269902620131,
      "learning_rate": 1.0834428858720608e-05,
      "loss": 0.3921,
      "step": 5197
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.607743423955505,
      "learning_rate": 1.083139054694286e-05,
      "loss": 0.3389,
      "step": 5198
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.4068623917725533,
      "learning_rate": 1.0828352157880489e-05,
      "loss": 0.2502,
      "step": 5199
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.8457690301453291,
      "learning_rate": 1.0825313691815928e-05,
      "loss": 0.3647,
      "step": 5200
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.5275440181495041,
      "learning_rate": 1.0822275149031635e-05,
      "loss": 0.2761,
      "step": 5201
    },
    {
      "epoch": 1.47,
      "grad_norm": 2.072498394002997,
      "learning_rate": 1.0819236529810062e-05,
      "loss": 0.4561,
      "step": 5202
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.8729061615862757,
      "learning_rate": 1.081619783443368e-05,
      "loss": 0.3926,
      "step": 5203
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.7085374660899826,
      "learning_rate": 1.0813159063184958e-05,
      "loss": 0.3076,
      "step": 5204
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.770663508498905,
      "learning_rate": 1.0810120216346368e-05,
      "loss": 0.3628,
      "step": 5205
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.8486571592162766,
      "learning_rate": 1.0807081294200413e-05,
      "loss": 0.3647,
      "step": 5206
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.9568300376541188,
      "learning_rate": 1.0804042297029567e-05,
      "loss": 0.3574,
      "step": 5207
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.6708134212376167,
      "learning_rate": 1.0801003225116341e-05,
      "loss": 0.3015,
      "step": 5208
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.8829400202974158,
      "learning_rate": 1.0797964078743241e-05,
      "loss": 0.3518,
      "step": 5209
    },
    {
      "epoch": 1.47,
      "grad_norm": 2.077538228338247,
      "learning_rate": 1.0794924858192779e-05,
      "loss": 0.3521,
      "step": 5210
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.6543444348768088,
      "learning_rate": 1.0791885563747472e-05,
      "loss": 0.335,
      "step": 5211
    },
    {
      "epoch": 1.47,
      "grad_norm": 2.0140808420213085,
      "learning_rate": 1.0788846195689856e-05,
      "loss": 0.3643,
      "step": 5212
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.6766657281249124,
      "learning_rate": 1.078580675430246e-05,
      "loss": 0.3335,
      "step": 5213
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.7495214675843163,
      "learning_rate": 1.0782767239867824e-05,
      "loss": 0.332,
      "step": 5214
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.6268743635925038,
      "learning_rate": 1.0779727652668496e-05,
      "loss": 0.2974,
      "step": 5215
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.6812963177446425,
      "learning_rate": 1.0776687992987038e-05,
      "loss": 0.3284,
      "step": 5216
    },
    {
      "epoch": 1.47,
      "grad_norm": 2.008137978021958,
      "learning_rate": 1.0773648261106005e-05,
      "loss": 0.415,
      "step": 5217
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.6058052298292422,
      "learning_rate": 1.0770608457307965e-05,
      "loss": 0.3818,
      "step": 5218
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.6501794125329923,
      "learning_rate": 1.0767568581875494e-05,
      "loss": 0.3303,
      "step": 5219
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.781322710216552,
      "learning_rate": 1.0764528635091179e-05,
      "loss": 0.3069,
      "step": 5220
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.9759752678254112,
      "learning_rate": 1.0761488617237597e-05,
      "loss": 0.4058,
      "step": 5221
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.860057004236979,
      "learning_rate": 1.075844852859735e-05,
      "loss": 0.3853,
      "step": 5222
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.7735913224413375,
      "learning_rate": 1.075540836945304e-05,
      "loss": 0.3376,
      "step": 5223
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.5505767011619425,
      "learning_rate": 1.0752368140087272e-05,
      "loss": 0.3188,
      "step": 5224
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.5560691598458842,
      "learning_rate": 1.0749327840782663e-05,
      "loss": 0.3091,
      "step": 5225
    },
    {
      "epoch": 1.48,
      "grad_norm": 2.073891434529925,
      "learning_rate": 1.0746287471821833e-05,
      "loss": 0.3284,
      "step": 5226
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.8629479190222973,
      "learning_rate": 1.074324703348741e-05,
      "loss": 0.3267,
      "step": 5227
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.648745909714456,
      "learning_rate": 1.0740206526062022e-05,
      "loss": 0.3296,
      "step": 5228
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.911004069483733,
      "learning_rate": 1.073716594982832e-05,
      "loss": 0.4617,
      "step": 5229
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.39222024915599,
      "learning_rate": 1.0734125305068943e-05,
      "loss": 0.292,
      "step": 5230
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.798048434692731,
      "learning_rate": 1.0731084592066548e-05,
      "loss": 0.353,
      "step": 5231
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.9231396527628695,
      "learning_rate": 1.072804381110379e-05,
      "loss": 0.3799,
      "step": 5232
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.9471742941588197,
      "learning_rate": 1.072500296246334e-05,
      "loss": 0.3535,
      "step": 5233
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.8550795472761628,
      "learning_rate": 1.0721962046427866e-05,
      "loss": 0.3857,
      "step": 5234
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.9781081857535208,
      "learning_rate": 1.0718921063280048e-05,
      "loss": 0.3833,
      "step": 5235
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.4845441009274583,
      "learning_rate": 1.0715880013302568e-05,
      "loss": 0.3176,
      "step": 5236
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.6547481743586367,
      "learning_rate": 1.0712838896778124e-05,
      "loss": 0.3306,
      "step": 5237
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.8679682693796897,
      "learning_rate": 1.0709797713989403e-05,
      "loss": 0.3569,
      "step": 5238
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.7584020197598618,
      "learning_rate": 1.0706756465219114e-05,
      "loss": 0.3516,
      "step": 5239
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.871281555488767,
      "learning_rate": 1.0703715150749967e-05,
      "loss": 0.3711,
      "step": 5240
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.752711253987863,
      "learning_rate": 1.0700673770864673e-05,
      "loss": 0.3467,
      "step": 5241
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.9176137468072516,
      "learning_rate": 1.069763232584596e-05,
      "loss": 0.354,
      "step": 5242
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.9325142968964866,
      "learning_rate": 1.0694590815976549e-05,
      "loss": 0.4053,
      "step": 5243
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.6363806250189776,
      "learning_rate": 1.0691549241539177e-05,
      "loss": 0.3438,
      "step": 5244
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.6550928280771802,
      "learning_rate": 1.0688507602816581e-05,
      "loss": 0.3188,
      "step": 5245
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.4366367669649278,
      "learning_rate": 1.068546590009151e-05,
      "loss": 0.2554,
      "step": 5246
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.870634274632981,
      "learning_rate": 1.0682424133646712e-05,
      "loss": 0.3379,
      "step": 5247
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.8191742464109029,
      "learning_rate": 1.0679382303764945e-05,
      "loss": 0.3311,
      "step": 5248
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.7476444890879566,
      "learning_rate": 1.0676340410728976e-05,
      "loss": 0.3284,
      "step": 5249
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.5358047178370071,
      "learning_rate": 1.0673298454821567e-05,
      "loss": 0.3008,
      "step": 5250
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.898948908046394,
      "learning_rate": 1.0670256436325499e-05,
      "loss": 0.3704,
      "step": 5251
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.9832723948370325,
      "learning_rate": 1.0667214355523552e-05,
      "loss": 0.4062,
      "step": 5252
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.545185847937556,
      "learning_rate": 1.0664172212698512e-05,
      "loss": 0.2568,
      "step": 5253
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.5313189300523593,
      "learning_rate": 1.0661130008133169e-05,
      "loss": 0.2578,
      "step": 5254
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.8957701629867565,
      "learning_rate": 1.0658087742110322e-05,
      "loss": 0.3774,
      "step": 5255
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.865010741362926,
      "learning_rate": 1.0655045414912777e-05,
      "loss": 0.3149,
      "step": 5256
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.9079447564409926,
      "learning_rate": 1.0652003026823344e-05,
      "loss": 0.3535,
      "step": 5257
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.9426386647791172,
      "learning_rate": 1.0648960578124831e-05,
      "loss": 0.3848,
      "step": 5258
    },
    {
      "epoch": 1.49,
      "grad_norm": 2.029404297936551,
      "learning_rate": 1.064591806910007e-05,
      "loss": 0.335,
      "step": 5259
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.8370586493605714,
      "learning_rate": 1.0642875500031878e-05,
      "loss": 0.3442,
      "step": 5260
    },
    {
      "epoch": 1.49,
      "grad_norm": 2.033032195460092,
      "learning_rate": 1.0639832871203094e-05,
      "loss": 0.4312,
      "step": 5261
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.836317738304516,
      "learning_rate": 1.0636790182896545e-05,
      "loss": 0.3184,
      "step": 5262
    },
    {
      "epoch": 1.49,
      "grad_norm": 2.000717500257213,
      "learning_rate": 1.063374743539509e-05,
      "loss": 0.3625,
      "step": 5263
    },
    {
      "epoch": 1.49,
      "grad_norm": 2.010190105506856,
      "learning_rate": 1.0630704628981561e-05,
      "loss": 0.3799,
      "step": 5264
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.635786580914021,
      "learning_rate": 1.0627661763938824e-05,
      "loss": 0.3071,
      "step": 5265
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.6545324722303385,
      "learning_rate": 1.0624618840549732e-05,
      "loss": 0.3123,
      "step": 5266
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.8221482832848577,
      "learning_rate": 1.0621575859097153e-05,
      "loss": 0.3484,
      "step": 5267
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.7834646122132194,
      "learning_rate": 1.0618532819863953e-05,
      "loss": 0.2925,
      "step": 5268
    },
    {
      "epoch": 1.49,
      "grad_norm": 2.0465834729951053,
      "learning_rate": 1.0615489723133015e-05,
      "loss": 0.4102,
      "step": 5269
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.787294866393007,
      "learning_rate": 1.0612446569187214e-05,
      "loss": 0.3979,
      "step": 5270
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.636764289489923,
      "learning_rate": 1.060940335830944e-05,
      "loss": 0.3174,
      "step": 5271
    },
    {
      "epoch": 1.49,
      "grad_norm": 2.1742028868803644,
      "learning_rate": 1.0606360090782578e-05,
      "loss": 0.4468,
      "step": 5272
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.9190339251292465,
      "learning_rate": 1.0603316766889537e-05,
      "loss": 0.3604,
      "step": 5273
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.55375630936681,
      "learning_rate": 1.0600273386913207e-05,
      "loss": 0.2893,
      "step": 5274
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.9784073153325863,
      "learning_rate": 1.0597229951136498e-05,
      "loss": 0.3711,
      "step": 5275
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.5513650176652694,
      "learning_rate": 1.0594186459842333e-05,
      "loss": 0.3042,
      "step": 5276
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.8872164499391386,
      "learning_rate": 1.0591142913313615e-05,
      "loss": 0.356,
      "step": 5277
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.9442860883238875,
      "learning_rate": 1.0588099311833275e-05,
      "loss": 0.3296,
      "step": 5278
    },
    {
      "epoch": 1.49,
      "grad_norm": 2.3136616256428013,
      "learning_rate": 1.058505565568424e-05,
      "loss": 0.5068,
      "step": 5279
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.9821169683511015,
      "learning_rate": 1.058201194514944e-05,
      "loss": 0.3931,
      "step": 5280
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.8424273623953802,
      "learning_rate": 1.0578968180511815e-05,
      "loss": 0.3357,
      "step": 5281
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.7606978803462805,
      "learning_rate": 1.057592436205431e-05,
      "loss": 0.3494,
      "step": 5282
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.5861828878056745,
      "learning_rate": 1.0572880490059874e-05,
      "loss": 0.3171,
      "step": 5283
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.7168009646568636,
      "learning_rate": 1.0569836564811456e-05,
      "loss": 0.3643,
      "step": 5284
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.747631968563414,
      "learning_rate": 1.0566792586592012e-05,
      "loss": 0.4175,
      "step": 5285
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.67505402818386,
      "learning_rate": 1.0563748555684511e-05,
      "loss": 0.3086,
      "step": 5286
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.6792425149869208,
      "learning_rate": 1.0560704472371919e-05,
      "loss": 0.3047,
      "step": 5287
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.800592476276209,
      "learning_rate": 1.0557660336937207e-05,
      "loss": 0.3594,
      "step": 5288
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.812040694821361,
      "learning_rate": 1.0554616149663355e-05,
      "loss": 0.3511,
      "step": 5289
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.7399281751388729,
      "learning_rate": 1.0551571910833344e-05,
      "loss": 0.2937,
      "step": 5290
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.7128675547754515,
      "learning_rate": 1.054852762073016e-05,
      "loss": 0.3601,
      "step": 5291
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.5787433800504487,
      "learning_rate": 1.0545483279636799e-05,
      "loss": 0.3076,
      "step": 5292
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.7885751002587875,
      "learning_rate": 1.0542438887836252e-05,
      "loss": 0.3574,
      "step": 5293
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.7749286589375057,
      "learning_rate": 1.0539394445611524e-05,
      "loss": 0.3433,
      "step": 5294
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.6811917184393865,
      "learning_rate": 1.0536349953245622e-05,
      "loss": 0.3213,
      "step": 5295
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.7758832103476871,
      "learning_rate": 1.0533305411021555e-05,
      "loss": 0.4131,
      "step": 5296
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.6557307835460064,
      "learning_rate": 1.0530260819222337e-05,
      "loss": 0.3381,
      "step": 5297
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.5561020661380418,
      "learning_rate": 1.0527216178130988e-05,
      "loss": 0.322,
      "step": 5298
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.6511187967722112,
      "learning_rate": 1.0524171488030537e-05,
      "loss": 0.3315,
      "step": 5299
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.574669962370916,
      "learning_rate": 1.0521126749204009e-05,
      "loss": 0.2764,
      "step": 5300
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.912748545718705,
      "learning_rate": 1.051808196193444e-05,
      "loss": 0.3784,
      "step": 5301
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.7866182601502194,
      "learning_rate": 1.0515037126504865e-05,
      "loss": 0.3386,
      "step": 5302
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.527848815166339,
      "learning_rate": 1.0511992243198335e-05,
      "loss": 0.3022,
      "step": 5303
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.731917887537752,
      "learning_rate": 1.0508947312297884e-05,
      "loss": 0.2944,
      "step": 5304
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.7530118009249023,
      "learning_rate": 1.050590233408657e-05,
      "loss": 0.3281,
      "step": 5305
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.931629942065713,
      "learning_rate": 1.0502857308847453e-05,
      "loss": 0.3604,
      "step": 5306
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.9108328083276476,
      "learning_rate": 1.0499812236863589e-05,
      "loss": 0.3521,
      "step": 5307
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.823099206180559,
      "learning_rate": 1.049676711841804e-05,
      "loss": 0.3555,
      "step": 5308
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.8727127924139988,
      "learning_rate": 1.0493721953793881e-05,
      "loss": 0.3682,
      "step": 5309
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.665286106394833,
      "learning_rate": 1.0490676743274181e-05,
      "loss": 0.3167,
      "step": 5310
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.9115330229981962,
      "learning_rate": 1.0487631487142018e-05,
      "loss": 0.3584,
      "step": 5311
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.6472831422012548,
      "learning_rate": 1.0484586185680477e-05,
      "loss": 0.3032,
      "step": 5312
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.7100039234350846,
      "learning_rate": 1.0481540839172641e-05,
      "loss": 0.3184,
      "step": 5313
    },
    {
      "epoch": 1.5,
      "grad_norm": 2.0497075446929216,
      "learning_rate": 1.04784954479016e-05,
      "loss": 0.3782,
      "step": 5314
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.7545574501457954,
      "learning_rate": 1.0475450012150447e-05,
      "loss": 0.3599,
      "step": 5315
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.815671717683854,
      "learning_rate": 1.0472404532202289e-05,
      "loss": 0.3149,
      "step": 5316
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.9421032730832681,
      "learning_rate": 1.0469359008340216e-05,
      "loss": 0.3198,
      "step": 5317
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.8632949386522455,
      "learning_rate": 1.0466313440847343e-05,
      "loss": 0.4136,
      "step": 5318
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.740287532980205,
      "learning_rate": 1.0463267830006779e-05,
      "loss": 0.3169,
      "step": 5319
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.6082642651034367,
      "learning_rate": 1.0460222176101635e-05,
      "loss": 0.314,
      "step": 5320
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.9510957284984538,
      "learning_rate": 1.0457176479415034e-05,
      "loss": 0.3965,
      "step": 5321
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.543994312480014,
      "learning_rate": 1.0454130740230098e-05,
      "loss": 0.2964,
      "step": 5322
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.908634867060869,
      "learning_rate": 1.0451084958829953e-05,
      "loss": 0.4263,
      "step": 5323
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.7092032578828964,
      "learning_rate": 1.0448039135497732e-05,
      "loss": 0.3599,
      "step": 5324
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.5818746584821388,
      "learning_rate": 1.0444993270516562e-05,
      "loss": 0.3643,
      "step": 5325
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.8339899490051894,
      "learning_rate": 1.044194736416959e-05,
      "loss": 0.3237,
      "step": 5326
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.9533232312985123,
      "learning_rate": 1.0438901416739955e-05,
      "loss": 0.4087,
      "step": 5327
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.9646954738456321,
      "learning_rate": 1.04358554285108e-05,
      "loss": 0.4253,
      "step": 5328
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.6271466882913765,
      "learning_rate": 1.0432809399765281e-05,
      "loss": 0.3059,
      "step": 5329
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.7428094058187653,
      "learning_rate": 1.0429763330786546e-05,
      "loss": 0.3425,
      "step": 5330
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.7528321508730182,
      "learning_rate": 1.0426717221857756e-05,
      "loss": 0.3611,
      "step": 5331
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.571357965094117,
      "learning_rate": 1.0423671073262067e-05,
      "loss": 0.2986,
      "step": 5332
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.7153614345423627,
      "learning_rate": 1.0420624885282653e-05,
      "loss": 0.3643,
      "step": 5333
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.6334500528843494,
      "learning_rate": 1.0417578658202672e-05,
      "loss": 0.3103,
      "step": 5334
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.5411323363160185,
      "learning_rate": 1.0414532392305301e-05,
      "loss": 0.2812,
      "step": 5335
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.4482461284644967,
      "learning_rate": 1.0411486087873717e-05,
      "loss": 0.2783,
      "step": 5336
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.6306850209665784,
      "learning_rate": 1.0408439745191096e-05,
      "loss": 0.2773,
      "step": 5337
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.6940356812181705,
      "learning_rate": 1.0405393364540618e-05,
      "loss": 0.344,
      "step": 5338
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.820371060156807,
      "learning_rate": 1.040234694620548e-05,
      "loss": 0.3633,
      "step": 5339
    },
    {
      "epoch": 1.51,
      "grad_norm": 2.038686103968118,
      "learning_rate": 1.0399300490468862e-05,
      "loss": 0.439,
      "step": 5340
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.5454521242167207,
      "learning_rate": 1.0396253997613964e-05,
      "loss": 0.3145,
      "step": 5341
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.7458595776033015,
      "learning_rate": 1.0393207467923973e-05,
      "loss": 0.3525,
      "step": 5342
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.7615771859480778,
      "learning_rate": 1.03901609016821e-05,
      "loss": 0.3821,
      "step": 5343
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.7792387989204195,
      "learning_rate": 1.0387114299171541e-05,
      "loss": 0.3159,
      "step": 5344
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.6706329844969638,
      "learning_rate": 1.0384067660675508e-05,
      "loss": 0.3262,
      "step": 5345
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.574141450613407,
      "learning_rate": 1.0381020986477209e-05,
      "loss": 0.3516,
      "step": 5346
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.7805425011459526,
      "learning_rate": 1.0377974276859853e-05,
      "loss": 0.3359,
      "step": 5347
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.8584323764318469,
      "learning_rate": 1.0374927532106667e-05,
      "loss": 0.335,
      "step": 5348
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.5464835129184313,
      "learning_rate": 1.0371880752500862e-05,
      "loss": 0.2964,
      "step": 5349
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.7156935661382742,
      "learning_rate": 1.0368833938325667e-05,
      "loss": 0.3386,
      "step": 5350
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.67122312874975,
      "learning_rate": 1.0365787089864303e-05,
      "loss": 0.3262,
      "step": 5351
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.827789577299359,
      "learning_rate": 1.0362740207400006e-05,
      "loss": 0.4126,
      "step": 5352
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.5194596625749734,
      "learning_rate": 1.0359693291216007e-05,
      "loss": 0.2573,
      "step": 5353
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.6340349241754326,
      "learning_rate": 1.0356646341595539e-05,
      "loss": 0.2681,
      "step": 5354
    },
    {
      "epoch": 1.51,
      "grad_norm": 2.0984899492610647,
      "learning_rate": 1.0353599358821845e-05,
      "loss": 0.4331,
      "step": 5355
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.7411756252931287,
      "learning_rate": 1.0350552343178164e-05,
      "loss": 0.2952,
      "step": 5356
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.6095349375849208,
      "learning_rate": 1.0347505294947744e-05,
      "loss": 0.3159,
      "step": 5357
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.7436204390087775,
      "learning_rate": 1.0344458214413833e-05,
      "loss": 0.3369,
      "step": 5358
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.7580845043581375,
      "learning_rate": 1.034141110185968e-05,
      "loss": 0.3301,
      "step": 5359
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.6459056062092476,
      "learning_rate": 1.0338363957568544e-05,
      "loss": 0.3179,
      "step": 5360
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.6578416352775687,
      "learning_rate": 1.0335316781823675e-05,
      "loss": 0.3286,
      "step": 5361
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.9785151077940515,
      "learning_rate": 1.033226957490834e-05,
      "loss": 0.4185,
      "step": 5362
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.9087973745472457,
      "learning_rate": 1.03292223371058e-05,
      "loss": 0.3706,
      "step": 5363
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.8691771575340195,
      "learning_rate": 1.0326175068699316e-05,
      "loss": 0.3477,
      "step": 5364
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.848973791360804,
      "learning_rate": 1.0323127769972165e-05,
      "loss": 0.4194,
      "step": 5365
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.7767396286801438,
      "learning_rate": 1.0320080441207616e-05,
      "loss": 0.3521,
      "step": 5366
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.73088856811453,
      "learning_rate": 1.031703308268894e-05,
      "loss": 0.3567,
      "step": 5367
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.8231322399325882,
      "learning_rate": 1.0313985694699415e-05,
      "loss": 0.3506,
      "step": 5368
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.7148844935592165,
      "learning_rate": 1.0310938277522326e-05,
      "loss": 0.3435,
      "step": 5369
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.7351262410083077,
      "learning_rate": 1.030789083144095e-05,
      "loss": 0.3057,
      "step": 5370
    },
    {
      "epoch": 1.52,
      "grad_norm": 2.107800372210972,
      "learning_rate": 1.0304843356738576e-05,
      "loss": 0.4404,
      "step": 5371
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.8909737122119639,
      "learning_rate": 1.0301795853698487e-05,
      "loss": 0.3237,
      "step": 5372
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.554889803755046,
      "learning_rate": 1.0298748322603982e-05,
      "loss": 0.3032,
      "step": 5373
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.6945528833488752,
      "learning_rate": 1.0295700763738345e-05,
      "loss": 0.2971,
      "step": 5374
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.6940529150540764,
      "learning_rate": 1.0292653177384878e-05,
      "loss": 0.3687,
      "step": 5375
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.7888614098350222,
      "learning_rate": 1.0289605563826876e-05,
      "loss": 0.3799,
      "step": 5376
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.5775206902393515,
      "learning_rate": 1.0286557923347642e-05,
      "loss": 0.3269,
      "step": 5377
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.6073043161642033,
      "learning_rate": 1.0283510256230478e-05,
      "loss": 0.4224,
      "step": 5378
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.687435808973918,
      "learning_rate": 1.028046256275869e-05,
      "loss": 0.3467,
      "step": 5379
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.8225996918210987,
      "learning_rate": 1.027741484321559e-05,
      "loss": 0.3389,
      "step": 5380
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.6720427106165607,
      "learning_rate": 1.0274367097884483e-05,
      "loss": 0.3511,
      "step": 5381
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.7556699248376442,
      "learning_rate": 1.0271319327048684e-05,
      "loss": 0.281,
      "step": 5382
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.5675554322267946,
      "learning_rate": 1.0268271530991509e-05,
      "loss": 0.3059,
      "step": 5383
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.631177088199582,
      "learning_rate": 1.026522370999628e-05,
      "loss": 0.3125,
      "step": 5384
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.9613923697962343,
      "learning_rate": 1.0262175864346307e-05,
      "loss": 0.458,
      "step": 5385
    },
    {
      "epoch": 1.52,
      "grad_norm": 2.1932255383363795,
      "learning_rate": 1.0259127994324923e-05,
      "loss": 0.4385,
      "step": 5386
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.5904529184854341,
      "learning_rate": 1.0256080100215448e-05,
      "loss": 0.2949,
      "step": 5387
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.6601373816652822,
      "learning_rate": 1.025303218230121e-05,
      "loss": 0.4102,
      "step": 5388
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.6905114441511795,
      "learning_rate": 1.0249984240865534e-05,
      "loss": 0.3638,
      "step": 5389
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.9288438721541767,
      "learning_rate": 1.024693627619176e-05,
      "loss": 0.3262,
      "step": 5390
    },
    {
      "epoch": 1.52,
      "grad_norm": 2.0033383411969443,
      "learning_rate": 1.0243888288563213e-05,
      "loss": 0.3745,
      "step": 5391
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.5673813053133994,
      "learning_rate": 1.0240840278263233e-05,
      "loss": 0.3059,
      "step": 5392
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.586403228317498,
      "learning_rate": 1.0237792245575158e-05,
      "loss": 0.2993,
      "step": 5393
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.5183790439589535,
      "learning_rate": 1.0234744190782326e-05,
      "loss": 0.3179,
      "step": 5394
    },
    {
      "epoch": 1.52,
      "grad_norm": 2.080316204973381,
      "learning_rate": 1.0231696114168077e-05,
      "loss": 0.4639,
      "step": 5395
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.603564703915673,
      "learning_rate": 1.022864801601576e-05,
      "loss": 0.3137,
      "step": 5396
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.7165872512230047,
      "learning_rate": 1.022559989660872e-05,
      "loss": 0.4302,
      "step": 5397
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.8362493217479736,
      "learning_rate": 1.0222551756230304e-05,
      "loss": 0.3652,
      "step": 5398
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.63565710815225,
      "learning_rate": 1.0219503595163857e-05,
      "loss": 0.3267,
      "step": 5399
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.531178433456537,
      "learning_rate": 1.0216455413692738e-05,
      "loss": 0.354,
      "step": 5400
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.723973611737417,
      "learning_rate": 1.0213407212100296e-05,
      "loss": 0.3711,
      "step": 5401
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.9474103413478685,
      "learning_rate": 1.0210358990669889e-05,
      "loss": 0.4253,
      "step": 5402
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.8735307645773431,
      "learning_rate": 1.0207310749684877e-05,
      "loss": 0.353,
      "step": 5403
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.8399783888865986,
      "learning_rate": 1.0204262489428611e-05,
      "loss": 0.3535,
      "step": 5404
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.684019780429249,
      "learning_rate": 1.020121421018446e-05,
      "loss": 0.3164,
      "step": 5405
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.7699541909979286,
      "learning_rate": 1.0198165912235784e-05,
      "loss": 0.3198,
      "step": 5406
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.5674048580835278,
      "learning_rate": 1.0195117595865947e-05,
      "loss": 0.2839,
      "step": 5407
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.8364637345598327,
      "learning_rate": 1.0192069261358313e-05,
      "loss": 0.3496,
      "step": 5408
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.889495691761023,
      "learning_rate": 1.0189020908996255e-05,
      "loss": 0.3584,
      "step": 5409
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.9367465981495295,
      "learning_rate": 1.0185972539063139e-05,
      "loss": 0.416,
      "step": 5410
    },
    {
      "epoch": 1.53,
      "grad_norm": 2.132998498674255,
      "learning_rate": 1.0182924151842337e-05,
      "loss": 0.4243,
      "step": 5411
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.8296197448151357,
      "learning_rate": 1.0179875747617221e-05,
      "loss": 0.3191,
      "step": 5412
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.7547107212031374,
      "learning_rate": 1.0176827326671168e-05,
      "loss": 0.3333,
      "step": 5413
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.792056233659187,
      "learning_rate": 1.017377888928755e-05,
      "loss": 0.3384,
      "step": 5414
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.736329471724082,
      "learning_rate": 1.017073043574975e-05,
      "loss": 0.3271,
      "step": 5415
    },
    {
      "epoch": 1.53,
      "grad_norm": 2.212135985448771,
      "learning_rate": 1.016768196634114e-05,
      "loss": 0.439,
      "step": 5416
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.8204439285066922,
      "learning_rate": 1.0164633481345108e-05,
      "loss": 0.3301,
      "step": 5417
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.849348955705315,
      "learning_rate": 1.0161584981045029e-05,
      "loss": 0.312,
      "step": 5418
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.677484232666149,
      "learning_rate": 1.0158536465724291e-05,
      "loss": 0.3264,
      "step": 5419
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.7826438711564832,
      "learning_rate": 1.0155487935666277e-05,
      "loss": 0.3716,
      "step": 5420
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.5834668740765223,
      "learning_rate": 1.0152439391154372e-05,
      "loss": 0.3037,
      "step": 5421
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.7129503990938355,
      "learning_rate": 1.0149390832471965e-05,
      "loss": 0.3369,
      "step": 5422
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.7338485765591363,
      "learning_rate": 1.0146342259902446e-05,
      "loss": 0.3042,
      "step": 5423
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.60640467966154,
      "learning_rate": 1.0143293673729202e-05,
      "loss": 0.2876,
      "step": 5424
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.732003427531947,
      "learning_rate": 1.0140245074235624e-05,
      "loss": 0.3662,
      "step": 5425
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.6411096736895343,
      "learning_rate": 1.013719646170511e-05,
      "loss": 0.2664,
      "step": 5426
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.8639603337823454,
      "learning_rate": 1.013414783642105e-05,
      "loss": 0.3652,
      "step": 5427
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.6744217330597422,
      "learning_rate": 1.0131099198666835e-05,
      "loss": 0.2581,
      "step": 5428
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.8461635766103766,
      "learning_rate": 1.0128050548725865e-05,
      "loss": 0.3486,
      "step": 5429
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.8770139427952328,
      "learning_rate": 1.0125001886881543e-05,
      "loss": 0.3584,
      "step": 5430
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.865612521217203,
      "learning_rate": 1.0121953213417256e-05,
      "loss": 0.4561,
      "step": 5431
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.9077331138846567,
      "learning_rate": 1.0118904528616411e-05,
      "loss": 0.3716,
      "step": 5432
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.6718945804118304,
      "learning_rate": 1.0115855832762408e-05,
      "loss": 0.2932,
      "step": 5433
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.8331623587110062,
      "learning_rate": 1.0112807126138646e-05,
      "loss": 0.3496,
      "step": 5434
    },
    {
      "epoch": 1.53,
      "grad_norm": 2.0667686168905237,
      "learning_rate": 1.0109758409028527e-05,
      "loss": 0.3223,
      "step": 5435
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.4899358936191063,
      "learning_rate": 1.0106709681715456e-05,
      "loss": 0.2861,
      "step": 5436
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.7298308343951667,
      "learning_rate": 1.0103660944482841e-05,
      "loss": 0.3118,
      "step": 5437
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.929739084127951,
      "learning_rate": 1.0100612197614076e-05,
      "loss": 0.4048,
      "step": 5438
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.758895879354861,
      "learning_rate": 1.0097563441392582e-05,
      "loss": 0.3247,
      "step": 5439
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.6973581813460341,
      "learning_rate": 1.0094514676101759e-05,
      "loss": 0.3833,
      "step": 5440
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.9022559287636027,
      "learning_rate": 1.0091465902025012e-05,
      "loss": 0.3748,
      "step": 5441
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.6796624966591325,
      "learning_rate": 1.0088417119445752e-05,
      "loss": 0.3267,
      "step": 5442
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.5442047550973854,
      "learning_rate": 1.0085368328647395e-05,
      "loss": 0.3064,
      "step": 5443
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.5615937267466797,
      "learning_rate": 1.008231952991334e-05,
      "loss": 0.2588,
      "step": 5444
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.6716873154455885,
      "learning_rate": 1.0079270723527005e-05,
      "loss": 0.3101,
      "step": 5445
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.8094834094965078,
      "learning_rate": 1.0076221909771805e-05,
      "loss": 0.3823,
      "step": 5446
    },
    {
      "epoch": 1.54,
      "grad_norm": 2.0514657460661336,
      "learning_rate": 1.0073173088931143e-05,
      "loss": 0.4722,
      "step": 5447
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.8640507752236428,
      "learning_rate": 1.0070124261288437e-05,
      "loss": 0.2637,
      "step": 5448
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.723619772943777,
      "learning_rate": 1.0067075427127103e-05,
      "loss": 0.3262,
      "step": 5449
    },
    {
      "epoch": 1.54,
      "grad_norm": 2.213831247188296,
      "learning_rate": 1.0064026586730553e-05,
      "loss": 0.4795,
      "step": 5450
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.9030745981139507,
      "learning_rate": 1.00609777403822e-05,
      "loss": 0.3433,
      "step": 5451
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.6987334733494652,
      "learning_rate": 1.0057928888365462e-05,
      "loss": 0.3286,
      "step": 5452
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.8156884401046818,
      "learning_rate": 1.0054880030963756e-05,
      "loss": 0.3623,
      "step": 5453
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.6345083681588328,
      "learning_rate": 1.0051831168460497e-05,
      "loss": 0.271,
      "step": 5454
    },
    {
      "epoch": 1.54,
      "grad_norm": 2.0453469077493693,
      "learning_rate": 1.0048782301139102e-05,
      "loss": 0.4375,
      "step": 5455
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.7472688563300562,
      "learning_rate": 1.0045733429282989e-05,
      "loss": 0.3274,
      "step": 5456
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.7040515828901153,
      "learning_rate": 1.0042684553175575e-05,
      "loss": 0.3242,
      "step": 5457
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.7907616617950368,
      "learning_rate": 1.0039635673100282e-05,
      "loss": 0.376,
      "step": 5458
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.5853066298057517,
      "learning_rate": 1.0036586789340518e-05,
      "loss": 0.2825,
      "step": 5459
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.5927960413153701,
      "learning_rate": 1.0033537902179716e-05,
      "loss": 0.3247,
      "step": 5460
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.8697328993386921,
      "learning_rate": 1.0030489011901285e-05,
      "loss": 0.3311,
      "step": 5461
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.6391913216548932,
      "learning_rate": 1.0027440118788649e-05,
      "loss": 0.3152,
      "step": 5462
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.6564306819312764,
      "learning_rate": 1.0024391223125226e-05,
      "loss": 0.3359,
      "step": 5463
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.9018381132815287,
      "learning_rate": 1.0021342325194441e-05,
      "loss": 0.3828,
      "step": 5464
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.5370630597463706,
      "learning_rate": 1.0018293425279706e-05,
      "loss": 0.2732,
      "step": 5465
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.704560104945675,
      "learning_rate": 1.0015244523664447e-05,
      "loss": 0.3457,
      "step": 5466
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.6299300110642423,
      "learning_rate": 1.0012195620632084e-05,
      "loss": 0.344,
      "step": 5467
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.698396323144471,
      "learning_rate": 1.0009146716466038e-05,
      "loss": 0.3291,
      "step": 5468
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.652549458644463,
      "learning_rate": 1.0006097811449726e-05,
      "loss": 0.3494,
      "step": 5469
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.6865059269928588,
      "learning_rate": 1.0003048905866577e-05,
      "loss": 0.3364,
      "step": 5470
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.8257210002487465,
      "learning_rate": 1e-05,
      "loss": 0.3545,
      "step": 5471
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.6226214726748407,
      "learning_rate": 9.996951094133426e-06,
      "loss": 0.3167,
      "step": 5472
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.7437114074940563,
      "learning_rate": 9.993902188550276e-06,
      "loss": 0.3411,
      "step": 5473
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.7411948741232643,
      "learning_rate": 9.990853283533968e-06,
      "loss": 0.335,
      "step": 5474
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.8443933244516413,
      "learning_rate": 9.987804379367919e-06,
      "loss": 0.3892,
      "step": 5475
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.6071674996175735,
      "learning_rate": 9.984755476335556e-06,
      "loss": 0.2754,
      "step": 5476
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.7608615151505356,
      "learning_rate": 9.981706574720296e-06,
      "loss": 0.3516,
      "step": 5477
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.8406212620814022,
      "learning_rate": 9.978657674805564e-06,
      "loss": 0.3867,
      "step": 5478
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.8332406822151783,
      "learning_rate": 9.975608776874775e-06,
      "loss": 0.4048,
      "step": 5479
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.9815726274276582,
      "learning_rate": 9.972559881211353e-06,
      "loss": 0.3491,
      "step": 5480
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.8730914394815246,
      "learning_rate": 9.969510988098716e-06,
      "loss": 0.3345,
      "step": 5481
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.7421633701753445,
      "learning_rate": 9.966462097820289e-06,
      "loss": 0.2803,
      "step": 5482
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.9383974027150233,
      "learning_rate": 9.963413210659485e-06,
      "loss": 0.3604,
      "step": 5483
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.812404458680124,
      "learning_rate": 9.960364326899723e-06,
      "loss": 0.3572,
      "step": 5484
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.9336197078806279,
      "learning_rate": 9.957315446824425e-06,
      "loss": 0.3953,
      "step": 5485
    },
    {
      "epoch": 1.55,
      "grad_norm": 2.0200315468745864,
      "learning_rate": 9.954266570717014e-06,
      "loss": 0.3643,
      "step": 5486
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.7429445150658343,
      "learning_rate": 9.951217698860902e-06,
      "loss": 0.3008,
      "step": 5487
    },
    {
      "epoch": 1.55,
      "grad_norm": 2.2189910716420056,
      "learning_rate": 9.948168831539505e-06,
      "loss": 0.3818,
      "step": 5488
    },
    {
      "epoch": 1.55,
      "grad_norm": 2.026916023142368,
      "learning_rate": 9.945119969036245e-06,
      "loss": 0.3774,
      "step": 5489
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.9861426441106123,
      "learning_rate": 9.942071111634538e-06,
      "loss": 0.4119,
      "step": 5490
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.7498049534098248,
      "learning_rate": 9.939022259617805e-06,
      "loss": 0.3206,
      "step": 5491
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.9016369743146577,
      "learning_rate": 9.93597341326945e-06,
      "loss": 0.3914,
      "step": 5492
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.914394344411035,
      "learning_rate": 9.9329245728729e-06,
      "loss": 0.3164,
      "step": 5493
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.744863711189231,
      "learning_rate": 9.929875738711565e-06,
      "loss": 0.3223,
      "step": 5494
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.7534349370339362,
      "learning_rate": 9.926826911068862e-06,
      "loss": 0.3271,
      "step": 5495
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.5685939773434558,
      "learning_rate": 9.9237780902282e-06,
      "loss": 0.3242,
      "step": 5496
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.597202658149189,
      "learning_rate": 9.920729276472996e-06,
      "loss": 0.2671,
      "step": 5497
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.8549036923358753,
      "learning_rate": 9.91768047008666e-06,
      "loss": 0.4165,
      "step": 5498
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.7040169478866762,
      "learning_rate": 9.91463167135261e-06,
      "loss": 0.2979,
      "step": 5499
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.6474242073498624,
      "learning_rate": 9.91158288055425e-06,
      "loss": 0.3037,
      "step": 5500
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.8366512235642667,
      "learning_rate": 9.90853409797499e-06,
      "loss": 0.3108,
      "step": 5501
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.4981276220734496,
      "learning_rate": 9.905485323898243e-06,
      "loss": 0.2935,
      "step": 5502
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.657327920695838,
      "learning_rate": 9.90243655860742e-06,
      "loss": 0.3333,
      "step": 5503
    },
    {
      "epoch": 1.55,
      "grad_norm": 2.0224193711973104,
      "learning_rate": 9.899387802385925e-06,
      "loss": 0.4004,
      "step": 5504
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.4786684724905987,
      "learning_rate": 9.896339055517164e-06,
      "loss": 0.2417,
      "step": 5505
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.7988218152255604,
      "learning_rate": 9.893290318284546e-06,
      "loss": 0.3604,
      "step": 5506
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.8185709523752471,
      "learning_rate": 9.890241590971475e-06,
      "loss": 0.3589,
      "step": 5507
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.794447250321781,
      "learning_rate": 9.88719287386136e-06,
      "loss": 0.3457,
      "step": 5508
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.9134804586815373,
      "learning_rate": 9.884144167237595e-06,
      "loss": 0.3032,
      "step": 5509
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.8663780715934024,
      "learning_rate": 9.88109547138359e-06,
      "loss": 0.365,
      "step": 5510
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.883915717616009,
      "learning_rate": 9.878046786582745e-06,
      "loss": 0.3555,
      "step": 5511
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.8963309029983277,
      "learning_rate": 9.874998113118462e-06,
      "loss": 0.3525,
      "step": 5512
    },
    {
      "epoch": 1.56,
      "grad_norm": 2.111777445119944,
      "learning_rate": 9.871949451274137e-06,
      "loss": 0.4502,
      "step": 5513
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.8062839549317045,
      "learning_rate": 9.868900801333168e-06,
      "loss": 0.2876,
      "step": 5514
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.7738954904062105,
      "learning_rate": 9.865852163578952e-06,
      "loss": 0.3303,
      "step": 5515
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.9648776237704777,
      "learning_rate": 9.862803538294894e-06,
      "loss": 0.4785,
      "step": 5516
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.7088642135730125,
      "learning_rate": 9.85975492576438e-06,
      "loss": 0.3206,
      "step": 5517
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.6696348413645785,
      "learning_rate": 9.856706326270801e-06,
      "loss": 0.3071,
      "step": 5518
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.6397190224900613,
      "learning_rate": 9.853657740097558e-06,
      "loss": 0.3223,
      "step": 5519
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.7268531298154202,
      "learning_rate": 9.850609167528038e-06,
      "loss": 0.3701,
      "step": 5520
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.7647932680367786,
      "learning_rate": 9.847560608845632e-06,
      "loss": 0.3818,
      "step": 5521
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.7465409506564922,
      "learning_rate": 9.844512064333726e-06,
      "loss": 0.3042,
      "step": 5522
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.6753059769847174,
      "learning_rate": 9.841463534275712e-06,
      "loss": 0.312,
      "step": 5523
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.7065209128622276,
      "learning_rate": 9.838415018954976e-06,
      "loss": 0.3872,
      "step": 5524
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.6371849946620718,
      "learning_rate": 9.835366518654897e-06,
      "loss": 0.3059,
      "step": 5525
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.5996217454396404,
      "learning_rate": 9.832318033658862e-06,
      "loss": 0.3486,
      "step": 5526
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.6342870838020302,
      "learning_rate": 9.829269564250254e-06,
      "loss": 0.3164,
      "step": 5527
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.9703075961145908,
      "learning_rate": 9.826221110712451e-06,
      "loss": 0.3687,
      "step": 5528
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.7621632498250572,
      "learning_rate": 9.823172673328837e-06,
      "loss": 0.3428,
      "step": 5529
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.795089820343496,
      "learning_rate": 9.820124252382784e-06,
      "loss": 0.332,
      "step": 5530
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.7607586300875169,
      "learning_rate": 9.817075848157666e-06,
      "loss": 0.2874,
      "step": 5531
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.8306464469783574,
      "learning_rate": 9.814027460936863e-06,
      "loss": 0.3462,
      "step": 5532
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.7940852305216035,
      "learning_rate": 9.81097909100375e-06,
      "loss": 0.3281,
      "step": 5533
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.4562807625779348,
      "learning_rate": 9.807930738641692e-06,
      "loss": 0.2327,
      "step": 5534
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.7778839272699067,
      "learning_rate": 9.804882404134057e-06,
      "loss": 0.3413,
      "step": 5535
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.6446087046213496,
      "learning_rate": 9.801834087764219e-06,
      "loss": 0.3691,
      "step": 5536
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.7166537943722426,
      "learning_rate": 9.798785789815545e-06,
      "loss": 0.3174,
      "step": 5537
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.797758439037462,
      "learning_rate": 9.79573751057139e-06,
      "loss": 0.3557,
      "step": 5538
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.852849278025158,
      "learning_rate": 9.792689250315126e-06,
      "loss": 0.3303,
      "step": 5539
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.9273838749794607,
      "learning_rate": 9.789641009330113e-06,
      "loss": 0.3799,
      "step": 5540
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.9265745114333586,
      "learning_rate": 9.786592787899707e-06,
      "loss": 0.3696,
      "step": 5541
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.772389130566527,
      "learning_rate": 9.783544586307266e-06,
      "loss": 0.3389,
      "step": 5542
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.8938135116241592,
      "learning_rate": 9.780496404836146e-06,
      "loss": 0.3276,
      "step": 5543
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.6531651384999733,
      "learning_rate": 9.7774482437697e-06,
      "loss": 0.2788,
      "step": 5544
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.7766564682821393,
      "learning_rate": 9.77440010339128e-06,
      "loss": 0.3633,
      "step": 5545
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.8073942429305463,
      "learning_rate": 9.771351983984241e-06,
      "loss": 0.3618,
      "step": 5546
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.671752713780309,
      "learning_rate": 9.768303885831924e-06,
      "loss": 0.2966,
      "step": 5547
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.7371254396438358,
      "learning_rate": 9.765255809217676e-06,
      "loss": 0.3154,
      "step": 5548
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.8809590838734958,
      "learning_rate": 9.762207754424845e-06,
      "loss": 0.3413,
      "step": 5549
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.7472907162339484,
      "learning_rate": 9.759159721736772e-06,
      "loss": 0.3096,
      "step": 5550
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.741538206738876,
      "learning_rate": 9.75611171143679e-06,
      "loss": 0.3423,
      "step": 5551
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.5734419741175545,
      "learning_rate": 9.753063723808243e-06,
      "loss": 0.3069,
      "step": 5552
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.9800547005223237,
      "learning_rate": 9.750015759134466e-06,
      "loss": 0.3682,
      "step": 5553
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.660966604181892,
      "learning_rate": 9.746967817698795e-06,
      "loss": 0.3032,
      "step": 5554
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.6783767391364033,
      "learning_rate": 9.743919899784555e-06,
      "loss": 0.2986,
      "step": 5555
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.9111382720290138,
      "learning_rate": 9.740872005675079e-06,
      "loss": 0.3726,
      "step": 5556
    },
    {
      "epoch": 1.57,
      "grad_norm": 2.2540101017107923,
      "learning_rate": 9.737824135653693e-06,
      "loss": 0.3691,
      "step": 5557
    },
    {
      "epoch": 1.57,
      "grad_norm": 2.0267646250692257,
      "learning_rate": 9.734776290003727e-06,
      "loss": 0.4165,
      "step": 5558
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.9586366110087494,
      "learning_rate": 9.731728469008493e-06,
      "loss": 0.3408,
      "step": 5559
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.6977211447870362,
      "learning_rate": 9.72868067295132e-06,
      "loss": 0.3716,
      "step": 5560
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.8152795525189223,
      "learning_rate": 9.72563290211552e-06,
      "loss": 0.3594,
      "step": 5561
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.7923164874325888,
      "learning_rate": 9.72258515678441e-06,
      "loss": 0.3203,
      "step": 5562
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.9185390879306685,
      "learning_rate": 9.719537437241311e-06,
      "loss": 0.4487,
      "step": 5563
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.8194566494684679,
      "learning_rate": 9.716489743769525e-06,
      "loss": 0.3281,
      "step": 5564
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.9527184588501352,
      "learning_rate": 9.713442076652359e-06,
      "loss": 0.385,
      "step": 5565
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.6878627685576595,
      "learning_rate": 9.710394436173126e-06,
      "loss": 0.3257,
      "step": 5566
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.825313199461077,
      "learning_rate": 9.707346822615127e-06,
      "loss": 0.3652,
      "step": 5567
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.846324667027293,
      "learning_rate": 9.704299236261658e-06,
      "loss": 0.3677,
      "step": 5568
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.9453829750066527,
      "learning_rate": 9.701251677396021e-06,
      "loss": 0.3354,
      "step": 5569
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.7914765323128865,
      "learning_rate": 9.698204146301513e-06,
      "loss": 0.2971,
      "step": 5570
    },
    {
      "epoch": 1.57,
      "grad_norm": 2.1367695785179555,
      "learning_rate": 9.69515664326143e-06,
      "loss": 0.4033,
      "step": 5571
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.7102798387185196,
      "learning_rate": 9.692109168559051e-06,
      "loss": 0.3594,
      "step": 5572
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.7233167631805435,
      "learning_rate": 9.689061722477677e-06,
      "loss": 0.3398,
      "step": 5573
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.614272957394539,
      "learning_rate": 9.686014305300585e-06,
      "loss": 0.3076,
      "step": 5574
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.8133980619497645,
      "learning_rate": 9.682966917311065e-06,
      "loss": 0.3975,
      "step": 5575
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.7521698798033845,
      "learning_rate": 9.679919558792388e-06,
      "loss": 0.3303,
      "step": 5576
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.9393245123174454,
      "learning_rate": 9.676872230027837e-06,
      "loss": 0.3628,
      "step": 5577
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.7273115852527767,
      "learning_rate": 9.673824931300684e-06,
      "loss": 0.2966,
      "step": 5578
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.6086371767821281,
      "learning_rate": 9.670777662894205e-06,
      "loss": 0.3022,
      "step": 5579
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.762940909188812,
      "learning_rate": 9.667730425091666e-06,
      "loss": 0.3335,
      "step": 5580
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.7665835145559503,
      "learning_rate": 9.664683218176328e-06,
      "loss": 0.3862,
      "step": 5581
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.423756775223865,
      "learning_rate": 9.661636042431459e-06,
      "loss": 0.2769,
      "step": 5582
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.591321852585858,
      "learning_rate": 9.658588898140322e-06,
      "loss": 0.3362,
      "step": 5583
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.7532106588647243,
      "learning_rate": 9.655541785586172e-06,
      "loss": 0.2783,
      "step": 5584
    },
    {
      "epoch": 1.58,
      "grad_norm": 2.0485338110268576,
      "learning_rate": 9.65249470505226e-06,
      "loss": 0.4585,
      "step": 5585
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.638102803390952,
      "learning_rate": 9.64944765682184e-06,
      "loss": 0.3027,
      "step": 5586
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.8356676234446707,
      "learning_rate": 9.646400641178157e-06,
      "loss": 0.3125,
      "step": 5587
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.6352706725434256,
      "learning_rate": 9.643353658404466e-06,
      "loss": 0.2834,
      "step": 5588
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.7071636104552401,
      "learning_rate": 9.640306708783997e-06,
      "loss": 0.4648,
      "step": 5589
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.7175909785072172,
      "learning_rate": 9.637259792599997e-06,
      "loss": 0.3506,
      "step": 5590
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.7043138709472703,
      "learning_rate": 9.634212910135697e-06,
      "loss": 0.3521,
      "step": 5591
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.6419555997515596,
      "learning_rate": 9.631166061674338e-06,
      "loss": 0.2854,
      "step": 5592
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.6644203499793144,
      "learning_rate": 9.62811924749914e-06,
      "loss": 0.3206,
      "step": 5593
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.6384599790375436,
      "learning_rate": 9.625072467893337e-06,
      "loss": 0.3911,
      "step": 5594
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.866546140610239,
      "learning_rate": 9.622025723140147e-06,
      "loss": 0.2939,
      "step": 5595
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.9162542314714064,
      "learning_rate": 9.618979013522796e-06,
      "loss": 0.4355,
      "step": 5596
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.7473660058474585,
      "learning_rate": 9.615932339324497e-06,
      "loss": 0.3271,
      "step": 5597
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.7073899118985612,
      "learning_rate": 9.612885700828462e-06,
      "loss": 0.3281,
      "step": 5598
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.4887295052500287,
      "learning_rate": 9.609839098317902e-06,
      "loss": 0.2686,
      "step": 5599
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.749227747067994,
      "learning_rate": 9.606792532076028e-06,
      "loss": 0.3838,
      "step": 5600
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.7868623971384443,
      "learning_rate": 9.603746002386043e-06,
      "loss": 0.3396,
      "step": 5601
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.7374839478471256,
      "learning_rate": 9.60069950953114e-06,
      "loss": 0.345,
      "step": 5602
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.395766305183318,
      "learning_rate": 9.597653053794521e-06,
      "loss": 0.2607,
      "step": 5603
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.774224132067328,
      "learning_rate": 9.59460663545938e-06,
      "loss": 0.3208,
      "step": 5604
    },
    {
      "epoch": 1.58,
      "grad_norm": 2.0366895133762726,
      "learning_rate": 9.591560254808909e-06,
      "loss": 0.458,
      "step": 5605
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.8625925442043956,
      "learning_rate": 9.588513912126286e-06,
      "loss": 0.3745,
      "step": 5606
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.9211976533255006,
      "learning_rate": 9.585467607694702e-06,
      "loss": 0.3711,
      "step": 5607
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.6276194705951,
      "learning_rate": 9.58242134179733e-06,
      "loss": 0.3215,
      "step": 5608
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.9381130568426677,
      "learning_rate": 9.579375114717352e-06,
      "loss": 0.3574,
      "step": 5609
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.8883846806866667,
      "learning_rate": 9.576328926737936e-06,
      "loss": 0.3093,
      "step": 5610
    },
    {
      "epoch": 1.58,
      "grad_norm": 2.12828267652341,
      "learning_rate": 9.573282778142246e-06,
      "loss": 0.4375,
      "step": 5611
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.4019126659774266,
      "learning_rate": 9.570236669213454e-06,
      "loss": 0.2598,
      "step": 5612
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.7642782265728871,
      "learning_rate": 9.567190600234722e-06,
      "loss": 0.3008,
      "step": 5613
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.5530441050573638,
      "learning_rate": 9.564144571489202e-06,
      "loss": 0.3267,
      "step": 5614
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.8231639445937424,
      "learning_rate": 9.561098583260047e-06,
      "loss": 0.345,
      "step": 5615
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.764580303192368,
      "learning_rate": 9.558052635830413e-06,
      "loss": 0.3164,
      "step": 5616
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.7792633181001174,
      "learning_rate": 9.555006729483438e-06,
      "loss": 0.376,
      "step": 5617
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.7725588087682291,
      "learning_rate": 9.551960864502275e-06,
      "loss": 0.3745,
      "step": 5618
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.737903603378558,
      "learning_rate": 9.548915041170049e-06,
      "loss": 0.2952,
      "step": 5619
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.644148748179616,
      "learning_rate": 9.545869259769904e-06,
      "loss": 0.2632,
      "step": 5620
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.8717746010706382,
      "learning_rate": 9.542823520584968e-06,
      "loss": 0.4092,
      "step": 5621
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.9879410247467364,
      "learning_rate": 9.539777823898368e-06,
      "loss": 0.3794,
      "step": 5622
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.789822082950624,
      "learning_rate": 9.536732169993225e-06,
      "loss": 0.3076,
      "step": 5623
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.6256767914686605,
      "learning_rate": 9.53368655915266e-06,
      "loss": 0.3086,
      "step": 5624
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.7566371738851516,
      "learning_rate": 9.530640991659785e-06,
      "loss": 0.3579,
      "step": 5625
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.8289409184630177,
      "learning_rate": 9.527595467797716e-06,
      "loss": 0.353,
      "step": 5626
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.749283074056311,
      "learning_rate": 9.524549987849555e-06,
      "loss": 0.333,
      "step": 5627
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.6958562512541833,
      "learning_rate": 9.521504552098402e-06,
      "loss": 0.3711,
      "step": 5628
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.736516688084063,
      "learning_rate": 9.518459160827359e-06,
      "loss": 0.3652,
      "step": 5629
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.8837876196026822,
      "learning_rate": 9.515413814319524e-06,
      "loss": 0.3564,
      "step": 5630
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.678401643221884,
      "learning_rate": 9.512368512857983e-06,
      "loss": 0.3083,
      "step": 5631
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.562690281522799,
      "learning_rate": 9.50932325672582e-06,
      "loss": 0.3296,
      "step": 5632
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.712583907542674,
      "learning_rate": 9.506278046206122e-06,
      "loss": 0.3193,
      "step": 5633
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.7902367868352778,
      "learning_rate": 9.50323288158196e-06,
      "loss": 0.3809,
      "step": 5634
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.8218848706821409,
      "learning_rate": 9.500187763136416e-06,
      "loss": 0.3774,
      "step": 5635
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.673315472101138,
      "learning_rate": 9.49714269115255e-06,
      "loss": 0.3418,
      "step": 5636
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.5147691657767812,
      "learning_rate": 9.494097665913432e-06,
      "loss": 0.2446,
      "step": 5637
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.7103831565437904,
      "learning_rate": 9.491052687702118e-06,
      "loss": 0.3323,
      "step": 5638
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.9874410734734693,
      "learning_rate": 9.488007756801672e-06,
      "loss": 0.4146,
      "step": 5639
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.8989418736462347,
      "learning_rate": 9.484962873495137e-06,
      "loss": 0.3965,
      "step": 5640
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.6314002414490383,
      "learning_rate": 9.481918038065561e-06,
      "loss": 0.3101,
      "step": 5641
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.781383155039068,
      "learning_rate": 9.478873250795991e-06,
      "loss": 0.3569,
      "step": 5642
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.836069577414162,
      "learning_rate": 9.475828511969466e-06,
      "loss": 0.3306,
      "step": 5643
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.684205320529051,
      "learning_rate": 9.472783821869015e-06,
      "loss": 0.2705,
      "step": 5644
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.752038082325636,
      "learning_rate": 9.469739180777666e-06,
      "loss": 0.3276,
      "step": 5645
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.7835915553215318,
      "learning_rate": 9.466694588978448e-06,
      "loss": 0.3467,
      "step": 5646
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.8190256045145057,
      "learning_rate": 9.463650046754383e-06,
      "loss": 0.3755,
      "step": 5647
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.844271178893795,
      "learning_rate": 9.460605554388479e-06,
      "loss": 0.3721,
      "step": 5648
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.7279846653667383,
      "learning_rate": 9.45756111216375e-06,
      "loss": 0.3357,
      "step": 5649
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.7140738797587134,
      "learning_rate": 9.454516720363203e-06,
      "loss": 0.2998,
      "step": 5650
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.496871761539267,
      "learning_rate": 9.45147237926984e-06,
      "loss": 0.302,
      "step": 5651
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.8541267467024072,
      "learning_rate": 9.44842808916666e-06,
      "loss": 0.3867,
      "step": 5652
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.6317241043481485,
      "learning_rate": 9.445383850336648e-06,
      "loss": 0.2988,
      "step": 5653
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.8772818622696754,
      "learning_rate": 9.442339663062795e-06,
      "loss": 0.4102,
      "step": 5654
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.7999262972392405,
      "learning_rate": 9.439295527628083e-06,
      "loss": 0.3799,
      "step": 5655
    },
    {
      "epoch": 1.6,
      "grad_norm": 2.039744456119282,
      "learning_rate": 9.436251444315492e-06,
      "loss": 0.3784,
      "step": 5656
    },
    {
      "epoch": 1.6,
      "grad_norm": 2.1084062910933223,
      "learning_rate": 9.433207413407991e-06,
      "loss": 0.4614,
      "step": 5657
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.7099795269252605,
      "learning_rate": 9.430163435188549e-06,
      "loss": 0.3286,
      "step": 5658
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.5189406344226446,
      "learning_rate": 9.42711950994013e-06,
      "loss": 0.3018,
      "step": 5659
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.9032864345617377,
      "learning_rate": 9.424075637945692e-06,
      "loss": 0.335,
      "step": 5660
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.6806110721313245,
      "learning_rate": 9.421031819488188e-06,
      "loss": 0.3047,
      "step": 5661
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.7287117743615181,
      "learning_rate": 9.417988054850561e-06,
      "loss": 0.324,
      "step": 5662
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.8408906406079888,
      "learning_rate": 9.414944344315765e-06,
      "loss": 0.3501,
      "step": 5663
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.8409460513530136,
      "learning_rate": 9.411900688166731e-06,
      "loss": 0.3254,
      "step": 5664
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.5325180448909195,
      "learning_rate": 9.40885708668639e-06,
      "loss": 0.3101,
      "step": 5665
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.6153434678941787,
      "learning_rate": 9.40581354015767e-06,
      "loss": 0.3242,
      "step": 5666
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.4508992848695184,
      "learning_rate": 9.402770048863502e-06,
      "loss": 0.2759,
      "step": 5667
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.6943103548134186,
      "learning_rate": 9.399726613086794e-06,
      "loss": 0.3271,
      "step": 5668
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.8106237196846753,
      "learning_rate": 9.396683233110468e-06,
      "loss": 0.3638,
      "step": 5669
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.7024762727698257,
      "learning_rate": 9.393639909217423e-06,
      "loss": 0.3384,
      "step": 5670
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.789426203517498,
      "learning_rate": 9.390596641690563e-06,
      "loss": 0.2959,
      "step": 5671
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.6837657682249865,
      "learning_rate": 9.387553430812786e-06,
      "loss": 0.3569,
      "step": 5672
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.6085970950326307,
      "learning_rate": 9.384510276866988e-06,
      "loss": 0.2896,
      "step": 5673
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.7286958029161308,
      "learning_rate": 9.381467180136049e-06,
      "loss": 0.3267,
      "step": 5674
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.703042350093264,
      "learning_rate": 9.37842414090285e-06,
      "loss": 0.3179,
      "step": 5675
    },
    {
      "epoch": 1.6,
      "grad_norm": 2.154698831784317,
      "learning_rate": 9.375381159450271e-06,
      "loss": 0.4258,
      "step": 5676
    },
    {
      "epoch": 1.6,
      "grad_norm": 2.0513475865026733,
      "learning_rate": 9.372338236061183e-06,
      "loss": 0.4414,
      "step": 5677
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.539285588358201,
      "learning_rate": 9.369295371018442e-06,
      "loss": 0.2571,
      "step": 5678
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.7228582284594745,
      "learning_rate": 9.366252564604914e-06,
      "loss": 0.3311,
      "step": 5679
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.806102780695767,
      "learning_rate": 9.363209817103455e-06,
      "loss": 0.3438,
      "step": 5680
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.7972687381658958,
      "learning_rate": 9.360167128796913e-06,
      "loss": 0.2605,
      "step": 5681
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.9417784966088358,
      "learning_rate": 9.357124499968124e-06,
      "loss": 0.3662,
      "step": 5682
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.607931104318841,
      "learning_rate": 9.354081930899935e-06,
      "loss": 0.3315,
      "step": 5683
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.9029054772016858,
      "learning_rate": 9.351039421875169e-06,
      "loss": 0.3882,
      "step": 5684
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.7573101067957444,
      "learning_rate": 9.347996973176661e-06,
      "loss": 0.3015,
      "step": 5685
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.5177885982726163,
      "learning_rate": 9.344954585087226e-06,
      "loss": 0.3025,
      "step": 5686
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.6004950593831384,
      "learning_rate": 9.34191225788968e-06,
      "loss": 0.3037,
      "step": 5687
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.6270264863841792,
      "learning_rate": 9.338869991866833e-06,
      "loss": 0.3281,
      "step": 5688
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.5324963333052535,
      "learning_rate": 9.335827787301492e-06,
      "loss": 0.2556,
      "step": 5689
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.788256679988976,
      "learning_rate": 9.332785644476452e-06,
      "loss": 0.374,
      "step": 5690
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.8472410816312672,
      "learning_rate": 9.329743563674505e-06,
      "loss": 0.3628,
      "step": 5691
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.7244612471855112,
      "learning_rate": 9.326701545178434e-06,
      "loss": 0.3132,
      "step": 5692
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.8295378301028147,
      "learning_rate": 9.323659589271028e-06,
      "loss": 0.3545,
      "step": 5693
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.6521433145942415,
      "learning_rate": 9.320617696235058e-06,
      "loss": 0.3333,
      "step": 5694
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.748213573608215,
      "learning_rate": 9.317575866353293e-06,
      "loss": 0.3713,
      "step": 5695
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.7838765922273347,
      "learning_rate": 9.314534099908492e-06,
      "loss": 0.3232,
      "step": 5696
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.5346701104327929,
      "learning_rate": 9.31149239718342e-06,
      "loss": 0.2324,
      "step": 5697
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.6302475341117821,
      "learning_rate": 9.308450758460828e-06,
      "loss": 0.3401,
      "step": 5698
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.5913788039142363,
      "learning_rate": 9.305409184023455e-06,
      "loss": 0.2708,
      "step": 5699
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.9163340977645815,
      "learning_rate": 9.302367674154043e-06,
      "loss": 0.4165,
      "step": 5700
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.7966755559287226,
      "learning_rate": 9.299326229135326e-06,
      "loss": 0.3538,
      "step": 5701
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.7996737013144355,
      "learning_rate": 9.296284849250038e-06,
      "loss": 0.312,
      "step": 5702
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.8289287427672138,
      "learning_rate": 9.293243534780887e-06,
      "loss": 0.3413,
      "step": 5703
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.872165659001665,
      "learning_rate": 9.290202286010602e-06,
      "loss": 0.3472,
      "step": 5704
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.8284912319211482,
      "learning_rate": 9.28716110322188e-06,
      "loss": 0.313,
      "step": 5705
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.5833270391717333,
      "learning_rate": 9.284119986697433e-06,
      "loss": 0.3406,
      "step": 5706
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.7858162678570897,
      "learning_rate": 9.281078936719958e-06,
      "loss": 0.301,
      "step": 5707
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.5947002510369968,
      "learning_rate": 9.278037953572138e-06,
      "loss": 0.3081,
      "step": 5708
    },
    {
      "epoch": 1.61,
      "grad_norm": 2.0478956008497953,
      "learning_rate": 9.274997037536663e-06,
      "loss": 0.3809,
      "step": 5709
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.7871952648072922,
      "learning_rate": 9.271956188896211e-06,
      "loss": 0.3516,
      "step": 5710
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.6545280646373617,
      "learning_rate": 9.268915407933457e-06,
      "loss": 0.2808,
      "step": 5711
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.5205517644205415,
      "learning_rate": 9.265874694931059e-06,
      "loss": 0.3171,
      "step": 5712
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.7433317025904482,
      "learning_rate": 9.262834050171683e-06,
      "loss": 0.3496,
      "step": 5713
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.984959125071603,
      "learning_rate": 9.259793473937977e-06,
      "loss": 0.3945,
      "step": 5714
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.9830145815784284,
      "learning_rate": 9.256752966512595e-06,
      "loss": 0.3267,
      "step": 5715
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.5993737881831884,
      "learning_rate": 9.253712528178169e-06,
      "loss": 0.2832,
      "step": 5716
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.555316688263623,
      "learning_rate": 9.25067215921734e-06,
      "loss": 0.3462,
      "step": 5717
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.7658980710654142,
      "learning_rate": 9.24763185991273e-06,
      "loss": 0.3525,
      "step": 5718
    },
    {
      "epoch": 1.62,
      "grad_norm": 2.082523073741213,
      "learning_rate": 9.244591630546964e-06,
      "loss": 0.3564,
      "step": 5719
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.7016395126122317,
      "learning_rate": 9.241551471402654e-06,
      "loss": 0.2837,
      "step": 5720
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.5212657123644653,
      "learning_rate": 9.238511382762408e-06,
      "loss": 0.2825,
      "step": 5721
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.8174157367697323,
      "learning_rate": 9.235471364908826e-06,
      "loss": 0.3789,
      "step": 5722
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.9792481706651135,
      "learning_rate": 9.232431418124507e-06,
      "loss": 0.4531,
      "step": 5723
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.6739027166247917,
      "learning_rate": 9.22939154269204e-06,
      "loss": 0.3022,
      "step": 5724
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.9695744304610114,
      "learning_rate": 9.226351738893999e-06,
      "loss": 0.4102,
      "step": 5725
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.7354721352993365,
      "learning_rate": 9.223312007012965e-06,
      "loss": 0.3418,
      "step": 5726
    },
    {
      "epoch": 1.62,
      "grad_norm": 2.1667050979597793,
      "learning_rate": 9.220272347331502e-06,
      "loss": 0.4548,
      "step": 5727
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.4892709728051707,
      "learning_rate": 9.217232760132181e-06,
      "loss": 0.2839,
      "step": 5728
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.7030867178664097,
      "learning_rate": 9.214193245697544e-06,
      "loss": 0.2988,
      "step": 5729
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.5513160562145067,
      "learning_rate": 9.211153804310146e-06,
      "loss": 0.3223,
      "step": 5730
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.6240411236555645,
      "learning_rate": 9.208114436252528e-06,
      "loss": 0.2886,
      "step": 5731
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.6942832253110864,
      "learning_rate": 9.205075141807226e-06,
      "loss": 0.2849,
      "step": 5732
    },
    {
      "epoch": 1.62,
      "grad_norm": 2.015350073017708,
      "learning_rate": 9.20203592125676e-06,
      "loss": 0.3848,
      "step": 5733
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.7449872969664462,
      "learning_rate": 9.19899677488366e-06,
      "loss": 0.3086,
      "step": 5734
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.7254912199160704,
      "learning_rate": 9.195957702970434e-06,
      "loss": 0.3481,
      "step": 5735
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.918056987600494,
      "learning_rate": 9.192918705799594e-06,
      "loss": 0.4229,
      "step": 5736
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.7941586460949541,
      "learning_rate": 9.189879783653633e-06,
      "loss": 0.3022,
      "step": 5737
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.8368601453008357,
      "learning_rate": 9.186840936815047e-06,
      "loss": 0.3267,
      "step": 5738
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.689914626744783,
      "learning_rate": 9.18380216556632e-06,
      "loss": 0.3501,
      "step": 5739
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.8951009512717452,
      "learning_rate": 9.180763470189938e-06,
      "loss": 0.395,
      "step": 5740
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.7951444068570752,
      "learning_rate": 9.17772485096837e-06,
      "loss": 0.3223,
      "step": 5741
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.8794553722911111,
      "learning_rate": 9.174686308184075e-06,
      "loss": 0.3857,
      "step": 5742
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.955854780139857,
      "learning_rate": 9.171647842119515e-06,
      "loss": 0.3884,
      "step": 5743
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.5427992171298335,
      "learning_rate": 9.16860945305714e-06,
      "loss": 0.29,
      "step": 5744
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.6498517733270166,
      "learning_rate": 9.165571141279397e-06,
      "loss": 0.28,
      "step": 5745
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.6697000450520134,
      "learning_rate": 9.162532907068713e-06,
      "loss": 0.2559,
      "step": 5746
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.6766272425041069,
      "learning_rate": 9.159494750707527e-06,
      "loss": 0.3965,
      "step": 5747
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.956643316260669,
      "learning_rate": 9.156456672478252e-06,
      "loss": 0.3135,
      "step": 5748
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.7357022854858417,
      "learning_rate": 9.153418672663313e-06,
      "loss": 0.3359,
      "step": 5749
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.9540490703676832,
      "learning_rate": 9.15038075154511e-06,
      "loss": 0.4038,
      "step": 5750
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.858432237151177,
      "learning_rate": 9.14734290940604e-06,
      "loss": 0.3418,
      "step": 5751
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.7261993475582922,
      "learning_rate": 9.144305146528502e-06,
      "loss": 0.3545,
      "step": 5752
    },
    {
      "epoch": 1.62,
      "grad_norm": 2.1986680011185245,
      "learning_rate": 9.141267463194883e-06,
      "loss": 0.3564,
      "step": 5753
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.693591465699814,
      "learning_rate": 9.138229859687553e-06,
      "loss": 0.3076,
      "step": 5754
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.6245875367210403,
      "learning_rate": 9.135192336288885e-06,
      "loss": 0.2634,
      "step": 5755
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.9913830081676718,
      "learning_rate": 9.132154893281244e-06,
      "loss": 0.3848,
      "step": 5756
    },
    {
      "epoch": 1.63,
      "grad_norm": 2.017235509784282,
      "learning_rate": 9.129117530946986e-06,
      "loss": 0.3779,
      "step": 5757
    },
    {
      "epoch": 1.63,
      "grad_norm": 2.003216250677568,
      "learning_rate": 9.126080249568457e-06,
      "loss": 0.3584,
      "step": 5758
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.6783160781205197,
      "learning_rate": 9.123043049427996e-06,
      "loss": 0.3289,
      "step": 5759
    },
    {
      "epoch": 1.63,
      "grad_norm": 2.0504253180799994,
      "learning_rate": 9.120005930807939e-06,
      "loss": 0.3296,
      "step": 5760
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.8041913928749793,
      "learning_rate": 9.116968893990609e-06,
      "loss": 0.3472,
      "step": 5761
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.8825400437555428,
      "learning_rate": 9.113931939258327e-06,
      "loss": 0.3643,
      "step": 5762
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.970334172794205,
      "learning_rate": 9.110895066893398e-06,
      "loss": 0.4092,
      "step": 5763
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.6316031814659497,
      "learning_rate": 9.10785827717813e-06,
      "loss": 0.2983,
      "step": 5764
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.8480640389533516,
      "learning_rate": 9.104821570394811e-06,
      "loss": 0.2959,
      "step": 5765
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.8698483955986043,
      "learning_rate": 9.101784946825739e-06,
      "loss": 0.356,
      "step": 5766
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.7382912210471653,
      "learning_rate": 9.098748406753181e-06,
      "loss": 0.343,
      "step": 5767
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.532539514647865,
      "learning_rate": 9.095711950459412e-06,
      "loss": 0.3113,
      "step": 5768
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.6611777792442066,
      "learning_rate": 9.0926755782267e-06,
      "loss": 0.2913,
      "step": 5769
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.6758664098761908,
      "learning_rate": 9.0896392903373e-06,
      "loss": 0.3391,
      "step": 5770
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.7297333313885344,
      "learning_rate": 9.086603087073457e-06,
      "loss": 0.3193,
      "step": 5771
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.9425030598951873,
      "learning_rate": 9.083566968717412e-06,
      "loss": 0.4072,
      "step": 5772
    },
    {
      "epoch": 1.63,
      "grad_norm": 2.0215526312016747,
      "learning_rate": 9.080530935551398e-06,
      "loss": 0.4033,
      "step": 5773
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.7762252782844161,
      "learning_rate": 9.077494987857644e-06,
      "loss": 0.3809,
      "step": 5774
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.7823509035310408,
      "learning_rate": 9.074459125918356e-06,
      "loss": 0.3696,
      "step": 5775
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.8381386158245967,
      "learning_rate": 9.071423350015747e-06,
      "loss": 0.3633,
      "step": 5776
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.6632417697405286,
      "learning_rate": 9.068387660432023e-06,
      "loss": 0.2449,
      "step": 5777
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.7574711783512094,
      "learning_rate": 9.06535205744937e-06,
      "loss": 0.3867,
      "step": 5778
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.6864577884340064,
      "learning_rate": 9.062316541349978e-06,
      "loss": 0.374,
      "step": 5779
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.6255918025144238,
      "learning_rate": 9.059281112416017e-06,
      "loss": 0.29,
      "step": 5780
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.84797508262636,
      "learning_rate": 9.056245770929659e-06,
      "loss": 0.3269,
      "step": 5781
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.6332732274188697,
      "learning_rate": 9.053210517173061e-06,
      "loss": 0.2998,
      "step": 5782
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.5657402786435135,
      "learning_rate": 9.050175351428381e-06,
      "loss": 0.3079,
      "step": 5783
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.5397819525844298,
      "learning_rate": 9.04714027397776e-06,
      "loss": 0.2463,
      "step": 5784
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.6439165454990037,
      "learning_rate": 9.044105285103327e-06,
      "loss": 0.3081,
      "step": 5785
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.8198631030139116,
      "learning_rate": 9.04107038508722e-06,
      "loss": 0.3262,
      "step": 5786
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.8454084485276656,
      "learning_rate": 9.038035574211553e-06,
      "loss": 0.3013,
      "step": 5787
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.9812134323045847,
      "learning_rate": 9.035000852758433e-06,
      "loss": 0.3687,
      "step": 5788
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.5935713753573113,
      "learning_rate": 9.031966221009966e-06,
      "loss": 0.3286,
      "step": 5789
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.4914359337422776,
      "learning_rate": 9.028931679248249e-06,
      "loss": 0.2314,
      "step": 5790
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.8345269483732083,
      "learning_rate": 9.025897227755367e-06,
      "loss": 0.335,
      "step": 5791
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.999608668774864,
      "learning_rate": 9.022862866813392e-06,
      "loss": 0.3267,
      "step": 5792
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.8882950497563964,
      "learning_rate": 9.019828596704394e-06,
      "loss": 0.3572,
      "step": 5793
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.5624625211640073,
      "learning_rate": 9.016794417710439e-06,
      "loss": 0.248,
      "step": 5794
    },
    {
      "epoch": 1.64,
      "grad_norm": 2.0244112400580314,
      "learning_rate": 9.013760330113575e-06,
      "loss": 0.4062,
      "step": 5795
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.7675162494126184,
      "learning_rate": 9.010726334195851e-06,
      "loss": 0.3477,
      "step": 5796
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.5265072150359553,
      "learning_rate": 9.007692430239294e-06,
      "loss": 0.2971,
      "step": 5797
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.6848026447216387,
      "learning_rate": 9.004658618525932e-06,
      "loss": 0.3135,
      "step": 5798
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.9247216275460794,
      "learning_rate": 9.001624899337785e-06,
      "loss": 0.3145,
      "step": 5799
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.960956670223678,
      "learning_rate": 8.998591272956866e-06,
      "loss": 0.4343,
      "step": 5800
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.463469550147598,
      "learning_rate": 8.995557739665172e-06,
      "loss": 0.2566,
      "step": 5801
    },
    {
      "epoch": 1.64,
      "grad_norm": 2.0546300859168065,
      "learning_rate": 8.99252429974469e-06,
      "loss": 0.3921,
      "step": 5802
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.6631468006642933,
      "learning_rate": 8.989490953477413e-06,
      "loss": 0.2751,
      "step": 5803
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.6033690487012493,
      "learning_rate": 8.98645770114531e-06,
      "loss": 0.2917,
      "step": 5804
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.6769980108615108,
      "learning_rate": 8.983424543030344e-06,
      "loss": 0.3157,
      "step": 5805
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.573618124792943,
      "learning_rate": 8.980391479414478e-06,
      "loss": 0.2695,
      "step": 5806
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.742158397424427,
      "learning_rate": 8.977358510579658e-06,
      "loss": 0.3137,
      "step": 5807
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.6144785166206197,
      "learning_rate": 8.974325636807826e-06,
      "loss": 0.2935,
      "step": 5808
    },
    {
      "epoch": 1.64,
      "grad_norm": 2.040206395000282,
      "learning_rate": 8.971292858380908e-06,
      "loss": 0.4297,
      "step": 5809
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.768002601599421,
      "learning_rate": 8.96826017558083e-06,
      "loss": 0.3721,
      "step": 5810
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.6290554886183128,
      "learning_rate": 8.9652275886895e-06,
      "loss": 0.2808,
      "step": 5811
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.6194603109884664,
      "learning_rate": 8.962195097988825e-06,
      "loss": 0.3657,
      "step": 5812
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.885221805281513,
      "learning_rate": 8.959162703760706e-06,
      "loss": 0.3242,
      "step": 5813
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.7147838885590356,
      "learning_rate": 8.95613040628702e-06,
      "loss": 0.3623,
      "step": 5814
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.5643558050173287,
      "learning_rate": 8.953098205849647e-06,
      "loss": 0.3096,
      "step": 5815
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.6508797901245253,
      "learning_rate": 8.950066102730456e-06,
      "loss": 0.2725,
      "step": 5816
    },
    {
      "epoch": 1.64,
      "grad_norm": 2.026717993972473,
      "learning_rate": 8.947034097211309e-06,
      "loss": 0.3914,
      "step": 5817
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.8858901263652308,
      "learning_rate": 8.944002189574047e-06,
      "loss": 0.3555,
      "step": 5818
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.8559487524750147,
      "learning_rate": 8.94097038010052e-06,
      "loss": 0.3611,
      "step": 5819
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.8957902656048526,
      "learning_rate": 8.937938669072557e-06,
      "loss": 0.3501,
      "step": 5820
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.520427781486123,
      "learning_rate": 8.93490705677198e-06,
      "loss": 0.2954,
      "step": 5821
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.8632162790083495,
      "learning_rate": 8.931875543480601e-06,
      "loss": 0.3608,
      "step": 5822
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.9293204312940637,
      "learning_rate": 8.928844129480228e-06,
      "loss": 0.3774,
      "step": 5823
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.8526774213645685,
      "learning_rate": 8.92581281505265e-06,
      "loss": 0.3662,
      "step": 5824
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.7104513243381647,
      "learning_rate": 8.922781600479663e-06,
      "loss": 0.3127,
      "step": 5825
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.7456243811816077,
      "learning_rate": 8.91975048604303e-06,
      "loss": 0.3289,
      "step": 5826
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.8199062237143668,
      "learning_rate": 8.916719472024528e-06,
      "loss": 0.3042,
      "step": 5827
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.741784555328843,
      "learning_rate": 8.91368855870591e-06,
      "loss": 0.3779,
      "step": 5828
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.718729420153302,
      "learning_rate": 8.91065774636893e-06,
      "loss": 0.3596,
      "step": 5829
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.9050258805447982,
      "learning_rate": 8.90762703529532e-06,
      "loss": 0.3521,
      "step": 5830
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.708092529042536,
      "learning_rate": 8.904596425766817e-06,
      "loss": 0.342,
      "step": 5831
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.604259667020751,
      "learning_rate": 8.901565918065134e-06,
      "loss": 0.2854,
      "step": 5832
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.8919279987705522,
      "learning_rate": 8.898535512471986e-06,
      "loss": 0.3677,
      "step": 5833
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.6561697475743462,
      "learning_rate": 8.895505209269078e-06,
      "loss": 0.3281,
      "step": 5834
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.6357745632433474,
      "learning_rate": 8.89247500873809e-06,
      "loss": 0.343,
      "step": 5835
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.7555801449090163,
      "learning_rate": 8.889444911160713e-06,
      "loss": 0.3223,
      "step": 5836
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.8794949279975237,
      "learning_rate": 8.88641491681862e-06,
      "loss": 0.3413,
      "step": 5837
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.7926548328936323,
      "learning_rate": 8.883385025993474e-06,
      "loss": 0.3555,
      "step": 5838
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.6382140865475212,
      "learning_rate": 8.880355238966923e-06,
      "loss": 0.2983,
      "step": 5839
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.8903715742120637,
      "learning_rate": 8.877325556020615e-06,
      "loss": 0.333,
      "step": 5840
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.8913786577591924,
      "learning_rate": 8.874295977436182e-06,
      "loss": 0.4175,
      "step": 5841
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.8308777061067312,
      "learning_rate": 8.871266503495255e-06,
      "loss": 0.3652,
      "step": 5842
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.9293348257641638,
      "learning_rate": 8.868237134479437e-06,
      "loss": 0.4209,
      "step": 5843
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.5691786442202926,
      "learning_rate": 8.865207870670342e-06,
      "loss": 0.3218,
      "step": 5844
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.8704809381301737,
      "learning_rate": 8.862178712349562e-06,
      "loss": 0.3901,
      "step": 5845
    },
    {
      "epoch": 1.65,
      "grad_norm": 2.0172437817597753,
      "learning_rate": 8.859149659798685e-06,
      "loss": 0.4058,
      "step": 5846
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.729945152617339,
      "learning_rate": 8.856120713299284e-06,
      "loss": 0.2837,
      "step": 5847
    },
    {
      "epoch": 1.65,
      "grad_norm": 2.183462439376854,
      "learning_rate": 8.853091873132921e-06,
      "loss": 0.3677,
      "step": 5848
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.8608477063953457,
      "learning_rate": 8.850063139581156e-06,
      "loss": 0.3594,
      "step": 5849
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.7438965267365958,
      "learning_rate": 8.847034512925536e-06,
      "loss": 0.3853,
      "step": 5850
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.6443983482967433,
      "learning_rate": 8.844005993447599e-06,
      "loss": 0.3059,
      "step": 5851
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.6442263247237974,
      "learning_rate": 8.840977581428863e-06,
      "loss": 0.2676,
      "step": 5852
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.7489597342602061,
      "learning_rate": 8.837949277150849e-06,
      "loss": 0.3403,
      "step": 5853
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.979462580738102,
      "learning_rate": 8.83492108089506e-06,
      "loss": 0.3213,
      "step": 5854
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.5304475717323267,
      "learning_rate": 8.831892992943e-06,
      "loss": 0.2664,
      "step": 5855
    },
    {
      "epoch": 1.65,
      "grad_norm": 2.1838126522501793,
      "learning_rate": 8.828865013576143e-06,
      "loss": 0.4604,
      "step": 5856
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.7672959601419378,
      "learning_rate": 8.825837143075973e-06,
      "loss": 0.3286,
      "step": 5857
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.891096887944942,
      "learning_rate": 8.822809381723952e-06,
      "loss": 0.415,
      "step": 5858
    },
    {
      "epoch": 1.65,
      "grad_norm": 2.004958895266214,
      "learning_rate": 8.81978172980154e-06,
      "loss": 0.4463,
      "step": 5859
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.649800289911441,
      "learning_rate": 8.816754187590175e-06,
      "loss": 0.2761,
      "step": 5860
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.6499804786434087,
      "learning_rate": 8.813726755371298e-06,
      "loss": 0.2825,
      "step": 5861
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.945687557825563,
      "learning_rate": 8.81069943342633e-06,
      "loss": 0.3977,
      "step": 5862
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.741027011280326,
      "learning_rate": 8.807672222036692e-06,
      "loss": 0.3125,
      "step": 5863
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.7826950047811156,
      "learning_rate": 8.804645121483781e-06,
      "loss": 0.355,
      "step": 5864
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.8896123886737997,
      "learning_rate": 8.801618132048992e-06,
      "loss": 0.4219,
      "step": 5865
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.459672222563807,
      "learning_rate": 8.798591254013712e-06,
      "loss": 0.2576,
      "step": 5866
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.7344523743829658,
      "learning_rate": 8.795564487659313e-06,
      "loss": 0.3843,
      "step": 5867
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.4458484936213285,
      "learning_rate": 8.792537833267161e-06,
      "loss": 0.3076,
      "step": 5868
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.7920416152960021,
      "learning_rate": 8.789511291118601e-06,
      "loss": 0.3418,
      "step": 5869
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.9610581032042729,
      "learning_rate": 8.786484861494984e-06,
      "loss": 0.4248,
      "step": 5870
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.7760504327624107,
      "learning_rate": 8.783458544677633e-06,
      "loss": 0.3411,
      "step": 5871
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.6429254168538474,
      "learning_rate": 8.780432340947879e-06,
      "loss": 0.2764,
      "step": 5872
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.5958837417373497,
      "learning_rate": 8.777406250587021e-06,
      "loss": 0.3176,
      "step": 5873
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.8386339712437996,
      "learning_rate": 8.77438027387637e-06,
      "loss": 0.3594,
      "step": 5874
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.6952493607184986,
      "learning_rate": 8.771354411097207e-06,
      "loss": 0.3198,
      "step": 5875
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.7766471429837412,
      "learning_rate": 8.768328662530818e-06,
      "loss": 0.2966,
      "step": 5876
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.5681371725523288,
      "learning_rate": 8.765303028458468e-06,
      "loss": 0.2793,
      "step": 5877
    },
    {
      "epoch": 1.66,
      "grad_norm": 2.0306896658220674,
      "learning_rate": 8.762277509161413e-06,
      "loss": 0.4087,
      "step": 5878
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.835554434676967,
      "learning_rate": 8.7592521049209e-06,
      "loss": 0.3408,
      "step": 5879
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.7947721071299392,
      "learning_rate": 8.756226816018172e-06,
      "loss": 0.3533,
      "step": 5880
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.7679415455833778,
      "learning_rate": 8.75320164273445e-06,
      "loss": 0.3062,
      "step": 5881
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.7506892206959535,
      "learning_rate": 8.750176585350945e-06,
      "loss": 0.2817,
      "step": 5882
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.9021893407205621,
      "learning_rate": 8.747151644148867e-06,
      "loss": 0.3213,
      "step": 5883
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.8674789873796374,
      "learning_rate": 8.744126819409405e-06,
      "loss": 0.3506,
      "step": 5884
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.7684215351701633,
      "learning_rate": 8.741102111413749e-06,
      "loss": 0.3462,
      "step": 5885
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.7324809793336327,
      "learning_rate": 8.738077520443061e-06,
      "loss": 0.313,
      "step": 5886
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.808914759003523,
      "learning_rate": 8.735053046778506e-06,
      "loss": 0.3157,
      "step": 5887
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.8637365513881725,
      "learning_rate": 8.732028690701235e-06,
      "loss": 0.3848,
      "step": 5888
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.691966319900858,
      "learning_rate": 8.729004452492388e-06,
      "loss": 0.3301,
      "step": 5889
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.670854424143492,
      "learning_rate": 8.725980332433089e-06,
      "loss": 0.3442,
      "step": 5890
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.7263880952675703,
      "learning_rate": 8.722956330804456e-06,
      "loss": 0.3179,
      "step": 5891
    },
    {
      "epoch": 1.66,
      "grad_norm": 2.094366299195598,
      "learning_rate": 8.719932447887594e-06,
      "loss": 0.3853,
      "step": 5892
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.8122966398643594,
      "learning_rate": 8.716908683963602e-06,
      "loss": 0.3411,
      "step": 5893
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.6913289517085355,
      "learning_rate": 8.713885039313562e-06,
      "loss": 0.2646,
      "step": 5894
    },
    {
      "epoch": 1.66,
      "grad_norm": 2.1079290880479227,
      "learning_rate": 8.710861514218545e-06,
      "loss": 0.3906,
      "step": 5895
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.843248090915413,
      "learning_rate": 8.707838108959617e-06,
      "loss": 0.3594,
      "step": 5896
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.7528957962968938,
      "learning_rate": 8.704814823817822e-06,
      "loss": 0.3506,
      "step": 5897
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.9350604759745786,
      "learning_rate": 8.701791659074206e-06,
      "loss": 0.3044,
      "step": 5898
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.6302813489113652,
      "learning_rate": 8.698768615009789e-06,
      "loss": 0.2732,
      "step": 5899
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.5365426084717453,
      "learning_rate": 8.695745691905599e-06,
      "loss": 0.2532,
      "step": 5900
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.736746768904932,
      "learning_rate": 8.692722890042632e-06,
      "loss": 0.3674,
      "step": 5901
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.6186590952493665,
      "learning_rate": 8.689700209701887e-06,
      "loss": 0.2627,
      "step": 5902
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.7432794302346004,
      "learning_rate": 8.686677651164345e-06,
      "loss": 0.3647,
      "step": 5903
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.7982134207036875,
      "learning_rate": 8.683655214710982e-06,
      "loss": 0.3857,
      "step": 5904
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.5246827725259406,
      "learning_rate": 8.680632900622752e-06,
      "loss": 0.2515,
      "step": 5905
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.7288283373340938,
      "learning_rate": 8.677610709180612e-06,
      "loss": 0.3818,
      "step": 5906
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.7637271459322172,
      "learning_rate": 8.674588640665495e-06,
      "loss": 0.3135,
      "step": 5907
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.8013372407135924,
      "learning_rate": 8.671566695358324e-06,
      "loss": 0.4045,
      "step": 5908
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.885051445864974,
      "learning_rate": 8.668544873540017e-06,
      "loss": 0.3389,
      "step": 5909
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.6033084053399898,
      "learning_rate": 8.665523175491484e-06,
      "loss": 0.2617,
      "step": 5910
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.5572669120485638,
      "learning_rate": 8.662501601493607e-06,
      "loss": 0.3203,
      "step": 5911
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.7542813385825773,
      "learning_rate": 8.659480151827267e-06,
      "loss": 0.3237,
      "step": 5912
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.712629702894739,
      "learning_rate": 8.65645882677334e-06,
      "loss": 0.3164,
      "step": 5913
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.6871337188704116,
      "learning_rate": 8.65343762661268e-06,
      "loss": 0.3069,
      "step": 5914
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.5652207664529634,
      "learning_rate": 8.650416551626126e-06,
      "loss": 0.3167,
      "step": 5915
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.9753950363300987,
      "learning_rate": 8.647395602094517e-06,
      "loss": 0.3574,
      "step": 5916
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.8200186641998715,
      "learning_rate": 8.64437477829868e-06,
      "loss": 0.3171,
      "step": 5917
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.8484165136364719,
      "learning_rate": 8.641354080519422e-06,
      "loss": 0.3574,
      "step": 5918
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.8347651671156113,
      "learning_rate": 8.638333509037537e-06,
      "loss": 0.3672,
      "step": 5919
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.83660782693055,
      "learning_rate": 8.635313064133817e-06,
      "loss": 0.4087,
      "step": 5920
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.7715013942777935,
      "learning_rate": 8.632292746089034e-06,
      "loss": 0.3223,
      "step": 5921
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.7686770446114777,
      "learning_rate": 8.629272555183956e-06,
      "loss": 0.3301,
      "step": 5922
    },
    {
      "epoch": 1.67,
      "grad_norm": 2.137534320537121,
      "learning_rate": 8.626252491699335e-06,
      "loss": 0.4043,
      "step": 5923
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.613378129033506,
      "learning_rate": 8.623232555915907e-06,
      "loss": 0.2646,
      "step": 5924
    },
    {
      "epoch": 1.67,
      "grad_norm": 2.0736421846510287,
      "learning_rate": 8.6202127481144e-06,
      "loss": 0.4321,
      "step": 5925
    },
    {
      "epoch": 1.67,
      "grad_norm": 2.1692453307665227,
      "learning_rate": 8.617193068575534e-06,
      "loss": 0.4741,
      "step": 5926
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.9662393212263156,
      "learning_rate": 8.61417351758001e-06,
      "loss": 0.3301,
      "step": 5927
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.7750883662276766,
      "learning_rate": 8.611154095408521e-06,
      "loss": 0.3235,
      "step": 5928
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.7233372861787382,
      "learning_rate": 8.608134802341745e-06,
      "loss": 0.3716,
      "step": 5929
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.8265375171776606,
      "learning_rate": 8.605115638660356e-06,
      "loss": 0.3516,
      "step": 5930
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.7188352294117253,
      "learning_rate": 8.602096604645009e-06,
      "loss": 0.282,
      "step": 5931
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.8710115452530318,
      "learning_rate": 8.599077700576342e-06,
      "loss": 0.4189,
      "step": 5932
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.9486171699727686,
      "learning_rate": 8.59605892673499e-06,
      "loss": 0.4624,
      "step": 5933
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.6528448546685206,
      "learning_rate": 8.593040283401576e-06,
      "loss": 0.3228,
      "step": 5934
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.8742641464141558,
      "learning_rate": 8.590021770856708e-06,
      "loss": 0.3965,
      "step": 5935
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.7735734651827755,
      "learning_rate": 8.587003389380977e-06,
      "loss": 0.3389,
      "step": 5936
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.665938342238145,
      "learning_rate": 8.58398513925497e-06,
      "loss": 0.3096,
      "step": 5937
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.561624078412844,
      "learning_rate": 8.580967020759257e-06,
      "loss": 0.3042,
      "step": 5938
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.9611452270239884,
      "learning_rate": 8.577949034174395e-06,
      "loss": 0.3462,
      "step": 5939
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.8901830536600261,
      "learning_rate": 8.57493117978094e-06,
      "loss": 0.3079,
      "step": 5940
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.9552084138598673,
      "learning_rate": 8.571913457859418e-06,
      "loss": 0.3691,
      "step": 5941
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.6719116538043866,
      "learning_rate": 8.56889586869035e-06,
      "loss": 0.355,
      "step": 5942
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.7459571066463735,
      "learning_rate": 8.565878412554251e-06,
      "loss": 0.3276,
      "step": 5943
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.6506991623248843,
      "learning_rate": 8.562861089731618e-06,
      "loss": 0.2988,
      "step": 5944
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.7243006350088084,
      "learning_rate": 8.559843900502934e-06,
      "loss": 0.3633,
      "step": 5945
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.8902593255652531,
      "learning_rate": 8.556826845148669e-06,
      "loss": 0.3291,
      "step": 5946
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.9498442893727768,
      "learning_rate": 8.55380992394929e-06,
      "loss": 0.3638,
      "step": 5947
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.8634699343009724,
      "learning_rate": 8.550793137185243e-06,
      "loss": 0.4028,
      "step": 5948
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.5534128755046168,
      "learning_rate": 8.547776485136957e-06,
      "loss": 0.2837,
      "step": 5949
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.7595684448532214,
      "learning_rate": 8.544759968084863e-06,
      "loss": 0.3286,
      "step": 5950
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.6310548504770928,
      "learning_rate": 8.541743586309366e-06,
      "loss": 0.3247,
      "step": 5951
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.5085317178665139,
      "learning_rate": 8.53872734009087e-06,
      "loss": 0.2854,
      "step": 5952
    },
    {
      "epoch": 1.68,
      "grad_norm": 2.07148959919668,
      "learning_rate": 8.535711229709749e-06,
      "loss": 0.3843,
      "step": 5953
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.5549113018481724,
      "learning_rate": 8.532695255446384e-06,
      "loss": 0.3174,
      "step": 5954
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.92528251014716,
      "learning_rate": 8.52967941758113e-06,
      "loss": 0.3682,
      "step": 5955
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.666715078678103,
      "learning_rate": 8.52666371639434e-06,
      "loss": 0.3149,
      "step": 5956
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.793217303493374,
      "learning_rate": 8.523648152166349e-06,
      "loss": 0.3254,
      "step": 5957
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.7653004525025884,
      "learning_rate": 8.520632725177468e-06,
      "loss": 0.3413,
      "step": 5958
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.8537353248969142,
      "learning_rate": 8.517617435708011e-06,
      "loss": 0.3413,
      "step": 5959
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.9881754609157398,
      "learning_rate": 8.51460228403828e-06,
      "loss": 0.3789,
      "step": 5960
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.8228633425134373,
      "learning_rate": 8.511587270448556e-06,
      "loss": 0.3804,
      "step": 5961
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.5940429677975982,
      "learning_rate": 8.508572395219104e-06,
      "loss": 0.2791,
      "step": 5962
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.8292468400692303,
      "learning_rate": 8.505557658630186e-06,
      "loss": 0.3401,
      "step": 5963
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.7455870256322072,
      "learning_rate": 8.502543060962043e-06,
      "loss": 0.2512,
      "step": 5964
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.798000110753008,
      "learning_rate": 8.499528602494914e-06,
      "loss": 0.3374,
      "step": 5965
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.8853470851231104,
      "learning_rate": 8.49651428350901e-06,
      "loss": 0.3149,
      "step": 5966
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.856783066828567,
      "learning_rate": 8.493500104284539e-06,
      "loss": 0.355,
      "step": 5967
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.827165151944094,
      "learning_rate": 8.490486065101698e-06,
      "loss": 0.3833,
      "step": 5968
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.806942865178122,
      "learning_rate": 8.487472166240665e-06,
      "loss": 0.2974,
      "step": 5969
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.8480272743521549,
      "learning_rate": 8.484458407981601e-06,
      "loss": 0.3291,
      "step": 5970
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.8440668451318751,
      "learning_rate": 8.481444790604668e-06,
      "loss": 0.2976,
      "step": 5971
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.7024562034533268,
      "learning_rate": 8.478431314390002e-06,
      "loss": 0.2734,
      "step": 5972
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.8085490564080096,
      "learning_rate": 8.475417979617732e-06,
      "loss": 0.3955,
      "step": 5973
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.478942458248738,
      "learning_rate": 8.472404786567974e-06,
      "loss": 0.325,
      "step": 5974
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.5932892751417718,
      "learning_rate": 8.469391735520824e-06,
      "loss": 0.2571,
      "step": 5975
    },
    {
      "epoch": 1.69,
      "grad_norm": 2.1184432629275776,
      "learning_rate": 8.466378826756373e-06,
      "loss": 0.3428,
      "step": 5976
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.5907506106990779,
      "learning_rate": 8.463366060554698e-06,
      "loss": 0.2732,
      "step": 5977
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.614978542306381,
      "learning_rate": 8.460353437195864e-06,
      "loss": 0.3184,
      "step": 5978
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.664562177502226,
      "learning_rate": 8.457340956959905e-06,
      "loss": 0.2869,
      "step": 5979
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.9054689445358823,
      "learning_rate": 8.454328620126871e-06,
      "loss": 0.3301,
      "step": 5980
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.9038423484404512,
      "learning_rate": 8.451316426976773e-06,
      "loss": 0.3867,
      "step": 5981
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.9523580774166593,
      "learning_rate": 8.448304377789628e-06,
      "loss": 0.3721,
      "step": 5982
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.747341174675607,
      "learning_rate": 8.445292472845423e-06,
      "loss": 0.3247,
      "step": 5983
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.874318356349631,
      "learning_rate": 8.442280712424146e-06,
      "loss": 0.373,
      "step": 5984
    },
    {
      "epoch": 1.69,
      "grad_norm": 2.016411698659357,
      "learning_rate": 8.439269096805758e-06,
      "loss": 0.3945,
      "step": 5985
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.820697700808939,
      "learning_rate": 8.436257626270225e-06,
      "loss": 0.3491,
      "step": 5986
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.7568342088965867,
      "learning_rate": 8.433246301097477e-06,
      "loss": 0.2983,
      "step": 5987
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.7700199298669346,
      "learning_rate": 8.430235121567444e-06,
      "loss": 0.3452,
      "step": 5988
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.7379998416950262,
      "learning_rate": 8.42722408796004e-06,
      "loss": 0.3677,
      "step": 5989
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.7446500300443164,
      "learning_rate": 8.424213200555171e-06,
      "loss": 0.3218,
      "step": 5990
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.757713389510954,
      "learning_rate": 8.42120245963272e-06,
      "loss": 0.3586,
      "step": 5991
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.952193970857859,
      "learning_rate": 8.418191865472559e-06,
      "loss": 0.3911,
      "step": 5992
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.7792977075234984,
      "learning_rate": 8.415181418354548e-06,
      "loss": 0.3306,
      "step": 5993
    },
    {
      "epoch": 1.69,
      "grad_norm": 2.1629772644100096,
      "learning_rate": 8.412171118558534e-06,
      "loss": 0.3799,
      "step": 5994
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.945438195327315,
      "learning_rate": 8.409160966364351e-06,
      "loss": 0.3677,
      "step": 5995
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.8170116467724162,
      "learning_rate": 8.406150962051813e-06,
      "loss": 0.3501,
      "step": 5996
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.8264387741986774,
      "learning_rate": 8.40314110590073e-06,
      "loss": 0.3076,
      "step": 5997
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.6916426720980506,
      "learning_rate": 8.400131398190887e-06,
      "loss": 0.3442,
      "step": 5998
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.6579215527060276,
      "learning_rate": 8.397121839202069e-06,
      "loss": 0.3242,
      "step": 5999
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.7093946494988188,
      "learning_rate": 8.394112429214032e-06,
      "loss": 0.3706,
      "step": 6000
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.7605864392288466,
      "learning_rate": 8.391103168506529e-06,
      "loss": 0.3223,
      "step": 6001
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.4984592604257376,
      "learning_rate": 8.388094057359295e-06,
      "loss": 0.3193,
      "step": 6002
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.6093289893688352,
      "learning_rate": 8.385085096052053e-06,
      "loss": 0.2979,
      "step": 6003
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.9147456414845632,
      "learning_rate": 8.38207628486451e-06,
      "loss": 0.3848,
      "step": 6004
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.8403718358318801,
      "learning_rate": 8.379067624076358e-06,
      "loss": 0.3254,
      "step": 6005
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.7588331960090577,
      "learning_rate": 8.376059113967279e-06,
      "loss": 0.3179,
      "step": 6006
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.732310158890937,
      "learning_rate": 8.373050754816942e-06,
      "loss": 0.3008,
      "step": 6007
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.7430228171760833,
      "learning_rate": 8.370042546904992e-06,
      "loss": 0.3242,
      "step": 6008
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.9012978424017497,
      "learning_rate": 8.36703449051107e-06,
      "loss": 0.3525,
      "step": 6009
    },
    {
      "epoch": 1.7,
      "grad_norm": 2.2376192556071586,
      "learning_rate": 8.364026585914802e-06,
      "loss": 0.4917,
      "step": 6010
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.9047315749316733,
      "learning_rate": 8.361018833395792e-06,
      "loss": 0.3628,
      "step": 6011
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.758488485160588,
      "learning_rate": 8.358011233233646e-06,
      "loss": 0.3984,
      "step": 6012
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.9799704889962415,
      "learning_rate": 8.355003785707932e-06,
      "loss": 0.3867,
      "step": 6013
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.998185824532213,
      "learning_rate": 8.351996491098227e-06,
      "loss": 0.3691,
      "step": 6014
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.6112235871188354,
      "learning_rate": 8.348989349684077e-06,
      "loss": 0.3223,
      "step": 6015
    },
    {
      "epoch": 1.7,
      "grad_norm": 2.098348878915678,
      "learning_rate": 8.345982361745029e-06,
      "loss": 0.4277,
      "step": 6016
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.9115707693945454,
      "learning_rate": 8.342975527560601e-06,
      "loss": 0.3857,
      "step": 6017
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.5717789027725235,
      "learning_rate": 8.339968847410301e-06,
      "loss": 0.2637,
      "step": 6018
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.7245164547294665,
      "learning_rate": 8.33696232157363e-06,
      "loss": 0.3203,
      "step": 6019
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.7098429831557294,
      "learning_rate": 8.33395595033007e-06,
      "loss": 0.2827,
      "step": 6020
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.7101000184162167,
      "learning_rate": 8.330949733959084e-06,
      "loss": 0.3286,
      "step": 6021
    },
    {
      "epoch": 1.7,
      "grad_norm": 2.183311945022186,
      "learning_rate": 8.327943672740126e-06,
      "loss": 0.4795,
      "step": 6022
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.9203778728694343,
      "learning_rate": 8.324937766952638e-06,
      "loss": 0.3291,
      "step": 6023
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.9379021551337599,
      "learning_rate": 8.32193201687604e-06,
      "loss": 0.3813,
      "step": 6024
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.712340883372593,
      "learning_rate": 8.31892642278974e-06,
      "loss": 0.3203,
      "step": 6025
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.8618820174016397,
      "learning_rate": 8.315920984973134e-06,
      "loss": 0.3582,
      "step": 6026
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.8692823686105673,
      "learning_rate": 8.312915703705603e-06,
      "loss": 0.3616,
      "step": 6027
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.6416842047958486,
      "learning_rate": 8.309910579266514e-06,
      "loss": 0.3137,
      "step": 6028
    },
    {
      "epoch": 1.7,
      "grad_norm": 2.026042649973368,
      "learning_rate": 8.306905611935221e-06,
      "loss": 0.4937,
      "step": 6029
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.9152523938215382,
      "learning_rate": 8.303900801991052e-06,
      "loss": 0.3652,
      "step": 6030
    },
    {
      "epoch": 1.7,
      "grad_norm": 2.1366449022759086,
      "learning_rate": 8.300896149713334e-06,
      "loss": 0.4507,
      "step": 6031
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.6245409177584875,
      "learning_rate": 8.297891655381375e-06,
      "loss": 0.3237,
      "step": 6032
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.7164759436278134,
      "learning_rate": 8.294887319274467e-06,
      "loss": 0.3047,
      "step": 6033
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.7319422822474053,
      "learning_rate": 8.291883141671888e-06,
      "loss": 0.3408,
      "step": 6034
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.8706379720694253,
      "learning_rate": 8.288879122852897e-06,
      "loss": 0.397,
      "step": 6035
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.7698876252959428,
      "learning_rate": 8.285875263096746e-06,
      "loss": 0.2988,
      "step": 6036
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.820868748707837,
      "learning_rate": 8.282871562682673e-06,
      "loss": 0.3867,
      "step": 6037
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.7235885141226464,
      "learning_rate": 8.27986802188989e-06,
      "loss": 0.3298,
      "step": 6038
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.7500104207327474,
      "learning_rate": 8.276864640997602e-06,
      "loss": 0.3467,
      "step": 6039
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.7785652784734507,
      "learning_rate": 8.273861420285e-06,
      "loss": 0.3367,
      "step": 6040
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.8278092900343925,
      "learning_rate": 8.27085836003126e-06,
      "loss": 0.3657,
      "step": 6041
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.717636563177335,
      "learning_rate": 8.267855460515536e-06,
      "loss": 0.3301,
      "step": 6042
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.7480240060967356,
      "learning_rate": 8.264852722016974e-06,
      "loss": 0.3955,
      "step": 6043
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.7613922910895932,
      "learning_rate": 8.261850144814707e-06,
      "loss": 0.3169,
      "step": 6044
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.654208501737968,
      "learning_rate": 8.258847729187845e-06,
      "loss": 0.2747,
      "step": 6045
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.7403207443236925,
      "learning_rate": 8.255845475415493e-06,
      "loss": 0.3093,
      "step": 6046
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.739816951267828,
      "learning_rate": 8.252843383776731e-06,
      "loss": 0.3647,
      "step": 6047
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.9421589377947424,
      "learning_rate": 8.249841454550626e-06,
      "loss": 0.4214,
      "step": 6048
    },
    {
      "epoch": 1.71,
      "grad_norm": 2.1713107999805334,
      "learning_rate": 8.246839688016235e-06,
      "loss": 0.4463,
      "step": 6049
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.9990966924456635,
      "learning_rate": 8.243838084452603e-06,
      "loss": 0.4282,
      "step": 6050
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.8059296810382754,
      "learning_rate": 8.240836644138743e-06,
      "loss": 0.3091,
      "step": 6051
    },
    {
      "epoch": 1.71,
      "grad_norm": 2.0046815872288786,
      "learning_rate": 8.237835367353668e-06,
      "loss": 0.4194,
      "step": 6052
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.7563059725726382,
      "learning_rate": 8.234834254376375e-06,
      "loss": 0.3569,
      "step": 6053
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.814328018936536,
      "learning_rate": 8.231833305485841e-06,
      "loss": 0.3799,
      "step": 6054
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.978053572201426,
      "learning_rate": 8.228832520961023e-06,
      "loss": 0.436,
      "step": 6055
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.8374617281563066,
      "learning_rate": 8.225831901080874e-06,
      "loss": 0.3486,
      "step": 6056
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.6667072920173267,
      "learning_rate": 8.222831446124327e-06,
      "loss": 0.3066,
      "step": 6057
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.683783072274359,
      "learning_rate": 8.2198311563703e-06,
      "loss": 0.2991,
      "step": 6058
    },
    {
      "epoch": 1.71,
      "grad_norm": 2.0976258404078023,
      "learning_rate": 8.216831032097689e-06,
      "loss": 0.4639,
      "step": 6059
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.7275183421800113,
      "learning_rate": 8.213831073585385e-06,
      "loss": 0.3293,
      "step": 6060
    },
    {
      "epoch": 1.71,
      "grad_norm": 2.0964433839384773,
      "learning_rate": 8.210831281112257e-06,
      "loss": 0.4028,
      "step": 6061
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.8874968541955643,
      "learning_rate": 8.207831654957162e-06,
      "loss": 0.3518,
      "step": 6062
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.9065419294601718,
      "learning_rate": 8.204832195398941e-06,
      "loss": 0.4629,
      "step": 6063
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.85678040458153,
      "learning_rate": 8.201832902716416e-06,
      "loss": 0.3584,
      "step": 6064
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.6067763831348612,
      "learning_rate": 8.198833777188396e-06,
      "loss": 0.3145,
      "step": 6065
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.8428501298014666,
      "learning_rate": 8.195834819093677e-06,
      "loss": 0.3477,
      "step": 6066
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.862326438347995,
      "learning_rate": 8.192836028711036e-06,
      "loss": 0.3989,
      "step": 6067
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.6486465069431409,
      "learning_rate": 8.189837406319233e-06,
      "loss": 0.3054,
      "step": 6068
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.9146800066847995,
      "learning_rate": 8.186838952197019e-06,
      "loss": 0.3853,
      "step": 6069
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.6299847665738887,
      "learning_rate": 8.183840666623123e-06,
      "loss": 0.2705,
      "step": 6070
    },
    {
      "epoch": 1.71,
      "grad_norm": 2.419444586758357,
      "learning_rate": 8.18084254987626e-06,
      "loss": 0.4321,
      "step": 6071
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.6016361381035122,
      "learning_rate": 8.177844602235128e-06,
      "loss": 0.2808,
      "step": 6072
    },
    {
      "epoch": 1.72,
      "grad_norm": 2.249973292882878,
      "learning_rate": 8.174846823978412e-06,
      "loss": 0.501,
      "step": 6073
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.84281083802159,
      "learning_rate": 8.171849215384786e-06,
      "loss": 0.3518,
      "step": 6074
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.8115057548022164,
      "learning_rate": 8.168851776732897e-06,
      "loss": 0.3093,
      "step": 6075
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.7961681669755252,
      "learning_rate": 8.16585450830138e-06,
      "loss": 0.373,
      "step": 6076
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.8744126360111828,
      "learning_rate": 8.162857410368859e-06,
      "loss": 0.3413,
      "step": 6077
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.7156737541420586,
      "learning_rate": 8.159860483213938e-06,
      "loss": 0.2896,
      "step": 6078
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.8122953103706305,
      "learning_rate": 8.15686372711521e-06,
      "loss": 0.2944,
      "step": 6079
    },
    {
      "epoch": 1.72,
      "grad_norm": 2.0282064044356827,
      "learning_rate": 8.153867142351242e-06,
      "loss": 0.3999,
      "step": 6080
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.6293383457760044,
      "learning_rate": 8.150870729200595e-06,
      "loss": 0.3413,
      "step": 6081
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.6955421249806484,
      "learning_rate": 8.147874487941809e-06,
      "loss": 0.3398,
      "step": 6082
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.822321591502878,
      "learning_rate": 8.144878418853412e-06,
      "loss": 0.2993,
      "step": 6083
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.771148968744845,
      "learning_rate": 8.141882522213913e-06,
      "loss": 0.3416,
      "step": 6084
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.8136783855368377,
      "learning_rate": 8.1388867983018e-06,
      "loss": 0.3464,
      "step": 6085
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.8739611535921852,
      "learning_rate": 8.135891247395554e-06,
      "loss": 0.3491,
      "step": 6086
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.9016339760808387,
      "learning_rate": 8.132895869773638e-06,
      "loss": 0.373,
      "step": 6087
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.668418401881678,
      "learning_rate": 8.129900665714498e-06,
      "loss": 0.2842,
      "step": 6088
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.529299699153575,
      "learning_rate": 8.126905635496557e-06,
      "loss": 0.2534,
      "step": 6089
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.8492203895295263,
      "learning_rate": 8.123910779398233e-06,
      "loss": 0.3394,
      "step": 6090
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.7985187439969113,
      "learning_rate": 8.120916097697918e-06,
      "loss": 0.3455,
      "step": 6091
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.7413824690156836,
      "learning_rate": 8.117921590674002e-06,
      "loss": 0.3662,
      "step": 6092
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.848005852503662,
      "learning_rate": 8.114927258604837e-06,
      "loss": 0.3813,
      "step": 6093
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.7602320053994558,
      "learning_rate": 8.111933101768779e-06,
      "loss": 0.2754,
      "step": 6094
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.8777119106999405,
      "learning_rate": 8.108939120444154e-06,
      "loss": 0.3853,
      "step": 6095
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.5832660583589064,
      "learning_rate": 8.105945314909287e-06,
      "loss": 0.2793,
      "step": 6096
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.7124230748674016,
      "learning_rate": 8.102951685442466e-06,
      "loss": 0.2908,
      "step": 6097
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.657987079809498,
      "learning_rate": 8.099958232321978e-06,
      "loss": 0.2998,
      "step": 6098
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.7006870271988286,
      "learning_rate": 8.09696495582609e-06,
      "loss": 0.312,
      "step": 6099
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.7928427943891079,
      "learning_rate": 8.093971856233051e-06,
      "loss": 0.3369,
      "step": 6100
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.7402419551610084,
      "learning_rate": 8.0909789338211e-06,
      "loss": 0.3237,
      "step": 6101
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.8674448828202668,
      "learning_rate": 8.087986188868441e-06,
      "loss": 0.3306,
      "step": 6102
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.6148982547350204,
      "learning_rate": 8.084993621653283e-06,
      "loss": 0.2737,
      "step": 6103
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.7159195173591286,
      "learning_rate": 8.08200123245381e-06,
      "loss": 0.2712,
      "step": 6104
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.812748919441024,
      "learning_rate": 8.079009021548193e-06,
      "loss": 0.3955,
      "step": 6105
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.6282901922523996,
      "learning_rate": 8.076016989214572e-06,
      "loss": 0.2788,
      "step": 6106
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.9092689654851227,
      "learning_rate": 8.07302513573109e-06,
      "loss": 0.3994,
      "step": 6107
    },
    {
      "epoch": 1.72,
      "grad_norm": 2.1494683547162476,
      "learning_rate": 8.070033461375857e-06,
      "loss": 0.4058,
      "step": 6108
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.6649575204679585,
      "learning_rate": 8.067041966426984e-06,
      "loss": 0.313,
      "step": 6109
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.9019899596842391,
      "learning_rate": 8.064050651162546e-06,
      "loss": 0.3748,
      "step": 6110
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.8810859872865753,
      "learning_rate": 8.061059515860616e-06,
      "loss": 0.3442,
      "step": 6111
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.9712990143262201,
      "learning_rate": 8.058068560799241e-06,
      "loss": 0.3662,
      "step": 6112
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.7856094614976632,
      "learning_rate": 8.05507778625646e-06,
      "loss": 0.3452,
      "step": 6113
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.9711216038844912,
      "learning_rate": 8.052087192510285e-06,
      "loss": 0.3604,
      "step": 6114
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.9622717634863596,
      "learning_rate": 8.04909677983872e-06,
      "loss": 0.375,
      "step": 6115
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.7036798313828057,
      "learning_rate": 8.046106548519743e-06,
      "loss": 0.3032,
      "step": 6116
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.7946501220949653,
      "learning_rate": 8.043116498831328e-06,
      "loss": 0.3621,
      "step": 6117
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.884086394631177,
      "learning_rate": 8.040126631051425e-06,
      "loss": 0.3955,
      "step": 6118
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.773738631270828,
      "learning_rate": 8.037136945457959e-06,
      "loss": 0.29,
      "step": 6119
    },
    {
      "epoch": 1.73,
      "grad_norm": 2.017639340531585,
      "learning_rate": 8.034147442328852e-06,
      "loss": 0.3257,
      "step": 6120
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.6483473109806375,
      "learning_rate": 8.031158121942001e-06,
      "loss": 0.3198,
      "step": 6121
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.9399740085230368,
      "learning_rate": 8.028168984575292e-06,
      "loss": 0.3459,
      "step": 6122
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.822895824074695,
      "learning_rate": 8.025180030506584e-06,
      "loss": 0.3379,
      "step": 6123
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.8738487883336157,
      "learning_rate": 8.022191260013727e-06,
      "loss": 0.3271,
      "step": 6124
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.923349482604321,
      "learning_rate": 8.019202673374554e-06,
      "loss": 0.3984,
      "step": 6125
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.9633458856042454,
      "learning_rate": 8.01621427086688e-06,
      "loss": 0.3872,
      "step": 6126
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.6232899958197378,
      "learning_rate": 8.013226052768498e-06,
      "loss": 0.3184,
      "step": 6127
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.7489580467971657,
      "learning_rate": 8.010238019357185e-06,
      "loss": 0.3579,
      "step": 6128
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.8320909650740427,
      "learning_rate": 8.00725017091071e-06,
      "loss": 0.4058,
      "step": 6129
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.7279134863358103,
      "learning_rate": 8.004262507706819e-06,
      "loss": 0.2969,
      "step": 6130
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.7257709256523102,
      "learning_rate": 8.001275030023234e-06,
      "loss": 0.3672,
      "step": 6131
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.6234727838213847,
      "learning_rate": 7.998287738137669e-06,
      "loss": 0.3279,
      "step": 6132
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.8496947151308343,
      "learning_rate": 7.995300632327816e-06,
      "loss": 0.3506,
      "step": 6133
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.682400359511411,
      "learning_rate": 7.99231371287135e-06,
      "loss": 0.3008,
      "step": 6134
    },
    {
      "epoch": 1.73,
      "grad_norm": 2.0911619601880247,
      "learning_rate": 7.989326980045937e-06,
      "loss": 0.4707,
      "step": 6135
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.79900348905239,
      "learning_rate": 7.98634043412921e-06,
      "loss": 0.2942,
      "step": 6136
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.6268031188392003,
      "learning_rate": 7.983354075398797e-06,
      "loss": 0.3145,
      "step": 6137
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.7218175257785213,
      "learning_rate": 7.980367904132303e-06,
      "loss": 0.3525,
      "step": 6138
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.7026896792034723,
      "learning_rate": 7.977381920607324e-06,
      "loss": 0.3481,
      "step": 6139
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.775208411000509,
      "learning_rate": 7.97439612510142e-06,
      "loss": 0.3201,
      "step": 6140
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.7708836782570663,
      "learning_rate": 7.971410517892155e-06,
      "loss": 0.3735,
      "step": 6141
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.792974091243189,
      "learning_rate": 7.968425099257062e-06,
      "loss": 0.3057,
      "step": 6142
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.681035640364061,
      "learning_rate": 7.965439869473664e-06,
      "loss": 0.3064,
      "step": 6143
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.4220255264556265,
      "learning_rate": 7.96245482881946e-06,
      "loss": 0.2812,
      "step": 6144
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.942856217691942,
      "learning_rate": 7.95946997757193e-06,
      "loss": 0.366,
      "step": 6145
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.8445851849225723,
      "learning_rate": 7.956485316008545e-06,
      "loss": 0.2959,
      "step": 6146
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.827174454416937,
      "learning_rate": 7.953500844406758e-06,
      "loss": 0.3213,
      "step": 6147
    },
    {
      "epoch": 1.74,
      "grad_norm": 2.1077491774137673,
      "learning_rate": 7.950516563043994e-06,
      "loss": 0.4385,
      "step": 6148
    },
    {
      "epoch": 1.74,
      "grad_norm": 2.0087072123235625,
      "learning_rate": 7.947532472197668e-06,
      "loss": 0.3931,
      "step": 6149
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.6753210970114556,
      "learning_rate": 7.944548572145178e-06,
      "loss": 0.2578,
      "step": 6150
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.8318969765157433,
      "learning_rate": 7.941564863163899e-06,
      "loss": 0.3154,
      "step": 6151
    },
    {
      "epoch": 1.74,
      "grad_norm": 2.030754284371302,
      "learning_rate": 7.938581345531197e-06,
      "loss": 0.4404,
      "step": 6152
    },
    {
      "epoch": 1.74,
      "grad_norm": 2.0193276462763103,
      "learning_rate": 7.935598019524406e-06,
      "loss": 0.4082,
      "step": 6153
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.7841127948871363,
      "learning_rate": 7.932614885420859e-06,
      "loss": 0.3301,
      "step": 6154
    },
    {
      "epoch": 1.74,
      "grad_norm": 2.005870935965378,
      "learning_rate": 7.929631943497858e-06,
      "loss": 0.3794,
      "step": 6155
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.8722649686925597,
      "learning_rate": 7.926649194032699e-06,
      "loss": 0.3333,
      "step": 6156
    },
    {
      "epoch": 1.74,
      "grad_norm": 2.213772307288036,
      "learning_rate": 7.923666637302643e-06,
      "loss": 0.4258,
      "step": 6157
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.7446312744321355,
      "learning_rate": 7.920684273584949e-06,
      "loss": 0.3149,
      "step": 6158
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.881118966958096,
      "learning_rate": 7.91770210315685e-06,
      "loss": 0.3762,
      "step": 6159
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.706449704188258,
      "learning_rate": 7.914720126295572e-06,
      "loss": 0.3306,
      "step": 6160
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.8140919636655986,
      "learning_rate": 7.911738343278303e-06,
      "loss": 0.3726,
      "step": 6161
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.8423372968978236,
      "learning_rate": 7.908756754382228e-06,
      "loss": 0.3101,
      "step": 6162
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.6924011293395131,
      "learning_rate": 7.905775359884514e-06,
      "loss": 0.2788,
      "step": 6163
    },
    {
      "epoch": 1.74,
      "grad_norm": 2.2856171740968856,
      "learning_rate": 7.902794160062303e-06,
      "loss": 0.3853,
      "step": 6164
    },
    {
      "epoch": 1.74,
      "grad_norm": 2.0886448437229883,
      "learning_rate": 7.899813155192723e-06,
      "loss": 0.3501,
      "step": 6165
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.9539521918026437,
      "learning_rate": 7.896832345552882e-06,
      "loss": 0.3452,
      "step": 6166
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.6810645055403566,
      "learning_rate": 7.893851731419872e-06,
      "loss": 0.3806,
      "step": 6167
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.89458733533898,
      "learning_rate": 7.890871313070768e-06,
      "loss": 0.3032,
      "step": 6168
    },
    {
      "epoch": 1.74,
      "grad_norm": 2.0513235276392376,
      "learning_rate": 7.88789109078262e-06,
      "loss": 0.3613,
      "step": 6169
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.9674273393335466,
      "learning_rate": 7.884911064832466e-06,
      "loss": 0.3911,
      "step": 6170
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.7814296592983794,
      "learning_rate": 7.881931235497324e-06,
      "loss": 0.3054,
      "step": 6171
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.6139361668952883,
      "learning_rate": 7.878951603054195e-06,
      "loss": 0.2751,
      "step": 6172
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.9872578972704007,
      "learning_rate": 7.875972167780063e-06,
      "loss": 0.3892,
      "step": 6173
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.9711222091124763,
      "learning_rate": 7.872992929951886e-06,
      "loss": 0.3726,
      "step": 6174
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.7235886552665112,
      "learning_rate": 7.870013889846608e-06,
      "loss": 0.3169,
      "step": 6175
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.9734534653945754,
      "learning_rate": 7.867035047741159e-06,
      "loss": 0.3896,
      "step": 6176
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.9040832103427485,
      "learning_rate": 7.86405640391245e-06,
      "loss": 0.3369,
      "step": 6177
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.5430313164951284,
      "learning_rate": 7.861077958637365e-06,
      "loss": 0.3706,
      "step": 6178
    },
    {
      "epoch": 1.74,
      "grad_norm": 2.101257047962753,
      "learning_rate": 7.858099712192774e-06,
      "loss": 0.3706,
      "step": 6179
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.912022991983066,
      "learning_rate": 7.855121664855535e-06,
      "loss": 0.3667,
      "step": 6180
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.8489035563930836,
      "learning_rate": 7.852143816902483e-06,
      "loss": 0.3333,
      "step": 6181
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.7461727780802083,
      "learning_rate": 7.849166168610424e-06,
      "loss": 0.3264,
      "step": 6182
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.8613490733447418,
      "learning_rate": 7.846188720256162e-06,
      "loss": 0.3679,
      "step": 6183
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.8628549190456163,
      "learning_rate": 7.843211472116476e-06,
      "loss": 0.3687,
      "step": 6184
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.9217997797123945,
      "learning_rate": 7.84023442446813e-06,
      "loss": 0.3914,
      "step": 6185
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.4839524245673206,
      "learning_rate": 7.837257577587853e-06,
      "loss": 0.2769,
      "step": 6186
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.5206253851639508,
      "learning_rate": 7.834280931752375e-06,
      "loss": 0.2607,
      "step": 6187
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.9172322563596038,
      "learning_rate": 7.8313044872384e-06,
      "loss": 0.4292,
      "step": 6188
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.7575732413156961,
      "learning_rate": 7.828328244322612e-06,
      "loss": 0.4126,
      "step": 6189
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.7970775019459444,
      "learning_rate": 7.825352203281682e-06,
      "loss": 0.3159,
      "step": 6190
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.6392468858122209,
      "learning_rate": 7.822376364392248e-06,
      "loss": 0.3079,
      "step": 6191
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.6323571562966617,
      "learning_rate": 7.819400727930947e-06,
      "loss": 0.2622,
      "step": 6192
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.903752663486083,
      "learning_rate": 7.816425294174385e-06,
      "loss": 0.3853,
      "step": 6193
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.9343079257273106,
      "learning_rate": 7.813450063399158e-06,
      "loss": 0.3716,
      "step": 6194
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.9615241353343402,
      "learning_rate": 7.81047503588183e-06,
      "loss": 0.3975,
      "step": 6195
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.8291233919497372,
      "learning_rate": 7.807500211898959e-06,
      "loss": 0.3521,
      "step": 6196
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.739749939768596,
      "learning_rate": 7.804525591727081e-06,
      "loss": 0.3188,
      "step": 6197
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.6715567289600628,
      "learning_rate": 7.801551175642715e-06,
      "loss": 0.283,
      "step": 6198
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.5141372659650503,
      "learning_rate": 7.798576963922347e-06,
      "loss": 0.2783,
      "step": 6199
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.834656807552232,
      "learning_rate": 7.79560295684246e-06,
      "loss": 0.3191,
      "step": 6200
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.9042184327375253,
      "learning_rate": 7.792629154679514e-06,
      "loss": 0.3281,
      "step": 6201
    },
    {
      "epoch": 1.75,
      "grad_norm": 2.005236299412832,
      "learning_rate": 7.789655557709951e-06,
      "loss": 0.4033,
      "step": 6202
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.6599404244769547,
      "learning_rate": 7.786682166210184e-06,
      "loss": 0.3262,
      "step": 6203
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.837250759029627,
      "learning_rate": 7.78370898045662e-06,
      "loss": 0.3384,
      "step": 6204
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.873416730549372,
      "learning_rate": 7.780736000725636e-06,
      "loss": 0.3367,
      "step": 6205
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.6949635514870036,
      "learning_rate": 7.7777632272936e-06,
      "loss": 0.313,
      "step": 6206
    },
    {
      "epoch": 1.75,
      "grad_norm": 2.0679116118772365,
      "learning_rate": 7.774790660436857e-06,
      "loss": 0.3477,
      "step": 6207
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.9683320146135794,
      "learning_rate": 7.771818300431729e-06,
      "loss": 0.395,
      "step": 6208
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.8473996163676045,
      "learning_rate": 7.768846147554517e-06,
      "loss": 0.3352,
      "step": 6209
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.6481914935434339,
      "learning_rate": 7.765874202081516e-06,
      "loss": 0.2896,
      "step": 6210
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.8401743895148437,
      "learning_rate": 7.76290246428899e-06,
      "loss": 0.3652,
      "step": 6211
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.8167282154592492,
      "learning_rate": 7.75993093445318e-06,
      "loss": 0.3911,
      "step": 6212
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.7425747177671222,
      "learning_rate": 7.75695961285032e-06,
      "loss": 0.3237,
      "step": 6213
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.7138763422745649,
      "learning_rate": 7.753988499756621e-06,
      "loss": 0.2524,
      "step": 6214
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.8151023723312034,
      "learning_rate": 7.75101759544827e-06,
      "loss": 0.2881,
      "step": 6215
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.7953565425079867,
      "learning_rate": 7.748046900201438e-06,
      "loss": 0.333,
      "step": 6216
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.9775944285944203,
      "learning_rate": 7.745076414292272e-06,
      "loss": 0.3882,
      "step": 6217
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.685292904334649,
      "learning_rate": 7.742106137996905e-06,
      "loss": 0.3274,
      "step": 6218
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.873571987429675,
      "learning_rate": 7.739136071591455e-06,
      "loss": 0.3701,
      "step": 6219
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.8800336475324504,
      "learning_rate": 7.736166215352004e-06,
      "loss": 0.325,
      "step": 6220
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.5579067174168517,
      "learning_rate": 7.733196569554632e-06,
      "loss": 0.2832,
      "step": 6221
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.9411153952810962,
      "learning_rate": 7.730227134475388e-06,
      "loss": 0.3232,
      "step": 6222
    },
    {
      "epoch": 1.76,
      "grad_norm": 2.019839016967519,
      "learning_rate": 7.727257910390309e-06,
      "loss": 0.4419,
      "step": 6223
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.846146790276831,
      "learning_rate": 7.724288897575409e-06,
      "loss": 0.3091,
      "step": 6224
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.6343226529240613,
      "learning_rate": 7.721320096306679e-06,
      "loss": 0.3267,
      "step": 6225
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.835222163986206,
      "learning_rate": 7.718351506860092e-06,
      "loss": 0.3335,
      "step": 6226
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.6788076712741153,
      "learning_rate": 7.71538312951161e-06,
      "loss": 0.2861,
      "step": 6227
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.829070222208214,
      "learning_rate": 7.712414964537167e-06,
      "loss": 0.4194,
      "step": 6228
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.6234999196189919,
      "learning_rate": 7.70944701221267e-06,
      "loss": 0.3008,
      "step": 6229
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.9958036696221455,
      "learning_rate": 7.706479272814024e-06,
      "loss": 0.3203,
      "step": 6230
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.8070512211035767,
      "learning_rate": 7.703511746617098e-06,
      "loss": 0.3044,
      "step": 6231
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.9307813121047213,
      "learning_rate": 7.700544433897757e-06,
      "loss": 0.3604,
      "step": 6232
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.7400546260419285,
      "learning_rate": 7.697577334931828e-06,
      "loss": 0.2681,
      "step": 6233
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.832684670554611,
      "learning_rate": 7.694610449995133e-06,
      "loss": 0.3643,
      "step": 6234
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.876791027372308,
      "learning_rate": 7.691643779363465e-06,
      "loss": 0.3308,
      "step": 6235
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.9151420234199756,
      "learning_rate": 7.688677323312608e-06,
      "loss": 0.3167,
      "step": 6236
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.499198256069341,
      "learning_rate": 7.685711082118306e-06,
      "loss": 0.241,
      "step": 6237
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.6009592530991512,
      "learning_rate": 7.682745056056307e-06,
      "loss": 0.2468,
      "step": 6238
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.8269210707274726,
      "learning_rate": 7.679779245402321e-06,
      "loss": 0.3599,
      "step": 6239
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.703130219932522,
      "learning_rate": 7.67681365043205e-06,
      "loss": 0.2869,
      "step": 6240
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.6613232236370694,
      "learning_rate": 7.673848271421166e-06,
      "loss": 0.2803,
      "step": 6241
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.6986803266856718,
      "learning_rate": 7.670883108645326e-06,
      "loss": 0.3252,
      "step": 6242
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.7006985178793748,
      "learning_rate": 7.667918162380166e-06,
      "loss": 0.3208,
      "step": 6243
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.7146343419976775,
      "learning_rate": 7.664953432901306e-06,
      "loss": 0.2749,
      "step": 6244
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.9687380904200762,
      "learning_rate": 7.661988920484344e-06,
      "loss": 0.3916,
      "step": 6245
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.7094241379340913,
      "learning_rate": 7.659024625404845e-06,
      "loss": 0.303,
      "step": 6246
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.796326161862449,
      "learning_rate": 7.656060547938375e-06,
      "loss": 0.2725,
      "step": 6247
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.6175359055943481,
      "learning_rate": 7.653096688360465e-06,
      "loss": 0.3081,
      "step": 6248
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.8301257354778053,
      "learning_rate": 7.650133046946633e-06,
      "loss": 0.3682,
      "step": 6249
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.7957369945515977,
      "learning_rate": 7.64716962397237e-06,
      "loss": 0.3569,
      "step": 6250
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.7339420169731232,
      "learning_rate": 7.644206419713155e-06,
      "loss": 0.3811,
      "step": 6251
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.6382408393059755,
      "learning_rate": 7.641243434444439e-06,
      "loss": 0.3135,
      "step": 6252
    },
    {
      "epoch": 1.77,
      "grad_norm": 2.05578769582899,
      "learning_rate": 7.63828066844166e-06,
      "loss": 0.3481,
      "step": 6253
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.6578617490272587,
      "learning_rate": 7.635318121980228e-06,
      "loss": 0.3589,
      "step": 6254
    },
    {
      "epoch": 1.77,
      "grad_norm": 2.1691610114320246,
      "learning_rate": 7.632355795335533e-06,
      "loss": 0.3804,
      "step": 6255
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.8968375434286713,
      "learning_rate": 7.629393688782954e-06,
      "loss": 0.3569,
      "step": 6256
    },
    {
      "epoch": 1.77,
      "grad_norm": 2.131505562861137,
      "learning_rate": 7.6264318025978455e-06,
      "loss": 0.4556,
      "step": 6257
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.5411525545731628,
      "learning_rate": 7.623470137055533e-06,
      "loss": 0.2742,
      "step": 6258
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.6755409563293675,
      "learning_rate": 7.620508692431327e-06,
      "loss": 0.3242,
      "step": 6259
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.8183276937899007,
      "learning_rate": 7.617547469000524e-06,
      "loss": 0.3394,
      "step": 6260
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.6338287600092878,
      "learning_rate": 7.6145864670383884e-06,
      "loss": 0.2561,
      "step": 6261
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.7856125435035124,
      "learning_rate": 7.611625686820177e-06,
      "loss": 0.292,
      "step": 6262
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.7839367333865763,
      "learning_rate": 7.608665128621111e-06,
      "loss": 0.374,
      "step": 6263
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.8611425460845268,
      "learning_rate": 7.6057047927164055e-06,
      "loss": 0.3503,
      "step": 6264
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.8333105822666413,
      "learning_rate": 7.602744679381241e-06,
      "loss": 0.3774,
      "step": 6265
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.8011201885142178,
      "learning_rate": 7.599784788890794e-06,
      "loss": 0.2961,
      "step": 6266
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.8058572869532994,
      "learning_rate": 7.596825121520202e-06,
      "loss": 0.3762,
      "step": 6267
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.9791702244862814,
      "learning_rate": 7.593865677544594e-06,
      "loss": 0.4106,
      "step": 6268
    },
    {
      "epoch": 1.77,
      "grad_norm": 2.028766791625086,
      "learning_rate": 7.590906457239073e-06,
      "loss": 0.4082,
      "step": 6269
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.5224408689194087,
      "learning_rate": 7.587947460878731e-06,
      "loss": 0.2615,
      "step": 6270
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.6601049467497475,
      "learning_rate": 7.584988688738622e-06,
      "loss": 0.2979,
      "step": 6271
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.7187901979235178,
      "learning_rate": 7.582030141093788e-06,
      "loss": 0.2708,
      "step": 6272
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.8875651747888846,
      "learning_rate": 7.5790718182192545e-06,
      "loss": 0.3408,
      "step": 6273
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.659586209074065,
      "learning_rate": 7.576113720390027e-06,
      "loss": 0.3269,
      "step": 6274
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.6427111352004458,
      "learning_rate": 7.573155847881076e-06,
      "loss": 0.311,
      "step": 6275
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.71107082381093,
      "learning_rate": 7.570198200967363e-06,
      "loss": 0.4331,
      "step": 6276
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.5758376397287976,
      "learning_rate": 7.567240779923827e-06,
      "loss": 0.2803,
      "step": 6277
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.737609819057367,
      "learning_rate": 7.564283585025383e-06,
      "loss": 0.3247,
      "step": 6278
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.6776337875659952,
      "learning_rate": 7.561326616546932e-06,
      "loss": 0.2832,
      "step": 6279
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.736275732656939,
      "learning_rate": 7.5583698747633394e-06,
      "loss": 0.2815,
      "step": 6280
    },
    {
      "epoch": 1.77,
      "grad_norm": 2.026941819202746,
      "learning_rate": 7.555413359949468e-06,
      "loss": 0.3787,
      "step": 6281
    },
    {
      "epoch": 1.77,
      "grad_norm": 2.054000398154599,
      "learning_rate": 7.552457072380143e-06,
      "loss": 0.4204,
      "step": 6282
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.7385476515948435,
      "learning_rate": 7.549501012330184e-06,
      "loss": 0.3113,
      "step": 6283
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.6582487999604623,
      "learning_rate": 7.546545180074374e-06,
      "loss": 0.3372,
      "step": 6284
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.940486550999383,
      "learning_rate": 7.543589575887482e-06,
      "loss": 0.4316,
      "step": 6285
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.6957493425988877,
      "learning_rate": 7.54063420004426e-06,
      "loss": 0.3286,
      "step": 6286
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.76847399088716,
      "learning_rate": 7.5376790528194354e-06,
      "loss": 0.396,
      "step": 6287
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.805671483585485,
      "learning_rate": 7.534724134487709e-06,
      "loss": 0.3081,
      "step": 6288
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.9806533420584593,
      "learning_rate": 7.531769445323767e-06,
      "loss": 0.3511,
      "step": 6289
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.6508488828606966,
      "learning_rate": 7.528814985602273e-06,
      "loss": 0.3394,
      "step": 6290
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.701397108636351,
      "learning_rate": 7.52586075559787e-06,
      "loss": 0.3308,
      "step": 6291
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.9976342863876315,
      "learning_rate": 7.522906755585171e-06,
      "loss": 0.374,
      "step": 6292
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.6958414456536801,
      "learning_rate": 7.5199529858387795e-06,
      "loss": 0.2917,
      "step": 6293
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.8918298877703366,
      "learning_rate": 7.516999446633277e-06,
      "loss": 0.3296,
      "step": 6294
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.7786277382295088,
      "learning_rate": 7.514046138243211e-06,
      "loss": 0.3127,
      "step": 6295
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.8015843312491238,
      "learning_rate": 7.511093060943125e-06,
      "loss": 0.3647,
      "step": 6296
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.7186702870609532,
      "learning_rate": 7.508140215007526e-06,
      "loss": 0.3171,
      "step": 6297
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.5656110200845468,
      "learning_rate": 7.505187600710903e-06,
      "loss": 0.2561,
      "step": 6298
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.7404408318221636,
      "learning_rate": 7.50223521832773e-06,
      "loss": 0.3525,
      "step": 6299
    },
    {
      "epoch": 1.78,
      "grad_norm": 2.1975434502100186,
      "learning_rate": 7.49928306813246e-06,
      "loss": 0.374,
      "step": 6300
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.8321823054078934,
      "learning_rate": 7.496331150399512e-06,
      "loss": 0.2798,
      "step": 6301
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.9388362584666583,
      "learning_rate": 7.493379465403292e-06,
      "loss": 0.4453,
      "step": 6302
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.5808441097277275,
      "learning_rate": 7.490428013418187e-06,
      "loss": 0.3208,
      "step": 6303
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.897624800109671,
      "learning_rate": 7.4874767947185586e-06,
      "loss": 0.322,
      "step": 6304
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.6545911816093957,
      "learning_rate": 7.4845258095787454e-06,
      "loss": 0.3057,
      "step": 6305
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.8296892318310418,
      "learning_rate": 7.481575058273063e-06,
      "loss": 0.3203,
      "step": 6306
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.7940301041426479,
      "learning_rate": 7.478624541075814e-06,
      "loss": 0.3184,
      "step": 6307
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.5267437618323672,
      "learning_rate": 7.475674258261274e-06,
      "loss": 0.3098,
      "step": 6308
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.7201132763053777,
      "learning_rate": 7.472724210103687e-06,
      "loss": 0.304,
      "step": 6309
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.7811534330439804,
      "learning_rate": 7.4697743968772906e-06,
      "loss": 0.3062,
      "step": 6310
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.7032420004604092,
      "learning_rate": 7.466824818856296e-06,
      "loss": 0.3262,
      "step": 6311
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.7558106536510594,
      "learning_rate": 7.463875476314888e-06,
      "loss": 0.3271,
      "step": 6312
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.8489870388707745,
      "learning_rate": 7.460926369527236e-06,
      "loss": 0.3923,
      "step": 6313
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.9314256543031099,
      "learning_rate": 7.45797749876748e-06,
      "loss": 0.3281,
      "step": 6314
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.9224158057938512,
      "learning_rate": 7.455028864309742e-06,
      "loss": 0.3364,
      "step": 6315
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.958092517913011,
      "learning_rate": 7.4520804664281195e-06,
      "loss": 0.4092,
      "step": 6316
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.8270566524934613,
      "learning_rate": 7.449132305396701e-06,
      "loss": 0.312,
      "step": 6317
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.8355113255587947,
      "learning_rate": 7.446184381489533e-06,
      "loss": 0.3313,
      "step": 6318
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.9412446553713754,
      "learning_rate": 7.443236694980649e-06,
      "loss": 0.3906,
      "step": 6319
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.992081280454665,
      "learning_rate": 7.440289246144067e-06,
      "loss": 0.3228,
      "step": 6320
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.5535538911650788,
      "learning_rate": 7.437342035253775e-06,
      "loss": 0.3203,
      "step": 6321
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.7214891082093506,
      "learning_rate": 7.434395062583735e-06,
      "loss": 0.3147,
      "step": 6322
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.8756707652220939,
      "learning_rate": 7.431448328407896e-06,
      "loss": 0.3821,
      "step": 6323
    },
    {
      "epoch": 1.79,
      "grad_norm": 2.1756783732654688,
      "learning_rate": 7.428501833000186e-06,
      "loss": 0.4087,
      "step": 6324
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.8253588925399815,
      "learning_rate": 7.4255555766345025e-06,
      "loss": 0.3418,
      "step": 6325
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.6953197830720925,
      "learning_rate": 7.422609559584719e-06,
      "loss": 0.3057,
      "step": 6326
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.7507988037835345,
      "learning_rate": 7.4196637821247e-06,
      "loss": 0.377,
      "step": 6327
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.808077415161811,
      "learning_rate": 7.416718244528275e-06,
      "loss": 0.3374,
      "step": 6328
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.83432995117331,
      "learning_rate": 7.413772947069262e-06,
      "loss": 0.3232,
      "step": 6329
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.8694402789745934,
      "learning_rate": 7.410827890021444e-06,
      "loss": 0.3647,
      "step": 6330
    },
    {
      "epoch": 1.79,
      "grad_norm": 2.1565217334698215,
      "learning_rate": 7.407883073658591e-06,
      "loss": 0.4087,
      "step": 6331
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.685484629224633,
      "learning_rate": 7.404938498254448e-06,
      "loss": 0.2827,
      "step": 6332
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.9378884045866156,
      "learning_rate": 7.401994164082738e-06,
      "loss": 0.3867,
      "step": 6333
    },
    {
      "epoch": 1.79,
      "grad_norm": 2.1158017539862963,
      "learning_rate": 7.399050071417163e-06,
      "loss": 0.4346,
      "step": 6334
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.8813811447186866,
      "learning_rate": 7.396106220531398e-06,
      "loss": 0.3721,
      "step": 6335
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.766087020215543,
      "learning_rate": 7.3931626116990984e-06,
      "loss": 0.3716,
      "step": 6336
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.6370981473255684,
      "learning_rate": 7.3902192451939e-06,
      "loss": 0.335,
      "step": 6337
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.5812677186768342,
      "learning_rate": 7.3872761212894105e-06,
      "loss": 0.2996,
      "step": 6338
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.7050089652467855,
      "learning_rate": 7.384333240259216e-06,
      "loss": 0.3379,
      "step": 6339
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.9856728278481728,
      "learning_rate": 7.381390602376882e-06,
      "loss": 0.4297,
      "step": 6340
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.7702214821514106,
      "learning_rate": 7.378448207915956e-06,
      "loss": 0.3333,
      "step": 6341
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.6386460543019903,
      "learning_rate": 7.3755060571499555e-06,
      "loss": 0.3032,
      "step": 6342
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.8820879301993025,
      "learning_rate": 7.372564150352373e-06,
      "loss": 0.3521,
      "step": 6343
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.747924646237497,
      "learning_rate": 7.369622487796689e-06,
      "loss": 0.3625,
      "step": 6344
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.7485848236949997,
      "learning_rate": 7.366681069756352e-06,
      "loss": 0.3657,
      "step": 6345
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.784365977527254,
      "learning_rate": 7.363739896504795e-06,
      "loss": 0.3088,
      "step": 6346
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.7826581385518148,
      "learning_rate": 7.360798968315419e-06,
      "loss": 0.3652,
      "step": 6347
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.5628676719673333,
      "learning_rate": 7.357858285461612e-06,
      "loss": 0.2749,
      "step": 6348
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.7130648651955185,
      "learning_rate": 7.35491784821673e-06,
      "loss": 0.3413,
      "step": 6349
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.6819130192764473,
      "learning_rate": 7.351977656854118e-06,
      "loss": 0.311,
      "step": 6350
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.5822837337186997,
      "learning_rate": 7.349037711647089e-06,
      "loss": 0.3052,
      "step": 6351
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.7633795461192106,
      "learning_rate": 7.346098012868928e-06,
      "loss": 0.3511,
      "step": 6352
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.8214165990557645,
      "learning_rate": 7.34315856079291e-06,
      "loss": 0.3594,
      "step": 6353
    },
    {
      "epoch": 1.79,
      "grad_norm": 2.1661417229218185,
      "learning_rate": 7.340219355692284e-06,
      "loss": 0.3706,
      "step": 6354
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.8024628900402002,
      "learning_rate": 7.337280397840272e-06,
      "loss": 0.3491,
      "step": 6355
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.8324078492171114,
      "learning_rate": 7.334341687510069e-06,
      "loss": 0.3804,
      "step": 6356
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.9219141482203648,
      "learning_rate": 7.331403224974859e-06,
      "loss": 0.3491,
      "step": 6357
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.6816033574517493,
      "learning_rate": 7.328465010507791e-06,
      "loss": 0.3633,
      "step": 6358
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.9315648203978477,
      "learning_rate": 7.325527044382004e-06,
      "loss": 0.386,
      "step": 6359
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.6286254988295845,
      "learning_rate": 7.322589326870597e-06,
      "loss": 0.3135,
      "step": 6360
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.8537552155944053,
      "learning_rate": 7.319651858246661e-06,
      "loss": 0.3164,
      "step": 6361
    },
    {
      "epoch": 1.8,
      "grad_norm": 2.021368636404815,
      "learning_rate": 7.316714638783257e-06,
      "loss": 0.3286,
      "step": 6362
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.615363283170116,
      "learning_rate": 7.313777668753426e-06,
      "loss": 0.2778,
      "step": 6363
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.904882116209489,
      "learning_rate": 7.310840948430178e-06,
      "loss": 0.376,
      "step": 6364
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.831206201436456,
      "learning_rate": 7.307904478086507e-06,
      "loss": 0.3491,
      "step": 6365
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.5638798159867655,
      "learning_rate": 7.304968257995384e-06,
      "loss": 0.2583,
      "step": 6366
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.6875323673496585,
      "learning_rate": 7.3020322884297565e-06,
      "loss": 0.3477,
      "step": 6367
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.718849315912924,
      "learning_rate": 7.299096569662549e-06,
      "loss": 0.2822,
      "step": 6368
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.8861953010261336,
      "learning_rate": 7.296161101966653e-06,
      "loss": 0.3306,
      "step": 6369
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.5703936136259626,
      "learning_rate": 7.293225885614948e-06,
      "loss": 0.3008,
      "step": 6370
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.997305529957675,
      "learning_rate": 7.290290920880287e-06,
      "loss": 0.3569,
      "step": 6371
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.7979135971675646,
      "learning_rate": 7.2873562080355045e-06,
      "loss": 0.3345,
      "step": 6372
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.9686589928169915,
      "learning_rate": 7.284421747353396e-06,
      "loss": 0.3999,
      "step": 6373
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.7832868820878085,
      "learning_rate": 7.281487539106752e-06,
      "loss": 0.2993,
      "step": 6374
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.9122953407060816,
      "learning_rate": 7.278553583568326e-06,
      "loss": 0.3652,
      "step": 6375
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.7337750377999765,
      "learning_rate": 7.275619881010861e-06,
      "loss": 0.3618,
      "step": 6376
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.6068737053829472,
      "learning_rate": 7.27268643170706e-06,
      "loss": 0.2927,
      "step": 6377
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.8668027881641733,
      "learning_rate": 7.269753235929617e-06,
      "loss": 0.3447,
      "step": 6378
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.8806694364580778,
      "learning_rate": 7.2668202939511946e-06,
      "loss": 0.3657,
      "step": 6379
    },
    {
      "epoch": 1.8,
      "grad_norm": 2.0341011365022776,
      "learning_rate": 7.263887606044437e-06,
      "loss": 0.4141,
      "step": 6380
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.9816554984330723,
      "learning_rate": 7.260955172481959e-06,
      "loss": 0.439,
      "step": 6381
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.922476727199818,
      "learning_rate": 7.258022993536352e-06,
      "loss": 0.3926,
      "step": 6382
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.8729847900091907,
      "learning_rate": 7.2550910694801905e-06,
      "loss": 0.3438,
      "step": 6383
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.7954021822088209,
      "learning_rate": 7.2521594005860205e-06,
      "loss": 0.3589,
      "step": 6384
    },
    {
      "epoch": 1.8,
      "grad_norm": 2.3230730977111578,
      "learning_rate": 7.249227987126368e-06,
      "loss": 0.5093,
      "step": 6385
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.9360382688572793,
      "learning_rate": 7.246296829373723e-06,
      "loss": 0.3574,
      "step": 6386
    },
    {
      "epoch": 1.8,
      "grad_norm": 2.1235652009361408,
      "learning_rate": 7.24336592760057e-06,
      "loss": 0.396,
      "step": 6387
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.8529878369453405,
      "learning_rate": 7.240435282079352e-06,
      "loss": 0.3677,
      "step": 6388
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.851117426072087,
      "learning_rate": 7.237504893082507e-06,
      "loss": 0.3921,
      "step": 6389
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.7481292187723982,
      "learning_rate": 7.234574760882431e-06,
      "loss": 0.3418,
      "step": 6390
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.8857274628466425,
      "learning_rate": 7.2316448857515076e-06,
      "loss": 0.3665,
      "step": 6391
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.8406561787428417,
      "learning_rate": 7.22871526796209e-06,
      "loss": 0.3701,
      "step": 6392
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.8496930886381215,
      "learning_rate": 7.225785907786516e-06,
      "loss": 0.3511,
      "step": 6393
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.870653186912155,
      "learning_rate": 7.222856805497087e-06,
      "loss": 0.4023,
      "step": 6394
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.627542651791187,
      "learning_rate": 7.219927961366091e-06,
      "loss": 0.3127,
      "step": 6395
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.754382589191206,
      "learning_rate": 7.216999375665785e-06,
      "loss": 0.3228,
      "step": 6396
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.9093756422818842,
      "learning_rate": 7.214071048668414e-06,
      "loss": 0.373,
      "step": 6397
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.8826994288007974,
      "learning_rate": 7.211142980646182e-06,
      "loss": 0.3696,
      "step": 6398
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.6951325440106035,
      "learning_rate": 7.208215171871277e-06,
      "loss": 0.3391,
      "step": 6399
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.7160541574630754,
      "learning_rate": 7.205287622615866e-06,
      "loss": 0.3716,
      "step": 6400
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.8626666288363742,
      "learning_rate": 7.20236033315209e-06,
      "loss": 0.3394,
      "step": 6401
    },
    {
      "epoch": 1.81,
      "grad_norm": 2.095504419401377,
      "learning_rate": 7.199433303752064e-06,
      "loss": 0.3279,
      "step": 6402
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.7278727956272577,
      "learning_rate": 7.196506534687875e-06,
      "loss": 0.374,
      "step": 6403
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.8093455880138876,
      "learning_rate": 7.193580026231596e-06,
      "loss": 0.354,
      "step": 6404
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.585406201737899,
      "learning_rate": 7.190653778655267e-06,
      "loss": 0.3071,
      "step": 6405
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.7310841689314875,
      "learning_rate": 7.187727792230912e-06,
      "loss": 0.3467,
      "step": 6406
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.534152656529068,
      "learning_rate": 7.184802067230518e-06,
      "loss": 0.2581,
      "step": 6407
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.7978460441944866,
      "learning_rate": 7.18187660392606e-06,
      "loss": 0.3613,
      "step": 6408
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.650464589694222,
      "learning_rate": 7.178951402589482e-06,
      "loss": 0.2939,
      "step": 6409
    },
    {
      "epoch": 1.81,
      "grad_norm": 2.2138628031413203,
      "learning_rate": 7.176026463492711e-06,
      "loss": 0.4185,
      "step": 6410
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.7096882305956056,
      "learning_rate": 7.173101786907638e-06,
      "loss": 0.2803,
      "step": 6411
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.9536482130273307,
      "learning_rate": 7.170177373106135e-06,
      "loss": 0.3423,
      "step": 6412
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.62497979625287,
      "learning_rate": 7.167253222360056e-06,
      "loss": 0.2998,
      "step": 6413
    },
    {
      "epoch": 1.81,
      "grad_norm": 2.1939395693963135,
      "learning_rate": 7.164329334941225e-06,
      "loss": 0.3774,
      "step": 6414
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.899472964559252,
      "learning_rate": 7.161405711121436e-06,
      "loss": 0.3896,
      "step": 6415
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.7225226070165174,
      "learning_rate": 7.158482351172465e-06,
      "loss": 0.2856,
      "step": 6416
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.8971789038303517,
      "learning_rate": 7.155559255366067e-06,
      "loss": 0.3057,
      "step": 6417
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.867056935885528,
      "learning_rate": 7.152636423973969e-06,
      "loss": 0.3875,
      "step": 6418
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.791661660053155,
      "learning_rate": 7.149713857267862e-06,
      "loss": 0.3018,
      "step": 6419
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.9953831466175027,
      "learning_rate": 7.146791555519431e-06,
      "loss": 0.3945,
      "step": 6420
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.9782544609498494,
      "learning_rate": 7.143869519000328e-06,
      "loss": 0.3901,
      "step": 6421
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.6897095034271825,
      "learning_rate": 7.140947747982179e-06,
      "loss": 0.3164,
      "step": 6422
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.7624598297751597,
      "learning_rate": 7.1380262427365885e-06,
      "loss": 0.3408,
      "step": 6423
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.6388227714479946,
      "learning_rate": 7.135105003535132e-06,
      "loss": 0.3052,
      "step": 6424
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.782189267133896,
      "learning_rate": 7.1321840306493625e-06,
      "loss": 0.3306,
      "step": 6425
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.6918783248004443,
      "learning_rate": 7.129263324350808e-06,
      "loss": 0.3032,
      "step": 6426
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.593735639266094,
      "learning_rate": 7.1263428849109805e-06,
      "loss": 0.2981,
      "step": 6427
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.8142727140637882,
      "learning_rate": 7.123422712601349e-06,
      "loss": 0.3584,
      "step": 6428
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.7600472552109663,
      "learning_rate": 7.1205028076933705e-06,
      "loss": 0.3218,
      "step": 6429
    },
    {
      "epoch": 1.82,
      "grad_norm": 2.1548155968821376,
      "learning_rate": 7.117583170458478e-06,
      "loss": 0.3584,
      "step": 6430
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.8410696356574585,
      "learning_rate": 7.114663801168073e-06,
      "loss": 0.3403,
      "step": 6431
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.642913321112393,
      "learning_rate": 7.111744700093531e-06,
      "loss": 0.2681,
      "step": 6432
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.8811470855325865,
      "learning_rate": 7.108825867506213e-06,
      "loss": 0.312,
      "step": 6433
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.996236959122384,
      "learning_rate": 7.105907303677446e-06,
      "loss": 0.3989,
      "step": 6434
    },
    {
      "epoch": 1.82,
      "grad_norm": 2.0664868126359637,
      "learning_rate": 7.102989008878538e-06,
      "loss": 0.3491,
      "step": 6435
    },
    {
      "epoch": 1.82,
      "grad_norm": 2.087545007293868,
      "learning_rate": 7.100070983380763e-06,
      "loss": 0.3994,
      "step": 6436
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.7158004814508914,
      "learning_rate": 7.097153227455379e-06,
      "loss": 0.282,
      "step": 6437
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.7082702989859455,
      "learning_rate": 7.0942357413736116e-06,
      "loss": 0.3049,
      "step": 6438
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.606890464268944,
      "learning_rate": 7.091318525406671e-06,
      "loss": 0.2925,
      "step": 6439
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.7296164797513331,
      "learning_rate": 7.0884015798257365e-06,
      "loss": 0.3198,
      "step": 6440
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.8895059513902697,
      "learning_rate": 7.0854849049019584e-06,
      "loss": 0.332,
      "step": 6441
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.8596430864576494,
      "learning_rate": 7.082568500906466e-06,
      "loss": 0.3408,
      "step": 6442
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.7357572919967892,
      "learning_rate": 7.0796523681103635e-06,
      "loss": 0.2991,
      "step": 6443
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.8845659429750004,
      "learning_rate": 7.076736506784734e-06,
      "loss": 0.3181,
      "step": 6444
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.6348987168906346,
      "learning_rate": 7.073820917200627e-06,
      "loss": 0.3491,
      "step": 6445
    },
    {
      "epoch": 1.82,
      "grad_norm": 2.07155776875727,
      "learning_rate": 7.070905599629068e-06,
      "loss": 0.3916,
      "step": 6446
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.7773353599982584,
      "learning_rate": 7.067990554341064e-06,
      "loss": 0.3481,
      "step": 6447
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.8917014517557036,
      "learning_rate": 7.065075781607594e-06,
      "loss": 0.3291,
      "step": 6448
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.6214951410033736,
      "learning_rate": 7.062161281699605e-06,
      "loss": 0.2996,
      "step": 6449
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.875621145387336,
      "learning_rate": 7.059247054888025e-06,
      "loss": 0.4282,
      "step": 6450
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.6004474368648869,
      "learning_rate": 7.056333101443761e-06,
      "loss": 0.2576,
      "step": 6451
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.810694331275202,
      "learning_rate": 7.053419421637686e-06,
      "loss": 0.3145,
      "step": 6452
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.7913303137678944,
      "learning_rate": 7.050506015740646e-06,
      "loss": 0.3711,
      "step": 6453
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.796126545288118,
      "learning_rate": 7.047592884023473e-06,
      "loss": 0.4175,
      "step": 6454
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.5012629140465665,
      "learning_rate": 7.04468002675696e-06,
      "loss": 0.3091,
      "step": 6455
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.6740696443947087,
      "learning_rate": 7.041767444211886e-06,
      "loss": 0.3423,
      "step": 6456
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.7748294673130338,
      "learning_rate": 7.038855136659002e-06,
      "loss": 0.3381,
      "step": 6457
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.9363264419542676,
      "learning_rate": 7.035943104369026e-06,
      "loss": 0.3662,
      "step": 6458
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.8698052945374264,
      "learning_rate": 7.033031347612655e-06,
      "loss": 0.3628,
      "step": 6459
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.6609375801208126,
      "learning_rate": 7.030119866660565e-06,
      "loss": 0.3113,
      "step": 6460
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.6162318692045319,
      "learning_rate": 7.027208661783402e-06,
      "loss": 0.2651,
      "step": 6461
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.7523594159722404,
      "learning_rate": 7.024297733251781e-06,
      "loss": 0.2559,
      "step": 6462
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.5094228496310689,
      "learning_rate": 7.021387081336302e-06,
      "loss": 0.2839,
      "step": 6463
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.7362960997356196,
      "learning_rate": 7.018476706307533e-06,
      "loss": 0.3484,
      "step": 6464
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.7301899873698554,
      "learning_rate": 7.01556660843602e-06,
      "loss": 0.3589,
      "step": 6465
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.9306755375653908,
      "learning_rate": 7.0126567879922735e-06,
      "loss": 0.3535,
      "step": 6466
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.7588587904317445,
      "learning_rate": 7.009747245246792e-06,
      "loss": 0.2817,
      "step": 6467
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.6040791470652458,
      "learning_rate": 7.006837980470038e-06,
      "loss": 0.2356,
      "step": 6468
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.6363094096853943,
      "learning_rate": 7.003928993932456e-06,
      "loss": 0.2942,
      "step": 6469
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.8211309763433605,
      "learning_rate": 7.001020285904454e-06,
      "loss": 0.3604,
      "step": 6470
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.8121155478721593,
      "learning_rate": 6.998111856656427e-06,
      "loss": 0.3188,
      "step": 6471
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.636305877284048,
      "learning_rate": 6.995203706458731e-06,
      "loss": 0.2944,
      "step": 6472
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.8932904875248806,
      "learning_rate": 6.9922958355817085e-06,
      "loss": 0.3621,
      "step": 6473
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.7513536039141804,
      "learning_rate": 6.989388244295669e-06,
      "loss": 0.2456,
      "step": 6474
    },
    {
      "epoch": 1.83,
      "grad_norm": 2.0801490059080514,
      "learning_rate": 6.9864809328708945e-06,
      "loss": 0.3574,
      "step": 6475
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.745565114894324,
      "learning_rate": 6.983573901577644e-06,
      "loss": 0.291,
      "step": 6476
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.8629358860319751,
      "learning_rate": 6.980667150686153e-06,
      "loss": 0.3557,
      "step": 6477
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.7701956377396613,
      "learning_rate": 6.977760680466628e-06,
      "loss": 0.3015,
      "step": 6478
    },
    {
      "epoch": 1.83,
      "grad_norm": 2.263650289897951,
      "learning_rate": 6.974854491189243e-06,
      "loss": 0.3794,
      "step": 6479
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.818031132836296,
      "learning_rate": 6.971948583124159e-06,
      "loss": 0.3091,
      "step": 6480
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.8878868112059954,
      "learning_rate": 6.969042956541503e-06,
      "loss": 0.3069,
      "step": 6481
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.938410354404707,
      "learning_rate": 6.9661376117113775e-06,
      "loss": 0.3779,
      "step": 6482
    },
    {
      "epoch": 1.83,
      "grad_norm": 2.1125265284066486,
      "learning_rate": 6.963232548903853e-06,
      "loss": 0.4487,
      "step": 6483
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.7042068056848716,
      "learning_rate": 6.960327768388987e-06,
      "loss": 0.3579,
      "step": 6484
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.892434757163528,
      "learning_rate": 6.957423270436796e-06,
      "loss": 0.3159,
      "step": 6485
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.8740713608856736,
      "learning_rate": 6.9545190553172836e-06,
      "loss": 0.3745,
      "step": 6486
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.7199374947607295,
      "learning_rate": 6.951615123300415e-06,
      "loss": 0.2793,
      "step": 6487
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.698986922201676,
      "learning_rate": 6.948711474656138e-06,
      "loss": 0.3044,
      "step": 6488
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.6600680597422375,
      "learning_rate": 6.945808109654366e-06,
      "loss": 0.291,
      "step": 6489
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.9013752912164061,
      "learning_rate": 6.9429050285650015e-06,
      "loss": 0.3701,
      "step": 6490
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.729592023067477,
      "learning_rate": 6.940002231657899e-06,
      "loss": 0.3425,
      "step": 6491
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.673155624053044,
      "learning_rate": 6.9370997192028995e-06,
      "loss": 0.3594,
      "step": 6492
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.540340831166815,
      "learning_rate": 6.934197491469818e-06,
      "loss": 0.3215,
      "step": 6493
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.9246685122841067,
      "learning_rate": 6.93129554872844e-06,
      "loss": 0.3232,
      "step": 6494
    },
    {
      "epoch": 1.83,
      "grad_norm": 2.1986889137768117,
      "learning_rate": 6.928393891248529e-06,
      "loss": 0.3999,
      "step": 6495
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.8250826820196506,
      "learning_rate": 6.92549251929981e-06,
      "loss": 0.4082,
      "step": 6496
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.857473474998123,
      "learning_rate": 6.922591433151995e-06,
      "loss": 0.3723,
      "step": 6497
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.9911238613119635,
      "learning_rate": 6.91969063307476e-06,
      "loss": 0.3433,
      "step": 6498
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.7329826482230195,
      "learning_rate": 6.916790119337766e-06,
      "loss": 0.3687,
      "step": 6499
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.6709207831666355,
      "learning_rate": 6.913889892210631e-06,
      "loss": 0.2688,
      "step": 6500
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.721756477416229,
      "learning_rate": 6.9109899519629605e-06,
      "loss": 0.3625,
      "step": 6501
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.6135815632953736,
      "learning_rate": 6.908090298864325e-06,
      "loss": 0.3015,
      "step": 6502
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.7541408764842645,
      "learning_rate": 6.905190933184275e-06,
      "loss": 0.3359,
      "step": 6503
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.8457542993330667,
      "learning_rate": 6.9022918551923245e-06,
      "loss": 0.3301,
      "step": 6504
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.836519086322503,
      "learning_rate": 6.899393065157972e-06,
      "loss": 0.3376,
      "step": 6505
    },
    {
      "epoch": 1.84,
      "grad_norm": 2.0053881665552202,
      "learning_rate": 6.896494563350681e-06,
      "loss": 0.3721,
      "step": 6506
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.80075898620906,
      "learning_rate": 6.893596350039896e-06,
      "loss": 0.3164,
      "step": 6507
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.6518656255544653,
      "learning_rate": 6.8906984254950235e-06,
      "loss": 0.2622,
      "step": 6508
    },
    {
      "epoch": 1.84,
      "grad_norm": 2.0584771412062497,
      "learning_rate": 6.887800789985452e-06,
      "loss": 0.4004,
      "step": 6509
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.5781517686945743,
      "learning_rate": 6.884903443780541e-06,
      "loss": 0.2793,
      "step": 6510
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.9611540654729607,
      "learning_rate": 6.882006387149625e-06,
      "loss": 0.3906,
      "step": 6511
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.8924132865398788,
      "learning_rate": 6.879109620362008e-06,
      "loss": 0.3901,
      "step": 6512
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.8676281988807437,
      "learning_rate": 6.876213143686965e-06,
      "loss": 0.3813,
      "step": 6513
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.8317432715877173,
      "learning_rate": 6.873316957393752e-06,
      "loss": 0.3665,
      "step": 6514
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.9512819505006975,
      "learning_rate": 6.87042106175159e-06,
      "loss": 0.4019,
      "step": 6515
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.7733348059885408,
      "learning_rate": 6.867525457029682e-06,
      "loss": 0.3203,
      "step": 6516
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.7848369039084235,
      "learning_rate": 6.864630143497191e-06,
      "loss": 0.3125,
      "step": 6517
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.672331059351868,
      "learning_rate": 6.861735121423264e-06,
      "loss": 0.3613,
      "step": 6518
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.985010523108,
      "learning_rate": 6.858840391077017e-06,
      "loss": 0.3979,
      "step": 6519
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.4933167547393624,
      "learning_rate": 6.8559459527275426e-06,
      "loss": 0.2651,
      "step": 6520
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.6570455476278558,
      "learning_rate": 6.853051806643898e-06,
      "loss": 0.2649,
      "step": 6521
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.678173534882247,
      "learning_rate": 6.850157953095117e-06,
      "loss": 0.3171,
      "step": 6522
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.883174043693147,
      "learning_rate": 6.84726439235021e-06,
      "loss": 0.4102,
      "step": 6523
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.7932479759377364,
      "learning_rate": 6.844371124678161e-06,
      "loss": 0.2986,
      "step": 6524
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.5365690707887019,
      "learning_rate": 6.841478150347918e-06,
      "loss": 0.2603,
      "step": 6525
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.6604125132348344,
      "learning_rate": 6.838585469628405e-06,
      "loss": 0.3562,
      "step": 6526
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.8305796102013026,
      "learning_rate": 6.8356930827885256e-06,
      "loss": 0.3289,
      "step": 6527
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.8061927924717123,
      "learning_rate": 6.832800990097148e-06,
      "loss": 0.3491,
      "step": 6528
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.703224566851172,
      "learning_rate": 6.829909191823121e-06,
      "loss": 0.3301,
      "step": 6529
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.93941996487063,
      "learning_rate": 6.827017688235255e-06,
      "loss": 0.3416,
      "step": 6530
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.4587201313642622,
      "learning_rate": 6.824126479602342e-06,
      "loss": 0.2747,
      "step": 6531
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.8746312945116077,
      "learning_rate": 6.821235566193143e-06,
      "loss": 0.3188,
      "step": 6532
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.8807808753334971,
      "learning_rate": 6.818344948276397e-06,
      "loss": 0.2922,
      "step": 6533
    },
    {
      "epoch": 1.85,
      "grad_norm": 2.0234939038783653,
      "learning_rate": 6.815454626120804e-06,
      "loss": 0.4785,
      "step": 6534
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.8030305770804056,
      "learning_rate": 6.812564599995042e-06,
      "loss": 0.3406,
      "step": 6535
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.6241948522264793,
      "learning_rate": 6.809674870167768e-06,
      "loss": 0.3137,
      "step": 6536
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.6512644255499513,
      "learning_rate": 6.80678543690761e-06,
      "loss": 0.3425,
      "step": 6537
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.6475626172711564,
      "learning_rate": 6.803896300483159e-06,
      "loss": 0.2983,
      "step": 6538
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.9243180860266897,
      "learning_rate": 6.8010074611629815e-06,
      "loss": 0.3513,
      "step": 6539
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.7200062434902292,
      "learning_rate": 6.798118919215625e-06,
      "loss": 0.3535,
      "step": 6540
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.7316326938007318,
      "learning_rate": 6.795230674909601e-06,
      "loss": 0.3496,
      "step": 6541
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.7171328636360244,
      "learning_rate": 6.7923427285133945e-06,
      "loss": 0.3228,
      "step": 6542
    },
    {
      "epoch": 1.85,
      "grad_norm": 2.1443198769489857,
      "learning_rate": 6.789455080295464e-06,
      "loss": 0.4326,
      "step": 6543
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.8102882280121124,
      "learning_rate": 6.786567730524243e-06,
      "loss": 0.3035,
      "step": 6544
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.6798843657086635,
      "learning_rate": 6.783680679468132e-06,
      "loss": 0.2908,
      "step": 6545
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.754896877734016,
      "learning_rate": 6.78079392739551e-06,
      "loss": 0.3113,
      "step": 6546
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.9575018804624653,
      "learning_rate": 6.777907474574718e-06,
      "loss": 0.3489,
      "step": 6547
    },
    {
      "epoch": 1.85,
      "grad_norm": 2.135456371913182,
      "learning_rate": 6.775021321274082e-06,
      "loss": 0.3918,
      "step": 6548
    },
    {
      "epoch": 1.85,
      "grad_norm": 2.0047598221256946,
      "learning_rate": 6.772135467761889e-06,
      "loss": 0.356,
      "step": 6549
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.826322392701368,
      "learning_rate": 6.769249914306408e-06,
      "loss": 0.3633,
      "step": 6550
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.7827881647028982,
      "learning_rate": 6.766364661175872e-06,
      "loss": 0.2937,
      "step": 6551
    },
    {
      "epoch": 1.85,
      "grad_norm": 2.174572514880288,
      "learning_rate": 6.763479708638485e-06,
      "loss": 0.4443,
      "step": 6552
    },
    {
      "epoch": 1.85,
      "grad_norm": 2.0684216722440762,
      "learning_rate": 6.7605950569624335e-06,
      "loss": 0.4199,
      "step": 6553
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.7685600608276904,
      "learning_rate": 6.757710706415872e-06,
      "loss": 0.3105,
      "step": 6554
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.9421460016978978,
      "learning_rate": 6.754826657266918e-06,
      "loss": 0.2998,
      "step": 6555
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.8800421414998942,
      "learning_rate": 6.7519429097836675e-06,
      "loss": 0.3472,
      "step": 6556
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.7651174382621617,
      "learning_rate": 6.749059464234193e-06,
      "loss": 0.3672,
      "step": 6557
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.88919778782801,
      "learning_rate": 6.746176320886537e-06,
      "loss": 0.3508,
      "step": 6558
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.8323306316639119,
      "learning_rate": 6.743293480008703e-06,
      "loss": 0.3535,
      "step": 6559
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.732435817323488,
      "learning_rate": 6.740410941868678e-06,
      "loss": 0.3054,
      "step": 6560
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.5722938266700663,
      "learning_rate": 6.737528706734423e-06,
      "loss": 0.2866,
      "step": 6561
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.8064379816000327,
      "learning_rate": 6.734646774873863e-06,
      "loss": 0.364,
      "step": 6562
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.593482357255299,
      "learning_rate": 6.731765146554891e-06,
      "loss": 0.2778,
      "step": 6563
    },
    {
      "epoch": 1.85,
      "grad_norm": 2.001267425942674,
      "learning_rate": 6.728883822045389e-06,
      "loss": 0.4092,
      "step": 6564
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.8335169286656023,
      "learning_rate": 6.726002801613189e-06,
      "loss": 0.3574,
      "step": 6565
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.8548006147482416,
      "learning_rate": 6.723122085526113e-06,
      "loss": 0.3462,
      "step": 6566
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.7125461984843464,
      "learning_rate": 6.720241674051948e-06,
      "loss": 0.2827,
      "step": 6567
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.568069972030013,
      "learning_rate": 6.717361567458449e-06,
      "loss": 0.3628,
      "step": 6568
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.6806893435783332,
      "learning_rate": 6.714481766013343e-06,
      "loss": 0.2854,
      "step": 6569
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.7660536220828853,
      "learning_rate": 6.711602269984339e-06,
      "loss": 0.3765,
      "step": 6570
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.9240509616693255,
      "learning_rate": 6.7087230796391035e-06,
      "loss": 0.3838,
      "step": 6571
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.8608161083406198,
      "learning_rate": 6.705844195245283e-06,
      "loss": 0.3384,
      "step": 6572
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.6802423074074793,
      "learning_rate": 6.702965617070492e-06,
      "loss": 0.3481,
      "step": 6573
    },
    {
      "epoch": 1.86,
      "grad_norm": 2.264546274004109,
      "learning_rate": 6.7000873453823225e-06,
      "loss": 0.4458,
      "step": 6574
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.886293040696927,
      "learning_rate": 6.697209380448333e-06,
      "loss": 0.4111,
      "step": 6575
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.5963231057068994,
      "learning_rate": 6.6943317225360474e-06,
      "loss": 0.3208,
      "step": 6576
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.876610891175688,
      "learning_rate": 6.691454371912974e-06,
      "loss": 0.3281,
      "step": 6577
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.6536571576081305,
      "learning_rate": 6.688577328846586e-06,
      "loss": 0.3022,
      "step": 6578
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.7214351399656052,
      "learning_rate": 6.685700593604329e-06,
      "loss": 0.3516,
      "step": 6579
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.5944391251950227,
      "learning_rate": 6.6828241664536145e-06,
      "loss": 0.3403,
      "step": 6580
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.7729050686089438,
      "learning_rate": 6.679948047661835e-06,
      "loss": 0.3345,
      "step": 6581
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.7741398380272482,
      "learning_rate": 6.677072237496347e-06,
      "loss": 0.2871,
      "step": 6582
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.7023017992806615,
      "learning_rate": 6.674196736224481e-06,
      "loss": 0.335,
      "step": 6583
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.6629442411809845,
      "learning_rate": 6.6713215441135424e-06,
      "loss": 0.2664,
      "step": 6584
    },
    {
      "epoch": 1.86,
      "grad_norm": 2.0440769316918646,
      "learning_rate": 6.668446661430801e-06,
      "loss": 0.4966,
      "step": 6585
    },
    {
      "epoch": 1.86,
      "grad_norm": 2.0069293938898927,
      "learning_rate": 6.665572088443497e-06,
      "loss": 0.439,
      "step": 6586
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.6524604172403987,
      "learning_rate": 6.662697825418853e-06,
      "loss": 0.3179,
      "step": 6587
    },
    {
      "epoch": 1.86,
      "grad_norm": 2.118615889197074,
      "learning_rate": 6.659823872624054e-06,
      "loss": 0.439,
      "step": 6588
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.8410423021625866,
      "learning_rate": 6.656950230326251e-06,
      "loss": 0.4238,
      "step": 6589
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.9438457715654485,
      "learning_rate": 6.65407689879258e-06,
      "loss": 0.3579,
      "step": 6590
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.6984240038039526,
      "learning_rate": 6.651203878290139e-06,
      "loss": 0.2944,
      "step": 6591
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.9174725477728898,
      "learning_rate": 6.648331169086002e-06,
      "loss": 0.3599,
      "step": 6592
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.8261911112437348,
      "learning_rate": 6.645458771447202e-06,
      "loss": 0.3638,
      "step": 6593
    },
    {
      "epoch": 1.86,
      "grad_norm": 2.02407787162618,
      "learning_rate": 6.642586685640761e-06,
      "loss": 0.3999,
      "step": 6594
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.9372893226957437,
      "learning_rate": 6.639714911933658e-06,
      "loss": 0.3667,
      "step": 6595
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.9017457761133343,
      "learning_rate": 6.636843450592854e-06,
      "loss": 0.3486,
      "step": 6596
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.6965899191646252,
      "learning_rate": 6.633972301885268e-06,
      "loss": 0.366,
      "step": 6597
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.6378828581727403,
      "learning_rate": 6.631101466077801e-06,
      "loss": 0.2964,
      "step": 6598
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.7597368413045342,
      "learning_rate": 6.628230943437319e-06,
      "loss": 0.2966,
      "step": 6599
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.7279263715715196,
      "learning_rate": 6.625360734230663e-06,
      "loss": 0.2661,
      "step": 6600
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.9321951379121258,
      "learning_rate": 6.6224908387246466e-06,
      "loss": 0.4585,
      "step": 6601
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.8445259218648105,
      "learning_rate": 6.619621257186039e-06,
      "loss": 0.3506,
      "step": 6602
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.7308490220967043,
      "learning_rate": 6.616751989881598e-06,
      "loss": 0.3179,
      "step": 6603
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.5894456289233554,
      "learning_rate": 6.613883037078048e-06,
      "loss": 0.3147,
      "step": 6604
    },
    {
      "epoch": 1.87,
      "grad_norm": 2.1857512925443694,
      "learning_rate": 6.6110143990420824e-06,
      "loss": 0.3877,
      "step": 6605
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.7073234117262361,
      "learning_rate": 6.608146076040358e-06,
      "loss": 0.3342,
      "step": 6606
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.6772092600261752,
      "learning_rate": 6.605278068339516e-06,
      "loss": 0.3064,
      "step": 6607
    },
    {
      "epoch": 1.87,
      "grad_norm": 2.006641569443639,
      "learning_rate": 6.602410376206154e-06,
      "loss": 0.457,
      "step": 6608
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.8593529969236626,
      "learning_rate": 6.599542999906858e-06,
      "loss": 0.4185,
      "step": 6609
    },
    {
      "epoch": 1.87,
      "grad_norm": 2.0525670509024785,
      "learning_rate": 6.596675939708166e-06,
      "loss": 0.4175,
      "step": 6610
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.930937945065672,
      "learning_rate": 6.593809195876597e-06,
      "loss": 0.3933,
      "step": 6611
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.756199129844216,
      "learning_rate": 6.5909427686786386e-06,
      "loss": 0.2812,
      "step": 6612
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.6062527869411203,
      "learning_rate": 6.588076658380754e-06,
      "loss": 0.3252,
      "step": 6613
    },
    {
      "epoch": 1.87,
      "grad_norm": 2.0325459966043575,
      "learning_rate": 6.585210865249363e-06,
      "loss": 0.3423,
      "step": 6614
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.898219620630347,
      "learning_rate": 6.58234538955087e-06,
      "loss": 0.3394,
      "step": 6615
    },
    {
      "epoch": 1.87,
      "grad_norm": 2.1438186907816488,
      "learning_rate": 6.579480231551644e-06,
      "loss": 0.3809,
      "step": 6616
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.5980096912862334,
      "learning_rate": 6.576615391518026e-06,
      "loss": 0.2888,
      "step": 6617
    },
    {
      "epoch": 1.87,
      "grad_norm": 2.148467147465418,
      "learning_rate": 6.573750869716327e-06,
      "loss": 0.4131,
      "step": 6618
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.8535019895295375,
      "learning_rate": 6.570886666412823e-06,
      "loss": 0.3521,
      "step": 6619
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.7602553257261728,
      "learning_rate": 6.5680227818737695e-06,
      "loss": 0.3281,
      "step": 6620
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.976587524858602,
      "learning_rate": 6.5651592163653885e-06,
      "loss": 0.439,
      "step": 6621
    },
    {
      "epoch": 1.87,
      "grad_norm": 2.0299517025435545,
      "learning_rate": 6.562295970153875e-06,
      "loss": 0.3804,
      "step": 6622
    },
    {
      "epoch": 1.87,
      "grad_norm": 2.0709741104689803,
      "learning_rate": 6.559433043505383e-06,
      "loss": 0.4116,
      "step": 6623
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.7743544933215039,
      "learning_rate": 6.556570436686052e-06,
      "loss": 0.3521,
      "step": 6624
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.8607992574289076,
      "learning_rate": 6.553708149961979e-06,
      "loss": 0.3447,
      "step": 6625
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.8424734220108385,
      "learning_rate": 6.550846183599249e-06,
      "loss": 0.3284,
      "step": 6626
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.937406322853396,
      "learning_rate": 6.5479845378638905e-06,
      "loss": 0.3716,
      "step": 6627
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.7901894202875541,
      "learning_rate": 6.545123213021927e-06,
      "loss": 0.3413,
      "step": 6628
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.8036313003790057,
      "learning_rate": 6.542262209339339e-06,
      "loss": 0.2825,
      "step": 6629
    },
    {
      "epoch": 1.87,
      "grad_norm": 2.0624602031081243,
      "learning_rate": 6.539401527082083e-06,
      "loss": 0.3716,
      "step": 6630
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.8534522364715542,
      "learning_rate": 6.536541166516079e-06,
      "loss": 0.4146,
      "step": 6631
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.8993287176520575,
      "learning_rate": 6.533681127907224e-06,
      "loss": 0.3606,
      "step": 6632
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.7501583125580344,
      "learning_rate": 6.5308214115213785e-06,
      "loss": 0.2932,
      "step": 6633
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.9687005643872617,
      "learning_rate": 6.527962017624383e-06,
      "loss": 0.4146,
      "step": 6634
    },
    {
      "epoch": 1.87,
      "grad_norm": 2.066113507959135,
      "learning_rate": 6.52510294648204e-06,
      "loss": 0.3389,
      "step": 6635
    },
    {
      "epoch": 1.87,
      "grad_norm": 2.0946976789163627,
      "learning_rate": 6.522244198360119e-06,
      "loss": 0.4307,
      "step": 6636
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.86554404972336,
      "learning_rate": 6.519385773524371e-06,
      "loss": 0.3701,
      "step": 6637
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.8189746897138066,
      "learning_rate": 6.516527672240504e-06,
      "loss": 0.3369,
      "step": 6638
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.7228568477099782,
      "learning_rate": 6.513669894774209e-06,
      "loss": 0.3359,
      "step": 6639
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.7420420482937964,
      "learning_rate": 6.510812441391131e-06,
      "loss": 0.3003,
      "step": 6640
    },
    {
      "epoch": 1.88,
      "grad_norm": 2.010824750763657,
      "learning_rate": 6.507955312356903e-06,
      "loss": 0.3955,
      "step": 6641
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.982374610468388,
      "learning_rate": 6.505098507937111e-06,
      "loss": 0.3767,
      "step": 6642
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.640253842029038,
      "learning_rate": 6.502242028397328e-06,
      "loss": 0.334,
      "step": 6643
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.5035528325789453,
      "learning_rate": 6.499385874003077e-06,
      "loss": 0.239,
      "step": 6644
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.7179638029627,
      "learning_rate": 6.496530045019869e-06,
      "loss": 0.3054,
      "step": 6645
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.8658250236051972,
      "learning_rate": 6.493674541713172e-06,
      "loss": 0.3569,
      "step": 6646
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.84036340195829,
      "learning_rate": 6.490819364348434e-06,
      "loss": 0.2949,
      "step": 6647
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.643124230203491,
      "learning_rate": 6.4879645131910626e-06,
      "loss": 0.2661,
      "step": 6648
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.8822157641226331,
      "learning_rate": 6.485109988506439e-06,
      "loss": 0.3127,
      "step": 6649
    },
    {
      "epoch": 1.88,
      "grad_norm": 2.1883003389454796,
      "learning_rate": 6.4822557905599156e-06,
      "loss": 0.3916,
      "step": 6650
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.7964175775973774,
      "learning_rate": 6.479401919616821e-06,
      "loss": 0.3242,
      "step": 6651
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.8621851094000321,
      "learning_rate": 6.476548375942437e-06,
      "loss": 0.4058,
      "step": 6652
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.8968893680576118,
      "learning_rate": 6.473695159802027e-06,
      "loss": 0.3472,
      "step": 6653
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.7051015988338865,
      "learning_rate": 6.470842271460823e-06,
      "loss": 0.2939,
      "step": 6654
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.5309226671019598,
      "learning_rate": 6.467989711184021e-06,
      "loss": 0.2627,
      "step": 6655
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.709448596620678,
      "learning_rate": 6.465137479236796e-06,
      "loss": 0.3083,
      "step": 6656
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.9218663338204693,
      "learning_rate": 6.4622855758842785e-06,
      "loss": 0.3804,
      "step": 6657
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.9548353973718762,
      "learning_rate": 6.459434001391582e-06,
      "loss": 0.3525,
      "step": 6658
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.6845217701010937,
      "learning_rate": 6.456582756023781e-06,
      "loss": 0.3081,
      "step": 6659
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.8813042886865627,
      "learning_rate": 6.4537318400459295e-06,
      "loss": 0.3455,
      "step": 6660
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.8805363458510462,
      "learning_rate": 6.450881253723035e-06,
      "loss": 0.3286,
      "step": 6661
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.8508238014148433,
      "learning_rate": 6.448030997320084e-06,
      "loss": 0.3442,
      "step": 6662
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.6697505219525315,
      "learning_rate": 6.445181071102034e-06,
      "loss": 0.2634,
      "step": 6663
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.8841239990266927,
      "learning_rate": 6.4423314753338115e-06,
      "loss": 0.3667,
      "step": 6664
    },
    {
      "epoch": 1.88,
      "grad_norm": 2.203646526033782,
      "learning_rate": 6.439482210280308e-06,
      "loss": 0.3999,
      "step": 6665
    },
    {
      "epoch": 1.88,
      "grad_norm": 2.0342566111645115,
      "learning_rate": 6.436633276206381e-06,
      "loss": 0.385,
      "step": 6666
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.7150288065562431,
      "learning_rate": 6.43378467337687e-06,
      "loss": 0.3225,
      "step": 6667
    },
    {
      "epoch": 1.88,
      "grad_norm": 2.1726127733336935,
      "learning_rate": 6.430936402056577e-06,
      "loss": 0.3914,
      "step": 6668
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.5494358043768701,
      "learning_rate": 6.428088462510262e-06,
      "loss": 0.2849,
      "step": 6669
    },
    {
      "epoch": 1.88,
      "grad_norm": 2.1330874201061842,
      "learning_rate": 6.425240855002674e-06,
      "loss": 0.3936,
      "step": 6670
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.8870561022473933,
      "learning_rate": 6.422393579798519e-06,
      "loss": 0.3352,
      "step": 6671
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.7603760716388284,
      "learning_rate": 6.4195466371624726e-06,
      "loss": 0.304,
      "step": 6672
    },
    {
      "epoch": 1.88,
      "grad_norm": 2.1022680744235145,
      "learning_rate": 6.416700027359189e-06,
      "loss": 0.4263,
      "step": 6673
    },
    {
      "epoch": 1.88,
      "grad_norm": 2.083697092166709,
      "learning_rate": 6.413853750653274e-06,
      "loss": 0.3843,
      "step": 6674
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.7711260931868806,
      "learning_rate": 6.41100780730932e-06,
      "loss": 0.3298,
      "step": 6675
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.8743219713907429,
      "learning_rate": 6.408162197591876e-06,
      "loss": 0.4331,
      "step": 6676
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.8669945322065344,
      "learning_rate": 6.405316921765472e-06,
      "loss": 0.3228,
      "step": 6677
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.780173528668004,
      "learning_rate": 6.402471980094594e-06,
      "loss": 0.3496,
      "step": 6678
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.7794248595573523,
      "learning_rate": 6.399627372843699e-06,
      "loss": 0.3296,
      "step": 6679
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.6908593610497729,
      "learning_rate": 6.396783100277224e-06,
      "loss": 0.373,
      "step": 6680
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.859169215810217,
      "learning_rate": 6.3939391626595704e-06,
      "loss": 0.3618,
      "step": 6681
    },
    {
      "epoch": 1.89,
      "grad_norm": 2.0387215863810138,
      "learning_rate": 6.391095560255098e-06,
      "loss": 0.3931,
      "step": 6682
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.7925925320520604,
      "learning_rate": 6.388252293328143e-06,
      "loss": 0.334,
      "step": 6683
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.90263329488295,
      "learning_rate": 6.385409362143018e-06,
      "loss": 0.355,
      "step": 6684
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.9778811299047743,
      "learning_rate": 6.382566766963993e-06,
      "loss": 0.3809,
      "step": 6685
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.5936278415947454,
      "learning_rate": 6.3797245080553075e-06,
      "loss": 0.2805,
      "step": 6686
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.837187976499243,
      "learning_rate": 6.376882585681174e-06,
      "loss": 0.4111,
      "step": 6687
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.7197870355207847,
      "learning_rate": 6.374041000105779e-06,
      "loss": 0.3232,
      "step": 6688
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.7165781081349785,
      "learning_rate": 6.371199751593264e-06,
      "loss": 0.3428,
      "step": 6689
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.6971153663055474,
      "learning_rate": 6.368358840407754e-06,
      "loss": 0.3186,
      "step": 6690
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.7950074969845689,
      "learning_rate": 6.365518266813327e-06,
      "loss": 0.3359,
      "step": 6691
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.7843146349563308,
      "learning_rate": 6.362678031074041e-06,
      "loss": 0.3157,
      "step": 6692
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.8547293109029601,
      "learning_rate": 6.35983813345392e-06,
      "loss": 0.3877,
      "step": 6693
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.7493308702010897,
      "learning_rate": 6.3569985742169594e-06,
      "loss": 0.3528,
      "step": 6694
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.940754803400101,
      "learning_rate": 6.354159353627114e-06,
      "loss": 0.3523,
      "step": 6695
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.7954283186059297,
      "learning_rate": 6.351320471948313e-06,
      "loss": 0.3628,
      "step": 6696
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.768314984073666,
      "learning_rate": 6.348481929444458e-06,
      "loss": 0.3281,
      "step": 6697
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.7010391609854916,
      "learning_rate": 6.345643726379416e-06,
      "loss": 0.2581,
      "step": 6698
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.9702552616900904,
      "learning_rate": 6.342805863017012e-06,
      "loss": 0.3799,
      "step": 6699
    },
    {
      "epoch": 1.89,
      "grad_norm": 2.077054620980844,
      "learning_rate": 6.339968339621056e-06,
      "loss": 0.3535,
      "step": 6700
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.996927116158408,
      "learning_rate": 6.337131156455323e-06,
      "loss": 0.3965,
      "step": 6701
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.8270515656329467,
      "learning_rate": 6.334294313783549e-06,
      "loss": 0.2708,
      "step": 6702
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.777195681289133,
      "learning_rate": 6.331457811869437e-06,
      "loss": 0.2815,
      "step": 6703
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.6017548974226126,
      "learning_rate": 6.32862165097667e-06,
      "loss": 0.2988,
      "step": 6704
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.5821394975972134,
      "learning_rate": 6.325785831368887e-06,
      "loss": 0.2896,
      "step": 6705
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.7186211473910247,
      "learning_rate": 6.322950353309705e-06,
      "loss": 0.3262,
      "step": 6706
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.959239370257076,
      "learning_rate": 6.320115217062708e-06,
      "loss": 0.3521,
      "step": 6707
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.8099102162420528,
      "learning_rate": 6.317280422891439e-06,
      "loss": 0.2998,
      "step": 6708
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.7349259799511556,
      "learning_rate": 6.314445971059416e-06,
      "loss": 0.3081,
      "step": 6709
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.8880648119055203,
      "learning_rate": 6.311611861830129e-06,
      "loss": 0.3735,
      "step": 6710
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.9919513916404623,
      "learning_rate": 6.3087780954670306e-06,
      "loss": 0.457,
      "step": 6711
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.6658844494530083,
      "learning_rate": 6.305944672233542e-06,
      "loss": 0.2778,
      "step": 6712
    },
    {
      "epoch": 1.9,
      "grad_norm": 2.1179060957500355,
      "learning_rate": 6.303111592393051e-06,
      "loss": 0.4312,
      "step": 6713
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.757316775149834,
      "learning_rate": 6.300278856208919e-06,
      "loss": 0.2627,
      "step": 6714
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.5892687336356512,
      "learning_rate": 6.2974464639444745e-06,
      "loss": 0.2864,
      "step": 6715
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.8456873757893333,
      "learning_rate": 6.294614415863005e-06,
      "loss": 0.3994,
      "step": 6716
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.9228015608888331,
      "learning_rate": 6.291782712227776e-06,
      "loss": 0.3809,
      "step": 6717
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.7782926728356416,
      "learning_rate": 6.28895135330202e-06,
      "loss": 0.3503,
      "step": 6718
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.6973157456753778,
      "learning_rate": 6.286120339348935e-06,
      "loss": 0.3188,
      "step": 6719
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.832897941481605,
      "learning_rate": 6.283289670631684e-06,
      "loss": 0.3647,
      "step": 6720
    },
    {
      "epoch": 1.9,
      "grad_norm": 2.2520255782481757,
      "learning_rate": 6.280459347413402e-06,
      "loss": 0.3389,
      "step": 6721
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.6340884869375438,
      "learning_rate": 6.277629369957191e-06,
      "loss": 0.3298,
      "step": 6722
    },
    {
      "epoch": 1.9,
      "grad_norm": 2.1583466310378907,
      "learning_rate": 6.274799738526125e-06,
      "loss": 0.3652,
      "step": 6723
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.8580766672774316,
      "learning_rate": 6.271970453383235e-06,
      "loss": 0.3865,
      "step": 6724
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.8276647873778749,
      "learning_rate": 6.2691415147915306e-06,
      "loss": 0.3296,
      "step": 6725
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.946186947280052,
      "learning_rate": 6.266312923013983e-06,
      "loss": 0.3723,
      "step": 6726
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.9134798080546398,
      "learning_rate": 6.263484678313536e-06,
      "loss": 0.3638,
      "step": 6727
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.6526030812211983,
      "learning_rate": 6.260656780953098e-06,
      "loss": 0.2483,
      "step": 6728
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.7448302898792187,
      "learning_rate": 6.25782923119554e-06,
      "loss": 0.334,
      "step": 6729
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.7646107471103325,
      "learning_rate": 6.2550020293037095e-06,
      "loss": 0.3064,
      "step": 6730
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.8377645053652947,
      "learning_rate": 6.2521751755404226e-06,
      "loss": 0.3315,
      "step": 6731
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.837743858098263,
      "learning_rate": 6.2493486701684556e-06,
      "loss": 0.3325,
      "step": 6732
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.9585888367765139,
      "learning_rate": 6.246522513450552e-06,
      "loss": 0.3755,
      "step": 6733
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.9149764072600322,
      "learning_rate": 6.243696705649432e-06,
      "loss": 0.3618,
      "step": 6734
    },
    {
      "epoch": 1.9,
      "grad_norm": 2.265694492367969,
      "learning_rate": 6.240871247027774e-06,
      "loss": 0.5112,
      "step": 6735
    },
    {
      "epoch": 1.9,
      "grad_norm": 2.0977049338289193,
      "learning_rate": 6.23804613784823e-06,
      "loss": 0.395,
      "step": 6736
    },
    {
      "epoch": 1.9,
      "grad_norm": 2.0503670632459547,
      "learning_rate": 6.235221378373415e-06,
      "loss": 0.4688,
      "step": 6737
    },
    {
      "epoch": 1.9,
      "grad_norm": 2.0305628899431816,
      "learning_rate": 6.232396968865916e-06,
      "loss": 0.333,
      "step": 6738
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.8282792903398855,
      "learning_rate": 6.229572909588282e-06,
      "loss": 0.3423,
      "step": 6739
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.6662879781565947,
      "learning_rate": 6.2267492008030395e-06,
      "loss": 0.3206,
      "step": 6740
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.8295807795374637,
      "learning_rate": 6.223925842772668e-06,
      "loss": 0.3135,
      "step": 6741
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.649377450081735,
      "learning_rate": 6.221102835759626e-06,
      "loss": 0.292,
      "step": 6742
    },
    {
      "epoch": 1.9,
      "grad_norm": 2.015303611019319,
      "learning_rate": 6.2182801800263325e-06,
      "loss": 0.416,
      "step": 6743
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.708877977400773,
      "learning_rate": 6.21545787583518e-06,
      "loss": 0.3105,
      "step": 6744
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.9389106706955779,
      "learning_rate": 6.212635923448526e-06,
      "loss": 0.3726,
      "step": 6745
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.5554107173204699,
      "learning_rate": 6.209814323128689e-06,
      "loss": 0.281,
      "step": 6746
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.7669037036223867,
      "learning_rate": 6.2069930751379635e-06,
      "loss": 0.3486,
      "step": 6747
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.7338847128306347,
      "learning_rate": 6.2041721797386075e-06,
      "loss": 0.3569,
      "step": 6748
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.6688614212238184,
      "learning_rate": 6.201351637192849e-06,
      "loss": 0.2397,
      "step": 6749
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.4633010651793996,
      "learning_rate": 6.198531447762875e-06,
      "loss": 0.2605,
      "step": 6750
    },
    {
      "epoch": 1.91,
      "grad_norm": 2.229058048447936,
      "learning_rate": 6.195711611710851e-06,
      "loss": 0.3718,
      "step": 6751
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.8341264852270671,
      "learning_rate": 6.192892129298898e-06,
      "loss": 0.3335,
      "step": 6752
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.7150819474489194,
      "learning_rate": 6.1900730007891185e-06,
      "loss": 0.3677,
      "step": 6753
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.8815858821390328,
      "learning_rate": 6.187254226443566e-06,
      "loss": 0.3005,
      "step": 6754
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.7233204323653935,
      "learning_rate": 6.184435806524274e-06,
      "loss": 0.303,
      "step": 6755
    },
    {
      "epoch": 1.91,
      "grad_norm": 2.0479006506394493,
      "learning_rate": 6.181617741293234e-06,
      "loss": 0.4175,
      "step": 6756
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.569937804143893,
      "learning_rate": 6.178800031012414e-06,
      "loss": 0.2905,
      "step": 6757
    },
    {
      "epoch": 1.91,
      "grad_norm": 2.039069710735134,
      "learning_rate": 6.175982675943739e-06,
      "loss": 0.3213,
      "step": 6758
    },
    {
      "epoch": 1.91,
      "grad_norm": 2.281759172667215,
      "learning_rate": 6.173165676349103e-06,
      "loss": 0.4863,
      "step": 6759
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.8066334404104976,
      "learning_rate": 6.1703490324903745e-06,
      "loss": 0.3325,
      "step": 6760
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.8097662225882698,
      "learning_rate": 6.167532744629383e-06,
      "loss": 0.3286,
      "step": 6761
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.8669018360819114,
      "learning_rate": 6.164716813027929e-06,
      "loss": 0.3259,
      "step": 6762
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.7912169939802123,
      "learning_rate": 6.161901237947768e-06,
      "loss": 0.3071,
      "step": 6763
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.8361860916934003,
      "learning_rate": 6.159086019650635e-06,
      "loss": 0.3743,
      "step": 6764
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.8050375513926296,
      "learning_rate": 6.15627115839823e-06,
      "loss": 0.3115,
      "step": 6765
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.865690567754256,
      "learning_rate": 6.1534566544522175e-06,
      "loss": 0.3286,
      "step": 6766
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.78063143284909,
      "learning_rate": 6.150642508074225e-06,
      "loss": 0.354,
      "step": 6767
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.8912404047300204,
      "learning_rate": 6.1478287195258545e-06,
      "loss": 0.2954,
      "step": 6768
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.9432827847508067,
      "learning_rate": 6.1450152890686684e-06,
      "loss": 0.4233,
      "step": 6769
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.7005760934732026,
      "learning_rate": 6.142202216964204e-06,
      "loss": 0.314,
      "step": 6770
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.8123899068945366,
      "learning_rate": 6.139389503473951e-06,
      "loss": 0.3323,
      "step": 6771
    },
    {
      "epoch": 1.91,
      "grad_norm": 2.037199593141389,
      "learning_rate": 6.13657714885938e-06,
      "loss": 0.3689,
      "step": 6772
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.7647626406187809,
      "learning_rate": 6.133765153381918e-06,
      "loss": 0.2688,
      "step": 6773
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.9050885822140429,
      "learning_rate": 6.130953517302972e-06,
      "loss": 0.3555,
      "step": 6774
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.7939082482852131,
      "learning_rate": 6.128142240883899e-06,
      "loss": 0.3306,
      "step": 6775
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.9404429140068715,
      "learning_rate": 6.12533132438603e-06,
      "loss": 0.3726,
      "step": 6776
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.8311460970437883,
      "learning_rate": 6.122520768070669e-06,
      "loss": 0.3105,
      "step": 6777
    },
    {
      "epoch": 1.91,
      "grad_norm": 2.046945114609409,
      "learning_rate": 6.119710572199075e-06,
      "loss": 0.4067,
      "step": 6778
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.742383672028282,
      "learning_rate": 6.116900737032484e-06,
      "loss": 0.2935,
      "step": 6779
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.7290471370680278,
      "learning_rate": 6.114091262832087e-06,
      "loss": 0.2605,
      "step": 6780
    },
    {
      "epoch": 1.91,
      "grad_norm": 2.0191803196272518,
      "learning_rate": 6.111282149859052e-06,
      "loss": 0.3662,
      "step": 6781
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.6992178413928034,
      "learning_rate": 6.108473398374509e-06,
      "loss": 0.3333,
      "step": 6782
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.7926522883068405,
      "learning_rate": 6.105665008639557e-06,
      "loss": 0.2974,
      "step": 6783
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.9202218236036355,
      "learning_rate": 6.102856980915253e-06,
      "loss": 0.3306,
      "step": 6784
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.8082873691871644,
      "learning_rate": 6.10004931546263e-06,
      "loss": 0.304,
      "step": 6785
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.4304619616030105,
      "learning_rate": 6.097242012542683e-06,
      "loss": 0.2554,
      "step": 6786
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.7666199772165172,
      "learning_rate": 6.094435072416379e-06,
      "loss": 0.3794,
      "step": 6787
    },
    {
      "epoch": 1.92,
      "grad_norm": 2.0267100102317004,
      "learning_rate": 6.091628495344639e-06,
      "loss": 0.4561,
      "step": 6788
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.9296143455166785,
      "learning_rate": 6.088822281588359e-06,
      "loss": 0.3755,
      "step": 6789
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.6871778993663324,
      "learning_rate": 6.0860164314084e-06,
      "loss": 0.2681,
      "step": 6790
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.9240807065158256,
      "learning_rate": 6.083210945065595e-06,
      "loss": 0.3394,
      "step": 6791
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.9628217360000006,
      "learning_rate": 6.08040582282073e-06,
      "loss": 0.3867,
      "step": 6792
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.91831608570659,
      "learning_rate": 6.077601064934565e-06,
      "loss": 0.4062,
      "step": 6793
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.7349572071762376,
      "learning_rate": 6.074796671667829e-06,
      "loss": 0.2705,
      "step": 6794
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.6794696812766339,
      "learning_rate": 6.071992643281209e-06,
      "loss": 0.2964,
      "step": 6795
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.604595526598498,
      "learning_rate": 6.06918898003537e-06,
      "loss": 0.3462,
      "step": 6796
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.971820092057661,
      "learning_rate": 6.066385682190928e-06,
      "loss": 0.3892,
      "step": 6797
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.519264364752603,
      "learning_rate": 6.063582750008474e-06,
      "loss": 0.282,
      "step": 6798
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.920957636542304,
      "learning_rate": 6.0607801837485665e-06,
      "loss": 0.385,
      "step": 6799
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.8393461260495776,
      "learning_rate": 6.05797798367173e-06,
      "loss": 0.3467,
      "step": 6800
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.9617170131373076,
      "learning_rate": 6.055176150038445e-06,
      "loss": 0.46,
      "step": 6801
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.7153412735757532,
      "learning_rate": 6.052374683109168e-06,
      "loss": 0.3145,
      "step": 6802
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.685038875237218,
      "learning_rate": 6.049573583144317e-06,
      "loss": 0.3345,
      "step": 6803
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.617574006448173,
      "learning_rate": 6.0467728504042844e-06,
      "loss": 0.3052,
      "step": 6804
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.6561919630950814,
      "learning_rate": 6.043972485149414e-06,
      "loss": 0.3279,
      "step": 6805
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.7403823400643312,
      "learning_rate": 6.041172487640024e-06,
      "loss": 0.3618,
      "step": 6806
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.8103993485666974,
      "learning_rate": 6.038372858136401e-06,
      "loss": 0.3672,
      "step": 6807
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.6915719836856,
      "learning_rate": 6.035573596898789e-06,
      "loss": 0.2961,
      "step": 6808
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.8092360804628103,
      "learning_rate": 6.0327747041874075e-06,
      "loss": 0.2874,
      "step": 6809
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.8079455888866176,
      "learning_rate": 6.029976180262431e-06,
      "loss": 0.353,
      "step": 6810
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.9202453682273446,
      "learning_rate": 6.027178025384011e-06,
      "loss": 0.4009,
      "step": 6811
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.7084311854984164,
      "learning_rate": 6.024380239812258e-06,
      "loss": 0.3162,
      "step": 6812
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.7908855323137902,
      "learning_rate": 6.0215828238072446e-06,
      "loss": 0.321,
      "step": 6813
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.8442388184804177,
      "learning_rate": 6.018785777629017e-06,
      "loss": 0.333,
      "step": 6814
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.6851824432061062,
      "learning_rate": 6.015989101537586e-06,
      "loss": 0.2979,
      "step": 6815
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.9620498863229785,
      "learning_rate": 6.013192795792922e-06,
      "loss": 0.3123,
      "step": 6816
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.7033032489658075,
      "learning_rate": 6.01039686065497e-06,
      "loss": 0.3435,
      "step": 6817
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.7964310092598308,
      "learning_rate": 6.007601296383629e-06,
      "loss": 0.3535,
      "step": 6818
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.5425770576990059,
      "learning_rate": 6.004806103238771e-06,
      "loss": 0.272,
      "step": 6819
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.8656992230035396,
      "learning_rate": 6.0020112814802355e-06,
      "loss": 0.3103,
      "step": 6820
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.9833255813931647,
      "learning_rate": 5.9992168313678264e-06,
      "loss": 0.3164,
      "step": 6821
    },
    {
      "epoch": 1.93,
      "grad_norm": 2.1162738006706805,
      "learning_rate": 5.996422753161304e-06,
      "loss": 0.5269,
      "step": 6822
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.7339713165040183,
      "learning_rate": 5.993629047120403e-06,
      "loss": 0.2705,
      "step": 6823
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.999848770568832,
      "learning_rate": 5.990835713504825e-06,
      "loss": 0.436,
      "step": 6824
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.6242814640882282,
      "learning_rate": 5.988042752574233e-06,
      "loss": 0.3057,
      "step": 6825
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.7928759978930047,
      "learning_rate": 5.9852501645882485e-06,
      "loss": 0.2764,
      "step": 6826
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.8789921565993906,
      "learning_rate": 5.9824579498064725e-06,
      "loss": 0.3726,
      "step": 6827
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.9557577043771366,
      "learning_rate": 5.979666108488464e-06,
      "loss": 0.3794,
      "step": 6828
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.7418603364845755,
      "learning_rate": 5.976874640893751e-06,
      "loss": 0.2805,
      "step": 6829
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.69154430375148,
      "learning_rate": 5.9740835472818145e-06,
      "loss": 0.3325,
      "step": 6830
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.84545710445257,
      "learning_rate": 5.971292827912117e-06,
      "loss": 0.3608,
      "step": 6831
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.7773441365176983,
      "learning_rate": 5.968502483044074e-06,
      "loss": 0.4058,
      "step": 6832
    },
    {
      "epoch": 1.93,
      "grad_norm": 2.082979187898115,
      "learning_rate": 5.965712512937074e-06,
      "loss": 0.3149,
      "step": 6833
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.9703223047898295,
      "learning_rate": 5.962922917850471e-06,
      "loss": 0.4194,
      "step": 6834
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.9159255931648365,
      "learning_rate": 5.9601336980435756e-06,
      "loss": 0.3569,
      "step": 6835
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.9700527158771743,
      "learning_rate": 5.957344853775668e-06,
      "loss": 0.356,
      "step": 6836
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.9305623311907711,
      "learning_rate": 5.954556385306001e-06,
      "loss": 0.3994,
      "step": 6837
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.5886764768621637,
      "learning_rate": 5.9517682928937825e-06,
      "loss": 0.2791,
      "step": 6838
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.8584023401206873,
      "learning_rate": 5.9489805767981845e-06,
      "loss": 0.3325,
      "step": 6839
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.5672713843239987,
      "learning_rate": 5.946193237278352e-06,
      "loss": 0.2668,
      "step": 6840
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.816568884764327,
      "learning_rate": 5.9434062745933926e-06,
      "loss": 0.3103,
      "step": 6841
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.8772204731662605,
      "learning_rate": 5.9406196890023786e-06,
      "loss": 0.3105,
      "step": 6842
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.8984408105540618,
      "learning_rate": 5.937833480764339e-06,
      "loss": 0.3462,
      "step": 6843
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.738414514188337,
      "learning_rate": 5.935047650138282e-06,
      "loss": 0.2874,
      "step": 6844
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.7183737004456772,
      "learning_rate": 5.93226219738317e-06,
      "loss": 0.3057,
      "step": 6845
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.898501795070127,
      "learning_rate": 5.929477122757938e-06,
      "loss": 0.3494,
      "step": 6846
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.8811997386769415,
      "learning_rate": 5.926692426521474e-06,
      "loss": 0.3701,
      "step": 6847
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.7829319206486889,
      "learning_rate": 5.923908108932648e-06,
      "loss": 0.3157,
      "step": 6848
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.8187831152371277,
      "learning_rate": 5.921124170250276e-06,
      "loss": 0.2993,
      "step": 6849
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.807172629700605,
      "learning_rate": 5.918340610733154e-06,
      "loss": 0.3818,
      "step": 6850
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.5908159074424173,
      "learning_rate": 5.91555743064004e-06,
      "loss": 0.2664,
      "step": 6851
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.7596248440773732,
      "learning_rate": 5.912774630229648e-06,
      "loss": 0.2808,
      "step": 6852
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.9701557717668154,
      "learning_rate": 5.90999220976066e-06,
      "loss": 0.3027,
      "step": 6853
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.8184643170128378,
      "learning_rate": 5.9072101694917305e-06,
      "loss": 0.3655,
      "step": 6854
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.785342449166781,
      "learning_rate": 5.904428509681473e-06,
      "loss": 0.3672,
      "step": 6855
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.9399606984420463,
      "learning_rate": 5.901647230588461e-06,
      "loss": 0.4072,
      "step": 6856
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.758623498900255,
      "learning_rate": 5.898866332471241e-06,
      "loss": 0.2751,
      "step": 6857
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.640195706458956,
      "learning_rate": 5.896085815588321e-06,
      "loss": 0.3281,
      "step": 6858
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.6639899349439757,
      "learning_rate": 5.893305680198175e-06,
      "loss": 0.2676,
      "step": 6859
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.8392135004425243,
      "learning_rate": 5.8905259265592315e-06,
      "loss": 0.3794,
      "step": 6860
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.9080499235101611,
      "learning_rate": 5.887746554929899e-06,
      "loss": 0.3521,
      "step": 6861
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.7724499358160868,
      "learning_rate": 5.884967565568539e-06,
      "loss": 0.3462,
      "step": 6862
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.6346677923785524,
      "learning_rate": 5.882188958733488e-06,
      "loss": 0.2703,
      "step": 6863
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.7739969167630536,
      "learning_rate": 5.87941073468303e-06,
      "loss": 0.3267,
      "step": 6864
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.4950626448712436,
      "learning_rate": 5.876632893675432e-06,
      "loss": 0.281,
      "step": 6865
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.8233357754851809,
      "learning_rate": 5.873855435968913e-06,
      "loss": 0.2949,
      "step": 6866
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.828723746483499,
      "learning_rate": 5.871078361821667e-06,
      "loss": 0.3237,
      "step": 6867
    },
    {
      "epoch": 1.94,
      "grad_norm": 2.0966951851207436,
      "learning_rate": 5.8683016714918416e-06,
      "loss": 0.4082,
      "step": 6868
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.6033239503204038,
      "learning_rate": 5.86552536523755e-06,
      "loss": 0.3176,
      "step": 6869
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.6616210818501629,
      "learning_rate": 5.8627494433168756e-06,
      "loss": 0.3105,
      "step": 6870
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.803801404360103,
      "learning_rate": 5.859973905987866e-06,
      "loss": 0.3433,
      "step": 6871
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.8231713496581672,
      "learning_rate": 5.8571987535085286e-06,
      "loss": 0.3423,
      "step": 6872
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.7984664603532885,
      "learning_rate": 5.8544239861368346e-06,
      "loss": 0.3198,
      "step": 6873
    },
    {
      "epoch": 1.94,
      "grad_norm": 2.028953698858567,
      "learning_rate": 5.851649604130723e-06,
      "loss": 0.3523,
      "step": 6874
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.8293526515013239,
      "learning_rate": 5.848875607748095e-06,
      "loss": 0.2974,
      "step": 6875
    },
    {
      "epoch": 1.94,
      "grad_norm": 2.2504737390447715,
      "learning_rate": 5.846101997246822e-06,
      "loss": 0.4136,
      "step": 6876
    },
    {
      "epoch": 1.94,
      "grad_norm": 2.0226858853673013,
      "learning_rate": 5.8433287728847244e-06,
      "loss": 0.4048,
      "step": 6877
    },
    {
      "epoch": 1.94,
      "grad_norm": 2.262620169765628,
      "learning_rate": 5.840555934919604e-06,
      "loss": 0.439,
      "step": 6878
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.838091718962648,
      "learning_rate": 5.837783483609214e-06,
      "loss": 0.3657,
      "step": 6879
    },
    {
      "epoch": 1.94,
      "grad_norm": 2.168976978946976,
      "learning_rate": 5.835011419211285e-06,
      "loss": 0.3789,
      "step": 6880
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.6775005938355703,
      "learning_rate": 5.83223974198349e-06,
      "loss": 0.3123,
      "step": 6881
    },
    {
      "epoch": 1.94,
      "grad_norm": 2.2622324992965366,
      "learning_rate": 5.829468452183489e-06,
      "loss": 0.5015,
      "step": 6882
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.6745346410256994,
      "learning_rate": 5.8266975500688945e-06,
      "loss": 0.3152,
      "step": 6883
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.8721726065236395,
      "learning_rate": 5.823927035897285e-06,
      "loss": 0.3594,
      "step": 6884
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.84229735063832,
      "learning_rate": 5.821156909926202e-06,
      "loss": 0.3213,
      "step": 6885
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.6667191886269863,
      "learning_rate": 5.818387172413151e-06,
      "loss": 0.3108,
      "step": 6886
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.775846680393937,
      "learning_rate": 5.815617823615599e-06,
      "loss": 0.3691,
      "step": 6887
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.6483908844125725,
      "learning_rate": 5.812848863790988e-06,
      "loss": 0.2981,
      "step": 6888
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.7550838403619218,
      "learning_rate": 5.810080293196709e-06,
      "loss": 0.325,
      "step": 6889
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.9207517921400934,
      "learning_rate": 5.807312112090129e-06,
      "loss": 0.313,
      "step": 6890
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.6836828025281771,
      "learning_rate": 5.804544320728567e-06,
      "loss": 0.3457,
      "step": 6891
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.7171580071019565,
      "learning_rate": 5.801776919369317e-06,
      "loss": 0.3081,
      "step": 6892
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.616800688391652,
      "learning_rate": 5.799009908269629e-06,
      "loss": 0.2593,
      "step": 6893
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.641871073650064,
      "learning_rate": 5.7962432876867215e-06,
      "loss": 0.2954,
      "step": 6894
    },
    {
      "epoch": 1.95,
      "grad_norm": 2.084497576915022,
      "learning_rate": 5.79347705787777e-06,
      "loss": 0.3853,
      "step": 6895
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.8320087572188097,
      "learning_rate": 5.790711219099925e-06,
      "loss": 0.3369,
      "step": 6896
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.956890587060912,
      "learning_rate": 5.787945771610296e-06,
      "loss": 0.3633,
      "step": 6897
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.6016934486402314,
      "learning_rate": 5.785180715665941e-06,
      "loss": 0.2981,
      "step": 6898
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.686899896992318,
      "learning_rate": 5.782416051523909e-06,
      "loss": 0.3208,
      "step": 6899
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.9650252618299178,
      "learning_rate": 5.779651779441192e-06,
      "loss": 0.3169,
      "step": 6900
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.6638491851686539,
      "learning_rate": 5.776887899674752e-06,
      "loss": 0.2876,
      "step": 6901
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.8937011968908826,
      "learning_rate": 5.774124412481517e-06,
      "loss": 0.394,
      "step": 6902
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.9272786783597553,
      "learning_rate": 5.771361318118374e-06,
      "loss": 0.3721,
      "step": 6903
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.4486196333364747,
      "learning_rate": 5.76859861684217e-06,
      "loss": 0.2544,
      "step": 6904
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.860359183077816,
      "learning_rate": 5.765836308909731e-06,
      "loss": 0.3545,
      "step": 6905
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.4375275950204949,
      "learning_rate": 5.763074394577835e-06,
      "loss": 0.2651,
      "step": 6906
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.6217273591005414,
      "learning_rate": 5.760312874103219e-06,
      "loss": 0.3525,
      "step": 6907
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.7107283460324503,
      "learning_rate": 5.757551747742592e-06,
      "loss": 0.3208,
      "step": 6908
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.603122612651583,
      "learning_rate": 5.754791015752624e-06,
      "loss": 0.29,
      "step": 6909
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.890929628439709,
      "learning_rate": 5.752030678389948e-06,
      "loss": 0.3521,
      "step": 6910
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.7894700651675735,
      "learning_rate": 5.749270735911159e-06,
      "loss": 0.3054,
      "step": 6911
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.7461662795693682,
      "learning_rate": 5.746511188572814e-06,
      "loss": 0.3462,
      "step": 6912
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.7696212118412318,
      "learning_rate": 5.743752036631443e-06,
      "loss": 0.3198,
      "step": 6913
    },
    {
      "epoch": 1.95,
      "grad_norm": 2.0239962117076558,
      "learning_rate": 5.740993280343529e-06,
      "loss": 0.3921,
      "step": 6914
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.7678326812440586,
      "learning_rate": 5.738234919965516e-06,
      "loss": 0.3999,
      "step": 6915
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.688018151924628,
      "learning_rate": 5.735476955753821e-06,
      "loss": 0.2783,
      "step": 6916
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.8122290049014393,
      "learning_rate": 5.732719387964822e-06,
      "loss": 0.3521,
      "step": 6917
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.97344766367834,
      "learning_rate": 5.729962216854853e-06,
      "loss": 0.3657,
      "step": 6918
    },
    {
      "epoch": 1.95,
      "grad_norm": 2.1057744674012504,
      "learning_rate": 5.727205442680218e-06,
      "loss": 0.4634,
      "step": 6919
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.727701060695548,
      "learning_rate": 5.724449065697182e-06,
      "loss": 0.3374,
      "step": 6920
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.8853297297066278,
      "learning_rate": 5.721693086161968e-06,
      "loss": 0.3274,
      "step": 6921
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.6115746926809886,
      "learning_rate": 5.7189375043307775e-06,
      "loss": 0.2358,
      "step": 6922
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.6437288197944322,
      "learning_rate": 5.71618232045976e-06,
      "loss": 0.2979,
      "step": 6923
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.854111175983552,
      "learning_rate": 5.713427534805025e-06,
      "loss": 0.3337,
      "step": 6924
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.857299114061296,
      "learning_rate": 5.710673147622664e-06,
      "loss": 0.3311,
      "step": 6925
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.8846589507020461,
      "learning_rate": 5.707919159168714e-06,
      "loss": 0.312,
      "step": 6926
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.9713216180944022,
      "learning_rate": 5.7051655696991825e-06,
      "loss": 0.4258,
      "step": 6927
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.6941544503783725,
      "learning_rate": 5.702412379470039e-06,
      "loss": 0.2876,
      "step": 6928
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.8634690317257867,
      "learning_rate": 5.6996595887372145e-06,
      "loss": 0.3643,
      "step": 6929
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.913857667298943,
      "learning_rate": 5.696907197756598e-06,
      "loss": 0.4287,
      "step": 6930
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.6686827399087054,
      "learning_rate": 5.694155206784061e-06,
      "loss": 0.2947,
      "step": 6931
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.7499772106758698,
      "learning_rate": 5.691403616075409e-06,
      "loss": 0.3359,
      "step": 6932
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.9734650067907618,
      "learning_rate": 5.688652425886434e-06,
      "loss": 0.3896,
      "step": 6933
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.6902212930451819,
      "learning_rate": 5.6859016364728795e-06,
      "loss": 0.2952,
      "step": 6934
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.854410924340719,
      "learning_rate": 5.683151248090455e-06,
      "loss": 0.3286,
      "step": 6935
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.769414923453271,
      "learning_rate": 5.68040126099483e-06,
      "loss": 0.3364,
      "step": 6936
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.9975382478546404,
      "learning_rate": 5.67765167544164e-06,
      "loss": 0.3997,
      "step": 6937
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.695517442001022,
      "learning_rate": 5.674902491686477e-06,
      "loss": 0.3091,
      "step": 6938
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.919570926902795,
      "learning_rate": 5.672153709984909e-06,
      "loss": 0.3721,
      "step": 6939
    },
    {
      "epoch": 1.96,
      "grad_norm": 2.172396295322597,
      "learning_rate": 5.669405330592457e-06,
      "loss": 0.4185,
      "step": 6940
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.779093138285216,
      "learning_rate": 5.666657353764594e-06,
      "loss": 0.3125,
      "step": 6941
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.8083758340943372,
      "learning_rate": 5.66390977975678e-06,
      "loss": 0.3604,
      "step": 6942
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.8774530795594058,
      "learning_rate": 5.66116260882442e-06,
      "loss": 0.4097,
      "step": 6943
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.7535996006625791,
      "learning_rate": 5.658415841222887e-06,
      "loss": 0.3008,
      "step": 6944
    },
    {
      "epoch": 1.96,
      "grad_norm": 2.0366183174435077,
      "learning_rate": 5.655669477207513e-06,
      "loss": 0.3716,
      "step": 6945
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.7451446928535,
      "learning_rate": 5.6529235170335995e-06,
      "loss": 0.3096,
      "step": 6946
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.8429706416359948,
      "learning_rate": 5.650177960956399e-06,
      "loss": 0.2979,
      "step": 6947
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.631001511886939,
      "learning_rate": 5.647432809231147e-06,
      "loss": 0.2468,
      "step": 6948
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.8186902015847832,
      "learning_rate": 5.644688062113012e-06,
      "loss": 0.3745,
      "step": 6949
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.705885704091943,
      "learning_rate": 5.6419437198571525e-06,
      "loss": 0.2874,
      "step": 6950
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.7524673286977503,
      "learning_rate": 5.639199782718674e-06,
      "loss": 0.3171,
      "step": 6951
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.745187428080099,
      "learning_rate": 5.6364562509526474e-06,
      "loss": 0.3291,
      "step": 6952
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.7267749256086296,
      "learning_rate": 5.6337131248141085e-06,
      "loss": 0.3479,
      "step": 6953
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.6680445847619723,
      "learning_rate": 5.630970404558053e-06,
      "loss": 0.291,
      "step": 6954
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.9426646834547425,
      "learning_rate": 5.628228090439434e-06,
      "loss": 0.4194,
      "step": 6955
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.541774578608236,
      "learning_rate": 5.625486182713181e-06,
      "loss": 0.2764,
      "step": 6956
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.7079012129087336,
      "learning_rate": 5.622744681634178e-06,
      "loss": 0.3105,
      "step": 6957
    },
    {
      "epoch": 1.96,
      "grad_norm": 2.027403304609562,
      "learning_rate": 5.620003587457258e-06,
      "loss": 0.4092,
      "step": 6958
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.9071741771742126,
      "learning_rate": 5.617262900437239e-06,
      "loss": 0.3323,
      "step": 6959
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.9096885768064873,
      "learning_rate": 5.6145226208288875e-06,
      "loss": 0.3394,
      "step": 6960
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.8042214356498851,
      "learning_rate": 5.611782748886935e-06,
      "loss": 0.3416,
      "step": 6961
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.6699524729472952,
      "learning_rate": 5.609043284866076e-06,
      "loss": 0.3223,
      "step": 6962
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.7485190929650665,
      "learning_rate": 5.606304229020965e-06,
      "loss": 0.354,
      "step": 6963
    },
    {
      "epoch": 1.97,
      "grad_norm": 2.2692321682698124,
      "learning_rate": 5.603565581606216e-06,
      "loss": 0.4006,
      "step": 6964
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.9000568826970243,
      "learning_rate": 5.600827342876422e-06,
      "loss": 0.3455,
      "step": 6965
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.7298341550871748,
      "learning_rate": 5.598089513086108e-06,
      "loss": 0.3564,
      "step": 6966
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.673116675865626,
      "learning_rate": 5.595352092489791e-06,
      "loss": 0.3357,
      "step": 6967
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.961346821675957,
      "learning_rate": 5.592615081341932e-06,
      "loss": 0.332,
      "step": 6968
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.6996329707148565,
      "learning_rate": 5.589878479896959e-06,
      "loss": 0.3523,
      "step": 6969
    },
    {
      "epoch": 1.97,
      "grad_norm": 2.0629203967043472,
      "learning_rate": 5.587142288409262e-06,
      "loss": 0.4053,
      "step": 6970
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.733011380091027,
      "learning_rate": 5.584406507133191e-06,
      "loss": 0.2969,
      "step": 6971
    },
    {
      "epoch": 1.97,
      "grad_norm": 2.020080843818175,
      "learning_rate": 5.581671136323059e-06,
      "loss": 0.3127,
      "step": 6972
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.8921768634546563,
      "learning_rate": 5.578936176233149e-06,
      "loss": 0.3716,
      "step": 6973
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.6723175686085356,
      "learning_rate": 5.576201627117691e-06,
      "loss": 0.2607,
      "step": 6974
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.8611655629828565,
      "learning_rate": 5.573467489230879e-06,
      "loss": 0.3254,
      "step": 6975
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.736039165929246,
      "learning_rate": 5.5707337628268864e-06,
      "loss": 0.2856,
      "step": 6976
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.8508178446447603,
      "learning_rate": 5.568000448159828e-06,
      "loss": 0.3711,
      "step": 6977
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.7423802402537525,
      "learning_rate": 5.56526754548379e-06,
      "loss": 0.3262,
      "step": 6978
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.825956966157743,
      "learning_rate": 5.562535055052818e-06,
      "loss": 0.3918,
      "step": 6979
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.7625101938582557,
      "learning_rate": 5.559802977120918e-06,
      "loss": 0.2791,
      "step": 6980
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.7532818440038245,
      "learning_rate": 5.557071311942057e-06,
      "loss": 0.3242,
      "step": 6981
    },
    {
      "epoch": 1.97,
      "grad_norm": 2.1284974495997644,
      "learning_rate": 5.554340059770178e-06,
      "loss": 0.4099,
      "step": 6982
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.9758152333893715,
      "learning_rate": 5.55160922085916e-06,
      "loss": 0.354,
      "step": 6983
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.973255401855351,
      "learning_rate": 5.54887879546286e-06,
      "loss": 0.4165,
      "step": 6984
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.9125417513871066,
      "learning_rate": 5.546148783835097e-06,
      "loss": 0.3965,
      "step": 6985
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.998990808042272,
      "learning_rate": 5.5434191862296485e-06,
      "loss": 0.3633,
      "step": 6986
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.5600321562464374,
      "learning_rate": 5.5406900029002504e-06,
      "loss": 0.2756,
      "step": 6987
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.9352177009522384,
      "learning_rate": 5.5379612341006054e-06,
      "loss": 0.4004,
      "step": 6988
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.9117037742473297,
      "learning_rate": 5.535232880084373e-06,
      "loss": 0.3271,
      "step": 6989
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.8822669971385242,
      "learning_rate": 5.532504941105176e-06,
      "loss": 0.3125,
      "step": 6990
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.7996498984551714,
      "learning_rate": 5.529777417416599e-06,
      "loss": 0.3457,
      "step": 6991
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.5980206658015248,
      "learning_rate": 5.527050309272187e-06,
      "loss": 0.2825,
      "step": 6992
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.7125419026436677,
      "learning_rate": 5.524323616925452e-06,
      "loss": 0.3286,
      "step": 6993
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.7714217010516942,
      "learning_rate": 5.52159734062986e-06,
      "loss": 0.356,
      "step": 6994
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.8347184659536693,
      "learning_rate": 5.5188714806388395e-06,
      "loss": 0.313,
      "step": 6995
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.8525705193310644,
      "learning_rate": 5.516146037205784e-06,
      "loss": 0.3594,
      "step": 6996
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.7198645355675641,
      "learning_rate": 5.513421010584044e-06,
      "loss": 0.3521,
      "step": 6997
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.7543968995315007,
      "learning_rate": 5.51069640102693e-06,
      "loss": 0.2976,
      "step": 6998
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.7279275775963099,
      "learning_rate": 5.507972208787728e-06,
      "loss": 0.2861,
      "step": 6999
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.8032282837376028,
      "learning_rate": 5.505248434119666e-06,
      "loss": 0.3096,
      "step": 7000
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.8244752552205663,
      "learning_rate": 5.502525077275937e-06,
      "loss": 0.3672,
      "step": 7001
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.6075468423437718,
      "learning_rate": 5.499802138509709e-06,
      "loss": 0.3469,
      "step": 7002
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.7354584825228518,
      "learning_rate": 5.497079618074097e-06,
      "loss": 0.3411,
      "step": 7003
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.606609906283914,
      "learning_rate": 5.494357516222184e-06,
      "loss": 0.2932,
      "step": 7004
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.7779735199061752,
      "learning_rate": 5.49163583320701e-06,
      "loss": 0.3992,
      "step": 7005
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.9112988030839106,
      "learning_rate": 5.488914569281578e-06,
      "loss": 0.3979,
      "step": 7006
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.5942494182878384,
      "learning_rate": 5.486193724698854e-06,
      "loss": 0.3159,
      "step": 7007
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.9716156138654726,
      "learning_rate": 5.48347329971176e-06,
      "loss": 0.4089,
      "step": 7008
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.6568709094209977,
      "learning_rate": 5.48075329457318e-06,
      "loss": 0.3091,
      "step": 7009
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.7739280296462252,
      "learning_rate": 5.478033709535968e-06,
      "loss": 0.3306,
      "step": 7010
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.9598022690879335,
      "learning_rate": 5.4753145448529284e-06,
      "loss": 0.3306,
      "step": 7011
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.7345806174716731,
      "learning_rate": 5.472595800776831e-06,
      "loss": 0.2998,
      "step": 7012
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.7596915203046697,
      "learning_rate": 5.469877477560403e-06,
      "loss": 0.3721,
      "step": 7013
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.8473644463544017,
      "learning_rate": 5.467159575456338e-06,
      "loss": 0.3345,
      "step": 7014
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.814607747129295,
      "learning_rate": 5.464442094717281e-06,
      "loss": 0.3428,
      "step": 7015
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.6795760845298293,
      "learning_rate": 5.46172503559586e-06,
      "loss": 0.3145,
      "step": 7016
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.8146958851808443,
      "learning_rate": 5.45900839834463e-06,
      "loss": 0.3491,
      "step": 7017
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.8654428964168723,
      "learning_rate": 5.45629218321613e-06,
      "loss": 0.3062,
      "step": 7018
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.954267112637655,
      "learning_rate": 5.453576390462861e-06,
      "loss": 0.3438,
      "step": 7019
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.794743968064144,
      "learning_rate": 5.4508610203372794e-06,
      "loss": 0.3267,
      "step": 7020
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.6181471763482733,
      "learning_rate": 5.448146073091787e-06,
      "loss": 0.2974,
      "step": 7021
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.727118994569276,
      "learning_rate": 5.445431548978775e-06,
      "loss": 0.3271,
      "step": 7022
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.6108561462213813,
      "learning_rate": 5.442717448250574e-06,
      "loss": 0.2939,
      "step": 7023
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.8768192950834608,
      "learning_rate": 5.440003771159485e-06,
      "loss": 0.4243,
      "step": 7024
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.5855608982701381,
      "learning_rate": 5.437290517957767e-06,
      "loss": 0.2803,
      "step": 7025
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.6745099728360604,
      "learning_rate": 5.434577688897637e-06,
      "loss": 0.2759,
      "step": 7026
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.6702257323775236,
      "learning_rate": 5.4318652842312725e-06,
      "loss": 0.2939,
      "step": 7027
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.928076280089567,
      "learning_rate": 5.429153304210821e-06,
      "loss": 0.322,
      "step": 7028
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.8256998966349391,
      "learning_rate": 5.42644174908838e-06,
      "loss": 0.3179,
      "step": 7029
    },
    {
      "epoch": 1.99,
      "grad_norm": 2.154731943356032,
      "learning_rate": 5.42373061911601e-06,
      "loss": 0.4004,
      "step": 7030
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.7755675941266111,
      "learning_rate": 5.421019914545735e-06,
      "loss": 0.3501,
      "step": 7031
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.9638770771996887,
      "learning_rate": 5.418309635629536e-06,
      "loss": 0.3223,
      "step": 7032
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.7607057516655518,
      "learning_rate": 5.415599782619356e-06,
      "loss": 0.3052,
      "step": 7033
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.9111183697972292,
      "learning_rate": 5.412890355767098e-06,
      "loss": 0.3274,
      "step": 7034
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.940240794507028,
      "learning_rate": 5.410181355324622e-06,
      "loss": 0.3357,
      "step": 7035
    },
    {
      "epoch": 1.99,
      "grad_norm": 2.0230652828486337,
      "learning_rate": 5.407472781543761e-06,
      "loss": 0.3911,
      "step": 7036
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.691471032125269,
      "learning_rate": 5.404764634676297e-06,
      "loss": 0.3059,
      "step": 7037
    },
    {
      "epoch": 1.99,
      "grad_norm": 2.043028728125996,
      "learning_rate": 5.402056914973965e-06,
      "loss": 0.3901,
      "step": 7038
    },
    {
      "epoch": 1.99,
      "grad_norm": 2.0063434823356756,
      "learning_rate": 5.399349622688479e-06,
      "loss": 0.3901,
      "step": 7039
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.86208224671136,
      "learning_rate": 5.3966427580715044e-06,
      "loss": 0.3032,
      "step": 7040
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.617381948865882,
      "learning_rate": 5.393936321374663e-06,
      "loss": 0.3091,
      "step": 7041
    },
    {
      "epoch": 1.99,
      "grad_norm": 2.0014128413583765,
      "learning_rate": 5.391230312849543e-06,
      "loss": 0.3452,
      "step": 7042
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.734508139804231,
      "learning_rate": 5.388524732747688e-06,
      "loss": 0.2654,
      "step": 7043
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.7734907085034033,
      "learning_rate": 5.385819581320602e-06,
      "loss": 0.2832,
      "step": 7044
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.7707724366130906,
      "learning_rate": 5.3831148588197615e-06,
      "loss": 0.3218,
      "step": 7045
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.831130143058288,
      "learning_rate": 5.3804105654965784e-06,
      "loss": 0.3398,
      "step": 7046
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.7482420483074401,
      "learning_rate": 5.3777067016024495e-06,
      "loss": 0.3113,
      "step": 7047
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.8898454407834242,
      "learning_rate": 5.375003267388719e-06,
      "loss": 0.3887,
      "step": 7048
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.637644317108901,
      "learning_rate": 5.372300263106693e-06,
      "loss": 0.2573,
      "step": 7049
    },
    {
      "epoch": 1.99,
      "grad_norm": 2.029318665874101,
      "learning_rate": 5.3695976890076375e-06,
      "loss": 0.395,
      "step": 7050
    },
    {
      "epoch": 1.99,
      "grad_norm": 2.0657877723244775,
      "learning_rate": 5.3668955453427795e-06,
      "loss": 0.3677,
      "step": 7051
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.8515449306435883,
      "learning_rate": 5.3641938323633025e-06,
      "loss": 0.3711,
      "step": 7052
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.8980034409113462,
      "learning_rate": 5.3614925503203586e-06,
      "loss": 0.3184,
      "step": 7053
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.7467537505349033,
      "learning_rate": 5.358791699465056e-06,
      "loss": 0.3257,
      "step": 7054
    },
    {
      "epoch": 1.99,
      "grad_norm": 2.0658185615598934,
      "learning_rate": 5.35609128004845e-06,
      "loss": 0.4448,
      "step": 7055
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.822665825827835,
      "learning_rate": 5.353391292321577e-06,
      "loss": 0.3237,
      "step": 7056
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.708504474799417,
      "learning_rate": 5.3506917365354205e-06,
      "loss": 0.323,
      "step": 7057
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.8028268162102488,
      "learning_rate": 5.347992612940927e-06,
      "loss": 0.2969,
      "step": 7058
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.766405551785207,
      "learning_rate": 5.345293921789e-06,
      "loss": 0.3506,
      "step": 7059
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.8226560610624163,
      "learning_rate": 5.3425956633305075e-06,
      "loss": 0.344,
      "step": 7060
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.8821449552789955,
      "learning_rate": 5.3398978378162696e-06,
      "loss": 0.3628,
      "step": 7061
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.7473748666889608,
      "learning_rate": 5.337200445497084e-06,
      "loss": 0.2925,
      "step": 7062
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.7885894022693984,
      "learning_rate": 5.33450348662368e-06,
      "loss": 0.3052,
      "step": 7063
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.816610807550774,
      "learning_rate": 5.331806961446774e-06,
      "loss": 0.2915,
      "step": 7064
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.9797817120249073,
      "learning_rate": 5.3291108702170244e-06,
      "loss": 0.3369,
      "step": 7065
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.8646975355813156,
      "learning_rate": 5.3264152131850585e-06,
      "loss": 0.3293,
      "step": 7066
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.850009548659363,
      "learning_rate": 5.323719990601459e-06,
      "loss": 0.3252,
      "step": 7067
    },
    {
      "epoch": 2.0,
      "grad_norm": 2.0633139829864504,
      "learning_rate": 5.321025202716769e-06,
      "loss": 0.3823,
      "step": 7068
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.812635728882181,
      "learning_rate": 5.318330849781485e-06,
      "loss": 0.3447,
      "step": 7069
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.8326539470611922,
      "learning_rate": 5.3156369320460796e-06,
      "loss": 0.356,
      "step": 7070
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.8343432704532163,
      "learning_rate": 5.312943449760975e-06,
      "loss": 0.3762,
      "step": 7071
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.6087372580038144,
      "learning_rate": 5.31025040317654e-06,
      "loss": 0.3171,
      "step": 7072
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.6865834680793974,
      "learning_rate": 5.307557792543128e-06,
      "loss": 0.3057,
      "step": 7073
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.690481324010927,
      "learning_rate": 5.304865618111034e-06,
      "loss": 0.2825,
      "step": 7074
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.955529911851475,
      "learning_rate": 5.302173880130519e-06,
      "loss": 0.3564,
      "step": 7075
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.9592345447312014,
      "learning_rate": 5.299482578851803e-06,
      "loss": 0.3472,
      "step": 7076
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.9371369454770666,
      "learning_rate": 5.296791714525064e-06,
      "loss": 0.3704,
      "step": 7077
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.5930506290179531,
      "learning_rate": 5.294101287400435e-06,
      "loss": 0.2844,
      "step": 7078
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.7436555565936778,
      "learning_rate": 5.291411297728027e-06,
      "loss": 0.29,
      "step": 7079
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.982044311659687,
      "learning_rate": 5.2887217457578856e-06,
      "loss": 0.3484,
      "step": 7080
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.7685726395623196,
      "learning_rate": 5.286032631740023e-06,
      "loss": 0.2664,
      "step": 7081
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.5670188459552712,
      "learning_rate": 5.283343955924427e-06,
      "loss": 0.3098,
      "step": 7082
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.3070866809826986,
      "learning_rate": 5.280655718561026e-06,
      "loss": 0.2451,
      "step": 7083
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.392621627307803,
      "learning_rate": 5.2779679198997145e-06,
      "loss": 0.2339,
      "step": 7084
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.183471002123739,
      "learning_rate": 5.275280560190346e-06,
      "loss": 0.1821,
      "step": 7085
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.320125864854389,
      "learning_rate": 5.27259363968273e-06,
      "loss": 0.2124,
      "step": 7086
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2320982810343621,
      "learning_rate": 5.269907158626639e-06,
      "loss": 0.178,
      "step": 7087
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.4845824734680795,
      "learning_rate": 5.267221117271812e-06,
      "loss": 0.2371,
      "step": 7088
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.231245183596714,
      "learning_rate": 5.264535515867924e-06,
      "loss": 0.2139,
      "step": 7089
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.3877729084861001,
      "learning_rate": 5.261850354664633e-06,
      "loss": 0.2146,
      "step": 7090
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.4306000385375965,
      "learning_rate": 5.259165633911549e-06,
      "loss": 0.1833,
      "step": 7091
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.4237948693840947,
      "learning_rate": 5.256481353858234e-06,
      "loss": 0.1921,
      "step": 7092
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.3240894471722435,
      "learning_rate": 5.253797514754214e-06,
      "loss": 0.1909,
      "step": 7093
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.4459938275375852,
      "learning_rate": 5.2511141168489764e-06,
      "loss": 0.2302,
      "step": 7094
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.4721378295928034,
      "learning_rate": 5.248431160391963e-06,
      "loss": 0.1956,
      "step": 7095
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.301908170989291,
      "learning_rate": 5.245748645632583e-06,
      "loss": 0.1451,
      "step": 7096
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.4021922696932194,
      "learning_rate": 5.24306657282019e-06,
      "loss": 0.1702,
      "step": 7097
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.5028906312427714,
      "learning_rate": 5.240384942204105e-06,
      "loss": 0.1813,
      "step": 7098
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.8585768538545024,
      "learning_rate": 5.237703754033616e-06,
      "loss": 0.2388,
      "step": 7099
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.584973857572281,
      "learning_rate": 5.235023008557955e-06,
      "loss": 0.1836,
      "step": 7100
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.6811451963206474,
      "learning_rate": 5.232342706026323e-06,
      "loss": 0.1799,
      "step": 7101
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.77705812559102,
      "learning_rate": 5.229662846687873e-06,
      "loss": 0.1802,
      "step": 7102
    },
    {
      "epoch": 2.01,
      "grad_norm": 2.0623695445094485,
      "learning_rate": 5.226983430791722e-06,
      "loss": 0.2219,
      "step": 7103
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.9834231797419095,
      "learning_rate": 5.22430445858694e-06,
      "loss": 0.1914,
      "step": 7104
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.9810275799396944,
      "learning_rate": 5.221625930322571e-06,
      "loss": 0.2207,
      "step": 7105
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.8307999101158203,
      "learning_rate": 5.2189478462475905e-06,
      "loss": 0.1621,
      "step": 7106
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.7843139757407434,
      "learning_rate": 5.216270206610959e-06,
      "loss": 0.176,
      "step": 7107
    },
    {
      "epoch": 2.01,
      "grad_norm": 2.13504951680848,
      "learning_rate": 5.213593011661584e-06,
      "loss": 0.1642,
      "step": 7108
    },
    {
      "epoch": 2.01,
      "grad_norm": 2.1807250799978264,
      "learning_rate": 5.2109162616483325e-06,
      "loss": 0.2378,
      "step": 7109
    },
    {
      "epoch": 2.01,
      "grad_norm": 2.0671836899347635,
      "learning_rate": 5.20823995682003e-06,
      "loss": 0.1987,
      "step": 7110
    },
    {
      "epoch": 2.01,
      "grad_norm": 2.0570829708019476,
      "learning_rate": 5.205564097425458e-06,
      "loss": 0.2153,
      "step": 7111
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.822925842307161,
      "learning_rate": 5.2028886837133605e-06,
      "loss": 0.1678,
      "step": 7112
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.5483095694595062,
      "learning_rate": 5.200213715932449e-06,
      "loss": 0.1412,
      "step": 7113
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.825981118464744,
      "learning_rate": 5.197539194331373e-06,
      "loss": 0.1885,
      "step": 7114
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.9983716102806108,
      "learning_rate": 5.194865119158752e-06,
      "loss": 0.2112,
      "step": 7115
    },
    {
      "epoch": 2.01,
      "grad_norm": 2.0116208047952924,
      "learning_rate": 5.192191490663168e-06,
      "loss": 0.219,
      "step": 7116
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.869974856481004,
      "learning_rate": 5.189518309093157e-06,
      "loss": 0.1713,
      "step": 7117
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.8834300292557187,
      "learning_rate": 5.18684557469721e-06,
      "loss": 0.1626,
      "step": 7118
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.958801960797903,
      "learning_rate": 5.184173287723782e-06,
      "loss": 0.2368,
      "step": 7119
    },
    {
      "epoch": 2.01,
      "grad_norm": 2.1431728074634226,
      "learning_rate": 5.1815014484212825e-06,
      "loss": 0.2371,
      "step": 7120
    },
    {
      "epoch": 2.01,
      "grad_norm": 2.221577106933762,
      "learning_rate": 5.178830057038079e-06,
      "loss": 0.2307,
      "step": 7121
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.76002821613519,
      "learning_rate": 5.1761591138225096e-06,
      "loss": 0.1846,
      "step": 7122
    },
    {
      "epoch": 2.01,
      "grad_norm": 2.0969207665847946,
      "learning_rate": 5.1734886190228496e-06,
      "loss": 0.1914,
      "step": 7123
    },
    {
      "epoch": 2.01,
      "grad_norm": 2.216663881867041,
      "learning_rate": 5.170818572887344e-06,
      "loss": 0.28,
      "step": 7124
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.6326036192728712,
      "learning_rate": 5.168148975664203e-06,
      "loss": 0.1848,
      "step": 7125
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.6788719936989531,
      "learning_rate": 5.165479827601583e-06,
      "loss": 0.1577,
      "step": 7126
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.7987796820863975,
      "learning_rate": 5.1628111289476025e-06,
      "loss": 0.165,
      "step": 7127
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.847364370342019,
      "learning_rate": 5.160142879950343e-06,
      "loss": 0.199,
      "step": 7128
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.7578444358226066,
      "learning_rate": 5.157475080857838e-06,
      "loss": 0.1481,
      "step": 7129
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.8314271237067117,
      "learning_rate": 5.154807731918081e-06,
      "loss": 0.1995,
      "step": 7130
    },
    {
      "epoch": 2.01,
      "grad_norm": 2.008093163295767,
      "learning_rate": 5.152140833379025e-06,
      "loss": 0.2185,
      "step": 7131
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.5815897971077877,
      "learning_rate": 5.1494743854885755e-06,
      "loss": 0.1824,
      "step": 7132
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.6923232595775253,
      "learning_rate": 5.1468083884946104e-06,
      "loss": 0.1581,
      "step": 7133
    },
    {
      "epoch": 2.01,
      "grad_norm": 2.251416439800534,
      "learning_rate": 5.144142842644954e-06,
      "loss": 0.2715,
      "step": 7134
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.971153748163251,
      "learning_rate": 5.14147774818738e-06,
      "loss": 0.1995,
      "step": 7135
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.6617631719944521,
      "learning_rate": 5.138813105369645e-06,
      "loss": 0.1735,
      "step": 7136
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.8360133439122412,
      "learning_rate": 5.136148914439441e-06,
      "loss": 0.2363,
      "step": 7137
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.671451913603592,
      "learning_rate": 5.133485175644426e-06,
      "loss": 0.2139,
      "step": 7138
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.9129185918958371,
      "learning_rate": 5.130821889232228e-06,
      "loss": 0.218,
      "step": 7139
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.7466733424978853,
      "learning_rate": 5.1281590554504095e-06,
      "loss": 0.1667,
      "step": 7140
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.7835989607716785,
      "learning_rate": 5.125496674546502e-06,
      "loss": 0.1899,
      "step": 7141
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.8029075697779386,
      "learning_rate": 5.122834746768004e-06,
      "loss": 0.1992,
      "step": 7142
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.5447040195627213,
      "learning_rate": 5.120173272362361e-06,
      "loss": 0.1719,
      "step": 7143
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.7515757499437874,
      "learning_rate": 5.117512251576978e-06,
      "loss": 0.2263,
      "step": 7144
    },
    {
      "epoch": 2.02,
      "grad_norm": 2.1201056012682944,
      "learning_rate": 5.114851684659219e-06,
      "loss": 0.2034,
      "step": 7145
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.9065354022524288,
      "learning_rate": 5.112191571856406e-06,
      "loss": 0.1755,
      "step": 7146
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.7468665001825718,
      "learning_rate": 5.109531913415819e-06,
      "loss": 0.1862,
      "step": 7147
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.709048312057623,
      "learning_rate": 5.1068727095846935e-06,
      "loss": 0.1741,
      "step": 7148
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.8851622143013687,
      "learning_rate": 5.104213960610223e-06,
      "loss": 0.2419,
      "step": 7149
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.9725393340481803,
      "learning_rate": 5.101555666739563e-06,
      "loss": 0.1884,
      "step": 7150
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.6873917213007728,
      "learning_rate": 5.098897828219831e-06,
      "loss": 0.16,
      "step": 7151
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.9173392407750294,
      "learning_rate": 5.09624044529808e-06,
      "loss": 0.1738,
      "step": 7152
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.6689998060481555,
      "learning_rate": 5.093583518221347e-06,
      "loss": 0.1757,
      "step": 7153
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.9423413887231562,
      "learning_rate": 5.090927047236611e-06,
      "loss": 0.1749,
      "step": 7154
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.8704688419928122,
      "learning_rate": 5.0882710325908106e-06,
      "loss": 0.179,
      "step": 7155
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.7999050677985158,
      "learning_rate": 5.085615474530854e-06,
      "loss": 0.2048,
      "step": 7156
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.6014960068505433,
      "learning_rate": 5.082960373303588e-06,
      "loss": 0.1497,
      "step": 7157
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.8043184214737247,
      "learning_rate": 5.0803057291558255e-06,
      "loss": 0.1772,
      "step": 7158
    },
    {
      "epoch": 2.02,
      "grad_norm": 2.183711398232285,
      "learning_rate": 5.0776515423343445e-06,
      "loss": 0.2424,
      "step": 7159
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.9321196873509756,
      "learning_rate": 5.074997813085873e-06,
      "loss": 0.2092,
      "step": 7160
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.8552209189238935,
      "learning_rate": 5.072344541657089e-06,
      "loss": 0.1902,
      "step": 7161
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.7459033914997184,
      "learning_rate": 5.069691728294643e-06,
      "loss": 0.166,
      "step": 7162
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.8659772643374195,
      "learning_rate": 5.067039373245136e-06,
      "loss": 0.1799,
      "step": 7163
    },
    {
      "epoch": 2.02,
      "grad_norm": 2.0258978870091533,
      "learning_rate": 5.064387476755124e-06,
      "loss": 0.2366,
      "step": 7164
    },
    {
      "epoch": 2.02,
      "grad_norm": 2.0949318398140337,
      "learning_rate": 5.061736039071124e-06,
      "loss": 0.2471,
      "step": 7165
    },
    {
      "epoch": 2.02,
      "grad_norm": 2.1298845762384815,
      "learning_rate": 5.059085060439608e-06,
      "loss": 0.2407,
      "step": 7166
    },
    {
      "epoch": 2.02,
      "grad_norm": 2.057920534904063,
      "learning_rate": 5.0564345411070025e-06,
      "loss": 0.2131,
      "step": 7167
    },
    {
      "epoch": 2.02,
      "grad_norm": 2.024626861154295,
      "learning_rate": 5.053784481319708e-06,
      "loss": 0.2107,
      "step": 7168
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.8338785752088733,
      "learning_rate": 5.0511348813240515e-06,
      "loss": 0.1956,
      "step": 7169
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.781962352519117,
      "learning_rate": 5.048485741366351e-06,
      "loss": 0.178,
      "step": 7170
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.9033572980212,
      "learning_rate": 5.045837061692857e-06,
      "loss": 0.2029,
      "step": 7171
    },
    {
      "epoch": 2.03,
      "grad_norm": 2.1889960463404208,
      "learning_rate": 5.043188842549789e-06,
      "loss": 0.2144,
      "step": 7172
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.5739444035366552,
      "learning_rate": 5.040541084183326e-06,
      "loss": 0.161,
      "step": 7173
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.8696321839816552,
      "learning_rate": 5.037893786839589e-06,
      "loss": 0.1978,
      "step": 7174
    },
    {
      "epoch": 2.03,
      "grad_norm": 2.2222664744317453,
      "learning_rate": 5.03524695076467e-06,
      "loss": 0.2334,
      "step": 7175
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.8092007180637217,
      "learning_rate": 5.032600576204616e-06,
      "loss": 0.1919,
      "step": 7176
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.6886010962022615,
      "learning_rate": 5.0299546634054345e-06,
      "loss": 0.1461,
      "step": 7177
    },
    {
      "epoch": 2.03,
      "grad_norm": 2.0422281931973494,
      "learning_rate": 5.0273092126130715e-06,
      "loss": 0.197,
      "step": 7178
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.77191642703467,
      "learning_rate": 5.024664224073454e-06,
      "loss": 0.1924,
      "step": 7179
    },
    {
      "epoch": 2.03,
      "grad_norm": 2.2338753463133023,
      "learning_rate": 5.0220196980324545e-06,
      "loss": 0.2603,
      "step": 7180
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.6515933374307332,
      "learning_rate": 5.0193756347359015e-06,
      "loss": 0.174,
      "step": 7181
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.6715866614165795,
      "learning_rate": 5.016732034429581e-06,
      "loss": 0.1663,
      "step": 7182
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.5810746370875886,
      "learning_rate": 5.014088897359242e-06,
      "loss": 0.1521,
      "step": 7183
    },
    {
      "epoch": 2.03,
      "grad_norm": 2.1185840464205663,
      "learning_rate": 5.011446223770578e-06,
      "loss": 0.1853,
      "step": 7184
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.7207851118428743,
      "learning_rate": 5.0088040139092605e-06,
      "loss": 0.1603,
      "step": 7185
    },
    {
      "epoch": 2.03,
      "grad_norm": 2.082841923340528,
      "learning_rate": 5.006162268020891e-06,
      "loss": 0.2092,
      "step": 7186
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.8910799035137238,
      "learning_rate": 5.0035209863510496e-06,
      "loss": 0.1992,
      "step": 7187
    },
    {
      "epoch": 2.03,
      "grad_norm": 2.1781273857807415,
      "learning_rate": 5.000880169145264e-06,
      "loss": 0.2085,
      "step": 7188
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.9384014810076229,
      "learning_rate": 4.99823981664902e-06,
      "loss": 0.2004,
      "step": 7189
    },
    {
      "epoch": 2.03,
      "grad_norm": 2.0981602806818196,
      "learning_rate": 4.995599929107758e-06,
      "loss": 0.2314,
      "step": 7190
    },
    {
      "epoch": 2.03,
      "grad_norm": 2.0828461131553833,
      "learning_rate": 4.992960506766881e-06,
      "loss": 0.2097,
      "step": 7191
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.971903134004784,
      "learning_rate": 4.990321549871738e-06,
      "loss": 0.1738,
      "step": 7192
    },
    {
      "epoch": 2.03,
      "grad_norm": 2.033909786727759,
      "learning_rate": 4.987683058667651e-06,
      "loss": 0.2014,
      "step": 7193
    },
    {
      "epoch": 2.03,
      "grad_norm": 2.0602140329228753,
      "learning_rate": 4.985045033399889e-06,
      "loss": 0.2483,
      "step": 7194
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.717065759510219,
      "learning_rate": 4.982407474313667e-06,
      "loss": 0.1699,
      "step": 7195
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.9476425270044684,
      "learning_rate": 4.979770381654181e-06,
      "loss": 0.2134,
      "step": 7196
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.6723966249304247,
      "learning_rate": 4.977133755666564e-06,
      "loss": 0.1674,
      "step": 7197
    },
    {
      "epoch": 2.03,
      "grad_norm": 2.034564945225967,
      "learning_rate": 4.9744975965959145e-06,
      "loss": 0.2427,
      "step": 7198
    },
    {
      "epoch": 2.03,
      "grad_norm": 2.0877912425013667,
      "learning_rate": 4.971861904687283e-06,
      "loss": 0.2395,
      "step": 7199
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.8009554159088643,
      "learning_rate": 4.9692266801856815e-06,
      "loss": 0.1763,
      "step": 7200
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.9413463795706294,
      "learning_rate": 4.96659192333607e-06,
      "loss": 0.2241,
      "step": 7201
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.9201870554106935,
      "learning_rate": 4.963957634383384e-06,
      "loss": 0.1571,
      "step": 7202
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.7870823742689954,
      "learning_rate": 4.961323813572485e-06,
      "loss": 0.1593,
      "step": 7203
    },
    {
      "epoch": 2.03,
      "grad_norm": 2.2529904028857364,
      "learning_rate": 4.958690461148222e-06,
      "loss": 0.2417,
      "step": 7204
    },
    {
      "epoch": 2.03,
      "grad_norm": 2.1998570932510826,
      "learning_rate": 4.9560575773553824e-06,
      "loss": 0.207,
      "step": 7205
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.9905724842632007,
      "learning_rate": 4.953425162438714e-06,
      "loss": 0.218,
      "step": 7206
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.4287254779666727,
      "learning_rate": 4.950793216642923e-06,
      "loss": 0.1248,
      "step": 7207
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.7601228338136128,
      "learning_rate": 4.948161740212669e-06,
      "loss": 0.2039,
      "step": 7208
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.6609405555890127,
      "learning_rate": 4.945530733392566e-06,
      "loss": 0.1731,
      "step": 7209
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.8457756042297038,
      "learning_rate": 4.942900196427195e-06,
      "loss": 0.1642,
      "step": 7210
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.889815960025252,
      "learning_rate": 4.940270129561088e-06,
      "loss": 0.2188,
      "step": 7211
    },
    {
      "epoch": 2.04,
      "grad_norm": 2.0927042933673037,
      "learning_rate": 4.937640533038718e-06,
      "loss": 0.231,
      "step": 7212
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.8200641664188209,
      "learning_rate": 4.9350114071045405e-06,
      "loss": 0.2166,
      "step": 7213
    },
    {
      "epoch": 2.04,
      "grad_norm": 2.499468739374447,
      "learning_rate": 4.932382752002951e-06,
      "loss": 0.2104,
      "step": 7214
    },
    {
      "epoch": 2.04,
      "grad_norm": 2.084645773217031,
      "learning_rate": 4.929754567978303e-06,
      "loss": 0.2251,
      "step": 7215
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.8735701032631475,
      "learning_rate": 4.92712685527491e-06,
      "loss": 0.1792,
      "step": 7216
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.8549524360922858,
      "learning_rate": 4.924499614137037e-06,
      "loss": 0.1864,
      "step": 7217
    },
    {
      "epoch": 2.04,
      "grad_norm": 2.0386149334549404,
      "learning_rate": 4.921872844808906e-06,
      "loss": 0.2324,
      "step": 7218
    },
    {
      "epoch": 2.04,
      "grad_norm": 2.306191471037278,
      "learning_rate": 4.919246547534709e-06,
      "loss": 0.2393,
      "step": 7219
    },
    {
      "epoch": 2.04,
      "grad_norm": 2.035091337640498,
      "learning_rate": 4.916620722558568e-06,
      "loss": 0.2522,
      "step": 7220
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.8638298461252207,
      "learning_rate": 4.913995370124578e-06,
      "loss": 0.2319,
      "step": 7221
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.8345229428765126,
      "learning_rate": 4.911370490476792e-06,
      "loss": 0.2109,
      "step": 7222
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.8231758654964714,
      "learning_rate": 4.908746083859214e-06,
      "loss": 0.17,
      "step": 7223
    },
    {
      "epoch": 2.04,
      "grad_norm": 2.2348569963674922,
      "learning_rate": 4.906122150515801e-06,
      "loss": 0.2231,
      "step": 7224
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.769423427175264,
      "learning_rate": 4.9034986906904715e-06,
      "loss": 0.1702,
      "step": 7225
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.7709230039467083,
      "learning_rate": 4.900875704627096e-06,
      "loss": 0.166,
      "step": 7226
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.6959012617055038,
      "learning_rate": 4.898253192569501e-06,
      "loss": 0.1901,
      "step": 7227
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.5350550257768332,
      "learning_rate": 4.8956311547614796e-06,
      "loss": 0.1266,
      "step": 7228
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.8948808061804985,
      "learning_rate": 4.893009591446759e-06,
      "loss": 0.1954,
      "step": 7229
    },
    {
      "epoch": 2.04,
      "grad_norm": 2.208242561512238,
      "learning_rate": 4.8903885028690454e-06,
      "loss": 0.1909,
      "step": 7230
    },
    {
      "epoch": 2.04,
      "grad_norm": 2.0113466549346173,
      "learning_rate": 4.887767889271987e-06,
      "loss": 0.1921,
      "step": 7231
    },
    {
      "epoch": 2.04,
      "grad_norm": 2.4038107342295523,
      "learning_rate": 4.885147750899192e-06,
      "loss": 0.2512,
      "step": 7232
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.773416606516532,
      "learning_rate": 4.882528087994223e-06,
      "loss": 0.177,
      "step": 7233
    },
    {
      "epoch": 2.04,
      "grad_norm": 2.302767490929523,
      "learning_rate": 4.8799089008006005e-06,
      "loss": 0.2292,
      "step": 7234
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.797390424158834,
      "learning_rate": 4.877290189561795e-06,
      "loss": 0.209,
      "step": 7235
    },
    {
      "epoch": 2.04,
      "grad_norm": 2.117045327400835,
      "learning_rate": 4.874671954521249e-06,
      "loss": 0.1873,
      "step": 7236
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.9210693961803988,
      "learning_rate": 4.872054195922338e-06,
      "loss": 0.1797,
      "step": 7237
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.8914195880678966,
      "learning_rate": 4.869436914008403e-06,
      "loss": 0.158,
      "step": 7238
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.8251749731860685,
      "learning_rate": 4.866820109022752e-06,
      "loss": 0.167,
      "step": 7239
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.5467235587194395,
      "learning_rate": 4.864203781208632e-06,
      "loss": 0.1455,
      "step": 7240
    },
    {
      "epoch": 2.04,
      "grad_norm": 2.1510424984968757,
      "learning_rate": 4.861587930809255e-06,
      "loss": 0.2432,
      "step": 7241
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.8169634658668088,
      "learning_rate": 4.858972558067784e-06,
      "loss": 0.1694,
      "step": 7242
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.8627005477282732,
      "learning_rate": 4.856357663227341e-06,
      "loss": 0.2075,
      "step": 7243
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.7159979182425624,
      "learning_rate": 4.8537432465309975e-06,
      "loss": 0.1256,
      "step": 7244
    },
    {
      "epoch": 2.05,
      "grad_norm": 2.6732625100280885,
      "learning_rate": 4.851129308221796e-06,
      "loss": 0.2583,
      "step": 7245
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.8876455464209128,
      "learning_rate": 4.848515848542709e-06,
      "loss": 0.2319,
      "step": 7246
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.7294105300861928,
      "learning_rate": 4.845902867736692e-06,
      "loss": 0.1565,
      "step": 7247
    },
    {
      "epoch": 2.05,
      "grad_norm": 2.1264014604222705,
      "learning_rate": 4.843290366046637e-06,
      "loss": 0.1963,
      "step": 7248
    },
    {
      "epoch": 2.05,
      "grad_norm": 2.1165640357103226,
      "learning_rate": 4.840678343715399e-06,
      "loss": 0.2219,
      "step": 7249
    },
    {
      "epoch": 2.05,
      "grad_norm": 2.0503507654268263,
      "learning_rate": 4.838066800985786e-06,
      "loss": 0.2004,
      "step": 7250
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.6770882767966575,
      "learning_rate": 4.8354557381005655e-06,
      "loss": 0.1973,
      "step": 7251
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.9006844597318358,
      "learning_rate": 4.832845155302449e-06,
      "loss": 0.1775,
      "step": 7252
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.665512143654297,
      "learning_rate": 4.830235052834127e-06,
      "loss": 0.1675,
      "step": 7253
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.6599824119515358,
      "learning_rate": 4.827625430938216e-06,
      "loss": 0.1595,
      "step": 7254
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.9264483500169667,
      "learning_rate": 4.8250162898573046e-06,
      "loss": 0.1642,
      "step": 7255
    },
    {
      "epoch": 2.05,
      "grad_norm": 2.0740830516106534,
      "learning_rate": 4.822407629833941e-06,
      "loss": 0.1941,
      "step": 7256
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.9535634374171982,
      "learning_rate": 4.81979945111062e-06,
      "loss": 0.1858,
      "step": 7257
    },
    {
      "epoch": 2.05,
      "grad_norm": 2.0784405509297814,
      "learning_rate": 4.8171917539297816e-06,
      "loss": 0.2158,
      "step": 7258
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.7043532973106055,
      "learning_rate": 4.814584538533848e-06,
      "loss": 0.1803,
      "step": 7259
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.928809288073434,
      "learning_rate": 4.811977805165174e-06,
      "loss": 0.2126,
      "step": 7260
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.5991497023509973,
      "learning_rate": 4.8093715540660745e-06,
      "loss": 0.137,
      "step": 7261
    },
    {
      "epoch": 2.05,
      "grad_norm": 2.02003359640203,
      "learning_rate": 4.806765785478833e-06,
      "loss": 0.2361,
      "step": 7262
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.7131831061467544,
      "learning_rate": 4.804160499645667e-06,
      "loss": 0.1525,
      "step": 7263
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.9861746298143992,
      "learning_rate": 4.801555696808758e-06,
      "loss": 0.2056,
      "step": 7264
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.803683273256488,
      "learning_rate": 4.798951377210253e-06,
      "loss": 0.1725,
      "step": 7265
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.8804855432623777,
      "learning_rate": 4.796347541092241e-06,
      "loss": 0.1958,
      "step": 7266
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.9744569908505194,
      "learning_rate": 4.7937441886967694e-06,
      "loss": 0.1479,
      "step": 7267
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.8027251119215317,
      "learning_rate": 4.791141320265842e-06,
      "loss": 0.151,
      "step": 7268
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.9390161477752772,
      "learning_rate": 4.788538936041415e-06,
      "loss": 0.1685,
      "step": 7269
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.833488873925283,
      "learning_rate": 4.7859370362654045e-06,
      "loss": 0.1637,
      "step": 7270
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.9685631841403002,
      "learning_rate": 4.783335621179675e-06,
      "loss": 0.2053,
      "step": 7271
    },
    {
      "epoch": 2.05,
      "grad_norm": 2.1233855500459082,
      "learning_rate": 4.780734691026051e-06,
      "loss": 0.23,
      "step": 7272
    },
    {
      "epoch": 2.05,
      "grad_norm": 2.0133344584084725,
      "learning_rate": 4.778134246046313e-06,
      "loss": 0.2212,
      "step": 7273
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.5109507656270813,
      "learning_rate": 4.775534286482197e-06,
      "loss": 0.1466,
      "step": 7274
    },
    {
      "epoch": 2.05,
      "grad_norm": 2.235935393524673,
      "learning_rate": 4.772934812575377e-06,
      "loss": 0.241,
      "step": 7275
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.9555562024882092,
      "learning_rate": 4.770335824567508e-06,
      "loss": 0.2307,
      "step": 7276
    },
    {
      "epoch": 2.06,
      "grad_norm": 2.1162412373011357,
      "learning_rate": 4.767737322700185e-06,
      "loss": 0.1936,
      "step": 7277
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.8180110719754683,
      "learning_rate": 4.765139307214956e-06,
      "loss": 0.1841,
      "step": 7278
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.7339802242898092,
      "learning_rate": 4.762541778353337e-06,
      "loss": 0.1628,
      "step": 7279
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.8754674477445972,
      "learning_rate": 4.75994473635678e-06,
      "loss": 0.2053,
      "step": 7280
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.7614784443200096,
      "learning_rate": 4.757348181466702e-06,
      "loss": 0.1799,
      "step": 7281
    },
    {
      "epoch": 2.06,
      "grad_norm": 2.1971309870274425,
      "learning_rate": 4.754752113924482e-06,
      "loss": 0.2197,
      "step": 7282
    },
    {
      "epoch": 2.06,
      "grad_norm": 2.0683986727702903,
      "learning_rate": 4.7521565339714415e-06,
      "loss": 0.2117,
      "step": 7283
    },
    {
      "epoch": 2.06,
      "grad_norm": 2.229089544510979,
      "learning_rate": 4.749561441848862e-06,
      "loss": 0.1967,
      "step": 7284
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.7449258470583218,
      "learning_rate": 4.746966837797977e-06,
      "loss": 0.1804,
      "step": 7285
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.802216285410644,
      "learning_rate": 4.744372722059978e-06,
      "loss": 0.1633,
      "step": 7286
    },
    {
      "epoch": 2.06,
      "grad_norm": 2.0724117131267095,
      "learning_rate": 4.741779094876009e-06,
      "loss": 0.2358,
      "step": 7287
    },
    {
      "epoch": 2.06,
      "grad_norm": 2.155676920310751,
      "learning_rate": 4.739185956487169e-06,
      "loss": 0.2068,
      "step": 7288
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.866474461467842,
      "learning_rate": 4.736593307134508e-06,
      "loss": 0.1687,
      "step": 7289
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.7261009530173241,
      "learning_rate": 4.7340011470590415e-06,
      "loss": 0.1699,
      "step": 7290
    },
    {
      "epoch": 2.06,
      "grad_norm": 2.308548317556166,
      "learning_rate": 4.7314094765017325e-06,
      "loss": 0.2158,
      "step": 7291
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.943677220866317,
      "learning_rate": 4.728818295703487e-06,
      "loss": 0.1824,
      "step": 7292
    },
    {
      "epoch": 2.06,
      "grad_norm": 2.0335299516923944,
      "learning_rate": 4.726227604905188e-06,
      "loss": 0.1707,
      "step": 7293
    },
    {
      "epoch": 2.06,
      "grad_norm": 3.1595595478467082,
      "learning_rate": 4.723637404347658e-06,
      "loss": 0.1874,
      "step": 7294
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.819984523932869,
      "learning_rate": 4.721047694271676e-06,
      "loss": 0.1726,
      "step": 7295
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.9975664929061814,
      "learning_rate": 4.718458474917979e-06,
      "loss": 0.1885,
      "step": 7296
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.8421081427997463,
      "learning_rate": 4.715869746527256e-06,
      "loss": 0.1827,
      "step": 7297
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.869699267554532,
      "learning_rate": 4.713281509340146e-06,
      "loss": 0.1951,
      "step": 7298
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.7183424575643031,
      "learning_rate": 4.7106937635972565e-06,
      "loss": 0.199,
      "step": 7299
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.9515051007241613,
      "learning_rate": 4.708106509539134e-06,
      "loss": 0.1942,
      "step": 7300
    },
    {
      "epoch": 2.06,
      "grad_norm": 2.2713263471849197,
      "learning_rate": 4.705519747406285e-06,
      "loss": 0.2388,
      "step": 7301
    },
    {
      "epoch": 2.06,
      "grad_norm": 2.017934092990872,
      "learning_rate": 4.702933477439172e-06,
      "loss": 0.2319,
      "step": 7302
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.9432595025556465,
      "learning_rate": 4.700347699878211e-06,
      "loss": 0.2275,
      "step": 7303
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.8590520310487553,
      "learning_rate": 4.697762414963768e-06,
      "loss": 0.196,
      "step": 7304
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.9112061452111353,
      "learning_rate": 4.695177622936169e-06,
      "loss": 0.1772,
      "step": 7305
    },
    {
      "epoch": 2.06,
      "grad_norm": 2.0358618223739953,
      "learning_rate": 4.692593324035688e-06,
      "loss": 0.1963,
      "step": 7306
    },
    {
      "epoch": 2.06,
      "grad_norm": 2.2869878977442677,
      "learning_rate": 4.690009518502564e-06,
      "loss": 0.189,
      "step": 7307
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.8694278171024485,
      "learning_rate": 4.687426206576983e-06,
      "loss": 0.1868,
      "step": 7308
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.710592675595815,
      "learning_rate": 4.684843388499075e-06,
      "loss": 0.1338,
      "step": 7309
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.8170856678837364,
      "learning_rate": 4.682261064508944e-06,
      "loss": 0.1946,
      "step": 7310
    },
    {
      "epoch": 2.06,
      "grad_norm": 2.098128488665753,
      "learning_rate": 4.679679234846636e-06,
      "loss": 0.2346,
      "step": 7311
    },
    {
      "epoch": 2.06,
      "grad_norm": 2.104195349051402,
      "learning_rate": 4.677097899752152e-06,
      "loss": 0.2954,
      "step": 7312
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.7448024916891176,
      "learning_rate": 4.674517059465449e-06,
      "loss": 0.1725,
      "step": 7313
    },
    {
      "epoch": 2.07,
      "grad_norm": 2.2481594872813986,
      "learning_rate": 4.671936714226438e-06,
      "loss": 0.2478,
      "step": 7314
    },
    {
      "epoch": 2.07,
      "grad_norm": 2.3213243039240674,
      "learning_rate": 4.66935686427498e-06,
      "loss": 0.2634,
      "step": 7315
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.974581333470498,
      "learning_rate": 4.666777509850899e-06,
      "loss": 0.2261,
      "step": 7316
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.772473495855579,
      "learning_rate": 4.664198651193969e-06,
      "loss": 0.1393,
      "step": 7317
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.8478589939433792,
      "learning_rate": 4.6616202885439056e-06,
      "loss": 0.1648,
      "step": 7318
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.9030146710463367,
      "learning_rate": 4.659042422140399e-06,
      "loss": 0.1936,
      "step": 7319
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.8670429804748865,
      "learning_rate": 4.656465052223079e-06,
      "loss": 0.1707,
      "step": 7320
    },
    {
      "epoch": 2.07,
      "grad_norm": 2.008072991078385,
      "learning_rate": 4.653888179031533e-06,
      "loss": 0.2104,
      "step": 7321
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.9706627222935273,
      "learning_rate": 4.6513118028053065e-06,
      "loss": 0.1748,
      "step": 7322
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.7007810168462383,
      "learning_rate": 4.64873592378389e-06,
      "loss": 0.1555,
      "step": 7323
    },
    {
      "epoch": 2.07,
      "grad_norm": 2.021510090282908,
      "learning_rate": 4.646160542206733e-06,
      "loss": 0.1653,
      "step": 7324
    },
    {
      "epoch": 2.07,
      "grad_norm": 2.012756410014875,
      "learning_rate": 4.6435856583132465e-06,
      "loss": 0.1921,
      "step": 7325
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.9174414571004108,
      "learning_rate": 4.641011272342775e-06,
      "loss": 0.1816,
      "step": 7326
    },
    {
      "epoch": 2.07,
      "grad_norm": 2.811040595285526,
      "learning_rate": 4.6384373845346375e-06,
      "loss": 0.21,
      "step": 7327
    },
    {
      "epoch": 2.07,
      "grad_norm": 2.2248261579211275,
      "learning_rate": 4.635863995128097e-06,
      "loss": 0.2449,
      "step": 7328
    },
    {
      "epoch": 2.07,
      "grad_norm": 2.0009460493595728,
      "learning_rate": 4.633291104362369e-06,
      "loss": 0.2219,
      "step": 7329
    },
    {
      "epoch": 2.07,
      "grad_norm": 2.1225726073363935,
      "learning_rate": 4.630718712476628e-06,
      "loss": 0.196,
      "step": 7330
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.583441038508293,
      "learning_rate": 4.628146819709997e-06,
      "loss": 0.1279,
      "step": 7331
    },
    {
      "epoch": 2.07,
      "grad_norm": 2.2548979360473465,
      "learning_rate": 4.6255754263015495e-06,
      "loss": 0.2317,
      "step": 7332
    },
    {
      "epoch": 2.07,
      "grad_norm": 2.5141423200191064,
      "learning_rate": 4.623004532490328e-06,
      "loss": 0.2595,
      "step": 7333
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.7626526520027275,
      "learning_rate": 4.6204341385153186e-06,
      "loss": 0.1456,
      "step": 7334
    },
    {
      "epoch": 2.07,
      "grad_norm": 2.0837856050954544,
      "learning_rate": 4.617864244615448e-06,
      "loss": 0.1851,
      "step": 7335
    },
    {
      "epoch": 2.07,
      "grad_norm": 2.0113789688971515,
      "learning_rate": 4.615294851029619e-06,
      "loss": 0.1731,
      "step": 7336
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.9083230650427598,
      "learning_rate": 4.612725957996677e-06,
      "loss": 0.1958,
      "step": 7337
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.7091760404098653,
      "learning_rate": 4.610157565755421e-06,
      "loss": 0.1741,
      "step": 7338
    },
    {
      "epoch": 2.07,
      "grad_norm": 2.0200570535613083,
      "learning_rate": 4.607589674544603e-06,
      "loss": 0.1636,
      "step": 7339
    },
    {
      "epoch": 2.07,
      "grad_norm": 2.296252855840107,
      "learning_rate": 4.6050222846029315e-06,
      "loss": 0.2585,
      "step": 7340
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.6402805853658213,
      "learning_rate": 4.602455396169062e-06,
      "loss": 0.1406,
      "step": 7341
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.972543979854082,
      "learning_rate": 4.59988900948162e-06,
      "loss": 0.1643,
      "step": 7342
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.802915220706626,
      "learning_rate": 4.597323124779155e-06,
      "loss": 0.1851,
      "step": 7343
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.9501864607097235,
      "learning_rate": 4.594757742300201e-06,
      "loss": 0.1472,
      "step": 7344
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.9915712908119598,
      "learning_rate": 4.592192862283228e-06,
      "loss": 0.1582,
      "step": 7345
    },
    {
      "epoch": 2.07,
      "grad_norm": 2.0601983521947522,
      "learning_rate": 4.589628484966661e-06,
      "loss": 0.1934,
      "step": 7346
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.8873587835199888,
      "learning_rate": 4.587064610588881e-06,
      "loss": 0.177,
      "step": 7347
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.963312070120315,
      "learning_rate": 4.5845012393882205e-06,
      "loss": 0.2205,
      "step": 7348
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.8962004522097209,
      "learning_rate": 4.581938371602964e-06,
      "loss": 0.1802,
      "step": 7349
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.6399547673897539,
      "learning_rate": 4.5793760074713565e-06,
      "loss": 0.177,
      "step": 7350
    },
    {
      "epoch": 2.08,
      "grad_norm": 2.1403810319959993,
      "learning_rate": 4.576814147231594e-06,
      "loss": 0.2087,
      "step": 7351
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.9596759733540186,
      "learning_rate": 4.574252791121808e-06,
      "loss": 0.2042,
      "step": 7352
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.9369846040268195,
      "learning_rate": 4.571691939380111e-06,
      "loss": 0.1621,
      "step": 7353
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.9072945627229922,
      "learning_rate": 4.569131592244552e-06,
      "loss": 0.2054,
      "step": 7354
    },
    {
      "epoch": 2.08,
      "grad_norm": 2.1705892848475297,
      "learning_rate": 4.566571749953135e-06,
      "loss": 0.2771,
      "step": 7355
    },
    {
      "epoch": 2.08,
      "grad_norm": 2.0481178353970386,
      "learning_rate": 4.564012412743819e-06,
      "loss": 0.1904,
      "step": 7356
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.7448399470840872,
      "learning_rate": 4.561453580854516e-06,
      "loss": 0.1943,
      "step": 7357
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.9135736109193628,
      "learning_rate": 4.558895254523086e-06,
      "loss": 0.1829,
      "step": 7358
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.9902946374191337,
      "learning_rate": 4.556337433987359e-06,
      "loss": 0.2246,
      "step": 7359
    },
    {
      "epoch": 2.08,
      "grad_norm": 2.0169783511339197,
      "learning_rate": 4.553780119485093e-06,
      "loss": 0.2261,
      "step": 7360
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.9127811766408807,
      "learning_rate": 4.551223311254013e-06,
      "loss": 0.189,
      "step": 7361
    },
    {
      "epoch": 2.08,
      "grad_norm": 2.159253129912973,
      "learning_rate": 4.5486670095318006e-06,
      "loss": 0.2021,
      "step": 7362
    },
    {
      "epoch": 2.08,
      "grad_norm": 2.6860491637684314,
      "learning_rate": 4.5461112145560845e-06,
      "loss": 0.1838,
      "step": 7363
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.8356888494508807,
      "learning_rate": 4.543555926564445e-06,
      "loss": 0.1917,
      "step": 7364
    },
    {
      "epoch": 2.08,
      "grad_norm": 2.0780224687745235,
      "learning_rate": 4.541001145794419e-06,
      "loss": 0.2021,
      "step": 7365
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.8822685356508182,
      "learning_rate": 4.538446872483492e-06,
      "loss": 0.2002,
      "step": 7366
    },
    {
      "epoch": 2.08,
      "grad_norm": 2.043549012676514,
      "learning_rate": 4.535893106869107e-06,
      "loss": 0.1998,
      "step": 7367
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.6120404438552005,
      "learning_rate": 4.533339849188656e-06,
      "loss": 0.1584,
      "step": 7368
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.8739755358445855,
      "learning_rate": 4.530787099679482e-06,
      "loss": 0.1714,
      "step": 7369
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.949081961543098,
      "learning_rate": 4.528234858578894e-06,
      "loss": 0.177,
      "step": 7370
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.715621211398721,
      "learning_rate": 4.525683126124137e-06,
      "loss": 0.1667,
      "step": 7371
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.7028071524351915,
      "learning_rate": 4.523131902552417e-06,
      "loss": 0.1621,
      "step": 7372
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.8718053201187486,
      "learning_rate": 4.520581188100891e-06,
      "loss": 0.1912,
      "step": 7373
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.998664628987699,
      "learning_rate": 4.5180309830066696e-06,
      "loss": 0.2104,
      "step": 7374
    },
    {
      "epoch": 2.08,
      "grad_norm": 2.4963313786109693,
      "learning_rate": 4.515481287506811e-06,
      "loss": 0.272,
      "step": 7375
    },
    {
      "epoch": 2.08,
      "grad_norm": 2.0605441773252973,
      "learning_rate": 4.512932101838343e-06,
      "loss": 0.1746,
      "step": 7376
    },
    {
      "epoch": 2.08,
      "grad_norm": 2.015280722604414,
      "learning_rate": 4.510383426238221e-06,
      "loss": 0.2002,
      "step": 7377
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.6637421593050383,
      "learning_rate": 4.507835260943365e-06,
      "loss": 0.1785,
      "step": 7378
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.787695612751076,
      "learning_rate": 4.505287606190658e-06,
      "loss": 0.1777,
      "step": 7379
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.8564078196437577,
      "learning_rate": 4.502740462216919e-06,
      "loss": 0.1948,
      "step": 7380
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.9990404966232311,
      "learning_rate": 4.500193829258928e-06,
      "loss": 0.2058,
      "step": 7381
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.9982068300881,
      "learning_rate": 4.497647707553414e-06,
      "loss": 0.1958,
      "step": 7382
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.9887689428662993,
      "learning_rate": 4.495102097337062e-06,
      "loss": 0.21,
      "step": 7383
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.9479631132246673,
      "learning_rate": 4.492556998846505e-06,
      "loss": 0.2004,
      "step": 7384
    },
    {
      "epoch": 2.09,
      "grad_norm": 2.2371556295204016,
      "learning_rate": 4.490012412318334e-06,
      "loss": 0.2061,
      "step": 7385
    },
    {
      "epoch": 2.09,
      "grad_norm": 2.178646411182086,
      "learning_rate": 4.487468337989083e-06,
      "loss": 0.2168,
      "step": 7386
    },
    {
      "epoch": 2.09,
      "grad_norm": 2.026612284310206,
      "learning_rate": 4.484924776095255e-06,
      "loss": 0.1907,
      "step": 7387
    },
    {
      "epoch": 2.09,
      "grad_norm": 2.022597561471462,
      "learning_rate": 4.48238172687329e-06,
      "loss": 0.2058,
      "step": 7388
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.8180045629449193,
      "learning_rate": 4.479839190559583e-06,
      "loss": 0.1414,
      "step": 7389
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.9174779501545658,
      "learning_rate": 4.477297167390487e-06,
      "loss": 0.1921,
      "step": 7390
    },
    {
      "epoch": 2.09,
      "grad_norm": 2.0930049410697005,
      "learning_rate": 4.474755657602303e-06,
      "loss": 0.2095,
      "step": 7391
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.8838414464529014,
      "learning_rate": 4.472214661431282e-06,
      "loss": 0.1676,
      "step": 7392
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.6171421144903815,
      "learning_rate": 4.469674179113641e-06,
      "loss": 0.1405,
      "step": 7393
    },
    {
      "epoch": 2.09,
      "grad_norm": 2.047344158979099,
      "learning_rate": 4.467134210885529e-06,
      "loss": 0.1982,
      "step": 7394
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.809600015992032,
      "learning_rate": 4.464594756983055e-06,
      "loss": 0.1801,
      "step": 7395
    },
    {
      "epoch": 2.09,
      "grad_norm": 2.0603707257679233,
      "learning_rate": 4.462055817642291e-06,
      "loss": 0.198,
      "step": 7396
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.8053177686894168,
      "learning_rate": 4.459517393099253e-06,
      "loss": 0.1921,
      "step": 7397
    },
    {
      "epoch": 2.09,
      "grad_norm": 2.1613538246812953,
      "learning_rate": 4.456979483589896e-06,
      "loss": 0.2328,
      "step": 7398
    },
    {
      "epoch": 2.09,
      "grad_norm": 2.811683357934006,
      "learning_rate": 4.454442089350151e-06,
      "loss": 0.2178,
      "step": 7399
    },
    {
      "epoch": 2.09,
      "grad_norm": 2.212512777227779,
      "learning_rate": 4.451905210615889e-06,
      "loss": 0.2466,
      "step": 7400
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.8740869497184431,
      "learning_rate": 4.4493688476229295e-06,
      "loss": 0.1833,
      "step": 7401
    },
    {
      "epoch": 2.09,
      "grad_norm": 2.112563831792361,
      "learning_rate": 4.4468330006070505e-06,
      "loss": 0.2095,
      "step": 7402
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.80016785563204,
      "learning_rate": 4.444297669803981e-06,
      "loss": 0.1692,
      "step": 7403
    },
    {
      "epoch": 2.09,
      "grad_norm": 2.2074149605735927,
      "learning_rate": 4.441762855449395e-06,
      "loss": 0.2207,
      "step": 7404
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.9148879064707667,
      "learning_rate": 4.439228557778933e-06,
      "loss": 0.1812,
      "step": 7405
    },
    {
      "epoch": 2.09,
      "grad_norm": 2.0655857827484994,
      "learning_rate": 4.436694777028175e-06,
      "loss": 0.2134,
      "step": 7406
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.8871293772396842,
      "learning_rate": 4.434161513432659e-06,
      "loss": 0.1919,
      "step": 7407
    },
    {
      "epoch": 2.09,
      "grad_norm": 2.0761081582511167,
      "learning_rate": 4.431628767227869e-06,
      "loss": 0.2234,
      "step": 7408
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.9404642380036659,
      "learning_rate": 4.4290965386492436e-06,
      "loss": 0.1892,
      "step": 7409
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.9310481960440269,
      "learning_rate": 4.426564827932185e-06,
      "loss": 0.1775,
      "step": 7410
    },
    {
      "epoch": 2.09,
      "grad_norm": 2.0146601324307882,
      "learning_rate": 4.424033635312025e-06,
      "loss": 0.1827,
      "step": 7411
    },
    {
      "epoch": 2.09,
      "grad_norm": 2.2079210115499333,
      "learning_rate": 4.421502961024059e-06,
      "loss": 0.2239,
      "step": 7412
    },
    {
      "epoch": 2.09,
      "grad_norm": 2.10255538170971,
      "learning_rate": 4.418972805303542e-06,
      "loss": 0.2021,
      "step": 7413
    },
    {
      "epoch": 2.09,
      "grad_norm": 2.1444707348450245,
      "learning_rate": 4.416443168385672e-06,
      "loss": 0.1597,
      "step": 7414
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.7942704304115293,
      "learning_rate": 4.413914050505591e-06,
      "loss": 0.1699,
      "step": 7415
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.8020445977260962,
      "learning_rate": 4.4113854518984085e-06,
      "loss": 0.1638,
      "step": 7416
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.7800694118198068,
      "learning_rate": 4.408857372799179e-06,
      "loss": 0.1754,
      "step": 7417
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.9678507549627022,
      "learning_rate": 4.406329813442907e-06,
      "loss": 0.1843,
      "step": 7418
    },
    {
      "epoch": 2.1,
      "grad_norm": 2.028648587924447,
      "learning_rate": 4.403802774064548e-06,
      "loss": 0.179,
      "step": 7419
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.8307851781989883,
      "learning_rate": 4.401276254899014e-06,
      "loss": 0.1479,
      "step": 7420
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.9654033562119586,
      "learning_rate": 4.39875025618116e-06,
      "loss": 0.2168,
      "step": 7421
    },
    {
      "epoch": 2.1,
      "grad_norm": 2.2839191946335227,
      "learning_rate": 4.396224778145808e-06,
      "loss": 0.1954,
      "step": 7422
    },
    {
      "epoch": 2.1,
      "grad_norm": 2.0048918384142334,
      "learning_rate": 4.393699821027716e-06,
      "loss": 0.1982,
      "step": 7423
    },
    {
      "epoch": 2.1,
      "grad_norm": 2.263053346544276,
      "learning_rate": 4.391175385061601e-06,
      "loss": 0.2561,
      "step": 7424
    },
    {
      "epoch": 2.1,
      "grad_norm": 2.3988315748862328,
      "learning_rate": 4.3886514704821315e-06,
      "loss": 0.2083,
      "step": 7425
    },
    {
      "epoch": 2.1,
      "grad_norm": 2.25695750297442,
      "learning_rate": 4.386128077523923e-06,
      "loss": 0.1991,
      "step": 7426
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.9377221326219007,
      "learning_rate": 4.383605206421549e-06,
      "loss": 0.1748,
      "step": 7427
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.886498889148439,
      "learning_rate": 4.38108285740953e-06,
      "loss": 0.177,
      "step": 7428
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.8354368377631467,
      "learning_rate": 4.378561030722335e-06,
      "loss": 0.1619,
      "step": 7429
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.5991939636784962,
      "learning_rate": 4.3760397265943965e-06,
      "loss": 0.139,
      "step": 7430
    },
    {
      "epoch": 2.1,
      "grad_norm": 2.124747980424631,
      "learning_rate": 4.37351894526009e-06,
      "loss": 0.187,
      "step": 7431
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.9936347099883982,
      "learning_rate": 4.370998686953733e-06,
      "loss": 0.1978,
      "step": 7432
    },
    {
      "epoch": 2.1,
      "grad_norm": 2.0954769866799006,
      "learning_rate": 4.368478951909614e-06,
      "loss": 0.1968,
      "step": 7433
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.9988569346173015,
      "learning_rate": 4.36595974036196e-06,
      "loss": 0.2285,
      "step": 7434
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.6756432883290957,
      "learning_rate": 4.363441052544953e-06,
      "loss": 0.1619,
      "step": 7435
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.9147491651186246,
      "learning_rate": 4.3609228886927265e-06,
      "loss": 0.2139,
      "step": 7436
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.740875940317975,
      "learning_rate": 4.3584052490393645e-06,
      "loss": 0.1785,
      "step": 7437
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.7902002007019466,
      "learning_rate": 4.355888133818897e-06,
      "loss": 0.1414,
      "step": 7438
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.7098445693400723,
      "learning_rate": 4.35337154326532e-06,
      "loss": 0.1517,
      "step": 7439
    },
    {
      "epoch": 2.1,
      "grad_norm": 2.1964735282643817,
      "learning_rate": 4.350855477612565e-06,
      "loss": 0.2139,
      "step": 7440
    },
    {
      "epoch": 2.1,
      "grad_norm": 2.0908370131130987,
      "learning_rate": 4.348339937094525e-06,
      "loss": 0.197,
      "step": 7441
    },
    {
      "epoch": 2.1,
      "grad_norm": 2.0418148211953686,
      "learning_rate": 4.345824921945039e-06,
      "loss": 0.2129,
      "step": 7442
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.6939346936083457,
      "learning_rate": 4.3433104323978956e-06,
      "loss": 0.1654,
      "step": 7443
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.8788181083838476,
      "learning_rate": 4.340796468686841e-06,
      "loss": 0.1721,
      "step": 7444
    },
    {
      "epoch": 2.1,
      "grad_norm": 2.0823048666954387,
      "learning_rate": 4.338283031045567e-06,
      "loss": 0.2051,
      "step": 7445
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.740099084442664,
      "learning_rate": 4.335770119707715e-06,
      "loss": 0.1559,
      "step": 7446
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.8782161338805525,
      "learning_rate": 4.333257734906889e-06,
      "loss": 0.1584,
      "step": 7447
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.7672659604317726,
      "learning_rate": 4.330745876876635e-06,
      "loss": 0.1791,
      "step": 7448
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.927842348722979,
      "learning_rate": 4.328234545850441e-06,
      "loss": 0.2148,
      "step": 7449
    },
    {
      "epoch": 2.1,
      "grad_norm": 2.3126617195545727,
      "learning_rate": 4.325723742061767e-06,
      "loss": 0.1836,
      "step": 7450
    },
    {
      "epoch": 2.1,
      "grad_norm": 2.3563680167371537,
      "learning_rate": 4.323213465744007e-06,
      "loss": 0.2632,
      "step": 7451
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.7629039229457377,
      "learning_rate": 4.320703717130516e-06,
      "loss": 0.1875,
      "step": 7452
    },
    {
      "epoch": 2.1,
      "grad_norm": 2.056694800138804,
      "learning_rate": 4.3181944964545915e-06,
      "loss": 0.2059,
      "step": 7453
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.8242545755368618,
      "learning_rate": 4.3156858039494895e-06,
      "loss": 0.1855,
      "step": 7454
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.8305569495968517,
      "learning_rate": 4.313177639848408e-06,
      "loss": 0.1917,
      "step": 7455
    },
    {
      "epoch": 2.11,
      "grad_norm": 2.2208505603340716,
      "learning_rate": 4.310670004384515e-06,
      "loss": 0.186,
      "step": 7456
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.8866114913983314,
      "learning_rate": 4.308162897790903e-06,
      "loss": 0.1838,
      "step": 7457
    },
    {
      "epoch": 2.11,
      "grad_norm": 2.169006745853913,
      "learning_rate": 4.305656320300628e-06,
      "loss": 0.2302,
      "step": 7458
    },
    {
      "epoch": 2.11,
      "grad_norm": 2.2911331910303265,
      "learning_rate": 4.303150272146706e-06,
      "loss": 0.2119,
      "step": 7459
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.9508144112120156,
      "learning_rate": 4.30064475356209e-06,
      "loss": 0.1724,
      "step": 7460
    },
    {
      "epoch": 2.11,
      "grad_norm": 2.183046956627005,
      "learning_rate": 4.29813976477969e-06,
      "loss": 0.2234,
      "step": 7461
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.7705026397918502,
      "learning_rate": 4.295635306032364e-06,
      "loss": 0.2019,
      "step": 7462
    },
    {
      "epoch": 2.11,
      "grad_norm": 2.4433404142093305,
      "learning_rate": 4.293131377552923e-06,
      "loss": 0.2522,
      "step": 7463
    },
    {
      "epoch": 2.11,
      "grad_norm": 2.0611889904456366,
      "learning_rate": 4.290627979574123e-06,
      "loss": 0.1819,
      "step": 7464
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.7784776011797832,
      "learning_rate": 4.2881251123286896e-06,
      "loss": 0.2214,
      "step": 7465
    },
    {
      "epoch": 2.11,
      "grad_norm": 2.08862535723185,
      "learning_rate": 4.2856227760492665e-06,
      "loss": 0.1667,
      "step": 7466
    },
    {
      "epoch": 2.11,
      "grad_norm": 2.3269264821445383,
      "learning_rate": 4.28312097096848e-06,
      "loss": 0.2439,
      "step": 7467
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.8163556773128544,
      "learning_rate": 4.280619697318888e-06,
      "loss": 0.182,
      "step": 7468
    },
    {
      "epoch": 2.11,
      "grad_norm": 2.254083424484172,
      "learning_rate": 4.278118955333007e-06,
      "loss": 0.1909,
      "step": 7469
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.9729003407958983,
      "learning_rate": 4.275618745243301e-06,
      "loss": 0.1721,
      "step": 7470
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.8817370348679083,
      "learning_rate": 4.273119067282184e-06,
      "loss": 0.1395,
      "step": 7471
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.9331982506432883,
      "learning_rate": 4.270619921682019e-06,
      "loss": 0.1611,
      "step": 7472
    },
    {
      "epoch": 2.11,
      "grad_norm": 2.086869664662594,
      "learning_rate": 4.268121308675132e-06,
      "loss": 0.2019,
      "step": 7473
    },
    {
      "epoch": 2.11,
      "grad_norm": 2.0675795763856137,
      "learning_rate": 4.265623228493781e-06,
      "loss": 0.2178,
      "step": 7474
    },
    {
      "epoch": 2.11,
      "grad_norm": 2.043906379271576,
      "learning_rate": 4.2631256813701815e-06,
      "loss": 0.228,
      "step": 7475
    },
    {
      "epoch": 2.11,
      "grad_norm": 2.239021679424122,
      "learning_rate": 4.260628667536508e-06,
      "loss": 0.1926,
      "step": 7476
    },
    {
      "epoch": 2.11,
      "grad_norm": 2.573739409490148,
      "learning_rate": 4.258132187224877e-06,
      "loss": 0.2151,
      "step": 7477
    },
    {
      "epoch": 2.11,
      "grad_norm": 2.1703140353258945,
      "learning_rate": 4.255636240667356e-06,
      "loss": 0.2451,
      "step": 7478
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.7698021990454735,
      "learning_rate": 4.253140828095964e-06,
      "loss": 0.1758,
      "step": 7479
    },
    {
      "epoch": 2.11,
      "grad_norm": 2.2531310229641646,
      "learning_rate": 4.2506459497426685e-06,
      "loss": 0.2383,
      "step": 7480
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.9949682621379405,
      "learning_rate": 4.2481516058393876e-06,
      "loss": 0.1547,
      "step": 7481
    },
    {
      "epoch": 2.11,
      "grad_norm": 2.161158794006704,
      "learning_rate": 4.2456577966180025e-06,
      "loss": 0.2109,
      "step": 7482
    },
    {
      "epoch": 2.11,
      "grad_norm": 2.141300341841473,
      "learning_rate": 4.243164522310317e-06,
      "loss": 0.1965,
      "step": 7483
    },
    {
      "epoch": 2.11,
      "grad_norm": 2.275595337997374,
      "learning_rate": 4.240671783148114e-06,
      "loss": 0.2336,
      "step": 7484
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.7941709157126742,
      "learning_rate": 4.238179579363109e-06,
      "loss": 0.1577,
      "step": 7485
    },
    {
      "epoch": 2.11,
      "grad_norm": 2.006562320793888,
      "learning_rate": 4.235687911186974e-06,
      "loss": 0.2009,
      "step": 7486
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.9858761684829116,
      "learning_rate": 4.2331967788513295e-06,
      "loss": 0.2148,
      "step": 7487
    },
    {
      "epoch": 2.11,
      "grad_norm": 2.0335313017803873,
      "learning_rate": 4.230706182587748e-06,
      "loss": 0.2034,
      "step": 7488
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.768418808937858,
      "learning_rate": 4.228216122627747e-06,
      "loss": 0.1692,
      "step": 7489
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.7703861930971658,
      "learning_rate": 4.225726599202808e-06,
      "loss": 0.1716,
      "step": 7490
    },
    {
      "epoch": 2.12,
      "grad_norm": 2.138728083179006,
      "learning_rate": 4.223237612544344e-06,
      "loss": 0.1714,
      "step": 7491
    },
    {
      "epoch": 2.12,
      "grad_norm": 2.1636016286931894,
      "learning_rate": 4.220749162883725e-06,
      "loss": 0.2397,
      "step": 7492
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.8921729531968332,
      "learning_rate": 4.218261250452281e-06,
      "loss": 0.1665,
      "step": 7493
    },
    {
      "epoch": 2.12,
      "grad_norm": 2.030438007012933,
      "learning_rate": 4.21577387548128e-06,
      "loss": 0.2401,
      "step": 7494
    },
    {
      "epoch": 2.12,
      "grad_norm": 2.0361275973717383,
      "learning_rate": 4.213287038201943e-06,
      "loss": 0.2212,
      "step": 7495
    },
    {
      "epoch": 2.12,
      "grad_norm": 2.2604498839449763,
      "learning_rate": 4.210800738845445e-06,
      "loss": 0.2505,
      "step": 7496
    },
    {
      "epoch": 2.12,
      "grad_norm": 2.2899500516378253,
      "learning_rate": 4.208314977642907e-06,
      "loss": 0.1974,
      "step": 7497
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.8153533975846534,
      "learning_rate": 4.205829754825396e-06,
      "loss": 0.1807,
      "step": 7498
    },
    {
      "epoch": 2.12,
      "grad_norm": 2.0450454592679983,
      "learning_rate": 4.203345070623947e-06,
      "loss": 0.1866,
      "step": 7499
    },
    {
      "epoch": 2.12,
      "grad_norm": 2.0253508442964017,
      "learning_rate": 4.200860925269519e-06,
      "loss": 0.2146,
      "step": 7500
    },
    {
      "epoch": 2.12,
      "grad_norm": 2.156989628260012,
      "learning_rate": 4.198377318993035e-06,
      "loss": 0.2085,
      "step": 7501
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.872971314960543,
      "learning_rate": 4.1958942520253735e-06,
      "loss": 0.1826,
      "step": 7502
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.8548003130289312,
      "learning_rate": 4.193411724597352e-06,
      "loss": 0.2002,
      "step": 7503
    },
    {
      "epoch": 2.12,
      "grad_norm": 2.045065625625272,
      "learning_rate": 4.1909297369397435e-06,
      "loss": 0.2202,
      "step": 7504
    },
    {
      "epoch": 2.12,
      "grad_norm": 2.194999853641754,
      "learning_rate": 4.188448289283269e-06,
      "loss": 0.1978,
      "step": 7505
    },
    {
      "epoch": 2.12,
      "grad_norm": 2.0390922312509097,
      "learning_rate": 4.185967381858599e-06,
      "loss": 0.1897,
      "step": 7506
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.7664998030780097,
      "learning_rate": 4.183487014896354e-06,
      "loss": 0.1543,
      "step": 7507
    },
    {
      "epoch": 2.12,
      "grad_norm": 2.03292077358598,
      "learning_rate": 4.1810071886271065e-06,
      "loss": 0.2063,
      "step": 7508
    },
    {
      "epoch": 2.12,
      "grad_norm": 2.240596420797422,
      "learning_rate": 4.178527903281371e-06,
      "loss": 0.1694,
      "step": 7509
    },
    {
      "epoch": 2.12,
      "grad_norm": 2.4238837758916243,
      "learning_rate": 4.176049159089626e-06,
      "loss": 0.2556,
      "step": 7510
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.9394138598679083,
      "learning_rate": 4.173570956282286e-06,
      "loss": 0.2029,
      "step": 7511
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.7686113321257988,
      "learning_rate": 4.171093295089723e-06,
      "loss": 0.157,
      "step": 7512
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.9271530404897643,
      "learning_rate": 4.168616175742255e-06,
      "loss": 0.1633,
      "step": 7513
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.6961930034643586,
      "learning_rate": 4.1661395984701495e-06,
      "loss": 0.1622,
      "step": 7514
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.7742147288214536,
      "learning_rate": 4.1636635635036235e-06,
      "loss": 0.1929,
      "step": 7515
    },
    {
      "epoch": 2.12,
      "grad_norm": 2.2371207110510607,
      "learning_rate": 4.161188071072854e-06,
      "loss": 0.178,
      "step": 7516
    },
    {
      "epoch": 2.12,
      "grad_norm": 2.150723010766719,
      "learning_rate": 4.158713121407949e-06,
      "loss": 0.1689,
      "step": 7517
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.7901048187117723,
      "learning_rate": 4.156238714738974e-06,
      "loss": 0.1777,
      "step": 7518
    },
    {
      "epoch": 2.12,
      "grad_norm": 2.1391110617788813,
      "learning_rate": 4.153764851295954e-06,
      "loss": 0.2139,
      "step": 7519
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.8622700525884925,
      "learning_rate": 4.1512915313088505e-06,
      "loss": 0.1677,
      "step": 7520
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.9107613548310347,
      "learning_rate": 4.148818755007581e-06,
      "loss": 0.1794,
      "step": 7521
    },
    {
      "epoch": 2.12,
      "grad_norm": 2.169796323567054,
      "learning_rate": 4.146346522622008e-06,
      "loss": 0.1868,
      "step": 7522
    },
    {
      "epoch": 2.12,
      "grad_norm": 2.0726533709147628,
      "learning_rate": 4.143874834381947e-06,
      "loss": 0.1605,
      "step": 7523
    },
    {
      "epoch": 2.12,
      "grad_norm": 2.199831720454316,
      "learning_rate": 4.141403690517163e-06,
      "loss": 0.2019,
      "step": 7524
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.9932940116045512,
      "learning_rate": 4.138933091257368e-06,
      "loss": 0.1455,
      "step": 7525
    },
    {
      "epoch": 2.13,
      "grad_norm": 2.0198542365687477,
      "learning_rate": 4.13646303683222e-06,
      "loss": 0.1797,
      "step": 7526
    },
    {
      "epoch": 2.13,
      "grad_norm": 2.2829471982943392,
      "learning_rate": 4.1339935274713404e-06,
      "loss": 0.2524,
      "step": 7527
    },
    {
      "epoch": 2.13,
      "grad_norm": 2.210724916811285,
      "learning_rate": 4.13152456340429e-06,
      "loss": 0.2493,
      "step": 7528
    },
    {
      "epoch": 2.13,
      "grad_norm": 2.085166566612497,
      "learning_rate": 4.129056144860567e-06,
      "loss": 0.2302,
      "step": 7529
    },
    {
      "epoch": 2.13,
      "grad_norm": 2.2265782907477476,
      "learning_rate": 4.126588272069645e-06,
      "loss": 0.2119,
      "step": 7530
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.8674010624846438,
      "learning_rate": 4.124120945260927e-06,
      "loss": 0.1841,
      "step": 7531
    },
    {
      "epoch": 2.13,
      "grad_norm": 2.210244383029525,
      "learning_rate": 4.121654164663769e-06,
      "loss": 0.2175,
      "step": 7532
    },
    {
      "epoch": 2.13,
      "grad_norm": 2.2835501336616444,
      "learning_rate": 4.119187930507489e-06,
      "loss": 0.239,
      "step": 7533
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.6939794662746386,
      "learning_rate": 4.116722243021333e-06,
      "loss": 0.1458,
      "step": 7534
    },
    {
      "epoch": 2.13,
      "grad_norm": 2.105469413439059,
      "learning_rate": 4.114257102434508e-06,
      "loss": 0.2009,
      "step": 7535
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.8895947643653277,
      "learning_rate": 4.111792508976175e-06,
      "loss": 0.2063,
      "step": 7536
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.8357202678649152,
      "learning_rate": 4.1093284628754355e-06,
      "loss": 0.2007,
      "step": 7537
    },
    {
      "epoch": 2.13,
      "grad_norm": 2.1071339086829934,
      "learning_rate": 4.106864964361343e-06,
      "loss": 0.2195,
      "step": 7538
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.9133666852407545,
      "learning_rate": 4.104402013662901e-06,
      "loss": 0.1808,
      "step": 7539
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.743360336926452,
      "learning_rate": 4.101939611009059e-06,
      "loss": 0.1667,
      "step": 7540
    },
    {
      "epoch": 2.13,
      "grad_norm": 2.374554993802437,
      "learning_rate": 4.09947775662872e-06,
      "loss": 0.2334,
      "step": 7541
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.9821912662976329,
      "learning_rate": 4.097016450750733e-06,
      "loss": 0.1748,
      "step": 7542
    },
    {
      "epoch": 2.13,
      "grad_norm": 2.269510132048565,
      "learning_rate": 4.094555693603891e-06,
      "loss": 0.2388,
      "step": 7543
    },
    {
      "epoch": 2.13,
      "grad_norm": 2.1726130050560224,
      "learning_rate": 4.092095485416952e-06,
      "loss": 0.2273,
      "step": 7544
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.8118133474470377,
      "learning_rate": 4.0896358264186095e-06,
      "loss": 0.1548,
      "step": 7545
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.7397523399387953,
      "learning_rate": 4.087176716837502e-06,
      "loss": 0.1884,
      "step": 7546
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.9990825754615913,
      "learning_rate": 4.0847181569022335e-06,
      "loss": 0.1655,
      "step": 7547
    },
    {
      "epoch": 2.13,
      "grad_norm": 2.624023538278871,
      "learning_rate": 4.0822601468413425e-06,
      "loss": 0.1865,
      "step": 7548
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.9133174380527893,
      "learning_rate": 4.079802686883318e-06,
      "loss": 0.1899,
      "step": 7549
    },
    {
      "epoch": 2.13,
      "grad_norm": 2.050131840196608,
      "learning_rate": 4.077345777256614e-06,
      "loss": 0.1545,
      "step": 7550
    },
    {
      "epoch": 2.13,
      "grad_norm": 2.2528258750321593,
      "learning_rate": 4.074889418189608e-06,
      "loss": 0.2151,
      "step": 7551
    },
    {
      "epoch": 2.13,
      "grad_norm": 2.3255794333726247,
      "learning_rate": 4.07243360991064e-06,
      "loss": 0.1992,
      "step": 7552
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.999014057578627,
      "learning_rate": 4.069978352648004e-06,
      "loss": 0.197,
      "step": 7553
    },
    {
      "epoch": 2.13,
      "grad_norm": 2.0140887547024944,
      "learning_rate": 4.067523646629938e-06,
      "loss": 0.2058,
      "step": 7554
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.5941951519707283,
      "learning_rate": 4.065069492084614e-06,
      "loss": 0.1453,
      "step": 7555
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.9161141922103038,
      "learning_rate": 4.062615889240176e-06,
      "loss": 0.1857,
      "step": 7556
    },
    {
      "epoch": 2.13,
      "grad_norm": 2.502977475188623,
      "learning_rate": 4.060162838324708e-06,
      "loss": 0.2354,
      "step": 7557
    },
    {
      "epoch": 2.13,
      "grad_norm": 2.510805657169908,
      "learning_rate": 4.057710339566238e-06,
      "loss": 0.2009,
      "step": 7558
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.992107257336878,
      "learning_rate": 4.055258393192746e-06,
      "loss": 0.2227,
      "step": 7559
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.7197881020222858,
      "learning_rate": 4.052806999432161e-06,
      "loss": 0.1697,
      "step": 7560
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.7819422128270648,
      "learning_rate": 4.050356158512357e-06,
      "loss": 0.1615,
      "step": 7561
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.978143796340364,
      "learning_rate": 4.047905870661172e-06,
      "loss": 0.1641,
      "step": 7562
    },
    {
      "epoch": 2.14,
      "grad_norm": 2.1738773980342936,
      "learning_rate": 4.045456136106363e-06,
      "loss": 0.2324,
      "step": 7563
    },
    {
      "epoch": 2.14,
      "grad_norm": 2.4601228966082553,
      "learning_rate": 4.043006955075667e-06,
      "loss": 0.2178,
      "step": 7564
    },
    {
      "epoch": 2.14,
      "grad_norm": 2.174842428997726,
      "learning_rate": 4.0405583277967506e-06,
      "loss": 0.2198,
      "step": 7565
    },
    {
      "epoch": 2.14,
      "grad_norm": 2.0853091590772497,
      "learning_rate": 4.038110254497234e-06,
      "loss": 0.2112,
      "step": 7566
    },
    {
      "epoch": 2.14,
      "grad_norm": 2.0343691814348155,
      "learning_rate": 4.0356627354046854e-06,
      "loss": 0.1877,
      "step": 7567
    },
    {
      "epoch": 2.14,
      "grad_norm": 2.287155215748969,
      "learning_rate": 4.033215770746622e-06,
      "loss": 0.1887,
      "step": 7568
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.8440430425919319,
      "learning_rate": 4.030769360750507e-06,
      "loss": 0.1888,
      "step": 7569
    },
    {
      "epoch": 2.14,
      "grad_norm": 2.137227695463478,
      "learning_rate": 4.028323505643762e-06,
      "loss": 0.1921,
      "step": 7570
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.8261937976862836,
      "learning_rate": 4.025878205653747e-06,
      "loss": 0.1643,
      "step": 7571
    },
    {
      "epoch": 2.14,
      "grad_norm": 2.257113312528663,
      "learning_rate": 4.023433461007764e-06,
      "loss": 0.2388,
      "step": 7572
    },
    {
      "epoch": 2.14,
      "grad_norm": 2.442368899035337,
      "learning_rate": 4.020989271933082e-06,
      "loss": 0.2415,
      "step": 7573
    },
    {
      "epoch": 2.14,
      "grad_norm": 2.106434934836089,
      "learning_rate": 4.0185456386569066e-06,
      "loss": 0.2092,
      "step": 7574
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.9842431667048765,
      "learning_rate": 4.016102561406392e-06,
      "loss": 0.217,
      "step": 7575
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.9816177364865777,
      "learning_rate": 4.013660040408643e-06,
      "loss": 0.1843,
      "step": 7576
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.8417109500859639,
      "learning_rate": 4.011218075890711e-06,
      "loss": 0.1912,
      "step": 7577
    },
    {
      "epoch": 2.14,
      "grad_norm": 2.3742067970234375,
      "learning_rate": 4.008776668079596e-06,
      "loss": 0.2312,
      "step": 7578
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.9905882485624367,
      "learning_rate": 4.006335817202256e-06,
      "loss": 0.2068,
      "step": 7579
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.8863306648450737,
      "learning_rate": 4.003895523485575e-06,
      "loss": 0.1776,
      "step": 7580
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.728892231352397,
      "learning_rate": 4.001455787156407e-06,
      "loss": 0.1421,
      "step": 7581
    },
    {
      "epoch": 2.14,
      "grad_norm": 2.048909326260258,
      "learning_rate": 3.999016608441544e-06,
      "loss": 0.188,
      "step": 7582
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.9796079972668146,
      "learning_rate": 3.996577987567727e-06,
      "loss": 0.1812,
      "step": 7583
    },
    {
      "epoch": 2.14,
      "grad_norm": 2.3524118261034337,
      "learning_rate": 3.994139924761646e-06,
      "loss": 0.191,
      "step": 7584
    },
    {
      "epoch": 2.14,
      "grad_norm": 2.26105439702915,
      "learning_rate": 3.991702420249941e-06,
      "loss": 0.2058,
      "step": 7585
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.843076189175083,
      "learning_rate": 3.989265474259192e-06,
      "loss": 0.229,
      "step": 7586
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.9097532621499091,
      "learning_rate": 3.986829087015941e-06,
      "loss": 0.1646,
      "step": 7587
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.8999826997583034,
      "learning_rate": 3.984393258746672e-06,
      "loss": 0.199,
      "step": 7588
    },
    {
      "epoch": 2.14,
      "grad_norm": 2.4483109465819415,
      "learning_rate": 3.981957989677803e-06,
      "loss": 0.3281,
      "step": 7589
    },
    {
      "epoch": 2.14,
      "grad_norm": 2.2864762199566853,
      "learning_rate": 3.979523280035723e-06,
      "loss": 0.167,
      "step": 7590
    },
    {
      "epoch": 2.14,
      "grad_norm": 2.0206962586672175,
      "learning_rate": 3.977089130046756e-06,
      "loss": 0.2065,
      "step": 7591
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.8529414708197005,
      "learning_rate": 3.974655539937176e-06,
      "loss": 0.1346,
      "step": 7592
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.934838499261001,
      "learning_rate": 3.972222509933204e-06,
      "loss": 0.1754,
      "step": 7593
    },
    {
      "epoch": 2.14,
      "grad_norm": 2.199623947367478,
      "learning_rate": 3.969790040261013e-06,
      "loss": 0.1826,
      "step": 7594
    },
    {
      "epoch": 2.14,
      "grad_norm": 2.2389011974894246,
      "learning_rate": 3.967358131146716e-06,
      "loss": 0.2043,
      "step": 7595
    },
    {
      "epoch": 2.15,
      "grad_norm": 2.1329605530897835,
      "learning_rate": 3.96492678281639e-06,
      "loss": 0.2312,
      "step": 7596
    },
    {
      "epoch": 2.15,
      "grad_norm": 2.014502690842627,
      "learning_rate": 3.962495995496037e-06,
      "loss": 0.1328,
      "step": 7597
    },
    {
      "epoch": 2.15,
      "grad_norm": 2.2972667445421817,
      "learning_rate": 3.960065769411619e-06,
      "loss": 0.207,
      "step": 7598
    },
    {
      "epoch": 2.15,
      "grad_norm": 2.1200168636130696,
      "learning_rate": 3.957636104789056e-06,
      "loss": 0.188,
      "step": 7599
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.8238396920678386,
      "learning_rate": 3.955207001854197e-06,
      "loss": 0.1437,
      "step": 7600
    },
    {
      "epoch": 2.15,
      "grad_norm": 2.1128355760866877,
      "learning_rate": 3.952778460832851e-06,
      "loss": 0.1753,
      "step": 7601
    },
    {
      "epoch": 2.15,
      "grad_norm": 2.2691955437159113,
      "learning_rate": 3.95035048195077e-06,
      "loss": 0.2483,
      "step": 7602
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.6451079571726646,
      "learning_rate": 3.947923065433654e-06,
      "loss": 0.1332,
      "step": 7603
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.7520982720415863,
      "learning_rate": 3.945496211507148e-06,
      "loss": 0.1802,
      "step": 7604
    },
    {
      "epoch": 2.15,
      "grad_norm": 2.070574511478381,
      "learning_rate": 3.943069920396859e-06,
      "loss": 0.2378,
      "step": 7605
    },
    {
      "epoch": 2.15,
      "grad_norm": 2.02199873826689,
      "learning_rate": 3.940644192328317e-06,
      "loss": 0.1853,
      "step": 7606
    },
    {
      "epoch": 2.15,
      "grad_norm": 2.0849213506076296,
      "learning_rate": 3.938219027527023e-06,
      "loss": 0.218,
      "step": 7607
    },
    {
      "epoch": 2.15,
      "grad_norm": 2.224766346719895,
      "learning_rate": 3.935794426218415e-06,
      "loss": 0.1873,
      "step": 7608
    },
    {
      "epoch": 2.15,
      "grad_norm": 2.1602156650420037,
      "learning_rate": 3.933370388627878e-06,
      "loss": 0.2061,
      "step": 7609
    },
    {
      "epoch": 2.15,
      "grad_norm": 2.1496676728626642,
      "learning_rate": 3.930946914980744e-06,
      "loss": 0.1921,
      "step": 7610
    },
    {
      "epoch": 2.15,
      "grad_norm": 2.028027566153744,
      "learning_rate": 3.9285240055023e-06,
      "loss": 0.2224,
      "step": 7611
    },
    {
      "epoch": 2.15,
      "grad_norm": 2.076625241762043,
      "learning_rate": 3.926101660417769e-06,
      "loss": 0.2251,
      "step": 7612
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.9555782556943915,
      "learning_rate": 3.9236798799523375e-06,
      "loss": 0.1592,
      "step": 7613
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.6819484096799489,
      "learning_rate": 3.921258664331122e-06,
      "loss": 0.135,
      "step": 7614
    },
    {
      "epoch": 2.15,
      "grad_norm": 2.14246642088301,
      "learning_rate": 3.9188380137791934e-06,
      "loss": 0.2134,
      "step": 7615
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.679887463542208,
      "learning_rate": 3.9164179285215785e-06,
      "loss": 0.1689,
      "step": 7616
    },
    {
      "epoch": 2.15,
      "grad_norm": 2.105469561344347,
      "learning_rate": 3.91399840878324e-06,
      "loss": 0.1848,
      "step": 7617
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.7326224919812963,
      "learning_rate": 3.911579454789093e-06,
      "loss": 0.1646,
      "step": 7618
    },
    {
      "epoch": 2.15,
      "grad_norm": 2.5711586990948847,
      "learning_rate": 3.909161066763999e-06,
      "loss": 0.241,
      "step": 7619
    },
    {
      "epoch": 2.15,
      "grad_norm": 2.0846760124901285,
      "learning_rate": 3.906743244932767e-06,
      "loss": 0.208,
      "step": 7620
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.9197020785826595,
      "learning_rate": 3.90432598952015e-06,
      "loss": 0.1823,
      "step": 7621
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.7768249875917899,
      "learning_rate": 3.9019093007508635e-06,
      "loss": 0.1619,
      "step": 7622
    },
    {
      "epoch": 2.15,
      "grad_norm": 2.1693210189106,
      "learning_rate": 3.899493178849544e-06,
      "loss": 0.1956,
      "step": 7623
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.7222773248099164,
      "learning_rate": 3.8970776240407995e-06,
      "loss": 0.1672,
      "step": 7624
    },
    {
      "epoch": 2.15,
      "grad_norm": 2.1192386898477933,
      "learning_rate": 3.8946626365491735e-06,
      "loss": 0.1875,
      "step": 7625
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.9835336446591008,
      "learning_rate": 3.892248216599158e-06,
      "loss": 0.2026,
      "step": 7626
    },
    {
      "epoch": 2.15,
      "grad_norm": 2.259726326698506,
      "learning_rate": 3.8898343644151945e-06,
      "loss": 0.2012,
      "step": 7627
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.9985222805550722,
      "learning_rate": 3.887421080221671e-06,
      "loss": 0.2017,
      "step": 7628
    },
    {
      "epoch": 2.15,
      "grad_norm": 2.0125946548801736,
      "learning_rate": 3.885008364242917e-06,
      "loss": 0.1848,
      "step": 7629
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.9252774100767511,
      "learning_rate": 3.882596216703226e-06,
      "loss": 0.2163,
      "step": 7630
    },
    {
      "epoch": 2.16,
      "grad_norm": 2.0479514847206763,
      "learning_rate": 3.880184637826816e-06,
      "loss": 0.1639,
      "step": 7631
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.9468108979048167,
      "learning_rate": 3.877773627837863e-06,
      "loss": 0.1989,
      "step": 7632
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.9714388018058473,
      "learning_rate": 3.875363186960499e-06,
      "loss": 0.1722,
      "step": 7633
    },
    {
      "epoch": 2.16,
      "grad_norm": 2.278101384072379,
      "learning_rate": 3.872953315418793e-06,
      "loss": 0.2544,
      "step": 7634
    },
    {
      "epoch": 2.16,
      "grad_norm": 2.1447639528689653,
      "learning_rate": 3.870544013436751e-06,
      "loss": 0.2133,
      "step": 7635
    },
    {
      "epoch": 2.16,
      "grad_norm": 2.2437059744697323,
      "learning_rate": 3.868135281238351e-06,
      "loss": 0.2058,
      "step": 7636
    },
    {
      "epoch": 2.16,
      "grad_norm": 2.244135936359934,
      "learning_rate": 3.865727119047497e-06,
      "loss": 0.1929,
      "step": 7637
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.8974324975984578,
      "learning_rate": 3.863319527088048e-06,
      "loss": 0.2173,
      "step": 7638
    },
    {
      "epoch": 2.16,
      "grad_norm": 2.165534510903707,
      "learning_rate": 3.860912505583819e-06,
      "loss": 0.1973,
      "step": 7639
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.8748384215511256,
      "learning_rate": 3.858506054758547e-06,
      "loss": 0.1851,
      "step": 7640
    },
    {
      "epoch": 2.16,
      "grad_norm": 2.112204849506047,
      "learning_rate": 3.856100174835945e-06,
      "loss": 0.2063,
      "step": 7641
    },
    {
      "epoch": 2.16,
      "grad_norm": 2.2905144302605236,
      "learning_rate": 3.853694866039652e-06,
      "loss": 0.2166,
      "step": 7642
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.779517338730405,
      "learning_rate": 3.851290128593265e-06,
      "loss": 0.166,
      "step": 7643
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.8265591757912596,
      "learning_rate": 3.848885962720321e-06,
      "loss": 0.146,
      "step": 7644
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.8205783462457648,
      "learning_rate": 3.84648236864431e-06,
      "loss": 0.2097,
      "step": 7645
    },
    {
      "epoch": 2.16,
      "grad_norm": 2.2809463718549594,
      "learning_rate": 3.8440793465886595e-06,
      "loss": 0.2751,
      "step": 7646
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.7014022408743126,
      "learning_rate": 3.841676896776764e-06,
      "loss": 0.1285,
      "step": 7647
    },
    {
      "epoch": 2.16,
      "grad_norm": 2.217823619016986,
      "learning_rate": 3.8392750194319385e-06,
      "loss": 0.1992,
      "step": 7648
    },
    {
      "epoch": 2.16,
      "grad_norm": 2.362154790544176,
      "learning_rate": 3.836873714777458e-06,
      "loss": 0.2041,
      "step": 7649
    },
    {
      "epoch": 2.16,
      "grad_norm": 2.198671763719858,
      "learning_rate": 3.834472983036551e-06,
      "loss": 0.2266,
      "step": 7650
    },
    {
      "epoch": 2.16,
      "grad_norm": 2.1525186159461036,
      "learning_rate": 3.832072824432385e-06,
      "loss": 0.2324,
      "step": 7651
    },
    {
      "epoch": 2.16,
      "grad_norm": 2.238147542225343,
      "learning_rate": 3.829673239188063e-06,
      "loss": 0.1958,
      "step": 7652
    },
    {
      "epoch": 2.16,
      "grad_norm": 2.4356129666016204,
      "learning_rate": 3.827274227526658e-06,
      "loss": 0.2219,
      "step": 7653
    },
    {
      "epoch": 2.16,
      "grad_norm": 2.0206671760868447,
      "learning_rate": 3.824875789671175e-06,
      "loss": 0.2097,
      "step": 7654
    },
    {
      "epoch": 2.16,
      "grad_norm": 2.3222291530936396,
      "learning_rate": 3.822477925844564e-06,
      "loss": 0.2236,
      "step": 7655
    },
    {
      "epoch": 2.16,
      "grad_norm": 2.164119085042346,
      "learning_rate": 3.820080636269737e-06,
      "loss": 0.2273,
      "step": 7656
    },
    {
      "epoch": 2.16,
      "grad_norm": 2.217632202722385,
      "learning_rate": 3.8176839211695314e-06,
      "loss": 0.2385,
      "step": 7657
    },
    {
      "epoch": 2.16,
      "grad_norm": 2.2639133918888024,
      "learning_rate": 3.815287780766743e-06,
      "loss": 0.2339,
      "step": 7658
    },
    {
      "epoch": 2.16,
      "grad_norm": 2.141495934049779,
      "learning_rate": 3.8128922152841188e-06,
      "loss": 0.1721,
      "step": 7659
    },
    {
      "epoch": 2.16,
      "grad_norm": 2.8941275258932597,
      "learning_rate": 3.8104972249443417e-06,
      "loss": 0.269,
      "step": 7660
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.7820258304776577,
      "learning_rate": 3.808102809970048e-06,
      "loss": 0.1543,
      "step": 7661
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.7829224995739241,
      "learning_rate": 3.805708970583817e-06,
      "loss": 0.1591,
      "step": 7662
    },
    {
      "epoch": 2.16,
      "grad_norm": 2.240530859841787,
      "learning_rate": 3.803315707008176e-06,
      "loss": 0.228,
      "step": 7663
    },
    {
      "epoch": 2.16,
      "grad_norm": 2.018531980568549,
      "learning_rate": 3.800923019465599e-06,
      "loss": 0.2031,
      "step": 7664
    },
    {
      "epoch": 2.16,
      "grad_norm": 2.094336840181632,
      "learning_rate": 3.798530908178506e-06,
      "loss": 0.2285,
      "step": 7665
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.4553984236983954,
      "learning_rate": 3.796139373369261e-06,
      "loss": 0.1328,
      "step": 7666
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.9130575062576893,
      "learning_rate": 3.7937484152601824e-06,
      "loss": 0.188,
      "step": 7667
    },
    {
      "epoch": 2.17,
      "grad_norm": 2.3027531535169237,
      "learning_rate": 3.79135803407353e-06,
      "loss": 0.2092,
      "step": 7668
    },
    {
      "epoch": 2.17,
      "grad_norm": 2.196832803633083,
      "learning_rate": 3.7889682300315e-06,
      "loss": 0.2224,
      "step": 7669
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.8216866233620026,
      "learning_rate": 3.7865790033562532e-06,
      "loss": 0.1613,
      "step": 7670
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.8628155152953636,
      "learning_rate": 3.7841903542698855e-06,
      "loss": 0.1831,
      "step": 7671
    },
    {
      "epoch": 2.17,
      "grad_norm": 2.1307328830037306,
      "learning_rate": 3.781802282994439e-06,
      "loss": 0.2007,
      "step": 7672
    },
    {
      "epoch": 2.17,
      "grad_norm": 2.5606726434803275,
      "learning_rate": 3.7794147897519128e-06,
      "loss": 0.2378,
      "step": 7673
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.710249609095553,
      "learning_rate": 3.7770278747642364e-06,
      "loss": 0.1523,
      "step": 7674
    },
    {
      "epoch": 2.17,
      "grad_norm": 2.2683089569837875,
      "learning_rate": 3.7746415382532907e-06,
      "loss": 0.2336,
      "step": 7675
    },
    {
      "epoch": 2.17,
      "grad_norm": 2.0958442773995287,
      "learning_rate": 3.7722557804409145e-06,
      "loss": 0.2297,
      "step": 7676
    },
    {
      "epoch": 2.17,
      "grad_norm": 2.2098270261183917,
      "learning_rate": 3.7698706015488783e-06,
      "loss": 0.1958,
      "step": 7677
    },
    {
      "epoch": 2.17,
      "grad_norm": 2.157680502816733,
      "learning_rate": 3.767486001798907e-06,
      "loss": 0.2388,
      "step": 7678
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.795857696171177,
      "learning_rate": 3.7651019814126656e-06,
      "loss": 0.1733,
      "step": 7679
    },
    {
      "epoch": 2.17,
      "grad_norm": 2.04801476745326,
      "learning_rate": 3.7627185406117707e-06,
      "loss": 0.2009,
      "step": 7680
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.8786698729319102,
      "learning_rate": 3.7603356796177824e-06,
      "loss": 0.1718,
      "step": 7681
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.8687360997416116,
      "learning_rate": 3.7579533986522065e-06,
      "loss": 0.1765,
      "step": 7682
    },
    {
      "epoch": 2.17,
      "grad_norm": 2.407706278897114,
      "learning_rate": 3.755571697936493e-06,
      "loss": 0.219,
      "step": 7683
    },
    {
      "epoch": 2.17,
      "grad_norm": 2.4926241589223084,
      "learning_rate": 3.7531905776920474e-06,
      "loss": 0.2766,
      "step": 7684
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.6609999186585196,
      "learning_rate": 3.7508100381402157e-06,
      "loss": 0.1425,
      "step": 7685
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.9092120356315974,
      "learning_rate": 3.7484300795022765e-06,
      "loss": 0.1763,
      "step": 7686
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.9544120534200546,
      "learning_rate": 3.7460507019994775e-06,
      "loss": 0.2322,
      "step": 7687
    },
    {
      "epoch": 2.17,
      "grad_norm": 2.0320817591899063,
      "learning_rate": 3.743671905852999e-06,
      "loss": 0.1992,
      "step": 7688
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.9757021715441816,
      "learning_rate": 3.741293691283968e-06,
      "loss": 0.2021,
      "step": 7689
    },
    {
      "epoch": 2.17,
      "grad_norm": 2.263395237247711,
      "learning_rate": 3.738916058513462e-06,
      "loss": 0.2114,
      "step": 7690
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.8394221614508635,
      "learning_rate": 3.7365390077624985e-06,
      "loss": 0.1647,
      "step": 7691
    },
    {
      "epoch": 2.17,
      "grad_norm": 2.095224515103355,
      "learning_rate": 3.734162539252043e-06,
      "loss": 0.186,
      "step": 7692
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.9623313523497645,
      "learning_rate": 3.7317866532030145e-06,
      "loss": 0.1655,
      "step": 7693
    },
    {
      "epoch": 2.17,
      "grad_norm": 2.0720643719309813,
      "learning_rate": 3.7294113498362695e-06,
      "loss": 0.2434,
      "step": 7694
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.962647786424514,
      "learning_rate": 3.7270366293726033e-06,
      "loss": 0.2141,
      "step": 7695
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.8628710029116495,
      "learning_rate": 3.7246624920327755e-06,
      "loss": 0.1543,
      "step": 7696
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.8999993601337997,
      "learning_rate": 3.722288938037478e-06,
      "loss": 0.1851,
      "step": 7697
    },
    {
      "epoch": 2.17,
      "grad_norm": 2.012038011895889,
      "learning_rate": 3.719915967607355e-06,
      "loss": 0.2068,
      "step": 7698
    },
    {
      "epoch": 2.17,
      "grad_norm": 2.244357463976042,
      "learning_rate": 3.71754358096299e-06,
      "loss": 0.2268,
      "step": 7699
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.8309486240049828,
      "learning_rate": 3.7151717783249175e-06,
      "loss": 0.196,
      "step": 7700
    },
    {
      "epoch": 2.17,
      "grad_norm": 2.4412838478403636,
      "learning_rate": 3.712800559913613e-06,
      "loss": 0.2393,
      "step": 7701
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.8405056611865103,
      "learning_rate": 3.7104299259495113e-06,
      "loss": 0.1843,
      "step": 7702
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.9748151008034314,
      "learning_rate": 3.7080598766529686e-06,
      "loss": 0.2285,
      "step": 7703
    },
    {
      "epoch": 2.18,
      "grad_norm": 2.065424134106734,
      "learning_rate": 3.7056904122443105e-06,
      "loss": 0.2224,
      "step": 7704
    },
    {
      "epoch": 2.18,
      "grad_norm": 2.154410668935536,
      "learning_rate": 3.7033215329437945e-06,
      "loss": 0.2234,
      "step": 7705
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.8749977794464439,
      "learning_rate": 3.7009532389716283e-06,
      "loss": 0.1937,
      "step": 7706
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.9250195936933303,
      "learning_rate": 3.6985855305479645e-06,
      "loss": 0.1508,
      "step": 7707
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.78366018602808,
      "learning_rate": 3.696218407892901e-06,
      "loss": 0.1956,
      "step": 7708
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.9638338636488752,
      "learning_rate": 3.6938518712264783e-06,
      "loss": 0.1575,
      "step": 7709
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.794376141840137,
      "learning_rate": 3.6914859207686916e-06,
      "loss": 0.2117,
      "step": 7710
    },
    {
      "epoch": 2.18,
      "grad_norm": 2.1999894300324025,
      "learning_rate": 3.689120556739475e-06,
      "loss": 0.1863,
      "step": 7711
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.8899613488848899,
      "learning_rate": 3.6867557793587005e-06,
      "loss": 0.207,
      "step": 7712
    },
    {
      "epoch": 2.18,
      "grad_norm": 2.3165223922476397,
      "learning_rate": 3.6843915888462022e-06,
      "loss": 0.2128,
      "step": 7713
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.976111057943877,
      "learning_rate": 3.682027985421749e-06,
      "loss": 0.1814,
      "step": 7714
    },
    {
      "epoch": 2.18,
      "grad_norm": 2.1702275922901677,
      "learning_rate": 3.6796649693050568e-06,
      "loss": 0.1926,
      "step": 7715
    },
    {
      "epoch": 2.18,
      "grad_norm": 2.1598916193900015,
      "learning_rate": 3.677302540715787e-06,
      "loss": 0.2002,
      "step": 7716
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.9393485138272317,
      "learning_rate": 3.674940699873547e-06,
      "loss": 0.1775,
      "step": 7717
    },
    {
      "epoch": 2.18,
      "grad_norm": 2.052075982505852,
      "learning_rate": 3.672579446997887e-06,
      "loss": 0.1946,
      "step": 7718
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.7444985941191673,
      "learning_rate": 3.6702187823083147e-06,
      "loss": 0.1698,
      "step": 7719
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.9898847315155517,
      "learning_rate": 3.6678587060242586e-06,
      "loss": 0.2042,
      "step": 7720
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.848841825091892,
      "learning_rate": 3.6654992183651193e-06,
      "loss": 0.2009,
      "step": 7721
    },
    {
      "epoch": 2.18,
      "grad_norm": 2.2284753274225504,
      "learning_rate": 3.663140319550227e-06,
      "loss": 0.1909,
      "step": 7722
    },
    {
      "epoch": 2.18,
      "grad_norm": 2.3035008533912205,
      "learning_rate": 3.66078200979886e-06,
      "loss": 0.1951,
      "step": 7723
    },
    {
      "epoch": 2.18,
      "grad_norm": 2.0546379602093734,
      "learning_rate": 3.6584242893302436e-06,
      "loss": 0.1594,
      "step": 7724
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.9064967866443097,
      "learning_rate": 3.6560671583635467e-06,
      "loss": 0.1851,
      "step": 7725
    },
    {
      "epoch": 2.18,
      "grad_norm": 2.526354808081644,
      "learning_rate": 3.6537106171178817e-06,
      "loss": 0.2229,
      "step": 7726
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.961393982967992,
      "learning_rate": 3.651354665812313e-06,
      "loss": 0.2085,
      "step": 7727
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.9900736582686467,
      "learning_rate": 3.648999304665849e-06,
      "loss": 0.2012,
      "step": 7728
    },
    {
      "epoch": 2.18,
      "grad_norm": 2.1252257539392208,
      "learning_rate": 3.6466445338974276e-06,
      "loss": 0.2007,
      "step": 7729
    },
    {
      "epoch": 2.18,
      "grad_norm": 2.1631901046411257,
      "learning_rate": 3.6442903537259556e-06,
      "loss": 0.24,
      "step": 7730
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.851584217110749,
      "learning_rate": 3.6419367643702684e-06,
      "loss": 0.1868,
      "step": 7731
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.8089924586795778,
      "learning_rate": 3.6395837660491538e-06,
      "loss": 0.1492,
      "step": 7732
    },
    {
      "epoch": 2.18,
      "grad_norm": 2.1764480045217907,
      "learning_rate": 3.637231358981341e-06,
      "loss": 0.1969,
      "step": 7733
    },
    {
      "epoch": 2.18,
      "grad_norm": 2.5929849204166415,
      "learning_rate": 3.6348795433855064e-06,
      "loss": 0.2419,
      "step": 7734
    },
    {
      "epoch": 2.18,
      "grad_norm": 2.2890833489101685,
      "learning_rate": 3.6325283194802675e-06,
      "loss": 0.2346,
      "step": 7735
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.9566254861691228,
      "learning_rate": 3.6301776874841997e-06,
      "loss": 0.177,
      "step": 7736
    },
    {
      "epoch": 2.18,
      "grad_norm": 2.0010269896177033,
      "learning_rate": 3.6278276476158025e-06,
      "loss": 0.168,
      "step": 7737
    },
    {
      "epoch": 2.19,
      "grad_norm": 2.354959799162431,
      "learning_rate": 3.625478200093534e-06,
      "loss": 0.2255,
      "step": 7738
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.9443938729739005,
      "learning_rate": 3.6231293451357994e-06,
      "loss": 0.1941,
      "step": 7739
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.8318442102301615,
      "learning_rate": 3.6207810829609414e-06,
      "loss": 0.1824,
      "step": 7740
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.8323132466700631,
      "learning_rate": 3.6184334137872514e-06,
      "loss": 0.1675,
      "step": 7741
    },
    {
      "epoch": 2.19,
      "grad_norm": 2.2100470543214428,
      "learning_rate": 3.6160863378329636e-06,
      "loss": 0.2554,
      "step": 7742
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.9308419493082514,
      "learning_rate": 3.613739855316257e-06,
      "loss": 0.2085,
      "step": 7743
    },
    {
      "epoch": 2.19,
      "grad_norm": 2.063375918173881,
      "learning_rate": 3.6113939664552567e-06,
      "loss": 0.2078,
      "step": 7744
    },
    {
      "epoch": 2.19,
      "grad_norm": 2.134124031974081,
      "learning_rate": 3.6090486714680396e-06,
      "loss": 0.2242,
      "step": 7745
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.8995566917214965,
      "learning_rate": 3.606703970572607e-06,
      "loss": 0.2102,
      "step": 7746
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.8632390781429466,
      "learning_rate": 3.6043598639869293e-06,
      "loss": 0.1858,
      "step": 7747
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.820095847675935,
      "learning_rate": 3.6020163519289077e-06,
      "loss": 0.1943,
      "step": 7748
    },
    {
      "epoch": 2.19,
      "grad_norm": 2.0409845568923197,
      "learning_rate": 3.5996734346163897e-06,
      "loss": 0.21,
      "step": 7749
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.9413886801007167,
      "learning_rate": 3.5973311122671695e-06,
      "loss": 0.2075,
      "step": 7750
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.8848408787292197,
      "learning_rate": 3.594989385098985e-06,
      "loss": 0.1929,
      "step": 7751
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.8307661680860317,
      "learning_rate": 3.592648253329516e-06,
      "loss": 0.183,
      "step": 7752
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.8401664458273117,
      "learning_rate": 3.590307717176401e-06,
      "loss": 0.1588,
      "step": 7753
    },
    {
      "epoch": 2.19,
      "grad_norm": 2.281005448869145,
      "learning_rate": 3.587967776857201e-06,
      "loss": 0.2505,
      "step": 7754
    },
    {
      "epoch": 2.19,
      "grad_norm": 2.038876334552482,
      "learning_rate": 3.585628432589432e-06,
      "loss": 0.2207,
      "step": 7755
    },
    {
      "epoch": 2.19,
      "grad_norm": 2.046972373190518,
      "learning_rate": 3.583289684590564e-06,
      "loss": 0.2122,
      "step": 7756
    },
    {
      "epoch": 2.19,
      "grad_norm": 2.121163959620324,
      "learning_rate": 3.5809515330779985e-06,
      "loss": 0.1973,
      "step": 7757
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.9174706275047173,
      "learning_rate": 3.5786139782690867e-06,
      "loss": 0.2278,
      "step": 7758
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.9215333289460204,
      "learning_rate": 3.5762770203811225e-06,
      "loss": 0.1943,
      "step": 7759
    },
    {
      "epoch": 2.19,
      "grad_norm": 2.0097373639424565,
      "learning_rate": 3.5739406596313474e-06,
      "loss": 0.2175,
      "step": 7760
    },
    {
      "epoch": 2.19,
      "grad_norm": 2.1046556174020528,
      "learning_rate": 3.57160489623694e-06,
      "loss": 0.2222,
      "step": 7761
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.6869635336790005,
      "learning_rate": 3.56926973041504e-06,
      "loss": 0.1562,
      "step": 7762
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.880500668691898,
      "learning_rate": 3.5669351623827077e-06,
      "loss": 0.14,
      "step": 7763
    },
    {
      "epoch": 2.19,
      "grad_norm": 2.3062378646198898,
      "learning_rate": 3.564601192356968e-06,
      "loss": 0.2194,
      "step": 7764
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.9930434605684746,
      "learning_rate": 3.5622678205547824e-06,
      "loss": 0.2163,
      "step": 7765
    },
    {
      "epoch": 2.19,
      "grad_norm": 2.1347549600159907,
      "learning_rate": 3.559935047193055e-06,
      "loss": 0.2473,
      "step": 7766
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.9920988763717324,
      "learning_rate": 3.557602872488638e-06,
      "loss": 0.1743,
      "step": 7767
    },
    {
      "epoch": 2.19,
      "grad_norm": 2.0375504366275177,
      "learning_rate": 3.5552712966583247e-06,
      "loss": 0.2086,
      "step": 7768
    },
    {
      "epoch": 2.19,
      "grad_norm": 2.0470437406773465,
      "learning_rate": 3.5529403199188517e-06,
      "loss": 0.2151,
      "step": 7769
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.730319209857267,
      "learning_rate": 3.5506099424869133e-06,
      "loss": 0.1635,
      "step": 7770
    },
    {
      "epoch": 2.19,
      "grad_norm": 2.0059380216453633,
      "learning_rate": 3.5482801645791266e-06,
      "loss": 0.1746,
      "step": 7771
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.9843615330379762,
      "learning_rate": 3.545950986412063e-06,
      "loss": 0.1664,
      "step": 7772
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.8944060833185101,
      "learning_rate": 3.5436224082022476e-06,
      "loss": 0.1597,
      "step": 7773
    },
    {
      "epoch": 2.2,
      "grad_norm": 2.251487655150563,
      "learning_rate": 3.5412944301661356e-06,
      "loss": 0.2173,
      "step": 7774
    },
    {
      "epoch": 2.2,
      "grad_norm": 2.018111220138519,
      "learning_rate": 3.5389670525201335e-06,
      "loss": 0.2065,
      "step": 7775
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.7967560212567915,
      "learning_rate": 3.5366402754805885e-06,
      "loss": 0.1887,
      "step": 7776
    },
    {
      "epoch": 2.2,
      "grad_norm": 2.2193101993627846,
      "learning_rate": 3.5343140992637946e-06,
      "loss": 0.2063,
      "step": 7777
    },
    {
      "epoch": 2.2,
      "grad_norm": 2.1683136425318743,
      "learning_rate": 3.531988524085991e-06,
      "loss": 0.2119,
      "step": 7778
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.9967458948049783,
      "learning_rate": 3.5296635501633558e-06,
      "loss": 0.1802,
      "step": 7779
    },
    {
      "epoch": 2.2,
      "grad_norm": 2.1674914273045194,
      "learning_rate": 3.5273391777120136e-06,
      "loss": 0.1921,
      "step": 7780
    },
    {
      "epoch": 2.2,
      "grad_norm": 2.44203883833334,
      "learning_rate": 3.525015406948039e-06,
      "loss": 0.2703,
      "step": 7781
    },
    {
      "epoch": 2.2,
      "grad_norm": 2.3044381075526306,
      "learning_rate": 3.522692238087445e-06,
      "loss": 0.1656,
      "step": 7782
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.7384474650594728,
      "learning_rate": 3.5203696713461866e-06,
      "loss": 0.1782,
      "step": 7783
    },
    {
      "epoch": 2.2,
      "grad_norm": 2.1233306918958332,
      "learning_rate": 3.5180477069401664e-06,
      "loss": 0.1804,
      "step": 7784
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.9438434071797484,
      "learning_rate": 3.515726345085232e-06,
      "loss": 0.2344,
      "step": 7785
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.9263721401137617,
      "learning_rate": 3.5134055859971674e-06,
      "loss": 0.1914,
      "step": 7786
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.8979871983703227,
      "learning_rate": 3.511085429891716e-06,
      "loss": 0.1663,
      "step": 7787
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.917079411939141,
      "learning_rate": 3.508765876984549e-06,
      "loss": 0.1846,
      "step": 7788
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.8564675381345985,
      "learning_rate": 3.5064469274912847e-06,
      "loss": 0.1617,
      "step": 7789
    },
    {
      "epoch": 2.2,
      "grad_norm": 2.529299925343539,
      "learning_rate": 3.504128581627497e-06,
      "loss": 0.2671,
      "step": 7790
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.920330245611529,
      "learning_rate": 3.5018108396086945e-06,
      "loss": 0.2087,
      "step": 7791
    },
    {
      "epoch": 2.2,
      "grad_norm": 2.1355403115406597,
      "learning_rate": 3.4994937016503206e-06,
      "loss": 0.175,
      "step": 7792
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.7062065477156416,
      "learning_rate": 3.4971771679677826e-06,
      "loss": 0.1553,
      "step": 7793
    },
    {
      "epoch": 2.2,
      "grad_norm": 2.367971293782869,
      "learning_rate": 3.494861238776418e-06,
      "loss": 0.2227,
      "step": 7794
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.966092042545038,
      "learning_rate": 3.492545914291512e-06,
      "loss": 0.1882,
      "step": 7795
    },
    {
      "epoch": 2.2,
      "grad_norm": 2.155190281249936,
      "learning_rate": 3.490231194728293e-06,
      "loss": 0.1726,
      "step": 7796
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.8591288407863433,
      "learning_rate": 3.4879170803019336e-06,
      "loss": 0.1968,
      "step": 7797
    },
    {
      "epoch": 2.2,
      "grad_norm": 2.065708837699305,
      "learning_rate": 3.485603571227545e-06,
      "loss": 0.2246,
      "step": 7798
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.5177105580651253,
      "learning_rate": 3.483290667720196e-06,
      "loss": 0.1157,
      "step": 7799
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.9639889228501761,
      "learning_rate": 3.480978369994885e-06,
      "loss": 0.1934,
      "step": 7800
    },
    {
      "epoch": 2.2,
      "grad_norm": 2.0802279285754786,
      "learning_rate": 3.478666678266559e-06,
      "loss": 0.1755,
      "step": 7801
    },
    {
      "epoch": 2.2,
      "grad_norm": 2.2580742882517386,
      "learning_rate": 3.4763555927501113e-06,
      "loss": 0.2109,
      "step": 7802
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.913532889208487,
      "learning_rate": 3.4740451136603737e-06,
      "loss": 0.1833,
      "step": 7803
    },
    {
      "epoch": 2.2,
      "grad_norm": 2.7329115251552913,
      "learning_rate": 3.4717352412121254e-06,
      "loss": 0.2388,
      "step": 7804
    },
    {
      "epoch": 2.2,
      "grad_norm": 2.0894096220578997,
      "learning_rate": 3.4694259756200888e-06,
      "loss": 0.1968,
      "step": 7805
    },
    {
      "epoch": 2.2,
      "grad_norm": 2.248362485069967,
      "learning_rate": 3.467117317098925e-06,
      "loss": 0.2366,
      "step": 7806
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.9085531078981164,
      "learning_rate": 3.4648092658632506e-06,
      "loss": 0.1729,
      "step": 7807
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.6246935233389763,
      "learning_rate": 3.4625018221276165e-06,
      "loss": 0.1707,
      "step": 7808
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.9478906671621374,
      "learning_rate": 3.4601949861065086e-06,
      "loss": 0.228,
      "step": 7809
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.7748273803630477,
      "learning_rate": 3.4578887580143793e-06,
      "loss": 0.1714,
      "step": 7810
    },
    {
      "epoch": 2.21,
      "grad_norm": 2.2227694835634897,
      "learning_rate": 3.4555831380656044e-06,
      "loss": 0.1819,
      "step": 7811
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.873366120387973,
      "learning_rate": 3.453278126474513e-06,
      "loss": 0.1782,
      "step": 7812
    },
    {
      "epoch": 2.21,
      "grad_norm": 2.405157918483297,
      "learning_rate": 3.4509737234553752e-06,
      "loss": 0.24,
      "step": 7813
    },
    {
      "epoch": 2.21,
      "grad_norm": 2.221877042716796,
      "learning_rate": 3.4486699292224022e-06,
      "loss": 0.2085,
      "step": 7814
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.7438077696271792,
      "learning_rate": 3.4463667439897486e-06,
      "loss": 0.1458,
      "step": 7815
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.8852699478645916,
      "learning_rate": 3.4440641679715204e-06,
      "loss": 0.177,
      "step": 7816
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.9195733879056565,
      "learning_rate": 3.4417622013817597e-06,
      "loss": 0.1731,
      "step": 7817
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.9685857776415598,
      "learning_rate": 3.4394608444344514e-06,
      "loss": 0.1754,
      "step": 7818
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.9848799733714668,
      "learning_rate": 3.437160097343526e-06,
      "loss": 0.1667,
      "step": 7819
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.7263246153100267,
      "learning_rate": 3.4348599603228584e-06,
      "loss": 0.1638,
      "step": 7820
    },
    {
      "epoch": 2.21,
      "grad_norm": 2.199745840965309,
      "learning_rate": 3.432560433586264e-06,
      "loss": 0.1985,
      "step": 7821
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.8791152419266715,
      "learning_rate": 3.4302615173475017e-06,
      "loss": 0.1545,
      "step": 7822
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.8881547562554521,
      "learning_rate": 3.4279632118202744e-06,
      "loss": 0.1956,
      "step": 7823
    },
    {
      "epoch": 2.21,
      "grad_norm": 2.182617789604594,
      "learning_rate": 3.4256655172182328e-06,
      "loss": 0.1975,
      "step": 7824
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.8866877291892377,
      "learning_rate": 3.423368433754969e-06,
      "loss": 0.187,
      "step": 7825
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.7320430382635716,
      "learning_rate": 3.4210719616440025e-06,
      "loss": 0.1432,
      "step": 7826
    },
    {
      "epoch": 2.21,
      "grad_norm": 2.156470851203495,
      "learning_rate": 3.418776101098823e-06,
      "loss": 0.219,
      "step": 7827
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.9768823007646565,
      "learning_rate": 3.416480852332844e-06,
      "loss": 0.2107,
      "step": 7828
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.8651912931745227,
      "learning_rate": 3.4141862155594283e-06,
      "loss": 0.1628,
      "step": 7829
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.82749487129536,
      "learning_rate": 3.411892190991882e-06,
      "loss": 0.199,
      "step": 7830
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.8912704020501308,
      "learning_rate": 3.4095987788434538e-06,
      "loss": 0.1777,
      "step": 7831
    },
    {
      "epoch": 2.21,
      "grad_norm": 2.0555017945001173,
      "learning_rate": 3.407305979327331e-06,
      "loss": 0.1987,
      "step": 7832
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.978912721575286,
      "learning_rate": 3.405013792656655e-06,
      "loss": 0.2068,
      "step": 7833
    },
    {
      "epoch": 2.21,
      "grad_norm": 2.149209872993488,
      "learning_rate": 3.402722219044505e-06,
      "loss": 0.1687,
      "step": 7834
    },
    {
      "epoch": 2.21,
      "grad_norm": 2.0287209549482323,
      "learning_rate": 3.40043125870389e-06,
      "loss": 0.1875,
      "step": 7835
    },
    {
      "epoch": 2.21,
      "grad_norm": 2.088455861971149,
      "learning_rate": 3.3981409118477847e-06,
      "loss": 0.1672,
      "step": 7836
    },
    {
      "epoch": 2.21,
      "grad_norm": 2.1607467116813344,
      "learning_rate": 3.3958511786890923e-06,
      "loss": 0.199,
      "step": 7837
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.7457266671609575,
      "learning_rate": 3.3935620594406614e-06,
      "loss": 0.1683,
      "step": 7838
    },
    {
      "epoch": 2.21,
      "grad_norm": 2.1318963836305165,
      "learning_rate": 3.3912735543152864e-06,
      "loss": 0.2041,
      "step": 7839
    },
    {
      "epoch": 2.21,
      "grad_norm": 2.038529680096669,
      "learning_rate": 3.3889856635257024e-06,
      "loss": 0.1755,
      "step": 7840
    },
    {
      "epoch": 2.21,
      "grad_norm": 2.458178384273272,
      "learning_rate": 3.3866983872845826e-06,
      "loss": 0.251,
      "step": 7841
    },
    {
      "epoch": 2.21,
      "grad_norm": 2.0324841498722916,
      "learning_rate": 3.3844117258045604e-06,
      "loss": 0.2048,
      "step": 7842
    },
    {
      "epoch": 2.21,
      "grad_norm": 2.100534411276142,
      "learning_rate": 3.382125679298185e-06,
      "loss": 0.2039,
      "step": 7843
    },
    {
      "epoch": 2.22,
      "grad_norm": 2.373875679138592,
      "learning_rate": 3.3798402479779747e-06,
      "loss": 0.2639,
      "step": 7844
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.858828084376985,
      "learning_rate": 3.3775554320563753e-06,
      "loss": 0.1741,
      "step": 7845
    },
    {
      "epoch": 2.22,
      "grad_norm": 2.0147955473332364,
      "learning_rate": 3.375271231745779e-06,
      "loss": 0.2446,
      "step": 7846
    },
    {
      "epoch": 2.22,
      "grad_norm": 2.155487609289334,
      "learning_rate": 3.372987647258521e-06,
      "loss": 0.1841,
      "step": 7847
    },
    {
      "epoch": 2.22,
      "grad_norm": 2.276830001675697,
      "learning_rate": 3.370704678806881e-06,
      "loss": 0.1738,
      "step": 7848
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.5848739877035245,
      "learning_rate": 3.368422326603075e-06,
      "loss": 0.1234,
      "step": 7849
    },
    {
      "epoch": 2.22,
      "grad_norm": 2.012509295082376,
      "learning_rate": 3.366140590859276e-06,
      "loss": 0.2124,
      "step": 7850
    },
    {
      "epoch": 2.22,
      "grad_norm": 2.020123075558816,
      "learning_rate": 3.3638594717875807e-06,
      "loss": 0.2053,
      "step": 7851
    },
    {
      "epoch": 2.22,
      "grad_norm": 2.1214171234971366,
      "learning_rate": 3.361578969600039e-06,
      "loss": 0.2269,
      "step": 7852
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.9164093517855003,
      "learning_rate": 3.359299084508648e-06,
      "loss": 0.1489,
      "step": 7853
    },
    {
      "epoch": 2.22,
      "grad_norm": 2.1399028066174917,
      "learning_rate": 3.3570198167253386e-06,
      "loss": 0.1882,
      "step": 7854
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.7196998748643406,
      "learning_rate": 3.354741166461989e-06,
      "loss": 0.1589,
      "step": 7855
    },
    {
      "epoch": 2.22,
      "grad_norm": 2.0171819614088626,
      "learning_rate": 3.352463133930417e-06,
      "loss": 0.1724,
      "step": 7856
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.9155980315987513,
      "learning_rate": 3.350185719342385e-06,
      "loss": 0.1702,
      "step": 7857
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.8730073602894466,
      "learning_rate": 3.347908922909594e-06,
      "loss": 0.1423,
      "step": 7858
    },
    {
      "epoch": 2.22,
      "grad_norm": 2.17525352163227,
      "learning_rate": 3.345632744843702e-06,
      "loss": 0.2627,
      "step": 7859
    },
    {
      "epoch": 2.22,
      "grad_norm": 2.2022397999471433,
      "learning_rate": 3.343357185356284e-06,
      "loss": 0.1978,
      "step": 7860
    },
    {
      "epoch": 2.22,
      "grad_norm": 2.182444190566135,
      "learning_rate": 3.3410822446588833e-06,
      "loss": 0.1931,
      "step": 7861
    },
    {
      "epoch": 2.22,
      "grad_norm": 2.147753307258201,
      "learning_rate": 3.33880792296297e-06,
      "loss": 0.2228,
      "step": 7862
    },
    {
      "epoch": 2.22,
      "grad_norm": 2.0473813510186507,
      "learning_rate": 3.3365342204799613e-06,
      "loss": 0.2087,
      "step": 7863
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.8942591759185916,
      "learning_rate": 3.3342611374212176e-06,
      "loss": 0.1775,
      "step": 7864
    },
    {
      "epoch": 2.22,
      "grad_norm": 2.0570253356766526,
      "learning_rate": 3.33198867399804e-06,
      "loss": 0.2051,
      "step": 7865
    },
    {
      "epoch": 2.22,
      "grad_norm": 2.228531378633083,
      "learning_rate": 3.3297168304216686e-06,
      "loss": 0.1895,
      "step": 7866
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.9316235369210084,
      "learning_rate": 3.3274456069033024e-06,
      "loss": 0.1908,
      "step": 7867
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.8451926275302761,
      "learning_rate": 3.3251750036540585e-06,
      "loss": 0.1302,
      "step": 7868
    },
    {
      "epoch": 2.22,
      "grad_norm": 2.1221910141722975,
      "learning_rate": 3.322905020885009e-06,
      "loss": 0.1819,
      "step": 7869
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.9604944269965465,
      "learning_rate": 3.3206356588071733e-06,
      "loss": 0.1815,
      "step": 7870
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.9952807406630753,
      "learning_rate": 3.3183669176315046e-06,
      "loss": 0.1975,
      "step": 7871
    },
    {
      "epoch": 2.22,
      "grad_norm": 2.2433161285656413,
      "learning_rate": 3.3160987975689017e-06,
      "loss": 0.2129,
      "step": 7872
    },
    {
      "epoch": 2.22,
      "grad_norm": 2.0690316551087253,
      "learning_rate": 3.3138312988302055e-06,
      "loss": 0.1804,
      "step": 7873
    },
    {
      "epoch": 2.22,
      "grad_norm": 2.067921493613672,
      "learning_rate": 3.3115644216261977e-06,
      "loss": 0.249,
      "step": 7874
    },
    {
      "epoch": 2.22,
      "grad_norm": 2.142828584250023,
      "learning_rate": 3.309298166167599e-06,
      "loss": 0.1897,
      "step": 7875
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.7282320618871636,
      "learning_rate": 3.30703253266509e-06,
      "loss": 0.1472,
      "step": 7876
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.7957660997474838,
      "learning_rate": 3.304767521329263e-06,
      "loss": 0.1973,
      "step": 7877
    },
    {
      "epoch": 2.22,
      "grad_norm": 2.023878539325994,
      "learning_rate": 3.302503132370681e-06,
      "loss": 0.1992,
      "step": 7878
    },
    {
      "epoch": 2.23,
      "grad_norm": 2.09892004028073,
      "learning_rate": 3.3002393659998357e-06,
      "loss": 0.238,
      "step": 7879
    },
    {
      "epoch": 2.23,
      "grad_norm": 2.359108147504145,
      "learning_rate": 3.2979762224271616e-06,
      "loss": 0.2871,
      "step": 7880
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.9122420129278284,
      "learning_rate": 3.295713701863036e-06,
      "loss": 0.1697,
      "step": 7881
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.9395039279039454,
      "learning_rate": 3.2934518045177798e-06,
      "loss": 0.1655,
      "step": 7882
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.991212707295546,
      "learning_rate": 3.291190530601651e-06,
      "loss": 0.2349,
      "step": 7883
    },
    {
      "epoch": 2.23,
      "grad_norm": 2.1344701344353405,
      "learning_rate": 3.288929880324865e-06,
      "loss": 0.1941,
      "step": 7884
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.9592502223329924,
      "learning_rate": 3.2866698538975572e-06,
      "loss": 0.1692,
      "step": 7885
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.9014531715536633,
      "learning_rate": 3.284410451529816e-06,
      "loss": 0.1941,
      "step": 7886
    },
    {
      "epoch": 2.23,
      "grad_norm": 2.377814122327505,
      "learning_rate": 3.2821516734316772e-06,
      "loss": 0.2422,
      "step": 7887
    },
    {
      "epoch": 2.23,
      "grad_norm": 2.1481083403066976,
      "learning_rate": 3.279893519813111e-06,
      "loss": 0.1946,
      "step": 7888
    },
    {
      "epoch": 2.23,
      "grad_norm": 2.1317121667589034,
      "learning_rate": 3.2776359908840306e-06,
      "loss": 0.181,
      "step": 7889
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.8696523849117317,
      "learning_rate": 3.275379086854292e-06,
      "loss": 0.1768,
      "step": 7890
    },
    {
      "epoch": 2.23,
      "grad_norm": 2.1820082210832914,
      "learning_rate": 3.2731228079336928e-06,
      "loss": 0.1863,
      "step": 7891
    },
    {
      "epoch": 2.23,
      "grad_norm": 2.2513330316569267,
      "learning_rate": 3.2708671543319712e-06,
      "loss": 0.1914,
      "step": 7892
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.9144953778158516,
      "learning_rate": 3.2686121262588165e-06,
      "loss": 0.1785,
      "step": 7893
    },
    {
      "epoch": 2.23,
      "grad_norm": 2.147718758331818,
      "learning_rate": 3.2663577239238443e-06,
      "loss": 0.2019,
      "step": 7894
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.9233234096132446,
      "learning_rate": 3.264103947536619e-06,
      "loss": 0.1765,
      "step": 7895
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.8138854964976898,
      "learning_rate": 3.2618507973066536e-06,
      "loss": 0.1768,
      "step": 7896
    },
    {
      "epoch": 2.23,
      "grad_norm": 2.2671047585625885,
      "learning_rate": 3.259598273443394e-06,
      "loss": 0.1826,
      "step": 7897
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.9577310717103011,
      "learning_rate": 3.257346376156233e-06,
      "loss": 0.1648,
      "step": 7898
    },
    {
      "epoch": 2.23,
      "grad_norm": 2.0455080222893938,
      "learning_rate": 3.2550951056545e-06,
      "loss": 0.1873,
      "step": 7899
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.9731313122089444,
      "learning_rate": 3.252844462147472e-06,
      "loss": 0.1938,
      "step": 7900
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.8730541559690215,
      "learning_rate": 3.2505944458443626e-06,
      "loss": 0.1729,
      "step": 7901
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.7373452888151417,
      "learning_rate": 3.248345056954332e-06,
      "loss": 0.1658,
      "step": 7902
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.906984638639604,
      "learning_rate": 3.2460962956864727e-06,
      "loss": 0.1511,
      "step": 7903
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.7205083626032838,
      "learning_rate": 3.243848162249835e-06,
      "loss": 0.1597,
      "step": 7904
    },
    {
      "epoch": 2.23,
      "grad_norm": 2.4724259396474397,
      "learning_rate": 3.2416006568533987e-06,
      "loss": 0.2124,
      "step": 7905
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.7895240443035298,
      "learning_rate": 3.239353779706086e-06,
      "loss": 0.1433,
      "step": 7906
    },
    {
      "epoch": 2.23,
      "grad_norm": 2.0317517317533733,
      "learning_rate": 3.2371075310167634e-06,
      "loss": 0.1921,
      "step": 7907
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.778082790490562,
      "learning_rate": 3.234861910994238e-06,
      "loss": 0.1382,
      "step": 7908
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.959552086095138,
      "learning_rate": 3.2326169198472555e-06,
      "loss": 0.1543,
      "step": 7909
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.8398014089250305,
      "learning_rate": 3.230372557784518e-06,
      "loss": 0.156,
      "step": 7910
    },
    {
      "epoch": 2.23,
      "grad_norm": 2.1101139904077777,
      "learning_rate": 3.2281288250146447e-06,
      "loss": 0.1953,
      "step": 7911
    },
    {
      "epoch": 2.23,
      "grad_norm": 2.166835051627543,
      "learning_rate": 3.2258857217462115e-06,
      "loss": 0.1941,
      "step": 7912
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.6578223299491228,
      "learning_rate": 3.22364324818774e-06,
      "loss": 0.1444,
      "step": 7913
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.7461278485166398,
      "learning_rate": 3.2214014045476815e-06,
      "loss": 0.161,
      "step": 7914
    },
    {
      "epoch": 2.24,
      "grad_norm": 2.1269664409900018,
      "learning_rate": 3.219160191034435e-06,
      "loss": 0.2139,
      "step": 7915
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.6949411989691339,
      "learning_rate": 3.2169196078563403e-06,
      "loss": 0.1582,
      "step": 7916
    },
    {
      "epoch": 2.24,
      "grad_norm": 2.326427447598449,
      "learning_rate": 3.2146796552216773e-06,
      "loss": 0.2285,
      "step": 7917
    },
    {
      "epoch": 2.24,
      "grad_norm": 2.502462358884763,
      "learning_rate": 3.212440333338668e-06,
      "loss": 0.2483,
      "step": 7918
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.8003762841877293,
      "learning_rate": 3.210201642415477e-06,
      "loss": 0.1599,
      "step": 7919
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.7785397781877246,
      "learning_rate": 3.2079635826602053e-06,
      "loss": 0.1597,
      "step": 7920
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.9063727372866617,
      "learning_rate": 3.205726154280905e-06,
      "loss": 0.157,
      "step": 7921
    },
    {
      "epoch": 2.24,
      "grad_norm": 2.0762735167556237,
      "learning_rate": 3.203489357485562e-06,
      "loss": 0.2009,
      "step": 7922
    },
    {
      "epoch": 2.24,
      "grad_norm": 2.2129476068865204,
      "learning_rate": 3.201253192482102e-06,
      "loss": 0.217,
      "step": 7923
    },
    {
      "epoch": 2.24,
      "grad_norm": 2.6124608739566026,
      "learning_rate": 3.1990176594783983e-06,
      "loss": 0.1924,
      "step": 7924
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.8122106327565672,
      "learning_rate": 3.196782758682261e-06,
      "loss": 0.2063,
      "step": 7925
    },
    {
      "epoch": 2.24,
      "grad_norm": 2.0829624137770675,
      "learning_rate": 3.194548490301439e-06,
      "loss": 0.2075,
      "step": 7926
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.9711340393777745,
      "learning_rate": 3.1923148545436357e-06,
      "loss": 0.2249,
      "step": 7927
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.9329137170913773,
      "learning_rate": 3.1900818516164766e-06,
      "loss": 0.1753,
      "step": 7928
    },
    {
      "epoch": 2.24,
      "grad_norm": 2.126279807640404,
      "learning_rate": 3.1878494817275374e-06,
      "loss": 0.1854,
      "step": 7929
    },
    {
      "epoch": 2.24,
      "grad_norm": 2.386044945526648,
      "learning_rate": 3.185617745084343e-06,
      "loss": 0.23,
      "step": 7930
    },
    {
      "epoch": 2.24,
      "grad_norm": 2.301080574507414,
      "learning_rate": 3.1833866418943503e-06,
      "loss": 0.2294,
      "step": 7931
    },
    {
      "epoch": 2.24,
      "grad_norm": 2.1739370237781546,
      "learning_rate": 3.1811561723649496e-06,
      "loss": 0.1819,
      "step": 7932
    },
    {
      "epoch": 2.24,
      "grad_norm": 2.1364966775378327,
      "learning_rate": 3.1789263367034918e-06,
      "loss": 0.2336,
      "step": 7933
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.766868810847437,
      "learning_rate": 3.1766971351172547e-06,
      "loss": 0.1765,
      "step": 7934
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.9623877547299304,
      "learning_rate": 3.174468567813461e-06,
      "loss": 0.2032,
      "step": 7935
    },
    {
      "epoch": 2.24,
      "grad_norm": 2.020111242196448,
      "learning_rate": 3.172240634999275e-06,
      "loss": 0.1746,
      "step": 7936
    },
    {
      "epoch": 2.24,
      "grad_norm": 2.0334384517171626,
      "learning_rate": 3.170013336881801e-06,
      "loss": 0.1868,
      "step": 7937
    },
    {
      "epoch": 2.24,
      "grad_norm": 2.197186585333072,
      "learning_rate": 3.167786673668082e-06,
      "loss": 0.2053,
      "step": 7938
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.956533469522605,
      "learning_rate": 3.1655606455651134e-06,
      "loss": 0.1926,
      "step": 7939
    },
    {
      "epoch": 2.24,
      "grad_norm": 2.1584385557693677,
      "learning_rate": 3.163335252779811e-06,
      "loss": 0.1675,
      "step": 7940
    },
    {
      "epoch": 2.24,
      "grad_norm": 2.039849524820674,
      "learning_rate": 3.161110495519053e-06,
      "loss": 0.2302,
      "step": 7941
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.9441617559361524,
      "learning_rate": 3.1588863739896457e-06,
      "loss": 0.1863,
      "step": 7942
    },
    {
      "epoch": 2.24,
      "grad_norm": 2.142475405268512,
      "learning_rate": 3.1566628883983395e-06,
      "loss": 0.2302,
      "step": 7943
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.678982483974594,
      "learning_rate": 3.154440038951827e-06,
      "loss": 0.1501,
      "step": 7944
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.8510260418331272,
      "learning_rate": 3.152217825856738e-06,
      "loss": 0.1952,
      "step": 7945
    },
    {
      "epoch": 2.24,
      "grad_norm": 2.502300043748547,
      "learning_rate": 3.1499962493196446e-06,
      "loss": 0.2275,
      "step": 7946
    },
    {
      "epoch": 2.24,
      "grad_norm": 2.3351559370579116,
      "learning_rate": 3.1477753095470654e-06,
      "loss": 0.2061,
      "step": 7947
    },
    {
      "epoch": 2.24,
      "grad_norm": 2.0264914781954544,
      "learning_rate": 3.1455550067454555e-06,
      "loss": 0.1809,
      "step": 7948
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.7809870761503486,
      "learning_rate": 3.143335341121202e-06,
      "loss": 0.1587,
      "step": 7949
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.8262416448386727,
      "learning_rate": 3.1411163128806497e-06,
      "loss": 0.197,
      "step": 7950
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.983030111454883,
      "learning_rate": 3.138897922230074e-06,
      "loss": 0.1807,
      "step": 7951
    },
    {
      "epoch": 2.25,
      "grad_norm": 2.3189323741836367,
      "learning_rate": 3.13668016937569e-06,
      "loss": 0.1716,
      "step": 7952
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.966111521885731,
      "learning_rate": 3.1344630545236576e-06,
      "loss": 0.2288,
      "step": 7953
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.9272996676316458,
      "learning_rate": 3.132246577880077e-06,
      "loss": 0.1418,
      "step": 7954
    },
    {
      "epoch": 2.25,
      "grad_norm": 2.037206186841154,
      "learning_rate": 3.1300307396509833e-06,
      "loss": 0.1809,
      "step": 7955
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.8959703035182154,
      "learning_rate": 3.1278155400423673e-06,
      "loss": 0.1633,
      "step": 7956
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.7048408601277039,
      "learning_rate": 3.125600979260136e-06,
      "loss": 0.1495,
      "step": 7957
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.64007638754271,
      "learning_rate": 3.123387057510162e-06,
      "loss": 0.1606,
      "step": 7958
    },
    {
      "epoch": 2.25,
      "grad_norm": 2.1738022823474683,
      "learning_rate": 3.121173774998245e-06,
      "loss": 0.2136,
      "step": 7959
    },
    {
      "epoch": 2.25,
      "grad_norm": 2.011891261070291,
      "learning_rate": 3.118961131930127e-06,
      "loss": 0.1663,
      "step": 7960
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.9456846174826845,
      "learning_rate": 3.1167491285114928e-06,
      "loss": 0.1504,
      "step": 7961
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.9449709445624057,
      "learning_rate": 3.1145377649479635e-06,
      "loss": 0.187,
      "step": 7962
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.9505797412194814,
      "learning_rate": 3.1123270414451035e-06,
      "loss": 0.1523,
      "step": 7963
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.706011853440649,
      "learning_rate": 3.110116958208422e-06,
      "loss": 0.1624,
      "step": 7964
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.6683546740946424,
      "learning_rate": 3.107907515443367e-06,
      "loss": 0.1328,
      "step": 7965
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.8109994581055102,
      "learning_rate": 3.1056987133553118e-06,
      "loss": 0.2161,
      "step": 7966
    },
    {
      "epoch": 2.25,
      "grad_norm": 2.2616576752648703,
      "learning_rate": 3.103490552149595e-06,
      "loss": 0.2117,
      "step": 7967
    },
    {
      "epoch": 2.25,
      "grad_norm": 2.324945058350115,
      "learning_rate": 3.1012830320314793e-06,
      "loss": 0.1521,
      "step": 7968
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.9718369552034802,
      "learning_rate": 3.0990761532061707e-06,
      "loss": 0.1594,
      "step": 7969
    },
    {
      "epoch": 2.25,
      "grad_norm": 2.1320182111653376,
      "learning_rate": 3.0968699158788185e-06,
      "loss": 0.1658,
      "step": 7970
    },
    {
      "epoch": 2.25,
      "grad_norm": 2.0009430975238507,
      "learning_rate": 3.0946643202545113e-06,
      "loss": 0.1733,
      "step": 7971
    },
    {
      "epoch": 2.25,
      "grad_norm": 2.0503958103928537,
      "learning_rate": 3.092459366538272e-06,
      "loss": 0.1875,
      "step": 7972
    },
    {
      "epoch": 2.25,
      "grad_norm": 2.2630950527568694,
      "learning_rate": 3.090255054935081e-06,
      "loss": 0.1948,
      "step": 7973
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.7778034477147802,
      "learning_rate": 3.0880513856498363e-06,
      "loss": 0.1807,
      "step": 7974
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.9732463961670437,
      "learning_rate": 3.0858483588873878e-06,
      "loss": 0.1777,
      "step": 7975
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.975045810446466,
      "learning_rate": 3.0836459748525316e-06,
      "loss": 0.1875,
      "step": 7976
    },
    {
      "epoch": 2.25,
      "grad_norm": 2.0828585276349774,
      "learning_rate": 3.081444233749994e-06,
      "loss": 0.1658,
      "step": 7977
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.8462950050554698,
      "learning_rate": 3.0792431357844444e-06,
      "loss": 0.1593,
      "step": 7978
    },
    {
      "epoch": 2.25,
      "grad_norm": 2.114695889549059,
      "learning_rate": 3.0770426811604946e-06,
      "loss": 0.2075,
      "step": 7979
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.924623749566638,
      "learning_rate": 3.0748428700826938e-06,
      "loss": 0.1692,
      "step": 7980
    },
    {
      "epoch": 2.25,
      "grad_norm": 2.157269786549259,
      "learning_rate": 3.07264370275553e-06,
      "loss": 0.2124,
      "step": 7981
    },
    {
      "epoch": 2.25,
      "grad_norm": 2.046868596954335,
      "learning_rate": 3.0704451793834433e-06,
      "loss": 0.1804,
      "step": 7982
    },
    {
      "epoch": 2.25,
      "grad_norm": 2.1700294861208693,
      "learning_rate": 3.0682473001707925e-06,
      "loss": 0.229,
      "step": 7983
    },
    {
      "epoch": 2.25,
      "grad_norm": 2.4660957266226227,
      "learning_rate": 3.0660500653218973e-06,
      "loss": 0.2222,
      "step": 7984
    },
    {
      "epoch": 2.26,
      "grad_norm": 2.001549676938995,
      "learning_rate": 3.0638534750410065e-06,
      "loss": 0.1765,
      "step": 7985
    },
    {
      "epoch": 2.26,
      "grad_norm": 2.0512547089527504,
      "learning_rate": 3.0616575295323105e-06,
      "loss": 0.197,
      "step": 7986
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.7667223239258496,
      "learning_rate": 3.059462228999941e-06,
      "loss": 0.1396,
      "step": 7987
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.7824068306646084,
      "learning_rate": 3.0572675736479696e-06,
      "loss": 0.1433,
      "step": 7988
    },
    {
      "epoch": 2.26,
      "grad_norm": 2.102787998812191,
      "learning_rate": 3.055073563680404e-06,
      "loss": 0.1941,
      "step": 7989
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.7280189696242736,
      "learning_rate": 3.0528801993012056e-06,
      "loss": 0.1848,
      "step": 7990
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.876357734552089,
      "learning_rate": 3.050687480714256e-06,
      "loss": 0.1807,
      "step": 7991
    },
    {
      "epoch": 2.26,
      "grad_norm": 2.100572323411406,
      "learning_rate": 3.0484954081233877e-06,
      "loss": 0.2045,
      "step": 7992
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.9142958062653297,
      "learning_rate": 3.046303981732376e-06,
      "loss": 0.1606,
      "step": 7993
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.9421062725537819,
      "learning_rate": 3.0441132017449305e-06,
      "loss": 0.1552,
      "step": 7994
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.971313683149759,
      "learning_rate": 3.0419230683647018e-06,
      "loss": 0.1677,
      "step": 7995
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.8638766681332375,
      "learning_rate": 3.03973358179528e-06,
      "loss": 0.1591,
      "step": 7996
    },
    {
      "epoch": 2.26,
      "grad_norm": 2.5032950669750527,
      "learning_rate": 3.0375447422401982e-06,
      "loss": 0.2437,
      "step": 7997
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.9190659584270584,
      "learning_rate": 3.0353565499029223e-06,
      "loss": 0.1664,
      "step": 7998
    },
    {
      "epoch": 2.26,
      "grad_norm": 2.1574012050491347,
      "learning_rate": 3.0331690049868733e-06,
      "loss": 0.2063,
      "step": 7999
    },
    {
      "epoch": 2.26,
      "grad_norm": 2.103519245258392,
      "learning_rate": 3.0309821076953893e-06,
      "loss": 0.2341,
      "step": 8000
    },
    {
      "epoch": 2.26,
      "grad_norm": 2.222911563384444,
      "learning_rate": 3.028795858231768e-06,
      "loss": 0.2224,
      "step": 8001
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.8653498813308256,
      "learning_rate": 3.026610256799238e-06,
      "loss": 0.1802,
      "step": 8002
    },
    {
      "epoch": 2.26,
      "grad_norm": 2.0327904229831515,
      "learning_rate": 3.0244253036009684e-06,
      "loss": 0.1725,
      "step": 8003
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.9527383860988106,
      "learning_rate": 3.022240998840068e-06,
      "loss": 0.1661,
      "step": 8004
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.9977090615324455,
      "learning_rate": 3.0200573427195877e-06,
      "loss": 0.1873,
      "step": 8005
    },
    {
      "epoch": 2.26,
      "grad_norm": 2.2760300430878617,
      "learning_rate": 3.017874335442512e-06,
      "loss": 0.2224,
      "step": 8006
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.912803775778812,
      "learning_rate": 3.0156919772117788e-06,
      "loss": 0.1813,
      "step": 8007
    },
    {
      "epoch": 2.26,
      "grad_norm": 2.0069989992980544,
      "learning_rate": 3.0135102682302477e-06,
      "loss": 0.1814,
      "step": 8008
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.7508895871538168,
      "learning_rate": 3.011329208700726e-06,
      "loss": 0.1677,
      "step": 8009
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.9259205078263386,
      "learning_rate": 3.0091487988259684e-06,
      "loss": 0.1997,
      "step": 8010
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.8918580261133338,
      "learning_rate": 3.006969038808658e-06,
      "loss": 0.1748,
      "step": 8011
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.9170793540271818,
      "learning_rate": 3.0047899288514213e-06,
      "loss": 0.1522,
      "step": 8012
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.8664740307615415,
      "learning_rate": 3.0026114691568255e-06,
      "loss": 0.1539,
      "step": 8013
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.8116605537110517,
      "learning_rate": 3.000433659927375e-06,
      "loss": 0.1536,
      "step": 8014
    },
    {
      "epoch": 2.26,
      "grad_norm": 2.3762741831242886,
      "learning_rate": 2.998256501365514e-06,
      "loss": 0.1755,
      "step": 8015
    },
    {
      "epoch": 2.26,
      "grad_norm": 2.0258795518650308,
      "learning_rate": 2.9960799936736353e-06,
      "loss": 0.1522,
      "step": 8016
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.7817840568703343,
      "learning_rate": 2.993904137054051e-06,
      "loss": 0.1475,
      "step": 8017
    },
    {
      "epoch": 2.26,
      "grad_norm": 2.472984197793089,
      "learning_rate": 2.9917289317090357e-06,
      "loss": 0.2146,
      "step": 8018
    },
    {
      "epoch": 2.26,
      "grad_norm": 2.4662979930348943,
      "learning_rate": 2.9895543778407875e-06,
      "loss": 0.2114,
      "step": 8019
    },
    {
      "epoch": 2.26,
      "grad_norm": 2.002106176485836,
      "learning_rate": 2.9873804756514513e-06,
      "loss": 0.1621,
      "step": 8020
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.5649495397605961,
      "learning_rate": 2.9852072253431073e-06,
      "loss": 0.1473,
      "step": 8021
    },
    {
      "epoch": 2.27,
      "grad_norm": 2.171043015468638,
      "learning_rate": 2.983034627117779e-06,
      "loss": 0.2041,
      "step": 8022
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.7488491041431489,
      "learning_rate": 2.9808626811774222e-06,
      "loss": 0.1278,
      "step": 8023
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.6962842562318445,
      "learning_rate": 2.9786913877239486e-06,
      "loss": 0.1921,
      "step": 8024
    },
    {
      "epoch": 2.27,
      "grad_norm": 2.015960393414949,
      "learning_rate": 2.976520746959187e-06,
      "loss": 0.187,
      "step": 8025
    },
    {
      "epoch": 2.27,
      "grad_norm": 2.373147449420495,
      "learning_rate": 2.9743507590849176e-06,
      "loss": 0.2241,
      "step": 8026
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.9943920274734035,
      "learning_rate": 2.9721814243028635e-06,
      "loss": 0.1667,
      "step": 8027
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.7742093852982563,
      "learning_rate": 2.970012742814684e-06,
      "loss": 0.1766,
      "step": 8028
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.8789534937652328,
      "learning_rate": 2.967844714821966e-06,
      "loss": 0.2086,
      "step": 8029
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.8896125709014853,
      "learning_rate": 2.965677340526254e-06,
      "loss": 0.1599,
      "step": 8030
    },
    {
      "epoch": 2.27,
      "grad_norm": 2.2725575134120026,
      "learning_rate": 2.963510620129021e-06,
      "loss": 0.2013,
      "step": 8031
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.8281463804903404,
      "learning_rate": 2.961344553831679e-06,
      "loss": 0.2031,
      "step": 8032
    },
    {
      "epoch": 2.27,
      "grad_norm": 2.3255291022893045,
      "learning_rate": 2.959179141835591e-06,
      "loss": 0.1958,
      "step": 8033
    },
    {
      "epoch": 2.27,
      "grad_norm": 2.5578107002911388,
      "learning_rate": 2.9570143843420394e-06,
      "loss": 0.2064,
      "step": 8034
    },
    {
      "epoch": 2.27,
      "grad_norm": 2.008693432521619,
      "learning_rate": 2.9548502815522573e-06,
      "loss": 0.2019,
      "step": 8035
    },
    {
      "epoch": 2.27,
      "grad_norm": 2.0920846116705656,
      "learning_rate": 2.952686833667423e-06,
      "loss": 0.2039,
      "step": 8036
    },
    {
      "epoch": 2.27,
      "grad_norm": 2.127776388658436,
      "learning_rate": 2.9505240408886417e-06,
      "loss": 0.1831,
      "step": 8037
    },
    {
      "epoch": 2.27,
      "grad_norm": 2.016063820812475,
      "learning_rate": 2.948361903416965e-06,
      "loss": 0.1831,
      "step": 8038
    },
    {
      "epoch": 2.27,
      "grad_norm": 2.016882481450657,
      "learning_rate": 2.9462004214533803e-06,
      "loss": 0.1875,
      "step": 8039
    },
    {
      "epoch": 2.27,
      "grad_norm": 2.1508354593644166,
      "learning_rate": 2.944039595198814e-06,
      "loss": 0.2068,
      "step": 8040
    },
    {
      "epoch": 2.27,
      "grad_norm": 2.103127021442027,
      "learning_rate": 2.9418794248541362e-06,
      "loss": 0.2197,
      "step": 8041
    },
    {
      "epoch": 2.27,
      "grad_norm": 2.245219258736531,
      "learning_rate": 2.9397199106201492e-06,
      "loss": 0.2241,
      "step": 8042
    },
    {
      "epoch": 2.27,
      "grad_norm": 2.345067962580227,
      "learning_rate": 2.937561052697597e-06,
      "loss": 0.2144,
      "step": 8043
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.986450520035037,
      "learning_rate": 2.935402851287168e-06,
      "loss": 0.2136,
      "step": 8044
    },
    {
      "epoch": 2.27,
      "grad_norm": 2.4784968143039805,
      "learning_rate": 2.933245306589485e-06,
      "loss": 0.2078,
      "step": 8045
    },
    {
      "epoch": 2.27,
      "grad_norm": 2.0049490373849395,
      "learning_rate": 2.9310884188051013e-06,
      "loss": 0.1934,
      "step": 8046
    },
    {
      "epoch": 2.27,
      "grad_norm": 2.401011614510166,
      "learning_rate": 2.9289321881345257e-06,
      "loss": 0.2024,
      "step": 8047
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.916686010873016,
      "learning_rate": 2.926776614778195e-06,
      "loss": 0.158,
      "step": 8048
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.7243981611780972,
      "learning_rate": 2.924621698936485e-06,
      "loss": 0.1678,
      "step": 8049
    },
    {
      "epoch": 2.27,
      "grad_norm": 2.176827682761328,
      "learning_rate": 2.9224674408097207e-06,
      "loss": 0.2161,
      "step": 8050
    },
    {
      "epoch": 2.27,
      "grad_norm": 2.1842503111177427,
      "learning_rate": 2.9203138405981514e-06,
      "loss": 0.2366,
      "step": 8051
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.7082769725117166,
      "learning_rate": 2.9181608985019705e-06,
      "loss": 0.1823,
      "step": 8052
    },
    {
      "epoch": 2.27,
      "grad_norm": 2.4173056427164146,
      "learning_rate": 2.9160086147213174e-06,
      "loss": 0.2896,
      "step": 8053
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.696985908091586,
      "learning_rate": 2.913856989456262e-06,
      "loss": 0.1694,
      "step": 8054
    },
    {
      "epoch": 2.27,
      "grad_norm": 2.120705550346593,
      "learning_rate": 2.911706022906816e-06,
      "loss": 0.2126,
      "step": 8055
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.7207238619689227,
      "learning_rate": 2.90955571527293e-06,
      "loss": 0.1553,
      "step": 8056
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.0657074669691613,
      "learning_rate": 2.907406066754492e-06,
      "loss": 0.2336,
      "step": 8057
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.2217086927026077,
      "learning_rate": 2.905257077551331e-06,
      "loss": 0.2375,
      "step": 8058
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.9880881845337872,
      "learning_rate": 2.9031087478632116e-06,
      "loss": 0.2009,
      "step": 8059
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.8767466536978132,
      "learning_rate": 2.900961077889837e-06,
      "loss": 0.1618,
      "step": 8060
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.0764208471800387,
      "learning_rate": 2.898814067830855e-06,
      "loss": 0.2148,
      "step": 8061
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.086544755639518,
      "learning_rate": 2.8966677178858506e-06,
      "loss": 0.2588,
      "step": 8062
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.156480910745955,
      "learning_rate": 2.894522028254334e-06,
      "loss": 0.2019,
      "step": 8063
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.4004171773959735,
      "learning_rate": 2.892376999135774e-06,
      "loss": 0.1875,
      "step": 8064
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.7899821824945064,
      "learning_rate": 2.890232630729567e-06,
      "loss": 0.1581,
      "step": 8065
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.6615610514066026,
      "learning_rate": 2.888088923235045e-06,
      "loss": 0.1577,
      "step": 8066
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.9970237426392345,
      "learning_rate": 2.885945876851495e-06,
      "loss": 0.1689,
      "step": 8067
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.1849905157852976,
      "learning_rate": 2.8838034917781187e-06,
      "loss": 0.2119,
      "step": 8068
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.184001287420521,
      "learning_rate": 2.8816617682140703e-06,
      "loss": 0.2393,
      "step": 8069
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.767486291314383,
      "learning_rate": 2.879520706358446e-06,
      "loss": 0.1631,
      "step": 8070
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.055642128566816,
      "learning_rate": 2.8773803064102758e-06,
      "loss": 0.2646,
      "step": 8071
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.7401877767095186,
      "learning_rate": 2.875240568568518e-06,
      "loss": 0.1514,
      "step": 8072
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.1703411817601976,
      "learning_rate": 2.873101493032089e-06,
      "loss": 0.1775,
      "step": 8073
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.9497917001330722,
      "learning_rate": 2.87096307999983e-06,
      "loss": 0.166,
      "step": 8074
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.173519550363379,
      "learning_rate": 2.868825329670524e-06,
      "loss": 0.2073,
      "step": 8075
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.234251106489173,
      "learning_rate": 2.866688242242892e-06,
      "loss": 0.2046,
      "step": 8076
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.279036188445893,
      "learning_rate": 2.8645518179155953e-06,
      "loss": 0.2568,
      "step": 8077
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.2367256293615747,
      "learning_rate": 2.8624160568872273e-06,
      "loss": 0.1705,
      "step": 8078
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.3396825594963317,
      "learning_rate": 2.860280959356336e-06,
      "loss": 0.2202,
      "step": 8079
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.374987535681457,
      "learning_rate": 2.8581465255213834e-06,
      "loss": 0.2495,
      "step": 8080
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.1472970096592396,
      "learning_rate": 2.8560127555807902e-06,
      "loss": 0.2053,
      "step": 8081
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.9882152375360092,
      "learning_rate": 2.853879649732908e-06,
      "loss": 0.1887,
      "step": 8082
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.412443386924402,
      "learning_rate": 2.8517472081760243e-06,
      "loss": 0.2085,
      "step": 8083
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.2632508122242196,
      "learning_rate": 2.849615431108368e-06,
      "loss": 0.2065,
      "step": 8084
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.2225148584328425,
      "learning_rate": 2.847484318728105e-06,
      "loss": 0.2158,
      "step": 8085
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.093872167910696,
      "learning_rate": 2.845353871233337e-06,
      "loss": 0.2141,
      "step": 8086
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.762036030054498,
      "learning_rate": 2.843224088822113e-06,
      "loss": 0.1866,
      "step": 8087
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.6429787932587714,
      "learning_rate": 2.841094971692414e-06,
      "loss": 0.1549,
      "step": 8088
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.0540364102048736,
      "learning_rate": 2.83896652004215e-06,
      "loss": 0.2209,
      "step": 8089
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.9248402989211337,
      "learning_rate": 2.836838734069187e-06,
      "loss": 0.2515,
      "step": 8090
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.599654967297613,
      "learning_rate": 2.834711613971317e-06,
      "loss": 0.131,
      "step": 8091
    },
    {
      "epoch": 2.29,
      "grad_norm": 2.1195446561304423,
      "learning_rate": 2.832585159946274e-06,
      "loss": 0.1564,
      "step": 8092
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.829820829462661,
      "learning_rate": 2.8304593721917283e-06,
      "loss": 0.1951,
      "step": 8093
    },
    {
      "epoch": 2.29,
      "grad_norm": 4.080842859839683,
      "learning_rate": 2.8283342509052915e-06,
      "loss": 0.1511,
      "step": 8094
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.820818334136093,
      "learning_rate": 2.8262097962845058e-06,
      "loss": 0.1654,
      "step": 8095
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.9310019345025038,
      "learning_rate": 2.8240860085268683e-06,
      "loss": 0.1667,
      "step": 8096
    },
    {
      "epoch": 2.29,
      "grad_norm": 2.1968783099863005,
      "learning_rate": 2.821962887829789e-06,
      "loss": 0.2134,
      "step": 8097
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.89848040797854,
      "learning_rate": 2.819840434390638e-06,
      "loss": 0.1719,
      "step": 8098
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.758934472374058,
      "learning_rate": 2.8177186484067143e-06,
      "loss": 0.1498,
      "step": 8099
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.9783599976686872,
      "learning_rate": 2.8155975300752524e-06,
      "loss": 0.1812,
      "step": 8100
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.587198298648475,
      "learning_rate": 2.8134770795934307e-06,
      "loss": 0.1565,
      "step": 8101
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.8205788653819417,
      "learning_rate": 2.811357297158361e-06,
      "loss": 0.1936,
      "step": 8102
    },
    {
      "epoch": 2.29,
      "grad_norm": 2.0773744854533356,
      "learning_rate": 2.809238182967092e-06,
      "loss": 0.2156,
      "step": 8103
    },
    {
      "epoch": 2.29,
      "grad_norm": 2.2178169418735116,
      "learning_rate": 2.807119737216619e-06,
      "loss": 0.2324,
      "step": 8104
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.7514016029497659,
      "learning_rate": 2.805001960103868e-06,
      "loss": 0.1555,
      "step": 8105
    },
    {
      "epoch": 2.29,
      "grad_norm": 2.1672581967539744,
      "learning_rate": 2.8028848518256967e-06,
      "loss": 0.1777,
      "step": 8106
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.9104939399370466,
      "learning_rate": 2.800768412578916e-06,
      "loss": 0.1847,
      "step": 8107
    },
    {
      "epoch": 2.29,
      "grad_norm": 2.1951267939978907,
      "learning_rate": 2.7986526425602623e-06,
      "loss": 0.2095,
      "step": 8108
    },
    {
      "epoch": 2.29,
      "grad_norm": 2.256708163573336,
      "learning_rate": 2.7965375419664154e-06,
      "loss": 0.1726,
      "step": 8109
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.8753829793211734,
      "learning_rate": 2.794423110993991e-06,
      "loss": 0.1624,
      "step": 8110
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.9510996100104703,
      "learning_rate": 2.7923093498395438e-06,
      "loss": 0.1941,
      "step": 8111
    },
    {
      "epoch": 2.29,
      "grad_norm": 2.4701173514900803,
      "learning_rate": 2.79019625869956e-06,
      "loss": 0.1992,
      "step": 8112
    },
    {
      "epoch": 2.29,
      "grad_norm": 2.071112222130357,
      "learning_rate": 2.7880838377704812e-06,
      "loss": 0.2148,
      "step": 8113
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.9461205160553625,
      "learning_rate": 2.7859720872486585e-06,
      "loss": 0.1875,
      "step": 8114
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.9943391607194167,
      "learning_rate": 2.7838610073304096e-06,
      "loss": 0.22,
      "step": 8115
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.8968740810379965,
      "learning_rate": 2.7817505982119708e-06,
      "loss": 0.2021,
      "step": 8116
    },
    {
      "epoch": 2.29,
      "grad_norm": 2.1349611550920775,
      "learning_rate": 2.779640860089523e-06,
      "loss": 0.1899,
      "step": 8117
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.832629699507038,
      "learning_rate": 2.777531793159185e-06,
      "loss": 0.1631,
      "step": 8118
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.9504728654362724,
      "learning_rate": 2.77542339761701e-06,
      "loss": 0.1606,
      "step": 8119
    },
    {
      "epoch": 2.29,
      "grad_norm": 2.500397720198831,
      "learning_rate": 2.7733156736589893e-06,
      "loss": 0.1873,
      "step": 8120
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.9583312338501875,
      "learning_rate": 2.7712086214810585e-06,
      "loss": 0.2017,
      "step": 8121
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.8524938630462346,
      "learning_rate": 2.769102241279087e-06,
      "loss": 0.17,
      "step": 8122
    },
    {
      "epoch": 2.29,
      "grad_norm": 2.047691248026839,
      "learning_rate": 2.7669965332488692e-06,
      "loss": 0.197,
      "step": 8123
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.9668686832974887,
      "learning_rate": 2.7648914975861573e-06,
      "loss": 0.2014,
      "step": 8124
    },
    {
      "epoch": 2.29,
      "grad_norm": 2.0026261324722148,
      "learning_rate": 2.7627871344866308e-06,
      "loss": 0.1902,
      "step": 8125
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.8501934258308577,
      "learning_rate": 2.760683444145905e-06,
      "loss": 0.1431,
      "step": 8126
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.971818152522406,
      "learning_rate": 2.7585804267595383e-06,
      "loss": 0.1753,
      "step": 8127
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.95317805418307,
      "learning_rate": 2.756478082523021e-06,
      "loss": 0.1523,
      "step": 8128
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.334489771896246,
      "learning_rate": 2.7543764116317806e-06,
      "loss": 0.23,
      "step": 8129
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.1172482617510964,
      "learning_rate": 2.7522754142811957e-06,
      "loss": 0.1831,
      "step": 8130
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.0009163786968807,
      "learning_rate": 2.7501750906665603e-06,
      "loss": 0.1736,
      "step": 8131
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.2025842852623683,
      "learning_rate": 2.748075440983118e-06,
      "loss": 0.2351,
      "step": 8132
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.655438762911239,
      "learning_rate": 2.7459764654260547e-06,
      "loss": 0.1587,
      "step": 8133
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.038730196455013,
      "learning_rate": 2.7438781641904834e-06,
      "loss": 0.1914,
      "step": 8134
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.127173159970969,
      "learning_rate": 2.74178053747146e-06,
      "loss": 0.2227,
      "step": 8135
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.031227635426293,
      "learning_rate": 2.7396835854639757e-06,
      "loss": 0.1852,
      "step": 8136
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.592233871521833,
      "learning_rate": 2.737587308362959e-06,
      "loss": 0.1534,
      "step": 8137
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.254025208008979,
      "learning_rate": 2.7354917063632735e-06,
      "loss": 0.2219,
      "step": 8138
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.9643394070549614,
      "learning_rate": 2.7333967796597317e-06,
      "loss": 0.1814,
      "step": 8139
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.3009421067061844,
      "learning_rate": 2.731302528447063e-06,
      "loss": 0.2651,
      "step": 8140
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.9255611070300271,
      "learning_rate": 2.729208952919954e-06,
      "loss": 0.1536,
      "step": 8141
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.0944516565023292,
      "learning_rate": 2.7271160532730167e-06,
      "loss": 0.2036,
      "step": 8142
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.8417014627644095,
      "learning_rate": 2.7250238297008026e-06,
      "loss": 0.1741,
      "step": 8143
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.7646646162392678,
      "learning_rate": 2.7229322823978033e-06,
      "loss": 0.1619,
      "step": 8144
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.8633922655091864,
      "learning_rate": 2.7208414115584436e-06,
      "loss": 0.1704,
      "step": 8145
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.503929924230758,
      "learning_rate": 2.7187512173770848e-06,
      "loss": 0.304,
      "step": 8146
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.8769088348332406,
      "learning_rate": 2.7166617000480367e-06,
      "loss": 0.1879,
      "step": 8147
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.9676392600441193,
      "learning_rate": 2.7145728597655286e-06,
      "loss": 0.1843,
      "step": 8148
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.152742337974171,
      "learning_rate": 2.712484696723735e-06,
      "loss": 0.1975,
      "step": 8149
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.0208829918070643,
      "learning_rate": 2.710397211116774e-06,
      "loss": 0.189,
      "step": 8150
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.113017969388576,
      "learning_rate": 2.708310403138692e-06,
      "loss": 0.1997,
      "step": 8151
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.8767036844404565,
      "learning_rate": 2.7062242729834743e-06,
      "loss": 0.1721,
      "step": 8152
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.2455780873476314,
      "learning_rate": 2.7041388208450457e-06,
      "loss": 0.1995,
      "step": 8153
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.1049687456927453,
      "learning_rate": 2.702054046917264e-06,
      "loss": 0.1975,
      "step": 8154
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.9376235518186673,
      "learning_rate": 2.699969951393925e-06,
      "loss": 0.1919,
      "step": 8155
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.8860689711711232,
      "learning_rate": 2.6978865344687697e-06,
      "loss": 0.1641,
      "step": 8156
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.4125428189201443,
      "learning_rate": 2.695803796335459e-06,
      "loss": 0.2415,
      "step": 8157
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.8754722951385665,
      "learning_rate": 2.6937217371876077e-06,
      "loss": 0.1855,
      "step": 8158
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.9403075995224046,
      "learning_rate": 2.691640357218759e-06,
      "loss": 0.1592,
      "step": 8159
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.8606789194800726,
      "learning_rate": 2.6895596566223937e-06,
      "loss": 0.1362,
      "step": 8160
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.356319578179288,
      "learning_rate": 2.687479635591931e-06,
      "loss": 0.1848,
      "step": 8161
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.0428128839207327,
      "learning_rate": 2.6854002943207245e-06,
      "loss": 0.1884,
      "step": 8162
    },
    {
      "epoch": 2.31,
      "grad_norm": 2.054820465876354,
      "learning_rate": 2.683321633002064e-06,
      "loss": 0.1873,
      "step": 8163
    },
    {
      "epoch": 2.31,
      "grad_norm": 2.317191379690964,
      "learning_rate": 2.681243651829187e-06,
      "loss": 0.2288,
      "step": 8164
    },
    {
      "epoch": 2.31,
      "grad_norm": 2.207477697337609,
      "learning_rate": 2.6791663509952504e-06,
      "loss": 0.2103,
      "step": 8165
    },
    {
      "epoch": 2.31,
      "grad_norm": 2.0160609733135475,
      "learning_rate": 2.677089730693356e-06,
      "loss": 0.184,
      "step": 8166
    },
    {
      "epoch": 2.31,
      "grad_norm": 2.696413583609775,
      "learning_rate": 2.675013791116551e-06,
      "loss": 0.2222,
      "step": 8167
    },
    {
      "epoch": 2.31,
      "grad_norm": 2.0905927459604534,
      "learning_rate": 2.672938532457807e-06,
      "loss": 0.2114,
      "step": 8168
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.9709932134778616,
      "learning_rate": 2.670863954910031e-06,
      "loss": 0.1782,
      "step": 8169
    },
    {
      "epoch": 2.31,
      "grad_norm": 2.1066084338847606,
      "learning_rate": 2.66879005866608e-06,
      "loss": 0.1877,
      "step": 8170
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.8285065495638122,
      "learning_rate": 2.6667168439187364e-06,
      "loss": 0.1583,
      "step": 8171
    },
    {
      "epoch": 2.31,
      "grad_norm": 2.0835217140469635,
      "learning_rate": 2.6646443108607234e-06,
      "loss": 0.2151,
      "step": 8172
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.7029450158120065,
      "learning_rate": 2.662572459684699e-06,
      "loss": 0.1541,
      "step": 8173
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.808340889395494,
      "learning_rate": 2.6605012905832605e-06,
      "loss": 0.1604,
      "step": 8174
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.7701488890516925,
      "learning_rate": 2.658430803748936e-06,
      "loss": 0.1646,
      "step": 8175
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.8717289635024132,
      "learning_rate": 2.656360999374201e-06,
      "loss": 0.1619,
      "step": 8176
    },
    {
      "epoch": 2.31,
      "grad_norm": 2.095842089081617,
      "learning_rate": 2.654291877651457e-06,
      "loss": 0.1991,
      "step": 8177
    },
    {
      "epoch": 2.31,
      "grad_norm": 2.0099209634167434,
      "learning_rate": 2.652223438773047e-06,
      "loss": 0.2078,
      "step": 8178
    },
    {
      "epoch": 2.31,
      "grad_norm": 2.199536338926006,
      "learning_rate": 2.6501556829312492e-06,
      "loss": 0.1775,
      "step": 8179
    },
    {
      "epoch": 2.31,
      "grad_norm": 2.1793840922608263,
      "learning_rate": 2.648088610318278e-06,
      "loss": 0.1812,
      "step": 8180
    },
    {
      "epoch": 2.31,
      "grad_norm": 2.1609925303261797,
      "learning_rate": 2.646022221126285e-06,
      "loss": 0.1609,
      "step": 8181
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.9383760884969063,
      "learning_rate": 2.6439565155473602e-06,
      "loss": 0.1782,
      "step": 8182
    },
    {
      "epoch": 2.31,
      "grad_norm": 2.2979581539499114,
      "learning_rate": 2.6418914937735228e-06,
      "loss": 0.2056,
      "step": 8183
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.9097044723051382,
      "learning_rate": 2.63982715599674e-06,
      "loss": 0.1792,
      "step": 8184
    },
    {
      "epoch": 2.31,
      "grad_norm": 2.517972331956169,
      "learning_rate": 2.637763502408909e-06,
      "loss": 0.1953,
      "step": 8185
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.939494072647246,
      "learning_rate": 2.6357005332018557e-06,
      "loss": 0.1709,
      "step": 8186
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.7236283551765668,
      "learning_rate": 2.6336382485673574e-06,
      "loss": 0.1689,
      "step": 8187
    },
    {
      "epoch": 2.31,
      "grad_norm": 2.2525197268856476,
      "learning_rate": 2.631576648697118e-06,
      "loss": 0.2253,
      "step": 8188
    },
    {
      "epoch": 2.31,
      "grad_norm": 2.1122289299627446,
      "learning_rate": 2.6295157337827827e-06,
      "loss": 0.1536,
      "step": 8189
    },
    {
      "epoch": 2.31,
      "grad_norm": 2.7492433999055015,
      "learning_rate": 2.6274555040159265e-06,
      "loss": 0.2903,
      "step": 8190
    },
    {
      "epoch": 2.31,
      "grad_norm": 2.361170906752086,
      "learning_rate": 2.625395959588067e-06,
      "loss": 0.2451,
      "step": 8191
    },
    {
      "epoch": 2.31,
      "grad_norm": 2.2915714570644923,
      "learning_rate": 2.623337100690654e-06,
      "loss": 0.2166,
      "step": 8192
    },
    {
      "epoch": 2.31,
      "grad_norm": 2.487153776320478,
      "learning_rate": 2.6212789275150796e-06,
      "loss": 0.2114,
      "step": 8193
    },
    {
      "epoch": 2.31,
      "grad_norm": 2.3216021548316736,
      "learning_rate": 2.6192214402526662e-06,
      "loss": 0.2065,
      "step": 8194
    },
    {
      "epoch": 2.31,
      "grad_norm": 2.508451992630998,
      "learning_rate": 2.6171646390946727e-06,
      "loss": 0.2322,
      "step": 8195
    },
    {
      "epoch": 2.31,
      "grad_norm": 2.395291435931027,
      "learning_rate": 2.6151085242322973e-06,
      "loss": 0.1897,
      "step": 8196
    },
    {
      "epoch": 2.31,
      "grad_norm": 2.1989628772564256,
      "learning_rate": 2.613053095856671e-06,
      "loss": 0.197,
      "step": 8197
    },
    {
      "epoch": 2.32,
      "grad_norm": 2.009657760359571,
      "learning_rate": 2.6109983541588655e-06,
      "loss": 0.1912,
      "step": 8198
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.8986249268057478,
      "learning_rate": 2.6089442993298854e-06,
      "loss": 0.1588,
      "step": 8199
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.9681257692426934,
      "learning_rate": 2.606890931560667e-06,
      "loss": 0.167,
      "step": 8200
    },
    {
      "epoch": 2.32,
      "grad_norm": 2.378763041102129,
      "learning_rate": 2.6048382510420954e-06,
      "loss": 0.2241,
      "step": 8201
    },
    {
      "epoch": 2.32,
      "grad_norm": 2.683082010472782,
      "learning_rate": 2.6027862579649856e-06,
      "loss": 0.2251,
      "step": 8202
    },
    {
      "epoch": 2.32,
      "grad_norm": 2.1650068342334188,
      "learning_rate": 2.6007349525200754e-06,
      "loss": 0.2095,
      "step": 8203
    },
    {
      "epoch": 2.32,
      "grad_norm": 2.101966890832883,
      "learning_rate": 2.5986843348980607e-06,
      "loss": 0.2004,
      "step": 8204
    },
    {
      "epoch": 2.32,
      "grad_norm": 2.2236521344760347,
      "learning_rate": 2.596634405289562e-06,
      "loss": 0.24,
      "step": 8205
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.8276472960125656,
      "learning_rate": 2.594585163885135e-06,
      "loss": 0.1787,
      "step": 8206
    },
    {
      "epoch": 2.32,
      "grad_norm": 2.2787849292549387,
      "learning_rate": 2.592536610875275e-06,
      "loss": 0.1692,
      "step": 8207
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.976890991481842,
      "learning_rate": 2.5904887464504115e-06,
      "loss": 0.1807,
      "step": 8208
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.8315686935625823,
      "learning_rate": 2.588441570800907e-06,
      "loss": 0.1554,
      "step": 8209
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.8488710289639008,
      "learning_rate": 2.5863950841170704e-06,
      "loss": 0.1672,
      "step": 8210
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.8784445683975275,
      "learning_rate": 2.5843492865891383e-06,
      "loss": 0.147,
      "step": 8211
    },
    {
      "epoch": 2.32,
      "grad_norm": 2.530326421250463,
      "learning_rate": 2.5823041784072767e-06,
      "loss": 0.2822,
      "step": 8212
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.9804215763574395,
      "learning_rate": 2.580259759761604e-06,
      "loss": 0.165,
      "step": 8213
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.8782527258493589,
      "learning_rate": 2.578216030842162e-06,
      "loss": 0.1753,
      "step": 8214
    },
    {
      "epoch": 2.32,
      "grad_norm": 2.281322640724523,
      "learning_rate": 2.576172991838933e-06,
      "loss": 0.2334,
      "step": 8215
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.8644510266255836,
      "learning_rate": 2.5741306429418355e-06,
      "loss": 0.1595,
      "step": 8216
    },
    {
      "epoch": 2.32,
      "grad_norm": 2.136596912630292,
      "learning_rate": 2.5720889843407205e-06,
      "loss": 0.2178,
      "step": 8217
    },
    {
      "epoch": 2.32,
      "grad_norm": 2.065625129055097,
      "learning_rate": 2.5700480162253748e-06,
      "loss": 0.2107,
      "step": 8218
    },
    {
      "epoch": 2.32,
      "grad_norm": 2.454337952772471,
      "learning_rate": 2.568007738785533e-06,
      "loss": 0.2031,
      "step": 8219
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.8676299712291453,
      "learning_rate": 2.5659681522108428e-06,
      "loss": 0.1777,
      "step": 8220
    },
    {
      "epoch": 2.32,
      "grad_norm": 2.2026024843929117,
      "learning_rate": 2.5639292566909103e-06,
      "loss": 0.1812,
      "step": 8221
    },
    {
      "epoch": 2.32,
      "grad_norm": 2.119470469866211,
      "learning_rate": 2.5618910524152652e-06,
      "loss": 0.1912,
      "step": 8222
    },
    {
      "epoch": 2.32,
      "grad_norm": 2.0384192710454045,
      "learning_rate": 2.5598535395733735e-06,
      "loss": 0.191,
      "step": 8223
    },
    {
      "epoch": 2.32,
      "grad_norm": 2.2206658440123848,
      "learning_rate": 2.557816718354641e-06,
      "loss": 0.2083,
      "step": 8224
    },
    {
      "epoch": 2.32,
      "grad_norm": 2.081974672253329,
      "learning_rate": 2.5557805889484055e-06,
      "loss": 0.2277,
      "step": 8225
    },
    {
      "epoch": 2.32,
      "grad_norm": 3.1709653438131764,
      "learning_rate": 2.5537451515439406e-06,
      "loss": 0.2273,
      "step": 8226
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.7419529097704394,
      "learning_rate": 2.551710406330462e-06,
      "loss": 0.1383,
      "step": 8227
    },
    {
      "epoch": 2.32,
      "grad_norm": 2.2966724261956837,
      "learning_rate": 2.549676353497116e-06,
      "loss": 0.2732,
      "step": 8228
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.8345068786596839,
      "learning_rate": 2.547642993232976e-06,
      "loss": 0.1907,
      "step": 8229
    },
    {
      "epoch": 2.32,
      "grad_norm": 2.216788730610408,
      "learning_rate": 2.5456103257270693e-06,
      "loss": 0.1526,
      "step": 8230
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.9806714235413885,
      "learning_rate": 2.5435783511683444e-06,
      "loss": 0.2083,
      "step": 8231
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.7455388582551927,
      "learning_rate": 2.5415470697456923e-06,
      "loss": 0.1732,
      "step": 8232
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.8443513013713617,
      "learning_rate": 2.5395164816479357e-06,
      "loss": 0.1509,
      "step": 8233
    },
    {
      "epoch": 2.33,
      "grad_norm": 2.23645407063221,
      "learning_rate": 2.5374865870638354e-06,
      "loss": 0.1899,
      "step": 8234
    },
    {
      "epoch": 2.33,
      "grad_norm": 2.301027421059833,
      "learning_rate": 2.535457386182083e-06,
      "loss": 0.1851,
      "step": 8235
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.9624018981900513,
      "learning_rate": 2.533428879191321e-06,
      "loss": 0.1868,
      "step": 8236
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.8325791420255462,
      "learning_rate": 2.5314010662801e-06,
      "loss": 0.1514,
      "step": 8237
    },
    {
      "epoch": 2.33,
      "grad_norm": 2.086333631883438,
      "learning_rate": 2.529373947636934e-06,
      "loss": 0.2101,
      "step": 8238
    },
    {
      "epoch": 2.33,
      "grad_norm": 2.2793379273217855,
      "learning_rate": 2.5273475234502565e-06,
      "loss": 0.1812,
      "step": 8239
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.832415707081909,
      "learning_rate": 2.5253217939084407e-06,
      "loss": 0.1523,
      "step": 8240
    },
    {
      "epoch": 2.33,
      "grad_norm": 2.0953035140222758,
      "learning_rate": 2.5232967591997946e-06,
      "loss": 0.2001,
      "step": 8241
    },
    {
      "epoch": 2.33,
      "grad_norm": 2.312835291894682,
      "learning_rate": 2.5212724195125616e-06,
      "loss": 0.2419,
      "step": 8242
    },
    {
      "epoch": 2.33,
      "grad_norm": 2.304733195598346,
      "learning_rate": 2.519248775034918e-06,
      "loss": 0.2161,
      "step": 8243
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.8265375789201477,
      "learning_rate": 2.5172258259549854e-06,
      "loss": 0.1677,
      "step": 8244
    },
    {
      "epoch": 2.33,
      "grad_norm": 2.273316890669747,
      "learning_rate": 2.5152035724608117e-06,
      "loss": 0.2222,
      "step": 8245
    },
    {
      "epoch": 2.33,
      "grad_norm": 2.2194721851915475,
      "learning_rate": 2.513182014740375e-06,
      "loss": 0.2067,
      "step": 8246
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.8426626136461883,
      "learning_rate": 2.511161152981604e-06,
      "loss": 0.1716,
      "step": 8247
    },
    {
      "epoch": 2.33,
      "grad_norm": 2.3369783060587492,
      "learning_rate": 2.5091409873723506e-06,
      "loss": 0.2256,
      "step": 8248
    },
    {
      "epoch": 2.33,
      "grad_norm": 2.248735289995527,
      "learning_rate": 2.5071215181004073e-06,
      "loss": 0.2156,
      "step": 8249
    },
    {
      "epoch": 2.33,
      "grad_norm": 2.0146090319368666,
      "learning_rate": 2.505102745353499e-06,
      "loss": 0.1951,
      "step": 8250
    },
    {
      "epoch": 2.33,
      "grad_norm": 2.3795968087347954,
      "learning_rate": 2.503084669319289e-06,
      "loss": 0.2156,
      "step": 8251
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.8307652900640539,
      "learning_rate": 2.5010672901853704e-06,
      "loss": 0.1655,
      "step": 8252
    },
    {
      "epoch": 2.33,
      "grad_norm": 2.166382237438904,
      "learning_rate": 2.499050608139284e-06,
      "loss": 0.1926,
      "step": 8253
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.709745220001688,
      "learning_rate": 2.4970346233684863e-06,
      "loss": 0.1584,
      "step": 8254
    },
    {
      "epoch": 2.33,
      "grad_norm": 2.201968637186567,
      "learning_rate": 2.4950193360603868e-06,
      "loss": 0.1843,
      "step": 8255
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.8466491235400895,
      "learning_rate": 2.493004746402322e-06,
      "loss": 0.1573,
      "step": 8256
    },
    {
      "epoch": 2.33,
      "grad_norm": 2.1315648282973454,
      "learning_rate": 2.490990854581563e-06,
      "loss": 0.189,
      "step": 8257
    },
    {
      "epoch": 2.33,
      "grad_norm": 2.435226057677687,
      "learning_rate": 2.488977660785319e-06,
      "loss": 0.2074,
      "step": 8258
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.8029125342582673,
      "learning_rate": 2.486965165200733e-06,
      "loss": 0.1643,
      "step": 8259
    },
    {
      "epoch": 2.33,
      "grad_norm": 2.0489302823206073,
      "learning_rate": 2.4849533680148787e-06,
      "loss": 0.1689,
      "step": 8260
    },
    {
      "epoch": 2.33,
      "grad_norm": 2.248901578751564,
      "learning_rate": 2.4829422694147796e-06,
      "loss": 0.2227,
      "step": 8261
    },
    {
      "epoch": 2.33,
      "grad_norm": 2.053716045979034,
      "learning_rate": 2.480931869587375e-06,
      "loss": 0.1997,
      "step": 8262
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.765513221980394,
      "learning_rate": 2.4789221687195473e-06,
      "loss": 0.157,
      "step": 8263
    },
    {
      "epoch": 2.33,
      "grad_norm": 2.1587436429663507,
      "learning_rate": 2.4769131669981217e-06,
      "loss": 0.1772,
      "step": 8264
    },
    {
      "epoch": 2.33,
      "grad_norm": 2.0871305591181804,
      "learning_rate": 2.4749048646098486e-06,
      "loss": 0.2039,
      "step": 8265
    },
    {
      "epoch": 2.33,
      "grad_norm": 2.2248830690030412,
      "learning_rate": 2.472897261741415e-06,
      "loss": 0.2002,
      "step": 8266
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.7673489675419072,
      "learning_rate": 2.470890358579444e-06,
      "loss": 0.1288,
      "step": 8267
    },
    {
      "epoch": 2.33,
      "grad_norm": 2.1340712756022477,
      "learning_rate": 2.468884155310497e-06,
      "loss": 0.2227,
      "step": 8268
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.0984765279379687,
      "learning_rate": 2.466878652121061e-06,
      "loss": 0.1714,
      "step": 8269
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.303009288377355,
      "learning_rate": 2.4648738491975745e-06,
      "loss": 0.2241,
      "step": 8270
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.2440470582687486,
      "learning_rate": 2.4628697467263916e-06,
      "loss": 0.1589,
      "step": 8271
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.0516368059818224,
      "learning_rate": 2.46086634489381e-06,
      "loss": 0.2236,
      "step": 8272
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.1323555396812224,
      "learning_rate": 2.458863643886067e-06,
      "loss": 0.2117,
      "step": 8273
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.1046555381524934,
      "learning_rate": 2.4568616438893287e-06,
      "loss": 0.218,
      "step": 8274
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.1554621363195503,
      "learning_rate": 2.454860345089698e-06,
      "loss": 0.2219,
      "step": 8275
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.192295783485569,
      "learning_rate": 2.452859747673212e-06,
      "loss": 0.1765,
      "step": 8276
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.7569135586063895,
      "learning_rate": 2.450859851825842e-06,
      "loss": 0.1726,
      "step": 8277
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.8075448320977738,
      "learning_rate": 2.448860657733495e-06,
      "loss": 0.15,
      "step": 8278
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.9421891937966094,
      "learning_rate": 2.4468621655820125e-06,
      "loss": 0.2227,
      "step": 8279
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.0833293628744265,
      "learning_rate": 2.4448643755571687e-06,
      "loss": 0.1912,
      "step": 8280
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.4085804410301477,
      "learning_rate": 2.4428672878446803e-06,
      "loss": 0.218,
      "step": 8281
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.7273717573931449,
      "learning_rate": 2.4408709026301903e-06,
      "loss": 0.1492,
      "step": 8282
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.910089247917832,
      "learning_rate": 2.438875220099278e-06,
      "loss": 0.1626,
      "step": 8283
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.3473698418211297,
      "learning_rate": 2.43688024043746e-06,
      "loss": 0.2583,
      "step": 8284
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.1736031097445845,
      "learning_rate": 2.4348859638301857e-06,
      "loss": 0.1888,
      "step": 8285
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.763774487614614,
      "learning_rate": 2.4328923904628376e-06,
      "loss": 0.1757,
      "step": 8286
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.9414795262404063,
      "learning_rate": 2.430899520520741e-06,
      "loss": 0.1782,
      "step": 8287
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.058761181239184,
      "learning_rate": 2.428907354189144e-06,
      "loss": 0.1646,
      "step": 8288
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.9705170738061326,
      "learning_rate": 2.4269158916532332e-06,
      "loss": 0.1699,
      "step": 8289
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.418105571855881,
      "learning_rate": 2.424925133098137e-06,
      "loss": 0.272,
      "step": 8290
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.7750384952195584,
      "learning_rate": 2.422935078708911e-06,
      "loss": 0.14,
      "step": 8291
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.9417637739467184,
      "learning_rate": 2.4209457286705475e-06,
      "loss": 0.175,
      "step": 8292
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.1914385148304976,
      "learning_rate": 2.418957083167972e-06,
      "loss": 0.1848,
      "step": 8293
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.0081167933221136,
      "learning_rate": 2.4169691423860454e-06,
      "loss": 0.1875,
      "step": 8294
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.8704167863156747,
      "learning_rate": 2.414981906509565e-06,
      "loss": 0.1553,
      "step": 8295
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.210958779869398,
      "learning_rate": 2.4129953757232584e-06,
      "loss": 0.196,
      "step": 8296
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.5074414075121956,
      "learning_rate": 2.411009550211789e-06,
      "loss": 0.2153,
      "step": 8297
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.4560003677340254,
      "learning_rate": 2.409024430159761e-06,
      "loss": 0.2351,
      "step": 8298
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.7312460097867253,
      "learning_rate": 2.4070400157517036e-06,
      "loss": 0.1643,
      "step": 8299
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.9052846723624883,
      "learning_rate": 2.4050563071720867e-06,
      "loss": 0.1892,
      "step": 8300
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.3597125451868344,
      "learning_rate": 2.403073304605311e-06,
      "loss": 0.2454,
      "step": 8301
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.1804183682826004,
      "learning_rate": 2.401091008235714e-06,
      "loss": 0.175,
      "step": 8302
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.111991054316142,
      "learning_rate": 2.399109418247563e-06,
      "loss": 0.1742,
      "step": 8303
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.9722953980119622,
      "learning_rate": 2.3971285348250705e-06,
      "loss": 0.2085,
      "step": 8304
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.009538790817952,
      "learning_rate": 2.3951483581523694e-06,
      "loss": 0.1779,
      "step": 8305
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.0889233157208196,
      "learning_rate": 2.393168888413533e-06,
      "loss": 0.183,
      "step": 8306
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.3544718174224557,
      "learning_rate": 2.3911901257925752e-06,
      "loss": 0.2112,
      "step": 8307
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.198288790190701,
      "learning_rate": 2.389212070473438e-06,
      "loss": 0.1733,
      "step": 8308
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.9306273799839542,
      "learning_rate": 2.3872347226399895e-06,
      "loss": 0.1541,
      "step": 8309
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.7450441052785455,
      "learning_rate": 2.3852580824760487e-06,
      "loss": 0.1372,
      "step": 8310
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.2241712903803723,
      "learning_rate": 2.383282150165358e-06,
      "loss": 0.2031,
      "step": 8311
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.0955979048735185,
      "learning_rate": 2.381306925891598e-06,
      "loss": 0.2063,
      "step": 8312
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.1145071446978547,
      "learning_rate": 2.3793324098383796e-06,
      "loss": 0.1799,
      "step": 8313
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.1443941357955323,
      "learning_rate": 2.3773586021892524e-06,
      "loss": 0.2329,
      "step": 8314
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.9577277224010838,
      "learning_rate": 2.375385503127695e-06,
      "loss": 0.1694,
      "step": 8315
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.1364598176787246,
      "learning_rate": 2.3734131128371283e-06,
      "loss": 0.2068,
      "step": 8316
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.2507532232215786,
      "learning_rate": 2.3714414315009005e-06,
      "loss": 0.1945,
      "step": 8317
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.2241036083459806,
      "learning_rate": 2.3694704593022942e-06,
      "loss": 0.2148,
      "step": 8318
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.348607287335995,
      "learning_rate": 2.367500196424529e-06,
      "loss": 0.2107,
      "step": 8319
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.8630997333767494,
      "learning_rate": 2.3655306430507563e-06,
      "loss": 0.1675,
      "step": 8320
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.116485339531672,
      "learning_rate": 2.3635617993640625e-06,
      "loss": 0.1692,
      "step": 8321
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.161433407464144,
      "learning_rate": 2.3615936655474693e-06,
      "loss": 0.1865,
      "step": 8322
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.2705451234069267,
      "learning_rate": 2.3596262417839256e-06,
      "loss": 0.21,
      "step": 8323
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.8067066181895033,
      "learning_rate": 2.357659528256326e-06,
      "loss": 0.1548,
      "step": 8324
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.2280373667354914,
      "learning_rate": 2.3556935251474946e-06,
      "loss": 0.1907,
      "step": 8325
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.528079156116986,
      "learning_rate": 2.3537282326401777e-06,
      "loss": 0.2139,
      "step": 8326
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.2084278306921226,
      "learning_rate": 2.351763650917074e-06,
      "loss": 0.2017,
      "step": 8327
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.3979163137213697,
      "learning_rate": 2.3497997801608054e-06,
      "loss": 0.2161,
      "step": 8328
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.182515700082025,
      "learning_rate": 2.3478366205539295e-06,
      "loss": 0.2224,
      "step": 8329
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.7841175084622674,
      "learning_rate": 2.345874172278939e-06,
      "loss": 0.1509,
      "step": 8330
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.3740246134933276,
      "learning_rate": 2.3439124355182584e-06,
      "loss": 0.2463,
      "step": 8331
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.9302353823617455,
      "learning_rate": 2.341951410454245e-06,
      "loss": 0.1661,
      "step": 8332
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.8446031366754359,
      "learning_rate": 2.3399910972692017e-06,
      "loss": 0.1646,
      "step": 8333
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.970282734216698,
      "learning_rate": 2.338031496145343e-06,
      "loss": 0.1512,
      "step": 8334
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.0801951333311473,
      "learning_rate": 2.33607260726484e-06,
      "loss": 0.1595,
      "step": 8335
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.4805142379276806,
      "learning_rate": 2.334114430809784e-06,
      "loss": 0.1738,
      "step": 8336
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.944421867245332,
      "learning_rate": 2.3321569669622034e-06,
      "loss": 0.1777,
      "step": 8337
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.1120093796344523,
      "learning_rate": 2.3302002159040613e-06,
      "loss": 0.1483,
      "step": 8338
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.4594037349285047,
      "learning_rate": 2.328244177817254e-06,
      "loss": 0.2734,
      "step": 8339
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.3149928634354158,
      "learning_rate": 2.326288852883607e-06,
      "loss": 0.1948,
      "step": 8340
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.0910761035484247,
      "learning_rate": 2.3243342412848923e-06,
      "loss": 0.1831,
      "step": 8341
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.2641284491319795,
      "learning_rate": 2.3223803432028046e-06,
      "loss": 0.1807,
      "step": 8342
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.7972122238885926,
      "learning_rate": 2.3204271588189685e-06,
      "loss": 0.1566,
      "step": 8343
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.100303218605271,
      "learning_rate": 2.3184746883149556e-06,
      "loss": 0.1992,
      "step": 8344
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.218768278203565,
      "learning_rate": 2.3165229318722627e-06,
      "loss": 0.1936,
      "step": 8345
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.8970780669868865,
      "learning_rate": 2.3145718896723204e-06,
      "loss": 0.1833,
      "step": 8346
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.092168160005882,
      "learning_rate": 2.312621561896494e-06,
      "loss": 0.1846,
      "step": 8347
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.192985396231806,
      "learning_rate": 2.3106719487260843e-06,
      "loss": 0.209,
      "step": 8348
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.5389863403092896,
      "learning_rate": 2.3087230503423207e-06,
      "loss": 0.2202,
      "step": 8349
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.869385670396692,
      "learning_rate": 2.306774866926377e-06,
      "loss": 0.1682,
      "step": 8350
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.54314564316564,
      "learning_rate": 2.304827398659342e-06,
      "loss": 0.2427,
      "step": 8351
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.1575815306926622,
      "learning_rate": 2.3028806457222565e-06,
      "loss": 0.2128,
      "step": 8352
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.2414265120115164,
      "learning_rate": 2.300934608296086e-06,
      "loss": 0.2534,
      "step": 8353
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.004584502235058,
      "learning_rate": 2.2989892865617313e-06,
      "loss": 0.1736,
      "step": 8354
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.7850760905893233,
      "learning_rate": 2.2970446807000237e-06,
      "loss": 0.1764,
      "step": 8355
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.0762012263279352,
      "learning_rate": 2.2951007908917334e-06,
      "loss": 0.1704,
      "step": 8356
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.9867412528027173,
      "learning_rate": 2.293157617317555e-06,
      "loss": 0.2078,
      "step": 8357
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.323291562549131,
      "learning_rate": 2.291215160158131e-06,
      "loss": 0.2571,
      "step": 8358
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.4506561060460053,
      "learning_rate": 2.289273419594027e-06,
      "loss": 0.2002,
      "step": 8359
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.866841707689372,
      "learning_rate": 2.287332395805737e-06,
      "loss": 0.1687,
      "step": 8360
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.192575716636076,
      "learning_rate": 2.2853920889737015e-06,
      "loss": 0.187,
      "step": 8361
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.0943102606689497,
      "learning_rate": 2.2834524992782867e-06,
      "loss": 0.2109,
      "step": 8362
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.9500677282742058,
      "learning_rate": 2.281513626899794e-06,
      "loss": 0.1843,
      "step": 8363
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.877894993725501,
      "learning_rate": 2.279575472018457e-06,
      "loss": 0.181,
      "step": 8364
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.184817798581257,
      "learning_rate": 2.2776380348144435e-06,
      "loss": 0.1987,
      "step": 8365
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.269247995021958,
      "learning_rate": 2.2757013154678496e-06,
      "loss": 0.2593,
      "step": 8366
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.2214956077891,
      "learning_rate": 2.2737653141587203e-06,
      "loss": 0.2747,
      "step": 8367
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.7793759564558713,
      "learning_rate": 2.2718300310670136e-06,
      "loss": 0.1687,
      "step": 8368
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.9356412940269732,
      "learning_rate": 2.26989546637263e-06,
      "loss": 0.1592,
      "step": 8369
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.9151219224963885,
      "learning_rate": 2.26796162025541e-06,
      "loss": 0.1602,
      "step": 8370
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.7971699292730232,
      "learning_rate": 2.2660284928951148e-06,
      "loss": 0.1399,
      "step": 8371
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.0837806022255525,
      "learning_rate": 2.2640960844714478e-06,
      "loss": 0.1755,
      "step": 8372
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.290324570424654,
      "learning_rate": 2.2621643951640413e-06,
      "loss": 0.2302,
      "step": 8373
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.848710825504877,
      "learning_rate": 2.2602334251524617e-06,
      "loss": 0.1554,
      "step": 8374
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.9049503168868758,
      "learning_rate": 2.258303174616204e-06,
      "loss": 0.1904,
      "step": 8375
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.7944100891978054,
      "learning_rate": 2.256373643734713e-06,
      "loss": 0.147,
      "step": 8376
    },
    {
      "epoch": 2.37,
      "grad_norm": 2.2185688857297605,
      "learning_rate": 2.25444483268734e-06,
      "loss": 0.2086,
      "step": 8377
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.9211438569065222,
      "learning_rate": 2.2525167416533945e-06,
      "loss": 0.1772,
      "step": 8378
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.9364410286049458,
      "learning_rate": 2.250589370812105e-06,
      "loss": 0.1702,
      "step": 8379
    },
    {
      "epoch": 2.37,
      "grad_norm": 2.1703243689848275,
      "learning_rate": 2.248662720342637e-06,
      "loss": 0.2559,
      "step": 8380
    },
    {
      "epoch": 2.37,
      "grad_norm": 2.0177561676329856,
      "learning_rate": 2.2467367904240877e-06,
      "loss": 0.1632,
      "step": 8381
    },
    {
      "epoch": 2.37,
      "grad_norm": 2.2732413836766354,
      "learning_rate": 2.2448115812354888e-06,
      "loss": 0.2183,
      "step": 8382
    },
    {
      "epoch": 2.37,
      "grad_norm": 2.171572801872392,
      "learning_rate": 2.2428870929558012e-06,
      "loss": 0.2178,
      "step": 8383
    },
    {
      "epoch": 2.37,
      "grad_norm": 2.106920824544671,
      "learning_rate": 2.2409633257639306e-06,
      "loss": 0.1755,
      "step": 8384
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.75119650309562,
      "learning_rate": 2.239040279838699e-06,
      "loss": 0.1597,
      "step": 8385
    },
    {
      "epoch": 2.37,
      "grad_norm": 2.105663342859142,
      "learning_rate": 2.2371179553588685e-06,
      "loss": 0.186,
      "step": 8386
    },
    {
      "epoch": 2.37,
      "grad_norm": 2.1851894854308482,
      "learning_rate": 2.2351963525031428e-06,
      "loss": 0.2202,
      "step": 8387
    },
    {
      "epoch": 2.37,
      "grad_norm": 2.052216953532893,
      "learning_rate": 2.233275471450146e-06,
      "loss": 0.1659,
      "step": 8388
    },
    {
      "epoch": 2.37,
      "grad_norm": 2.1723569562705114,
      "learning_rate": 2.231355312378439e-06,
      "loss": 0.1975,
      "step": 8389
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.8013306652918586,
      "learning_rate": 2.229435875466519e-06,
      "loss": 0.1664,
      "step": 8390
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.9865316254609233,
      "learning_rate": 2.2275171608928124e-06,
      "loss": 0.2,
      "step": 8391
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.9296604534671247,
      "learning_rate": 2.225599168835677e-06,
      "loss": 0.1777,
      "step": 8392
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.8334043852550554,
      "learning_rate": 2.2236818994734144e-06,
      "loss": 0.167,
      "step": 8393
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.7234234608528645,
      "learning_rate": 2.221765352984239e-06,
      "loss": 0.1511,
      "step": 8394
    },
    {
      "epoch": 2.37,
      "grad_norm": 2.3427329510296944,
      "learning_rate": 2.2198495295463184e-06,
      "loss": 0.2456,
      "step": 8395
    },
    {
      "epoch": 2.37,
      "grad_norm": 2.153262410271014,
      "learning_rate": 2.2179344293377415e-06,
      "loss": 0.2185,
      "step": 8396
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.7932742590955753,
      "learning_rate": 2.2160200525365326e-06,
      "loss": 0.1641,
      "step": 8397
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.7682298117804478,
      "learning_rate": 2.214106399320648e-06,
      "loss": 0.1951,
      "step": 8398
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.8811789972745225,
      "learning_rate": 2.2121934698679793e-06,
      "loss": 0.1731,
      "step": 8399
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.989729404926689,
      "learning_rate": 2.2102812643563455e-06,
      "loss": 0.1655,
      "step": 8400
    },
    {
      "epoch": 2.37,
      "grad_norm": 2.3339760116598085,
      "learning_rate": 2.20836978296351e-06,
      "loss": 0.2534,
      "step": 8401
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.8478788258552634,
      "learning_rate": 2.2064590258671524e-06,
      "loss": 0.1882,
      "step": 8402
    },
    {
      "epoch": 2.37,
      "grad_norm": 2.0376931439698867,
      "learning_rate": 2.2045489932448937e-06,
      "loss": 0.1743,
      "step": 8403
    },
    {
      "epoch": 2.37,
      "grad_norm": 2.2713919366082855,
      "learning_rate": 2.2026396852742936e-06,
      "loss": 0.2407,
      "step": 8404
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.8749359555792355,
      "learning_rate": 2.2007311021328337e-06,
      "loss": 0.1912,
      "step": 8405
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.984855019993052,
      "learning_rate": 2.198823243997933e-06,
      "loss": 0.1863,
      "step": 8406
    },
    {
      "epoch": 2.37,
      "grad_norm": 2.132981472062537,
      "learning_rate": 2.196916111046944e-06,
      "loss": 0.2314,
      "step": 8407
    },
    {
      "epoch": 2.37,
      "grad_norm": 2.482659044690007,
      "learning_rate": 2.1950097034571495e-06,
      "loss": 0.2402,
      "step": 8408
    },
    {
      "epoch": 2.37,
      "grad_norm": 2.3181221812162875,
      "learning_rate": 2.193104021405762e-06,
      "loss": 0.2013,
      "step": 8409
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.1663798669349035,
      "learning_rate": 2.191199065069941e-06,
      "loss": 0.2207,
      "step": 8410
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.9472117096226453,
      "learning_rate": 2.1892948346267583e-06,
      "loss": 0.1631,
      "step": 8411
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.8697046555214611,
      "learning_rate": 2.187391330253228e-06,
      "loss": 0.1804,
      "step": 8412
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.5462151318016946,
      "learning_rate": 2.1854885521263016e-06,
      "loss": 0.2075,
      "step": 8413
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.3277756054831764,
      "learning_rate": 2.183586500422856e-06,
      "loss": 0.2327,
      "step": 8414
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.1464037158847447,
      "learning_rate": 2.1816851753197023e-06,
      "loss": 0.189,
      "step": 8415
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.309515063237885,
      "learning_rate": 2.1797845769935856e-06,
      "loss": 0.2056,
      "step": 8416
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.012766900536498,
      "learning_rate": 2.1778847056211803e-06,
      "loss": 0.2205,
      "step": 8417
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.9967373945460043,
      "learning_rate": 2.175985561379097e-06,
      "loss": 0.207,
      "step": 8418
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.345839768221673,
      "learning_rate": 2.174087144443875e-06,
      "loss": 0.2219,
      "step": 8419
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.259839098763812,
      "learning_rate": 2.1721894549919863e-06,
      "loss": 0.1825,
      "step": 8420
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.228061638663215,
      "learning_rate": 2.1702924931998426e-06,
      "loss": 0.2151,
      "step": 8421
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.8619896977169985,
      "learning_rate": 2.168396259243781e-06,
      "loss": 0.1841,
      "step": 8422
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.0207540604236907,
      "learning_rate": 2.166500753300065e-06,
      "loss": 0.1803,
      "step": 8423
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.0260672288191883,
      "learning_rate": 2.1646059755449058e-06,
      "loss": 0.1699,
      "step": 8424
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.9361575260754336,
      "learning_rate": 2.1627119261544348e-06,
      "loss": 0.1832,
      "step": 8425
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.985421740787267,
      "learning_rate": 2.1608186053047177e-06,
      "loss": 0.1743,
      "step": 8426
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.2626172831460742,
      "learning_rate": 2.1589260131717614e-06,
      "loss": 0.2634,
      "step": 8427
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.030755317815769,
      "learning_rate": 2.157034149931492e-06,
      "loss": 0.2085,
      "step": 8428
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.111798125005657,
      "learning_rate": 2.155143015759773e-06,
      "loss": 0.1907,
      "step": 8429
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.1043613083542945,
      "learning_rate": 2.1532526108324047e-06,
      "loss": 0.209,
      "step": 8430
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.0432212602822024,
      "learning_rate": 2.151362935325115e-06,
      "loss": 0.1956,
      "step": 8431
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.9754478309936068,
      "learning_rate": 2.149473989413564e-06,
      "loss": 0.2109,
      "step": 8432
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.9086745002680319,
      "learning_rate": 2.1475857732733464e-06,
      "loss": 0.1912,
      "step": 8433
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.0505943143382397,
      "learning_rate": 2.1456982870799848e-06,
      "loss": 0.1685,
      "step": 8434
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.8583326003935794,
      "learning_rate": 2.1438115310089393e-06,
      "loss": 0.1672,
      "step": 8435
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.380715243488299,
      "learning_rate": 2.1419255052355983e-06,
      "loss": 0.2349,
      "step": 8436
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.015188220940018,
      "learning_rate": 2.14004020993528e-06,
      "loss": 0.1968,
      "step": 8437
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.7228940450741876,
      "learning_rate": 2.138155645283244e-06,
      "loss": 0.144,
      "step": 8438
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.908944893195969,
      "learning_rate": 2.1362718114546777e-06,
      "loss": 0.1826,
      "step": 8439
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.0647133971882288,
      "learning_rate": 2.1343887086246893e-06,
      "loss": 0.2126,
      "step": 8440
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.1652152235631155,
      "learning_rate": 2.1325063369683374e-06,
      "loss": 0.1989,
      "step": 8441
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.9546887143004492,
      "learning_rate": 2.1306246966606025e-06,
      "loss": 0.1519,
      "step": 8442
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.0497212516216816,
      "learning_rate": 2.128743787876393e-06,
      "loss": 0.1965,
      "step": 8443
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.317236260614886,
      "learning_rate": 2.126863610790566e-06,
      "loss": 0.2302,
      "step": 8444
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.060034103122643,
      "learning_rate": 2.124984165577889e-06,
      "loss": 0.2006,
      "step": 8445
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.7360493210387453,
      "learning_rate": 2.1231054524130746e-06,
      "loss": 0.1797,
      "step": 8446
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.508557966401493,
      "learning_rate": 2.121227471470768e-06,
      "loss": 0.2325,
      "step": 8447
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.202400082273728,
      "learning_rate": 2.1193502229255436e-06,
      "loss": 0.2009,
      "step": 8448
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.8665408231202942,
      "learning_rate": 2.1174737069519e-06,
      "loss": 0.1522,
      "step": 8449
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.9993779911272673,
      "learning_rate": 2.1155979237242817e-06,
      "loss": 0.1702,
      "step": 8450
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.4213041430219557,
      "learning_rate": 2.1137228734170558e-06,
      "loss": 0.2314,
      "step": 8451
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.7378613943867527,
      "learning_rate": 2.1118485562045244e-06,
      "loss": 0.1593,
      "step": 8452
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.8344200303758191,
      "learning_rate": 2.109974972260921e-06,
      "loss": 0.2014,
      "step": 8453
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.0770233478500324,
      "learning_rate": 2.108102121760409e-06,
      "loss": 0.1802,
      "step": 8454
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.2362045413131324,
      "learning_rate": 2.1062300048770847e-06,
      "loss": 0.2063,
      "step": 8455
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.9512385221098831,
      "learning_rate": 2.104358621784983e-06,
      "loss": 0.2117,
      "step": 8456
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.6155782991289107,
      "learning_rate": 2.102487972658056e-06,
      "loss": 0.2549,
      "step": 8457
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.6702414696343115,
      "learning_rate": 2.1006180576702017e-06,
      "loss": 0.1198,
      "step": 8458
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.313045789037182,
      "learning_rate": 2.0987488769952436e-06,
      "loss": 0.1997,
      "step": 8459
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.343358250367896,
      "learning_rate": 2.0968804308069324e-06,
      "loss": 0.1953,
      "step": 8460
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.190146185708426,
      "learning_rate": 2.095012719278966e-06,
      "loss": 0.1938,
      "step": 8461
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.964941898590686,
      "learning_rate": 2.0931457425849555e-06,
      "loss": 0.1592,
      "step": 8462
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.0074471693779947,
      "learning_rate": 2.09127950089845e-06,
      "loss": 0.1711,
      "step": 8463
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.123828153275395,
      "learning_rate": 2.089413994392938e-06,
      "loss": 0.2136,
      "step": 8464
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.0242652861846766,
      "learning_rate": 2.087549223241835e-06,
      "loss": 0.1891,
      "step": 8465
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.961535730623286,
      "learning_rate": 2.085685187618478e-06,
      "loss": 0.1976,
      "step": 8466
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.152586237265174,
      "learning_rate": 2.0838218876961524e-06,
      "loss": 0.1782,
      "step": 8467
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.7607812142335084,
      "learning_rate": 2.081959323648065e-06,
      "loss": 0.1521,
      "step": 8468
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.1911507380970114,
      "learning_rate": 2.0800974956473565e-06,
      "loss": 0.2053,
      "step": 8469
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.101340183045589,
      "learning_rate": 2.0782364038670986e-06,
      "loss": 0.176,
      "step": 8470
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.2888703286782897,
      "learning_rate": 2.0763760484802966e-06,
      "loss": 0.177,
      "step": 8471
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.023022005149693,
      "learning_rate": 2.074516429659882e-06,
      "loss": 0.1599,
      "step": 8472
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.048552329527157,
      "learning_rate": 2.0726575475787293e-06,
      "loss": 0.1963,
      "step": 8473
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.3610897472555976,
      "learning_rate": 2.070799402409628e-06,
      "loss": 0.1836,
      "step": 8474
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.1543763465556514,
      "learning_rate": 2.068941994325315e-06,
      "loss": 0.2271,
      "step": 8475
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.036891258077894,
      "learning_rate": 2.06708532349845e-06,
      "loss": 0.1958,
      "step": 8476
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.1250399989596755,
      "learning_rate": 2.065229390101624e-06,
      "loss": 0.175,
      "step": 8477
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.194816486396495,
      "learning_rate": 2.063374194307364e-06,
      "loss": 0.1931,
      "step": 8478
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.8767134603198496,
      "learning_rate": 2.0615197362881234e-06,
      "loss": 0.1707,
      "step": 8479
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.3342995962613307,
      "learning_rate": 2.0596660162162872e-06,
      "loss": 0.2212,
      "step": 8480
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.0046063386733888,
      "learning_rate": 2.057813034264181e-06,
      "loss": 0.1726,
      "step": 8481
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.002255236958309,
      "learning_rate": 2.0559607906040524e-06,
      "loss": 0.187,
      "step": 8482
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.023008638266615,
      "learning_rate": 2.054109285408077e-06,
      "loss": 0.1748,
      "step": 8483
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.964161803284419,
      "learning_rate": 2.0522585188483745e-06,
      "loss": 0.1902,
      "step": 8484
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.9972836526947897,
      "learning_rate": 2.0504084910969856e-06,
      "loss": 0.1863,
      "step": 8485
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.222903876707252,
      "learning_rate": 2.0485592023258872e-06,
      "loss": 0.2075,
      "step": 8486
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.3273685526228136,
      "learning_rate": 2.046710652706985e-06,
      "loss": 0.2336,
      "step": 8487
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.521625376121161,
      "learning_rate": 2.0448628424121173e-06,
      "loss": 0.2251,
      "step": 8488
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.9339759386384976,
      "learning_rate": 2.0430157716130506e-06,
      "loss": 0.1897,
      "step": 8489
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.9953744257915014,
      "learning_rate": 2.041169440481493e-06,
      "loss": 0.1752,
      "step": 8490
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.8963999649190337,
      "learning_rate": 2.0393238491890655e-06,
      "loss": 0.1799,
      "step": 8491
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.9848118880160024,
      "learning_rate": 2.0374789979073408e-06,
      "loss": 0.1995,
      "step": 8492
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.0257916063309414,
      "learning_rate": 2.035634886807808e-06,
      "loss": 0.1832,
      "step": 8493
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.000640452490954,
      "learning_rate": 2.0337915160618936e-06,
      "loss": 0.1821,
      "step": 8494
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.2592091951568967,
      "learning_rate": 2.0319488858409552e-06,
      "loss": 0.2461,
      "step": 8495
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.997707477104376,
      "learning_rate": 2.030106996316279e-06,
      "loss": 0.1885,
      "step": 8496
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.9250807933670968,
      "learning_rate": 2.0282658476590823e-06,
      "loss": 0.2042,
      "step": 8497
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.885841929899634,
      "learning_rate": 2.0264254400405192e-06,
      "loss": 0.1631,
      "step": 8498
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.100473902071926,
      "learning_rate": 2.024585773631671e-06,
      "loss": 0.1768,
      "step": 8499
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.314171155796096,
      "learning_rate": 2.022746848603543e-06,
      "loss": 0.2427,
      "step": 8500
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.6691343005422326,
      "learning_rate": 2.020908665127086e-06,
      "loss": 0.17,
      "step": 8501
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.121091611621433,
      "learning_rate": 2.019071223373171e-06,
      "loss": 0.1709,
      "step": 8502
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.098615914076585,
      "learning_rate": 2.0172345235126043e-06,
      "loss": 0.1987,
      "step": 8503
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.6100595208693869,
      "learning_rate": 2.0153985657161223e-06,
      "loss": 0.1521,
      "step": 8504
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.9005044407525178,
      "learning_rate": 2.0135633501543916e-06,
      "loss": 0.1882,
      "step": 8505
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.163215842944791,
      "learning_rate": 2.0117288769980092e-06,
      "loss": 0.2273,
      "step": 8506
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.1954146315154834,
      "learning_rate": 2.009895146417512e-06,
      "loss": 0.2656,
      "step": 8507
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.1107179398266926,
      "learning_rate": 2.0080621585833516e-06,
      "loss": 0.1588,
      "step": 8508
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.937906327255436,
      "learning_rate": 2.0062299136659203e-06,
      "loss": 0.1711,
      "step": 8509
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.256372167034367,
      "learning_rate": 2.0043984118355464e-06,
      "loss": 0.2256,
      "step": 8510
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.3569701656505146,
      "learning_rate": 2.0025676532624794e-06,
      "loss": 0.2004,
      "step": 8511
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.293915656557336,
      "learning_rate": 2.000737638116903e-06,
      "loss": 0.1907,
      "step": 8512
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.9723708247261476,
      "learning_rate": 1.9989083665689335e-06,
      "loss": 0.1635,
      "step": 8513
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.8192974638562045,
      "learning_rate": 1.997079838788616e-06,
      "loss": 0.1482,
      "step": 8514
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.2434041707596313,
      "learning_rate": 1.9952520549459254e-06,
      "loss": 0.2,
      "step": 8515
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.117564653930923,
      "learning_rate": 1.993425015210777e-06,
      "loss": 0.1956,
      "step": 8516
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.1957059791192624,
      "learning_rate": 1.9915987197529973e-06,
      "loss": 0.2336,
      "step": 8517
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.408638336941353,
      "learning_rate": 1.989773168742366e-06,
      "loss": 0.2393,
      "step": 8518
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.1088597155037587,
      "learning_rate": 1.9879483623485786e-06,
      "loss": 0.2075,
      "step": 8519
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.250170120451138,
      "learning_rate": 1.986124300741267e-06,
      "loss": 0.2434,
      "step": 8520
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.256269495858848,
      "learning_rate": 1.9843009840899917e-06,
      "loss": 0.2551,
      "step": 8521
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.2384633468854105,
      "learning_rate": 1.982478412564246e-06,
      "loss": 0.1912,
      "step": 8522
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.3544946705641965,
      "learning_rate": 1.980656586333449e-06,
      "loss": 0.2019,
      "step": 8523
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.2460055968225014,
      "learning_rate": 1.978835505566965e-06,
      "loss": 0.1837,
      "step": 8524
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.1515099582582007,
      "learning_rate": 1.977015170434069e-06,
      "loss": 0.199,
      "step": 8525
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.2900674724280408,
      "learning_rate": 1.975195581103976e-06,
      "loss": 0.2068,
      "step": 8526
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.2133595871665848,
      "learning_rate": 1.9733767377458377e-06,
      "loss": 0.2344,
      "step": 8527
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.7768026164897903,
      "learning_rate": 1.971558640528728e-06,
      "loss": 0.1648,
      "step": 8528
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.0590370942567846,
      "learning_rate": 1.969741289621653e-06,
      "loss": 0.1914,
      "step": 8529
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.5703561425678845,
      "learning_rate": 1.967924685193552e-06,
      "loss": 0.2854,
      "step": 8530
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.1855750866382158,
      "learning_rate": 1.9661088274132924e-06,
      "loss": 0.2473,
      "step": 8531
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.9988281589265666,
      "learning_rate": 1.9642937164496712e-06,
      "loss": 0.161,
      "step": 8532
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.665003254962688,
      "learning_rate": 1.962479352471426e-06,
      "loss": 0.1752,
      "step": 8533
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.407644757223966,
      "learning_rate": 1.960665735647206e-06,
      "loss": 0.2201,
      "step": 8534
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.9042791126663485,
      "learning_rate": 1.9588528661456087e-06,
      "loss": 0.157,
      "step": 8535
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.034296346545941,
      "learning_rate": 1.9570407441351548e-06,
      "loss": 0.1697,
      "step": 8536
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.0882775501091686,
      "learning_rate": 1.955229369784295e-06,
      "loss": 0.238,
      "step": 8537
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.222162334417886,
      "learning_rate": 1.9534187432614114e-06,
      "loss": 0.2416,
      "step": 8538
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.0742157469146467,
      "learning_rate": 1.9516088647348164e-06,
      "loss": 0.2437,
      "step": 8539
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.754019255514677,
      "learning_rate": 1.9497997343727513e-06,
      "loss": 0.1936,
      "step": 8540
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.9680633254802227,
      "learning_rate": 1.947991352343398e-06,
      "loss": 0.1733,
      "step": 8541
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.990762987107702,
      "learning_rate": 1.9461837188148513e-06,
      "loss": 0.2002,
      "step": 8542
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.8405146816338183,
      "learning_rate": 1.944376833955147e-06,
      "loss": 0.166,
      "step": 8543
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.101194471392853,
      "learning_rate": 1.9425706979322544e-06,
      "loss": 0.1716,
      "step": 8544
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.378568399422728,
      "learning_rate": 1.940765310914069e-06,
      "loss": 0.2166,
      "step": 8545
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.0779397080373623,
      "learning_rate": 1.9389606730684084e-06,
      "loss": 0.1818,
      "step": 8546
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.0968591271000085,
      "learning_rate": 1.937156784563037e-06,
      "loss": 0.1959,
      "step": 8547
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.1280071772520723,
      "learning_rate": 1.9353536455656385e-06,
      "loss": 0.1854,
      "step": 8548
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.353648423555922,
      "learning_rate": 1.9335512562438263e-06,
      "loss": 0.2671,
      "step": 8549
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.712647839346249,
      "learning_rate": 1.9317496167651563e-06,
      "loss": 0.1393,
      "step": 8550
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.0409515287212243,
      "learning_rate": 1.929948727297096e-06,
      "loss": 0.1965,
      "step": 8551
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.7794397503729136,
      "learning_rate": 1.928148588007055e-06,
      "loss": 0.175,
      "step": 8552
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.9999743650928177,
      "learning_rate": 1.9263491990623763e-06,
      "loss": 0.1891,
      "step": 8553
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.3561913282105946,
      "learning_rate": 1.9245505606303238e-06,
      "loss": 0.2317,
      "step": 8554
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.001533823609447,
      "learning_rate": 1.9227526728780978e-06,
      "loss": 0.2119,
      "step": 8555
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.292699792190749,
      "learning_rate": 1.920955535972825e-06,
      "loss": 0.2178,
      "step": 8556
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.4006529112804764,
      "learning_rate": 1.9191591500815642e-06,
      "loss": 0.2278,
      "step": 8557
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.96478222238123,
      "learning_rate": 1.9173635153713066e-06,
      "loss": 0.1777,
      "step": 8558
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.4045817222488153,
      "learning_rate": 1.9155686320089684e-06,
      "loss": 0.2922,
      "step": 8559
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.1652660966322785,
      "learning_rate": 1.9137745001613984e-06,
      "loss": 0.1968,
      "step": 8560
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.041736804095437,
      "learning_rate": 1.911981119995381e-06,
      "loss": 0.1919,
      "step": 8561
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.117702604833684,
      "learning_rate": 1.9101884916776236e-06,
      "loss": 0.1609,
      "step": 8562
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.148809746479407,
      "learning_rate": 1.9083966153747603e-06,
      "loss": 0.1931,
      "step": 8563
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.1495939162092217,
      "learning_rate": 1.9066054912533683e-06,
      "loss": 0.2271,
      "step": 8564
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.9060787484031927,
      "learning_rate": 1.9048151194799435e-06,
      "loss": 0.1543,
      "step": 8565
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.1432128684121925,
      "learning_rate": 1.9030255002209153e-06,
      "loss": 0.2241,
      "step": 8566
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.425793329404496,
      "learning_rate": 1.901236633642649e-06,
      "loss": 0.22,
      "step": 8567
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.7036760587224022,
      "learning_rate": 1.8994485199114287e-06,
      "loss": 0.151,
      "step": 8568
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.051288487820651,
      "learning_rate": 1.8976611591934734e-06,
      "loss": 0.1711,
      "step": 8569
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.8255391462152624,
      "learning_rate": 1.8958745516549382e-06,
      "loss": 0.1729,
      "step": 8570
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.0906847937256923,
      "learning_rate": 1.8940886974619e-06,
      "loss": 0.1561,
      "step": 8571
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.7944869899960239,
      "learning_rate": 1.8923035967803704e-06,
      "loss": 0.1628,
      "step": 8572
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.319163225708441,
      "learning_rate": 1.8905192497762881e-06,
      "loss": 0.2087,
      "step": 8573
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.332048301847129,
      "learning_rate": 1.8887356566155236e-06,
      "loss": 0.2246,
      "step": 8574
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.9558126541397787,
      "learning_rate": 1.8869528174638752e-06,
      "loss": 0.1547,
      "step": 8575
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.9605853582661488,
      "learning_rate": 1.885170732487074e-06,
      "loss": 0.1941,
      "step": 8576
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.494054990042046,
      "learning_rate": 1.8833894018507758e-06,
      "loss": 0.2356,
      "step": 8577
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.2764761747448903,
      "learning_rate": 1.8816088257205767e-06,
      "loss": 0.2285,
      "step": 8578
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.0499083471715234,
      "learning_rate": 1.8798290042619949e-06,
      "loss": 0.2148,
      "step": 8579
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.7190660526814057,
      "learning_rate": 1.8780499376404715e-06,
      "loss": 0.1444,
      "step": 8580
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.9995042203758175,
      "learning_rate": 1.8762716260213943e-06,
      "loss": 0.2109,
      "step": 8581
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.9068311414317138,
      "learning_rate": 1.8744940695700686e-06,
      "loss": 0.1763,
      "step": 8582
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.101690172418446,
      "learning_rate": 1.8727172684517325e-06,
      "loss": 0.2025,
      "step": 8583
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.4204097705088694,
      "learning_rate": 1.870941222831556e-06,
      "loss": 0.2148,
      "step": 8584
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.120219746678706,
      "learning_rate": 1.869165932874636e-06,
      "loss": 0.1873,
      "step": 8585
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.2175864473815863,
      "learning_rate": 1.867391398745999e-06,
      "loss": 0.1853,
      "step": 8586
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.2655700984206852,
      "learning_rate": 1.8656176206106059e-06,
      "loss": 0.2073,
      "step": 8587
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.7485835908619907,
      "learning_rate": 1.863844598633343e-06,
      "loss": 0.1505,
      "step": 8588
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.9048852182706564,
      "learning_rate": 1.8620723329790269e-06,
      "loss": 0.1936,
      "step": 8589
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.091294997167436,
      "learning_rate": 1.8603008238124043e-06,
      "loss": 0.2012,
      "step": 8590
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.921079915120468,
      "learning_rate": 1.8585300712981514e-06,
      "loss": 0.1903,
      "step": 8591
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.42290033097323,
      "learning_rate": 1.8567600756008753e-06,
      "loss": 0.228,
      "step": 8592
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.3108270117139993,
      "learning_rate": 1.8549908368851099e-06,
      "loss": 0.2163,
      "step": 8593
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.113767310122784,
      "learning_rate": 1.8532223553153194e-06,
      "loss": 0.1948,
      "step": 8594
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.8262405760254348,
      "learning_rate": 1.8514546310559044e-06,
      "loss": 0.165,
      "step": 8595
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.9832550471142836,
      "learning_rate": 1.8496876642711882e-06,
      "loss": 0.1951,
      "step": 8596
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.9427711422165312,
      "learning_rate": 1.8479214551254176e-06,
      "loss": 0.1689,
      "step": 8597
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.095728716985854,
      "learning_rate": 1.8461560037827842e-06,
      "loss": 0.1794,
      "step": 8598
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.187206536394345,
      "learning_rate": 1.8443913104073984e-06,
      "loss": 0.2283,
      "step": 8599
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.065840878577499,
      "learning_rate": 1.842627375163305e-06,
      "loss": 0.1633,
      "step": 8600
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.9436934734572615,
      "learning_rate": 1.8408641982144738e-06,
      "loss": 0.1665,
      "step": 8601
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.7926648552753326,
      "learning_rate": 1.8391017797248079e-06,
      "loss": 0.1697,
      "step": 8602
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.8350177458012233,
      "learning_rate": 1.8373401198581353e-06,
      "loss": 0.1912,
      "step": 8603
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.6159033439247006,
      "learning_rate": 1.8355792187782228e-06,
      "loss": 0.1543,
      "step": 8604
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.14302179640534,
      "learning_rate": 1.8338190766487606e-06,
      "loss": 0.1951,
      "step": 8605
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.1225120190039077,
      "learning_rate": 1.8320596936333613e-06,
      "loss": 0.1921,
      "step": 8606
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.242236607955735,
      "learning_rate": 1.8303010698955803e-06,
      "loss": 0.2195,
      "step": 8607
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.8468607058900464,
      "learning_rate": 1.828543205598895e-06,
      "loss": 0.1758,
      "step": 8608
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.9530328825752261,
      "learning_rate": 1.8267861009067124e-06,
      "loss": 0.1616,
      "step": 8609
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.313749360167037,
      "learning_rate": 1.8250297559823716e-06,
      "loss": 0.207,
      "step": 8610
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.1462529358341205,
      "learning_rate": 1.8232741709891376e-06,
      "loss": 0.187,
      "step": 8611
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.2374170774612376,
      "learning_rate": 1.8215193460902047e-06,
      "loss": 0.1924,
      "step": 8612
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.802903574920175,
      "learning_rate": 1.8197652814487054e-06,
      "loss": 0.2501,
      "step": 8613
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.229903298852395,
      "learning_rate": 1.818011977227686e-06,
      "loss": 0.2561,
      "step": 8614
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.8281401934340953,
      "learning_rate": 1.8162594335901363e-06,
      "loss": 0.1414,
      "step": 8615
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.009137774338367,
      "learning_rate": 1.814507650698969e-06,
      "loss": 0.1672,
      "step": 8616
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.4206687560945244,
      "learning_rate": 1.812756628717025e-06,
      "loss": 0.2307,
      "step": 8617
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.8765163741887045,
      "learning_rate": 1.8110063678070778e-06,
      "loss": 0.1494,
      "step": 8618
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.9750652736915981,
      "learning_rate": 1.809256868131828e-06,
      "loss": 0.2043,
      "step": 8619
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.4410467869199586,
      "learning_rate": 1.8075081298539032e-06,
      "loss": 0.2683,
      "step": 8620
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.9573125815734709,
      "learning_rate": 1.8057601531358693e-06,
      "loss": 0.1599,
      "step": 8621
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.984370947396166,
      "learning_rate": 1.8040129381402137e-06,
      "loss": 0.1644,
      "step": 8622
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.7649317778799576,
      "learning_rate": 1.802266485029347e-06,
      "loss": 0.172,
      "step": 8623
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.7148986711349359,
      "learning_rate": 1.8005207939656255e-06,
      "loss": 0.1434,
      "step": 8624
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.4263319271211437,
      "learning_rate": 1.7987758651113218e-06,
      "loss": 0.1636,
      "step": 8625
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.1878121093569085,
      "learning_rate": 1.797031698628643e-06,
      "loss": 0.241,
      "step": 8626
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.9886662449814494,
      "learning_rate": 1.7952882946797212e-06,
      "loss": 0.1855,
      "step": 8627
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.2774420954822694,
      "learning_rate": 1.7935456534266227e-06,
      "loss": 0.1926,
      "step": 8628
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.1927370059276154,
      "learning_rate": 1.7918037750313366e-06,
      "loss": 0.2137,
      "step": 8629
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.4345714864093884,
      "learning_rate": 1.7900626596557924e-06,
      "loss": 0.1707,
      "step": 8630
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.7862787937006441,
      "learning_rate": 1.7883223074618316e-06,
      "loss": 0.1384,
      "step": 8631
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.2234260070344245,
      "learning_rate": 1.7865827186112429e-06,
      "loss": 0.2051,
      "step": 8632
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.269329841138603,
      "learning_rate": 1.7848438932657309e-06,
      "loss": 0.208,
      "step": 8633
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.2364058899151726,
      "learning_rate": 1.7831058315869343e-06,
      "loss": 0.2119,
      "step": 8634
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.4855548906144684,
      "learning_rate": 1.7813685337364205e-06,
      "loss": 0.2302,
      "step": 8635
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.06549951689499,
      "learning_rate": 1.7796319998756872e-06,
      "loss": 0.1897,
      "step": 8636
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.1402857418736296,
      "learning_rate": 1.777896230166155e-06,
      "loss": 0.1819,
      "step": 8637
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.980931314850875,
      "learning_rate": 1.776161224769185e-06,
      "loss": 0.2258,
      "step": 8638
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.9943333273338875,
      "learning_rate": 1.774426983846058e-06,
      "loss": 0.1921,
      "step": 8639
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.8978414890616873,
      "learning_rate": 1.7726935075579798e-06,
      "loss": 0.1843,
      "step": 8640
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.7758370315212781,
      "learning_rate": 1.770960796066099e-06,
      "loss": 0.1824,
      "step": 8641
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.2507505678318163,
      "learning_rate": 1.7692288495314836e-06,
      "loss": 0.2371,
      "step": 8642
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.7122604082347126,
      "learning_rate": 1.7674976681151302e-06,
      "loss": 0.1405,
      "step": 8643
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.3081284058288434,
      "learning_rate": 1.765767251977969e-06,
      "loss": 0.2322,
      "step": 8644
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.2358805390473444,
      "learning_rate": 1.7640376012808536e-06,
      "loss": 0.2368,
      "step": 8645
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.027939336795965,
      "learning_rate": 1.7623087161845699e-06,
      "loss": 0.1958,
      "step": 8646
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.3600455477007625,
      "learning_rate": 1.760580596849838e-06,
      "loss": 0.2179,
      "step": 8647
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.8958631740234508,
      "learning_rate": 1.758853243437293e-06,
      "loss": 0.1929,
      "step": 8648
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.7518060871995218,
      "learning_rate": 1.7571266561075073e-06,
      "loss": 0.1306,
      "step": 8649
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.2480411890072136,
      "learning_rate": 1.7554008350209862e-06,
      "loss": 0.2192,
      "step": 8650
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.9977793779717203,
      "learning_rate": 1.7536757803381576e-06,
      "loss": 0.175,
      "step": 8651
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.4118945235335714,
      "learning_rate": 1.751951492219378e-06,
      "loss": 0.2603,
      "step": 8652
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.3155510511226076,
      "learning_rate": 1.750227970824936e-06,
      "loss": 0.2102,
      "step": 8653
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.460479819566498,
      "learning_rate": 1.7485052163150452e-06,
      "loss": 0.2153,
      "step": 8654
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.331687631940217,
      "learning_rate": 1.746783228849851e-06,
      "loss": 0.2097,
      "step": 8655
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.672215918399645,
      "learning_rate": 1.7450620085894255e-06,
      "loss": 0.1653,
      "step": 8656
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.215909620469304,
      "learning_rate": 1.7433415556937693e-06,
      "loss": 0.1941,
      "step": 8657
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.2578620874536823,
      "learning_rate": 1.741621870322817e-06,
      "loss": 0.2104,
      "step": 8658
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.9941418306108099,
      "learning_rate": 1.7399029526364254e-06,
      "loss": 0.2275,
      "step": 8659
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.840403003301056,
      "learning_rate": 1.7381848027943815e-06,
      "loss": 0.1256,
      "step": 8660
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.216835007506854,
      "learning_rate": 1.7364674209564025e-06,
      "loss": 0.2366,
      "step": 8661
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.1553124962324732,
      "learning_rate": 1.7347508072821317e-06,
      "loss": 0.1934,
      "step": 8662
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.075937011207465,
      "learning_rate": 1.7330349619311415e-06,
      "loss": 0.1609,
      "step": 8663
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.8667428622060758,
      "learning_rate": 1.731319885062941e-06,
      "loss": 0.1628,
      "step": 8664
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.2093096058317507,
      "learning_rate": 1.7296055768369524e-06,
      "loss": 0.1873,
      "step": 8665
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.9662501460194304,
      "learning_rate": 1.7278920374125362e-06,
      "loss": 0.1899,
      "step": 8666
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.2610908161796206,
      "learning_rate": 1.726179266948984e-06,
      "loss": 0.2312,
      "step": 8667
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.923057499670177,
      "learning_rate": 1.7244672656055105e-06,
      "loss": 0.1312,
      "step": 8668
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.8933227609879824,
      "learning_rate": 1.7227560335412597e-06,
      "loss": 0.1813,
      "step": 8669
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.928891206952228,
      "learning_rate": 1.721045570915304e-06,
      "loss": 0.1707,
      "step": 8670
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.842976444973187,
      "learning_rate": 1.7193358778866464e-06,
      "loss": 0.1707,
      "step": 8671
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.478751715408715,
      "learning_rate": 1.7176269546142166e-06,
      "loss": 0.2219,
      "step": 8672
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.9526077356226281,
      "learning_rate": 1.715918801256874e-06,
      "loss": 0.1692,
      "step": 8673
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.8167932952875605,
      "learning_rate": 1.7142114179734004e-06,
      "loss": 0.1564,
      "step": 8674
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.2384125616185915,
      "learning_rate": 1.712504804922519e-06,
      "loss": 0.2607,
      "step": 8675
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.034125856243393,
      "learning_rate": 1.7107989622628706e-06,
      "loss": 0.1584,
      "step": 8676
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.299518794127182,
      "learning_rate": 1.7090938901530264e-06,
      "loss": 0.1785,
      "step": 8677
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.247399752981874,
      "learning_rate": 1.7073895887514869e-06,
      "loss": 0.1782,
      "step": 8678
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.954582453043549,
      "learning_rate": 1.7056860582166823e-06,
      "loss": 0.1771,
      "step": 8679
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.4025658957786193,
      "learning_rate": 1.703983298706966e-06,
      "loss": 0.176,
      "step": 8680
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.0561833366202626,
      "learning_rate": 1.7022813103806324e-06,
      "loss": 0.1758,
      "step": 8681
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.565318706862935,
      "learning_rate": 1.700580093395886e-06,
      "loss": 0.1677,
      "step": 8682
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.6304800637901775,
      "learning_rate": 1.6988796479108716e-06,
      "loss": 0.291,
      "step": 8683
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.898784120431102,
      "learning_rate": 1.6971799740836625e-06,
      "loss": 0.1824,
      "step": 8684
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.271267191016911,
      "learning_rate": 1.6954810720722581e-06,
      "loss": 0.1903,
      "step": 8685
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.240794745546972,
      "learning_rate": 1.6937829420345775e-06,
      "loss": 0.2131,
      "step": 8686
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.8824434175333016,
      "learning_rate": 1.6920855841284844e-06,
      "loss": 0.1777,
      "step": 8687
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.726254400529733,
      "learning_rate": 1.6903889985117594e-06,
      "loss": 0.1367,
      "step": 8688
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.29850924414045,
      "learning_rate": 1.688693185342114e-06,
      "loss": 0.2375,
      "step": 8689
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.09659171107697,
      "learning_rate": 1.6869981447771876e-06,
      "loss": 0.1621,
      "step": 8690
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.076115681869422,
      "learning_rate": 1.6853038769745466e-06,
      "loss": 0.199,
      "step": 8691
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.8544950221338643,
      "learning_rate": 1.6836103820916926e-06,
      "loss": 0.1351,
      "step": 8692
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.0500235672879623,
      "learning_rate": 1.6819176602860454e-06,
      "loss": 0.2061,
      "step": 8693
    },
    {
      "epoch": 2.46,
      "grad_norm": 2.068370767064867,
      "learning_rate": 1.68022571171496e-06,
      "loss": 0.2144,
      "step": 8694
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.8156606249302292,
      "learning_rate": 1.6785345365357153e-06,
      "loss": 0.1411,
      "step": 8695
    },
    {
      "epoch": 2.46,
      "grad_norm": 2.2156979050893018,
      "learning_rate": 1.6768441349055197e-06,
      "loss": 0.2395,
      "step": 8696
    },
    {
      "epoch": 2.46,
      "grad_norm": 2.2175209907712263,
      "learning_rate": 1.675154506981509e-06,
      "loss": 0.2144,
      "step": 8697
    },
    {
      "epoch": 2.46,
      "grad_norm": 2.0004204894719604,
      "learning_rate": 1.673465652920755e-06,
      "loss": 0.1736,
      "step": 8698
    },
    {
      "epoch": 2.46,
      "grad_norm": 2.162665193186894,
      "learning_rate": 1.6717775728802432e-06,
      "loss": 0.1912,
      "step": 8699
    },
    {
      "epoch": 2.46,
      "grad_norm": 2.217186330118761,
      "learning_rate": 1.670090267016895e-06,
      "loss": 0.2036,
      "step": 8700
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.815222310314905,
      "learning_rate": 1.6684037354875626e-06,
      "loss": 0.1575,
      "step": 8701
    },
    {
      "epoch": 2.46,
      "grad_norm": 2.182873886603968,
      "learning_rate": 1.6667179784490251e-06,
      "loss": 0.1777,
      "step": 8702
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.9090551955514063,
      "learning_rate": 1.6650329960579792e-06,
      "loss": 0.1907,
      "step": 8703
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.9271264247696072,
      "learning_rate": 1.6633487884710663e-06,
      "loss": 0.1736,
      "step": 8704
    },
    {
      "epoch": 2.46,
      "grad_norm": 4.747413906625156,
      "learning_rate": 1.6616653558448437e-06,
      "loss": 0.1819,
      "step": 8705
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.731258982055223,
      "learning_rate": 1.6599826983358002e-06,
      "loss": 0.1503,
      "step": 8706
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.6672323707578598,
      "learning_rate": 1.6583008161003544e-06,
      "loss": 0.1489,
      "step": 8707
    },
    {
      "epoch": 2.46,
      "grad_norm": 2.173302503300288,
      "learning_rate": 1.65661970929485e-06,
      "loss": 0.1963,
      "step": 8708
    },
    {
      "epoch": 2.46,
      "grad_norm": 2.54342445402906,
      "learning_rate": 1.6549393780755574e-06,
      "loss": 0.2195,
      "step": 8709
    },
    {
      "epoch": 2.46,
      "grad_norm": 2.0075569037056824,
      "learning_rate": 1.653259822598683e-06,
      "loss": 0.2229,
      "step": 8710
    },
    {
      "epoch": 2.46,
      "grad_norm": 2.297138330712357,
      "learning_rate": 1.6515810430203516e-06,
      "loss": 0.2334,
      "step": 8711
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.9171739734180036,
      "learning_rate": 1.649903039496621e-06,
      "loss": 0.1982,
      "step": 8712
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.917704738215456,
      "learning_rate": 1.6482258121834737e-06,
      "loss": 0.1736,
      "step": 8713
    },
    {
      "epoch": 2.46,
      "grad_norm": 2.0384839268262835,
      "learning_rate": 1.6465493612368233e-06,
      "loss": 0.2148,
      "step": 8714
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.7503438196732246,
      "learning_rate": 1.6448736868125093e-06,
      "loss": 0.1412,
      "step": 8715
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.970563584117771,
      "learning_rate": 1.6431987890663004e-06,
      "loss": 0.1837,
      "step": 8716
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.792701249103225,
      "learning_rate": 1.6415246681538887e-06,
      "loss": 0.1768,
      "step": 8717
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.9668405274631071,
      "learning_rate": 1.6398513242309022e-06,
      "loss": 0.1521,
      "step": 8718
    },
    {
      "epoch": 2.46,
      "grad_norm": 2.399082272030279,
      "learning_rate": 1.638178757452894e-06,
      "loss": 0.2017,
      "step": 8719
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.9807434636820636,
      "learning_rate": 1.6365069679753331e-06,
      "loss": 0.1615,
      "step": 8720
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.8123515896213027,
      "learning_rate": 1.6348359559536353e-06,
      "loss": 0.1819,
      "step": 8721
    },
    {
      "epoch": 2.46,
      "grad_norm": 2.081330055465298,
      "learning_rate": 1.6331657215431319e-06,
      "loss": 0.1549,
      "step": 8722
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.8023792132690848,
      "learning_rate": 1.631496264899085e-06,
      "loss": 0.156,
      "step": 8723
    },
    {
      "epoch": 2.46,
      "grad_norm": 2.452361409309166,
      "learning_rate": 1.6298275861766843e-06,
      "loss": 0.2437,
      "step": 8724
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.9509367217149036,
      "learning_rate": 1.6281596855310478e-06,
      "loss": 0.1809,
      "step": 8725
    },
    {
      "epoch": 2.46,
      "grad_norm": 2.1717179579163255,
      "learning_rate": 1.626492563117217e-06,
      "loss": 0.1992,
      "step": 8726
    },
    {
      "epoch": 2.46,
      "grad_norm": 2.0711517217652915,
      "learning_rate": 1.624826219090172e-06,
      "loss": 0.1685,
      "step": 8727
    },
    {
      "epoch": 2.46,
      "grad_norm": 2.134880077581215,
      "learning_rate": 1.6231606536048083e-06,
      "loss": 0.2002,
      "step": 8728
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.4494564934922542,
      "learning_rate": 1.6214958668159552e-06,
      "loss": 0.2073,
      "step": 8729
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.428900719033579,
      "learning_rate": 1.619831858878368e-06,
      "loss": 0.2646,
      "step": 8730
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.2550570636085983,
      "learning_rate": 1.6181686299467303e-06,
      "loss": 0.2139,
      "step": 8731
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.0487415180484785,
      "learning_rate": 1.6165061801756531e-06,
      "loss": 0.1587,
      "step": 8732
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.923333659064565,
      "learning_rate": 1.614844509719674e-06,
      "loss": 0.1936,
      "step": 8733
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.5010500179112154,
      "learning_rate": 1.6131836187332584e-06,
      "loss": 0.2188,
      "step": 8734
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.9688607607031916,
      "learning_rate": 1.6115235073708024e-06,
      "loss": 0.1754,
      "step": 8735
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.190429768602841,
      "learning_rate": 1.6098641757866285e-06,
      "loss": 0.1902,
      "step": 8736
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.7783891874910218,
      "learning_rate": 1.6082056241349787e-06,
      "loss": 0.1609,
      "step": 8737
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.360777510571592,
      "learning_rate": 1.606547852570034e-06,
      "loss": 0.2379,
      "step": 8738
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.2315856435878407,
      "learning_rate": 1.604890861245898e-06,
      "loss": 0.184,
      "step": 8739
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.397834493739424,
      "learning_rate": 1.6032346503166007e-06,
      "loss": 0.2258,
      "step": 8740
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.9188160582748974,
      "learning_rate": 1.6015792199361003e-06,
      "loss": 0.1687,
      "step": 8741
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.0886704606676907,
      "learning_rate": 1.5999245702582833e-06,
      "loss": 0.1665,
      "step": 8742
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.238837824971813,
      "learning_rate": 1.5982707014369603e-06,
      "loss": 0.2186,
      "step": 8743
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.7983547518995686,
      "learning_rate": 1.5966176136258794e-06,
      "loss": 0.172,
      "step": 8744
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.151383993029726,
      "learning_rate": 1.5949653069787018e-06,
      "loss": 0.1907,
      "step": 8745
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.0981571165891166,
      "learning_rate": 1.5933137816490229e-06,
      "loss": 0.1885,
      "step": 8746
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.1322928123104132,
      "learning_rate": 1.5916630377903696e-06,
      "loss": 0.2089,
      "step": 8747
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.8842879503301753,
      "learning_rate": 1.5900130755561916e-06,
      "loss": 0.158,
      "step": 8748
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.3493106137730346,
      "learning_rate": 1.588363895099866e-06,
      "loss": 0.22,
      "step": 8749
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.7481804230222233,
      "learning_rate": 1.5867154965746956e-06,
      "loss": 0.1594,
      "step": 8750
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.9805537164386793,
      "learning_rate": 1.585067880133916e-06,
      "loss": 0.2114,
      "step": 8751
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.454876093566846,
      "learning_rate": 1.583421045930682e-06,
      "loss": 0.2034,
      "step": 8752
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.0440738755534333,
      "learning_rate": 1.5817749941180893e-06,
      "loss": 0.2095,
      "step": 8753
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.217006637921471,
      "learning_rate": 1.580129724849141e-06,
      "loss": 0.229,
      "step": 8754
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.972416051208057,
      "learning_rate": 1.5784852382767856e-06,
      "loss": 0.1582,
      "step": 8755
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.711077001966315,
      "learning_rate": 1.5768415345538911e-06,
      "loss": 0.2241,
      "step": 8756
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.7911885129744174,
      "learning_rate": 1.575198613833252e-06,
      "loss": 0.1145,
      "step": 8757
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.360645248339212,
      "learning_rate": 1.5735564762675914e-06,
      "loss": 0.2131,
      "step": 8758
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.8023863011271177,
      "learning_rate": 1.5719151220095596e-06,
      "loss": 0.1851,
      "step": 8759
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.3118998151480694,
      "learning_rate": 1.5702745512117323e-06,
      "loss": 0.2273,
      "step": 8760
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.2882733952344396,
      "learning_rate": 1.5686347640266208e-06,
      "loss": 0.2224,
      "step": 8761
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.996846346619261,
      "learning_rate": 1.56699576060665e-06,
      "loss": 0.2031,
      "step": 8762
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.1033391939065496,
      "learning_rate": 1.5653575411041788e-06,
      "loss": 0.1355,
      "step": 8763
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.1432796517608192,
      "learning_rate": 1.5637201056714967e-06,
      "loss": 0.1904,
      "step": 8764
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.3188252429072334,
      "learning_rate": 1.5620834544608166e-06,
      "loss": 0.2512,
      "step": 8765
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.7939263335470745,
      "learning_rate": 1.5604475876242775e-06,
      "loss": 0.1575,
      "step": 8766
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.0250824034848898,
      "learning_rate": 1.558812505313947e-06,
      "loss": 0.2117,
      "step": 8767
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.194741174361029,
      "learning_rate": 1.5571782076818197e-06,
      "loss": 0.2173,
      "step": 8768
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.0742410469497115,
      "learning_rate": 1.5555446948798147e-06,
      "loss": 0.2205,
      "step": 8769
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.372051630985507,
      "learning_rate": 1.553911967059788e-06,
      "loss": 0.2329,
      "step": 8770
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.1074905202859604,
      "learning_rate": 1.5522800243735037e-06,
      "loss": 0.1797,
      "step": 8771
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.9540189189904156,
      "learning_rate": 1.5506488669726738e-06,
      "loss": 0.2227,
      "step": 8772
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.8962720784363498,
      "learning_rate": 1.549018495008925e-06,
      "loss": 0.1448,
      "step": 8773
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.7699195343257241,
      "learning_rate": 1.5473889086338134e-06,
      "loss": 0.1541,
      "step": 8774
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.267100284473143,
      "learning_rate": 1.5457601079988226e-06,
      "loss": 0.2,
      "step": 8775
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.9424369539826987,
      "learning_rate": 1.5441320932553627e-06,
      "loss": 0.1777,
      "step": 8776
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.2998685744966814,
      "learning_rate": 1.5425048645547703e-06,
      "loss": 0.1692,
      "step": 8777
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.1377882639647905,
      "learning_rate": 1.5408784220483152e-06,
      "loss": 0.1987,
      "step": 8778
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.8444549916171495,
      "learning_rate": 1.5392527658871813e-06,
      "loss": 0.1484,
      "step": 8779
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.598610749315101,
      "learning_rate": 1.537627896222489e-06,
      "loss": 0.2493,
      "step": 8780
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.0964801930524875,
      "learning_rate": 1.5360038132052869e-06,
      "loss": 0.1924,
      "step": 8781
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.3066546046655034,
      "learning_rate": 1.5343805169865434e-06,
      "loss": 0.2258,
      "step": 8782
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5634220798451304,
      "learning_rate": 1.5327580077171589e-06,
      "loss": 0.1284,
      "step": 8783
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.8490729611041365,
      "learning_rate": 1.5311362855479584e-06,
      "loss": 0.1656,
      "step": 8784
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.295586958814283,
      "learning_rate": 1.5295153506296944e-06,
      "loss": 0.2629,
      "step": 8785
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.85072608280291,
      "learning_rate": 1.5278952031130445e-06,
      "loss": 0.1658,
      "step": 8786
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.771435849721423,
      "learning_rate": 1.5262758431486213e-06,
      "loss": 0.1672,
      "step": 8787
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.0901022172984893,
      "learning_rate": 1.524657270886951e-06,
      "loss": 0.1982,
      "step": 8788
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.373559353878017,
      "learning_rate": 1.5230394864784925e-06,
      "loss": 0.1799,
      "step": 8789
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.3195577625321255,
      "learning_rate": 1.5214224900736375e-06,
      "loss": 0.1995,
      "step": 8790
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.45929275709626,
      "learning_rate": 1.5198062818226967e-06,
      "loss": 0.2867,
      "step": 8791
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.4424836890618065,
      "learning_rate": 1.5181908618759101e-06,
      "loss": 0.262,
      "step": 8792
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.2173179577949487,
      "learning_rate": 1.5165762303834442e-06,
      "loss": 0.2507,
      "step": 8793
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.095590900155366,
      "learning_rate": 1.5149623874953922e-06,
      "loss": 0.2278,
      "step": 8794
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.241510732638494,
      "learning_rate": 1.5133493333617755e-06,
      "loss": 0.1699,
      "step": 8795
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.9288395460078116,
      "learning_rate": 1.5117370681325393e-06,
      "loss": 0.187,
      "step": 8796
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.980380693193512,
      "learning_rate": 1.5101255919575552e-06,
      "loss": 0.1996,
      "step": 8797
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.2396013332050364,
      "learning_rate": 1.5085149049866277e-06,
      "loss": 0.2097,
      "step": 8798
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.433670369884965,
      "learning_rate": 1.5069050073694813e-06,
      "loss": 0.2063,
      "step": 8799
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.951019647666109,
      "learning_rate": 1.5052958992557687e-06,
      "loss": 0.1768,
      "step": 8800
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.9834402905107713,
      "learning_rate": 1.5036875807950712e-06,
      "loss": 0.189,
      "step": 8801
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.649524522637361,
      "learning_rate": 1.502080052136894e-06,
      "loss": 0.2524,
      "step": 8802
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.9634978475178904,
      "learning_rate": 1.5004733134306692e-06,
      "loss": 0.1548,
      "step": 8803
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.8554212340884644,
      "learning_rate": 1.4988673648257624e-06,
      "loss": 0.1565,
      "step": 8804
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.7403019495555234,
      "learning_rate": 1.4972622064714515e-06,
      "loss": 0.1708,
      "step": 8805
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.8782783545933102,
      "learning_rate": 1.495657838516953e-06,
      "loss": 0.1637,
      "step": 8806
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.6803317388333319,
      "learning_rate": 1.4940542611114073e-06,
      "loss": 0.16,
      "step": 8807
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.9304227174211819,
      "learning_rate": 1.4924514744038787e-06,
      "loss": 0.1724,
      "step": 8808
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.368332985588615,
      "learning_rate": 1.4908494785433603e-06,
      "loss": 0.2603,
      "step": 8809
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.204330477182204,
      "learning_rate": 1.4892482736787717e-06,
      "loss": 0.2137,
      "step": 8810
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.9952337814852839,
      "learning_rate": 1.487647859958956e-06,
      "loss": 0.1956,
      "step": 8811
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.4603228861620825,
      "learning_rate": 1.4860482375326857e-06,
      "loss": 0.1741,
      "step": 8812
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.3752206359019126,
      "learning_rate": 1.4844494065486592e-06,
      "loss": 0.2537,
      "step": 8813
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.9181790300435726,
      "learning_rate": 1.4828513671554978e-06,
      "loss": 0.1904,
      "step": 8814
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.0855202739885184,
      "learning_rate": 1.4812541195017593e-06,
      "loss": 0.2332,
      "step": 8815
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.7392704555805363,
      "learning_rate": 1.4796576637359194e-06,
      "loss": 0.1704,
      "step": 8816
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.082854137639599,
      "learning_rate": 1.478062000006375e-06,
      "loss": 0.1636,
      "step": 8817
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.1579135091181056,
      "learning_rate": 1.4764671284614629e-06,
      "loss": 0.1799,
      "step": 8818
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.7655868903041543,
      "learning_rate": 1.474873049249439e-06,
      "loss": 0.3052,
      "step": 8819
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.3164829481678044,
      "learning_rate": 1.4732797625184814e-06,
      "loss": 0.2148,
      "step": 8820
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.2772210493492455,
      "learning_rate": 1.4716872684167082e-06,
      "loss": 0.2019,
      "step": 8821
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.201479090005346,
      "learning_rate": 1.4700955670921468e-06,
      "loss": 0.1948,
      "step": 8822
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.346802999903961,
      "learning_rate": 1.4685046586927598e-06,
      "loss": 0.2581,
      "step": 8823
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.443209244500412,
      "learning_rate": 1.46691454336644e-06,
      "loss": 0.2065,
      "step": 8824
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.2086400315481085,
      "learning_rate": 1.465325221260998e-06,
      "loss": 0.2031,
      "step": 8825
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.8632252114303587,
      "learning_rate": 1.463736692524176e-06,
      "loss": 0.1638,
      "step": 8826
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.1678293780924682,
      "learning_rate": 1.4621489573036407e-06,
      "loss": 0.207,
      "step": 8827
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.0721883075753356,
      "learning_rate": 1.4605620157469835e-06,
      "loss": 0.1832,
      "step": 8828
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.1986266227472444,
      "learning_rate": 1.4589758680017263e-06,
      "loss": 0.1924,
      "step": 8829
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.311019321289068,
      "learning_rate": 1.4573905142153134e-06,
      "loss": 0.2412,
      "step": 8830
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.9969291338026824,
      "learning_rate": 1.4558059545351144e-06,
      "loss": 0.1819,
      "step": 8831
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.1381046794288805,
      "learning_rate": 1.4542221891084307e-06,
      "loss": 0.1736,
      "step": 8832
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.305491880252154,
      "learning_rate": 1.4526392180824888e-06,
      "loss": 0.1549,
      "step": 8833
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.4871111621062187,
      "learning_rate": 1.45105704160443e-06,
      "loss": 0.2065,
      "step": 8834
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.896379796324328,
      "learning_rate": 1.4494756598213377e-06,
      "loss": 0.1552,
      "step": 8835
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.429703420340253,
      "learning_rate": 1.4478950728802132e-06,
      "loss": 0.2358,
      "step": 8836
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.2576006780034223,
      "learning_rate": 1.4463152809279824e-06,
      "loss": 0.2034,
      "step": 8837
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.8418102880255143,
      "learning_rate": 1.4447362841115075e-06,
      "loss": 0.2087,
      "step": 8838
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.859016544884904,
      "learning_rate": 1.4431580825775604e-06,
      "loss": 0.1812,
      "step": 8839
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.3764568162749464,
      "learning_rate": 1.44158067647285e-06,
      "loss": 0.2373,
      "step": 8840
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.4382728260428386,
      "learning_rate": 1.440004065944014e-06,
      "loss": 0.2676,
      "step": 8841
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.134069422196982,
      "learning_rate": 1.4384282511376102e-06,
      "loss": 0.218,
      "step": 8842
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.0413133297811625,
      "learning_rate": 1.4368532322001161e-06,
      "loss": 0.2007,
      "step": 8843
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.20937246622789,
      "learning_rate": 1.4352790092779511e-06,
      "loss": 0.1853,
      "step": 8844
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.3388580179584015,
      "learning_rate": 1.4337055825174506e-06,
      "loss": 0.2021,
      "step": 8845
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.204211775468109,
      "learning_rate": 1.4321329520648752e-06,
      "loss": 0.1816,
      "step": 8846
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.6932482183081237,
      "learning_rate": 1.4305611180664157e-06,
      "loss": 0.1368,
      "step": 8847
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.9662719252529923,
      "learning_rate": 1.4289900806681866e-06,
      "loss": 0.1868,
      "step": 8848
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.3794867858061513,
      "learning_rate": 1.4274198400162265e-06,
      "loss": 0.187,
      "step": 8849
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.0369541983994477,
      "learning_rate": 1.4258503962565096e-06,
      "loss": 0.1753,
      "step": 8850
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.809418185689096,
      "learning_rate": 1.424281749534918e-06,
      "loss": 0.1527,
      "step": 8851
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.1125672460148595,
      "learning_rate": 1.4227138999972801e-06,
      "loss": 0.2139,
      "step": 8852
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.0906264201734763,
      "learning_rate": 1.4211468477893352e-06,
      "loss": 0.2273,
      "step": 8853
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.195852919279068,
      "learning_rate": 1.4195805930567552e-06,
      "loss": 0.2227,
      "step": 8854
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.756136101086953,
      "learning_rate": 1.4180151359451367e-06,
      "loss": 0.1543,
      "step": 8855
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.265118585900519,
      "learning_rate": 1.416450476600001e-06,
      "loss": 0.2278,
      "step": 8856
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.8359532368704534,
      "learning_rate": 1.414886615166794e-06,
      "loss": 0.1687,
      "step": 8857
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.8113292330825332,
      "learning_rate": 1.4133235517908938e-06,
      "loss": 0.1438,
      "step": 8858
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.2377982507900636,
      "learning_rate": 1.4117612866176022e-06,
      "loss": 0.2102,
      "step": 8859
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.8888013225835552,
      "learning_rate": 1.4101998197921352e-06,
      "loss": 0.1674,
      "step": 8860
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.137647676671807,
      "learning_rate": 1.4086391514596532e-06,
      "loss": 0.2134,
      "step": 8861
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.0908382318376875,
      "learning_rate": 1.407079281765229e-06,
      "loss": 0.1532,
      "step": 8862
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.660538703801642,
      "learning_rate": 1.4055202108538657e-06,
      "loss": 0.1448,
      "step": 8863
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.8485160667908842,
      "learning_rate": 1.4039619388704928e-06,
      "loss": 0.177,
      "step": 8864
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.219775364646382,
      "learning_rate": 1.4024044659599633e-06,
      "loss": 0.2283,
      "step": 8865
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.8732434660378425,
      "learning_rate": 1.4008477922670571e-06,
      "loss": 0.2119,
      "step": 8866
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.01690277758928,
      "learning_rate": 1.399291917936484e-06,
      "loss": 0.2041,
      "step": 8867
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.4045048418339143,
      "learning_rate": 1.3977368431128679e-06,
      "loss": 0.2126,
      "step": 8868
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.1148726324396736,
      "learning_rate": 1.3961825679407726e-06,
      "loss": 0.1768,
      "step": 8869
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.238042307751252,
      "learning_rate": 1.3946290925646788e-06,
      "loss": 0.2212,
      "step": 8870
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.9336321285153646,
      "learning_rate": 1.3930764171289935e-06,
      "loss": 0.1799,
      "step": 8871
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.04783195988911,
      "learning_rate": 1.391524541778052e-06,
      "loss": 0.1758,
      "step": 8872
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.107069967988312,
      "learning_rate": 1.3899734666561138e-06,
      "loss": 0.1777,
      "step": 8873
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.317356901466061,
      "learning_rate": 1.388423191907361e-06,
      "loss": 0.2336,
      "step": 8874
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.084780866939736,
      "learning_rate": 1.3868737176759105e-06,
      "loss": 0.2155,
      "step": 8875
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.224101968600388,
      "learning_rate": 1.3853250441057975e-06,
      "loss": 0.2241,
      "step": 8876
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.2391812448881234,
      "learning_rate": 1.3837771713409776e-06,
      "loss": 0.2153,
      "step": 8877
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.1535899777478775,
      "learning_rate": 1.3822300995253445e-06,
      "loss": 0.1748,
      "step": 8878
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.9353163617035234,
      "learning_rate": 1.3806838288027113e-06,
      "loss": 0.1682,
      "step": 8879
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.6275011153051038,
      "learning_rate": 1.379138359316814e-06,
      "loss": 0.1575,
      "step": 8880
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.9538563230825197,
      "learning_rate": 1.3775936912113187e-06,
      "loss": 0.2115,
      "step": 8881
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.097891096704142,
      "learning_rate": 1.3760498246298138e-06,
      "loss": 0.241,
      "step": 8882
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.7383619785962159,
      "learning_rate": 1.3745067597158123e-06,
      "loss": 0.1304,
      "step": 8883
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.1963645989917384,
      "learning_rate": 1.372964496612763e-06,
      "loss": 0.1764,
      "step": 8884
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.9909623764251267,
      "learning_rate": 1.3714230354640234e-06,
      "loss": 0.1731,
      "step": 8885
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.9322009648960328,
      "learning_rate": 1.3698823764128867e-06,
      "loss": 0.1692,
      "step": 8886
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.3491582405901057,
      "learning_rate": 1.3683425196025734e-06,
      "loss": 0.2034,
      "step": 8887
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.0935361993570103,
      "learning_rate": 1.366803465176223e-06,
      "loss": 0.2388,
      "step": 8888
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.0967443567881134,
      "learning_rate": 1.365265213276905e-06,
      "loss": 0.1595,
      "step": 8889
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.1991549045337884,
      "learning_rate": 1.363727764047612e-06,
      "loss": 0.2034,
      "step": 8890
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.155574166454167,
      "learning_rate": 1.362191117631263e-06,
      "loss": 0.1846,
      "step": 8891
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.8078917565812558,
      "learning_rate": 1.360655274170698e-06,
      "loss": 0.1743,
      "step": 8892
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.0054746795941996,
      "learning_rate": 1.359120233808695e-06,
      "loss": 0.176,
      "step": 8893
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.164926038410454,
      "learning_rate": 1.357585996687939e-06,
      "loss": 0.1833,
      "step": 8894
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.838221715123701,
      "learning_rate": 1.3560525629510567e-06,
      "loss": 0.1511,
      "step": 8895
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.093285852246565,
      "learning_rate": 1.3545199327405922e-06,
      "loss": 0.2089,
      "step": 8896
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.2154187175634514,
      "learning_rate": 1.3529881061990147e-06,
      "loss": 0.2229,
      "step": 8897
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.9920822304462988,
      "learning_rate": 1.3514570834687203e-06,
      "loss": 0.1943,
      "step": 8898
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.143059384524886,
      "learning_rate": 1.3499268646920317e-06,
      "loss": 0.209,
      "step": 8899
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.669439294529817,
      "learning_rate": 1.3483974500111907e-06,
      "loss": 0.2573,
      "step": 8900
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.8807266522629227,
      "learning_rate": 1.3468688395683783e-06,
      "loss": 0.1589,
      "step": 8901
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.9205438185582808,
      "learning_rate": 1.3453410335056837e-06,
      "loss": 0.1741,
      "step": 8902
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.9632937237501238,
      "learning_rate": 1.3438140319651283e-06,
      "loss": 0.199,
      "step": 8903
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.8862106299285406,
      "learning_rate": 1.3422878350886658e-06,
      "loss": 0.1699,
      "step": 8904
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.3685087162432352,
      "learning_rate": 1.3407624430181644e-06,
      "loss": 0.2485,
      "step": 8905
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.9914402609661432,
      "learning_rate": 1.3392378558954233e-06,
      "loss": 0.168,
      "step": 8906
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.1919396249207588,
      "learning_rate": 1.3377140738621663e-06,
      "loss": 0.2222,
      "step": 8907
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.962201473727107,
      "learning_rate": 1.33619109706004e-06,
      "loss": 0.1746,
      "step": 8908
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.1486388180284237,
      "learning_rate": 1.3346689256306155e-06,
      "loss": 0.23,
      "step": 8909
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.9976239899442563,
      "learning_rate": 1.3331475597153988e-06,
      "loss": 0.2115,
      "step": 8910
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.78810049170648,
      "learning_rate": 1.331626999455804e-06,
      "loss": 0.1621,
      "step": 8911
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.957088201895097,
      "learning_rate": 1.3301072449931862e-06,
      "loss": 0.1941,
      "step": 8912
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.460235537661321,
      "learning_rate": 1.3285882964688168e-06,
      "loss": 0.1807,
      "step": 8913
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.0775304358285633,
      "learning_rate": 1.3270701540238962e-06,
      "loss": 0.1898,
      "step": 8914
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.2484471518725937,
      "learning_rate": 1.325552817799547e-06,
      "loss": 0.2529,
      "step": 8915
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.1987686623817524,
      "learning_rate": 1.3240362879368184e-06,
      "loss": 0.1814,
      "step": 8916
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.918977296289576,
      "learning_rate": 1.3225205645766815e-06,
      "loss": 0.1843,
      "step": 8917
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.9799748686991214,
      "learning_rate": 1.3210056478600431e-06,
      "loss": 0.1875,
      "step": 8918
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.248169115560358,
      "learning_rate": 1.3194915379277195e-06,
      "loss": 0.2307,
      "step": 8919
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.9356791270786182,
      "learning_rate": 1.3179782349204618e-06,
      "loss": 0.1641,
      "step": 8920
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.136298974193166,
      "learning_rate": 1.3164657389789459e-06,
      "loss": 0.2042,
      "step": 8921
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.722725851087367,
      "learning_rate": 1.314954050243772e-06,
      "loss": 0.1578,
      "step": 8922
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.0177242231433743,
      "learning_rate": 1.3134431688554572e-06,
      "loss": 0.1687,
      "step": 8923
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.0598866348003626,
      "learning_rate": 1.3119330949544573e-06,
      "loss": 0.196,
      "step": 8924
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.3576104576406043,
      "learning_rate": 1.3104238286811433e-06,
      "loss": 0.2039,
      "step": 8925
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.22527051015395,
      "learning_rate": 1.3089153701758128e-06,
      "loss": 0.1985,
      "step": 8926
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.9564555286828014,
      "learning_rate": 1.307407719578696e-06,
      "loss": 0.1724,
      "step": 8927
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.3631393971040917,
      "learning_rate": 1.305900877029932e-06,
      "loss": 0.2095,
      "step": 8928
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.098174444384442,
      "learning_rate": 1.3043948426696019e-06,
      "loss": 0.1799,
      "step": 8929
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.091826098702823,
      "learning_rate": 1.3028896166377003e-06,
      "loss": 0.179,
      "step": 8930
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.8773206816044516,
      "learning_rate": 1.301385199074151e-06,
      "loss": 0.1722,
      "step": 8931
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.06765921375073,
      "learning_rate": 1.2998815901188033e-06,
      "loss": 0.187,
      "step": 8932
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.230345508564583,
      "learning_rate": 1.2983787899114286e-06,
      "loss": 0.2246,
      "step": 8933
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.6612549735356712,
      "learning_rate": 1.296876798591723e-06,
      "loss": 0.146,
      "step": 8934
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.450215758851347,
      "learning_rate": 1.2953756162993158e-06,
      "loss": 0.3115,
      "step": 8935
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.9225887819428904,
      "learning_rate": 1.2938752431737467e-06,
      "loss": 0.176,
      "step": 8936
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.1207857854614787,
      "learning_rate": 1.2923756793544895e-06,
      "loss": 0.2195,
      "step": 8937
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.263195665701127,
      "learning_rate": 1.290876924980944e-06,
      "loss": 0.2236,
      "step": 8938
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.1583875657683165,
      "learning_rate": 1.2893789801924328e-06,
      "loss": 0.1951,
      "step": 8939
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.9376590166548995,
      "learning_rate": 1.2878818451281939e-06,
      "loss": 0.1747,
      "step": 8940
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.9031448561806674,
      "learning_rate": 1.2863855199274079e-06,
      "loss": 0.1875,
      "step": 8941
    },
    {
      "epoch": 2.53,
      "grad_norm": 2.327773950964909,
      "learning_rate": 1.2848900047291657e-06,
      "loss": 0.1895,
      "step": 8942
    },
    {
      "epoch": 2.53,
      "grad_norm": 2.0504708608494235,
      "learning_rate": 1.2833952996724864e-06,
      "loss": 0.2224,
      "step": 8943
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.776236470007933,
      "learning_rate": 1.281901404896323e-06,
      "loss": 0.1917,
      "step": 8944
    },
    {
      "epoch": 2.53,
      "grad_norm": 2.232381901763988,
      "learning_rate": 1.280408320539538e-06,
      "loss": 0.2024,
      "step": 8945
    },
    {
      "epoch": 2.53,
      "grad_norm": 2.793997811890734,
      "learning_rate": 1.2789160467409244e-06,
      "loss": 0.2368,
      "step": 8946
    },
    {
      "epoch": 2.53,
      "grad_norm": 2.296088680366618,
      "learning_rate": 1.2774245836392085e-06,
      "loss": 0.2473,
      "step": 8947
    },
    {
      "epoch": 2.53,
      "grad_norm": 2.1139839001114438,
      "learning_rate": 1.2759339313730302e-06,
      "loss": 0.1975,
      "step": 8948
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.84771370714577,
      "learning_rate": 1.2744440900809584e-06,
      "loss": 0.174,
      "step": 8949
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.6392391546688778,
      "learning_rate": 1.2729550599014862e-06,
      "loss": 0.1395,
      "step": 8950
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.8060735032579749,
      "learning_rate": 1.2714668409730312e-06,
      "loss": 0.1631,
      "step": 8951
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.7766342331921272,
      "learning_rate": 1.2699794334339356e-06,
      "loss": 0.165,
      "step": 8952
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.8270860037885388,
      "learning_rate": 1.268492837422467e-06,
      "loss": 0.1562,
      "step": 8953
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.7479546980473226,
      "learning_rate": 1.2670070530768131e-06,
      "loss": 0.121,
      "step": 8954
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.7434354404250745,
      "learning_rate": 1.2655220805350953e-06,
      "loss": 0.1411,
      "step": 8955
    },
    {
      "epoch": 2.53,
      "grad_norm": 2.1131812844563984,
      "learning_rate": 1.2640379199353536e-06,
      "loss": 0.2034,
      "step": 8956
    },
    {
      "epoch": 2.53,
      "grad_norm": 2.0797597796066563,
      "learning_rate": 1.2625545714155474e-06,
      "loss": 0.2366,
      "step": 8957
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.9940873733398925,
      "learning_rate": 1.2610720351135718e-06,
      "loss": 0.188,
      "step": 8958
    },
    {
      "epoch": 2.53,
      "grad_norm": 2.0588389523088733,
      "learning_rate": 1.259590311167238e-06,
      "loss": 0.1483,
      "step": 8959
    },
    {
      "epoch": 2.53,
      "grad_norm": 2.128315446152783,
      "learning_rate": 1.2581093997142846e-06,
      "loss": 0.1517,
      "step": 8960
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.8880591035955554,
      "learning_rate": 1.256629300892379e-06,
      "loss": 0.1643,
      "step": 8961
    },
    {
      "epoch": 2.53,
      "grad_norm": 2.5069965791534674,
      "learning_rate": 1.2551500148391026e-06,
      "loss": 0.2427,
      "step": 8962
    },
    {
      "epoch": 2.53,
      "grad_norm": 2.10667817165374,
      "learning_rate": 1.2536715416919676e-06,
      "loss": 0.1708,
      "step": 8963
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.9869029722665354,
      "learning_rate": 1.252193881588415e-06,
      "loss": 0.1501,
      "step": 8964
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.9975323904640931,
      "learning_rate": 1.2507170346658027e-06,
      "loss": 0.1853,
      "step": 8965
    },
    {
      "epoch": 2.53,
      "grad_norm": 2.32967036961373,
      "learning_rate": 1.2492410010614154e-06,
      "loss": 0.2441,
      "step": 8966
    },
    {
      "epoch": 2.53,
      "grad_norm": 2.0914729821076063,
      "learning_rate": 1.2477657809124632e-06,
      "loss": 0.1794,
      "step": 8967
    },
    {
      "epoch": 2.53,
      "grad_norm": 2.035333884944253,
      "learning_rate": 1.246291374356081e-06,
      "loss": 0.1838,
      "step": 8968
    },
    {
      "epoch": 2.53,
      "grad_norm": 2.145673137968265,
      "learning_rate": 1.244817781529326e-06,
      "loss": 0.1509,
      "step": 8969
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.970267986928022,
      "learning_rate": 1.2433450025691807e-06,
      "loss": 0.1895,
      "step": 8970
    },
    {
      "epoch": 2.53,
      "grad_norm": 2.152045709013826,
      "learning_rate": 1.24187303761255e-06,
      "loss": 0.21,
      "step": 8971
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.9571415489475545,
      "learning_rate": 1.2404018867962697e-06,
      "loss": 0.165,
      "step": 8972
    },
    {
      "epoch": 2.53,
      "grad_norm": 2.1780617676956386,
      "learning_rate": 1.2389315502570965e-06,
      "loss": 0.1763,
      "step": 8973
    },
    {
      "epoch": 2.53,
      "grad_norm": 2.4194951361510655,
      "learning_rate": 1.2374620281317019e-06,
      "loss": 0.1926,
      "step": 8974
    },
    {
      "epoch": 2.53,
      "grad_norm": 2.23099866797671,
      "learning_rate": 1.2359933205566987e-06,
      "loss": 0.1888,
      "step": 8975
    },
    {
      "epoch": 2.53,
      "grad_norm": 2.379593558791617,
      "learning_rate": 1.2345254276686114e-06,
      "loss": 0.2424,
      "step": 8976
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.3627664660911676,
      "learning_rate": 1.2330583496038929e-06,
      "loss": 0.2488,
      "step": 8977
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.3134742836022086,
      "learning_rate": 1.2315920864989218e-06,
      "loss": 0.2214,
      "step": 8978
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.7132692644938654,
      "learning_rate": 1.230126638489998e-06,
      "loss": 0.1479,
      "step": 8979
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.9989338039833084,
      "learning_rate": 1.2286620057133459e-06,
      "loss": 0.1787,
      "step": 8980
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.8669389031636423,
      "learning_rate": 1.2271981883051187e-06,
      "loss": 0.1598,
      "step": 8981
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.7350311468548423,
      "learning_rate": 1.22573518640139e-06,
      "loss": 0.2695,
      "step": 8982
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.106412647102995,
      "learning_rate": 1.2242730001381532e-06,
      "loss": 0.2356,
      "step": 8983
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.4443693184465647,
      "learning_rate": 1.2228116296513348e-06,
      "loss": 0.1963,
      "step": 8984
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.08457974727499,
      "learning_rate": 1.221351075076781e-06,
      "loss": 0.187,
      "step": 8985
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.0065950035466646,
      "learning_rate": 1.2198913365502606e-06,
      "loss": 0.155,
      "step": 8986
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.1652306272406987,
      "learning_rate": 1.218432414207471e-06,
      "loss": 0.2041,
      "step": 8987
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.1683590676779794,
      "learning_rate": 1.216974308184029e-06,
      "loss": 0.1882,
      "step": 8988
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.216265518152155,
      "learning_rate": 1.2155170186154753e-06,
      "loss": 0.1719,
      "step": 8989
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.903697390028462,
      "learning_rate": 1.2140605456372856e-06,
      "loss": 0.1426,
      "step": 8990
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.874284953191223,
      "learning_rate": 1.2126048893848396e-06,
      "loss": 0.1826,
      "step": 8991
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.8134231696977037,
      "learning_rate": 1.2111500499934613e-06,
      "loss": 0.17,
      "step": 8992
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.239719617923601,
      "learning_rate": 1.2096960275983872e-06,
      "loss": 0.2073,
      "step": 8993
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.2138147181546257,
      "learning_rate": 1.208242822334781e-06,
      "loss": 0.2163,
      "step": 8994
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.7086028424220545,
      "learning_rate": 1.206790434337729e-06,
      "loss": 0.1611,
      "step": 8995
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.260071236203762,
      "learning_rate": 1.2053388637422437e-06,
      "loss": 0.2043,
      "step": 8996
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.8625088001266534,
      "learning_rate": 1.203888110683259e-06,
      "loss": 0.1462,
      "step": 8997
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.006070012201298,
      "learning_rate": 1.2024381752956372e-06,
      "loss": 0.1699,
      "step": 8998
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.7355590508666554,
      "learning_rate": 1.2009890577141625e-06,
      "loss": 0.1339,
      "step": 8999
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.7445469491124372,
      "learning_rate": 1.1995407580735364e-06,
      "loss": 0.3,
      "step": 9000
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.5165204013026083,
      "learning_rate": 1.1980932765083964e-06,
      "loss": 0.241,
      "step": 9001
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.0422973408024983,
      "learning_rate": 1.196646613153295e-06,
      "loss": 0.1868,
      "step": 9002
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.148906547394862,
      "learning_rate": 1.1952007681427124e-06,
      "loss": 0.2498,
      "step": 9003
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.046605488740996,
      "learning_rate": 1.1937557416110512e-06,
      "loss": 0.167,
      "step": 9004
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.6886123188020312,
      "learning_rate": 1.1923115336926394e-06,
      "loss": 0.1466,
      "step": 9005
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.7211901388562618,
      "learning_rate": 1.1908681445217263e-06,
      "loss": 0.1389,
      "step": 9006
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.1366452921996046,
      "learning_rate": 1.189425574232491e-06,
      "loss": 0.1907,
      "step": 9007
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.95264743274222,
      "learning_rate": 1.1879838229590269e-06,
      "loss": 0.2056,
      "step": 9008
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.9733413409612035,
      "learning_rate": 1.1865428908353606e-06,
      "loss": 0.1434,
      "step": 9009
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.049055794713297,
      "learning_rate": 1.1851027779954373e-06,
      "loss": 0.1848,
      "step": 9010
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.812699479078422,
      "learning_rate": 1.1836634845731288e-06,
      "loss": 0.1682,
      "step": 9011
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.718474153475324,
      "learning_rate": 1.1822250107022271e-06,
      "loss": 0.1472,
      "step": 9012
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.2481519681524427,
      "learning_rate": 1.1807873565164507e-06,
      "loss": 0.1918,
      "step": 9013
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.8920298888826836,
      "learning_rate": 1.1793505221494405e-06,
      "loss": 0.1558,
      "step": 9014
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.830170819676475,
      "learning_rate": 1.1779145077347653e-06,
      "loss": 0.163,
      "step": 9015
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.179429548480052,
      "learning_rate": 1.176479313405916e-06,
      "loss": 0.1992,
      "step": 9016
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.1942749355219617,
      "learning_rate": 1.175044939296297e-06,
      "loss": 0.1708,
      "step": 9017
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.2231966784124833,
      "learning_rate": 1.173611385539254e-06,
      "loss": 0.2312,
      "step": 9018
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.9605462317931044,
      "learning_rate": 1.1721786522680445e-06,
      "loss": 0.1792,
      "step": 9019
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.544133695336184,
      "learning_rate": 1.1707467396158524e-06,
      "loss": 0.2595,
      "step": 9020
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.13463838489652,
      "learning_rate": 1.1693156477157863e-06,
      "loss": 0.1716,
      "step": 9021
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.9260106387305747,
      "learning_rate": 1.167885376700879e-06,
      "loss": 0.1553,
      "step": 9022
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.9519494578946734,
      "learning_rate": 1.1664559267040821e-06,
      "loss": 0.151,
      "step": 9023
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.2600941047900065,
      "learning_rate": 1.1650272978582823e-06,
      "loss": 0.1816,
      "step": 9024
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.058789996237396,
      "learning_rate": 1.1635994902962767e-06,
      "loss": 0.2222,
      "step": 9025
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.20596090325446,
      "learning_rate": 1.1621725041507904e-06,
      "loss": 0.2268,
      "step": 9026
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.225571003418639,
      "learning_rate": 1.1607463395544782e-06,
      "loss": 0.22,
      "step": 9027
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.3410623003876307,
      "learning_rate": 1.159320996639911e-06,
      "loss": 0.2336,
      "step": 9028
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.0620401734485214,
      "learning_rate": 1.1578964755395883e-06,
      "loss": 0.1707,
      "step": 9029
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.1721733541013957,
      "learning_rate": 1.1564727763859306e-06,
      "loss": 0.2329,
      "step": 9030
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.1941777611966344,
      "learning_rate": 1.1550498993112812e-06,
      "loss": 0.1849,
      "step": 9031
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.0582469591897867,
      "learning_rate": 1.1536278444479066e-06,
      "loss": 0.2029,
      "step": 9032
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.2001165694943734,
      "learning_rate": 1.1522066119280062e-06,
      "loss": 0.1656,
      "step": 9033
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.0776663927242582,
      "learning_rate": 1.1507862018836846e-06,
      "loss": 0.1936,
      "step": 9034
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.207955450910741,
      "learning_rate": 1.1493666144469894e-06,
      "loss": 0.1904,
      "step": 9035
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.2315391205671746,
      "learning_rate": 1.1479478497498796e-06,
      "loss": 0.1711,
      "step": 9036
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.362556781868599,
      "learning_rate": 1.146529907924241e-06,
      "loss": 0.2793,
      "step": 9037
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.8779099071808454,
      "learning_rate": 1.1451127891018832e-06,
      "loss": 0.1792,
      "step": 9038
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.918727426263548,
      "learning_rate": 1.143696493414539e-06,
      "loss": 0.1707,
      "step": 9039
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.9770066566529096,
      "learning_rate": 1.1422810209938627e-06,
      "loss": 0.1572,
      "step": 9040
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.0542304521543464,
      "learning_rate": 1.1408663719714418e-06,
      "loss": 0.188,
      "step": 9041
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.8742618875799844,
      "learning_rate": 1.1394525464787708e-06,
      "loss": 0.1816,
      "step": 9042
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.305179388639778,
      "learning_rate": 1.138039544647279e-06,
      "loss": 0.2239,
      "step": 9043
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.194286697850772,
      "learning_rate": 1.1366273666083194e-06,
      "loss": 0.2114,
      "step": 9044
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.975360606374602,
      "learning_rate": 1.1352160124931644e-06,
      "loss": 0.1848,
      "step": 9045
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.072391625368963,
      "learning_rate": 1.1338054824330092e-06,
      "loss": 0.1926,
      "step": 9046
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.157828627726227,
      "learning_rate": 1.1323957765589766e-06,
      "loss": 0.1965,
      "step": 9047
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.5561691956128656,
      "learning_rate": 1.1309868950021085e-06,
      "loss": 0.2002,
      "step": 9048
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.1026221625177643,
      "learning_rate": 1.1295788378933713e-06,
      "loss": 0.1692,
      "step": 9049
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.102746802827122,
      "learning_rate": 1.1281716053636616e-06,
      "loss": 0.23,
      "step": 9050
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.217733167157291,
      "learning_rate": 1.1267651975437844e-06,
      "loss": 0.2273,
      "step": 9051
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.9885878833038304,
      "learning_rate": 1.125359614564483e-06,
      "loss": 0.1863,
      "step": 9052
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.1044989798391245,
      "learning_rate": 1.1239548565564173e-06,
      "loss": 0.1824,
      "step": 9053
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.779540574575928,
      "learning_rate": 1.1225509236501698e-06,
      "loss": 0.1669,
      "step": 9054
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.4605089526604735,
      "learning_rate": 1.121147815976248e-06,
      "loss": 0.1685,
      "step": 9055
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.5924954803263427,
      "learning_rate": 1.119745533665083e-06,
      "loss": 0.2666,
      "step": 9056
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.0521226658513974,
      "learning_rate": 1.1183440768470255e-06,
      "loss": 0.2246,
      "step": 9057
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.470314827913713,
      "learning_rate": 1.1169434456523598e-06,
      "loss": 0.2061,
      "step": 9058
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.0725061715989446,
      "learning_rate": 1.1155436402112785e-06,
      "loss": 0.1995,
      "step": 9059
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.476886401304468,
      "learning_rate": 1.1141446606539063e-06,
      "loss": 0.2273,
      "step": 9060
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.9429465426381969,
      "learning_rate": 1.1127465071102938e-06,
      "loss": 0.1476,
      "step": 9061
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.313080886587548,
      "learning_rate": 1.1113491797104093e-06,
      "loss": 0.2251,
      "step": 9062
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.1001543920742365,
      "learning_rate": 1.109952678584144e-06,
      "loss": 0.2375,
      "step": 9063
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.9289489063967025,
      "learning_rate": 1.108557003861317e-06,
      "loss": 0.1858,
      "step": 9064
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.082079434288383,
      "learning_rate": 1.107162155671665e-06,
      "loss": 0.2117,
      "step": 9065
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.2095390772086105,
      "learning_rate": 1.1057681341448533e-06,
      "loss": 0.1819,
      "step": 9066
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.003938334557138,
      "learning_rate": 1.1043749394104665e-06,
      "loss": 0.1797,
      "step": 9067
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.004598401235914,
      "learning_rate": 1.1029825715980115e-06,
      "loss": 0.1329,
      "step": 9068
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.4388400507164505,
      "learning_rate": 1.1015910308369239e-06,
      "loss": 0.1902,
      "step": 9069
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.1471701600807185,
      "learning_rate": 1.1002003172565579e-06,
      "loss": 0.1802,
      "step": 9070
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.0616243524636793,
      "learning_rate": 1.0988104309861913e-06,
      "loss": 0.1833,
      "step": 9071
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.245462690772643,
      "learning_rate": 1.0974213721550264e-06,
      "loss": 0.1987,
      "step": 9072
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.340826400550566,
      "learning_rate": 1.0960331408921865e-06,
      "loss": 0.1978,
      "step": 9073
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.0349857490120575,
      "learning_rate": 1.0946457373267183e-06,
      "loss": 0.1948,
      "step": 9074
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.821732014177836,
      "learning_rate": 1.0932591615875975e-06,
      "loss": 0.1746,
      "step": 9075
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.9377997150372204,
      "learning_rate": 1.0918734138037113e-06,
      "loss": 0.1831,
      "step": 9076
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.070446717157587,
      "learning_rate": 1.0904884941038784e-06,
      "loss": 0.1926,
      "step": 9077
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.0641396516083934,
      "learning_rate": 1.0891044026168407e-06,
      "loss": 0.1536,
      "step": 9078
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.84812198735668,
      "learning_rate": 1.0877211394712617e-06,
      "loss": 0.1826,
      "step": 9079
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.8459472549083802,
      "learning_rate": 1.086338704795722e-06,
      "loss": 0.1504,
      "step": 9080
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.32266169861206,
      "learning_rate": 1.0849570987187341e-06,
      "loss": 0.2048,
      "step": 9081
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.0666203165626564,
      "learning_rate": 1.08357632136873e-06,
      "loss": 0.1672,
      "step": 9082
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.2372026079753735,
      "learning_rate": 1.0821963728740626e-06,
      "loss": 0.2483,
      "step": 9083
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.0012602775730555,
      "learning_rate": 1.0808172533630113e-06,
      "loss": 0.1758,
      "step": 9084
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.8382828912926028,
      "learning_rate": 1.0794389629637747e-06,
      "loss": 0.165,
      "step": 9085
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.0436271392719982,
      "learning_rate": 1.078061501804476e-06,
      "loss": 0.2202,
      "step": 9086
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.1783993846029306,
      "learning_rate": 1.076684870013165e-06,
      "loss": 0.2515,
      "step": 9087
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.4075949232290164,
      "learning_rate": 1.075309067717808e-06,
      "loss": 0.2268,
      "step": 9088
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.3303310239679362,
      "learning_rate": 1.0739340950462996e-06,
      "loss": 0.2202,
      "step": 9089
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.3274666386362157,
      "learning_rate": 1.0725599521264518e-06,
      "loss": 0.2029,
      "step": 9090
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.6779064669352626,
      "learning_rate": 1.071186639086005e-06,
      "loss": 0.126,
      "step": 9091
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.2562241006798014,
      "learning_rate": 1.0698141560526198e-06,
      "loss": 0.2085,
      "step": 9092
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.1995151695903954,
      "learning_rate": 1.068442503153878e-06,
      "loss": 0.2175,
      "step": 9093
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.005777629794316,
      "learning_rate": 1.0670716805172865e-06,
      "loss": 0.1992,
      "step": 9094
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.9966226952808093,
      "learning_rate": 1.0657016882702764e-06,
      "loss": 0.1873,
      "step": 9095
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.7961483603254673,
      "learning_rate": 1.0643325265402016e-06,
      "loss": 0.1987,
      "step": 9096
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.923872059104108,
      "learning_rate": 1.062964195454329e-06,
      "loss": 0.1807,
      "step": 9097
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.4974530915252817,
      "learning_rate": 1.061596695139865e-06,
      "loss": 0.1909,
      "step": 9098
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.2343988013325813,
      "learning_rate": 1.0602300257239262e-06,
      "loss": 0.2126,
      "step": 9099
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.263020576727418,
      "learning_rate": 1.0588641873335558e-06,
      "loss": 0.2231,
      "step": 9100
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.406234387520461,
      "learning_rate": 1.0574991800957203e-06,
      "loss": 0.2207,
      "step": 9101
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.9393631643003781,
      "learning_rate": 1.0561350041373086e-06,
      "loss": 0.1543,
      "step": 9102
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.9788766741905985,
      "learning_rate": 1.0547716595851298e-06,
      "loss": 0.2166,
      "step": 9103
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.174728890598107,
      "learning_rate": 1.0534091465659212e-06,
      "loss": 0.1556,
      "step": 9104
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.9498304175699237,
      "learning_rate": 1.0520474652063395e-06,
      "loss": 0.1757,
      "step": 9105
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.377154169080504,
      "learning_rate": 1.0506866156329632e-06,
      "loss": 0.2466,
      "step": 9106
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.016482360993597,
      "learning_rate": 1.0493265979722944e-06,
      "loss": 0.1931,
      "step": 9107
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.9797053350568143,
      "learning_rate": 1.0479674123507588e-06,
      "loss": 0.1975,
      "step": 9108
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.129834755287697,
      "learning_rate": 1.046609058894703e-06,
      "loss": 0.2004,
      "step": 9109
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.8975626698138572,
      "learning_rate": 1.0452515377303974e-06,
      "loss": 0.1694,
      "step": 9110
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.9895859543354337,
      "learning_rate": 1.0438948489840327e-06,
      "loss": 0.1565,
      "step": 9111
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.2169662949632256,
      "learning_rate": 1.0425389927817298e-06,
      "loss": 0.2529,
      "step": 9112
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.34630645190229,
      "learning_rate": 1.0411839692495241e-06,
      "loss": 0.2053,
      "step": 9113
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.0225312138454923,
      "learning_rate": 1.0398297785133727e-06,
      "loss": 0.1823,
      "step": 9114
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.015441408328001,
      "learning_rate": 1.0384764206991638e-06,
      "loss": 0.1858,
      "step": 9115
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.9699578408025735,
      "learning_rate": 1.0371238959327001e-06,
      "loss": 0.1775,
      "step": 9116
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.9353674384980577,
      "learning_rate": 1.0357722043397122e-06,
      "loss": 0.1787,
      "step": 9117
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.137293461853823,
      "learning_rate": 1.0344213460458496e-06,
      "loss": 0.1941,
      "step": 9118
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.2269022625691273,
      "learning_rate": 1.0330713211766864e-06,
      "loss": 0.2202,
      "step": 9119
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.8900173226484591,
      "learning_rate": 1.0317221298577163e-06,
      "loss": 0.134,
      "step": 9120
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.0722015589396996,
      "learning_rate": 1.0303737722143614e-06,
      "loss": 0.1997,
      "step": 9121
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.0823370100022762,
      "learning_rate": 1.0290262483719637e-06,
      "loss": 0.1978,
      "step": 9122
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.091182071412084,
      "learning_rate": 1.0276795584557796e-06,
      "loss": 0.1912,
      "step": 9123
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.0230919382161447,
      "learning_rate": 1.0263337025910015e-06,
      "loss": 0.2163,
      "step": 9124
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.0563177839372844,
      "learning_rate": 1.0249886809027355e-06,
      "loss": 0.1829,
      "step": 9125
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.2715993159094072,
      "learning_rate": 1.0236444935160129e-06,
      "loss": 0.2051,
      "step": 9126
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.1894277803978377,
      "learning_rate": 1.022301140555787e-06,
      "loss": 0.1921,
      "step": 9127
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.8232291389339654,
      "learning_rate": 1.0209586221469336e-06,
      "loss": 0.1655,
      "step": 9128
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.000809028518031,
      "learning_rate": 1.0196169384142495e-06,
      "loss": 0.1736,
      "step": 9129
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.3175794609573996,
      "learning_rate": 1.0182760894824607e-06,
      "loss": 0.1829,
      "step": 9130
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.0766387834289857,
      "learning_rate": 1.0169360754762013e-06,
      "loss": 0.1985,
      "step": 9131
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.0633304511354664,
      "learning_rate": 1.0155968965200435e-06,
      "loss": 0.1494,
      "step": 9132
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.8440313630107106,
      "learning_rate": 1.014258552738473e-06,
      "loss": 0.1836,
      "step": 9133
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.8663431029529027,
      "learning_rate": 1.0129210442559011e-06,
      "loss": 0.187,
      "step": 9134
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.2418099461595213,
      "learning_rate": 1.0115843711966577e-06,
      "loss": 0.2192,
      "step": 9135
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.000887713966838,
      "learning_rate": 1.0102485336849998e-06,
      "loss": 0.1467,
      "step": 9136
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.2147692132316554,
      "learning_rate": 1.0089135318451026e-06,
      "loss": 0.1838,
      "step": 9137
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.8372830789699488,
      "learning_rate": 1.00757936580107e-06,
      "loss": 0.1646,
      "step": 9138
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.3508529800974514,
      "learning_rate": 1.0062460356769189e-06,
      "loss": 0.2129,
      "step": 9139
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.1334506570159313,
      "learning_rate": 1.0049135415965926e-06,
      "loss": 0.238,
      "step": 9140
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.1791342112083423,
      "learning_rate": 1.0035818836839618e-06,
      "loss": 0.1941,
      "step": 9141
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.022060899721312,
      "learning_rate": 1.0022510620628145e-06,
      "loss": 0.1526,
      "step": 9142
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.0196773330569826,
      "learning_rate": 1.000921076856859e-06,
      "loss": 0.1617,
      "step": 9143
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.9578837341574948,
      "learning_rate": 9.995919281897304e-07,
      "loss": 0.1611,
      "step": 9144
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.088438624194711,
      "learning_rate": 9.982636161849824e-07,
      "loss": 0.1711,
      "step": 9145
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.1362811731754947,
      "learning_rate": 9.969361409660927e-07,
      "loss": 0.1735,
      "step": 9146
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.0690123462668595,
      "learning_rate": 9.95609502656465e-07,
      "loss": 0.2126,
      "step": 9147
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.98779689031093,
      "learning_rate": 9.942837013794149e-07,
      "loss": 0.2227,
      "step": 9148
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.0266552178019595,
      "learning_rate": 9.929587372581917e-07,
      "loss": 0.199,
      "step": 9149
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.1847912213061584,
      "learning_rate": 9.916346104159602e-07,
      "loss": 0.2295,
      "step": 9150
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.887306624152074,
      "learning_rate": 9.903113209758098e-07,
      "loss": 0.1948,
      "step": 9151
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.7602537164089957,
      "learning_rate": 9.889888690607485e-07,
      "loss": 0.1482,
      "step": 9152
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.3678416480282154,
      "learning_rate": 9.876672547937117e-07,
      "loss": 0.2019,
      "step": 9153
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.2703590599552457,
      "learning_rate": 9.86346478297552e-07,
      "loss": 0.1775,
      "step": 9154
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.1358142505929028,
      "learning_rate": 9.85026539695051e-07,
      "loss": 0.2197,
      "step": 9155
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.7396207865197868,
      "learning_rate": 9.83707439108903e-07,
      "loss": 0.147,
      "step": 9156
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.028987945060404,
      "learning_rate": 9.8238917666173e-07,
      "loss": 0.1912,
      "step": 9157
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.149515809920645,
      "learning_rate": 9.810717524760783e-07,
      "loss": 0.1794,
      "step": 9158
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.7457498984124789,
      "learning_rate": 9.79755166674411e-07,
      "loss": 0.1536,
      "step": 9159
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.9626124154176823,
      "learning_rate": 9.784394193791169e-07,
      "loss": 0.1647,
      "step": 9160
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.145075969268152,
      "learning_rate": 9.77124510712505e-07,
      "loss": 0.1775,
      "step": 9161
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.6714663479493495,
      "learning_rate": 9.758104407968073e-07,
      "loss": 0.2219,
      "step": 9162
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.1164793746247463,
      "learning_rate": 9.74497209754175e-07,
      "loss": 0.2067,
      "step": 9163
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.1149659127229676,
      "learning_rate": 9.731848177066905e-07,
      "loss": 0.2019,
      "step": 9164
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.773832807796695,
      "learning_rate": 9.718732647763419e-07,
      "loss": 0.1663,
      "step": 9165
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.6322924689379776,
      "learning_rate": 9.705625510850557e-07,
      "loss": 0.1281,
      "step": 9166
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.2144080742084067,
      "learning_rate": 9.692526767546727e-07,
      "loss": 0.1785,
      "step": 9167
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.419448264841267,
      "learning_rate": 9.679436419069555e-07,
      "loss": 0.2234,
      "step": 9168
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.9328377698431383,
      "learning_rate": 9.666354466635908e-07,
      "loss": 0.198,
      "step": 9169
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.846319788089335,
      "learning_rate": 9.653280911461837e-07,
      "loss": 0.1677,
      "step": 9170
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.0657111910651937,
      "learning_rate": 9.640215754762638e-07,
      "loss": 0.1821,
      "step": 9171
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.9899869450852394,
      "learning_rate": 9.627158997752883e-07,
      "loss": 0.2046,
      "step": 9172
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.3342710120887413,
      "learning_rate": 9.614110641646235e-07,
      "loss": 0.1978,
      "step": 9173
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.106029565829106,
      "learning_rate": 9.601070687655667e-07,
      "loss": 0.2129,
      "step": 9174
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.516195744448746,
      "learning_rate": 9.588039136993366e-07,
      "loss": 0.1907,
      "step": 9175
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.9800105441070956,
      "learning_rate": 9.575015990870717e-07,
      "loss": 0.17,
      "step": 9176
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.7518384815943016,
      "learning_rate": 9.562001250498333e-07,
      "loss": 0.1565,
      "step": 9177
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.018737158425432,
      "learning_rate": 9.54899491708603e-07,
      "loss": 0.1807,
      "step": 9178
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.0333530479449347,
      "learning_rate": 9.535996991842855e-07,
      "loss": 0.2122,
      "step": 9179
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.1680951580101078,
      "learning_rate": 9.523007475977064e-07,
      "loss": 0.2078,
      "step": 9180
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.7276233042799924,
      "learning_rate": 9.51002637069619e-07,
      "loss": 0.1543,
      "step": 9181
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.827877796227608,
      "learning_rate": 9.49705367720688e-07,
      "loss": 0.1653,
      "step": 9182
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.264757054213485,
      "learning_rate": 9.484089396715057e-07,
      "loss": 0.1912,
      "step": 9183
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.0148153638703516,
      "learning_rate": 9.47113353042588e-07,
      "loss": 0.2388,
      "step": 9184
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.893736982448097,
      "learning_rate": 9.458186079543697e-07,
      "loss": 0.2261,
      "step": 9185
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.873531932072915,
      "learning_rate": 9.445247045272077e-07,
      "loss": 0.1545,
      "step": 9186
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.0188921357951006,
      "learning_rate": 9.432316428813826e-07,
      "loss": 0.229,
      "step": 9187
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.9692965317390172,
      "learning_rate": 9.419394231370926e-07,
      "loss": 0.1677,
      "step": 9188
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.6261441726337023,
      "learning_rate": 9.406480454144617e-07,
      "loss": 0.1217,
      "step": 9189
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.173197350764491,
      "learning_rate": 9.393575098335339e-07,
      "loss": 0.2134,
      "step": 9190
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.2811273267795644,
      "learning_rate": 9.380678165142732e-07,
      "loss": 0.2046,
      "step": 9191
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.0854767806982775,
      "learning_rate": 9.367789655765703e-07,
      "loss": 0.1819,
      "step": 9192
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.9735096795799547,
      "learning_rate": 9.354909571402349e-07,
      "loss": 0.1693,
      "step": 9193
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.8378618072952224,
      "learning_rate": 9.342037913249957e-07,
      "loss": 0.155,
      "step": 9194
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.862618926227194,
      "learning_rate": 9.32917468250506e-07,
      "loss": 0.1533,
      "step": 9195
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.7743952227870379,
      "learning_rate": 9.316319880363411e-07,
      "loss": 0.1664,
      "step": 9196
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.8390294760167205,
      "learning_rate": 9.303473508019944e-07,
      "loss": 0.1487,
      "step": 9197
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.8062869904306729,
      "learning_rate": 9.290635566668893e-07,
      "loss": 0.1674,
      "step": 9198
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.159750135554323,
      "learning_rate": 9.277806057503592e-07,
      "loss": 0.1489,
      "step": 9199
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.8486566700931728,
      "learning_rate": 9.264984981716663e-07,
      "loss": 0.165,
      "step": 9200
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.8563007017215403,
      "learning_rate": 9.252172340499943e-07,
      "loss": 0.1674,
      "step": 9201
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.3066248682873454,
      "learning_rate": 9.23936813504448e-07,
      "loss": 0.207,
      "step": 9202
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.8800114837137214,
      "learning_rate": 9.22657236654051e-07,
      "loss": 0.1842,
      "step": 9203
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.220957552069896,
      "learning_rate": 9.213785036177525e-07,
      "loss": 0.2114,
      "step": 9204
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.1686170046350015,
      "learning_rate": 9.201006145144198e-07,
      "loss": 0.1814,
      "step": 9205
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.8427129198887826,
      "learning_rate": 9.188235694628445e-07,
      "loss": 0.1863,
      "step": 9206
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.9978389264077137,
      "learning_rate": 9.175473685817371e-07,
      "loss": 0.2112,
      "step": 9207
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.8679072029624537,
      "learning_rate": 9.162720119897306e-07,
      "loss": 0.1636,
      "step": 9208
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.0560320559269103,
      "learning_rate": 9.149974998053823e-07,
      "loss": 0.1672,
      "step": 9209
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.360697392366747,
      "learning_rate": 9.137238321471675e-07,
      "loss": 0.2366,
      "step": 9210
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.6383845835630924,
      "learning_rate": 9.124510091334849e-07,
      "loss": 0.2319,
      "step": 9211
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.227887844704817,
      "learning_rate": 9.111790308826529e-07,
      "loss": 0.2126,
      "step": 9212
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.027138861664731,
      "learning_rate": 9.099078975129116e-07,
      "loss": 0.1606,
      "step": 9213
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.7073646557730418,
      "learning_rate": 9.086376091424243e-07,
      "loss": 0.1646,
      "step": 9214
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.0862201681997843,
      "learning_rate": 9.073681658892775e-07,
      "loss": 0.1978,
      "step": 9215
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.9697136044304007,
      "learning_rate": 9.060995678714712e-07,
      "loss": 0.1765,
      "step": 9216
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.1353347054443623,
      "learning_rate": 9.048318152069346e-07,
      "loss": 0.2456,
      "step": 9217
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.233481998051491,
      "learning_rate": 9.035649080135167e-07,
      "loss": 0.1809,
      "step": 9218
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.226766074573422,
      "learning_rate": 9.022988464089888e-07,
      "loss": 0.2,
      "step": 9219
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.1324568389537277,
      "learning_rate": 9.010336305110345e-07,
      "loss": 0.1865,
      "step": 9220
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.145532290867972,
      "learning_rate": 8.997692604372743e-07,
      "loss": 0.2007,
      "step": 9221
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.8737553751603362,
      "learning_rate": 8.985057363052374e-07,
      "loss": 0.1962,
      "step": 9222
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.1663375929894775,
      "learning_rate": 8.972430582323788e-07,
      "loss": 0.1958,
      "step": 9223
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.335728780319002,
      "learning_rate": 8.959812263360779e-07,
      "loss": 0.2273,
      "step": 9224
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.099721483877859,
      "learning_rate": 8.947202407336286e-07,
      "loss": 0.2073,
      "step": 9225
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.8268085920877417,
      "learning_rate": 8.934601015422506e-07,
      "loss": 0.1812,
      "step": 9226
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.1635142105606975,
      "learning_rate": 8.92200808879089e-07,
      "loss": 0.2131,
      "step": 9227
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.237442007189189,
      "learning_rate": 8.909423628611991e-07,
      "loss": 0.198,
      "step": 9228
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.209002964349039,
      "learning_rate": 8.896847636055672e-07,
      "loss": 0.2695,
      "step": 9229
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.993451312061583,
      "learning_rate": 8.884280112290977e-07,
      "loss": 0.2046,
      "step": 9230
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.2442778885406187,
      "learning_rate": 8.871721058486149e-07,
      "loss": 0.2119,
      "step": 9231
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.534221740143823,
      "learning_rate": 8.859170475808665e-07,
      "loss": 0.2756,
      "step": 9232
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.092644464711049,
      "learning_rate": 8.846628365425203e-07,
      "loss": 0.2095,
      "step": 9233
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.237551965756002,
      "learning_rate": 8.83409472850163e-07,
      "loss": 0.2,
      "step": 9234
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.9606952062459067,
      "learning_rate": 8.821569566203103e-07,
      "loss": 0.1567,
      "step": 9235
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.317408681276493,
      "learning_rate": 8.809052879693925e-07,
      "loss": 0.2085,
      "step": 9236
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.0131495887277904,
      "learning_rate": 8.796544670137574e-07,
      "loss": 0.1528,
      "step": 9237
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.057349711055123,
      "learning_rate": 8.784044938696856e-07,
      "loss": 0.1946,
      "step": 9238
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.0205808801539957,
      "learning_rate": 8.771553686533684e-07,
      "loss": 0.1727,
      "step": 9239
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.6420223311655273,
      "learning_rate": 8.759070914809253e-07,
      "loss": 0.1516,
      "step": 9240
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.0120544468470216,
      "learning_rate": 8.746596624683922e-07,
      "loss": 0.1921,
      "step": 9241
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.9484250361462396,
      "learning_rate": 8.734130817317277e-07,
      "loss": 0.1597,
      "step": 9242
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.0705609007982493,
      "learning_rate": 8.721673493868111e-07,
      "loss": 0.2249,
      "step": 9243
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.0605364896494236,
      "learning_rate": 8.709224655494475e-07,
      "loss": 0.1824,
      "step": 9244
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.4184967845437675,
      "learning_rate": 8.696784303353534e-07,
      "loss": 0.2117,
      "step": 9245
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.0789898620777576,
      "learning_rate": 8.684352438601762e-07,
      "loss": 0.1929,
      "step": 9246
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.2602422100515396,
      "learning_rate": 8.671929062394802e-07,
      "loss": 0.2139,
      "step": 9247
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.9137459715967948,
      "learning_rate": 8.659514175887495e-07,
      "loss": 0.1505,
      "step": 9248
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.8101778049839554,
      "learning_rate": 8.647107780233921e-07,
      "loss": 0.1582,
      "step": 9249
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.032193093727047,
      "learning_rate": 8.634709876587344e-07,
      "loss": 0.1965,
      "step": 9250
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.515543222680265,
      "learning_rate": 8.622320466100242e-07,
      "loss": 0.281,
      "step": 9251
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.330950347935421,
      "learning_rate": 8.60993954992434e-07,
      "loss": 0.2205,
      "step": 9252
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.9460694595549168,
      "learning_rate": 8.59756712921056e-07,
      "loss": 0.1821,
      "step": 9253
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.276455927288235,
      "learning_rate": 8.585203205108949e-07,
      "loss": 0.1663,
      "step": 9254
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.3019541933273904,
      "learning_rate": 8.572847778768912e-07,
      "loss": 0.176,
      "step": 9255
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.8202817258752069,
      "learning_rate": 8.560500851338949e-07,
      "loss": 0.1794,
      "step": 9256
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.9858929667701497,
      "learning_rate": 8.548162423966832e-07,
      "loss": 0.1718,
      "step": 9257
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.265227170716061,
      "learning_rate": 8.53583249779949e-07,
      "loss": 0.215,
      "step": 9258
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.191175262956175,
      "learning_rate": 8.523511073983127e-07,
      "loss": 0.1753,
      "step": 9259
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.0377379874477244,
      "learning_rate": 8.511198153663069e-07,
      "loss": 0.1843,
      "step": 9260
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.0036801304839575,
      "learning_rate": 8.498893737983982e-07,
      "loss": 0.1453,
      "step": 9261
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.8346781960134446,
      "learning_rate": 8.486597828089594e-07,
      "loss": 0.1729,
      "step": 9262
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.159946316104897,
      "learning_rate": 8.474310425122923e-07,
      "loss": 0.2249,
      "step": 9263
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.2979126149170868,
      "learning_rate": 8.462031530226211e-07,
      "loss": 0.1613,
      "step": 9264
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.033291441707344,
      "learning_rate": 8.449761144540869e-07,
      "loss": 0.1571,
      "step": 9265
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.2845185660168443,
      "learning_rate": 8.437499269207538e-07,
      "loss": 0.1938,
      "step": 9266
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.98898883892938,
      "learning_rate": 8.425245905366052e-07,
      "loss": 0.1372,
      "step": 9267
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.217625915676039,
      "learning_rate": 8.413001054155467e-07,
      "loss": 0.2369,
      "step": 9268
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.1443765736651943,
      "learning_rate": 8.400764716714016e-07,
      "loss": 0.176,
      "step": 9269
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.9395722211372486,
      "learning_rate": 8.388536894179234e-07,
      "loss": 0.1899,
      "step": 9270
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.556459464058939,
      "learning_rate": 8.376317587687721e-07,
      "loss": 0.2319,
      "step": 9271
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.052070346076939,
      "learning_rate": 8.364106798375416e-07,
      "loss": 0.2031,
      "step": 9272
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.9989916692341234,
      "learning_rate": 8.351904527377397e-07,
      "loss": 0.1504,
      "step": 9273
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.8177548501251675,
      "learning_rate": 8.339710775827958e-07,
      "loss": 0.1628,
      "step": 9274
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.0131743634727655,
      "learning_rate": 8.327525544860626e-07,
      "loss": 0.2141,
      "step": 9275
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.4026838050655996,
      "learning_rate": 8.315348835608095e-07,
      "loss": 0.2205,
      "step": 9276
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.268729448311132,
      "learning_rate": 8.303180649202303e-07,
      "loss": 0.2045,
      "step": 9277
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.1393897324783064,
      "learning_rate": 8.291020986774412e-07,
      "loss": 0.2302,
      "step": 9278
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.1185726860263228,
      "learning_rate": 8.278869849454718e-07,
      "loss": 0.2136,
      "step": 9279
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.9986765339135826,
      "learning_rate": 8.266727238372763e-07,
      "loss": 0.1724,
      "step": 9280
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.062234840040308,
      "learning_rate": 8.254593154657353e-07,
      "loss": 0.2017,
      "step": 9281
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.526196390126751,
      "learning_rate": 8.242467599436432e-07,
      "loss": 0.2395,
      "step": 9282
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.1741530639734132,
      "learning_rate": 8.230350573837165e-07,
      "loss": 0.2454,
      "step": 9283
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.8040665663580175,
      "learning_rate": 8.218242078985917e-07,
      "loss": 0.1559,
      "step": 9284
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.7390208458790954,
      "learning_rate": 8.206142116008298e-07,
      "loss": 0.1509,
      "step": 9285
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.038275257417152,
      "learning_rate": 8.194050686029065e-07,
      "loss": 0.1804,
      "step": 9286
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.7222283828366536,
      "learning_rate": 8.181967790172274e-07,
      "loss": 0.113,
      "step": 9287
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.9313797612945327,
      "learning_rate": 8.16989342956106e-07,
      "loss": 0.1793,
      "step": 9288
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.228783354639993,
      "learning_rate": 8.157827605317892e-07,
      "loss": 0.179,
      "step": 9289
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.8042038783768013,
      "learning_rate": 8.145770318564361e-07,
      "loss": 0.2058,
      "step": 9290
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.6150870628116856,
      "learning_rate": 8.133721570421305e-07,
      "loss": 0.1892,
      "step": 9291
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.3361680674071588,
      "learning_rate": 8.121681362008737e-07,
      "loss": 0.2827,
      "step": 9292
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.0033357183522127,
      "learning_rate": 8.109649694445898e-07,
      "loss": 0.2493,
      "step": 9293
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.170161032597862,
      "learning_rate": 8.097626568851224e-07,
      "loss": 0.1553,
      "step": 9294
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.9504308385050964,
      "learning_rate": 8.085611986342423e-07,
      "loss": 0.2256,
      "step": 9295
    },
    {
      "epoch": 2.63,
      "grad_norm": 2.3122810521840584,
      "learning_rate": 8.073605948036267e-07,
      "loss": 0.2256,
      "step": 9296
    },
    {
      "epoch": 2.63,
      "grad_norm": 2.239148294340403,
      "learning_rate": 8.061608455048841e-07,
      "loss": 0.167,
      "step": 9297
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.9690920418631352,
      "learning_rate": 8.049619508495454e-07,
      "loss": 0.1741,
      "step": 9298
    },
    {
      "epoch": 2.63,
      "grad_norm": 2.0529123406362237,
      "learning_rate": 8.037639109490524e-07,
      "loss": 0.1761,
      "step": 9299
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.7821372864138012,
      "learning_rate": 8.025667259147773e-07,
      "loss": 0.1527,
      "step": 9300
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.790106020376895,
      "learning_rate": 8.013703958580044e-07,
      "loss": 0.1611,
      "step": 9301
    },
    {
      "epoch": 2.63,
      "grad_norm": 2.1576436075457264,
      "learning_rate": 8.001749208899445e-07,
      "loss": 0.186,
      "step": 9302
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.9307112009226772,
      "learning_rate": 7.989803011217256e-07,
      "loss": 0.1868,
      "step": 9303
    },
    {
      "epoch": 2.63,
      "grad_norm": 2.241613430367479,
      "learning_rate": 7.977865366644011e-07,
      "loss": 0.2283,
      "step": 9304
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.9121575896034937,
      "learning_rate": 7.965936276289366e-07,
      "loss": 0.1699,
      "step": 9305
    },
    {
      "epoch": 2.63,
      "grad_norm": 2.189160505404498,
      "learning_rate": 7.954015741262255e-07,
      "loss": 0.1771,
      "step": 9306
    },
    {
      "epoch": 2.63,
      "grad_norm": 2.02486574377325,
      "learning_rate": 7.942103762670783e-07,
      "loss": 0.16,
      "step": 9307
    },
    {
      "epoch": 2.63,
      "grad_norm": 2.045608322578948,
      "learning_rate": 7.930200341622274e-07,
      "loss": 0.2021,
      "step": 9308
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.9688929239246193,
      "learning_rate": 7.918305479223243e-07,
      "loss": 0.1914,
      "step": 9309
    },
    {
      "epoch": 2.63,
      "grad_norm": 2.2879052656277254,
      "learning_rate": 7.906419176579416e-07,
      "loss": 0.1912,
      "step": 9310
    },
    {
      "epoch": 2.63,
      "grad_norm": 2.1195456048498365,
      "learning_rate": 7.8945414347957e-07,
      "loss": 0.1669,
      "step": 9311
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.699741278143081,
      "learning_rate": 7.882672254976298e-07,
      "loss": 0.1389,
      "step": 9312
    },
    {
      "epoch": 2.63,
      "grad_norm": 2.2101839057376043,
      "learning_rate": 7.870811638224485e-07,
      "loss": 0.1802,
      "step": 9313
    },
    {
      "epoch": 2.63,
      "grad_norm": 2.0806499812129995,
      "learning_rate": 7.8589595856428e-07,
      "loss": 0.1819,
      "step": 9314
    },
    {
      "epoch": 2.63,
      "grad_norm": 2.0187099320635644,
      "learning_rate": 7.847116098333029e-07,
      "loss": 0.2151,
      "step": 9315
    },
    {
      "epoch": 2.63,
      "grad_norm": 2.1930369762556863,
      "learning_rate": 7.835281177396126e-07,
      "loss": 0.1865,
      "step": 9316
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.7056629033861155,
      "learning_rate": 7.823454823932186e-07,
      "loss": 0.167,
      "step": 9317
    },
    {
      "epoch": 2.63,
      "grad_norm": 2.130767355335622,
      "learning_rate": 7.811637039040621e-07,
      "loss": 0.176,
      "step": 9318
    },
    {
      "epoch": 2.63,
      "grad_norm": 2.146524434614594,
      "learning_rate": 7.799827823819972e-07,
      "loss": 0.182,
      "step": 9319
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.6214902832446423,
      "learning_rate": 7.788027179367997e-07,
      "loss": 0.1312,
      "step": 9320
    },
    {
      "epoch": 2.63,
      "grad_norm": 2.3744607682171286,
      "learning_rate": 7.776235106781704e-07,
      "loss": 0.2087,
      "step": 9321
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.9779444396314545,
      "learning_rate": 7.764451607157208e-07,
      "loss": 0.1802,
      "step": 9322
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.930084259003556,
      "learning_rate": 7.752676681589899e-07,
      "loss": 0.175,
      "step": 9323
    },
    {
      "epoch": 2.63,
      "grad_norm": 2.510061750957616,
      "learning_rate": 7.740910331174378e-07,
      "loss": 0.2646,
      "step": 9324
    },
    {
      "epoch": 2.63,
      "grad_norm": 2.0330292780359924,
      "learning_rate": 7.729152557004405e-07,
      "loss": 0.1642,
      "step": 9325
    },
    {
      "epoch": 2.63,
      "grad_norm": 2.2320133772453032,
      "learning_rate": 7.717403360172959e-07,
      "loss": 0.2065,
      "step": 9326
    },
    {
      "epoch": 2.63,
      "grad_norm": 2.1682559320021673,
      "learning_rate": 7.705662741772235e-07,
      "loss": 0.2234,
      "step": 9327
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.987918341348484,
      "learning_rate": 7.693930702893626e-07,
      "loss": 0.1968,
      "step": 9328
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.9639099123390007,
      "learning_rate": 7.682207244627704e-07,
      "loss": 0.1619,
      "step": 9329
    },
    {
      "epoch": 2.63,
      "grad_norm": 2.0833140145874682,
      "learning_rate": 7.670492368064275e-07,
      "loss": 0.1704,
      "step": 9330
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.1433024925739748,
      "learning_rate": 7.658786074292312e-07,
      "loss": 0.2214,
      "step": 9331
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.530997796991467,
      "learning_rate": 7.647088364400046e-07,
      "loss": 0.2346,
      "step": 9332
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.9445498675583754,
      "learning_rate": 7.635399239474872e-07,
      "loss": 0.1597,
      "step": 9333
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.021614729153346,
      "learning_rate": 7.623718700603356e-07,
      "loss": 0.1793,
      "step": 9334
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.5254718790300137,
      "learning_rate": 7.612046748871327e-07,
      "loss": 0.2043,
      "step": 9335
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.111458097384965,
      "learning_rate": 7.600383385363797e-07,
      "loss": 0.2051,
      "step": 9336
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.419600448283422,
      "learning_rate": 7.58872861116493e-07,
      "loss": 0.2092,
      "step": 9337
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.9012523941596053,
      "learning_rate": 7.577082427358207e-07,
      "loss": 0.1965,
      "step": 9338
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.01973227695898,
      "learning_rate": 7.56544483502617e-07,
      "loss": 0.1685,
      "step": 9339
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.28632521102436,
      "learning_rate": 7.553815835250644e-07,
      "loss": 0.2153,
      "step": 9340
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.9986262379040158,
      "learning_rate": 7.542195429112664e-07,
      "loss": 0.2043,
      "step": 9341
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.9819111554218716,
      "learning_rate": 7.530583617692433e-07,
      "loss": 0.1697,
      "step": 9342
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.987492927769711,
      "learning_rate": 7.518980402069354e-07,
      "loss": 0.2047,
      "step": 9343
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.1035312373190536,
      "learning_rate": 7.507385783322052e-07,
      "loss": 0.1372,
      "step": 9344
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.2041988973944906,
      "learning_rate": 7.495799762528333e-07,
      "loss": 0.2272,
      "step": 9345
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.964234512717421,
      "learning_rate": 7.484222340765235e-07,
      "loss": 0.2117,
      "step": 9346
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.9125437076468685,
      "learning_rate": 7.472653519108952e-07,
      "loss": 0.199,
      "step": 9347
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.2994035557285395,
      "learning_rate": 7.461093298634892e-07,
      "loss": 0.2056,
      "step": 9348
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.9021883566478446,
      "learning_rate": 7.449541680417704e-07,
      "loss": 0.1655,
      "step": 9349
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.883497625105576,
      "learning_rate": 7.437998665531221e-07,
      "loss": 0.183,
      "step": 9350
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.2757152330675985,
      "learning_rate": 7.426464255048393e-07,
      "loss": 0.2336,
      "step": 9351
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.6738476971835592,
      "learning_rate": 7.414938450041497e-07,
      "loss": 0.1421,
      "step": 9352
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.3384903220513875,
      "learning_rate": 7.403421251581933e-07,
      "loss": 0.2156,
      "step": 9353
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.0795190267012704,
      "learning_rate": 7.391912660740319e-07,
      "loss": 0.1948,
      "step": 9354
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.1596667141006245,
      "learning_rate": 7.38041267858649e-07,
      "loss": 0.1836,
      "step": 9355
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.2529079095426825,
      "learning_rate": 7.368921306189447e-07,
      "loss": 0.2209,
      "step": 9356
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.439196244441858,
      "learning_rate": 7.35743854461739e-07,
      "loss": 0.1515,
      "step": 9357
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.2545790689710112,
      "learning_rate": 7.345964394937788e-07,
      "loss": 0.2751,
      "step": 9358
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.1125756648079195,
      "learning_rate": 7.334498858217231e-07,
      "loss": 0.1615,
      "step": 9359
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.8317392127914796,
      "learning_rate": 7.323041935521502e-07,
      "loss": 0.165,
      "step": 9360
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.3447220043794332,
      "learning_rate": 7.31159362791567e-07,
      "loss": 0.2366,
      "step": 9361
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.8767365768837219,
      "learning_rate": 7.300153936463927e-07,
      "loss": 0.1553,
      "step": 9362
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.7478815603426516,
      "learning_rate": 7.288722862229691e-07,
      "loss": 0.1429,
      "step": 9363
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.551491938262081,
      "learning_rate": 7.277300406275567e-07,
      "loss": 0.2427,
      "step": 9364
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.8638663647992524,
      "learning_rate": 7.265886569663372e-07,
      "loss": 0.1539,
      "step": 9365
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.9156812939129966,
      "learning_rate": 7.254481353454102e-07,
      "loss": 0.1532,
      "step": 9366
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5110671615237783,
      "learning_rate": 7.243084758708007e-07,
      "loss": 0.1375,
      "step": 9367
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.086193024367188,
      "learning_rate": 7.231696786484443e-07,
      "loss": 0.1877,
      "step": 9368
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.9819529932075044,
      "learning_rate": 7.22031743784205e-07,
      "loss": 0.2085,
      "step": 9369
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.083293448116741,
      "learning_rate": 7.208946713838638e-07,
      "loss": 0.1865,
      "step": 9370
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.364017419330158,
      "learning_rate": 7.197584615531184e-07,
      "loss": 0.1877,
      "step": 9371
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.1334403869090726,
      "learning_rate": 7.186231143975908e-07,
      "loss": 0.2021,
      "step": 9372
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.090074306569928,
      "learning_rate": 7.17488630022819e-07,
      "loss": 0.1469,
      "step": 9373
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.9234530588321845,
      "learning_rate": 7.16355008534263e-07,
      "loss": 0.1428,
      "step": 9374
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.618232610514562,
      "learning_rate": 7.152222500373052e-07,
      "loss": 0.2834,
      "step": 9375
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.351628733796816,
      "learning_rate": 7.140903546372446e-07,
      "loss": 0.2263,
      "step": 9376
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.4149749267790392,
      "learning_rate": 7.12959322439295e-07,
      "loss": 0.1831,
      "step": 9377
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.8539317450558888,
      "learning_rate": 7.118291535485999e-07,
      "loss": 0.1638,
      "step": 9378
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.24953888545964,
      "learning_rate": 7.106998480702165e-07,
      "loss": 0.1785,
      "step": 9379
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.7798731964379741,
      "learning_rate": 7.095714061091241e-07,
      "loss": 0.1577,
      "step": 9380
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.243861842039552,
      "learning_rate": 7.084438277702188e-07,
      "loss": 0.2219,
      "step": 9381
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.011344902298796,
      "learning_rate": 7.073171131583201e-07,
      "loss": 0.1943,
      "step": 9382
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.0718792701541333,
      "learning_rate": 7.06191262378163e-07,
      "loss": 0.2087,
      "step": 9383
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.3540201625163686,
      "learning_rate": 7.050662755344096e-07,
      "loss": 0.209,
      "step": 9384
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.0097049708950045,
      "learning_rate": 7.039421527316304e-07,
      "loss": 0.1619,
      "step": 9385
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.1142551133966143,
      "learning_rate": 7.028188940743275e-07,
      "loss": 0.1792,
      "step": 9386
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.1339899261431943,
      "learning_rate": 7.016964996669129e-07,
      "loss": 0.1841,
      "step": 9387
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.8601334922240749,
      "learning_rate": 7.005749696137254e-07,
      "loss": 0.1688,
      "step": 9388
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.099487929333852,
      "learning_rate": 6.994543040190183e-07,
      "loss": 0.2041,
      "step": 9389
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.1999938963712187,
      "learning_rate": 6.983345029869681e-07,
      "loss": 0.2148,
      "step": 9390
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.5166786786983235,
      "learning_rate": 6.972155666216684e-07,
      "loss": 0.2266,
      "step": 9391
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.5345255846761363,
      "learning_rate": 6.960974950271348e-07,
      "loss": 0.2131,
      "step": 9392
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.064160911205337,
      "learning_rate": 6.949802883073031e-07,
      "loss": 0.1785,
      "step": 9393
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.4013292266070034,
      "learning_rate": 6.938639465660213e-07,
      "loss": 0.2195,
      "step": 9394
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.4594761680033517,
      "learning_rate": 6.927484699070675e-07,
      "loss": 0.194,
      "step": 9395
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.9965920712657974,
      "learning_rate": 6.91633858434132e-07,
      "loss": 0.1948,
      "step": 9396
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.0814646630535365,
      "learning_rate": 6.905201122508299e-07,
      "loss": 0.168,
      "step": 9397
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.172052734906305,
      "learning_rate": 6.894072314606892e-07,
      "loss": 0.2113,
      "step": 9398
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.190598645006827,
      "learning_rate": 6.882952161671652e-07,
      "loss": 0.2195,
      "step": 9399
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.2975379220901897,
      "learning_rate": 6.871840664736251e-07,
      "loss": 0.2375,
      "step": 9400
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.0498190103866305,
      "learning_rate": 6.860737824833652e-07,
      "loss": 0.2163,
      "step": 9401
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.43507069399866,
      "learning_rate": 6.849643642995873e-07,
      "loss": 0.2402,
      "step": 9402
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.2132822596066903,
      "learning_rate": 6.83855812025429e-07,
      "loss": 0.1809,
      "step": 9403
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.133050368963143,
      "learning_rate": 6.827481257639346e-07,
      "loss": 0.1995,
      "step": 9404
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.122585435533304,
      "learning_rate": 6.816413056180748e-07,
      "loss": 0.1824,
      "step": 9405
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.240371007725225,
      "learning_rate": 6.805353516907376e-07,
      "loss": 0.2058,
      "step": 9406
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.3500081020105053,
      "learning_rate": 6.794302640847294e-07,
      "loss": 0.2253,
      "step": 9407
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.121819746036731,
      "learning_rate": 6.78326042902776e-07,
      "loss": 0.2195,
      "step": 9408
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.664895661244437,
      "learning_rate": 6.772226882475275e-07,
      "loss": 0.1824,
      "step": 9409
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.0362531276833016,
      "learning_rate": 6.761202002215506e-07,
      "loss": 0.1536,
      "step": 9410
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.9754798452545128,
      "learning_rate": 6.750185789273234e-07,
      "loss": 0.1792,
      "step": 9411
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.309800080329592,
      "learning_rate": 6.739178244672584e-07,
      "loss": 0.1614,
      "step": 9412
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.9761894567951495,
      "learning_rate": 6.72817936943676e-07,
      "loss": 0.1912,
      "step": 9413
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.0496023200664877,
      "learning_rate": 6.717189164588212e-07,
      "loss": 0.1747,
      "step": 9414
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.1564636571066864,
      "learning_rate": 6.706207631148564e-07,
      "loss": 0.2136,
      "step": 9415
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.8575500358579038,
      "learning_rate": 6.695234770138648e-07,
      "loss": 0.1521,
      "step": 9416
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.9241129491819136,
      "learning_rate": 6.684270582578455e-07,
      "loss": 0.1841,
      "step": 9417
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.7556684571879586,
      "learning_rate": 6.673315069487252e-07,
      "loss": 0.259,
      "step": 9418
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.1918218606486715,
      "learning_rate": 6.662368231883388e-07,
      "loss": 0.1936,
      "step": 9419
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.9800712454697984,
      "learning_rate": 6.65143007078447e-07,
      "loss": 0.2058,
      "step": 9420
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.446150635516572,
      "learning_rate": 6.640500587207333e-07,
      "loss": 0.2104,
      "step": 9421
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.209100778724458,
      "learning_rate": 6.629579782167928e-07,
      "loss": 0.2209,
      "step": 9422
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.315768300628363,
      "learning_rate": 6.618667656681444e-07,
      "loss": 0.23,
      "step": 9423
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.8211496878813636,
      "learning_rate": 6.607764211762247e-07,
      "loss": 0.1847,
      "step": 9424
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.8161987094820542,
      "learning_rate": 6.596869448423903e-07,
      "loss": 0.1719,
      "step": 9425
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.266156281983945,
      "learning_rate": 6.585983367679171e-07,
      "loss": 0.2307,
      "step": 9426
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.419325587758742,
      "learning_rate": 6.57510597054003e-07,
      "loss": 0.2217,
      "step": 9427
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.6123683783489144,
      "learning_rate": 6.564237258017558e-07,
      "loss": 0.2253,
      "step": 9428
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.9805184095791646,
      "learning_rate": 6.55337723112216e-07,
      "loss": 0.1499,
      "step": 9429
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.9761050756014968,
      "learning_rate": 6.542525890863338e-07,
      "loss": 0.1738,
      "step": 9430
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.9381770694078309,
      "learning_rate": 6.531683238249809e-07,
      "loss": 0.1672,
      "step": 9431
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.888885482275301,
      "learning_rate": 6.520849274289498e-07,
      "loss": 0.1805,
      "step": 9432
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.823186649259057,
      "learning_rate": 6.510023999989501e-07,
      "loss": 0.1736,
      "step": 9433
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.8615584426148106,
      "learning_rate": 6.499207416356113e-07,
      "loss": 0.1572,
      "step": 9434
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.809133188813127,
      "learning_rate": 6.488399524394851e-07,
      "loss": 0.1611,
      "step": 9435
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.899702977075847,
      "learning_rate": 6.477600325110378e-07,
      "loss": 0.1626,
      "step": 9436
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.4935664779049973,
      "learning_rate": 6.466809819506548e-07,
      "loss": 0.261,
      "step": 9437
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.661027705965545,
      "learning_rate": 6.456028008586468e-07,
      "loss": 0.2412,
      "step": 9438
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.0391715541756166,
      "learning_rate": 6.445254893352381e-07,
      "loss": 0.2029,
      "step": 9439
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.9828955536424617,
      "learning_rate": 6.434490474805743e-07,
      "loss": 0.1613,
      "step": 9440
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.132039324060041,
      "learning_rate": 6.423734753947175e-07,
      "loss": 0.1914,
      "step": 9441
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.1773792616261014,
      "learning_rate": 6.412987731776532e-07,
      "loss": 0.2278,
      "step": 9442
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.1031090966468327,
      "learning_rate": 6.402249409292815e-07,
      "loss": 0.1619,
      "step": 9443
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.0883747051877317,
      "learning_rate": 6.391519787494282e-07,
      "loss": 0.2131,
      "step": 9444
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.006078026333115,
      "learning_rate": 6.380798867378291e-07,
      "loss": 0.209,
      "step": 9445
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.2060754312060507,
      "learning_rate": 6.370086649941465e-07,
      "loss": 0.2527,
      "step": 9446
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.4835832663321122,
      "learning_rate": 6.359383136179598e-07,
      "loss": 0.1685,
      "step": 9447
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.8935905111992193,
      "learning_rate": 6.348688327087671e-07,
      "loss": 0.1982,
      "step": 9448
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.2478172256610036,
      "learning_rate": 6.338002223659834e-07,
      "loss": 0.1885,
      "step": 9449
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.6697234901293432,
      "learning_rate": 6.327324826889469e-07,
      "loss": 0.129,
      "step": 9450
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.0396897191642904,
      "learning_rate": 6.316656137769095e-07,
      "loss": 0.1682,
      "step": 9451
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.1585140333292454,
      "learning_rate": 6.305996157290528e-07,
      "loss": 0.1562,
      "step": 9452
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.2184693653294976,
      "learning_rate": 6.295344886444632e-07,
      "loss": 0.2251,
      "step": 9453
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.017254475306312,
      "learning_rate": 6.284702326221537e-07,
      "loss": 0.1815,
      "step": 9454
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.8868622065326084,
      "learning_rate": 6.274068477610584e-07,
      "loss": 0.1621,
      "step": 9455
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.7162206648755414,
      "learning_rate": 6.263443341600284e-07,
      "loss": 0.1633,
      "step": 9456
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.208201391047842,
      "learning_rate": 6.252826919178278e-07,
      "loss": 0.1831,
      "step": 9457
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5766133833225144,
      "learning_rate": 6.242219211331512e-07,
      "loss": 0.1426,
      "step": 9458
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.9924395233150052,
      "learning_rate": 6.23162021904603e-07,
      "loss": 0.1749,
      "step": 9459
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.753297750623373,
      "learning_rate": 6.221029943307099e-07,
      "loss": 0.1722,
      "step": 9460
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.4295753662214974,
      "learning_rate": 6.210448385099177e-07,
      "loss": 0.2859,
      "step": 9461
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.340834421857068,
      "learning_rate": 6.199875545405898e-07,
      "loss": 0.2046,
      "step": 9462
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.8128358772022366,
      "learning_rate": 6.189311425210087e-07,
      "loss": 0.167,
      "step": 9463
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.0054746617943406,
      "learning_rate": 6.178756025493804e-07,
      "loss": 0.1855,
      "step": 9464
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.8830931872350112,
      "learning_rate": 6.16820934723823e-07,
      "loss": 0.1538,
      "step": 9465
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.765703473097823,
      "learning_rate": 6.157671391423769e-07,
      "loss": 0.1376,
      "step": 9466
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.300453715590132,
      "learning_rate": 6.147142159030017e-07,
      "loss": 0.178,
      "step": 9467
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.8929377996802472,
      "learning_rate": 6.136621651035756e-07,
      "loss": 0.1956,
      "step": 9468
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.090668461929228,
      "learning_rate": 6.126109868418951e-07,
      "loss": 0.2064,
      "step": 9469
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.5168430592140125,
      "learning_rate": 6.115606812156749e-07,
      "loss": 0.1987,
      "step": 9470
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.9063312743228276,
      "learning_rate": 6.105112483225495e-07,
      "loss": 0.214,
      "step": 9471
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.0897164813663616,
      "learning_rate": 6.094626882600751e-07,
      "loss": 0.2505,
      "step": 9472
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.2043381139140212,
      "learning_rate": 6.084150011257239e-07,
      "loss": 0.2146,
      "step": 9473
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.9133654174857033,
      "learning_rate": 6.073681870168813e-07,
      "loss": 0.1522,
      "step": 9474
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.6407088003010393,
      "learning_rate": 6.063222460308649e-07,
      "loss": 0.1519,
      "step": 9475
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.1853718794171337,
      "learning_rate": 6.052771782648981e-07,
      "loss": 0.2021,
      "step": 9476
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.0591407365964165,
      "learning_rate": 6.04232983816132e-07,
      "loss": 0.1829,
      "step": 9477
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.2314069495446898,
      "learning_rate": 6.031896627816314e-07,
      "loss": 0.2166,
      "step": 9478
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.21216917834901,
      "learning_rate": 6.021472152583818e-07,
      "loss": 0.2322,
      "step": 9479
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.342269562314909,
      "learning_rate": 6.01105641343287e-07,
      "loss": 0.2354,
      "step": 9480
    },
    {
      "epoch": 2.68,
      "grad_norm": 3.3323742888088344,
      "learning_rate": 6.000649411331705e-07,
      "loss": 0.2214,
      "step": 9481
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.013755940702662,
      "learning_rate": 5.99025114724775e-07,
      "loss": 0.1892,
      "step": 9482
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.5420428419259173,
      "learning_rate": 5.979861622147587e-07,
      "loss": 0.2004,
      "step": 9483
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.1271529615412237,
      "learning_rate": 5.969480836997032e-07,
      "loss": 0.207,
      "step": 9484
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.7172487008922224,
      "learning_rate": 5.959108792761048e-07,
      "loss": 0.1399,
      "step": 9485
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.2279823170533493,
      "learning_rate": 5.948745490403806e-07,
      "loss": 0.2024,
      "step": 9486
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.288629846663494,
      "learning_rate": 5.938390930888671e-07,
      "loss": 0.2466,
      "step": 9487
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.18713675066617,
      "learning_rate": 5.92804511517815e-07,
      "loss": 0.2493,
      "step": 9488
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5298255533371838,
      "learning_rate": 5.917708044234017e-07,
      "loss": 0.1284,
      "step": 9489
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.118671904923738,
      "learning_rate": 5.907379719017181e-07,
      "loss": 0.1958,
      "step": 9490
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.2536503092437936,
      "learning_rate": 5.897060140487709e-07,
      "loss": 0.2283,
      "step": 9491
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.164897737243496,
      "learning_rate": 5.886749309604922e-07,
      "loss": 0.207,
      "step": 9492
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.064491489206781,
      "learning_rate": 5.876447227327298e-07,
      "loss": 0.1941,
      "step": 9493
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.8487465963148255,
      "learning_rate": 5.866153894612492e-07,
      "loss": 0.166,
      "step": 9494
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.1578160536764552,
      "learning_rate": 5.855869312417362e-07,
      "loss": 0.2285,
      "step": 9495
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.0409267163842473,
      "learning_rate": 5.845593481697931e-07,
      "loss": 0.1704,
      "step": 9496
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.007426294574285,
      "learning_rate": 5.835326403409414e-07,
      "loss": 0.2313,
      "step": 9497
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.5289749022998245,
      "learning_rate": 5.825068078506257e-07,
      "loss": 0.1746,
      "step": 9498
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.6116113042227447,
      "learning_rate": 5.814818507942055e-07,
      "loss": 0.2686,
      "step": 9499
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.24914543810552,
      "learning_rate": 5.804577692669533e-07,
      "loss": 0.17,
      "step": 9500
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.2578815038134628,
      "learning_rate": 5.794345633640718e-07,
      "loss": 0.219,
      "step": 9501
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.1003294002101356,
      "learning_rate": 5.784122331806751e-07,
      "loss": 0.1943,
      "step": 9502
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.764244153557264,
      "learning_rate": 5.77390778811796e-07,
      "loss": 0.1476,
      "step": 9503
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.9698807580171633,
      "learning_rate": 5.763702003523874e-07,
      "loss": 0.1631,
      "step": 9504
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.0601330373378115,
      "learning_rate": 5.753504978973212e-07,
      "loss": 0.1776,
      "step": 9505
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.9105581017105118,
      "learning_rate": 5.743316715413849e-07,
      "loss": 0.156,
      "step": 9506
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.6087983156550507,
      "learning_rate": 5.733137213792928e-07,
      "loss": 0.1243,
      "step": 9507
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.334907988056662,
      "learning_rate": 5.722966475056646e-07,
      "loss": 0.1948,
      "step": 9508
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.853266757147694,
      "learning_rate": 5.712804500150493e-07,
      "loss": 0.1664,
      "step": 9509
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.9801163130461246,
      "learning_rate": 5.702651290019112e-07,
      "loss": 0.1766,
      "step": 9510
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.7890394222405264,
      "learning_rate": 5.692506845606327e-07,
      "loss": 0.1658,
      "step": 9511
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.481336733837336,
      "learning_rate": 5.682371167855127e-07,
      "loss": 0.2166,
      "step": 9512
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.3602765361916287,
      "learning_rate": 5.672244257707738e-07,
      "loss": 0.208,
      "step": 9513
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.3400746268460595,
      "learning_rate": 5.662126116105504e-07,
      "loss": 0.1959,
      "step": 9514
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.03007668868566,
      "learning_rate": 5.65201674398902e-07,
      "loss": 0.189,
      "step": 9515
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.2703096764649517,
      "learning_rate": 5.641916142298043e-07,
      "loss": 0.2078,
      "step": 9516
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.155416764324972,
      "learning_rate": 5.631824311971456e-07,
      "loss": 0.162,
      "step": 9517
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.09097314215533,
      "learning_rate": 5.621741253947432e-07,
      "loss": 0.1818,
      "step": 9518
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.580640347763246,
      "learning_rate": 5.611666969163243e-07,
      "loss": 0.2244,
      "step": 9519
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.288357367228842,
      "learning_rate": 5.601601458555406e-07,
      "loss": 0.2666,
      "step": 9520
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.8386844511777547,
      "learning_rate": 5.591544723059561e-07,
      "loss": 0.1344,
      "step": 9521
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.123964556996071,
      "learning_rate": 5.581496763610594e-07,
      "loss": 0.1924,
      "step": 9522
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.213195871936022,
      "learning_rate": 5.571457581142514e-07,
      "loss": 0.2161,
      "step": 9523
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.180465590723732,
      "learning_rate": 5.561427176588586e-07,
      "loss": 0.2173,
      "step": 9524
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.0591562399649974,
      "learning_rate": 5.551405550881173e-07,
      "loss": 0.1868,
      "step": 9525
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.171389462067622,
      "learning_rate": 5.541392704951909e-07,
      "loss": 0.182,
      "step": 9526
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.240814355206638,
      "learning_rate": 5.53138863973155e-07,
      "loss": 0.1738,
      "step": 9527
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.969834812094462,
      "learning_rate": 5.521393356150062e-07,
      "loss": 0.2188,
      "step": 9528
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.8215133068937108,
      "learning_rate": 5.51140685513658e-07,
      "loss": 0.137,
      "step": 9529
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.0716516719873437,
      "learning_rate": 5.501429137619452e-07,
      "loss": 0.1876,
      "step": 9530
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.9102373796574166,
      "learning_rate": 5.491460204526156e-07,
      "loss": 0.1685,
      "step": 9531
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.0271095361947467,
      "learning_rate": 5.481500056783429e-07,
      "loss": 0.2412,
      "step": 9532
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.0777172647894435,
      "learning_rate": 5.471548695317131e-07,
      "loss": 0.1929,
      "step": 9533
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.7650330724435281,
      "learning_rate": 5.461606121052299e-07,
      "loss": 0.139,
      "step": 9534
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.403180966979003,
      "learning_rate": 5.451672334913216e-07,
      "loss": 0.2378,
      "step": 9535
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.067576405477721,
      "learning_rate": 5.441747337823289e-07,
      "loss": 0.2085,
      "step": 9536
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.0328762976130506,
      "learning_rate": 5.431831130705123e-07,
      "loss": 0.2046,
      "step": 9537
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.9842425842098177,
      "learning_rate": 5.421923714480537e-07,
      "loss": 0.1461,
      "step": 9538
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.3650694532268726,
      "learning_rate": 5.412025090070483e-07,
      "loss": 0.2408,
      "step": 9539
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.986016020054448,
      "learning_rate": 5.402135258395114e-07,
      "loss": 0.1665,
      "step": 9540
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.7634412247183378,
      "learning_rate": 5.392254220373816e-07,
      "loss": 0.1523,
      "step": 9541
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.377210082756363,
      "learning_rate": 5.382381976925044e-07,
      "loss": 0.1975,
      "step": 9542
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.2137589128575446,
      "learning_rate": 5.372518528966575e-07,
      "loss": 0.2007,
      "step": 9543
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.1559237564112617,
      "learning_rate": 5.362663877415252e-07,
      "loss": 0.2202,
      "step": 9544
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.0806658145939716,
      "learning_rate": 5.352818023187167e-07,
      "loss": 0.1636,
      "step": 9545
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.905426830862656,
      "learning_rate": 5.342980967197564e-07,
      "loss": 0.178,
      "step": 9546
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.286269773609108,
      "learning_rate": 5.33315271036089e-07,
      "loss": 0.2358,
      "step": 9547
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.2667074632913926,
      "learning_rate": 5.323333253590734e-07,
      "loss": 0.2505,
      "step": 9548
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.956214201446175,
      "learning_rate": 5.313522597799947e-07,
      "loss": 0.1667,
      "step": 9549
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.7511594394079455,
      "learning_rate": 5.303720743900475e-07,
      "loss": 0.2351,
      "step": 9550
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.5972433014227887,
      "learning_rate": 5.293927692803458e-07,
      "loss": 0.2361,
      "step": 9551
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.4627309735977625,
      "learning_rate": 5.284143445419288e-07,
      "loss": 0.2385,
      "step": 9552
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.0840721637999198,
      "learning_rate": 5.274368002657482e-07,
      "loss": 0.1619,
      "step": 9553
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.3757684522981837,
      "learning_rate": 5.264601365426736e-07,
      "loss": 0.2285,
      "step": 9554
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.089426543079111,
      "learning_rate": 5.254843534634934e-07,
      "loss": 0.2056,
      "step": 9555
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.7920312899012683,
      "learning_rate": 5.245094511189163e-07,
      "loss": 0.1602,
      "step": 9556
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.2597041529315454,
      "learning_rate": 5.235354295995665e-07,
      "loss": 0.2329,
      "step": 9557
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.7628916075371286,
      "learning_rate": 5.225622889959892e-07,
      "loss": 0.1407,
      "step": 9558
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.0367789666570935,
      "learning_rate": 5.215900293986431e-07,
      "loss": 0.1776,
      "step": 9559
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.226235271443943,
      "learning_rate": 5.206186508979083e-07,
      "loss": 0.2515,
      "step": 9560
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.9594714133276756,
      "learning_rate": 5.196481535840847e-07,
      "loss": 0.1655,
      "step": 9561
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.9336202188235905,
      "learning_rate": 5.186785375473869e-07,
      "loss": 0.1851,
      "step": 9562
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.1449601934263964,
      "learning_rate": 5.17709802877947e-07,
      "loss": 0.2112,
      "step": 9563
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.142708027815568,
      "learning_rate": 5.1674194966582e-07,
      "loss": 0.2229,
      "step": 9564
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.3320850768201242,
      "learning_rate": 5.157749780009735e-07,
      "loss": 0.1914,
      "step": 9565
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.8543638096025574,
      "learning_rate": 5.14808887973296e-07,
      "loss": 0.1857,
      "step": 9566
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.0464589608854493,
      "learning_rate": 5.138436796725942e-07,
      "loss": 0.2043,
      "step": 9567
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.7878185717340656,
      "learning_rate": 5.1287935318859e-07,
      "loss": 0.1522,
      "step": 9568
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.223988158281457,
      "learning_rate": 5.119159086109293e-07,
      "loss": 0.1879,
      "step": 9569
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.065786174718044,
      "learning_rate": 5.109533460291694e-07,
      "loss": 0.186,
      "step": 9570
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.6866864424554073,
      "learning_rate": 5.099916655327907e-07,
      "loss": 0.1442,
      "step": 9571
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.9593459318469422,
      "learning_rate": 5.090308672111866e-07,
      "loss": 0.1798,
      "step": 9572
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.0384585635167096,
      "learning_rate": 5.08070951153673e-07,
      "loss": 0.1736,
      "step": 9573
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.2374579557043894,
      "learning_rate": 5.07111917449481e-07,
      "loss": 0.189,
      "step": 9574
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.2033673801373435,
      "learning_rate": 5.061537661877636e-07,
      "loss": 0.2063,
      "step": 9575
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.8339688872199487,
      "learning_rate": 5.051964974575851e-07,
      "loss": 0.2129,
      "step": 9576
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.200767688278932,
      "learning_rate": 5.042401113479312e-07,
      "loss": 0.1768,
      "step": 9577
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.893432903520031,
      "learning_rate": 5.032846079477105e-07,
      "loss": 0.1536,
      "step": 9578
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.7149475923217508,
      "learning_rate": 5.02329987345741e-07,
      "loss": 0.2881,
      "step": 9579
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.9711315991349394,
      "learning_rate": 5.01376249630764e-07,
      "loss": 0.1814,
      "step": 9580
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.0504236127016986,
      "learning_rate": 5.004233948914383e-07,
      "loss": 0.1844,
      "step": 9581
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.2124942693641576,
      "learning_rate": 4.994714232163378e-07,
      "loss": 0.1926,
      "step": 9582
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.4048776724436034,
      "learning_rate": 4.98520334693956e-07,
      "loss": 0.1877,
      "step": 9583
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.9413490777072369,
      "learning_rate": 4.975701294127067e-07,
      "loss": 0.1356,
      "step": 9584
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.9455321599190292,
      "learning_rate": 4.966208074609158e-07,
      "loss": 0.1997,
      "step": 9585
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.9755508992847324,
      "learning_rate": 4.956723689268339e-07,
      "loss": 0.1881,
      "step": 9586
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.1845943467844795,
      "learning_rate": 4.947248138986249e-07,
      "loss": 0.2173,
      "step": 9587
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.8347377859339538,
      "learning_rate": 4.937781424643728e-07,
      "loss": 0.1368,
      "step": 9588
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.0897871011034876,
      "learning_rate": 4.928323547120772e-07,
      "loss": 0.1862,
      "step": 9589
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.174583912600107,
      "learning_rate": 4.918874507296578e-07,
      "loss": 0.1843,
      "step": 9590
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.068525230842122,
      "learning_rate": 4.909434306049487e-07,
      "loss": 0.1927,
      "step": 9591
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.9100514235453119,
      "learning_rate": 4.900002944257098e-07,
      "loss": 0.1531,
      "step": 9592
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.9320384242388786,
      "learning_rate": 4.890580422796087e-07,
      "loss": 0.2153,
      "step": 9593
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.3253996271371875,
      "learning_rate": 4.881166742542365e-07,
      "loss": 0.2043,
      "step": 9594
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.2399159944111013,
      "learning_rate": 4.871761904371019e-07,
      "loss": 0.2224,
      "step": 9595
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.4085631477804172,
      "learning_rate": 4.86236590915633e-07,
      "loss": 0.2207,
      "step": 9596
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.3926002318256354,
      "learning_rate": 4.852978757771664e-07,
      "loss": 0.2388,
      "step": 9597
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.6678764556094683,
      "learning_rate": 4.843600451089702e-07,
      "loss": 0.1506,
      "step": 9598
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5532234392095943,
      "learning_rate": 4.834230989982214e-07,
      "loss": 0.1675,
      "step": 9599
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.0046745253916467,
      "learning_rate": 4.824870375320156e-07,
      "loss": 0.1569,
      "step": 9600
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.239576951113229,
      "learning_rate": 4.81551860797369e-07,
      "loss": 0.2214,
      "step": 9601
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.2474067413231764,
      "learning_rate": 4.806175688812142e-07,
      "loss": 0.1951,
      "step": 9602
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.0006636483256948,
      "learning_rate": 4.796841618703984e-07,
      "loss": 0.1897,
      "step": 9603
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.9067459049087097,
      "learning_rate": 4.787516398516934e-07,
      "loss": 0.1702,
      "step": 9604
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.271717356627037,
      "learning_rate": 4.77820002911783e-07,
      "loss": 0.1997,
      "step": 9605
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.8318219510304454,
      "learning_rate": 4.768892511372703e-07,
      "loss": 0.156,
      "step": 9606
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.7201078526728182,
      "learning_rate": 4.7595938461467706e-07,
      "loss": 0.1354,
      "step": 9607
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.0069723241556634,
      "learning_rate": 4.7503040343044205e-07,
      "loss": 0.1667,
      "step": 9608
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.0632423698372873,
      "learning_rate": 4.741023076709217e-07,
      "loss": 0.1475,
      "step": 9609
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.6993826160580274,
      "learning_rate": 4.731750974223892e-07,
      "loss": 0.1655,
      "step": 9610
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.112980212405654,
      "learning_rate": 4.7224877277103673e-07,
      "loss": 0.1875,
      "step": 9611
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.8905124761916507,
      "learning_rate": 4.7132333380297546e-07,
      "loss": 0.1846,
      "step": 9612
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.261695005538431,
      "learning_rate": 4.703987806042332e-07,
      "loss": 0.2068,
      "step": 9613
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.9211666750569398,
      "learning_rate": 4.6947511326074893e-07,
      "loss": 0.1754,
      "step": 9614
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.639442065057153,
      "learning_rate": 4.6855233185839175e-07,
      "loss": 0.1656,
      "step": 9615
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.750790025042636,
      "learning_rate": 4.676304364829398e-07,
      "loss": 0.1487,
      "step": 9616
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.1276303232556892,
      "learning_rate": 4.6670942722009004e-07,
      "loss": 0.1985,
      "step": 9617
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.9746050068970429,
      "learning_rate": 4.657893041554584e-07,
      "loss": 0.1472,
      "step": 9618
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.3556450899021573,
      "learning_rate": 4.6487006737457765e-07,
      "loss": 0.2114,
      "step": 9619
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.0816737876192475,
      "learning_rate": 4.639517169628971e-07,
      "loss": 0.1934,
      "step": 9620
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.2440716937258487,
      "learning_rate": 4.6303425300578964e-07,
      "loss": 0.2527,
      "step": 9621
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.016122137052886,
      "learning_rate": 4.6211767558853484e-07,
      "loss": 0.2144,
      "step": 9622
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.230804702729258,
      "learning_rate": 4.6120198479634117e-07,
      "loss": 0.2002,
      "step": 9623
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.3752798277903797,
      "learning_rate": 4.6028718071432834e-07,
      "loss": 0.2141,
      "step": 9624
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.1108153602230497,
      "learning_rate": 4.5937326342753384e-07,
      "loss": 0.1504,
      "step": 9625
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.9313034347355649,
      "learning_rate": 4.5846023302091424e-07,
      "loss": 0.2231,
      "step": 9626
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.0377838582682415,
      "learning_rate": 4.575480895793438e-07,
      "loss": 0.1724,
      "step": 9627
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.1295389301848466,
      "learning_rate": 4.5663683318761255e-07,
      "loss": 0.1936,
      "step": 9628
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.0198933701624235,
      "learning_rate": 4.557264639304315e-07,
      "loss": 0.1853,
      "step": 9629
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.295637318195062,
      "learning_rate": 4.548169818924275e-07,
      "loss": 0.2329,
      "step": 9630
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.2105598306441814,
      "learning_rate": 4.5390838715813956e-07,
      "loss": 0.1987,
      "step": 9631
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.3158620364604845,
      "learning_rate": 4.5300067981203346e-07,
      "loss": 0.249,
      "step": 9632
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.9850960031257752,
      "learning_rate": 4.520938599384872e-07,
      "loss": 0.1632,
      "step": 9633
    },
    {
      "epoch": 2.72,
      "grad_norm": 3.1992705500778675,
      "learning_rate": 4.511879276217967e-07,
      "loss": 0.2629,
      "step": 9634
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.388419325580103,
      "learning_rate": 4.5028288294617583e-07,
      "loss": 0.217,
      "step": 9635
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.1519997099196555,
      "learning_rate": 4.4937872599575605e-07,
      "loss": 0.1946,
      "step": 9636
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.9950332132063162,
      "learning_rate": 4.484754568545857e-07,
      "loss": 0.1824,
      "step": 9637
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.984314438362962,
      "learning_rate": 4.475730756066332e-07,
      "loss": 0.1836,
      "step": 9638
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.9972289127599394,
      "learning_rate": 4.4667158233577925e-07,
      "loss": 0.1812,
      "step": 9639
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.0593179590762163,
      "learning_rate": 4.4577097712582897e-07,
      "loss": 0.158,
      "step": 9640
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.870084612203608,
      "learning_rate": 4.4487126006049764e-07,
      "loss": 0.1372,
      "step": 9641
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.3479335886078876,
      "learning_rate": 4.4397243122342284e-07,
      "loss": 0.2078,
      "step": 9642
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.1269610450473238,
      "learning_rate": 4.430744906981577e-07,
      "loss": 0.1938,
      "step": 9643
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.8210190713800543,
      "learning_rate": 4.421774385681743e-07,
      "loss": 0.15,
      "step": 9644
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.9529571795695195,
      "learning_rate": 4.412812749168582e-07,
      "loss": 0.1729,
      "step": 9645
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.144047084240625,
      "learning_rate": 4.403859998275184e-07,
      "loss": 0.1997,
      "step": 9646
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.0314460592376067,
      "learning_rate": 4.394916133833782e-07,
      "loss": 0.208,
      "step": 9647
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.7650520646837413,
      "learning_rate": 4.385981156675756e-07,
      "loss": 0.2803,
      "step": 9648
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.093014855162224,
      "learning_rate": 4.377055067631697e-07,
      "loss": 0.2092,
      "step": 9649
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.189101945201727,
      "learning_rate": 4.3681378675313747e-07,
      "loss": 0.217,
      "step": 9650
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.318363617594386,
      "learning_rate": 4.3592295572037037e-07,
      "loss": 0.1781,
      "step": 9651
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.9436809072499763,
      "learning_rate": 4.350330137476777e-07,
      "loss": 0.1768,
      "step": 9652
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.9329215265390767,
      "learning_rate": 4.3414396091778774e-07,
      "loss": 0.2019,
      "step": 9653
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.9269446828444483,
      "learning_rate": 4.3325579731334444e-07,
      "loss": 0.1598,
      "step": 9654
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.688753559809884,
      "learning_rate": 4.323685230169128e-07,
      "loss": 0.1456,
      "step": 9655
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.085316376494535,
      "learning_rate": 4.314821381109702e-07,
      "loss": 0.1782,
      "step": 9656
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.2411949481312297,
      "learning_rate": 4.305966426779118e-07,
      "loss": 0.2375,
      "step": 9657
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.9142007889474497,
      "learning_rate": 4.2971203680005404e-07,
      "loss": 0.155,
      "step": 9658
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.100309375901361,
      "learning_rate": 4.2882832055962885e-07,
      "loss": 0.2283,
      "step": 9659
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.336826470772415,
      "learning_rate": 4.279454940387828e-07,
      "loss": 0.1752,
      "step": 9660
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.447083993486794,
      "learning_rate": 4.270635573195836e-07,
      "loss": 0.2258,
      "step": 9661
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.9247920338190385,
      "learning_rate": 4.261825104840145e-07,
      "loss": 0.1899,
      "step": 9662
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.0033950498493893,
      "learning_rate": 4.253023536139733e-07,
      "loss": 0.1833,
      "step": 9663
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.0849211103140877,
      "learning_rate": 4.244230867912835e-07,
      "loss": 0.21,
      "step": 9664
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.158187792840094,
      "learning_rate": 4.2354471009767415e-07,
      "loss": 0.1973,
      "step": 9665
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.0421803728499297,
      "learning_rate": 4.226672236148022e-07,
      "loss": 0.1417,
      "step": 9666
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.2704061295122657,
      "learning_rate": 4.217906274242345e-07,
      "loss": 0.2075,
      "step": 9667
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.3679519017357937,
      "learning_rate": 4.209149216074593e-07,
      "loss": 0.1929,
      "step": 9668
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.3629053835374596,
      "learning_rate": 4.2004010624588033e-07,
      "loss": 0.1772,
      "step": 9669
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.34125094380043,
      "learning_rate": 4.191661814208181e-07,
      "loss": 0.2025,
      "step": 9670
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.1710333509258115,
      "learning_rate": 4.1829314721351213e-07,
      "loss": 0.1978,
      "step": 9671
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.0447589953248695,
      "learning_rate": 4.1742100370511853e-07,
      "loss": 0.1794,
      "step": 9672
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.207362060958273,
      "learning_rate": 4.1654975097671025e-07,
      "loss": 0.2076,
      "step": 9673
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.0429499214232694,
      "learning_rate": 4.1567938910927475e-07,
      "loss": 0.1899,
      "step": 9674
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.6515567527436747,
      "learning_rate": 4.1480991818372284e-07,
      "loss": 0.1713,
      "step": 9675
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.080118939342619,
      "learning_rate": 4.1394133828087654e-07,
      "loss": 0.1943,
      "step": 9676
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.9877458220675301,
      "learning_rate": 4.130736494814802e-07,
      "loss": 0.1725,
      "step": 9677
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.9280293385217961,
      "learning_rate": 4.1220685186619037e-07,
      "loss": 0.1685,
      "step": 9678
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.1672672448729413,
      "learning_rate": 4.113409455155837e-07,
      "loss": 0.2021,
      "step": 9679
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.9542904947742605,
      "learning_rate": 4.1047593051015245e-07,
      "loss": 0.1897,
      "step": 9680
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.9139551686684644,
      "learning_rate": 4.0961180693031123e-07,
      "loss": 0.1633,
      "step": 9681
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.145592007052411,
      "learning_rate": 4.087485748563813e-07,
      "loss": 0.197,
      "step": 9682
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.8591656477034637,
      "learning_rate": 4.0788623436861077e-07,
      "loss": 0.1549,
      "step": 9683
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.9726136235388816,
      "learning_rate": 4.0702478554716094e-07,
      "loss": 0.1801,
      "step": 9684
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.048999246374509,
      "learning_rate": 4.0616422847211013e-07,
      "loss": 0.1941,
      "step": 9685
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.3829216771580604,
      "learning_rate": 4.053045632234542e-07,
      "loss": 0.1978,
      "step": 9686
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.458523548995245,
      "learning_rate": 4.0444578988110715e-07,
      "loss": 0.2371,
      "step": 9687
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.5682161128205134,
      "learning_rate": 4.0358790852489616e-07,
      "loss": 0.2419,
      "step": 9688
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.1139262761166147,
      "learning_rate": 4.0273091923457297e-07,
      "loss": 0.1823,
      "step": 9689
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.1016873402860003,
      "learning_rate": 4.018748220897994e-07,
      "loss": 0.2153,
      "step": 9690
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.892126817283096,
      "learning_rate": 4.0101961717015416e-07,
      "loss": 0.1382,
      "step": 9691
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.3430216223553373,
      "learning_rate": 4.0016530455514013e-07,
      "loss": 0.2292,
      "step": 9692
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.12253335299019,
      "learning_rate": 3.9931188432417057e-07,
      "loss": 0.176,
      "step": 9693
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.9589499577314342,
      "learning_rate": 3.9845935655657866e-07,
      "loss": 0.2075,
      "step": 9694
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.960538261796585,
      "learning_rate": 3.976077213316132e-07,
      "loss": 0.1533,
      "step": 9695
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.249845132316336,
      "learning_rate": 3.967569787284409e-07,
      "loss": 0.2368,
      "step": 9696
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.4094886053649804,
      "learning_rate": 3.95907128826144e-07,
      "loss": 0.2014,
      "step": 9697
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.0679673957952818,
      "learning_rate": 3.9505817170372606e-07,
      "loss": 0.1794,
      "step": 9698
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.075091500267179,
      "learning_rate": 3.942101074401028e-07,
      "loss": 0.1788,
      "step": 9699
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.107303913462306,
      "learning_rate": 3.933629361141078e-07,
      "loss": 0.1949,
      "step": 9700
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.060936273223081,
      "learning_rate": 3.9251665780449587e-07,
      "loss": 0.1726,
      "step": 9701
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.0523943569995033,
      "learning_rate": 3.91671272589933e-07,
      "loss": 0.1704,
      "step": 9702
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.1681505650673976,
      "learning_rate": 3.908267805490051e-07,
      "loss": 0.1794,
      "step": 9703
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.0967058596496035,
      "learning_rate": 3.899831817602151e-07,
      "loss": 0.1908,
      "step": 9704
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.1819803003156957,
      "learning_rate": 3.8914047630198237e-07,
      "loss": 0.2021,
      "step": 9705
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.6978081655846675,
      "learning_rate": 3.8829866425264317e-07,
      "loss": 0.1722,
      "step": 9706
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.9365755482462077,
      "learning_rate": 3.874577456904516e-07,
      "loss": 0.1775,
      "step": 9707
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.103523952505833,
      "learning_rate": 3.866177206935751e-07,
      "loss": 0.1655,
      "step": 9708
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.2526268778268532,
      "learning_rate": 3.857785893401056e-07,
      "loss": 0.209,
      "step": 9709
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.9738442130806175,
      "learning_rate": 3.849403517080452e-07,
      "loss": 0.1606,
      "step": 9710
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.2978833581211258,
      "learning_rate": 3.8410300787531385e-07,
      "loss": 0.1497,
      "step": 9711
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.6585936150036678,
      "learning_rate": 3.832665579197503e-07,
      "loss": 0.2153,
      "step": 9712
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.1683097770235094,
      "learning_rate": 3.824310019191102e-07,
      "loss": 0.1813,
      "step": 9713
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.9651955407170898,
      "learning_rate": 3.815963399510647e-07,
      "loss": 0.1448,
      "step": 9714
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.1577288711712703,
      "learning_rate": 3.80762572093204e-07,
      "loss": 0.1855,
      "step": 9715
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.8113557074542195,
      "learning_rate": 3.799296984230316e-07,
      "loss": 0.1644,
      "step": 9716
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.400829727371474,
      "learning_rate": 3.790977190179701e-07,
      "loss": 0.2407,
      "step": 9717
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.1217391741790053,
      "learning_rate": 3.782666339553598e-07,
      "loss": 0.1874,
      "step": 9718
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.3334679253285793,
      "learning_rate": 3.774364433124578e-07,
      "loss": 0.2146,
      "step": 9719
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.285791472530287,
      "learning_rate": 3.7660714716643563e-07,
      "loss": 0.1831,
      "step": 9720
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.291674948770997,
      "learning_rate": 3.75778745594384e-07,
      "loss": 0.1975,
      "step": 9721
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.1303767446616604,
      "learning_rate": 3.749512386733101e-07,
      "loss": 0.1915,
      "step": 9722
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.9312011173284709,
      "learning_rate": 3.741246264801357e-07,
      "loss": 0.1558,
      "step": 9723
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.8543851085133523,
      "learning_rate": 3.7329890909170275e-07,
      "loss": 0.1776,
      "step": 9724
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.537042454487436,
      "learning_rate": 3.7247408658476756e-07,
      "loss": 0.241,
      "step": 9725
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.7364501618260813,
      "learning_rate": 3.7165015903600553e-07,
      "loss": 0.1373,
      "step": 9726
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.8749807840865644,
      "learning_rate": 3.708271265220087e-07,
      "loss": 0.1559,
      "step": 9727
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.0667240530879174,
      "learning_rate": 3.700049891192792e-07,
      "loss": 0.1846,
      "step": 9728
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.6051158055641936,
      "learning_rate": 3.691837469042481e-07,
      "loss": 0.2256,
      "step": 9729
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.9457484163017333,
      "learning_rate": 3.683633999532521e-07,
      "loss": 0.1821,
      "step": 9730
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.0693625911380966,
      "learning_rate": 3.6754394834255023e-07,
      "loss": 0.1948,
      "step": 9731
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.7560585291242443,
      "learning_rate": 3.6672539214832157e-07,
      "loss": 0.1462,
      "step": 9732
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.426968384241231,
      "learning_rate": 3.659077314466519e-07,
      "loss": 0.228,
      "step": 9733
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.4865589940258297,
      "learning_rate": 3.650909663135505e-07,
      "loss": 0.2285,
      "step": 9734
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.9448254938938174,
      "learning_rate": 3.642750968249442e-07,
      "loss": 0.1744,
      "step": 9735
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.3871998853812157,
      "learning_rate": 3.63460123056677e-07,
      "loss": 0.2805,
      "step": 9736
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.8774571207030573,
      "learning_rate": 3.626460450845015e-07,
      "loss": 0.1707,
      "step": 9737
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.1037835077031817,
      "learning_rate": 3.6183286298409724e-07,
      "loss": 0.1941,
      "step": 9738
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.8864016622248578,
      "learning_rate": 3.6102057683105596e-07,
      "loss": 0.1633,
      "step": 9739
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.177914492493905,
      "learning_rate": 3.60209186700885e-07,
      "loss": 0.2053,
      "step": 9740
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.365559117724198,
      "learning_rate": 3.5939869266901073e-07,
      "loss": 0.1895,
      "step": 9741
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.141265553571874,
      "learning_rate": 3.58589094810774e-07,
      "loss": 0.196,
      "step": 9742
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.9607449802761414,
      "learning_rate": 3.5778039320143456e-07,
      "loss": 0.166,
      "step": 9743
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.53138132975312,
      "learning_rate": 3.5697258791617007e-07,
      "loss": 0.2124,
      "step": 9744
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.8496139148194166,
      "learning_rate": 3.561656790300683e-07,
      "loss": 0.1479,
      "step": 9745
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.101071721688005,
      "learning_rate": 3.553596666181414e-07,
      "loss": 0.1747,
      "step": 9746
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.9187025865852816,
      "learning_rate": 3.545545507553139e-07,
      "loss": 0.188,
      "step": 9747
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.022872021972572,
      "learning_rate": 3.537503315164259e-07,
      "loss": 0.1963,
      "step": 9748
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.046706699698077,
      "learning_rate": 3.529470089762421e-07,
      "loss": 0.1833,
      "step": 9749
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.8197413667304705,
      "learning_rate": 3.521445832094328e-07,
      "loss": 0.1527,
      "step": 9750
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.1853342481845215,
      "learning_rate": 3.5134305429058935e-07,
      "loss": 0.2312,
      "step": 9751
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.7298963734476085,
      "learning_rate": 3.505424222942244e-07,
      "loss": 0.1796,
      "step": 9752
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.290261981756512,
      "learning_rate": 3.497426872947629e-07,
      "loss": 0.1926,
      "step": 9753
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.0005909093178436,
      "learning_rate": 3.48943849366542e-07,
      "loss": 0.2041,
      "step": 9754
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.057749385715865,
      "learning_rate": 3.481459085838268e-07,
      "loss": 0.1371,
      "step": 9755
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.8812353506044717,
      "learning_rate": 3.473488650207879e-07,
      "loss": 0.1912,
      "step": 9756
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.0763231168393315,
      "learning_rate": 3.465527187515194e-07,
      "loss": 0.2026,
      "step": 9757
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.8324549497949956,
      "learning_rate": 3.4575746985002877e-07,
      "loss": 0.1599,
      "step": 9758
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.0131162339844044,
      "learning_rate": 3.4496311839024133e-07,
      "loss": 0.1896,
      "step": 9759
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.919200396617859,
      "learning_rate": 3.441696644459969e-07,
      "loss": 0.1843,
      "step": 9760
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.0977005157401116,
      "learning_rate": 3.433771080910575e-07,
      "loss": 0.207,
      "step": 9761
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.6176719379783893,
      "learning_rate": 3.4258544939909324e-07,
      "loss": 0.2068,
      "step": 9762
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.3110408554357162,
      "learning_rate": 3.4179468844369847e-07,
      "loss": 0.172,
      "step": 9763
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.9081405914074447,
      "learning_rate": 3.4100482529838e-07,
      "loss": 0.1665,
      "step": 9764
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.018459044283163,
      "learning_rate": 3.4021586003656236e-07,
      "loss": 0.1812,
      "step": 9765
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.1515965502440775,
      "learning_rate": 3.394277927315859e-07,
      "loss": 0.1775,
      "step": 9766
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.8958520377282078,
      "learning_rate": 3.386406234567086e-07,
      "loss": 0.1403,
      "step": 9767
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.0465559748929123,
      "learning_rate": 3.37854352285103e-07,
      "loss": 0.1731,
      "step": 9768
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.8691703842576854,
      "learning_rate": 3.370689792898618e-07,
      "loss": 0.1948,
      "step": 9769
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.120609061601115,
      "learning_rate": 3.362845045439911e-07,
      "loss": 0.2046,
      "step": 9770
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.25935604529972,
      "learning_rate": 3.3550092812041244e-07,
      "loss": 0.2126,
      "step": 9771
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.291912994979665,
      "learning_rate": 3.347182500919677e-07,
      "loss": 0.2119,
      "step": 9772
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.8008969807034363,
      "learning_rate": 3.33936470531413e-07,
      "loss": 0.1592,
      "step": 9773
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.1819159061580686,
      "learning_rate": 3.3315558951142133e-07,
      "loss": 0.2107,
      "step": 9774
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.027514363747552,
      "learning_rate": 3.3237560710458137e-07,
      "loss": 0.2175,
      "step": 9775
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.0900348190519753,
      "learning_rate": 3.3159652338339953e-07,
      "loss": 0.2192,
      "step": 9776
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.146246055403962,
      "learning_rate": 3.3081833842029563e-07,
      "loss": 0.2368,
      "step": 9777
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.9225255291995942,
      "learning_rate": 3.300410522876141e-07,
      "loss": 0.1766,
      "step": 9778
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.233548377821219,
      "learning_rate": 3.292646650576037e-07,
      "loss": 0.1515,
      "step": 9779
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.27981707536632,
      "learning_rate": 3.284891768024401e-07,
      "loss": 0.2361,
      "step": 9780
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.1421433576743465,
      "learning_rate": 3.2771458759421005e-07,
      "loss": 0.208,
      "step": 9781
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.7353254599838146,
      "learning_rate": 3.269408975049182e-07,
      "loss": 0.1736,
      "step": 9782
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.9957541552659408,
      "learning_rate": 3.261681066064859e-07,
      "loss": 0.2046,
      "step": 9783
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.260196985979763,
      "learning_rate": 3.25396214970749e-07,
      "loss": 0.217,
      "step": 9784
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.224303122453485,
      "learning_rate": 3.2462522266946127e-07,
      "loss": 0.2354,
      "step": 9785
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.0534480176268994,
      "learning_rate": 3.238551297742953e-07,
      "loss": 0.2141,
      "step": 9786
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.9852502080384469,
      "learning_rate": 3.230859363568373e-07,
      "loss": 0.168,
      "step": 9787
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.68449031987444,
      "learning_rate": 3.2231764248858656e-07,
      "loss": 0.1874,
      "step": 9788
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.681867673162143,
      "learning_rate": 3.215502482409649e-07,
      "loss": 0.1835,
      "step": 9789
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.9556099546218353,
      "learning_rate": 3.207837536853087e-07,
      "loss": 0.163,
      "step": 9790
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.2202005557998974,
      "learning_rate": 3.2001815889286856e-07,
      "loss": 0.1895,
      "step": 9791
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.7263206813163186,
      "learning_rate": 3.1925346393481327e-07,
      "loss": 0.167,
      "step": 9792
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.0139220242143794,
      "learning_rate": 3.184896688822281e-07,
      "loss": 0.2021,
      "step": 9793
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.9512634323597693,
      "learning_rate": 3.1772677380611185e-07,
      "loss": 0.1649,
      "step": 9794
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.2529853865027043,
      "learning_rate": 3.1696477877738664e-07,
      "loss": 0.2688,
      "step": 9795
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.8475542379203973,
      "learning_rate": 3.1620368386688137e-07,
      "loss": 0.1997,
      "step": 9796
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.0768812206494984,
      "learning_rate": 3.154434891453473e-07,
      "loss": 0.1836,
      "step": 9797
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.9048084170425617,
      "learning_rate": 3.1468419468345223e-07,
      "loss": 0.2002,
      "step": 9798
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.93068471873497,
      "learning_rate": 3.1392580055177867e-07,
      "loss": 0.1777,
      "step": 9799
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.9641633614308989,
      "learning_rate": 3.131683068208247e-07,
      "loss": 0.187,
      "step": 9800
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.0926383329967324,
      "learning_rate": 3.1241171356100606e-07,
      "loss": 0.2302,
      "step": 9801
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.9335326393983874,
      "learning_rate": 3.1165602084265446e-07,
      "loss": 0.1736,
      "step": 9802
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.0553689256522603,
      "learning_rate": 3.109012287360158e-07,
      "loss": 0.2068,
      "step": 9803
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.527947469813647,
      "learning_rate": 3.1014733731125955e-07,
      "loss": 0.2039,
      "step": 9804
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.3249299148781497,
      "learning_rate": 3.093943466384597e-07,
      "loss": 0.228,
      "step": 9805
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.097982050961876,
      "learning_rate": 3.0864225678761684e-07,
      "loss": 0.2148,
      "step": 9806
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.238915445480791,
      "learning_rate": 3.0789106782864285e-07,
      "loss": 0.1812,
      "step": 9807
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.9916220982712198,
      "learning_rate": 3.071407798313686e-07,
      "loss": 0.1897,
      "step": 9808
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.424930131253101,
      "learning_rate": 3.0639139286553707e-07,
      "loss": 0.2458,
      "step": 9809
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.7075285929638269,
      "learning_rate": 3.0564290700081044e-07,
      "loss": 0.1462,
      "step": 9810
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.288953879054582,
      "learning_rate": 3.0489532230676744e-07,
      "loss": 0.2063,
      "step": 9811
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.247425454600148,
      "learning_rate": 3.041486388529036e-07,
      "loss": 0.2283,
      "step": 9812
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.9206098623692023,
      "learning_rate": 3.0340285670862667e-07,
      "loss": 0.1445,
      "step": 9813
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.8760291780286071,
      "learning_rate": 3.026579759432635e-07,
      "loss": 0.1593,
      "step": 9814
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.8318544109045243,
      "learning_rate": 3.019139966260587e-07,
      "loss": 0.1589,
      "step": 9815
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.027897939606409,
      "learning_rate": 3.0117091882617025e-07,
      "loss": 0.1577,
      "step": 9816
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.019999895433679,
      "learning_rate": 3.0042874261267395e-07,
      "loss": 0.1536,
      "step": 9817
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.2207965291994083,
      "learning_rate": 2.996874680545603e-07,
      "loss": 0.1836,
      "step": 9818
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.8050425018895426,
      "learning_rate": 2.989470952207385e-07,
      "loss": 0.162,
      "step": 9819
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.976690833876653,
      "learning_rate": 2.9820762418002916e-07,
      "loss": 0.1758,
      "step": 9820
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.9930377605535314,
      "learning_rate": 2.9746905500117604e-07,
      "loss": 0.1816,
      "step": 9821
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.1620598285547104,
      "learning_rate": 2.967313877528322e-07,
      "loss": 0.2407,
      "step": 9822
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.0306013803945757,
      "learning_rate": 2.959946225035726e-07,
      "loss": 0.1809,
      "step": 9823
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.9065503909806567,
      "learning_rate": 2.9525875932188365e-07,
      "loss": 0.156,
      "step": 9824
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.9865226178770785,
      "learning_rate": 2.945237982761706e-07,
      "loss": 0.1993,
      "step": 9825
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.230660731320299,
      "learning_rate": 2.937897394347544e-07,
      "loss": 0.1821,
      "step": 9826
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.212742962219973,
      "learning_rate": 2.930565828658716e-07,
      "loss": 0.2104,
      "step": 9827
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.2801055541056447,
      "learning_rate": 2.9232432863767424e-07,
      "loss": 0.2136,
      "step": 9828
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.9836120529606889,
      "learning_rate": 2.915929768182335e-07,
      "loss": 0.1936,
      "step": 9829
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.134608871803313,
      "learning_rate": 2.908625274755339e-07,
      "loss": 0.1782,
      "step": 9830
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.0562581454411824,
      "learning_rate": 2.901329806774744e-07,
      "loss": 0.189,
      "step": 9831
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.01794192729282,
      "learning_rate": 2.8940433649187525e-07,
      "loss": 0.1824,
      "step": 9832
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.0765018287801116,
      "learning_rate": 2.8867659498647e-07,
      "loss": 0.1775,
      "step": 9833
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.3938280315775438,
      "learning_rate": 2.8794975622890573e-07,
      "loss": 0.2234,
      "step": 9834
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.8086793447802438,
      "learning_rate": 2.8722382028675055e-07,
      "loss": 0.1641,
      "step": 9835
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.008509143273793,
      "learning_rate": 2.864987872274849e-07,
      "loss": 0.2014,
      "step": 9836
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.3375280151253963,
      "learning_rate": 2.8577465711850605e-07,
      "loss": 0.2051,
      "step": 9837
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.727964588561468,
      "learning_rate": 2.8505143002713007e-07,
      "loss": 0.1562,
      "step": 9838
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.955225181279314,
      "learning_rate": 2.843291060205855e-07,
      "loss": 0.1956,
      "step": 9839
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.069734429990153,
      "learning_rate": 2.8360768516601745e-07,
      "loss": 0.167,
      "step": 9840
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.0834260082852176,
      "learning_rate": 2.8288716753049007e-07,
      "loss": 0.2202,
      "step": 9841
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.729253419275287,
      "learning_rate": 2.821675531809809e-07,
      "loss": 0.1753,
      "step": 9842
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.221368158289456,
      "learning_rate": 2.814488421843831e-07,
      "loss": 0.2012,
      "step": 9843
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.025867193113482,
      "learning_rate": 2.8073103460750653e-07,
      "loss": 0.1874,
      "step": 9844
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.0786511772990224,
      "learning_rate": 2.800141305170789e-07,
      "loss": 0.1711,
      "step": 9845
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.8351489122992084,
      "learning_rate": 2.7929812997974036e-07,
      "loss": 0.1611,
      "step": 9846
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.014493611849932,
      "learning_rate": 2.785830330620509e-07,
      "loss": 0.1873,
      "step": 9847
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.8673270822392194,
      "learning_rate": 2.7786883983048294e-07,
      "loss": 0.2747,
      "step": 9848
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.0653426858331745,
      "learning_rate": 2.771555503514289e-07,
      "loss": 0.2106,
      "step": 9849
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.125236483656711,
      "learning_rate": 2.764431646911947e-07,
      "loss": 0.167,
      "step": 9850
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.2347477171199204,
      "learning_rate": 2.757316829159995e-07,
      "loss": 0.1936,
      "step": 9851
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.9859560387428363,
      "learning_rate": 2.750211050919849e-07,
      "loss": 0.2014,
      "step": 9852
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.3048942648587736,
      "learning_rate": 2.7431143128520243e-07,
      "loss": 0.2725,
      "step": 9853
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.262951538438897,
      "learning_rate": 2.7360266156162274e-07,
      "loss": 0.2239,
      "step": 9854
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.95984181236519,
      "learning_rate": 2.728947959871353e-07,
      "loss": 0.1652,
      "step": 9855
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.2325472932840933,
      "learning_rate": 2.721878346275364e-07,
      "loss": 0.186,
      "step": 9856
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.9134922204038747,
      "learning_rate": 2.714817775485468e-07,
      "loss": 0.1715,
      "step": 9857
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.4371700263167764,
      "learning_rate": 2.707766248158006e-07,
      "loss": 0.2236,
      "step": 9858
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.041032402159659,
      "learning_rate": 2.7007237649484763e-07,
      "loss": 0.2107,
      "step": 9859
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.286658622853031,
      "learning_rate": 2.693690326511533e-07,
      "loss": 0.2396,
      "step": 9860
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.1056723704109537,
      "learning_rate": 2.686665933500987e-07,
      "loss": 0.1763,
      "step": 9861
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.7425570053799746,
      "learning_rate": 2.6796505865698263e-07,
      "loss": 0.1626,
      "step": 9862
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.7415404217400472,
      "learning_rate": 2.672644286370163e-07,
      "loss": 0.1582,
      "step": 9863
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.3065542438344426,
      "learning_rate": 2.665647033553309e-07,
      "loss": 0.2112,
      "step": 9864
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.1170681218925322,
      "learning_rate": 2.6586588287697114e-07,
      "loss": 0.2114,
      "step": 9865
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.032222181791107,
      "learning_rate": 2.651679672668983e-07,
      "loss": 0.2183,
      "step": 9866
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.232418758846575,
      "learning_rate": 2.6447095658999054e-07,
      "loss": 0.1837,
      "step": 9867
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.20865353941169,
      "learning_rate": 2.6377485091103825e-07,
      "loss": 0.2058,
      "step": 9868
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.262233752225806,
      "learning_rate": 2.630796502947519e-07,
      "loss": 0.1602,
      "step": 9869
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.9747267688619294,
      "learning_rate": 2.6238535480575533e-07,
      "loss": 0.1809,
      "step": 9870
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.9675465396609293,
      "learning_rate": 2.616919645085902e-07,
      "loss": 0.1594,
      "step": 9871
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.0417809582300146,
      "learning_rate": 2.609994794677118e-07,
      "loss": 0.2002,
      "step": 9872
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.0583892782907087,
      "learning_rate": 2.6030789974749285e-07,
      "loss": 0.171,
      "step": 9873
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.0972568781170366,
      "learning_rate": 2.596172254122209e-07,
      "loss": 0.2043,
      "step": 9874
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.002687492541802,
      "learning_rate": 2.589274565261002e-07,
      "loss": 0.1743,
      "step": 9875
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.0327044421795963,
      "learning_rate": 2.582385931532505e-07,
      "loss": 0.1982,
      "step": 9876
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.288203367676447,
      "learning_rate": 2.575506353577084e-07,
      "loss": 0.1821,
      "step": 9877
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.087984964276376,
      "learning_rate": 2.5686358320342387e-07,
      "loss": 0.1763,
      "step": 9878
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.0739125923266326,
      "learning_rate": 2.5617743675426354e-07,
      "loss": 0.2053,
      "step": 9879
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.4200246697370176,
      "learning_rate": 2.55492196074012e-07,
      "loss": 0.2378,
      "step": 9880
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.182815967017807,
      "learning_rate": 2.5480786122636713e-07,
      "loss": 0.2156,
      "step": 9881
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.353094192577421,
      "learning_rate": 2.5412443227494365e-07,
      "loss": 0.2727,
      "step": 9882
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.384287357582218,
      "learning_rate": 2.534419092832718e-07,
      "loss": 0.2456,
      "step": 9883
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.2862727060551613,
      "learning_rate": 2.527602923147998e-07,
      "loss": 0.2241,
      "step": 9884
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.8155537113277427,
      "learning_rate": 2.520795814328847e-07,
      "loss": 0.153,
      "step": 9885
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.1717864690122903,
      "learning_rate": 2.513997767008092e-07,
      "loss": 0.2029,
      "step": 9886
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.2954381474832735,
      "learning_rate": 2.507208781817638e-07,
      "loss": 0.1882,
      "step": 9887
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.7593401009077692,
      "learning_rate": 2.500428859388593e-07,
      "loss": 0.146,
      "step": 9888
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.218008059039308,
      "learning_rate": 2.4936580003512066e-07,
      "loss": 0.1768,
      "step": 9889
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.9151361792195052,
      "learning_rate": 2.4868962053348764e-07,
      "loss": 0.1941,
      "step": 9890
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.1498549565096576,
      "learning_rate": 2.4801434749681553e-07,
      "loss": 0.1736,
      "step": 9891
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.0892046759084506,
      "learning_rate": 2.473399809878807e-07,
      "loss": 0.1969,
      "step": 9892
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.165429855281194,
      "learning_rate": 2.466665210693686e-07,
      "loss": 0.2231,
      "step": 9893
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.1567179589224903,
      "learning_rate": 2.459939678038803e-07,
      "loss": 0.1814,
      "step": 9894
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.0677102180393168,
      "learning_rate": 2.453223212539391e-07,
      "loss": 0.1807,
      "step": 9895
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.138013213555738,
      "learning_rate": 2.446515814819794e-07,
      "loss": 0.1729,
      "step": 9896
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.0949387217770408,
      "learning_rate": 2.4398174855035037e-07,
      "loss": 0.1934,
      "step": 9897
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.610610461806681,
      "learning_rate": 2.4331282252132103e-07,
      "loss": 0.2793,
      "step": 9898
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.883908365522868,
      "learning_rate": 2.4264480345707053e-07,
      "loss": 0.1589,
      "step": 9899
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.9571057979395354,
      "learning_rate": 2.419776914196981e-07,
      "loss": 0.1758,
      "step": 9900
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.7936415816908413,
      "learning_rate": 2.413114864712196e-07,
      "loss": 0.1685,
      "step": 9901
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.16984369888595,
      "learning_rate": 2.4064618867356003e-07,
      "loss": 0.251,
      "step": 9902
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.0583920346855438,
      "learning_rate": 2.399817980885677e-07,
      "loss": 0.1848,
      "step": 9903
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.3211286673298828,
      "learning_rate": 2.3931831477800207e-07,
      "loss": 0.2217,
      "step": 9904
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.342525538738455,
      "learning_rate": 2.3865573880353933e-07,
      "loss": 0.2468,
      "step": 9905
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.371900357253875,
      "learning_rate": 2.3799407022677022e-07,
      "loss": 0.2454,
      "step": 9906
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.049530164974525,
      "learning_rate": 2.373333091092056e-07,
      "loss": 0.1497,
      "step": 9907
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.323982724419571,
      "learning_rate": 2.3667345551226406e-07,
      "loss": 0.2046,
      "step": 9908
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.04775280921912,
      "learning_rate": 2.3601450949728876e-07,
      "loss": 0.1724,
      "step": 9909
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.1313058037239956,
      "learning_rate": 2.3535647112553295e-07,
      "loss": 0.1951,
      "step": 9910
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.1121595611735144,
      "learning_rate": 2.3469934045816435e-07,
      "loss": 0.1836,
      "step": 9911
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.7375339047704517,
      "learning_rate": 2.3404311755627184e-07,
      "loss": 0.1685,
      "step": 9912
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.374508213239011,
      "learning_rate": 2.3338780248085557e-07,
      "loss": 0.2471,
      "step": 9913
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.3766408430180834,
      "learning_rate": 2.3273339529283123e-07,
      "loss": 0.2661,
      "step": 9914
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.298435572528657,
      "learning_rate": 2.320798960530335e-07,
      "loss": 0.2273,
      "step": 9915
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.0804794193416787,
      "learning_rate": 2.314273048222093e-07,
      "loss": 0.2124,
      "step": 9916
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.033523213420363,
      "learning_rate": 2.307756216610224e-07,
      "loss": 0.1902,
      "step": 9917
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.8026594564512815,
      "learning_rate": 2.301248466300543e-07,
      "loss": 0.1722,
      "step": 9918
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.9511947482206096,
      "learning_rate": 2.294749797897955e-07,
      "loss": 0.1467,
      "step": 9919
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.5051336000566846,
      "learning_rate": 2.288260212006599e-07,
      "loss": 0.2209,
      "step": 9920
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.9518125750726585,
      "learning_rate": 2.2817797092297256e-07,
      "loss": 0.1768,
      "step": 9921
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.2624148200468275,
      "learning_rate": 2.2753082901697644e-07,
      "loss": 0.2278,
      "step": 9922
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.1045698093781557,
      "learning_rate": 2.2688459554282673e-07,
      "loss": 0.23,
      "step": 9923
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.706459783378022,
      "learning_rate": 2.2623927056059647e-07,
      "loss": 0.1292,
      "step": 9924
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.192330727377167,
      "learning_rate": 2.2559485413027438e-07,
      "loss": 0.2222,
      "step": 9925
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.8554389537823166,
      "learning_rate": 2.2495134631176585e-07,
      "loss": 0.1711,
      "step": 9926
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.6026919762615406,
      "learning_rate": 2.243087471648886e-07,
      "loss": 0.1454,
      "step": 9927
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.118362485648301,
      "learning_rate": 2.2366705674937596e-07,
      "loss": 0.1855,
      "step": 9928
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.2177300426386477,
      "learning_rate": 2.230262751248813e-07,
      "loss": 0.217,
      "step": 9929
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.8883842133755104,
      "learning_rate": 2.2238640235097032e-07,
      "loss": 0.1709,
      "step": 9930
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.0247576880094833,
      "learning_rate": 2.217474384871221e-07,
      "loss": 0.1804,
      "step": 9931
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.010402599615679,
      "learning_rate": 2.2110938359273583e-07,
      "loss": 0.1604,
      "step": 9932
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.1549695001359015,
      "learning_rate": 2.20472237727124e-07,
      "loss": 0.1626,
      "step": 9933
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.2664947461794114,
      "learning_rate": 2.1983600094951153e-07,
      "loss": 0.1766,
      "step": 9934
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.114527723129432,
      "learning_rate": 2.192006733190466e-07,
      "loss": 0.1898,
      "step": 9935
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.4654551183658824,
      "learning_rate": 2.1856625489478532e-07,
      "loss": 0.271,
      "step": 9936
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.2073216507457865,
      "learning_rate": 2.1793274573570166e-07,
      "loss": 0.1626,
      "step": 9937
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.1066437270127376,
      "learning_rate": 2.1730014590068625e-07,
      "loss": 0.1781,
      "step": 9938
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.5826983349962322,
      "learning_rate": 2.1666845544854542e-07,
      "loss": 0.188,
      "step": 9939
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.6999699693447814,
      "learning_rate": 2.1603767443799994e-07,
      "loss": 0.1807,
      "step": 9940
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.1612261779174933,
      "learning_rate": 2.1540780292768516e-07,
      "loss": 0.2256,
      "step": 9941
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.6923861100162763,
      "learning_rate": 2.1477884097615308e-07,
      "loss": 0.168,
      "step": 9942
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.0661081764756957,
      "learning_rate": 2.1415078864187034e-07,
      "loss": 0.1833,
      "step": 9943
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.1429200221607516,
      "learning_rate": 2.135236459832213e-07,
      "loss": 0.2024,
      "step": 9944
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.4848037166440187,
      "learning_rate": 2.1289741305850154e-07,
      "loss": 0.2393,
      "step": 9945
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.3132570438393985,
      "learning_rate": 2.1227208992592675e-07,
      "loss": 0.2368,
      "step": 9946
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.181750660801956,
      "learning_rate": 2.1164767664362485e-07,
      "loss": 0.2202,
      "step": 9947
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.8772033857094046,
      "learning_rate": 2.1102417326964165e-07,
      "loss": 0.1451,
      "step": 9948
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.400030537957562,
      "learning_rate": 2.104015798619352e-07,
      "loss": 0.2148,
      "step": 9949
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.30173928003969,
      "learning_rate": 2.097798964783826e-07,
      "loss": 0.2192,
      "step": 9950
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.032771367956775,
      "learning_rate": 2.091591231767709e-07,
      "loss": 0.2046,
      "step": 9951
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.182578311166938,
      "learning_rate": 2.085392600148106e-07,
      "loss": 0.1591,
      "step": 9952
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.3870886449120174,
      "learning_rate": 2.0792030705012013e-07,
      "loss": 0.2192,
      "step": 9953
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.034910875210514,
      "learning_rate": 2.0730226434023671e-07,
      "loss": 0.1863,
      "step": 9954
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.268597090686894,
      "learning_rate": 2.066851319426133e-07,
      "loss": 0.2,
      "step": 9955
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.12179927887074,
      "learning_rate": 2.0606890991461737e-07,
      "loss": 0.1689,
      "step": 9956
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.45369517699329,
      "learning_rate": 2.0545359831353195e-07,
      "loss": 0.2168,
      "step": 9957
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.088499148901687,
      "learning_rate": 2.0483919719655466e-07,
      "loss": 0.1499,
      "step": 9958
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.1638039525883768,
      "learning_rate": 2.0422570662079866e-07,
      "loss": 0.1682,
      "step": 9959
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.07390544774975,
      "learning_rate": 2.0361312664329502e-07,
      "loss": 0.1838,
      "step": 9960
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.8511677288571724,
      "learning_rate": 2.0300145732098596e-07,
      "loss": 0.1974,
      "step": 9961
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.337553444823108,
      "learning_rate": 2.0239069871073157e-07,
      "loss": 0.241,
      "step": 9962
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.4659925326645857,
      "learning_rate": 2.0178085086930865e-07,
      "loss": 0.1904,
      "step": 9963
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.3887370259426755,
      "learning_rate": 2.0117191385340629e-07,
      "loss": 0.1887,
      "step": 9964
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.2299883191066336,
      "learning_rate": 2.005638877196303e-07,
      "loss": 0.2363,
      "step": 9965
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.130496042308426,
      "learning_rate": 1.999567725245022e-07,
      "loss": 0.1995,
      "step": 9966
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.1922057259690355,
      "learning_rate": 1.9935056832445676e-07,
      "loss": 0.1521,
      "step": 9967
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.0089073459017723,
      "learning_rate": 1.9874527517584784e-07,
      "loss": 0.1746,
      "step": 9968
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.021967030650473,
      "learning_rate": 1.9814089313494157e-07,
      "loss": 0.1761,
      "step": 9969
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.217505444483152,
      "learning_rate": 1.975374222579207e-07,
      "loss": 0.2063,
      "step": 9970
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5120036530278973,
      "learning_rate": 1.9693486260088047e-07,
      "loss": 0.1276,
      "step": 9971
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.1731036924303417,
      "learning_rate": 1.9633321421983708e-07,
      "loss": 0.1703,
      "step": 9972
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.9107959402145762,
      "learning_rate": 1.957324771707181e-07,
      "loss": 0.1595,
      "step": 9973
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.4361383457991663,
      "learning_rate": 1.9513265150936433e-07,
      "loss": 0.2185,
      "step": 9974
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.8308743078160437,
      "learning_rate": 1.945337372915368e-07,
      "loss": 0.1444,
      "step": 9975
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.0196626065942316,
      "learning_rate": 1.9393573457290983e-07,
      "loss": 0.1599,
      "step": 9976
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.9575309454218424,
      "learning_rate": 1.9333864340907116e-07,
      "loss": 0.1855,
      "step": 9977
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.9489262831725525,
      "learning_rate": 1.9274246385552753e-07,
      "loss": 0.1584,
      "step": 9978
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.9323951559121886,
      "learning_rate": 1.921471959676957e-07,
      "loss": 0.1678,
      "step": 9979
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.0742619151380373,
      "learning_rate": 1.9155283980091366e-07,
      "loss": 0.1965,
      "step": 9980
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.7256390251814877,
      "learning_rate": 1.909593954104294e-07,
      "loss": 0.1215,
      "step": 9981
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.441267531276463,
      "learning_rate": 1.9036686285141105e-07,
      "loss": 0.2288,
      "step": 9982
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.4221386168570267,
      "learning_rate": 1.8977524217893782e-07,
      "loss": 0.2164,
      "step": 9983
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.2776694589705038,
      "learning_rate": 1.891845334480058e-07,
      "loss": 0.1666,
      "step": 9984
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.884949593153742,
      "learning_rate": 1.8859473671352546e-07,
      "loss": 0.1505,
      "step": 9985
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.9145011545381745,
      "learning_rate": 1.8800585203032517e-07,
      "loss": 0.1907,
      "step": 9986
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.9532724514134987,
      "learning_rate": 1.874178794531456e-07,
      "loss": 0.1804,
      "step": 9987
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.7899222788672224,
      "learning_rate": 1.86830819036643e-07,
      "loss": 0.2134,
      "step": 9988
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.1081295796378625,
      "learning_rate": 1.8624467083539154e-07,
      "loss": 0.2114,
      "step": 9989
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.325234072075245,
      "learning_rate": 1.8565943490387761e-07,
      "loss": 0.1938,
      "step": 9990
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.941230557035138,
      "learning_rate": 1.85075111296501e-07,
      "loss": 0.1781,
      "step": 9991
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.34894209904748,
      "learning_rate": 1.8449170006758278e-07,
      "loss": 0.1946,
      "step": 9992
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.1846014709806156,
      "learning_rate": 1.8390920127135613e-07,
      "loss": 0.1956,
      "step": 9993
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.9981699138676607,
      "learning_rate": 1.833276149619667e-07,
      "loss": 0.1814,
      "step": 9994
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.9004118242072423,
      "learning_rate": 1.8274694119347901e-07,
      "loss": 0.1965,
      "step": 9995
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.3874389099797666,
      "learning_rate": 1.8216718001987098e-07,
      "loss": 0.2063,
      "step": 9996
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.70135450988083,
      "learning_rate": 1.815883314950373e-07,
      "loss": 0.135,
      "step": 9997
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.0963494229638417,
      "learning_rate": 1.81010395672786e-07,
      "loss": 0.2017,
      "step": 9998
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.9654657704111804,
      "learning_rate": 1.804333726068408e-07,
      "loss": 0.1609,
      "step": 9999
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.7890308643143997,
      "learning_rate": 1.79857262350841e-07,
      "loss": 0.1515,
      "step": 10000
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.117723615253159,
      "learning_rate": 1.7928206495834043e-07,
      "loss": 0.1624,
      "step": 10001
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.802640198099071,
      "learning_rate": 1.787077804828097e-07,
      "loss": 0.1627,
      "step": 10002
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.1747840570518076,
      "learning_rate": 1.7813440897763158e-07,
      "loss": 0.1798,
      "step": 10003
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.9990354060719164,
      "learning_rate": 1.7756195049610682e-07,
      "loss": 0.198,
      "step": 10004
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.4216890172668766,
      "learning_rate": 1.769904050914495e-07,
      "loss": 0.248,
      "step": 10005
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.0715187776776047,
      "learning_rate": 1.7641977281679046e-07,
      "loss": 0.1814,
      "step": 10006
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.0006734218659368,
      "learning_rate": 1.7585005372517504e-07,
      "loss": 0.1934,
      "step": 10007
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.2241469175536555,
      "learning_rate": 1.7528124786956092e-07,
      "loss": 0.2317,
      "step": 10008
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.5420115030524935,
      "learning_rate": 1.7471335530282574e-07,
      "loss": 0.2556,
      "step": 10009
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.2298581718811477,
      "learning_rate": 1.741463760777584e-07,
      "loss": 0.2294,
      "step": 10010
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.045103578988092,
      "learning_rate": 1.7358031024706456e-07,
      "loss": 0.1941,
      "step": 10011
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.2104335269551267,
      "learning_rate": 1.7301515786336541e-07,
      "loss": 0.1899,
      "step": 10012
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.501788789403587,
      "learning_rate": 1.7245091897919564e-07,
      "loss": 0.2236,
      "step": 10013
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.938702686042517,
      "learning_rate": 1.7188759364700658e-07,
      "loss": 0.1643,
      "step": 10014
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.8866242346899036,
      "learning_rate": 1.7132518191916413e-07,
      "loss": 0.1775,
      "step": 10015
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.7643674600748676,
      "learning_rate": 1.7076368384794872e-07,
      "loss": 0.1692,
      "step": 10016
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.037276126557447,
      "learning_rate": 1.7020309948555525e-07,
      "loss": 0.1794,
      "step": 10017
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.7982541848328308,
      "learning_rate": 1.6964342888409646e-07,
      "loss": 0.1799,
      "step": 10018
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.8800256433167415,
      "learning_rate": 1.6908467209559853e-07,
      "loss": 0.1924,
      "step": 10019
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.8266048109948687,
      "learning_rate": 1.685268291719999e-07,
      "loss": 0.1471,
      "step": 10020
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.5191067628851593,
      "learning_rate": 1.6796990016515914e-07,
      "loss": 0.2117,
      "step": 10021
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.1299429159814425,
      "learning_rate": 1.6741388512684586e-07,
      "loss": 0.1909,
      "step": 10022
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.1691151357879894,
      "learning_rate": 1.6685878410874768e-07,
      "loss": 0.2168,
      "step": 10023
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.032325561263658,
      "learning_rate": 1.663045971624666e-07,
      "loss": 0.1938,
      "step": 10024
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.01754350394184,
      "learning_rate": 1.657513243395159e-07,
      "loss": 0.2058,
      "step": 10025
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.1084454666591315,
      "learning_rate": 1.6519896569132886e-07,
      "loss": 0.1755,
      "step": 10026
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.063567563000209,
      "learning_rate": 1.646475212692511e-07,
      "loss": 0.198,
      "step": 10027
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.9181674997212943,
      "learning_rate": 1.640969911245438e-07,
      "loss": 0.1559,
      "step": 10028
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.2624437693280908,
      "learning_rate": 1.6354737530838494e-07,
      "loss": 0.2178,
      "step": 10029
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.1496299330016706,
      "learning_rate": 1.6299867387186363e-07,
      "loss": 0.1646,
      "step": 10030
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.9875180832741646,
      "learning_rate": 1.6245088686598686e-07,
      "loss": 0.1938,
      "step": 10031
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.291780868852863,
      "learning_rate": 1.6190401434167725e-07,
      "loss": 0.2041,
      "step": 10032
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.1598247662914654,
      "learning_rate": 1.6135805634976966e-07,
      "loss": 0.2135,
      "step": 10033
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.9592620397532248,
      "learning_rate": 1.608130129410157e-07,
      "loss": 0.1697,
      "step": 10034
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.9784403007064288,
      "learning_rate": 1.6026888416608267e-07,
      "loss": 0.2085,
      "step": 10035
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.1115151786551736,
      "learning_rate": 1.5972567007555008e-07,
      "loss": 0.1885,
      "step": 10036
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.6883789307856496,
      "learning_rate": 1.591833707199153e-07,
      "loss": 0.1196,
      "step": 10037
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.0334415274225006,
      "learning_rate": 1.5864198614959025e-07,
      "loss": 0.1875,
      "step": 10038
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.3073826128913884,
      "learning_rate": 1.5810151641489912e-07,
      "loss": 0.2476,
      "step": 10039
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.8599696140262805,
      "learning_rate": 1.5756196156608393e-07,
      "loss": 0.2173,
      "step": 10040
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.120134594900854,
      "learning_rate": 1.5702332165330348e-07,
      "loss": 0.166,
      "step": 10041
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.166414425466234,
      "learning_rate": 1.5648559672662322e-07,
      "loss": 0.1628,
      "step": 10042
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.1550305190645505,
      "learning_rate": 1.559487868360343e-07,
      "loss": 0.1672,
      "step": 10043
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.940320868359509,
      "learning_rate": 1.554128920314346e-07,
      "loss": 0.1621,
      "step": 10044
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.185365741038542,
      "learning_rate": 1.5487791236264095e-07,
      "loss": 0.2302,
      "step": 10045
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.8634503032903593,
      "learning_rate": 1.543438478793846e-07,
      "loss": 0.1772,
      "step": 10046
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.2305115130056725,
      "learning_rate": 1.5381069863131037e-07,
      "loss": 0.1909,
      "step": 10047
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.8603165099194248,
      "learning_rate": 1.5327846466797857e-07,
      "loss": 0.189,
      "step": 10048
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.9648146666794721,
      "learning_rate": 1.5274714603886742e-07,
      "loss": 0.1713,
      "step": 10049
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.9767136802272531,
      "learning_rate": 1.5221674279336408e-07,
      "loss": 0.1775,
      "step": 10050
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.396174767037977,
      "learning_rate": 1.5168725498077574e-07,
      "loss": 0.2083,
      "step": 10051
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.7791831535198384,
      "learning_rate": 1.5115868265032195e-07,
      "loss": 0.1545,
      "step": 10052
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.4077306944006724,
      "learning_rate": 1.5063102585113786e-07,
      "loss": 0.2039,
      "step": 10053
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.3730448282431076,
      "learning_rate": 1.5010428463227423e-07,
      "loss": 0.2134,
      "step": 10054
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.8397253630469546,
      "learning_rate": 1.495784590426963e-07,
      "loss": 0.1697,
      "step": 10055
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.094386755154773,
      "learning_rate": 1.4905354913128279e-07,
      "loss": 0.1821,
      "step": 10056
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.137332507889821,
      "learning_rate": 1.4852955494682798e-07,
      "loss": 0.1721,
      "step": 10057
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.2619468165827277,
      "learning_rate": 1.4800647653804289e-07,
      "loss": 0.1793,
      "step": 10058
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.084182952334979,
      "learning_rate": 1.4748431395355088e-07,
      "loss": 0.1759,
      "step": 10059
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.19301582122914,
      "learning_rate": 1.4696306724189312e-07,
      "loss": 0.1865,
      "step": 10060
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.4724612750425115,
      "learning_rate": 1.4644273645152196e-07,
      "loss": 0.2261,
      "step": 10061
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.072770982822197,
      "learning_rate": 1.4592332163080648e-07,
      "loss": 0.2212,
      "step": 10062
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.9663144423394074,
      "learning_rate": 1.4540482282803136e-07,
      "loss": 0.1982,
      "step": 10063
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.1056983984280238,
      "learning_rate": 1.4488724009139588e-07,
      "loss": 0.2078,
      "step": 10064
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.956822833098971,
      "learning_rate": 1.4437057346901152e-07,
      "loss": 0.1565,
      "step": 10065
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.3139588632261208,
      "learning_rate": 1.4385482300890873e-07,
      "loss": 0.2422,
      "step": 10066
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.6666017917306808,
      "learning_rate": 1.4333998875903032e-07,
      "loss": 0.1179,
      "step": 10067
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.1699349711925504,
      "learning_rate": 1.4282607076723355e-07,
      "loss": 0.2036,
      "step": 10068
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.4089097614576556,
      "learning_rate": 1.423130690812924e-07,
      "loss": 0.2617,
      "step": 10069
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.216377126559869,
      "learning_rate": 1.4180098374889429e-07,
      "loss": 0.2346,
      "step": 10070
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.1492055293172485,
      "learning_rate": 1.4128981481764115e-07,
      "loss": 0.1938,
      "step": 10071
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.5767944562857124,
      "learning_rate": 1.4077956233505163e-07,
      "loss": 0.2507,
      "step": 10072
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.9883490166734568,
      "learning_rate": 1.402702263485567e-07,
      "loss": 0.1693,
      "step": 10073
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.9651447737791625,
      "learning_rate": 1.3976180690550402e-07,
      "loss": 0.167,
      "step": 10074
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.127398345175242,
      "learning_rate": 1.3925430405315577e-07,
      "loss": 0.2134,
      "step": 10075
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.0744102377106866,
      "learning_rate": 1.3874771783868758e-07,
      "loss": 0.2058,
      "step": 10076
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.3908005647686412,
      "learning_rate": 1.3824204830918952e-07,
      "loss": 0.2466,
      "step": 10077
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.7224478533636052,
      "learning_rate": 1.3773729551167182e-07,
      "loss": 0.1395,
      "step": 10078
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.8345457661097455,
      "learning_rate": 1.3723345949305245e-07,
      "loss": 0.1571,
      "step": 10079
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.160803722814032,
      "learning_rate": 1.367305403001673e-07,
      "loss": 0.2178,
      "step": 10080
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.9832594423732115,
      "learning_rate": 1.3622853797976786e-07,
      "loss": 0.1611,
      "step": 10081
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.4922857459663597,
      "learning_rate": 1.3572745257851792e-07,
      "loss": 0.2266,
      "step": 10082
    },
    {
      "epoch": 2.85,
      "grad_norm": 5.97955551965406,
      "learning_rate": 1.3522728414299911e-07,
      "loss": 0.1899,
      "step": 10083
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.045574965768253,
      "learning_rate": 1.3472803271970536e-07,
      "loss": 0.2104,
      "step": 10084
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.6125057422601183,
      "learning_rate": 1.342296983550462e-07,
      "loss": 0.2693,
      "step": 10085
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.6118642751784193,
      "learning_rate": 1.3373228109534675e-07,
      "loss": 0.2307,
      "step": 10086
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.7215919161286148,
      "learning_rate": 1.3323578098684565e-07,
      "loss": 0.1442,
      "step": 10087
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.7961157162129582,
      "learning_rate": 1.3274019807569593e-07,
      "loss": 0.1493,
      "step": 10088
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.0589992916357867,
      "learning_rate": 1.3224553240796633e-07,
      "loss": 0.1423,
      "step": 10089
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.8858190973473847,
      "learning_rate": 1.3175178402964116e-07,
      "loss": 0.1664,
      "step": 10090
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.652777268984285,
      "learning_rate": 1.3125895298661705e-07,
      "loss": 0.2515,
      "step": 10091
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.6229745063981844,
      "learning_rate": 1.3076703932470958e-07,
      "loss": 0.2561,
      "step": 10092
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.107233457701216,
      "learning_rate": 1.3027604308964215e-07,
      "loss": 0.2037,
      "step": 10093
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.9253300382310146,
      "learning_rate": 1.2978596432705826e-07,
      "loss": 0.2544,
      "step": 10094
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.9846277734361004,
      "learning_rate": 1.292968030825159e-07,
      "loss": 0.2299,
      "step": 10095
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.8598333676288648,
      "learning_rate": 1.288085594014865e-07,
      "loss": 0.1489,
      "step": 10096
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.9979201241151383,
      "learning_rate": 1.2832123332935598e-07,
      "loss": 0.1948,
      "step": 10097
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.6768156753821486,
      "learning_rate": 1.2783482491142474e-07,
      "loss": 0.1519,
      "step": 10098
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.0640764476995863,
      "learning_rate": 1.2734933419290996e-07,
      "loss": 0.1864,
      "step": 10099
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.8482531023317386,
      "learning_rate": 1.2686476121894e-07,
      "loss": 0.1917,
      "step": 10100
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.8301537902481,
      "learning_rate": 1.263811060345621e-07,
      "loss": 0.1603,
      "step": 10101
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.1756088904181907,
      "learning_rate": 1.2589836868473259e-07,
      "loss": 0.2393,
      "step": 10102
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.9494249357351021,
      "learning_rate": 1.2541654921432998e-07,
      "loss": 0.1643,
      "step": 10103
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.094888297743711,
      "learning_rate": 1.2493564766814292e-07,
      "loss": 0.1694,
      "step": 10104
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.2627655798060857,
      "learning_rate": 1.244556640908712e-07,
      "loss": 0.1912,
      "step": 10105
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.0929568812668453,
      "learning_rate": 1.2397659852713684e-07,
      "loss": 0.1691,
      "step": 10106
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.272508714488953,
      "learning_rate": 1.2349845102147317e-07,
      "loss": 0.1669,
      "step": 10107
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.1928182654413693,
      "learning_rate": 1.2302122161832464e-07,
      "loss": 0.2332,
      "step": 10108
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.3757492941579232,
      "learning_rate": 1.2254491036205797e-07,
      "loss": 0.1799,
      "step": 10109
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.1599308661933727,
      "learning_rate": 1.220695172969477e-07,
      "loss": 0.2222,
      "step": 10110
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.8797925487291995,
      "learning_rate": 1.2159504246718522e-07,
      "loss": 0.1511,
      "step": 10111
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.1455027013616443,
      "learning_rate": 1.2112148591687743e-07,
      "loss": 0.1956,
      "step": 10112
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.1546680245173135,
      "learning_rate": 1.2064884769004692e-07,
      "loss": 0.2166,
      "step": 10113
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.3852236069390567,
      "learning_rate": 1.201771278306263e-07,
      "loss": 0.2009,
      "step": 10114
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.9768187889337854,
      "learning_rate": 1.1970632638246827e-07,
      "loss": 0.1614,
      "step": 10115
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.1914809843963012,
      "learning_rate": 1.192364433893378e-07,
      "loss": 0.199,
      "step": 10116
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.237624212818116,
      "learning_rate": 1.1876747889491225e-07,
      "loss": 0.2151,
      "step": 10117
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5998909381899133,
      "learning_rate": 1.1829943294278778e-07,
      "loss": 0.1323,
      "step": 10118
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.3016315622326133,
      "learning_rate": 1.1783230557647075e-07,
      "loss": 0.188,
      "step": 10119
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.437392194114886,
      "learning_rate": 1.1736609683938749e-07,
      "loss": 0.2502,
      "step": 10120
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.316733491548517,
      "learning_rate": 1.1690080677487558e-07,
      "loss": 0.2065,
      "step": 10121
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.1157723151545627,
      "learning_rate": 1.1643643542618488e-07,
      "loss": 0.2134,
      "step": 10122
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.274856955963372,
      "learning_rate": 1.1597298283648529e-07,
      "loss": 0.2188,
      "step": 10123
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.0422756565161846,
      "learning_rate": 1.1551044904885678e-07,
      "loss": 0.2234,
      "step": 10124
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.216121292083567,
      "learning_rate": 1.1504883410629608e-07,
      "loss": 0.2156,
      "step": 10125
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.9639633040113864,
      "learning_rate": 1.1458813805171665e-07,
      "loss": 0.1904,
      "step": 10126
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.060050486649897,
      "learning_rate": 1.1412836092793977e-07,
      "loss": 0.197,
      "step": 10127
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.9110175494984936,
      "learning_rate": 1.1366950277770794e-07,
      "loss": 0.1658,
      "step": 10128
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.8358713265389972,
      "learning_rate": 1.1321156364367591e-07,
      "loss": 0.1543,
      "step": 10129
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.9343781993384765,
      "learning_rate": 1.1275454356841298e-07,
      "loss": 0.1923,
      "step": 10130
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.2889811804163327,
      "learning_rate": 1.1229844259440182e-07,
      "loss": 0.1929,
      "step": 10131
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.079270363105347,
      "learning_rate": 1.1184326076404073e-07,
      "loss": 0.1968,
      "step": 10132
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.4091062374318293,
      "learning_rate": 1.1138899811964477e-07,
      "loss": 0.1864,
      "step": 10133
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.1610693006820343,
      "learning_rate": 1.1093565470343904e-07,
      "loss": 0.2397,
      "step": 10134
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.2777135885633797,
      "learning_rate": 1.1048323055756649e-07,
      "loss": 0.1764,
      "step": 10135
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.0263225210360707,
      "learning_rate": 1.1003172572408349e-07,
      "loss": 0.1799,
      "step": 10136
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.738934908208488,
      "learning_rate": 1.0958114024496202e-07,
      "loss": 0.2317,
      "step": 10137
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.7499292274923828,
      "learning_rate": 1.0913147416208636e-07,
      "loss": 0.1575,
      "step": 10138
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.18573909121046,
      "learning_rate": 1.086827275172575e-07,
      "loss": 0.1689,
      "step": 10139
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.3121721073431787,
      "learning_rate": 1.0823490035218986e-07,
      "loss": 0.1865,
      "step": 10140
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.1885739904363533,
      "learning_rate": 1.0778799270851348e-07,
      "loss": 0.2019,
      "step": 10141
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.8601860274384547,
      "learning_rate": 1.0734200462777178e-07,
      "loss": 0.2827,
      "step": 10142
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.6140984383733148,
      "learning_rate": 1.068969361514216e-07,
      "loss": 0.1611,
      "step": 10143
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.1381989895571385,
      "learning_rate": 1.0645278732083763e-07,
      "loss": 0.2258,
      "step": 10144
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.386893091428355,
      "learning_rate": 1.0600955817730573e-07,
      "loss": 0.272,
      "step": 10145
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.9884557202166804,
      "learning_rate": 1.055672487620285e-07,
      "loss": 0.1429,
      "step": 10146
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.023588971655985,
      "learning_rate": 1.0512585911612416e-07,
      "loss": 0.1743,
      "step": 10147
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.0803140008418195,
      "learning_rate": 1.0468538928061878e-07,
      "loss": 0.1743,
      "step": 10148
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.967688492021202,
      "learning_rate": 1.0424583929646181e-07,
      "loss": 0.2048,
      "step": 10149
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.935923758652674,
      "learning_rate": 1.038072092045117e-07,
      "loss": 0.197,
      "step": 10150
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.3117235645232137,
      "learning_rate": 1.033694990455425e-07,
      "loss": 0.1521,
      "step": 10151
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.0934115427085245,
      "learning_rate": 1.0293270886024276e-07,
      "loss": 0.1821,
      "step": 10152
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.4209770426046013,
      "learning_rate": 1.0249683868921667e-07,
      "loss": 0.209,
      "step": 10153
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.968204061081371,
      "learning_rate": 1.0206188857298182e-07,
      "loss": 0.1527,
      "step": 10154
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.960952591984709,
      "learning_rate": 1.0162785855197032e-07,
      "loss": 0.1422,
      "step": 10155
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.9720913222230463,
      "learning_rate": 1.0119474866652767e-07,
      "loss": 0.1606,
      "step": 10156
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.0444438551097486,
      "learning_rate": 1.0076255895691611e-07,
      "loss": 0.1587,
      "step": 10157
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.8981996201278857,
      "learning_rate": 1.0033128946331128e-07,
      "loss": 0.207,
      "step": 10158
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.879870713601382,
      "learning_rate": 9.990094022580332e-08,
      "loss": 0.1721,
      "step": 10159
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.1651800352943016,
      "learning_rate": 9.947151128439692e-08,
      "loss": 0.1895,
      "step": 10160
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.1971777313923386,
      "learning_rate": 9.904300267901012e-08,
      "loss": 0.2095,
      "step": 10161
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.2594816115812577,
      "learning_rate": 9.861541444947554e-08,
      "loss": 0.2312,
      "step": 10162
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.1470887619156844,
      "learning_rate": 9.818874663554356e-08,
      "loss": 0.2012,
      "step": 10163
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.4290570616104756,
      "learning_rate": 9.776299927687694e-08,
      "loss": 0.2695,
      "step": 10164
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.0030854517209793,
      "learning_rate": 9.733817241304844e-08,
      "loss": 0.1877,
      "step": 10165
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.298548812225808,
      "learning_rate": 9.691426608355203e-08,
      "loss": 0.1958,
      "step": 10166
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.73092652691286,
      "learning_rate": 9.649128032779287e-08,
      "loss": 0.1356,
      "step": 10167
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.9009113244029163,
      "learning_rate": 9.606921518509172e-08,
      "loss": 0.1729,
      "step": 10168
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.7469835890457415,
      "learning_rate": 9.564807069468163e-08,
      "loss": 0.1467,
      "step": 10169
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.9025748900705013,
      "learning_rate": 9.52278468957124e-08,
      "loss": 0.1638,
      "step": 10170
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.1932963013781825,
      "learning_rate": 9.480854382724613e-08,
      "loss": 0.2153,
      "step": 10171
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.1942580067714914,
      "learning_rate": 9.439016152826275e-08,
      "loss": 0.1903,
      "step": 10172
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.3701712881049417,
      "learning_rate": 9.397270003765224e-08,
      "loss": 0.2642,
      "step": 10173
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.63331766603075,
      "learning_rate": 9.355615939422135e-08,
      "loss": 0.1411,
      "step": 10174
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.2908108587897944,
      "learning_rate": 9.314053963669245e-08,
      "loss": 0.1619,
      "step": 10175
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.9894036076291008,
      "learning_rate": 9.272584080370018e-08,
      "loss": 0.1713,
      "step": 10176
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.4122389770011456,
      "learning_rate": 9.231206293379257e-08,
      "loss": 0.2178,
      "step": 10177
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.790432870161936,
      "learning_rate": 9.189920606543556e-08,
      "loss": 0.1639,
      "step": 10178
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.8688946035501535,
      "learning_rate": 9.148727023700731e-08,
      "loss": 0.1482,
      "step": 10179
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.1724574455962395,
      "learning_rate": 9.107625548679944e-08,
      "loss": 0.2544,
      "step": 10180
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5742738126473603,
      "learning_rate": 9.066616185302246e-08,
      "loss": 0.1272,
      "step": 10181
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.7743566817709695,
      "learning_rate": 9.025698937379368e-08,
      "loss": 0.1498,
      "step": 10182
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.8140558301511613,
      "learning_rate": 8.984873808715155e-08,
      "loss": 0.1605,
      "step": 10183
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.175273531348597,
      "learning_rate": 8.944140803104573e-08,
      "loss": 0.2009,
      "step": 10184
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.6005075638544506,
      "learning_rate": 8.903499924334147e-08,
      "loss": 0.2361,
      "step": 10185
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.0911112884509437,
      "learning_rate": 8.862951176181744e-08,
      "loss": 0.1619,
      "step": 10186
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.0140583438368242,
      "learning_rate": 8.822494562416684e-08,
      "loss": 0.1958,
      "step": 10187
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.427871024042754,
      "learning_rate": 8.782130086799734e-08,
      "loss": 0.2146,
      "step": 10188
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.335451637710792,
      "learning_rate": 8.741857753083228e-08,
      "loss": 0.2151,
      "step": 10189
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.2754541516317635,
      "learning_rate": 8.701677565010725e-08,
      "loss": 0.1681,
      "step": 10190
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.005674696750985,
      "learning_rate": 8.661589526317238e-08,
      "loss": 0.1505,
      "step": 10191
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.9131402326908165,
      "learning_rate": 8.621593640729343e-08,
      "loss": 0.1282,
      "step": 10192
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.162410316313604,
      "learning_rate": 8.581689911965063e-08,
      "loss": 0.1721,
      "step": 10193
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.1016495547689233,
      "learning_rate": 8.541878343733656e-08,
      "loss": 0.198,
      "step": 10194
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.0542349119477157,
      "learning_rate": 8.502158939736049e-08,
      "loss": 0.192,
      "step": 10195
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.9846617202318186,
      "learning_rate": 8.46253170366429e-08,
      "loss": 0.1975,
      "step": 10196
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.2616980666241693,
      "learning_rate": 8.422996639202318e-08,
      "loss": 0.2229,
      "step": 10197
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.081059277613427,
      "learning_rate": 8.383553750025198e-08,
      "loss": 0.167,
      "step": 10198
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.2111285861095085,
      "learning_rate": 8.344203039799214e-08,
      "loss": 0.1987,
      "step": 10199
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.7997776123214921,
      "learning_rate": 8.304944512182666e-08,
      "loss": 0.1902,
      "step": 10200
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.5806915253884344,
      "learning_rate": 8.265778170824746e-08,
      "loss": 0.2468,
      "step": 10201
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.6025428830033643,
      "learning_rate": 8.226704019366427e-08,
      "loss": 0.1512,
      "step": 10202
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.7726833734638379,
      "learning_rate": 8.187722061439806e-08,
      "loss": 0.1626,
      "step": 10203
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.9561361867306246,
      "learning_rate": 8.148832300668763e-08,
      "loss": 0.1943,
      "step": 10204
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.485270828406276,
      "learning_rate": 8.110034740668293e-08,
      "loss": 0.2507,
      "step": 10205
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.9089027269277066,
      "learning_rate": 8.071329385045068e-08,
      "loss": 0.1685,
      "step": 10206
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.7048965325278171,
      "learning_rate": 8.032716237396987e-08,
      "loss": 0.1425,
      "step": 10207
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.0876778104736204,
      "learning_rate": 7.9941953013134e-08,
      "loss": 0.1803,
      "step": 10208
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.052336102718799,
      "learning_rate": 7.955766580375334e-08,
      "loss": 0.1772,
      "step": 10209
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.0675169019212567,
      "learning_rate": 7.91743007815493e-08,
      "loss": 0.2209,
      "step": 10210
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.006279569339244,
      "learning_rate": 7.879185798215894e-08,
      "loss": 0.1699,
      "step": 10211
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.2488098898488955,
      "learning_rate": 7.841033744113268e-08,
      "loss": 0.2173,
      "step": 10212
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.040357443930855,
      "learning_rate": 7.802973919393775e-08,
      "loss": 0.1765,
      "step": 10213
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.6705434476239747,
      "learning_rate": 7.765006327595248e-08,
      "loss": 0.2032,
      "step": 10214
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.1366365595005186,
      "learning_rate": 7.727130972247199e-08,
      "loss": 0.1826,
      "step": 10215
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.7575836978218313,
      "learning_rate": 7.689347856870366e-08,
      "loss": 0.1699,
      "step": 10216
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.8852997729989898,
      "learning_rate": 7.651656984977051e-08,
      "loss": 0.199,
      "step": 10217
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.135316398222401,
      "learning_rate": 7.614058360070897e-08,
      "loss": 0.1957,
      "step": 10218
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.7605099079580258,
      "learning_rate": 7.576551985647107e-08,
      "loss": 0.1545,
      "step": 10219
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.0328237219313716,
      "learning_rate": 7.539137865192003e-08,
      "loss": 0.167,
      "step": 10220
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.9526192630765946,
      "learning_rate": 7.501816002183803e-08,
      "loss": 0.186,
      "step": 10221
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.328881232136776,
      "learning_rate": 7.464586400091623e-08,
      "loss": 0.2053,
      "step": 10222
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.5069358037467984,
      "learning_rate": 7.427449062376468e-08,
      "loss": 0.238,
      "step": 10223
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.9158248355033272,
      "learning_rate": 7.390403992490358e-08,
      "loss": 0.1829,
      "step": 10224
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.186485365916272,
      "learning_rate": 7.353451193877092e-08,
      "loss": 0.1699,
      "step": 10225
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.3384668985239743,
      "learning_rate": 7.316590669971813e-08,
      "loss": 0.217,
      "step": 10226
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.6999024210617835,
      "learning_rate": 7.279822424200889e-08,
      "loss": 0.1609,
      "step": 10227
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.1427522551200946,
      "learning_rate": 7.243146459982142e-08,
      "loss": 0.1818,
      "step": 10228
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.129965165908953,
      "learning_rate": 7.206562780725068e-08,
      "loss": 0.1794,
      "step": 10229
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.716683036939545,
      "learning_rate": 7.17007138983028e-08,
      "loss": 0.1427,
      "step": 10230
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.684382188767298,
      "learning_rate": 7.133672290690064e-08,
      "loss": 0.1605,
      "step": 10231
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.9221775474981289,
      "learning_rate": 7.097365486688158e-08,
      "loss": 0.1902,
      "step": 10232
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.0651118053778363,
      "learning_rate": 7.061150981199194e-08,
      "loss": 0.1677,
      "step": 10233
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.0559077198735984,
      "learning_rate": 7.025028777589926e-08,
      "loss": 0.1672,
      "step": 10234
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.234267413288985,
      "learning_rate": 6.988998879218111e-08,
      "loss": 0.2534,
      "step": 10235
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.9884986947931487,
      "learning_rate": 6.953061289433072e-08,
      "loss": 0.176,
      "step": 10236
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.302550419642787,
      "learning_rate": 6.917216011575357e-08,
      "loss": 0.1724,
      "step": 10237
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.225840430538814,
      "learning_rate": 6.8814630489773e-08,
      "loss": 0.1703,
      "step": 10238
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.0383954525825185,
      "learning_rate": 6.845802404962243e-08,
      "loss": 0.178,
      "step": 10239
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.9119666045883306,
      "learning_rate": 6.810234082845313e-08,
      "loss": 0.1605,
      "step": 10240
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.026626965718169,
      "learning_rate": 6.774758085932642e-08,
      "loss": 0.2148,
      "step": 10241
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.8092589465156128,
      "learning_rate": 6.73937441752226e-08,
      "loss": 0.1433,
      "step": 10242
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.2670804372512143,
      "learning_rate": 6.704083080903201e-08,
      "loss": 0.186,
      "step": 10243
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.8843526874092478,
      "learning_rate": 6.668884079356287e-08,
      "loss": 0.1663,
      "step": 10244
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.021853471540036,
      "learning_rate": 6.633777416153232e-08,
      "loss": 0.1663,
      "step": 10245
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.0157121807343024,
      "learning_rate": 6.59876309455787e-08,
      "loss": 0.2048,
      "step": 10246
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.4156873687700324,
      "learning_rate": 6.56384111782482e-08,
      "loss": 0.2368,
      "step": 10247
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.123093889202099,
      "learning_rate": 6.529011489200377e-08,
      "loss": 0.1846,
      "step": 10248
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.0305685930596646,
      "learning_rate": 6.494274211922392e-08,
      "loss": 0.1887,
      "step": 10249
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.095868802539364,
      "learning_rate": 6.459629289219838e-08,
      "loss": 0.217,
      "step": 10250
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.067802857195718,
      "learning_rate": 6.425076724313251e-08,
      "loss": 0.1897,
      "step": 10251
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.864866280811196,
      "learning_rate": 6.390616520414617e-08,
      "loss": 0.1587,
      "step": 10252
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.9845847742443106,
      "learning_rate": 6.356248680727484e-08,
      "loss": 0.1895,
      "step": 10253
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.7673339407154514,
      "learning_rate": 6.321973208446298e-08,
      "loss": 0.1704,
      "step": 10254
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.164196618334475,
      "learning_rate": 6.287790106757396e-08,
      "loss": 0.1934,
      "step": 10255
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.4324615361823305,
      "learning_rate": 6.253699378838462e-08,
      "loss": 0.2131,
      "step": 10256
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.3677701187233153,
      "learning_rate": 6.219701027858405e-08,
      "loss": 0.2419,
      "step": 10257
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.5098294498776297,
      "learning_rate": 6.185795056977695e-08,
      "loss": 0.2141,
      "step": 10258
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.35766545099938,
      "learning_rate": 6.151981469348034e-08,
      "loss": 0.2659,
      "step": 10259
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.0122684971829594,
      "learning_rate": 6.118260268112908e-08,
      "loss": 0.1383,
      "step": 10260
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.8973259869251902,
      "learning_rate": 6.084631456406919e-08,
      "loss": 0.1814,
      "step": 10261
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.0762220687737374,
      "learning_rate": 6.051095037356013e-08,
      "loss": 0.2119,
      "step": 10262
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.2088955605320844,
      "learning_rate": 6.017651014077807e-08,
      "loss": 0.1831,
      "step": 10263
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.201603562143851,
      "learning_rate": 5.984299389681148e-08,
      "loss": 0.1963,
      "step": 10264
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.0988988754400806,
      "learning_rate": 5.951040167266331e-08,
      "loss": 0.1938,
      "step": 10265
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.1463466461718874,
      "learning_rate": 5.9178733499251073e-08,
      "loss": 0.1926,
      "step": 10266
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.048035216679142,
      "learning_rate": 5.884798940740566e-08,
      "loss": 0.1957,
      "step": 10267
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.039723912370669,
      "learning_rate": 5.8518169427873584e-08,
      "loss": 0.1968,
      "step": 10268
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.1621987660482076,
      "learning_rate": 5.8189273591312546e-08,
      "loss": 0.207,
      "step": 10269
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.1514568228591524,
      "learning_rate": 5.786130192829809e-08,
      "loss": 0.231,
      "step": 10270
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.1356982448860165,
      "learning_rate": 5.753425446931582e-08,
      "loss": 0.2114,
      "step": 10271
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.954252700347012,
      "learning_rate": 5.7208131244769206e-08,
      "loss": 0.1809,
      "step": 10272
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.9245085613799777,
      "learning_rate": 5.688293228497399e-08,
      "loss": 0.1812,
      "step": 10273
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.0630551369229324,
      "learning_rate": 5.655865762015822e-08,
      "loss": 0.2085,
      "step": 10274
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.8482087802079783,
      "learning_rate": 5.623530728046889e-08,
      "loss": 0.2117,
      "step": 10275
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.2834819809191083,
      "learning_rate": 5.591288129596195e-08,
      "loss": 0.2009,
      "step": 10276
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.963461686140153,
      "learning_rate": 5.559137969660899e-08,
      "loss": 0.1858,
      "step": 10277
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.7396519923050657,
      "learning_rate": 5.527080251229833e-08,
      "loss": 0.1763,
      "step": 10278
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.622148248591686,
      "learning_rate": 5.495114977282945e-08,
      "loss": 0.1461,
      "step": 10279
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.8866333568998446,
      "learning_rate": 5.4632421507916366e-08,
      "loss": 0.1548,
      "step": 10280
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.0528886728986557,
      "learning_rate": 5.43146177471876e-08,
      "loss": 0.1667,
      "step": 10281
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.2851815799532256,
      "learning_rate": 5.3997738520186194e-08,
      "loss": 0.1721,
      "step": 10282
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.009505287964675,
      "learning_rate": 5.3681783856367466e-08,
      "loss": 0.1816,
      "step": 10283
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.2986153574908097,
      "learning_rate": 5.336675378510348e-08,
      "loss": 0.2104,
      "step": 10284
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.2882391450860746,
      "learning_rate": 5.305264833567747e-08,
      "loss": 0.2109,
      "step": 10285
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.9828467371170209,
      "learning_rate": 5.273946753728942e-08,
      "loss": 0.1743,
      "step": 10286
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.517956335414278,
      "learning_rate": 5.2427211419051605e-08,
      "loss": 0.2349,
      "step": 10287
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.0915480129707205,
      "learning_rate": 5.2115880009990796e-08,
      "loss": 0.2065,
      "step": 10288
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.297995889043212,
      "learning_rate": 5.1805473339047173e-08,
      "loss": 0.1934,
      "step": 10289
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.9491980843970669,
      "learning_rate": 5.1495991435076555e-08,
      "loss": 0.2019,
      "step": 10290
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.120956074726143,
      "learning_rate": 5.118743432684814e-08,
      "loss": 0.1849,
      "step": 10291
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.0472456775188226,
      "learning_rate": 5.0879802043043434e-08,
      "loss": 0.2349,
      "step": 10292
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.066370834210357,
      "learning_rate": 5.0573094612260676e-08,
      "loss": 0.1968,
      "step": 10293
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.215033449820876,
      "learning_rate": 5.0267312063009275e-08,
      "loss": 0.1787,
      "step": 10294
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.290019596739523,
      "learning_rate": 4.99624544237165e-08,
      "loss": 0.186,
      "step": 10295
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.1188596783972584,
      "learning_rate": 4.9658521722719676e-08,
      "loss": 0.1833,
      "step": 10296
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.8078693758448334,
      "learning_rate": 4.9355513988272874e-08,
      "loss": 0.1383,
      "step": 10297
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.014625668476874,
      "learning_rate": 4.905343124854245e-08,
      "loss": 0.1737,
      "step": 10298
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.0934436258338196,
      "learning_rate": 4.8752273531609276e-08,
      "loss": 0.1821,
      "step": 10299
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.099371067424443,
      "learning_rate": 4.845204086546984e-08,
      "loss": 0.1919,
      "step": 10300
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.481674785607535,
      "learning_rate": 4.815273327803183e-08,
      "loss": 0.2288,
      "step": 10301
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.127211493782066,
      "learning_rate": 4.785435079711853e-08,
      "loss": 0.2039,
      "step": 10302
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.159726251597992,
      "learning_rate": 4.7556893450466656e-08,
      "loss": 0.214,
      "step": 10303
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.9640531127803609,
      "learning_rate": 4.7260361265729635e-08,
      "loss": 0.1635,
      "step": 10304
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.1682641250826356,
      "learning_rate": 4.696475427047098e-08,
      "loss": 0.197,
      "step": 10305
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.9166421671001856,
      "learning_rate": 4.66700724921687e-08,
      "loss": 0.175,
      "step": 10306
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.3406380117426187,
      "learning_rate": 4.6376315958218676e-08,
      "loss": 0.2129,
      "step": 10307
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.1384175710886484,
      "learning_rate": 4.608348469592461e-08,
      "loss": 0.2292,
      "step": 10308
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.1714774847507035,
      "learning_rate": 4.5791578732510276e-08,
      "loss": 0.1985,
      "step": 10309
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.3542233165616144,
      "learning_rate": 4.5500598095110645e-08,
      "loss": 0.1995,
      "step": 10310
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.9339937214721616,
      "learning_rate": 4.5210542810771864e-08,
      "loss": 0.1814,
      "step": 10311
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.419122568615259,
      "learning_rate": 4.492141290646124e-08,
      "loss": 0.24,
      "step": 10312
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.9081002826844677,
      "learning_rate": 4.463320840905283e-08,
      "loss": 0.1578,
      "step": 10313
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.1441806721095595,
      "learning_rate": 4.4345929345337436e-08,
      "loss": 0.2063,
      "step": 10314
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.864568569664085,
      "learning_rate": 4.405957574202147e-08,
      "loss": 0.1367,
      "step": 10315
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.9347555517727353,
      "learning_rate": 4.377414762572474e-08,
      "loss": 0.156,
      "step": 10316
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.7272870816351515,
      "learning_rate": 4.348964502297714e-08,
      "loss": 0.3262,
      "step": 10317
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.9525051729314857,
      "learning_rate": 4.320606796022864e-08,
      "loss": 0.1591,
      "step": 10318
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.1489893917060865,
      "learning_rate": 4.292341646383813e-08,
      "loss": 0.198,
      "step": 10319
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.129465045163355,
      "learning_rate": 4.264169056008016e-08,
      "loss": 0.1592,
      "step": 10320
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.963974698337736,
      "learning_rate": 4.23608902751449e-08,
      "loss": 0.1522,
      "step": 10321
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.045885534661652,
      "learning_rate": 4.208101563513367e-08,
      "loss": 0.1799,
      "step": 10322
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.910120022084214,
      "learning_rate": 4.1802066666064564e-08,
      "loss": 0.1685,
      "step": 10323
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.0708094505322183,
      "learning_rate": 4.152404339386795e-08,
      "loss": 0.2074,
      "step": 10324
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.328340799088711,
      "learning_rate": 4.1246945844387595e-08,
      "loss": 0.1924,
      "step": 10325
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.772781440234467,
      "learning_rate": 4.097077404338179e-08,
      "loss": 0.157,
      "step": 10326
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.1240480024644284,
      "learning_rate": 4.069552801652443e-08,
      "loss": 0.2017,
      "step": 10327
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.167986055089776,
      "learning_rate": 4.042120778939951e-08,
      "loss": 0.1682,
      "step": 10328
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.102554498526923,
      "learning_rate": 4.014781338751106e-08,
      "loss": 0.1892,
      "step": 10329
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.4497233312688214,
      "learning_rate": 3.987534483626987e-08,
      "loss": 0.208,
      "step": 10330
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.966166402658228,
      "learning_rate": 3.9603802161005675e-08,
      "loss": 0.176,
      "step": 10331
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.7078663115546426,
      "learning_rate": 3.933318538695941e-08,
      "loss": 0.1084,
      "step": 10332
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.7383537527564388,
      "learning_rate": 3.906349453928981e-08,
      "loss": 0.1365,
      "step": 10333
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.9807477127388904,
      "learning_rate": 3.8794729643064635e-08,
      "loss": 0.1904,
      "step": 10334
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.1055195223146335,
      "learning_rate": 3.852689072326832e-08,
      "loss": 0.1727,
      "step": 10335
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.1035838517843186,
      "learning_rate": 3.8259977804797624e-08,
      "loss": 0.1792,
      "step": 10336
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.0061103002089724,
      "learning_rate": 3.799399091246603e-08,
      "loss": 0.1936,
      "step": 10337
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.11797886257108,
      "learning_rate": 3.772893007100042e-08,
      "loss": 0.198,
      "step": 10338
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.1140143990359404,
      "learning_rate": 3.7464795305036664e-08,
      "loss": 0.171,
      "step": 10339
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.218057087690192,
      "learning_rate": 3.720158663913065e-08,
      "loss": 0.2505,
      "step": 10340
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.8241590215934727,
      "learning_rate": 3.693930409774948e-08,
      "loss": 0.1693,
      "step": 10341
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.1747728029506663,
      "learning_rate": 3.6677947705273664e-08,
      "loss": 0.1902,
      "step": 10342
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.4743913849647234,
      "learning_rate": 3.641751748600042e-08,
      "loss": 0.2102,
      "step": 10343
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.1234759372862553,
      "learning_rate": 3.6158013464137056e-08,
      "loss": 0.1938,
      "step": 10344
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.4843702152910083,
      "learning_rate": 3.589943566380649e-08,
      "loss": 0.2385,
      "step": 10345
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.228402722680877,
      "learning_rate": 3.5641784109047286e-08,
      "loss": 0.2166,
      "step": 10346
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.7862922051294787,
      "learning_rate": 3.538505882380916e-08,
      "loss": 0.151,
      "step": 10347
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.9332652398206869,
      "learning_rate": 3.5129259831956366e-08,
      "loss": 0.145,
      "step": 10348
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.2124640931641775,
      "learning_rate": 3.4874387157268766e-08,
      "loss": 0.229,
      "step": 10349
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.967718592933581,
      "learning_rate": 3.4620440823438517e-08,
      "loss": 0.1309,
      "step": 10350
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.0102737127013652,
      "learning_rate": 3.436742085407119e-08,
      "loss": 0.2278,
      "step": 10351
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.0084728961869045,
      "learning_rate": 3.411532727268796e-08,
      "loss": 0.2073,
      "step": 10352
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.9450565632800754,
      "learning_rate": 3.386416010272342e-08,
      "loss": 0.2,
      "step": 10353
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.346231276636432,
      "learning_rate": 3.361391936752445e-08,
      "loss": 0.241,
      "step": 10354
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.0037424879411483,
      "learning_rate": 3.336460509035355e-08,
      "loss": 0.1926,
      "step": 10355
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.16065459873302,
      "learning_rate": 3.311621729438552e-08,
      "loss": 0.2192,
      "step": 10356
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.4090669470124695,
      "learning_rate": 3.2868756002712997e-08,
      "loss": 0.2103,
      "step": 10357
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.207981282421412,
      "learning_rate": 3.262222123833647e-08,
      "loss": 0.1918,
      "step": 10358
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.8777444845983255,
      "learning_rate": 3.2376613024175384e-08,
      "loss": 0.182,
      "step": 10359
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.2790584584205575,
      "learning_rate": 3.2131931383059256e-08,
      "loss": 0.2063,
      "step": 10360
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.2859174815445704,
      "learning_rate": 3.1888176337734334e-08,
      "loss": 0.1982,
      "step": 10361
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.0534889115391457,
      "learning_rate": 3.1645347910860266e-08,
      "loss": 0.1902,
      "step": 10362
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.19937440318037,
      "learning_rate": 3.140344612500901e-08,
      "loss": 0.1958,
      "step": 10363
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.140461546495638,
      "learning_rate": 3.1162471002668113e-08,
      "loss": 0.1681,
      "step": 10364
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.1752959845096878,
      "learning_rate": 3.092242256623634e-08,
      "loss": 0.2183,
      "step": 10365
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.9914822425807928,
      "learning_rate": 3.0683300838030285e-08,
      "loss": 0.1404,
      "step": 10366
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.1199115913392417,
      "learning_rate": 3.044510584027771e-08,
      "loss": 0.2102,
      "step": 10367
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.2075722236543776,
      "learning_rate": 3.02078375951198e-08,
      "loss": 0.1782,
      "step": 10368
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.3702866031437657,
      "learning_rate": 2.997149612461447e-08,
      "loss": 0.2612,
      "step": 10369
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.0729634532663312,
      "learning_rate": 2.9736081450730813e-08,
      "loss": 0.2019,
      "step": 10370
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.006819173443776,
      "learning_rate": 2.950159359535132e-08,
      "loss": 0.1956,
      "step": 10371
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.911723724163603,
      "learning_rate": 2.9268032580275218e-08,
      "loss": 0.2036,
      "step": 10372
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.277799494878002,
      "learning_rate": 2.9035398427212925e-08,
      "loss": 0.2131,
      "step": 10373
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.7527344955429043,
      "learning_rate": 2.880369115778936e-08,
      "loss": 0.1748,
      "step": 10374
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.964777045225264,
      "learning_rate": 2.8572910793546183e-08,
      "loss": 0.1819,
      "step": 10375
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.5053435882307955,
      "learning_rate": 2.8343057355932902e-08,
      "loss": 0.2754,
      "step": 10376
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.2613059376823794,
      "learning_rate": 2.8114130866319088e-08,
      "loss": 0.2004,
      "step": 10377
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.285031250493038,
      "learning_rate": 2.788613134598328e-08,
      "loss": 0.2417,
      "step": 10378
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.3299495663586973,
      "learning_rate": 2.7659058816121855e-08,
      "loss": 0.2012,
      "step": 10379
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.061554339672505,
      "learning_rate": 2.7432913297841256e-08,
      "loss": 0.2217,
      "step": 10380
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.8230548615391076,
      "learning_rate": 2.7207694812164677e-08,
      "loss": 0.1294,
      "step": 10381
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.400283417391316,
      "learning_rate": 2.6983403380026474e-08,
      "loss": 0.2305,
      "step": 10382
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.14179132918744,
      "learning_rate": 2.676003902227886e-08,
      "loss": 0.1792,
      "step": 10383
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.101482099508795,
      "learning_rate": 2.6537601759684118e-08,
      "loss": 0.1874,
      "step": 10384
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.0426520193959403,
      "learning_rate": 2.6316091612920146e-08,
      "loss": 0.1919,
      "step": 10385
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.3060336353404987,
      "learning_rate": 2.6095508602577147e-08,
      "loss": 0.2214,
      "step": 10386
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.576992821351692,
      "learning_rate": 2.5875852749160936e-08,
      "loss": 0.172,
      "step": 10387
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.5275905424380536,
      "learning_rate": 2.5657124073089622e-08,
      "loss": 0.188,
      "step": 10388
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.326527090733095,
      "learning_rate": 2.5439322594696946e-08,
      "loss": 0.2051,
      "step": 10389
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.1229560336154236,
      "learning_rate": 2.5222448334227822e-08,
      "loss": 0.1746,
      "step": 10390
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.878828384714626,
      "learning_rate": 2.50065013118439e-08,
      "loss": 0.1702,
      "step": 10391
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.1625054376641333,
      "learning_rate": 2.4791481547619123e-08,
      "loss": 0.1681,
      "step": 10392
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.1783202669292323,
      "learning_rate": 2.4577389061539724e-08,
      "loss": 0.2083,
      "step": 10393
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.1971436715083765,
      "learning_rate": 2.4364223873509785e-08,
      "loss": 0.2,
      "step": 10394
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.7894232536971963,
      "learning_rate": 2.4151986003343453e-08,
      "loss": 0.1346,
      "step": 10395
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.3213135869005015,
      "learning_rate": 2.39406754707705e-08,
      "loss": 0.231,
      "step": 10396
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.9192618956740843,
      "learning_rate": 2.3730292295433e-08,
      "loss": 0.2089,
      "step": 10397
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.0507757933796347,
      "learning_rate": 2.352083649688863e-08,
      "loss": 0.1631,
      "step": 10398
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.951756452574515,
      "learning_rate": 2.3312308094607382e-08,
      "loss": 0.1697,
      "step": 10399
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.8188635205132917,
      "learning_rate": 2.3104707107974857e-08,
      "loss": 0.1714,
      "step": 10400
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.9491588281754613,
      "learning_rate": 2.2898033556288946e-08,
      "loss": 0.1882,
      "step": 10401
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.252811297021543,
      "learning_rate": 2.2692287458760953e-08,
      "loss": 0.2139,
      "step": 10402
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.2715007474441866,
      "learning_rate": 2.248746883451669e-08,
      "loss": 0.1919,
      "step": 10403
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.020366330745265,
      "learning_rate": 2.2283577702596482e-08,
      "loss": 0.2001,
      "step": 10404
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.107414715473795,
      "learning_rate": 2.2080614081954054e-08,
      "loss": 0.1709,
      "step": 10405
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.78108722891675,
      "learning_rate": 2.187857799145432e-08,
      "loss": 0.1562,
      "step": 10406
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.8622848781534909,
      "learning_rate": 2.167746944988114e-08,
      "loss": 0.1516,
      "step": 10407
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.5734342969835398,
      "learning_rate": 2.1477288475926227e-08,
      "loss": 0.2458,
      "step": 10408
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.3753801241075285,
      "learning_rate": 2.1278035088200255e-08,
      "loss": 0.2068,
      "step": 10409
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.040938734837763,
      "learning_rate": 2.1079709305226183e-08,
      "loss": 0.1724,
      "step": 10410
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.0950424802441194,
      "learning_rate": 2.088231114543704e-08,
      "loss": 0.1838,
      "step": 10411
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.9982654603554806,
      "learning_rate": 2.0685840627184817e-08,
      "loss": 0.1851,
      "step": 10412
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.908914228448754,
      "learning_rate": 2.049029776873268e-08,
      "loss": 0.1611,
      "step": 10413
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.263399588129245,
      "learning_rate": 2.0295682588257205e-08,
      "loss": 0.1936,
      "step": 10414
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.2269566214277035,
      "learning_rate": 2.010199510385058e-08,
      "loss": 0.2146,
      "step": 10415
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.1375174446665763,
      "learning_rate": 1.9909235333517296e-08,
      "loss": 0.1833,
      "step": 10416
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.0743604773561515,
      "learning_rate": 1.9717403295175242e-08,
      "loss": 0.1946,
      "step": 10417
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.0363820912983575,
      "learning_rate": 1.952649900665793e-08,
      "loss": 0.2374,
      "step": 10418
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.1385625898510736,
      "learning_rate": 1.9336522485710053e-08,
      "loss": 0.2212,
      "step": 10419
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.478151626598777,
      "learning_rate": 1.914747374999304e-08,
      "loss": 0.2356,
      "step": 10420
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.4934922609679364,
      "learning_rate": 1.89593528170795e-08,
      "loss": 0.2012,
      "step": 10421
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.3557373585027817,
      "learning_rate": 1.877215970445767e-08,
      "loss": 0.2581,
      "step": 10422
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.8959080114076696,
      "learning_rate": 1.8585894429528073e-08,
      "loss": 0.1667,
      "step": 10423
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.301934443604737,
      "learning_rate": 1.8400557009605746e-08,
      "loss": 0.1812,
      "step": 10424
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.9705372305977331,
      "learning_rate": 1.821614746191913e-08,
      "loss": 0.1929,
      "step": 10425
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.8828275705719266,
      "learning_rate": 1.8032665803612294e-08,
      "loss": 0.1633,
      "step": 10426
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.063844701559011,
      "learning_rate": 1.7850112051738255e-08,
      "loss": 0.1895,
      "step": 10427
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.252014024712502,
      "learning_rate": 1.7668486223269e-08,
      "loss": 0.1486,
      "step": 10428
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.8147088520633283,
      "learning_rate": 1.7487788335087686e-08,
      "loss": 0.1393,
      "step": 10429
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.0440044271758264,
      "learning_rate": 1.7308018403991988e-08,
      "loss": 0.1515,
      "step": 10430
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.254656017201782,
      "learning_rate": 1.7129176446692986e-08,
      "loss": 0.2068,
      "step": 10431
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.1544185556347033,
      "learning_rate": 1.6951262479815155e-08,
      "loss": 0.1985,
      "step": 10432
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.1392390407872037,
      "learning_rate": 1.6774276519896383e-08,
      "loss": 0.1824,
      "step": 10433
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.9849484039733705,
      "learning_rate": 1.6598218583390168e-08,
      "loss": 0.1626,
      "step": 10434
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.9603729189649977,
      "learning_rate": 1.6423088686662313e-08,
      "loss": 0.1885,
      "step": 10435
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.8144473540984865,
      "learning_rate": 1.624888684599202e-08,
      "loss": 0.1576,
      "step": 10436
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.3499695559943787,
      "learning_rate": 1.6075613077574103e-08,
      "loss": 0.1515,
      "step": 10437
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.2775872793542824,
      "learning_rate": 1.5903267397514576e-08,
      "loss": 0.2205,
      "step": 10438
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.3778038823604595,
      "learning_rate": 1.5731849821833955e-08,
      "loss": 0.2048,
      "step": 10439
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.054596817936993,
      "learning_rate": 1.556136036646838e-08,
      "loss": 0.1608,
      "step": 10440
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.3173007026169077,
      "learning_rate": 1.5391799047266287e-08,
      "loss": 0.2061,
      "step": 10441
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.1814459632618965,
      "learning_rate": 1.522316587998729e-08,
      "loss": 0.2062,
      "step": 10442
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.1263840160153364,
      "learning_rate": 1.5055460880311068e-08,
      "loss": 0.1868,
      "step": 10443
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.9464557945796563,
      "learning_rate": 1.4888684063824044e-08,
      "loss": 0.1776,
      "step": 10444
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.8907605571415869,
      "learning_rate": 1.4722835446030481e-08,
      "loss": 0.1456,
      "step": 10445
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.8457632707708267,
      "learning_rate": 1.4557915042346937e-08,
      "loss": 0.1785,
      "step": 10446
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.2389706414371937,
      "learning_rate": 1.4393922868105591e-08,
      "loss": 0.2449,
      "step": 10447
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.138899281289238,
      "learning_rate": 1.4230858938549808e-08,
      "loss": 0.1873,
      "step": 10448
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.3006868339475597,
      "learning_rate": 1.4068723268837459e-08,
      "loss": 0.2551,
      "step": 10449
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.270626370702469,
      "learning_rate": 1.3907515874042044e-08,
      "loss": 0.1968,
      "step": 10450
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.042469663417937,
      "learning_rate": 1.3747236769147133e-08,
      "loss": 0.1863,
      "step": 10451
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.187549355746638,
      "learning_rate": 1.3587885969051917e-08,
      "loss": 0.188,
      "step": 10452
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.9232008717522466,
      "learning_rate": 1.3429463488571216e-08,
      "loss": 0.1599,
      "step": 10453
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.191338059970624,
      "learning_rate": 1.3271969342431023e-08,
      "loss": 0.2222,
      "step": 10454
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.8229226067862248,
      "learning_rate": 1.3115403545270744e-08,
      "loss": 0.1635,
      "step": 10455
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.241784889142369,
      "learning_rate": 1.29597661116454e-08,
      "loss": 0.2289,
      "step": 10456
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.5527586533571855,
      "learning_rate": 1.2805057056022307e-08,
      "loss": 0.2317,
      "step": 10457
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.32313922178345,
      "learning_rate": 1.2651276392783297e-08,
      "loss": 0.218,
      "step": 10458
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.376823779365707,
      "learning_rate": 1.2498424136223597e-08,
      "loss": 0.219,
      "step": 10459
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.965436162076801,
      "learning_rate": 1.2346500300551844e-08,
      "loss": 0.1801,
      "step": 10460
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.9889896850033675,
      "learning_rate": 1.2195504899890076e-08,
      "loss": 0.1451,
      "step": 10461
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.9195635938136841,
      "learning_rate": 1.2045437948275952e-08,
      "loss": 0.1672,
      "step": 10462
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.7917668696783973,
      "learning_rate": 1.1896299459658311e-08,
      "loss": 0.1378,
      "step": 10463
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.9690964314295065,
      "learning_rate": 1.1748089447901623e-08,
      "loss": 0.199,
      "step": 10464
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.6837634763139058,
      "learning_rate": 1.1600807926782642e-08,
      "loss": 0.1237,
      "step": 10465
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.3018429281809256,
      "learning_rate": 1.145445490999153e-08,
      "loss": 0.2126,
      "step": 10466
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.7594504339460058,
      "learning_rate": 1.130903041113518e-08,
      "loss": 0.1244,
      "step": 10467
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.130810174740198,
      "learning_rate": 1.1164534443730557e-08,
      "loss": 0.2053,
      "step": 10468
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.9479385998140235,
      "learning_rate": 1.1020967021210249e-08,
      "loss": 0.2097,
      "step": 10469
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.193920673590348,
      "learning_rate": 1.0878328156919139e-08,
      "loss": 0.1714,
      "step": 10470
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.005587033349348,
      "learning_rate": 1.0736617864117727e-08,
      "loss": 0.1804,
      "step": 10471
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.014424274749688,
      "learning_rate": 1.0595836155978811e-08,
      "loss": 0.1794,
      "step": 10472
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.3769662192750567,
      "learning_rate": 1.0455983045588591e-08,
      "loss": 0.208,
      "step": 10473
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.2351272317344044,
      "learning_rate": 1.0317058545948888e-08,
      "loss": 0.2268,
      "step": 10474
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.1882261694993153,
      "learning_rate": 1.0179062669972705e-08,
      "loss": 0.1973,
      "step": 10475
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.8043640821934046,
      "learning_rate": 1.0041995430488671e-08,
      "loss": 0.151,
      "step": 10476
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.9307470549877912,
      "learning_rate": 9.905856840238815e-09,
      "loss": 0.166,
      "step": 10477
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.9733393565585988,
      "learning_rate": 9.770646911876347e-09,
      "loss": 0.1755,
      "step": 10478
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.3551666165380993,
      "learning_rate": 9.636365657971215e-09,
      "loss": 0.2253,
      "step": 10479
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.4178358810284144,
      "learning_rate": 9.503013091006763e-09,
      "loss": 0.2034,
      "step": 10480
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.020786421132535,
      "learning_rate": 9.370589223378635e-09,
      "loss": 0.1807,
      "step": 10481
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.9871150538132383,
      "learning_rate": 9.239094067396982e-09,
      "loss": 0.157,
      "step": 10482
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.0435114609510023,
      "learning_rate": 9.108527635284248e-09,
      "loss": 0.207,
      "step": 10483
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.9608885091139843,
      "learning_rate": 8.978889939178503e-09,
      "loss": 0.1581,
      "step": 10484
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.2003415370063992,
      "learning_rate": 8.850180991131219e-09,
      "loss": 0.2009,
      "step": 10485
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.0758925283987595,
      "learning_rate": 8.722400803106157e-09,
      "loss": 0.2072,
      "step": 10486
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.3348752615672357,
      "learning_rate": 8.595549386981595e-09,
      "loss": 0.2559,
      "step": 10487
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.9880313436576353,
      "learning_rate": 8.469626754549209e-09,
      "loss": 0.1824,
      "step": 10488
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.1809485048145145,
      "learning_rate": 8.344632917515194e-09,
      "loss": 0.25,
      "step": 10489
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.155945628355034,
      "learning_rate": 8.220567887498033e-09,
      "loss": 0.188,
      "step": 10490
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.340754884014186,
      "learning_rate": 8.09743167603072e-09,
      "loss": 0.1799,
      "step": 10491
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.740903924658684,
      "learning_rate": 7.975224294560769e-09,
      "loss": 0.1497,
      "step": 10492
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.1179958008135076,
      "learning_rate": 7.853945754447977e-09,
      "loss": 0.1855,
      "step": 10493
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.5308635188861084,
      "learning_rate": 7.733596066965555e-09,
      "loss": 0.2344,
      "step": 10494
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.364154222801278,
      "learning_rate": 7.614175243301213e-09,
      "loss": 0.2239,
      "step": 10495
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.4485035490633775,
      "learning_rate": 7.495683294556078e-09,
      "loss": 0.2195,
      "step": 10496
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.2395181532624,
      "learning_rate": 7.378120231745778e-09,
      "loss": 0.2031,
      "step": 10497
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.0260790936849893,
      "learning_rate": 7.26148606579713e-09,
      "loss": 0.1763,
      "step": 10498
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.1467965415056027,
      "learning_rate": 7.145780807553681e-09,
      "loss": 0.2058,
      "step": 10499
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.055187785582228,
      "learning_rate": 7.031004467771274e-09,
      "loss": 0.176,
      "step": 10500
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.265017374636939,
      "learning_rate": 6.91715705711804e-09,
      "loss": 0.2163,
      "step": 10501
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.1040790768032127,
      "learning_rate": 6.804238586177736e-09,
      "loss": 0.2124,
      "step": 10502
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.185901092565601,
      "learning_rate": 6.692249065447521e-09,
      "loss": 0.2085,
      "step": 10503
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.912191121919801,
      "learning_rate": 6.5811885053368444e-09,
      "loss": 0.1376,
      "step": 10504
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.8865396421653917,
      "learning_rate": 6.471056916170782e-09,
      "loss": 0.1926,
      "step": 10505
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.9244445460855197,
      "learning_rate": 6.361854308185589e-09,
      "loss": 0.1659,
      "step": 10506
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.1605159677195203,
      "learning_rate": 6.253580691534255e-09,
      "loss": 0.1643,
      "step": 10507
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.1299828463819837,
      "learning_rate": 6.146236076279843e-09,
      "loss": 0.2122,
      "step": 10508
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.9747475048034575,
      "learning_rate": 6.039820472403257e-09,
      "loss": 0.1755,
      "step": 10509
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.7630347798013921,
      "learning_rate": 5.934333889794364e-09,
      "loss": 0.1382,
      "step": 10510
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.2590438776497552,
      "learning_rate": 5.8297763382597625e-09,
      "loss": 0.2017,
      "step": 10511
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.945632040735695,
      "learning_rate": 5.726147827519457e-09,
      "loss": 0.1665,
      "step": 10512
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.9096229427385905,
      "learning_rate": 5.623448367205741e-09,
      "loss": 0.1552,
      "step": 10513
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.3456748905299425,
      "learning_rate": 5.521677966866534e-09,
      "loss": 0.1758,
      "step": 10514
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.8680935932123675,
      "learning_rate": 5.4208366359620455e-09,
      "loss": 0.1729,
      "step": 10515
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.8196279007024,
      "learning_rate": 5.3209243838647784e-09,
      "loss": 0.1707,
      "step": 10516
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.1675404821845965,
      "learning_rate": 5.221941219863969e-09,
      "loss": 0.189,
      "step": 10517
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.9938242890047648,
      "learning_rate": 5.123887153161145e-09,
      "loss": 0.1902,
      "step": 10518
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.278595183404665,
      "learning_rate": 5.026762192870127e-09,
      "loss": 0.1978,
      "step": 10519
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.9349937465444775,
      "learning_rate": 4.93056634801925e-09,
      "loss": 0.1764,
      "step": 10520
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.6013034711393603,
      "learning_rate": 4.83529962755247e-09,
      "loss": 0.1324,
      "step": 10521
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.9417877375554478,
      "learning_rate": 4.740962040323815e-09,
      "loss": 0.1909,
      "step": 10522
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.1059840162387666,
      "learning_rate": 4.647553595102938e-09,
      "loss": 0.186,
      "step": 10523
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.719241508893175,
      "learning_rate": 4.555074300574003e-09,
      "loss": 0.1621,
      "step": 10524
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.030459300777842,
      "learning_rate": 4.463524165333466e-09,
      "loss": 0.1902,
      "step": 10525
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.8100249297835957,
      "learning_rate": 4.372903197891188e-09,
      "loss": 0.1552,
      "step": 10526
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.0745207046984215,
      "learning_rate": 4.283211406670429e-09,
      "loss": 0.165,
      "step": 10527
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.2919695265514375,
      "learning_rate": 4.194448800011186e-09,
      "loss": 0.2131,
      "step": 10528
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.8960303060117347,
      "learning_rate": 4.1066153861624155e-09,
      "loss": 0.1758,
      "step": 10529
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.4784291555372824,
      "learning_rate": 4.019711173289809e-09,
      "loss": 0.23,
      "step": 10530
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.4675636559943186,
      "learning_rate": 3.933736169471347e-09,
      "loss": 0.2471,
      "step": 10531
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.9281887226439605,
      "learning_rate": 3.8486903826995266e-09,
      "loss": 0.1699,
      "step": 10532
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.1112111595054843,
      "learning_rate": 3.764573820880246e-09,
      "loss": 0.1853,
      "step": 10533
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.7955197419445403,
      "learning_rate": 3.6813864918328057e-09,
      "loss": 0.1577,
      "step": 10534
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.1150362743231965,
      "learning_rate": 3.5991284032899087e-09,
      "loss": 0.2148,
      "step": 10535
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.3955333065388,
      "learning_rate": 3.5177995628976613e-09,
      "loss": 0.231,
      "step": 10536
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.6712585750273399,
      "learning_rate": 3.437399978216682e-09,
      "loss": 0.1522,
      "step": 10537
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.2401938612672754,
      "learning_rate": 3.357929656722103e-09,
      "loss": 0.2178,
      "step": 10538
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.97468396805549,
      "learning_rate": 3.2793886057991277e-09,
      "loss": 0.1606,
      "step": 10539
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.34314336445277,
      "learning_rate": 3.201776832749692e-09,
      "loss": 0.2032,
      "step": 10540
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.126409558822197,
      "learning_rate": 3.125094344789137e-09,
      "loss": 0.2064,
      "step": 10541
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.1368827448785215,
      "learning_rate": 3.049341149045093e-09,
      "loss": 0.1997,
      "step": 10542
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.225650853962125,
      "learning_rate": 2.974517252558595e-09,
      "loss": 0.1902,
      "step": 10543
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.1683364883815934,
      "learning_rate": 2.9006226622874114e-09,
      "loss": 0.2224,
      "step": 10544
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.0054659586879358,
      "learning_rate": 2.8276573850982703e-09,
      "loss": 0.23,
      "step": 10545
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.8872614379601031,
      "learning_rate": 2.755621427774635e-09,
      "loss": 0.1665,
      "step": 10546
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.059541538698177,
      "learning_rate": 2.6845147970144816e-09,
      "loss": 0.1776,
      "step": 10547
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.217840345454584,
      "learning_rate": 2.6143374994247463e-09,
      "loss": 0.2087,
      "step": 10548
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.1391940520921358,
      "learning_rate": 2.5450895415324306e-09,
      "loss": 0.2021,
      "step": 10549
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.9409160393041762,
      "learning_rate": 2.4767709297712772e-09,
      "loss": 0.1821,
      "step": 10550
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.940567666348083,
      "learning_rate": 2.4093816704950924e-09,
      "loss": 0.2092,
      "step": 10551
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.6739448079330745,
      "learning_rate": 2.3429217699655337e-09,
      "loss": 0.1266,
      "step": 10552
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.0721092717631913,
      "learning_rate": 2.277391234363213e-09,
      "loss": 0.1841,
      "step": 10553
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.045709146132886,
      "learning_rate": 2.2127900697777038e-09,
      "loss": 0.1707,
      "step": 10554
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.935459103553954,
      "learning_rate": 2.1491182822153124e-09,
      "loss": 0.1809,
      "step": 10555
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.8561680344386435,
      "learning_rate": 2.086375877593527e-09,
      "loss": 0.1663,
      "step": 10556
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.7273763406370626,
      "learning_rate": 2.0245628617465706e-09,
      "loss": 0.2378,
      "step": 10557
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.1701486459801798,
      "learning_rate": 1.963679240419847e-09,
      "loss": 0.2156,
      "step": 10558
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.0109302789098846,
      "learning_rate": 1.9037250192732728e-09,
      "loss": 0.1995,
      "step": 10559
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.89701798864612,
      "learning_rate": 1.8447002038779471e-09,
      "loss": 0.1531,
      "step": 10560
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.1392130057898915,
      "learning_rate": 1.7866047997239234e-09,
      "loss": 0.2454,
      "step": 10561
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.0227288110773394,
      "learning_rate": 1.7294388122102157e-09,
      "loss": 0.1874,
      "step": 10562
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.9435659837283257,
      "learning_rate": 1.6732022466503516e-09,
      "loss": 0.1857,
      "step": 10563
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.426431934731805,
      "learning_rate": 1.6178951082712613e-09,
      "loss": 0.2446,
      "step": 10564
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.185707067286829,
      "learning_rate": 1.5635174022166077e-09,
      "loss": 0.1919,
      "step": 10565
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.9387018432265273,
      "learning_rate": 1.5100691335401262e-09,
      "loss": 0.187,
      "step": 10566
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.081682385176414,
      "learning_rate": 1.4575503072100649e-09,
      "loss": 0.1748,
      "step": 10567
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.366021592307441,
      "learning_rate": 1.4059609281080743e-09,
      "loss": 0.1846,
      "step": 10568
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.446541424485981,
      "learning_rate": 1.355301001030318e-09,
      "loss": 0.2349,
      "step": 10569
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.1234380777191966,
      "learning_rate": 1.305570530686362e-09,
      "loss": 0.2227,
      "step": 10570
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.166273421796514,
      "learning_rate": 1.256769521699175e-09,
      "loss": 0.2534,
      "step": 10571
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.8700723924012908,
      "learning_rate": 1.2088979786040179e-09,
      "loss": 0.1704,
      "step": 10572
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.350500893769988,
      "learning_rate": 1.1619559058517748e-09,
      "loss": 0.2158,
      "step": 10573
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.260663946645462,
      "learning_rate": 1.1159433078067327e-09,
      "loss": 0.2351,
      "step": 10574
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.123581519895594,
      "learning_rate": 1.0708601887454706e-09,
      "loss": 0.2112,
      "step": 10575
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.2077718395881165,
      "learning_rate": 1.02670655285797e-09,
      "loss": 0.2166,
      "step": 10576
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.8696942353974408,
      "learning_rate": 9.83482404249836e-10,
      "loss": 0.177,
      "step": 10577
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.8138263073277547,
      "learning_rate": 9.411877469389652e-10,
      "loss": 0.1862,
      "step": 10578
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.334589016514252,
      "learning_rate": 8.998225848566577e-10,
      "loss": 0.2087,
      "step": 10579
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.22840747522748,
      "learning_rate": 8.593869218487261e-10,
      "loss": 0.1965,
      "step": 10580
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.1636180463787746,
      "learning_rate": 8.198807616732752e-10,
      "loss": 0.1946,
      "step": 10581
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.9593946879767445,
      "learning_rate": 7.813041080029226e-10,
      "loss": 0.198,
      "step": 10582
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.2534402242009772,
      "learning_rate": 7.43656964423689e-10,
      "loss": 0.1958,
      "step": 10583
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.4838708631268718,
      "learning_rate": 7.069393344361075e-10,
      "loss": 0.1702,
      "step": 10584
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.8885450628441662,
      "learning_rate": 6.711512214518934e-10,
      "loss": 0.181,
      "step": 10585
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.749265248183381,
      "learning_rate": 6.36292628798385e-10,
      "loss": 0.1653,
      "step": 10586
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.139057830765931,
      "learning_rate": 6.023635597163236e-10,
      "loss": 0.1921,
      "step": 10587
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.1860165817348083,
      "learning_rate": 5.693640173598525e-10,
      "loss": 0.2031,
      "step": 10588
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.0539075129708095,
      "learning_rate": 5.372940047965181e-10,
      "loss": 0.1617,
      "step": 10589
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.8100589201649582,
      "learning_rate": 5.061535250061589e-10,
      "loss": 0.1566,
      "step": 10590
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.109018949812905,
      "learning_rate": 4.759425808853468e-10,
      "loss": 0.1821,
      "step": 10591
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.107754629984994,
      "learning_rate": 4.466611752418359e-10,
      "loss": 0.196,
      "step": 10592
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.7731985360116365,
      "learning_rate": 4.183093107967828e-10,
      "loss": 0.166,
      "step": 10593
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.0283192470951934,
      "learning_rate": 3.9088699018585695e-10,
      "loss": 0.1674,
      "step": 10594
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.276345601995626,
      "learning_rate": 3.6439421595924065e-10,
      "loss": 0.1853,
      "step": 10595
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.113017291353295,
      "learning_rate": 3.388309905794085e-10,
      "loss": 0.21,
      "step": 10596
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.909182339928852,
      "learning_rate": 3.1419731642223785e-10,
      "loss": 0.1924,
      "step": 10597
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.326324813219362,
      "learning_rate": 2.9049319577811873e-10,
      "loss": 0.2386,
      "step": 10598
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.488045602239734,
      "learning_rate": 2.677186308497337e-10,
      "loss": 0.2349,
      "step": 10599
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.143301424239074,
      "learning_rate": 2.45873623754278e-10,
      "loss": 0.1719,
      "step": 10600
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.2502964606511777,
      "learning_rate": 2.2495817652345987e-10,
      "loss": 0.2229,
      "step": 10601
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.3364633259603464,
      "learning_rate": 2.0497229110016948e-10,
      "loss": 0.2471,
      "step": 10602
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.343237994523007,
      "learning_rate": 1.8591596934292022e-10,
      "loss": 0.2412,
      "step": 10603
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.3712029988208596,
      "learning_rate": 1.6778921302362805e-10,
      "loss": 0.2349,
      "step": 10604
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.9055037374099135,
      "learning_rate": 1.505920238265013e-10,
      "loss": 0.1819,
      "step": 10605
    },
    {
      "epoch": 3.0,
      "grad_norm": 2.492818490139754,
      "learning_rate": 1.343244033513713e-10,
      "loss": 0.2322,
      "step": 10606
    },
    {
      "epoch": 3.0,
      "grad_norm": 2.097762568733801,
      "learning_rate": 1.1898635310925167e-10,
      "loss": 0.21,
      "step": 10607
    },
    {
      "epoch": 3.0,
      "grad_norm": 2.0392259726865976,
      "learning_rate": 1.045778745256687e-10,
      "loss": 0.16,
      "step": 10608
    },
    {
      "epoch": 3.0,
      "grad_norm": 2.0261319460700253,
      "learning_rate": 9.109896894066161e-11,
      "loss": 0.1927,
      "step": 10609
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.915460737181471,
      "learning_rate": 7.85496376076722e-11,
      "loss": 0.1687,
      "step": 10610
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.7932181939261618,
      "learning_rate": 6.692988169243463e-11,
      "loss": 0.1591,
      "step": 10611
    },
    {
      "epoch": 3.0,
      "grad_norm": 2.198102017547231,
      "learning_rate": 5.623970227630615e-11,
      "loss": 0.2041,
      "step": 10612
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.8248882468230934,
      "learning_rate": 4.64791003507159e-11,
      "loss": 0.1589,
      "step": 10613
    },
    {
      "epoch": 3.0,
      "grad_norm": 2.0265639660617847,
      "learning_rate": 3.764807682604676e-11,
      "loss": 0.2314,
      "step": 10614
    },
    {
      "epoch": 3.0,
      "grad_norm": 2.043950967429426,
      "learning_rate": 2.9746632520533116e-11,
      "loss": 0.1487,
      "step": 10615
    },
    {
      "epoch": 3.0,
      "grad_norm": 2.053179744046229,
      "learning_rate": 2.2774768170252813e-11,
      "loss": 0.1797,
      "step": 10616
    },
    {
      "epoch": 3.0,
      "grad_norm": 2.116391463588622,
      "learning_rate": 1.673248442246589e-11,
      "loss": 0.1816,
      "step": 10617
    },
    {
      "epoch": 3.0,
      "grad_norm": 2.158162322680363,
      "learning_rate": 1.1619781838945188e-11,
      "loss": 0.2351,
      "step": 10618
    },
    {
      "epoch": 3.0,
      "grad_norm": 2.023459583181967,
      "learning_rate": 7.436660894866165e-12,
      "loss": 0.1978,
      "step": 10619
    },
    {
      "epoch": 3.0,
      "grad_norm": 2.5457865999816303,
      "learning_rate": 4.1831219799171044e-12,
      "loss": 0.2742,
      "step": 10620
    },
    {
      "epoch": 3.0,
      "grad_norm": 2.4300828283728495,
      "learning_rate": 1.859165397188889e-12,
      "loss": 0.2461,
      "step": 10621
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.9787753821093632,
      "learning_rate": 4.647913598443409e-13,
      "loss": 0.1799,
      "step": 10622
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.7878699513737746,
      "learning_rate": 0.0,
      "loss": 0.1761,
      "step": 10623
    },
    {
      "epoch": 3.0,
      "step": 10623,
      "total_flos": 278869684842496.0,
      "train_loss": 0.34584171956265297,
      "train_runtime": 106932.4783,
      "train_samples_per_second": 1.589,
      "train_steps_per_second": 0.099
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 10623,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "total_flos": 278869684842496.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}