{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.6385242993969493,
  "eval_steps": 500,
  "global_step": 900,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.000709471443774388,
      "grad_norm": 265.86757613893764,
      "learning_rate": 0.0,
      "loss": 2.1209,
      "step": 1
    },
    {
      "epoch": 0.001418942887548776,
      "grad_norm": 256.9364781809131,
      "learning_rate": 2.1276595744680852e-07,
      "loss": 1.9508,
      "step": 2
    },
    {
      "epoch": 0.002128414331323164,
      "grad_norm": 261.72185657120684,
      "learning_rate": 4.2553191489361704e-07,
      "loss": 2.1995,
      "step": 3
    },
    {
      "epoch": 0.002837885775097552,
      "grad_norm": 261.0961921742672,
      "learning_rate": 6.382978723404255e-07,
      "loss": 2.0081,
      "step": 4
    },
    {
      "epoch": 0.0035473572188719402,
      "grad_norm": 255.1019965019492,
      "learning_rate": 8.510638297872341e-07,
      "loss": 2.0828,
      "step": 5
    },
    {
      "epoch": 0.004256828662646328,
      "grad_norm": 201.9040896756704,
      "learning_rate": 1.0638297872340427e-06,
      "loss": 1.6403,
      "step": 6
    },
    {
      "epoch": 0.004966300106420717,
      "grad_norm": 198.32057240838185,
      "learning_rate": 1.276595744680851e-06,
      "loss": 1.6037,
      "step": 7
    },
    {
      "epoch": 0.005675771550195104,
      "grad_norm": 91.64675018232528,
      "learning_rate": 1.4893617021276596e-06,
      "loss": 1.0155,
      "step": 8
    },
    {
      "epoch": 0.006385242993969493,
      "grad_norm": 71.73605818031255,
      "learning_rate": 1.7021276595744682e-06,
      "loss": 0.8153,
      "step": 9
    },
    {
      "epoch": 0.0070947144377438804,
      "grad_norm": 24.941283646351614,
      "learning_rate": 1.9148936170212763e-06,
      "loss": 0.3811,
      "step": 10
    },
    {
      "epoch": 0.007804185881518269,
      "grad_norm": 31.071639133351898,
      "learning_rate": 2.1276595744680853e-06,
      "loss": 0.3851,
      "step": 11
    },
    {
      "epoch": 0.008513657325292657,
      "grad_norm": 20.77631646223411,
      "learning_rate": 2.340425531914894e-06,
      "loss": 0.2891,
      "step": 12
    },
    {
      "epoch": 0.009223128769067045,
      "grad_norm": 13.477671793796729,
      "learning_rate": 2.553191489361702e-06,
      "loss": 0.2528,
      "step": 13
    },
    {
      "epoch": 0.009932600212841433,
      "grad_norm": 2.6555809333853024,
      "learning_rate": 2.7659574468085106e-06,
      "loss": 0.1542,
      "step": 14
    },
    {
      "epoch": 0.010642071656615822,
      "grad_norm": 2.898054397668401,
      "learning_rate": 2.978723404255319e-06,
      "loss": 0.1762,
      "step": 15
    },
    {
      "epoch": 0.011351543100390209,
      "grad_norm": 2.1504034279865385,
      "learning_rate": 3.1914893617021277e-06,
      "loss": 0.1403,
      "step": 16
    },
    {
      "epoch": 0.012061014544164597,
      "grad_norm": 5.259173556940105,
      "learning_rate": 3.4042553191489363e-06,
      "loss": 0.1561,
      "step": 17
    },
    {
      "epoch": 0.012770485987938986,
      "grad_norm": 4.169332828012237,
      "learning_rate": 3.6170212765957445e-06,
      "loss": 0.1632,
      "step": 18
    },
    {
      "epoch": 0.013479957431713374,
      "grad_norm": 1.9000129489744462,
      "learning_rate": 3.829787234042553e-06,
      "loss": 0.1667,
      "step": 19
    },
    {
      "epoch": 0.014189428875487761,
      "grad_norm": 7.769485079957688,
      "learning_rate": 4.042553191489362e-06,
      "loss": 0.1893,
      "step": 20
    },
    {
      "epoch": 0.01489890031926215,
      "grad_norm": 11.504242044440007,
      "learning_rate": 4.255319148936171e-06,
      "loss": 0.2085,
      "step": 21
    },
    {
      "epoch": 0.015608371763036538,
      "grad_norm": 7.145142567820805,
      "learning_rate": 4.468085106382979e-06,
      "loss": 0.1913,
      "step": 22
    },
    {
      "epoch": 0.016317843206810925,
      "grad_norm": 2.736186486725912,
      "learning_rate": 4.680851063829788e-06,
      "loss": 0.1741,
      "step": 23
    },
    {
      "epoch": 0.017027314650585313,
      "grad_norm": 5.7916893585600695,
      "learning_rate": 4.893617021276596e-06,
      "loss": 0.1408,
      "step": 24
    },
    {
      "epoch": 0.0177367860943597,
      "grad_norm": 4.68647394202743,
      "learning_rate": 5.106382978723404e-06,
      "loss": 0.1632,
      "step": 25
    },
    {
      "epoch": 0.01844625753813409,
      "grad_norm": 2.0524513403445757,
      "learning_rate": 5.319148936170213e-06,
      "loss": 0.1563,
      "step": 26
    },
    {
      "epoch": 0.01915572898190848,
      "grad_norm": 1.588252902269814,
      "learning_rate": 5.531914893617021e-06,
      "loss": 0.1582,
      "step": 27
    },
    {
      "epoch": 0.019865200425682867,
      "grad_norm": 3.5755825255555225,
      "learning_rate": 5.74468085106383e-06,
      "loss": 0.1451,
      "step": 28
    },
    {
      "epoch": 0.020574671869457255,
      "grad_norm": 7.715437458161659,
      "learning_rate": 5.957446808510638e-06,
      "loss": 0.157,
      "step": 29
    },
    {
      "epoch": 0.021284143313231644,
      "grad_norm": 19.70487906339096,
      "learning_rate": 6.1702127659574465e-06,
      "loss": 0.1725,
      "step": 30
    },
    {
      "epoch": 0.02199361475700603,
      "grad_norm": 2.21235512385039,
      "learning_rate": 6.3829787234042555e-06,
      "loss": 0.1318,
      "step": 31
    },
    {
      "epoch": 0.022703086200780417,
      "grad_norm": 5.021653379215672,
      "learning_rate": 6.5957446808510645e-06,
      "loss": 0.1795,
      "step": 32
    },
    {
      "epoch": 0.023412557644554806,
      "grad_norm": 25.495061679606632,
      "learning_rate": 6.808510638297873e-06,
      "loss": 0.1586,
      "step": 33
    },
    {
      "epoch": 0.024122029088329194,
      "grad_norm": 4.7481186730757035,
      "learning_rate": 7.021276595744681e-06,
      "loss": 0.1552,
      "step": 34
    },
    {
      "epoch": 0.024831500532103583,
      "grad_norm": 3.6399107060391365,
      "learning_rate": 7.234042553191489e-06,
      "loss": 0.1573,
      "step": 35
    },
    {
      "epoch": 0.02554097197587797,
      "grad_norm": 3.3754400837055707,
      "learning_rate": 7.446808510638298e-06,
      "loss": 0.149,
      "step": 36
    },
    {
      "epoch": 0.02625044341965236,
      "grad_norm": 2.8787349588257842,
      "learning_rate": 7.659574468085105e-06,
      "loss": 0.154,
      "step": 37
    },
    {
      "epoch": 0.026959914863426748,
      "grad_norm": 2.7404452323712416,
      "learning_rate": 7.872340425531914e-06,
      "loss": 0.1563,
      "step": 38
    },
    {
      "epoch": 0.027669386307201137,
      "grad_norm": 1.8896448248412065,
      "learning_rate": 8.085106382978723e-06,
      "loss": 0.1568,
      "step": 39
    },
    {
      "epoch": 0.028378857750975522,
      "grad_norm": 6.6357403507533395,
      "learning_rate": 8.297872340425532e-06,
      "loss": 0.1608,
      "step": 40
    },
    {
      "epoch": 0.02908832919474991,
      "grad_norm": 5.2141592606599545,
      "learning_rate": 8.510638297872341e-06,
      "loss": 0.1495,
      "step": 41
    },
    {
      "epoch": 0.0297978006385243,
      "grad_norm": 2.653621690332451,
      "learning_rate": 8.723404255319149e-06,
      "loss": 0.1746,
      "step": 42
    },
    {
      "epoch": 0.030507272082298687,
      "grad_norm": 2.1695013496806483,
      "learning_rate": 8.936170212765958e-06,
      "loss": 0.1501,
      "step": 43
    },
    {
      "epoch": 0.031216743526073076,
      "grad_norm": 1.9610946251979264,
      "learning_rate": 9.148936170212767e-06,
      "loss": 0.1614,
      "step": 44
    },
    {
      "epoch": 0.03192621496984746,
      "grad_norm": 3.806116355988534,
      "learning_rate": 9.361702127659576e-06,
      "loss": 0.1649,
      "step": 45
    },
    {
      "epoch": 0.03263568641362185,
      "grad_norm": 1.6308337200888532,
      "learning_rate": 9.574468085106385e-06,
      "loss": 0.1528,
      "step": 46
    },
    {
      "epoch": 0.03334515785739624,
      "grad_norm": 1.0471741398584304,
      "learning_rate": 9.787234042553192e-06,
      "loss": 0.151,
      "step": 47
    },
    {
      "epoch": 0.034054629301170626,
      "grad_norm": 2.1098801420979885,
      "learning_rate": 9.999999999999999e-06,
      "loss": 0.15,
      "step": 48
    },
    {
      "epoch": 0.034764100744945015,
      "grad_norm": 1.4498470002731845,
      "learning_rate": 1.0212765957446808e-05,
      "loss": 0.1385,
      "step": 49
    },
    {
      "epoch": 0.0354735721887194,
      "grad_norm": 2.4843540113556632,
      "learning_rate": 1.0425531914893617e-05,
      "loss": 0.1822,
      "step": 50
    },
    {
      "epoch": 0.03618304363249379,
      "grad_norm": 4.7022928228876335,
      "learning_rate": 1.0638297872340426e-05,
      "loss": 0.1465,
      "step": 51
    },
    {
      "epoch": 0.03689251507626818,
      "grad_norm": 3.6925599893471666,
      "learning_rate": 1.0851063829787235e-05,
      "loss": 0.1301,
      "step": 52
    },
    {
      "epoch": 0.03760198652004257,
      "grad_norm": 1.680217688963427,
      "learning_rate": 1.1063829787234042e-05,
      "loss": 0.1495,
      "step": 53
    },
    {
      "epoch": 0.03831145796381696,
      "grad_norm": 7.334120820499479,
      "learning_rate": 1.1276595744680851e-05,
      "loss": 0.1686,
      "step": 54
    },
    {
      "epoch": 0.039020929407591345,
      "grad_norm": 3.0183779232825234,
      "learning_rate": 1.148936170212766e-05,
      "loss": 0.1567,
      "step": 55
    },
    {
      "epoch": 0.039730400851365734,
      "grad_norm": 1.2525087005082511,
      "learning_rate": 1.170212765957447e-05,
      "loss": 0.1444,
      "step": 56
    },
    {
      "epoch": 0.04043987229514012,
      "grad_norm": 1.425039005914065,
      "learning_rate": 1.1914893617021277e-05,
      "loss": 0.1484,
      "step": 57
    },
    {
      "epoch": 0.04114934373891451,
      "grad_norm": 2.1985450386586933,
      "learning_rate": 1.2127659574468084e-05,
      "loss": 0.1502,
      "step": 58
    },
    {
      "epoch": 0.0418588151826889,
      "grad_norm": 3.249197889697388,
      "learning_rate": 1.2340425531914893e-05,
      "loss": 0.1473,
      "step": 59
    },
    {
      "epoch": 0.04256828662646329,
      "grad_norm": 2.0906804833194443,
      "learning_rate": 1.2553191489361702e-05,
      "loss": 0.1415,
      "step": 60
    },
    {
      "epoch": 0.043277758070237676,
      "grad_norm": 6.008820861881305,
      "learning_rate": 1.2765957446808511e-05,
      "loss": 0.1528,
      "step": 61
    },
    {
      "epoch": 0.04398722951401206,
      "grad_norm": 3.510113211053961,
      "learning_rate": 1.297872340425532e-05,
      "loss": 0.1659,
      "step": 62
    },
    {
      "epoch": 0.044696700957786446,
      "grad_norm": 3.310262198116976,
      "learning_rate": 1.3191489361702129e-05,
      "loss": 0.1318,
      "step": 63
    },
    {
      "epoch": 0.045406172401560835,
      "grad_norm": 2.998742157037864,
      "learning_rate": 1.3404255319148936e-05,
      "loss": 0.1444,
      "step": 64
    },
    {
      "epoch": 0.04611564384533522,
      "grad_norm": 3.037454924126702,
      "learning_rate": 1.3617021276595745e-05,
      "loss": 0.1534,
      "step": 65
    },
    {
      "epoch": 0.04682511528910961,
      "grad_norm": 3.0210908271546746,
      "learning_rate": 1.3829787234042554e-05,
      "loss": 0.1402,
      "step": 66
    },
    {
      "epoch": 0.047534586732884,
      "grad_norm": 5.520540860172146,
      "learning_rate": 1.4042553191489362e-05,
      "loss": 0.1425,
      "step": 67
    },
    {
      "epoch": 0.04824405817665839,
      "grad_norm": 2.8578795762705336,
      "learning_rate": 1.425531914893617e-05,
      "loss": 0.1441,
      "step": 68
    },
    {
      "epoch": 0.04895352962043278,
      "grad_norm": 3.0093954042315523,
      "learning_rate": 1.4468085106382978e-05,
      "loss": 0.1209,
      "step": 69
    },
    {
      "epoch": 0.049663001064207166,
      "grad_norm": 4.7761651201536,
      "learning_rate": 1.4680851063829787e-05,
      "loss": 0.1368,
      "step": 70
    },
    {
      "epoch": 0.050372472507981554,
      "grad_norm": 1.479679010072238,
      "learning_rate": 1.4893617021276596e-05,
      "loss": 0.1228,
      "step": 71
    },
    {
      "epoch": 0.05108194395175594,
      "grad_norm": 1.4562601345388664,
      "learning_rate": 1.5106382978723403e-05,
      "loss": 0.1504,
      "step": 72
    },
    {
      "epoch": 0.05179141539553033,
      "grad_norm": 1.008711424425401,
      "learning_rate": 1.531914893617021e-05,
      "loss": 0.1354,
      "step": 73
    },
    {
      "epoch": 0.05250088683930472,
      "grad_norm": 1.1921518196444796,
      "learning_rate": 1.553191489361702e-05,
      "loss": 0.1272,
      "step": 74
    },
    {
      "epoch": 0.05321035828307911,
      "grad_norm": 2.3190578291261152,
      "learning_rate": 1.574468085106383e-05,
      "loss": 0.1307,
      "step": 75
    },
    {
      "epoch": 0.053919829726853497,
      "grad_norm": 0.8942079530943704,
      "learning_rate": 1.5957446808510637e-05,
      "loss": 0.1215,
      "step": 76
    },
    {
      "epoch": 0.054629301170627885,
      "grad_norm": 2.0576355533811554,
      "learning_rate": 1.6170212765957446e-05,
      "loss": 0.1362,
      "step": 77
    },
    {
      "epoch": 0.05533877261440227,
      "grad_norm": 2.2622110548173677,
      "learning_rate": 1.6382978723404255e-05,
      "loss": 0.1899,
      "step": 78
    },
    {
      "epoch": 0.056048244058176655,
      "grad_norm": 2.374559556727405,
      "learning_rate": 1.6595744680851064e-05,
      "loss": 0.1611,
      "step": 79
    },
    {
      "epoch": 0.056757715501951043,
      "grad_norm": 5.447170725571408,
      "learning_rate": 1.6808510638297873e-05,
      "loss": 0.1514,
      "step": 80
    },
    {
      "epoch": 0.05746718694572543,
      "grad_norm": 2.5429748228117814,
      "learning_rate": 1.7021276595744682e-05,
      "loss": 0.1468,
      "step": 81
    },
    {
      "epoch": 0.05817665838949982,
      "grad_norm": 2.7415968193046902,
      "learning_rate": 1.723404255319149e-05,
      "loss": 0.1445,
      "step": 82
    },
    {
      "epoch": 0.05888612983327421,
      "grad_norm": 3.0382016331376094,
      "learning_rate": 1.7446808510638297e-05,
      "loss": 0.158,
      "step": 83
    },
    {
      "epoch": 0.0595956012770486,
      "grad_norm": 1.6325149171380335,
      "learning_rate": 1.7659574468085106e-05,
      "loss": 0.1253,
      "step": 84
    },
    {
      "epoch": 0.060305072720822986,
      "grad_norm": 3.316327985423067,
      "learning_rate": 1.7872340425531915e-05,
      "loss": 0.1418,
      "step": 85
    },
    {
      "epoch": 0.061014544164597374,
      "grad_norm": 6.588395895312288,
      "learning_rate": 1.8085106382978724e-05,
      "loss": 0.1455,
      "step": 86
    },
    {
      "epoch": 0.06172401560837176,
      "grad_norm": 1.5641033467639194,
      "learning_rate": 1.8297872340425533e-05,
      "loss": 0.139,
      "step": 87
    },
    {
      "epoch": 0.06243348705214615,
      "grad_norm": 1.6297579083033085,
      "learning_rate": 1.8510638297872342e-05,
      "loss": 0.1396,
      "step": 88
    },
    {
      "epoch": 0.06314295849592054,
      "grad_norm": 6.970523654466422,
      "learning_rate": 1.872340425531915e-05,
      "loss": 0.1568,
      "step": 89
    },
    {
      "epoch": 0.06385242993969492,
      "grad_norm": 1.271092780714167,
      "learning_rate": 1.893617021276596e-05,
      "loss": 0.147,
      "step": 90
    },
    {
      "epoch": 0.06456190138346932,
      "grad_norm": 9.422206811592059,
      "learning_rate": 1.914893617021277e-05,
      "loss": 0.1603,
      "step": 91
    },
    {
      "epoch": 0.0652713728272437,
      "grad_norm": 6.043928089777284,
      "learning_rate": 1.9361702127659575e-05,
      "loss": 0.1731,
      "step": 92
    },
    {
      "epoch": 0.0659808442710181,
      "grad_norm": 1.6440555166858408,
      "learning_rate": 1.9574468085106384e-05,
      "loss": 0.1323,
      "step": 93
    },
    {
      "epoch": 0.06669031571479248,
      "grad_norm": 3.722316535878048,
      "learning_rate": 1.978723404255319e-05,
      "loss": 0.123,
      "step": 94
    },
    {
      "epoch": 0.06739978715856687,
      "grad_norm": 0.876265361813727,
      "learning_rate": 1.9999999999999998e-05,
      "loss": 0.1492,
      "step": 95
    },
    {
      "epoch": 0.06810925860234125,
      "grad_norm": 2.9790618574403775,
      "learning_rate": 2.0212765957446807e-05,
      "loss": 0.1422,
      "step": 96
    },
    {
      "epoch": 0.06881873004611565,
      "grad_norm": 4.427704225480461,
      "learning_rate": 2.0425531914893616e-05,
      "loss": 0.1263,
      "step": 97
    },
    {
      "epoch": 0.06952820148989003,
      "grad_norm": 2.574380283469407,
      "learning_rate": 2.0638297872340425e-05,
      "loss": 0.1027,
      "step": 98
    },
    {
      "epoch": 0.07023767293366442,
      "grad_norm": 2.1421248571266838,
      "learning_rate": 2.0851063829787234e-05,
      "loss": 0.1315,
      "step": 99
    },
    {
      "epoch": 0.0709471443774388,
      "grad_norm": 5.305110629941279,
      "learning_rate": 2.1063829787234043e-05,
      "loss": 0.1971,
      "step": 100
    },
    {
      "epoch": 0.0716566158212132,
      "grad_norm": 5.642883274297945,
      "learning_rate": 2.1276595744680852e-05,
      "loss": 0.1875,
      "step": 101
    },
    {
      "epoch": 0.07236608726498758,
      "grad_norm": 1.7507710973299895,
      "learning_rate": 2.148936170212766e-05,
      "loss": 0.15,
      "step": 102
    },
    {
      "epoch": 0.07307555870876198,
      "grad_norm": 9.627116387162864,
      "learning_rate": 2.170212765957447e-05,
      "loss": 0.1866,
      "step": 103
    },
    {
      "epoch": 0.07378503015253636,
      "grad_norm": 9.749256984312629,
      "learning_rate": 2.191489361702128e-05,
      "loss": 0.191,
      "step": 104
    },
    {
      "epoch": 0.07449450159631076,
      "grad_norm": 7.887554277397637,
      "learning_rate": 2.2127659574468085e-05,
      "loss": 0.1667,
      "step": 105
    },
    {
      "epoch": 0.07520397304008514,
      "grad_norm": 5.470026091817564,
      "learning_rate": 2.2340425531914894e-05,
      "loss": 0.1784,
      "step": 106
    },
    {
      "epoch": 0.07591344448385952,
      "grad_norm": 4.3382421070052315,
      "learning_rate": 2.2553191489361703e-05,
      "loss": 0.1822,
      "step": 107
    },
    {
      "epoch": 0.07662291592763391,
      "grad_norm": 10.137184968638856,
      "learning_rate": 2.2765957446808512e-05,
      "loss": 0.1862,
      "step": 108
    },
    {
      "epoch": 0.0773323873714083,
      "grad_norm": 10.090500044101372,
      "learning_rate": 2.297872340425532e-05,
      "loss": 0.1892,
      "step": 109
    },
    {
      "epoch": 0.07804185881518269,
      "grad_norm": 4.447131231901499,
      "learning_rate": 2.319148936170213e-05,
      "loss": 0.1718,
      "step": 110
    },
    {
      "epoch": 0.07875133025895707,
      "grad_norm": 2.750599704413433,
      "learning_rate": 2.340425531914894e-05,
      "loss": 0.1589,
      "step": 111
    },
    {
      "epoch": 0.07946080170273147,
      "grad_norm": 5.03388412323486,
      "learning_rate": 2.3617021276595744e-05,
      "loss": 0.1634,
      "step": 112
    },
    {
      "epoch": 0.08017027314650585,
      "grad_norm": 7.506997785980621,
      "learning_rate": 2.3829787234042553e-05,
      "loss": 0.18,
      "step": 113
    },
    {
      "epoch": 0.08087974459028024,
      "grad_norm": 1.6324595253459384,
      "learning_rate": 2.4042553191489362e-05,
      "loss": 0.1529,
      "step": 114
    },
    {
      "epoch": 0.08158921603405463,
      "grad_norm": 4.93334287619821,
      "learning_rate": 2.4255319148936168e-05,
      "loss": 0.1891,
      "step": 115
    },
    {
      "epoch": 0.08229868747782902,
      "grad_norm": 2.2288648427809856,
      "learning_rate": 2.4468085106382977e-05,
      "loss": 0.1309,
      "step": 116
    },
    {
      "epoch": 0.0830081589216034,
      "grad_norm": 2.9523096736073926,
      "learning_rate": 2.4680851063829786e-05,
      "loss": 0.1422,
      "step": 117
    },
    {
      "epoch": 0.0837176303653778,
      "grad_norm": 1.9461700932872816,
      "learning_rate": 2.4893617021276595e-05,
      "loss": 0.1133,
      "step": 118
    },
    {
      "epoch": 0.08442710180915218,
      "grad_norm": 3.311754017421373,
      "learning_rate": 2.5106382978723404e-05,
      "loss": 0.1257,
      "step": 119
    },
    {
      "epoch": 0.08513657325292658,
      "grad_norm": 1.3693780728561118,
      "learning_rate": 2.5319148936170213e-05,
      "loss": 0.132,
      "step": 120
    },
    {
      "epoch": 0.08584604469670096,
      "grad_norm": 2.014986619725941,
      "learning_rate": 2.5531914893617022e-05,
      "loss": 0.1167,
      "step": 121
    },
    {
      "epoch": 0.08655551614047535,
      "grad_norm": 0.9852224915415894,
      "learning_rate": 2.574468085106383e-05,
      "loss": 0.1552,
      "step": 122
    },
    {
      "epoch": 0.08726498758424973,
      "grad_norm": 2.9682789311444537,
      "learning_rate": 2.595744680851064e-05,
      "loss": 0.1444,
      "step": 123
    },
    {
      "epoch": 0.08797445902802412,
      "grad_norm": 5.666004456761068,
      "learning_rate": 2.617021276595745e-05,
      "loss": 0.1675,
      "step": 124
    },
    {
      "epoch": 0.08868393047179851,
      "grad_norm": 7.081569238554394,
      "learning_rate": 2.6382978723404258e-05,
      "loss": 0.1783,
      "step": 125
    },
    {
      "epoch": 0.08939340191557289,
      "grad_norm": 1.8891227257145333,
      "learning_rate": 2.6595744680851064e-05,
      "loss": 0.1562,
      "step": 126
    },
    {
      "epoch": 0.09010287335934729,
      "grad_norm": 5.700800787109856,
      "learning_rate": 2.6808510638297873e-05,
      "loss": 0.1637,
      "step": 127
    },
    {
      "epoch": 0.09081234480312167,
      "grad_norm": 3.2343887142846954,
      "learning_rate": 2.702127659574468e-05,
      "loss": 0.129,
      "step": 128
    },
    {
      "epoch": 0.09152181624689606,
      "grad_norm": 7.308286740876094,
      "learning_rate": 2.723404255319149e-05,
      "loss": 0.174,
      "step": 129
    },
    {
      "epoch": 0.09223128769067045,
      "grad_norm": 2.953334456994522,
      "learning_rate": 2.74468085106383e-05,
      "loss": 0.1665,
      "step": 130
    },
    {
      "epoch": 0.09294075913444484,
      "grad_norm": 7.863579465079018,
      "learning_rate": 2.765957446808511e-05,
      "loss": 0.1649,
      "step": 131
    },
    {
      "epoch": 0.09365023057821922,
      "grad_norm": 10.116666038793914,
      "learning_rate": 2.7872340425531914e-05,
      "loss": 0.1966,
      "step": 132
    },
    {
      "epoch": 0.09435970202199362,
      "grad_norm": 10.40037657354269,
      "learning_rate": 2.8085106382978723e-05,
      "loss": 0.2054,
      "step": 133
    },
    {
      "epoch": 0.095069173465768,
      "grad_norm": 2.6057955754631337,
      "learning_rate": 2.8297872340425532e-05,
      "loss": 0.1495,
      "step": 134
    },
    {
      "epoch": 0.0957786449095424,
      "grad_norm": 2.964782006302471,
      "learning_rate": 2.851063829787234e-05,
      "loss": 0.1206,
      "step": 135
    },
    {
      "epoch": 0.09648811635331678,
      "grad_norm": 5.959133910397657,
      "learning_rate": 2.872340425531915e-05,
      "loss": 0.1816,
      "step": 136
    },
    {
      "epoch": 0.09719758779709117,
      "grad_norm": 3.55759144795398,
      "learning_rate": 2.8936170212765956e-05,
      "loss": 0.1586,
      "step": 137
    },
    {
      "epoch": 0.09790705924086555,
      "grad_norm": 1.1399173415846195,
      "learning_rate": 2.9148936170212765e-05,
      "loss": 0.1593,
      "step": 138
    },
    {
      "epoch": 0.09861653068463995,
      "grad_norm": 5.892411282592997,
      "learning_rate": 2.9361702127659574e-05,
      "loss": 0.1739,
      "step": 139
    },
    {
      "epoch": 0.09932600212841433,
      "grad_norm": 3.5540231225780694,
      "learning_rate": 2.9574468085106383e-05,
      "loss": 0.1511,
      "step": 140
    },
    {
      "epoch": 0.10003547357218871,
      "grad_norm": 1.3627037465057075,
      "learning_rate": 2.9787234042553192e-05,
      "loss": 0.1602,
      "step": 141
    },
    {
      "epoch": 0.10074494501596311,
      "grad_norm": 4.359910075426451,
      "learning_rate": 3e-05,
      "loss": 0.1699,
      "step": 142
    },
    {
      "epoch": 0.10145441645973749,
      "grad_norm": 7.221440560225048,
      "learning_rate": 2.9999989686275375e-05,
      "loss": 0.1921,
      "step": 143
    },
    {
      "epoch": 0.10216388790351189,
      "grad_norm": 5.3795984911945895,
      "learning_rate": 2.9999958745115665e-05,
      "loss": 0.1447,
      "step": 144
    },
    {
      "epoch": 0.10287335934728627,
      "grad_norm": 2.1612613655477424,
      "learning_rate": 2.9999907176563438e-05,
      "loss": 0.1478,
      "step": 145
    },
    {
      "epoch": 0.10358283079106066,
      "grad_norm": 5.934694080913587,
      "learning_rate": 2.9999834980689596e-05,
      "loss": 0.1732,
      "step": 146
    },
    {
      "epoch": 0.10429230223483504,
      "grad_norm": 8.428719380615808,
      "learning_rate": 2.999974215759343e-05,
      "loss": 0.1705,
      "step": 147
    },
    {
      "epoch": 0.10500177367860944,
      "grad_norm": 6.494392785741008,
      "learning_rate": 2.9999628707402574e-05,
      "loss": 0.1711,
      "step": 148
    },
    {
      "epoch": 0.10571124512238382,
      "grad_norm": 2.7373565298115503,
      "learning_rate": 2.999949463027305e-05,
      "loss": 0.1227,
      "step": 149
    },
    {
      "epoch": 0.10642071656615822,
      "grad_norm": 3.7095970629570596,
      "learning_rate": 2.9999339926389234e-05,
      "loss": 0.1541,
      "step": 150
    },
    {
      "epoch": 0.1071301880099326,
      "grad_norm": 7.237626192320179,
      "learning_rate": 2.9999164595963868e-05,
      "loss": 0.1769,
      "step": 151
    },
    {
      "epoch": 0.10783965945370699,
      "grad_norm": 5.161108573073184,
      "learning_rate": 2.9998968639238063e-05,
      "loss": 0.1667,
      "step": 152
    },
    {
      "epoch": 0.10854913089748137,
      "grad_norm": 2.9879622868437554,
      "learning_rate": 2.9998752056481292e-05,
      "loss": 0.1275,
      "step": 153
    },
    {
      "epoch": 0.10925860234125577,
      "grad_norm": 3.99129452194833,
      "learning_rate": 2.999851484799138e-05,
      "loss": 0.1236,
      "step": 154
    },
    {
      "epoch": 0.10996807378503015,
      "grad_norm": 5.920708898998149,
      "learning_rate": 2.9998257014094543e-05,
      "loss": 0.1717,
      "step": 155
    },
    {
      "epoch": 0.11067754522880455,
      "grad_norm": 8.512696528821294,
      "learning_rate": 2.999797855514534e-05,
      "loss": 0.1788,
      "step": 156
    },
    {
      "epoch": 0.11138701667257893,
      "grad_norm": 6.145053826415381,
      "learning_rate": 2.9997679471526692e-05,
      "loss": 0.1438,
      "step": 157
    },
    {
      "epoch": 0.11209648811635331,
      "grad_norm": 1.94217343320799,
      "learning_rate": 2.9997359763649892e-05,
      "loss": 0.157,
      "step": 158
    },
    {
      "epoch": 0.1128059595601277,
      "grad_norm": 1.6490655551209275,
      "learning_rate": 2.999701943195459e-05,
      "loss": 0.1242,
      "step": 159
    },
    {
      "epoch": 0.11351543100390209,
      "grad_norm": 3.350746503853587,
      "learning_rate": 2.9996658476908798e-05,
      "loss": 0.1469,
      "step": 160
    },
    {
      "epoch": 0.11422490244767648,
      "grad_norm": 3.335340051179254,
      "learning_rate": 2.9996276899008885e-05,
      "loss": 0.1493,
      "step": 161
    },
    {
      "epoch": 0.11493437389145086,
      "grad_norm": 0.9701252939826595,
      "learning_rate": 2.9995874698779587e-05,
      "loss": 0.1111,
      "step": 162
    },
    {
      "epoch": 0.11564384533522526,
      "grad_norm": 1.9147948348418735,
      "learning_rate": 2.9995451876773996e-05,
      "loss": 0.113,
      "step": 163
    },
    {
      "epoch": 0.11635331677899964,
      "grad_norm": 2.4415174657014447,
      "learning_rate": 2.9995008433573562e-05,
      "loss": 0.1508,
      "step": 164
    },
    {
      "epoch": 0.11706278822277404,
      "grad_norm": 1.6248197891915095,
      "learning_rate": 2.9994544369788082e-05,
      "loss": 0.1412,
      "step": 165
    },
    {
      "epoch": 0.11777225966654842,
      "grad_norm": 1.4976210318170662,
      "learning_rate": 2.999405968605573e-05,
      "loss": 0.122,
      "step": 166
    },
    {
      "epoch": 0.11848173111032281,
      "grad_norm": 1.1043965470742263,
      "learning_rate": 2.999355438304302e-05,
      "loss": 0.1429,
      "step": 167
    },
    {
      "epoch": 0.1191912025540972,
      "grad_norm": 0.9655524034474914,
      "learning_rate": 2.999302846144483e-05,
      "loss": 0.1346,
      "step": 168
    },
    {
      "epoch": 0.11990067399787159,
      "grad_norm": 1.6724108084246572,
      "learning_rate": 2.999248192198438e-05,
      "loss": 0.1502,
      "step": 169
    },
    {
      "epoch": 0.12061014544164597,
      "grad_norm": 0.6941077898344812,
      "learning_rate": 2.999191476541326e-05,
      "loss": 0.1366,
      "step": 170
    },
    {
      "epoch": 0.12131961688542037,
      "grad_norm": 3.2514684923862847,
      "learning_rate": 2.99913269925114e-05,
      "loss": 0.1134,
      "step": 171
    },
    {
      "epoch": 0.12202908832919475,
      "grad_norm": 5.4406959050041825,
      "learning_rate": 2.9990718604087083e-05,
      "loss": 0.1601,
      "step": 172
    },
    {
      "epoch": 0.12273855977296914,
      "grad_norm": 1.0046057813699996,
      "learning_rate": 2.9990089600976943e-05,
      "loss": 0.1588,
      "step": 173
    },
    {
      "epoch": 0.12344803121674353,
      "grad_norm": 1.2977013630239098,
      "learning_rate": 2.9989439984045962e-05,
      "loss": 0.1541,
      "step": 174
    },
    {
      "epoch": 0.12415750266051792,
      "grad_norm": 1.2466706411929727,
      "learning_rate": 2.998876975418747e-05,
      "loss": 0.1339,
      "step": 175
    },
    {
      "epoch": 0.1248669741042923,
      "grad_norm": 1.285065283020804,
      "learning_rate": 2.998807891232314e-05,
      "loss": 0.1325,
      "step": 176
    },
    {
      "epoch": 0.12557644554806668,
      "grad_norm": 1.0978226862258122,
      "learning_rate": 2.9987367459402993e-05,
      "loss": 0.141,
      "step": 177
    },
    {
      "epoch": 0.12628591699184108,
      "grad_norm": 1.539644970508895,
      "learning_rate": 2.9986635396405396e-05,
      "loss": 0.1082,
      "step": 178
    },
    {
      "epoch": 0.12699538843561547,
      "grad_norm": 0.8790121904376443,
      "learning_rate": 2.9985882724337055e-05,
      "loss": 0.1263,
      "step": 179
    },
    {
      "epoch": 0.12770485987938984,
      "grad_norm": 5.888354998159072,
      "learning_rate": 2.9985109444233012e-05,
      "loss": 0.1816,
      "step": 180
    },
    {
      "epoch": 0.12841433132316424,
      "grad_norm": 6.283291135812899,
      "learning_rate": 2.9984315557156654e-05,
      "loss": 0.1732,
      "step": 181
    },
    {
      "epoch": 0.12912380276693863,
      "grad_norm": 5.67047076772938,
      "learning_rate": 2.9983501064199715e-05,
      "loss": 0.1738,
      "step": 182
    },
    {
      "epoch": 0.12983327421071303,
      "grad_norm": 2.318419128866347,
      "learning_rate": 2.9982665966482243e-05,
      "loss": 0.1361,
      "step": 183
    },
    {
      "epoch": 0.1305427456544874,
      "grad_norm": 4.044186313532784,
      "learning_rate": 2.9981810265152638e-05,
      "loss": 0.1836,
      "step": 184
    },
    {
      "epoch": 0.1312522170982618,
      "grad_norm": 1.5522070412375442,
      "learning_rate": 2.9980933961387633e-05,
      "loss": 0.1706,
      "step": 185
    },
    {
      "epoch": 0.1319616885420362,
      "grad_norm": 3.9437944692153404,
      "learning_rate": 2.9980037056392288e-05,
      "loss": 0.1451,
      "step": 186
    },
    {
      "epoch": 0.13267115998581058,
      "grad_norm": 4.170515043293512,
      "learning_rate": 2.9979119551399988e-05,
      "loss": 0.1455,
      "step": 187
    },
    {
      "epoch": 0.13338063142958495,
      "grad_norm": 6.827705965643639,
      "learning_rate": 2.997818144767246e-05,
      "loss": 0.1712,
      "step": 188
    },
    {
      "epoch": 0.13409010287335935,
      "grad_norm": 0.7233782653093402,
      "learning_rate": 2.997722274649974e-05,
      "loss": 0.1506,
      "step": 189
    },
    {
      "epoch": 0.13479957431713374,
      "grad_norm": 4.118660841208239,
      "learning_rate": 2.9976243449200214e-05,
      "loss": 0.1546,
      "step": 190
    },
    {
      "epoch": 0.13550904576090814,
      "grad_norm": 5.529836536170726,
      "learning_rate": 2.997524355712056e-05,
      "loss": 0.1738,
      "step": 191
    },
    {
      "epoch": 0.1362185172046825,
      "grad_norm": 0.826889581023787,
      "learning_rate": 2.99742230716358e-05,
      "loss": 0.153,
      "step": 192
    },
    {
      "epoch": 0.1369279886484569,
      "grad_norm": 0.8498557920323406,
      "learning_rate": 2.997318199414927e-05,
      "loss": 0.1296,
      "step": 193
    },
    {
      "epoch": 0.1376374600922313,
      "grad_norm": 1.249135926994551,
      "learning_rate": 2.9972120326092613e-05,
      "loss": 0.128,
      "step": 194
    },
    {
      "epoch": 0.13834693153600566,
      "grad_norm": 8.206661775611908,
      "learning_rate": 2.9971038068925807e-05,
      "loss": 0.1792,
      "step": 195
    },
    {
      "epoch": 0.13905640297978006,
      "grad_norm": 6.085655991232297,
      "learning_rate": 2.9969935224137123e-05,
      "loss": 0.164,
      "step": 196
    },
    {
      "epoch": 0.13976587442355445,
      "grad_norm": 3.629154253059896,
      "learning_rate": 2.996881179324316e-05,
      "loss": 0.1397,
      "step": 197
    },
    {
      "epoch": 0.14047534586732885,
      "grad_norm": 1.2500476704771517,
      "learning_rate": 2.9967667777788818e-05,
      "loss": 0.1314,
      "step": 198
    },
    {
      "epoch": 0.14118481731110322,
      "grad_norm": 1.2512114220052228,
      "learning_rate": 2.99665031793473e-05,
      "loss": 0.179,
      "step": 199
    },
    {
      "epoch": 0.1418942887548776,
      "grad_norm": 2.1689630016652277,
      "learning_rate": 2.996531799952012e-05,
      "loss": 0.1304,
      "step": 200
    },
    {
      "epoch": 0.142603760198652,
      "grad_norm": 0.5537360236154936,
      "learning_rate": 2.9964112239937095e-05,
      "loss": 0.1103,
      "step": 201
    },
    {
      "epoch": 0.1433132316424264,
      "grad_norm": 2.1726248938535893,
      "learning_rate": 2.9962885902256344e-05,
      "loss": 0.1426,
      "step": 202
    },
    {
      "epoch": 0.14402270308620077,
      "grad_norm": 1.8049875872110488,
      "learning_rate": 2.9961638988164278e-05,
      "loss": 0.1645,
      "step": 203
    },
    {
      "epoch": 0.14473217452997517,
      "grad_norm": 1.5908502191048672,
      "learning_rate": 2.9960371499375615e-05,
      "loss": 0.1557,
      "step": 204
    },
    {
      "epoch": 0.14544164597374956,
      "grad_norm": 3.8328343525418402,
      "learning_rate": 2.9959083437633348e-05,
      "loss": 0.1435,
      "step": 205
    },
    {
      "epoch": 0.14615111741752396,
      "grad_norm": 0.9170631387728143,
      "learning_rate": 2.9957774804708776e-05,
      "loss": 0.131,
      "step": 206
    },
    {
      "epoch": 0.14686058886129832,
      "grad_norm": 0.516744620723013,
      "learning_rate": 2.9956445602401484e-05,
      "loss": 0.1467,
      "step": 207
    },
    {
      "epoch": 0.14757006030507272,
      "grad_norm": 1.1105168458368342,
      "learning_rate": 2.9955095832539343e-05,
      "loss": 0.1322,
      "step": 208
    },
    {
      "epoch": 0.14827953174884712,
      "grad_norm": 3.0557907501699098,
      "learning_rate": 2.9953725496978505e-05,
      "loss": 0.1494,
      "step": 209
    },
    {
      "epoch": 0.1489890031926215,
      "grad_norm": 1.4032502166139988,
      "learning_rate": 2.995233459760341e-05,
      "loss": 0.127,
      "step": 210
    },
    {
      "epoch": 0.14969847463639588,
      "grad_norm": 1.7576632481698398,
      "learning_rate": 2.9950923136326766e-05,
      "loss": 0.1413,
      "step": 211
    },
    {
      "epoch": 0.15040794608017027,
      "grad_norm": 2.1268817081467626,
      "learning_rate": 2.994949111508957e-05,
      "loss": 0.1097,
      "step": 212
    },
    {
      "epoch": 0.15111741752394467,
      "grad_norm": 2.2655473724165196,
      "learning_rate": 2.994803853586107e-05,
      "loss": 0.1619,
      "step": 213
    },
    {
      "epoch": 0.15182688896771904,
      "grad_norm": 1.8065357901680783,
      "learning_rate": 2.9946565400638818e-05,
      "loss": 0.1313,
      "step": 214
    },
    {
      "epoch": 0.15253636041149343,
      "grad_norm": 9.598841410309852,
      "learning_rate": 2.9945071711448603e-05,
      "loss": 0.2176,
      "step": 215
    },
    {
      "epoch": 0.15324583185526783,
      "grad_norm": 9.200294812878173,
      "learning_rate": 2.99435574703445e-05,
      "loss": 0.1787,
      "step": 216
    },
    {
      "epoch": 0.15395530329904222,
      "grad_norm": 9.621237210209282,
      "learning_rate": 2.9942022679408826e-05,
      "loss": 0.1582,
      "step": 217
    },
    {
      "epoch": 0.1546647747428166,
      "grad_norm": 6.3393844273573015,
      "learning_rate": 2.9940467340752182e-05,
      "loss": 0.1408,
      "step": 218
    },
    {
      "epoch": 0.155374246186591,
      "grad_norm": 4.840795961730082,
      "learning_rate": 2.9938891456513402e-05,
      "loss": 0.1333,
      "step": 219
    },
    {
      "epoch": 0.15608371763036538,
      "grad_norm": 3.720650273334849,
      "learning_rate": 2.9937295028859593e-05,
      "loss": 0.1525,
      "step": 220
    },
    {
      "epoch": 0.15679318907413978,
      "grad_norm": 4.234884490683812,
      "learning_rate": 2.99356780599861e-05,
      "loss": 0.1505,
      "step": 221
    },
    {
      "epoch": 0.15750266051791414,
      "grad_norm": 6.382042655623221,
      "learning_rate": 2.993404055211652e-05,
      "loss": 0.1736,
      "step": 222
    },
    {
      "epoch": 0.15821213196168854,
      "grad_norm": 3.9937541495319535,
      "learning_rate": 2.993238250750269e-05,
      "loss": 0.1497,
      "step": 223
    },
    {
      "epoch": 0.15892160340546294,
      "grad_norm": 3.7307886212055057,
      "learning_rate": 2.99307039284247e-05,
      "loss": 0.1617,
      "step": 224
    },
    {
      "epoch": 0.15963107484923733,
      "grad_norm": 0.6089181681224992,
      "learning_rate": 2.9929004817190857e-05,
      "loss": 0.1508,
      "step": 225
    },
    {
      "epoch": 0.1603405462930117,
      "grad_norm": 6.491861137387537,
      "learning_rate": 2.992728517613773e-05,
      "loss": 0.163,
      "step": 226
    },
    {
      "epoch": 0.1610500177367861,
      "grad_norm": 7.29299471998691,
      "learning_rate": 2.9925545007630103e-05,
      "loss": 0.1609,
      "step": 227
    },
    {
      "epoch": 0.1617594891805605,
      "grad_norm": 5.558761928431095,
      "learning_rate": 2.9923784314060985e-05,
      "loss": 0.1522,
      "step": 228
    },
    {
      "epoch": 0.16246896062433486,
      "grad_norm": 4.676737637689696,
      "learning_rate": 2.9922003097851625e-05,
      "loss": 0.156,
      "step": 229
    },
    {
      "epoch": 0.16317843206810925,
      "grad_norm": 1.109245410519303,
      "learning_rate": 2.9920201361451477e-05,
      "loss": 0.1188,
      "step": 230
    },
    {
      "epoch": 0.16388790351188365,
      "grad_norm": 1.7134861440689493,
      "learning_rate": 2.9918379107338235e-05,
      "loss": 0.1818,
      "step": 231
    },
    {
      "epoch": 0.16459737495565804,
      "grad_norm": 2.86248653043296,
      "learning_rate": 2.9916536338017787e-05,
      "loss": 0.1429,
      "step": 232
    },
    {
      "epoch": 0.1653068463994324,
      "grad_norm": 1.522341404032908,
      "learning_rate": 2.9914673056024243e-05,
      "loss": 0.1335,
      "step": 233
    },
    {
      "epoch": 0.1660163178432068,
      "grad_norm": 2.704891450197095,
      "learning_rate": 2.9912789263919928e-05,
      "loss": 0.1257,
      "step": 234
    },
    {
      "epoch": 0.1667257892869812,
      "grad_norm": 2.445565592103712,
      "learning_rate": 2.9910884964295353e-05,
      "loss": 0.117,
      "step": 235
    },
    {
      "epoch": 0.1674352607307556,
      "grad_norm": 1.5433295964894533,
      "learning_rate": 2.9908960159769243e-05,
      "loss": 0.1637,
      "step": 236
    },
    {
      "epoch": 0.16814473217452997,
      "grad_norm": 1.6964902329542633,
      "learning_rate": 2.9907014852988522e-05,
      "loss": 0.1207,
      "step": 237
    },
    {
      "epoch": 0.16885420361830436,
      "grad_norm": 2.1934180065676308,
      "learning_rate": 2.9905049046628303e-05,
      "loss": 0.1234,
      "step": 238
    },
    {
      "epoch": 0.16956367506207876,
      "grad_norm": 1.5659832880818358,
      "learning_rate": 2.9903062743391894e-05,
      "loss": 0.1526,
      "step": 239
    },
    {
      "epoch": 0.17027314650585315,
      "grad_norm": 0.6711661348290532,
      "learning_rate": 2.9901055946010774e-05,
      "loss": 0.1268,
      "step": 240
    },
    {
      "epoch": 0.17098261794962752,
      "grad_norm": 2.5141664232357654,
      "learning_rate": 2.9899028657244637e-05,
      "loss": 0.1428,
      "step": 241
    },
    {
      "epoch": 0.17169208939340191,
      "grad_norm": 1.6852753183414608,
      "learning_rate": 2.9896980879881316e-05,
      "loss": 0.1458,
      "step": 242
    },
    {
      "epoch": 0.1724015608371763,
      "grad_norm": 1.2287530689573898,
      "learning_rate": 2.9894912616736852e-05,
      "loss": 0.1509,
      "step": 243
    },
    {
      "epoch": 0.1731110322809507,
      "grad_norm": 2.15797008368182,
      "learning_rate": 2.9892823870655445e-05,
      "loss": 0.1491,
      "step": 244
    },
    {
      "epoch": 0.17382050372472507,
      "grad_norm": 0.5934695840028902,
      "learning_rate": 2.9890714644509457e-05,
      "loss": 0.1367,
      "step": 245
    },
    {
      "epoch": 0.17452997516849947,
      "grad_norm": 0.8632115516284793,
      "learning_rate": 2.9888584941199413e-05,
      "loss": 0.1279,
      "step": 246
    },
    {
      "epoch": 0.17523944661227386,
      "grad_norm": 1.1785833654497613,
      "learning_rate": 2.988643476365401e-05,
      "loss": 0.1187,
      "step": 247
    },
    {
      "epoch": 0.17594891805604823,
      "grad_norm": 0.4934583830601857,
      "learning_rate": 2.98842641148301e-05,
      "loss": 0.0923,
      "step": 248
    },
    {
      "epoch": 0.17665838949982263,
      "grad_norm": 2.3866116216635747,
      "learning_rate": 2.9882072997712665e-05,
      "loss": 0.1388,
      "step": 249
    },
    {
      "epoch": 0.17736786094359702,
      "grad_norm": 1.4356621377885848,
      "learning_rate": 2.9879861415314855e-05,
      "loss": 0.1391,
      "step": 250
    },
    {
      "epoch": 0.17807733238737142,
      "grad_norm": 1.1545314991373943,
      "learning_rate": 2.987762937067796e-05,
      "loss": 0.1281,
      "step": 251
    },
    {
      "epoch": 0.17878680383114579,
      "grad_norm": 1.0356937318996207,
      "learning_rate": 2.9875376866871404e-05,
      "loss": 0.1167,
      "step": 252
    },
    {
      "epoch": 0.17949627527492018,
      "grad_norm": 2.9560047482944567,
      "learning_rate": 2.9873103906992745e-05,
      "loss": 0.1305,
      "step": 253
    },
    {
      "epoch": 0.18020574671869458,
      "grad_norm": 1.6347565687572132,
      "learning_rate": 2.9870810494167677e-05,
      "loss": 0.1126,
      "step": 254
    },
    {
      "epoch": 0.18091521816246897,
      "grad_norm": 3.338203557091556,
      "learning_rate": 2.9868496631550018e-05,
      "loss": 0.1433,
      "step": 255
    },
    {
      "epoch": 0.18162468960624334,
      "grad_norm": 1.4815978427913965,
      "learning_rate": 2.9866162322321703e-05,
      "loss": 0.116,
      "step": 256
    },
    {
      "epoch": 0.18233416105001773,
      "grad_norm": 0.5335135172108745,
      "learning_rate": 2.9863807569692787e-05,
      "loss": 0.142,
      "step": 257
    },
    {
      "epoch": 0.18304363249379213,
      "grad_norm": 0.7726140131518255,
      "learning_rate": 2.9861432376901444e-05,
      "loss": 0.1174,
      "step": 258
    },
    {
      "epoch": 0.18375310393756653,
      "grad_norm": 1.087347338174722,
      "learning_rate": 2.9859036747213955e-05,
      "loss": 0.1226,
      "step": 259
    },
    {
      "epoch": 0.1844625753813409,
      "grad_norm": 2.1907412461793307,
      "learning_rate": 2.9856620683924694e-05,
      "loss": 0.1433,
      "step": 260
    },
    {
      "epoch": 0.1851720468251153,
      "grad_norm": 2.053242402620567,
      "learning_rate": 2.9854184190356146e-05,
      "loss": 0.1345,
      "step": 261
    },
    {
      "epoch": 0.18588151826888968,
      "grad_norm": 0.6595206308982481,
      "learning_rate": 2.985172726985889e-05,
      "loss": 0.1359,
      "step": 262
    },
    {
      "epoch": 0.18659098971266405,
      "grad_norm": 2.928310521090971,
      "learning_rate": 2.984924992581159e-05,
      "loss": 0.1364,
      "step": 263
    },
    {
      "epoch": 0.18730046115643845,
      "grad_norm": 3.1265595749971524,
      "learning_rate": 2.9846752161620994e-05,
      "loss": 0.1261,
      "step": 264
    },
    {
      "epoch": 0.18800993260021284,
      "grad_norm": 0.9746166109729698,
      "learning_rate": 2.9844233980721944e-05,
      "loss": 0.1491,
      "step": 265
    },
    {
      "epoch": 0.18871940404398724,
      "grad_norm": 4.2207538483174805,
      "learning_rate": 2.9841695386577345e-05,
      "loss": 0.1479,
      "step": 266
    },
    {
      "epoch": 0.1894288754877616,
      "grad_norm": 3.0801986262170593,
      "learning_rate": 2.983913638267818e-05,
      "loss": 0.1366,
      "step": 267
    },
    {
      "epoch": 0.190138346931536,
      "grad_norm": 5.454712342237991,
      "learning_rate": 2.9836556972543502e-05,
      "loss": 0.1543,
      "step": 268
    },
    {
      "epoch": 0.1908478183753104,
      "grad_norm": 2.5964897962881768,
      "learning_rate": 2.9833957159720414e-05,
      "loss": 0.1309,
      "step": 269
    },
    {
      "epoch": 0.1915572898190848,
      "grad_norm": 1.6001161689430843,
      "learning_rate": 2.9831336947784083e-05,
      "loss": 0.1304,
      "step": 270
    },
    {
      "epoch": 0.19226676126285916,
      "grad_norm": 3.636452756146101,
      "learning_rate": 2.982869634033773e-05,
      "loss": 0.1372,
      "step": 271
    },
    {
      "epoch": 0.19297623270663355,
      "grad_norm": 3.0436634447896096,
      "learning_rate": 2.9826035341012623e-05,
      "loss": 0.1276,
      "step": 272
    },
    {
      "epoch": 0.19368570415040795,
      "grad_norm": 1.945146608882815,
      "learning_rate": 2.9823353953468073e-05,
      "loss": 0.1466,
      "step": 273
    },
    {
      "epoch": 0.19439517559418235,
      "grad_norm": 0.9953220714535693,
      "learning_rate": 2.982065218139142e-05,
      "loss": 0.1419,
      "step": 274
    },
    {
      "epoch": 0.1951046470379567,
      "grad_norm": 4.152907494451121,
      "learning_rate": 2.981793002849804e-05,
      "loss": 0.1487,
      "step": 275
    },
    {
      "epoch": 0.1958141184817311,
      "grad_norm": 7.202163788627905,
      "learning_rate": 2.9815187498531348e-05,
      "loss": 0.1616,
      "step": 276
    },
    {
      "epoch": 0.1965235899255055,
      "grad_norm": 7.888111914020001,
      "learning_rate": 2.9812424595262763e-05,
      "loss": 0.1892,
      "step": 277
    },
    {
      "epoch": 0.1972330613692799,
      "grad_norm": 4.432936829143072,
      "learning_rate": 2.9809641322491727e-05,
      "loss": 0.1529,
      "step": 278
    },
    {
      "epoch": 0.19794253281305427,
      "grad_norm": 3.506292642573495,
      "learning_rate": 2.9806837684045697e-05,
      "loss": 0.1278,
      "step": 279
    },
    {
      "epoch": 0.19865200425682866,
      "grad_norm": 1.3127662675754346,
      "learning_rate": 2.9804013683780137e-05,
      "loss": 0.1271,
      "step": 280
    },
    {
      "epoch": 0.19936147570060306,
      "grad_norm": 4.370928442712123,
      "learning_rate": 2.98011693255785e-05,
      "loss": 0.1514,
      "step": 281
    },
    {
      "epoch": 0.20007094714437743,
      "grad_norm": 6.4793207972969515,
      "learning_rate": 2.979830461335225e-05,
      "loss": 0.1657,
      "step": 282
    },
    {
      "epoch": 0.20078041858815182,
      "grad_norm": 3.671625812708481,
      "learning_rate": 2.9795419551040836e-05,
      "loss": 0.1394,
      "step": 283
    },
    {
      "epoch": 0.20148989003192622,
      "grad_norm": 3.9538699689628647,
      "learning_rate": 2.9792514142611683e-05,
      "loss": 0.125,
      "step": 284
    },
    {
      "epoch": 0.2021993614757006,
      "grad_norm": 0.5715475190224388,
      "learning_rate": 2.9789588392060205e-05,
      "loss": 0.1169,
      "step": 285
    },
    {
      "epoch": 0.20290883291947498,
      "grad_norm": 4.741581016171732,
      "learning_rate": 2.9786642303409785e-05,
      "loss": 0.1379,
      "step": 286
    },
    {
      "epoch": 0.20361830436324937,
      "grad_norm": 3.5228025054256613,
      "learning_rate": 2.9783675880711783e-05,
      "loss": 0.1374,
      "step": 287
    },
    {
      "epoch": 0.20432777580702377,
      "grad_norm": 3.1024930941820537,
      "learning_rate": 2.9780689128045504e-05,
      "loss": 0.1663,
      "step": 288
    },
    {
      "epoch": 0.20503724725079817,
      "grad_norm": 3.064924835016314,
      "learning_rate": 2.9777682049518226e-05,
      "loss": 0.1604,
      "step": 289
    },
    {
      "epoch": 0.20574671869457253,
      "grad_norm": 2.6375689865745677,
      "learning_rate": 2.9774654649265173e-05,
      "loss": 0.1263,
      "step": 290
    },
    {
      "epoch": 0.20645619013834693,
      "grad_norm": 1.610926099534686,
      "learning_rate": 2.977160693144952e-05,
      "loss": 0.1261,
      "step": 291
    },
    {
      "epoch": 0.20716566158212132,
      "grad_norm": 4.594815416228411,
      "learning_rate": 2.976853890026236e-05,
      "loss": 0.1441,
      "step": 292
    },
    {
      "epoch": 0.20787513302589572,
      "grad_norm": 2.957834181803716,
      "learning_rate": 2.976545055992276e-05,
      "loss": 0.1336,
      "step": 293
    },
    {
      "epoch": 0.2085846044696701,
      "grad_norm": 0.484227879101729,
      "learning_rate": 2.976234191467767e-05,
      "loss": 0.1265,
      "step": 294
    },
    {
      "epoch": 0.20929407591344448,
      "grad_norm": 4.130668950574084,
      "learning_rate": 2.9759212968802e-05,
      "loss": 0.1345,
      "step": 295
    },
    {
      "epoch": 0.21000354735721888,
      "grad_norm": 6.451025714470222,
      "learning_rate": 2.9756063726598554e-05,
      "loss": 0.1967,
      "step": 296
    },
    {
      "epoch": 0.21071301880099327,
      "grad_norm": 4.992745848175137,
      "learning_rate": 2.9752894192398057e-05,
      "loss": 0.1506,
      "step": 297
    },
    {
      "epoch": 0.21142249024476764,
      "grad_norm": 2.9879713003816724,
      "learning_rate": 2.974970437055914e-05,
      "loss": 0.1262,
      "step": 298
    },
    {
      "epoch": 0.21213196168854204,
      "grad_norm": 1.0028224046748944,
      "learning_rate": 2.974649426546832e-05,
      "loss": 0.1342,
      "step": 299
    },
    {
      "epoch": 0.21284143313231643,
      "grad_norm": 2.8480083319182325,
      "learning_rate": 2.974326388154002e-05,
      "loss": 0.1305,
      "step": 300
    },
    {
      "epoch": 0.2135509045760908,
      "grad_norm": 3.5670682109244027,
      "learning_rate": 2.9740013223216548e-05,
      "loss": 0.139,
      "step": 301
    },
    {
      "epoch": 0.2142603760198652,
      "grad_norm": 4.212023817298028,
      "learning_rate": 2.9736742294968088e-05,
      "loss": 0.13,
      "step": 302
    },
    {
      "epoch": 0.2149698474636396,
      "grad_norm": 3.587089019566683,
      "learning_rate": 2.9733451101292698e-05,
      "loss": 0.1265,
      "step": 303
    },
    {
      "epoch": 0.21567931890741399,
      "grad_norm": 2.45071936937159,
      "learning_rate": 2.9730139646716308e-05,
      "loss": 0.1382,
      "step": 304
    },
    {
      "epoch": 0.21638879035118835,
      "grad_norm": 4.056735872956379,
      "learning_rate": 2.9726807935792714e-05,
      "loss": 0.1686,
      "step": 305
    },
    {
      "epoch": 0.21709826179496275,
      "grad_norm": 2.97078857249005,
      "learning_rate": 2.9723455973103554e-05,
      "loss": 0.1313,
      "step": 306
    },
    {
      "epoch": 0.21780773323873714,
      "grad_norm": 3.5121573353028808,
      "learning_rate": 2.9720083763258326e-05,
      "loss": 0.1112,
      "step": 307
    },
    {
      "epoch": 0.21851720468251154,
      "grad_norm": 3.18766256760675,
      "learning_rate": 2.971669131089438e-05,
      "loss": 0.113,
      "step": 308
    },
    {
      "epoch": 0.2192266761262859,
      "grad_norm": 1.0054652212444466,
      "learning_rate": 2.9713278620676876e-05,
      "loss": 0.1408,
      "step": 309
    },
    {
      "epoch": 0.2199361475700603,
      "grad_norm": 3.0557497330346415,
      "learning_rate": 2.970984569729883e-05,
      "loss": 0.1144,
      "step": 310
    },
    {
      "epoch": 0.2206456190138347,
      "grad_norm": 3.9053591865675474,
      "learning_rate": 2.970639254548107e-05,
      "loss": 0.1267,
      "step": 311
    },
    {
      "epoch": 0.2213550904576091,
      "grad_norm": 2.9315526595020285,
      "learning_rate": 2.970291916997224e-05,
      "loss": 0.1339,
      "step": 312
    },
    {
      "epoch": 0.22206456190138346,
      "grad_norm": 3.537025860153353,
      "learning_rate": 2.9699425575548802e-05,
      "loss": 0.1377,
      "step": 313
    },
    {
      "epoch": 0.22277403334515786,
      "grad_norm": 0.646576810328437,
      "learning_rate": 2.9695911767015024e-05,
      "loss": 0.1005,
      "step": 314
    },
    {
      "epoch": 0.22348350478893225,
      "grad_norm": 3.0375309551799776,
      "learning_rate": 2.9692377749202953e-05,
      "loss": 0.1321,
      "step": 315
    },
    {
      "epoch": 0.22419297623270662,
      "grad_norm": 2.160070387267167,
      "learning_rate": 2.9688823526972453e-05,
      "loss": 0.1141,
      "step": 316
    },
    {
      "epoch": 0.22490244767648102,
      "grad_norm": 3.6804450262657395,
      "learning_rate": 2.9685249105211154e-05,
      "loss": 0.1494,
      "step": 317
    },
    {
      "epoch": 0.2256119191202554,
      "grad_norm": 1.6218354531769252,
      "learning_rate": 2.9681654488834472e-05,
      "loss": 0.1351,
      "step": 318
    },
    {
      "epoch": 0.2263213905640298,
      "grad_norm": 1.0380972366514776,
      "learning_rate": 2.9678039682785594e-05,
      "loss": 0.0999,
      "step": 319
    },
    {
      "epoch": 0.22703086200780417,
      "grad_norm": 2.369119083792405,
      "learning_rate": 2.9674404692035458e-05,
      "loss": 0.1286,
      "step": 320
    },
    {
      "epoch": 0.22774033345157857,
      "grad_norm": 1.3776602694633155,
      "learning_rate": 2.9670749521582778e-05,
      "loss": 0.1197,
      "step": 321
    },
    {
      "epoch": 0.22844980489535296,
      "grad_norm": 1.503611884765329,
      "learning_rate": 2.9667074176454013e-05,
      "loss": 0.1108,
      "step": 322
    },
    {
      "epoch": 0.22915927633912736,
      "grad_norm": 1.9343104036169902,
      "learning_rate": 2.9663378661703356e-05,
      "loss": 0.1386,
      "step": 323
    },
    {
      "epoch": 0.22986874778290173,
      "grad_norm": 1.3450282783534633,
      "learning_rate": 2.965966298241275e-05,
      "loss": 0.1091,
      "step": 324
    },
    {
      "epoch": 0.23057821922667612,
      "grad_norm": 1.1669298363795306,
      "learning_rate": 2.965592714369185e-05,
      "loss": 0.1382,
      "step": 325
    },
    {
      "epoch": 0.23128769067045052,
      "grad_norm": 0.7340938703579164,
      "learning_rate": 2.9652171150678055e-05,
      "loss": 0.1027,
      "step": 326
    },
    {
      "epoch": 0.23199716211422491,
      "grad_norm": 1.0767838107063834,
      "learning_rate": 2.9648395008536464e-05,
      "loss": 0.1167,
      "step": 327
    },
    {
      "epoch": 0.23270663355799928,
      "grad_norm": 1.0586204854666508,
      "learning_rate": 2.964459872245989e-05,
      "loss": 0.106,
      "step": 328
    },
    {
      "epoch": 0.23341610500177368,
      "grad_norm": 1.7488457522091547,
      "learning_rate": 2.9640782297668845e-05,
      "loss": 0.1174,
      "step": 329
    },
    {
      "epoch": 0.23412557644554807,
      "grad_norm": 1.447672995304521,
      "learning_rate": 2.9636945739411533e-05,
      "loss": 0.1267,
      "step": 330
    },
    {
      "epoch": 0.23483504788932247,
      "grad_norm": 2.005280272097454,
      "learning_rate": 2.9633089052963858e-05,
      "loss": 0.1238,
      "step": 331
    },
    {
      "epoch": 0.23554451933309684,
      "grad_norm": 1.5411679400453127,
      "learning_rate": 2.9629212243629387e-05,
      "loss": 0.1141,
      "step": 332
    },
    {
      "epoch": 0.23625399077687123,
      "grad_norm": 0.6149683068305483,
      "learning_rate": 2.9625315316739365e-05,
      "loss": 0.1137,
      "step": 333
    },
    {
      "epoch": 0.23696346222064563,
      "grad_norm": 0.6354682310356766,
      "learning_rate": 2.9621398277652703e-05,
      "loss": 0.1523,
      "step": 334
    },
    {
      "epoch": 0.23767293366442,
      "grad_norm": 0.4258397242446468,
      "learning_rate": 2.9617461131755974e-05,
      "loss": 0.1004,
      "step": 335
    },
    {
      "epoch": 0.2383824051081944,
      "grad_norm": 0.845449244365038,
      "learning_rate": 2.961350388446339e-05,
      "loss": 0.1508,
      "step": 336
    },
    {
      "epoch": 0.23909187655196878,
      "grad_norm": 1.1920803712120955,
      "learning_rate": 2.9609526541216817e-05,
      "loss": 0.1069,
      "step": 337
    },
    {
      "epoch": 0.23980134799574318,
      "grad_norm": 1.4617256282688038,
      "learning_rate": 2.960552910748575e-05,
      "loss": 0.11,
      "step": 338
    },
    {
      "epoch": 0.24051081943951755,
      "grad_norm": 1.4313623063982495,
      "learning_rate": 2.9601511588767314e-05,
      "loss": 0.1151,
      "step": 339
    },
    {
      "epoch": 0.24122029088329194,
      "grad_norm": 0.5953978626552301,
      "learning_rate": 2.959747399058625e-05,
      "loss": 0.127,
      "step": 340
    },
    {
      "epoch": 0.24192976232706634,
      "grad_norm": 2.977298376039244,
      "learning_rate": 2.9593416318494916e-05,
      "loss": 0.1247,
      "step": 341
    },
    {
      "epoch": 0.24263923377084073,
      "grad_norm": 2.341990467465989,
      "learning_rate": 2.9589338578073277e-05,
      "loss": 0.1257,
      "step": 342
    },
    {
      "epoch": 0.2433487052146151,
      "grad_norm": 0.9654180090165114,
      "learning_rate": 2.9585240774928887e-05,
      "loss": 0.1399,
      "step": 343
    },
    {
      "epoch": 0.2440581766583895,
      "grad_norm": 0.4976154855602877,
      "learning_rate": 2.9581122914696898e-05,
      "loss": 0.1194,
      "step": 344
    },
    {
      "epoch": 0.2447676481021639,
      "grad_norm": 0.8093574207224504,
      "learning_rate": 2.957698500304004e-05,
      "loss": 0.1245,
      "step": 345
    },
    {
      "epoch": 0.2454771195459383,
      "grad_norm": 2.4496597624457164,
      "learning_rate": 2.9572827045648614e-05,
      "loss": 0.1109,
      "step": 346
    },
    {
      "epoch": 0.24618659098971266,
      "grad_norm": 3.831466220403665,
      "learning_rate": 2.9568649048240496e-05,
      "loss": 0.1679,
      "step": 347
    },
    {
      "epoch": 0.24689606243348705,
      "grad_norm": 1.4317595227825495,
      "learning_rate": 2.9564451016561107e-05,
      "loss": 0.1596,
      "step": 348
    },
    {
      "epoch": 0.24760553387726145,
      "grad_norm": 0.6483178119650296,
      "learning_rate": 2.9560232956383433e-05,
      "loss": 0.1171,
      "step": 349
    },
    {
      "epoch": 0.24831500532103584,
      "grad_norm": 2.0583845057044936,
      "learning_rate": 2.9555994873507992e-05,
      "loss": 0.1423,
      "step": 350
    },
    {
      "epoch": 0.2490244767648102,
      "grad_norm": 3.879641589013307,
      "learning_rate": 2.955173677376284e-05,
      "loss": 0.1276,
      "step": 351
    },
    {
      "epoch": 0.2497339482085846,
      "grad_norm": 1.3769005993971457,
      "learning_rate": 2.954745866300356e-05,
      "loss": 0.1163,
      "step": 352
    },
    {
      "epoch": 0.250443419652359,
      "grad_norm": 1.8376354282810619,
      "learning_rate": 2.9543160547113254e-05,
      "loss": 0.153,
      "step": 353
    },
    {
      "epoch": 0.25115289109613337,
      "grad_norm": 1.9844824326325108,
      "learning_rate": 2.953884243200253e-05,
      "loss": 0.0909,
      "step": 354
    },
    {
      "epoch": 0.2518623625399078,
      "grad_norm": 1.8353955039268464,
      "learning_rate": 2.9534504323609504e-05,
      "loss": 0.134,
      "step": 355
    },
    {
      "epoch": 0.25257183398368216,
      "grad_norm": 1.0110178319854,
      "learning_rate": 2.9530146227899787e-05,
      "loss": 0.1708,
      "step": 356
    },
    {
      "epoch": 0.2532813054274565,
      "grad_norm": 1.0238365805291345,
      "learning_rate": 2.9525768150866465e-05,
      "loss": 0.141,
      "step": 357
    },
    {
      "epoch": 0.25399077687123095,
      "grad_norm": 0.41824371907324315,
      "learning_rate": 2.952137009853011e-05,
      "loss": 0.1349,
      "step": 358
    },
    {
      "epoch": 0.2547002483150053,
      "grad_norm": 0.41464086641304115,
      "learning_rate": 2.9516952076938768e-05,
      "loss": 0.1321,
      "step": 359
    },
    {
      "epoch": 0.2554097197587797,
      "grad_norm": 0.7932581870199277,
      "learning_rate": 2.9512514092167933e-05,
      "loss": 0.1424,
      "step": 360
    },
    {
      "epoch": 0.2561191912025541,
      "grad_norm": 0.47130216597421964,
      "learning_rate": 2.9508056150320565e-05,
      "loss": 0.124,
      "step": 361
    },
    {
      "epoch": 0.2568286626463285,
      "grad_norm": 3.5373335369279273,
      "learning_rate": 2.950357825752706e-05,
      "loss": 0.1225,
      "step": 362
    },
    {
      "epoch": 0.2575381340901029,
      "grad_norm": 2.4843113969349346,
      "learning_rate": 2.9499080419945243e-05,
      "loss": 0.1242,
      "step": 363
    },
    {
      "epoch": 0.25824760553387727,
      "grad_norm": 0.858679728851391,
      "learning_rate": 2.9494562643760385e-05,
      "loss": 0.1068,
      "step": 364
    },
    {
      "epoch": 0.25895707697765163,
      "grad_norm": 1.3485967573023745,
      "learning_rate": 2.9490024935185167e-05,
      "loss": 0.1185,
      "step": 365
    },
    {
      "epoch": 0.25966654842142606,
      "grad_norm": 1.0040234129610184,
      "learning_rate": 2.9485467300459673e-05,
      "loss": 0.1098,
      "step": 366
    },
    {
      "epoch": 0.2603760198652004,
      "grad_norm": 1.0006381662096229,
      "learning_rate": 2.9480889745851396e-05,
      "loss": 0.1285,
      "step": 367
    },
    {
      "epoch": 0.2610854913089748,
      "grad_norm": 1.9738191133612824,
      "learning_rate": 2.9476292277655227e-05,
      "loss": 0.151,
      "step": 368
    },
    {
      "epoch": 0.2617949627527492,
      "grad_norm": 2.50954516418147,
      "learning_rate": 2.947167490219343e-05,
      "loss": 0.1281,
      "step": 369
    },
    {
      "epoch": 0.2625044341965236,
      "grad_norm": 2.366897710500404,
      "learning_rate": 2.946703762581565e-05,
      "loss": 0.1194,
      "step": 370
    },
    {
      "epoch": 0.26321390564029795,
      "grad_norm": 1.5891040342564644,
      "learning_rate": 2.9462380454898898e-05,
      "loss": 0.1201,
      "step": 371
    },
    {
      "epoch": 0.2639233770840724,
      "grad_norm": 0.7103642176294669,
      "learning_rate": 2.9457703395847552e-05,
      "loss": 0.122,
      "step": 372
    },
    {
      "epoch": 0.26463284852784674,
      "grad_norm": 1.5882807551220135,
      "learning_rate": 2.9453006455093324e-05,
      "loss": 0.132,
      "step": 373
    },
    {
      "epoch": 0.26534231997162117,
      "grad_norm": 3.506348182270578,
      "learning_rate": 2.944828963909528e-05,
      "loss": 0.1625,
      "step": 374
    },
    {
      "epoch": 0.26605179141539553,
      "grad_norm": 2.2167828603365187,
      "learning_rate": 2.9443552954339807e-05,
      "loss": 0.1281,
      "step": 375
    },
    {
      "epoch": 0.2667612628591699,
      "grad_norm": 0.7834681043304855,
      "learning_rate": 2.9438796407340624e-05,
      "loss": 0.1132,
      "step": 376
    },
    {
      "epoch": 0.2674707343029443,
      "grad_norm": 1.41517623538854,
      "learning_rate": 2.9434020004638757e-05,
      "loss": 0.1227,
      "step": 377
    },
    {
      "epoch": 0.2681802057467187,
      "grad_norm": 4.763582671829927,
      "learning_rate": 2.9429223752802543e-05,
      "loss": 0.1385,
      "step": 378
    },
    {
      "epoch": 0.26888967719049306,
      "grad_norm": 4.517989156270013,
      "learning_rate": 2.9424407658427605e-05,
      "loss": 0.1289,
      "step": 379
    },
    {
      "epoch": 0.2695991486342675,
      "grad_norm": 1.8668066600612216,
      "learning_rate": 2.9419571728136866e-05,
      "loss": 0.0931,
      "step": 380
    },
    {
      "epoch": 0.27030862007804185,
      "grad_norm": 2.4308728966392588,
      "learning_rate": 2.941471596858052e-05,
      "loss": 0.1127,
      "step": 381
    },
    {
      "epoch": 0.2710180915218163,
      "grad_norm": 2.1639101101062224,
      "learning_rate": 2.9409840386436024e-05,
      "loss": 0.1401,
      "step": 382
    },
    {
      "epoch": 0.27172756296559064,
      "grad_norm": 0.8909030254301304,
      "learning_rate": 2.94049449884081e-05,
      "loss": 0.0897,
      "step": 383
    },
    {
      "epoch": 0.272437034409365,
      "grad_norm": 1.0477063909729751,
      "learning_rate": 2.9400029781228722e-05,
      "loss": 0.1438,
      "step": 384
    },
    {
      "epoch": 0.27314650585313943,
      "grad_norm": 0.8902571869086673,
      "learning_rate": 2.9395094771657108e-05,
      "loss": 0.1001,
      "step": 385
    },
    {
      "epoch": 0.2738559772969138,
      "grad_norm": 1.448303545838668,
      "learning_rate": 2.9390139966479692e-05,
      "loss": 0.1141,
      "step": 386
    },
    {
      "epoch": 0.27456544874068817,
      "grad_norm": 1.1357321306457262,
      "learning_rate": 2.938516537251014e-05,
      "loss": 0.1046,
      "step": 387
    },
    {
      "epoch": 0.2752749201844626,
      "grad_norm": 5.65741028667349,
      "learning_rate": 2.9380170996589345e-05,
      "loss": 0.155,
      "step": 388
    },
    {
      "epoch": 0.27598439162823696,
      "grad_norm": 3.861699101319741,
      "learning_rate": 2.9375156845585374e-05,
      "loss": 0.1288,
      "step": 389
    },
    {
      "epoch": 0.2766938630720113,
      "grad_norm": 0.67675696886419,
      "learning_rate": 2.937012292639351e-05,
      "loss": 0.1302,
      "step": 390
    },
    {
      "epoch": 0.27740333451578575,
      "grad_norm": 1.4877319351369076,
      "learning_rate": 2.936506924593621e-05,
      "loss": 0.1207,
      "step": 391
    },
    {
      "epoch": 0.2781128059595601,
      "grad_norm": 1.2697475227356827,
      "learning_rate": 2.9359995811163114e-05,
      "loss": 0.1348,
      "step": 392
    },
    {
      "epoch": 0.27882227740333454,
      "grad_norm": 1.8358469663153398,
      "learning_rate": 2.9354902629051022e-05,
      "loss": 0.1361,
      "step": 393
    },
    {
      "epoch": 0.2795317488471089,
      "grad_norm": 1.5695843972913994,
      "learning_rate": 2.934978970660389e-05,
      "loss": 0.1009,
      "step": 394
    },
    {
      "epoch": 0.2802412202908833,
      "grad_norm": 0.5297163047032748,
      "learning_rate": 2.9344657050852825e-05,
      "loss": 0.142,
      "step": 395
    },
    {
      "epoch": 0.2809506917346577,
      "grad_norm": 1.1116725849481155,
      "learning_rate": 2.9339504668856064e-05,
      "loss": 0.133,
      "step": 396
    },
    {
      "epoch": 0.28166016317843207,
      "grad_norm": 0.7324661972256294,
      "learning_rate": 2.9334332567698973e-05,
      "loss": 0.1275,
      "step": 397
    },
    {
      "epoch": 0.28236963462220643,
      "grad_norm": 1.304101752615871,
      "learning_rate": 2.9329140754494035e-05,
      "loss": 0.1285,
      "step": 398
    },
    {
      "epoch": 0.28307910606598086,
      "grad_norm": 1.5052152972197361,
      "learning_rate": 2.9323929236380846e-05,
      "loss": 0.1254,
      "step": 399
    },
    {
      "epoch": 0.2837885775097552,
      "grad_norm": 1.5007070683932355,
      "learning_rate": 2.9318698020526084e-05,
      "loss": 0.136,
      "step": 400
    },
    {
      "epoch": 0.28449804895352965,
      "grad_norm": 1.2063360023678396,
      "learning_rate": 2.9313447114123532e-05,
      "loss": 0.1042,
      "step": 401
    },
    {
      "epoch": 0.285207520397304,
      "grad_norm": 0.7864089832257156,
      "learning_rate": 2.9308176524394045e-05,
      "loss": 0.1156,
      "step": 402
    },
    {
      "epoch": 0.2859169918410784,
      "grad_norm": 1.9480436539268096,
      "learning_rate": 2.9302886258585544e-05,
      "loss": 0.129,
      "step": 403
    },
    {
      "epoch": 0.2866264632848528,
      "grad_norm": 0.7817654765904787,
      "learning_rate": 2.9297576323973002e-05,
      "loss": 0.1115,
      "step": 404
    },
    {
      "epoch": 0.2873359347286272,
      "grad_norm": 1.2570405051850626,
      "learning_rate": 2.9292246727858457e-05,
      "loss": 0.0873,
      "step": 405
    },
    {
      "epoch": 0.28804540617240154,
      "grad_norm": 0.9784274165620066,
      "learning_rate": 2.9286897477570963e-05,
      "loss": 0.0862,
      "step": 406
    },
    {
      "epoch": 0.28875487761617596,
      "grad_norm": 1.3869879764201865,
      "learning_rate": 2.9281528580466618e-05,
      "loss": 0.1266,
      "step": 407
    },
    {
      "epoch": 0.28946434905995033,
      "grad_norm": 0.6665714564879087,
      "learning_rate": 2.9276140043928533e-05,
      "loss": 0.1319,
      "step": 408
    },
    {
      "epoch": 0.2901738205037247,
      "grad_norm": 0.7728563261780564,
      "learning_rate": 2.9270731875366825e-05,
      "loss": 0.1211,
      "step": 409
    },
    {
      "epoch": 0.2908832919474991,
      "grad_norm": 1.443579438086873,
      "learning_rate": 2.926530408221861e-05,
      "loss": 0.1419,
      "step": 410
    },
    {
      "epoch": 0.2915927633912735,
      "grad_norm": 0.6300572426399286,
      "learning_rate": 2.9259856671947978e-05,
      "loss": 0.0941,
      "step": 411
    },
    {
      "epoch": 0.2923022348350479,
      "grad_norm": 1.2666826561346438,
      "learning_rate": 2.9254389652046026e-05,
      "loss": 0.1176,
      "step": 412
    },
    {
      "epoch": 0.2930117062788223,
      "grad_norm": 2.852699353603261,
      "learning_rate": 2.924890303003079e-05,
      "loss": 0.1158,
      "step": 413
    },
    {
      "epoch": 0.29372117772259665,
      "grad_norm": 1.409721384356643,
      "learning_rate": 2.9243396813447265e-05,
      "loss": 0.1132,
      "step": 414
    },
    {
      "epoch": 0.2944306491663711,
      "grad_norm": 1.4840237921522614,
      "learning_rate": 2.9237871009867413e-05,
      "loss": 0.1174,
      "step": 415
    },
    {
      "epoch": 0.29514012061014544,
      "grad_norm": 2.920300665219914,
      "learning_rate": 2.9232325626890104e-05,
      "loss": 0.1402,
      "step": 416
    },
    {
      "epoch": 0.2958495920539198,
      "grad_norm": 2.0945564732858046,
      "learning_rate": 2.922676067214115e-05,
      "loss": 0.1194,
      "step": 417
    },
    {
      "epoch": 0.29655906349769423,
      "grad_norm": 3.785652600147428,
      "learning_rate": 2.9221176153273266e-05,
      "loss": 0.1601,
      "step": 418
    },
    {
      "epoch": 0.2972685349414686,
      "grad_norm": 1.7688322332829445,
      "learning_rate": 2.9215572077966085e-05,
      "loss": 0.1458,
      "step": 419
    },
    {
      "epoch": 0.297978006385243,
      "grad_norm": 1.214969690380084,
      "learning_rate": 2.920994845392613e-05,
      "loss": 0.1268,
      "step": 420
    },
    {
      "epoch": 0.2986874778290174,
      "grad_norm": 3.724258388362578,
      "learning_rate": 2.9204305288886788e-05,
      "loss": 0.1259,
      "step": 421
    },
    {
      "epoch": 0.29939694927279176,
      "grad_norm": 1.3752465754977663,
      "learning_rate": 2.919864259060834e-05,
      "loss": 0.1208,
      "step": 422
    },
    {
      "epoch": 0.3001064207165662,
      "grad_norm": 0.42692572508085563,
      "learning_rate": 2.9192960366877925e-05,
      "loss": 0.1153,
      "step": 423
    },
    {
      "epoch": 0.30081589216034055,
      "grad_norm": 1.8136871243949773,
      "learning_rate": 2.9187258625509518e-05,
      "loss": 0.1165,
      "step": 424
    },
    {
      "epoch": 0.3015253636041149,
      "grad_norm": 0.7748878262808272,
      "learning_rate": 2.9181537374343958e-05,
      "loss": 0.1264,
      "step": 425
    },
    {
      "epoch": 0.30223483504788934,
      "grad_norm": 2.611048369550947,
      "learning_rate": 2.9175796621248883e-05,
      "loss": 0.1378,
      "step": 426
    },
    {
      "epoch": 0.3029443064916637,
      "grad_norm": 0.6025666339471265,
      "learning_rate": 2.917003637411878e-05,
      "loss": 0.1061,
      "step": 427
    },
    {
      "epoch": 0.3036537779354381,
      "grad_norm": 2.726109616999527,
      "learning_rate": 2.9164256640874924e-05,
      "loss": 0.1413,
      "step": 428
    },
    {
      "epoch": 0.3043632493792125,
      "grad_norm": 1.3606865675089705,
      "learning_rate": 2.9158457429465394e-05,
      "loss": 0.1104,
      "step": 429
    },
    {
      "epoch": 0.30507272082298686,
      "grad_norm": 0.8802791707996749,
      "learning_rate": 2.9152638747865047e-05,
      "loss": 0.1296,
      "step": 430
    },
    {
      "epoch": 0.3057821922667613,
      "grad_norm": 0.8589442882441348,
      "learning_rate": 2.9146800604075526e-05,
      "loss": 0.1271,
      "step": 431
    },
    {
      "epoch": 0.30649166371053566,
      "grad_norm": 3.273544291678086,
      "learning_rate": 2.914094300612523e-05,
      "loss": 0.1387,
      "step": 432
    },
    {
      "epoch": 0.30720113515431,
      "grad_norm": 1.5045262793650462,
      "learning_rate": 2.9135065962069314e-05,
      "loss": 0.1414,
      "step": 433
    },
    {
      "epoch": 0.30791060659808445,
      "grad_norm": 0.734076000747538,
      "learning_rate": 2.9129169479989665e-05,
      "loss": 0.1154,
      "step": 434
    },
    {
      "epoch": 0.3086200780418588,
      "grad_norm": 3.7833323986694882,
      "learning_rate": 2.9123253567994925e-05,
      "loss": 0.1326,
      "step": 435
    },
    {
      "epoch": 0.3093295494856332,
      "grad_norm": 2.192228327755835,
      "learning_rate": 2.9117318234220425e-05,
      "loss": 0.1269,
      "step": 436
    },
    {
      "epoch": 0.3100390209294076,
      "grad_norm": 1.5546002303577293,
      "learning_rate": 2.911136348682822e-05,
      "loss": 0.1135,
      "step": 437
    },
    {
      "epoch": 0.310748492373182,
      "grad_norm": 1.130438794878335,
      "learning_rate": 2.9105389334007067e-05,
      "loss": 0.1274,
      "step": 438
    },
    {
      "epoch": 0.31145796381695634,
      "grad_norm": 1.1457972542843764,
      "learning_rate": 2.9099395783972392e-05,
      "loss": 0.1042,
      "step": 439
    },
    {
      "epoch": 0.31216743526073076,
      "grad_norm": 1.666814189686001,
      "learning_rate": 2.9093382844966315e-05,
      "loss": 0.1316,
      "step": 440
    },
    {
      "epoch": 0.31287690670450513,
      "grad_norm": 1.5079787467677475,
      "learning_rate": 2.90873505252576e-05,
      "loss": 0.1191,
      "step": 441
    },
    {
      "epoch": 0.31358637814827955,
      "grad_norm": 2.237784958259717,
      "learning_rate": 2.9081298833141678e-05,
      "loss": 0.134,
      "step": 442
    },
    {
      "epoch": 0.3142958495920539,
      "grad_norm": 3.3782849256322582,
      "learning_rate": 2.907522777694061e-05,
      "loss": 0.1129,
      "step": 443
    },
    {
      "epoch": 0.3150053210358283,
      "grad_norm": 2.531064255370323,
      "learning_rate": 2.9069137365003092e-05,
      "loss": 0.1373,
      "step": 444
    },
    {
      "epoch": 0.3157147924796027,
      "grad_norm": 0.7317744106761535,
      "learning_rate": 2.9063027605704434e-05,
      "loss": 0.0991,
      "step": 445
    },
    {
      "epoch": 0.3164242639233771,
      "grad_norm": 1.782380867213432,
      "learning_rate": 2.9056898507446553e-05,
      "loss": 0.1208,
      "step": 446
    },
    {
      "epoch": 0.31713373536715145,
      "grad_norm": 2.401405892283787,
      "learning_rate": 2.9050750078657956e-05,
      "loss": 0.1117,
      "step": 447
    },
    {
      "epoch": 0.31784320681092587,
      "grad_norm": 2.1801395152392304,
      "learning_rate": 2.9044582327793745e-05,
      "loss": 0.1083,
      "step": 448
    },
    {
      "epoch": 0.31855267825470024,
      "grad_norm": 0.6376062377860501,
      "learning_rate": 2.9038395263335577e-05,
      "loss": 0.0866,
      "step": 449
    },
    {
      "epoch": 0.31926214969847466,
      "grad_norm": 1.8980870479476657,
      "learning_rate": 2.903218889379168e-05,
      "loss": 0.1096,
      "step": 450
    },
    {
      "epoch": 0.31997162114224903,
      "grad_norm": 0.8441546985189771,
      "learning_rate": 2.902596322769682e-05,
      "loss": 0.1269,
      "step": 451
    },
    {
      "epoch": 0.3206810925860234,
      "grad_norm": 3.097844971673133,
      "learning_rate": 2.901971827361231e-05,
      "loss": 0.1442,
      "step": 452
    },
    {
      "epoch": 0.3213905640297978,
      "grad_norm": 1.0097737198624575,
      "learning_rate": 2.901345404012598e-05,
      "loss": 0.0995,
      "step": 453
    },
    {
      "epoch": 0.3221000354735722,
      "grad_norm": 0.7485286675934367,
      "learning_rate": 2.9007170535852176e-05,
      "loss": 0.1243,
      "step": 454
    },
    {
      "epoch": 0.32280950691734656,
      "grad_norm": 3.974146052255681,
      "learning_rate": 2.9000867769431734e-05,
      "loss": 0.1158,
      "step": 455
    },
    {
      "epoch": 0.323518978361121,
      "grad_norm": 5.153830169966646,
      "learning_rate": 2.8994545749532003e-05,
      "loss": 0.1243,
      "step": 456
    },
    {
      "epoch": 0.32422844980489535,
      "grad_norm": 4.522082737227656,
      "learning_rate": 2.8988204484846774e-05,
      "loss": 0.1452,
      "step": 457
    },
    {
      "epoch": 0.3249379212486697,
      "grad_norm": 3.816894032844365,
      "learning_rate": 2.8981843984096338e-05,
      "loss": 0.1378,
      "step": 458
    },
    {
      "epoch": 0.32564739269244414,
      "grad_norm": 1.3357144795947238,
      "learning_rate": 2.897546425602741e-05,
      "loss": 0.0992,
      "step": 459
    },
    {
      "epoch": 0.3263568641362185,
      "grad_norm": 1.7656239034092869,
      "learning_rate": 2.8969065309413158e-05,
      "loss": 0.1159,
      "step": 460
    },
    {
      "epoch": 0.32706633557999293,
      "grad_norm": 2.6594704803126126,
      "learning_rate": 2.8962647153053187e-05,
      "loss": 0.1298,
      "step": 461
    },
    {
      "epoch": 0.3277758070237673,
      "grad_norm": 3.470766271987268,
      "learning_rate": 2.8956209795773502e-05,
      "loss": 0.1348,
      "step": 462
    },
    {
      "epoch": 0.32848527846754166,
      "grad_norm": 2.426484992688522,
      "learning_rate": 2.894975324642653e-05,
      "loss": 0.1102,
      "step": 463
    },
    {
      "epoch": 0.3291947499113161,
      "grad_norm": 2.0087837128641763,
      "learning_rate": 2.894327751389107e-05,
      "loss": 0.1322,
      "step": 464
    },
    {
      "epoch": 0.32990422135509045,
      "grad_norm": 2.1044358389626665,
      "learning_rate": 2.8936782607072316e-05,
      "loss": 0.1286,
      "step": 465
    },
    {
      "epoch": 0.3306136927988648,
      "grad_norm": 2.9930476942786894,
      "learning_rate": 2.8930268534901823e-05,
      "loss": 0.1251,
      "step": 466
    },
    {
      "epoch": 0.33132316424263925,
      "grad_norm": 2.1925012098009247,
      "learning_rate": 2.892373530633751e-05,
      "loss": 0.1347,
      "step": 467
    },
    {
      "epoch": 0.3320326356864136,
      "grad_norm": 0.9158124499144366,
      "learning_rate": 2.891718293036363e-05,
      "loss": 0.1087,
      "step": 468
    },
    {
      "epoch": 0.33274210713018804,
      "grad_norm": 0.9890067798699078,
      "learning_rate": 2.8910611415990766e-05,
      "loss": 0.0855,
      "step": 469
    },
    {
      "epoch": 0.3334515785739624,
      "grad_norm": 1.2152282929004945,
      "learning_rate": 2.890402077225583e-05,
      "loss": 0.1302,
      "step": 470
    },
    {
      "epoch": 0.33416105001773677,
      "grad_norm": 3.1961672858202066,
      "learning_rate": 2.8897411008222026e-05,
      "loss": 0.1174,
      "step": 471
    },
    {
      "epoch": 0.3348705214615112,
      "grad_norm": 3.4748868366483716,
      "learning_rate": 2.8890782132978866e-05,
      "loss": 0.1117,
      "step": 472
    },
    {
      "epoch": 0.33557999290528556,
      "grad_norm": 1.3038400253004467,
      "learning_rate": 2.888413415564213e-05,
      "loss": 0.1051,
      "step": 473
    },
    {
      "epoch": 0.33628946434905993,
      "grad_norm": 0.6255093137268427,
      "learning_rate": 2.887746708535388e-05,
      "loss": 0.111,
      "step": 474
    },
    {
      "epoch": 0.33699893579283435,
      "grad_norm": 1.34689232058292,
      "learning_rate": 2.8870780931282412e-05,
      "loss": 0.1131,
      "step": 475
    },
    {
      "epoch": 0.3377084072366087,
      "grad_norm": 2.4879808829860743,
      "learning_rate": 2.8864075702622295e-05,
      "loss": 0.1444,
      "step": 476
    },
    {
      "epoch": 0.3384178786803831,
      "grad_norm": 2.2842096171823973,
      "learning_rate": 2.8857351408594303e-05,
      "loss": 0.116,
      "step": 477
    },
    {
      "epoch": 0.3391273501241575,
      "grad_norm": 1.6083058694068209,
      "learning_rate": 2.885060805844545e-05,
      "loss": 0.1155,
      "step": 478
    },
    {
      "epoch": 0.3398368215679319,
      "grad_norm": 1.1674603168326303,
      "learning_rate": 2.884384566144893e-05,
      "loss": 0.1089,
      "step": 479
    },
    {
      "epoch": 0.3405462930117063,
      "grad_norm": 0.7719232134916407,
      "learning_rate": 2.8837064226904145e-05,
      "loss": 0.1399,
      "step": 480
    },
    {
      "epoch": 0.34125576445548067,
      "grad_norm": 1.5218085739053615,
      "learning_rate": 2.883026376413668e-05,
      "loss": 0.1114,
      "step": 481
    },
    {
      "epoch": 0.34196523589925504,
      "grad_norm": 0.9109862142531445,
      "learning_rate": 2.8823444282498288e-05,
      "loss": 0.1179,
      "step": 482
    },
    {
      "epoch": 0.34267470734302946,
      "grad_norm": 0.5573121486290471,
      "learning_rate": 2.8816605791366855e-05,
      "loss": 0.0988,
      "step": 483
    },
    {
      "epoch": 0.34338417878680383,
      "grad_norm": 1.3377823727573568,
      "learning_rate": 2.880974830014643e-05,
      "loss": 0.0776,
      "step": 484
    },
    {
      "epoch": 0.3440936502305782,
      "grad_norm": 2.018482834568401,
      "learning_rate": 2.880287181826719e-05,
      "loss": 0.0855,
      "step": 485
    },
    {
      "epoch": 0.3448031216743526,
      "grad_norm": 1.916645708122162,
      "learning_rate": 2.879597635518541e-05,
      "loss": 0.1209,
      "step": 486
    },
    {
      "epoch": 0.345512593118127,
      "grad_norm": 2.1276787207181584,
      "learning_rate": 2.8789061920383475e-05,
      "loss": 0.0953,
      "step": 487
    },
    {
      "epoch": 0.3462220645619014,
      "grad_norm": 1.0368175822313435,
      "learning_rate": 2.878212852336987e-05,
      "loss": 0.1058,
      "step": 488
    },
    {
      "epoch": 0.3469315360056758,
      "grad_norm": 0.9901952589766785,
      "learning_rate": 2.8775176173679152e-05,
      "loss": 0.0836,
      "step": 489
    },
    {
      "epoch": 0.34764100744945015,
      "grad_norm": 0.8114828297928477,
      "learning_rate": 2.8768204880871925e-05,
      "loss": 0.1259,
      "step": 490
    },
    {
      "epoch": 0.34835047889322457,
      "grad_norm": 1.0284217348053095,
      "learning_rate": 2.8761214654534864e-05,
      "loss": 0.1081,
      "step": 491
    },
    {
      "epoch": 0.34905995033699894,
      "grad_norm": 1.848307395122552,
      "learning_rate": 2.875420550428067e-05,
      "loss": 0.1098,
      "step": 492
    },
    {
      "epoch": 0.3497694217807733,
      "grad_norm": 0.9573856105115914,
      "learning_rate": 2.8747177439748066e-05,
      "loss": 0.1065,
      "step": 493
    },
    {
      "epoch": 0.3504788932245477,
      "grad_norm": 0.5044734493627655,
      "learning_rate": 2.874013047060179e-05,
      "loss": 0.0841,
      "step": 494
    },
    {
      "epoch": 0.3511883646683221,
      "grad_norm": 0.7347140554247538,
      "learning_rate": 2.8733064606532574e-05,
      "loss": 0.0939,
      "step": 495
    },
    {
      "epoch": 0.35189783611209646,
      "grad_norm": 1.374773099981302,
      "learning_rate": 2.8725979857257138e-05,
      "loss": 0.0899,
      "step": 496
    },
    {
      "epoch": 0.3526073075558709,
      "grad_norm": 0.805310727253721,
      "learning_rate": 2.8718876232518165e-05,
      "loss": 0.0809,
      "step": 497
    },
    {
      "epoch": 0.35331677899964525,
      "grad_norm": 1.2252311956008417,
      "learning_rate": 2.8711753742084305e-05,
      "loss": 0.1218,
      "step": 498
    },
    {
      "epoch": 0.3540262504434197,
      "grad_norm": 0.5543857641099807,
      "learning_rate": 2.8704612395750133e-05,
      "loss": 0.1312,
      "step": 499
    },
    {
      "epoch": 0.35473572188719404,
      "grad_norm": 0.5165937866079755,
      "learning_rate": 2.869745220333618e-05,
      "loss": 0.0948,
      "step": 500
    },
    {
      "epoch": 0.3554451933309684,
      "grad_norm": 1.5884838029755333,
      "learning_rate": 2.8690273174688875e-05,
      "loss": 0.1275,
      "step": 501
    },
    {
      "epoch": 0.35615466477474284,
      "grad_norm": 1.1139505346254182,
      "learning_rate": 2.8683075319680553e-05,
      "loss": 0.1279,
      "step": 502
    },
    {
      "epoch": 0.3568641362185172,
      "grad_norm": 2.235958534086228,
      "learning_rate": 2.8675858648209442e-05,
      "loss": 0.1138,
      "step": 503
    },
    {
      "epoch": 0.35757360766229157,
      "grad_norm": 1.2953694992928817,
      "learning_rate": 2.8668623170199644e-05,
      "loss": 0.1286,
      "step": 504
    },
    {
      "epoch": 0.358283079106066,
      "grad_norm": 0.5678186538404929,
      "learning_rate": 2.8661368895601117e-05,
      "loss": 0.1394,
      "step": 505
    },
    {
      "epoch": 0.35899255054984036,
      "grad_norm": 2.71690689842914,
      "learning_rate": 2.865409583438968e-05,
      "loss": 0.1227,
      "step": 506
    },
    {
      "epoch": 0.3597020219936148,
      "grad_norm": 0.9839957303416551,
      "learning_rate": 2.8646803996566975e-05,
      "loss": 0.1258,
      "step": 507
    },
    {
      "epoch": 0.36041149343738915,
      "grad_norm": 0.6000833258292534,
      "learning_rate": 2.8639493392160473e-05,
      "loss": 0.0974,
      "step": 508
    },
    {
      "epoch": 0.3611209648811635,
      "grad_norm": 2.2142042594567224,
      "learning_rate": 2.863216403122345e-05,
      "loss": 0.1388,
      "step": 509
    },
    {
      "epoch": 0.36183043632493794,
      "grad_norm": 0.6554134256882517,
      "learning_rate": 2.8624815923834968e-05,
      "loss": 0.1015,
      "step": 510
    },
    {
      "epoch": 0.3625399077687123,
      "grad_norm": 0.9363173463439146,
      "learning_rate": 2.861744908009988e-05,
      "loss": 0.1249,
      "step": 511
    },
    {
      "epoch": 0.3632493792124867,
      "grad_norm": 0.5303245395389136,
      "learning_rate": 2.8610063510148796e-05,
      "loss": 0.1127,
      "step": 512
    },
    {
      "epoch": 0.3639588506562611,
      "grad_norm": 0.7952493442861119,
      "learning_rate": 2.8602659224138078e-05,
      "loss": 0.0966,
      "step": 513
    },
    {
      "epoch": 0.36466832210003547,
      "grad_norm": 0.7047199263011409,
      "learning_rate": 2.8595236232249836e-05,
      "loss": 0.0931,
      "step": 514
    },
    {
      "epoch": 0.36537779354380984,
      "grad_norm": 0.8888203524849596,
      "learning_rate": 2.858779454469189e-05,
      "loss": 0.1277,
      "step": 515
    },
    {
      "epoch": 0.36608726498758426,
      "grad_norm": 0.9081730555446176,
      "learning_rate": 2.858033417169778e-05,
      "loss": 0.1367,
      "step": 516
    },
    {
      "epoch": 0.3667967364313586,
      "grad_norm": 0.7404612859804995,
      "learning_rate": 2.8572855123526734e-05,
      "loss": 0.1271,
      "step": 517
    },
    {
      "epoch": 0.36750620787513305,
      "grad_norm": 2.100314961167394,
      "learning_rate": 2.8565357410463664e-05,
      "loss": 0.1094,
      "step": 518
    },
    {
      "epoch": 0.3682156793189074,
      "grad_norm": 1.1291899766290312,
      "learning_rate": 2.8557841042819147e-05,
      "loss": 0.1014,
      "step": 519
    },
    {
      "epoch": 0.3689251507626818,
      "grad_norm": 1.24535798498572,
      "learning_rate": 2.8550306030929425e-05,
      "loss": 0.1031,
      "step": 520
    },
    {
      "epoch": 0.3696346222064562,
      "grad_norm": 0.43112369779790843,
      "learning_rate": 2.854275238515636e-05,
      "loss": 0.0978,
      "step": 521
    },
    {
      "epoch": 0.3703440936502306,
      "grad_norm": 0.6300528409343545,
      "learning_rate": 2.8535180115887452e-05,
      "loss": 0.116,
      "step": 522
    },
    {
      "epoch": 0.37105356509400494,
      "grad_norm": 1.097212192645177,
      "learning_rate": 2.8527589233535814e-05,
      "loss": 0.1265,
      "step": 523
    },
    {
      "epoch": 0.37176303653777937,
      "grad_norm": 0.9501550373910891,
      "learning_rate": 2.851997974854014e-05,
      "loss": 0.1111,
      "step": 524
    },
    {
      "epoch": 0.37247250798155374,
      "grad_norm": 2.5055877732516585,
      "learning_rate": 2.8512351671364715e-05,
      "loss": 0.1262,
      "step": 525
    },
    {
      "epoch": 0.3731819794253281,
      "grad_norm": 1.4302636774870898,
      "learning_rate": 2.8504705012499393e-05,
      "loss": 0.1036,
      "step": 526
    },
    {
      "epoch": 0.3738914508691025,
      "grad_norm": 1.2276458748509624,
      "learning_rate": 2.8497039782459575e-05,
      "loss": 0.1162,
      "step": 527
    },
    {
      "epoch": 0.3746009223128769,
      "grad_norm": 0.7137109773755391,
      "learning_rate": 2.8489355991786215e-05,
      "loss": 0.0956,
      "step": 528
    },
    {
      "epoch": 0.3753103937566513,
      "grad_norm": 1.008888417467383,
      "learning_rate": 2.8481653651045766e-05,
      "loss": 0.0871,
      "step": 529
    },
    {
      "epoch": 0.3760198652004257,
      "grad_norm": 2.0246132601013116,
      "learning_rate": 2.847393277083021e-05,
      "loss": 0.13,
      "step": 530
    },
    {
      "epoch": 0.37672933664420005,
      "grad_norm": 2.1011438801827764,
      "learning_rate": 2.8466193361757024e-05,
      "loss": 0.1288,
      "step": 531
    },
    {
      "epoch": 0.3774388080879745,
      "grad_norm": 1.2671506346777859,
      "learning_rate": 2.8458435434469146e-05,
      "loss": 0.1102,
      "step": 532
    },
    {
      "epoch": 0.37814827953174884,
      "grad_norm": 0.6555487248726319,
      "learning_rate": 2.8450658999635006e-05,
      "loss": 0.1077,
      "step": 533
    },
    {
      "epoch": 0.3788577509755232,
      "grad_norm": 1.085286850768489,
      "learning_rate": 2.844286406794846e-05,
      "loss": 0.1106,
      "step": 534
    },
    {
      "epoch": 0.37956722241929763,
      "grad_norm": 0.9792362061459019,
      "learning_rate": 2.8435050650128824e-05,
      "loss": 0.1007,
      "step": 535
    },
    {
      "epoch": 0.380276693863072,
      "grad_norm": 1.2797079551443928,
      "learning_rate": 2.8427218756920812e-05,
      "loss": 0.0733,
      "step": 536
    },
    {
      "epoch": 0.3809861653068464,
      "grad_norm": 0.8833476768907377,
      "learning_rate": 2.8419368399094566e-05,
      "loss": 0.1084,
      "step": 537
    },
    {
      "epoch": 0.3816956367506208,
      "grad_norm": 1.077460291285719,
      "learning_rate": 2.8411499587445597e-05,
      "loss": 0.1078,
      "step": 538
    },
    {
      "epoch": 0.38240510819439516,
      "grad_norm": 0.5947485052041173,
      "learning_rate": 2.840361233279482e-05,
      "loss": 0.0906,
      "step": 539
    },
    {
      "epoch": 0.3831145796381696,
      "grad_norm": 0.5399505172653348,
      "learning_rate": 2.8395706645988488e-05,
      "loss": 0.09,
      "step": 540
    },
    {
      "epoch": 0.38382405108194395,
      "grad_norm": 0.8633163431177144,
      "learning_rate": 2.838778253789822e-05,
      "loss": 0.1071,
      "step": 541
    },
    {
      "epoch": 0.3845335225257183,
      "grad_norm": 1.6609420512087403,
      "learning_rate": 2.837984001942095e-05,
      "loss": 0.1114,
      "step": 542
    },
    {
      "epoch": 0.38524299396949274,
      "grad_norm": 0.7600386944333442,
      "learning_rate": 2.837187910147894e-05,
      "loss": 0.0989,
      "step": 543
    },
    {
      "epoch": 0.3859524654132671,
      "grad_norm": 1.440240990351525,
      "learning_rate": 2.8363899795019753e-05,
      "loss": 0.1065,
      "step": 544
    },
    {
      "epoch": 0.3866619368570415,
      "grad_norm": 0.8111575848469673,
      "learning_rate": 2.8355902111016244e-05,
      "loss": 0.104,
      "step": 545
    },
    {
      "epoch": 0.3873714083008159,
      "grad_norm": 0.5681118067015966,
      "learning_rate": 2.8347886060466523e-05,
      "loss": 0.1151,
      "step": 546
    },
    {
      "epoch": 0.38808087974459027,
      "grad_norm": 1.7155617220772919,
      "learning_rate": 2.8339851654393978e-05,
      "loss": 0.1089,
      "step": 547
    },
    {
      "epoch": 0.3887903511883647,
      "grad_norm": 1.7465767462695343,
      "learning_rate": 2.833179890384722e-05,
      "loss": 0.1214,
      "step": 548
    },
    {
      "epoch": 0.38949982263213906,
      "grad_norm": 0.5932015328688405,
      "learning_rate": 2.832372781990011e-05,
      "loss": 0.1173,
      "step": 549
    },
    {
      "epoch": 0.3902092940759134,
      "grad_norm": 1.4005153635349499,
      "learning_rate": 2.831563841365169e-05,
      "loss": 0.1077,
      "step": 550
    },
    {
      "epoch": 0.39091876551968785,
      "grad_norm": 1.3474483961218888,
      "learning_rate": 2.8307530696226224e-05,
      "loss": 0.0849,
      "step": 551
    },
    {
      "epoch": 0.3916282369634622,
      "grad_norm": 0.5599425920689892,
      "learning_rate": 2.8299404678773145e-05,
      "loss": 0.1079,
      "step": 552
    },
    {
      "epoch": 0.3923377084072366,
      "grad_norm": 0.8790334152733548,
      "learning_rate": 2.8291260372467064e-05,
      "loss": 0.1224,
      "step": 553
    },
    {
      "epoch": 0.393047179851011,
      "grad_norm": 0.941455289026995,
      "learning_rate": 2.8283097788507713e-05,
      "loss": 0.0981,
      "step": 554
    },
    {
      "epoch": 0.3937566512947854,
      "grad_norm": 0.8317601911521915,
      "learning_rate": 2.8274916938119987e-05,
      "loss": 0.0984,
      "step": 555
    },
    {
      "epoch": 0.3944661227385598,
      "grad_norm": 1.315951573310521,
      "learning_rate": 2.826671783255389e-05,
      "loss": 0.098,
      "step": 556
    },
    {
      "epoch": 0.39517559418233417,
      "grad_norm": 0.8093031174705544,
      "learning_rate": 2.8258500483084534e-05,
      "loss": 0.1232,
      "step": 557
    },
    {
      "epoch": 0.39588506562610853,
      "grad_norm": 2.111748407910676,
      "learning_rate": 2.825026490101212e-05,
      "loss": 0.1419,
      "step": 558
    },
    {
      "epoch": 0.39659453706988296,
      "grad_norm": 0.7546197406653664,
      "learning_rate": 2.8242011097661904e-05,
      "loss": 0.1186,
      "step": 559
    },
    {
      "epoch": 0.3973040085136573,
      "grad_norm": 1.2792560412735465,
      "learning_rate": 2.8233739084384228e-05,
      "loss": 0.1373,
      "step": 560
    },
    {
      "epoch": 0.3980134799574317,
      "grad_norm": 0.9387966041624755,
      "learning_rate": 2.8225448872554454e-05,
      "loss": 0.1317,
      "step": 561
    },
    {
      "epoch": 0.3987229514012061,
      "grad_norm": 1.324051537579158,
      "learning_rate": 2.8217140473572974e-05,
      "loss": 0.096,
      "step": 562
    },
    {
      "epoch": 0.3994324228449805,
      "grad_norm": 2.2184673432521613,
      "learning_rate": 2.8208813898865203e-05,
      "loss": 0.1355,
      "step": 563
    },
    {
      "epoch": 0.40014189428875485,
      "grad_norm": 2.2977328985321988,
      "learning_rate": 2.820046915988153e-05,
      "loss": 0.1245,
      "step": 564
    },
    {
      "epoch": 0.4008513657325293,
      "grad_norm": 0.8970425250647217,
      "learning_rate": 2.8192106268097336e-05,
      "loss": 0.116,
      "step": 565
    },
    {
      "epoch": 0.40156083717630364,
      "grad_norm": 2.7080964283232833,
      "learning_rate": 2.818372523501297e-05,
      "loss": 0.1156,
      "step": 566
    },
    {
      "epoch": 0.40227030862007807,
      "grad_norm": 4.297554105324148,
      "learning_rate": 2.8175326072153714e-05,
      "loss": 0.1405,
      "step": 567
    },
    {
      "epoch": 0.40297978006385243,
      "grad_norm": 3.9335382319654926,
      "learning_rate": 2.816690879106979e-05,
      "loss": 0.1165,
      "step": 568
    },
    {
      "epoch": 0.4036892515076268,
      "grad_norm": 4.734404491257155,
      "learning_rate": 2.8158473403336335e-05,
      "loss": 0.163,
      "step": 569
    },
    {
      "epoch": 0.4043987229514012,
      "grad_norm": 1.2641693545150974,
      "learning_rate": 2.8150019920553386e-05,
      "loss": 0.0983,
      "step": 570
    },
    {
      "epoch": 0.4051081943951756,
      "grad_norm": 0.7388020809360326,
      "learning_rate": 2.8141548354345855e-05,
      "loss": 0.1233,
      "step": 571
    },
    {
      "epoch": 0.40581766583894996,
      "grad_norm": 3.1179398092120216,
      "learning_rate": 2.813305871636354e-05,
      "loss": 0.1234,
      "step": 572
    },
    {
      "epoch": 0.4065271372827244,
      "grad_norm": 5.628199527445323,
      "learning_rate": 2.812455101828107e-05,
      "loss": 0.1377,
      "step": 573
    },
    {
      "epoch": 0.40723660872649875,
      "grad_norm": 3.3535594744498725,
      "learning_rate": 2.8116025271797923e-05,
      "loss": 0.1162,
      "step": 574
    },
    {
      "epoch": 0.4079460801702732,
      "grad_norm": 3.2499716790640676,
      "learning_rate": 2.8107481488638392e-05,
      "loss": 0.1022,
      "step": 575
    },
    {
      "epoch": 0.40865555161404754,
      "grad_norm": 2.823878852476875,
      "learning_rate": 2.8098919680551574e-05,
      "loss": 0.1196,
      "step": 576
    },
    {
      "epoch": 0.4093650230578219,
      "grad_norm": 1.4480570093450607,
      "learning_rate": 2.809033985931135e-05,
      "loss": 0.1131,
      "step": 577
    },
    {
      "epoch": 0.41007449450159633,
      "grad_norm": 0.8497223022259704,
      "learning_rate": 2.8081742036716385e-05,
      "loss": 0.1178,
      "step": 578
    },
    {
      "epoch": 0.4107839659453707,
      "grad_norm": 3.970096324971072,
      "learning_rate": 2.8073126224590073e-05,
      "loss": 0.1515,
      "step": 579
    },
    {
      "epoch": 0.41149343738914507,
      "grad_norm": 4.597521022934096,
      "learning_rate": 2.806449243478058e-05,
      "loss": 0.1467,
      "step": 580
    },
    {
      "epoch": 0.4122029088329195,
      "grad_norm": 5.45376769391393,
      "learning_rate": 2.805584067916076e-05,
      "loss": 0.1382,
      "step": 581
    },
    {
      "epoch": 0.41291238027669386,
      "grad_norm": 3.7527216143697766,
      "learning_rate": 2.8047170969628203e-05,
      "loss": 0.1224,
      "step": 582
    },
    {
      "epoch": 0.4136218517204682,
      "grad_norm": 3.110343863636251,
      "learning_rate": 2.8038483318105168e-05,
      "loss": 0.1084,
      "step": 583
    },
    {
      "epoch": 0.41433132316424265,
      "grad_norm": 0.8632185334036093,
      "learning_rate": 2.8029777736538598e-05,
      "loss": 0.0942,
      "step": 584
    },
    {
      "epoch": 0.415040794608017,
      "grad_norm": 1.632466927853409,
      "learning_rate": 2.802105423690009e-05,
      "loss": 0.0954,
      "step": 585
    },
    {
      "epoch": 0.41575026605179144,
      "grad_norm": 2.85259810688113,
      "learning_rate": 2.8012312831185875e-05,
      "loss": 0.1028,
      "step": 586
    },
    {
      "epoch": 0.4164597374955658,
      "grad_norm": 4.517740177335467,
      "learning_rate": 2.8003553531416818e-05,
      "loss": 0.14,
      "step": 587
    },
    {
      "epoch": 0.4171692089393402,
      "grad_norm": 2.983832501458289,
      "learning_rate": 2.799477634963838e-05,
      "loss": 0.0995,
      "step": 588
    },
    {
      "epoch": 0.4178786803831146,
      "grad_norm": 2.5750949400952554,
      "learning_rate": 2.798598129792063e-05,
      "loss": 0.1044,
      "step": 589
    },
    {
      "epoch": 0.41858815182688897,
      "grad_norm": 0.7340883922803195,
      "learning_rate": 2.797716838835819e-05,
      "loss": 0.0943,
      "step": 590
    },
    {
      "epoch": 0.41929762327066333,
      "grad_norm": 3.550188614082691,
      "learning_rate": 2.7968337633070252e-05,
      "loss": 0.1404,
      "step": 591
    },
    {
      "epoch": 0.42000709471443776,
      "grad_norm": 2.2835904132382807,
      "learning_rate": 2.7959489044200553e-05,
      "loss": 0.1078,
      "step": 592
    },
    {
      "epoch": 0.4207165661582121,
      "grad_norm": 3.253665348401309,
      "learning_rate": 2.795062263391734e-05,
      "loss": 0.1174,
      "step": 593
    },
    {
      "epoch": 0.42142603760198655,
      "grad_norm": 3.368908937873198,
      "learning_rate": 2.7941738414413374e-05,
      "loss": 0.1252,
      "step": 594
    },
    {
      "epoch": 0.4221355090457609,
      "grad_norm": 2.9843700331588243,
      "learning_rate": 2.793283639790592e-05,
      "loss": 0.1609,
      "step": 595
    },
    {
      "epoch": 0.4228449804895353,
      "grad_norm": 1.0247980106954122,
      "learning_rate": 2.7923916596636692e-05,
      "loss": 0.1041,
      "step": 596
    },
    {
      "epoch": 0.4235544519333097,
      "grad_norm": 1.883606813121272,
      "learning_rate": 2.7914979022871874e-05,
      "loss": 0.0839,
      "step": 597
    },
    {
      "epoch": 0.4242639233770841,
      "grad_norm": 3.492624813108691,
      "learning_rate": 2.790602368890209e-05,
      "loss": 0.1249,
      "step": 598
    },
    {
      "epoch": 0.42497339482085844,
      "grad_norm": 2.3319145331768736,
      "learning_rate": 2.789705060704239e-05,
      "loss": 0.0968,
      "step": 599
    },
    {
      "epoch": 0.42568286626463286,
      "grad_norm": 3.900032674331407,
      "learning_rate": 2.7888059789632228e-05,
      "loss": 0.1286,
      "step": 600
    },
    {
      "epoch": 0.42639233770840723,
      "grad_norm": 3.732480001308575,
      "learning_rate": 2.787905124903544e-05,
      "loss": 0.1133,
      "step": 601
    },
    {
      "epoch": 0.4271018091521816,
      "grad_norm": 1.182456635337205,
      "learning_rate": 2.7870024997640245e-05,
      "loss": 0.0954,
      "step": 602
    },
    {
      "epoch": 0.427811280595956,
      "grad_norm": 0.8826676596585253,
      "learning_rate": 2.7860981047859206e-05,
      "loss": 0.0982,
      "step": 603
    },
    {
      "epoch": 0.4285207520397304,
      "grad_norm": 1.3708148431705451,
      "learning_rate": 2.785191941212924e-05,
      "loss": 0.1016,
      "step": 604
    },
    {
      "epoch": 0.4292302234835048,
      "grad_norm": 1.9938187159664562,
      "learning_rate": 2.784284010291157e-05,
      "loss": 0.0952,
      "step": 605
    },
    {
      "epoch": 0.4299396949272792,
      "grad_norm": 2.549694179989554,
      "learning_rate": 2.7833743132691728e-05,
      "loss": 0.0986,
      "step": 606
    },
    {
      "epoch": 0.43064916637105355,
      "grad_norm": 1.273947499594232,
      "learning_rate": 2.7824628513979543e-05,
      "loss": 0.1046,
      "step": 607
    },
    {
      "epoch": 0.43135863781482797,
      "grad_norm": 0.6480794756125879,
      "learning_rate": 2.7815496259309088e-05,
      "loss": 0.0965,
      "step": 608
    },
    {
      "epoch": 0.43206810925860234,
      "grad_norm": 1.322629548390438,
      "learning_rate": 2.7806346381238716e-05,
      "loss": 0.097,
      "step": 609
    },
    {
      "epoch": 0.4327775807023767,
      "grad_norm": 1.8914047859341188,
      "learning_rate": 2.7797178892350997e-05,
      "loss": 0.1121,
      "step": 610
    },
    {
      "epoch": 0.43348705214615113,
      "grad_norm": 1.6872304604984099,
      "learning_rate": 2.778799380525273e-05,
      "loss": 0.1122,
      "step": 611
    },
    {
      "epoch": 0.4341965235899255,
      "grad_norm": 1.4023364845021251,
      "learning_rate": 2.7778791132574908e-05,
      "loss": 0.0937,
      "step": 612
    },
    {
      "epoch": 0.43490599503369987,
      "grad_norm": 1.3432790039581282,
      "learning_rate": 2.776957088697271e-05,
      "loss": 0.095,
      "step": 613
    },
    {
      "epoch": 0.4356154664774743,
      "grad_norm": 1.6450219360687317,
      "learning_rate": 2.776033308112547e-05,
      "loss": 0.1179,
      "step": 614
    },
    {
      "epoch": 0.43632493792124866,
      "grad_norm": 1.4925111555843509,
      "learning_rate": 2.775107772773669e-05,
      "loss": 0.121,
      "step": 615
    },
    {
      "epoch": 0.4370344093650231,
      "grad_norm": 0.6091261752379071,
      "learning_rate": 2.7741804839533995e-05,
      "loss": 0.1124,
      "step": 616
    },
    {
      "epoch": 0.43774388080879745,
      "grad_norm": 1.3348260825825455,
      "learning_rate": 2.773251442926911e-05,
      "loss": 0.1123,
      "step": 617
    },
    {
      "epoch": 0.4384533522525718,
      "grad_norm": 1.4050404684276534,
      "learning_rate": 2.772320650971787e-05,
      "loss": 0.1152,
      "step": 618
    },
    {
      "epoch": 0.43916282369634624,
      "grad_norm": 1.5497208983174737,
      "learning_rate": 2.7713881093680184e-05,
      "loss": 0.1038,
      "step": 619
    },
    {
      "epoch": 0.4398722951401206,
      "grad_norm": 2.9245270126728498,
      "learning_rate": 2.7704538193980024e-05,
      "loss": 0.1302,
      "step": 620
    },
    {
      "epoch": 0.440581766583895,
      "grad_norm": 1.2975316073853007,
      "learning_rate": 2.76951778234654e-05,
      "loss": 0.0888,
      "step": 621
    },
    {
      "epoch": 0.4412912380276694,
      "grad_norm": 1.1457202601732912,
      "learning_rate": 2.7685799995008355e-05,
      "loss": 0.0892,
      "step": 622
    },
    {
      "epoch": 0.44200070947144376,
      "grad_norm": 0.5443262686739241,
      "learning_rate": 2.7676404721504927e-05,
      "loss": 0.111,
      "step": 623
    },
    {
      "epoch": 0.4427101809152182,
      "grad_norm": 1.141138130325628,
      "learning_rate": 2.7666992015875152e-05,
      "loss": 0.1077,
      "step": 624
    },
    {
      "epoch": 0.44341965235899256,
      "grad_norm": 1.8625071274478626,
      "learning_rate": 2.7657561891063042e-05,
      "loss": 0.1246,
      "step": 625
    },
    {
      "epoch": 0.4441291238027669,
      "grad_norm": 0.3468592158620779,
      "learning_rate": 2.7648114360036557e-05,
      "loss": 0.0762,
      "step": 626
    },
    {
      "epoch": 0.44483859524654135,
      "grad_norm": 0.9503409611669715,
      "learning_rate": 2.7638649435787596e-05,
      "loss": 0.0858,
      "step": 627
    },
    {
      "epoch": 0.4455480666903157,
      "grad_norm": 2.743854441330861,
      "learning_rate": 2.7629167131331967e-05,
      "loss": 0.1308,
      "step": 628
    },
    {
      "epoch": 0.4462575381340901,
      "grad_norm": 0.6940548708286597,
      "learning_rate": 2.7619667459709394e-05,
      "loss": 0.1041,
      "step": 629
    },
    {
      "epoch": 0.4469670095778645,
      "grad_norm": 0.8457511365226246,
      "learning_rate": 2.7610150433983482e-05,
      "loss": 0.12,
      "step": 630
    },
    {
      "epoch": 0.4476764810216389,
      "grad_norm": 0.44706855167796616,
      "learning_rate": 2.7600616067241682e-05,
      "loss": 0.0776,
      "step": 631
    },
    {
      "epoch": 0.44838595246541324,
      "grad_norm": 2.21544742425388,
      "learning_rate": 2.7591064372595315e-05,
      "loss": 0.1178,
      "step": 632
    },
    {
      "epoch": 0.44909542390918766,
      "grad_norm": 0.8699432131758461,
      "learning_rate": 2.758149536317952e-05,
      "loss": 0.0758,
      "step": 633
    },
    {
      "epoch": 0.44980489535296203,
      "grad_norm": 1.2955684825232407,
      "learning_rate": 2.7571909052153242e-05,
      "loss": 0.1148,
      "step": 634
    },
    {
      "epoch": 0.45051436679673645,
      "grad_norm": 1.6574835544271427,
      "learning_rate": 2.756230545269923e-05,
      "loss": 0.0682,
      "step": 635
    },
    {
      "epoch": 0.4512238382405108,
      "grad_norm": 0.8770082079938105,
      "learning_rate": 2.7552684578024e-05,
      "loss": 0.1117,
      "step": 636
    },
    {
      "epoch": 0.4519333096842852,
      "grad_norm": 1.6069917385521764,
      "learning_rate": 2.7543046441357816e-05,
      "loss": 0.0762,
      "step": 637
    },
    {
      "epoch": 0.4526427811280596,
      "grad_norm": 0.7516135636403152,
      "learning_rate": 2.7533391055954703e-05,
      "loss": 0.0807,
      "step": 638
    },
    {
      "epoch": 0.453352252571834,
      "grad_norm": 0.6562967000937064,
      "learning_rate": 2.752371843509239e-05,
      "loss": 0.0951,
      "step": 639
    },
    {
      "epoch": 0.45406172401560835,
      "grad_norm": 0.656113015480493,
      "learning_rate": 2.7514028592072304e-05,
      "loss": 0.0989,
      "step": 640
    },
    {
      "epoch": 0.45477119545938277,
      "grad_norm": 1.0869205414112095,
      "learning_rate": 2.7504321540219566e-05,
      "loss": 0.0854,
      "step": 641
    },
    {
      "epoch": 0.45548066690315714,
      "grad_norm": 1.0508451584907033,
      "learning_rate": 2.7494597292882957e-05,
      "loss": 0.1365,
      "step": 642
    },
    {
      "epoch": 0.45619013834693156,
      "grad_norm": 0.6067761335725533,
      "learning_rate": 2.7484855863434902e-05,
      "loss": 0.0915,
      "step": 643
    },
    {
      "epoch": 0.45689960979070593,
      "grad_norm": 0.5291156473760684,
      "learning_rate": 2.7475097265271463e-05,
      "loss": 0.093,
      "step": 644
    },
    {
      "epoch": 0.4576090812344803,
      "grad_norm": 0.8052577911588888,
      "learning_rate": 2.74653215118123e-05,
      "loss": 0.0876,
      "step": 645
    },
    {
      "epoch": 0.4583185526782547,
      "grad_norm": 0.7666571788994913,
      "learning_rate": 2.745552861650067e-05,
      "loss": 0.0778,
      "step": 646
    },
    {
      "epoch": 0.4590280241220291,
      "grad_norm": 0.5675264353523128,
      "learning_rate": 2.7445718592803408e-05,
      "loss": 0.1117,
      "step": 647
    },
    {
      "epoch": 0.45973749556580346,
      "grad_norm": 0.7353816580307925,
      "learning_rate": 2.7435891454210892e-05,
      "loss": 0.0877,
      "step": 648
    },
    {
      "epoch": 0.4604469670095779,
      "grad_norm": 0.67985377504387,
      "learning_rate": 2.742604721423705e-05,
      "loss": 0.11,
      "step": 649
    },
    {
      "epoch": 0.46115643845335225,
      "grad_norm": 0.7259751686604848,
      "learning_rate": 2.7416185886419314e-05,
      "loss": 0.1028,
      "step": 650
    },
    {
      "epoch": 0.4618659098971266,
      "grad_norm": 0.9583862089614943,
      "learning_rate": 2.740630748431862e-05,
      "loss": 0.1294,
      "step": 651
    },
    {
      "epoch": 0.46257538134090104,
      "grad_norm": 0.94549034452099,
      "learning_rate": 2.7396412021519387e-05,
      "loss": 0.0952,
      "step": 652
    },
    {
      "epoch": 0.4632848527846754,
      "grad_norm": 2.3992201318316426,
      "learning_rate": 2.738649951162949e-05,
      "loss": 0.109,
      "step": 653
    },
    {
      "epoch": 0.46399432422844983,
      "grad_norm": 0.756627398185069,
      "learning_rate": 2.737656996828025e-05,
      "loss": 0.101,
      "step": 654
    },
    {
      "epoch": 0.4647037956722242,
      "grad_norm": 0.7821077123247125,
      "learning_rate": 2.7366623405126405e-05,
      "loss": 0.1106,
      "step": 655
    },
    {
      "epoch": 0.46541326711599856,
      "grad_norm": 0.8648200971159972,
      "learning_rate": 2.735665983584611e-05,
      "loss": 0.1236,
      "step": 656
    },
    {
      "epoch": 0.466122738559773,
      "grad_norm": 2.1583369007079405,
      "learning_rate": 2.7346679274140898e-05,
      "loss": 0.0959,
      "step": 657
    },
    {
      "epoch": 0.46683221000354735,
      "grad_norm": 1.1598479792129288,
      "learning_rate": 2.733668173373567e-05,
      "loss": 0.1039,
      "step": 658
    },
    {
      "epoch": 0.4675416814473217,
      "grad_norm": 0.6789386955180678,
      "learning_rate": 2.7326667228378677e-05,
      "loss": 0.1234,
      "step": 659
    },
    {
      "epoch": 0.46825115289109615,
      "grad_norm": 1.2921029008843232,
      "learning_rate": 2.73166357718415e-05,
      "loss": 0.0823,
      "step": 660
    },
    {
      "epoch": 0.4689606243348705,
      "grad_norm": 1.1886676966512852,
      "learning_rate": 2.7306587377919028e-05,
      "loss": 0.0951,
      "step": 661
    },
    {
      "epoch": 0.46967009577864494,
      "grad_norm": 0.6851883682983636,
      "learning_rate": 2.7296522060429447e-05,
      "loss": 0.1008,
      "step": 662
    },
    {
      "epoch": 0.4703795672224193,
      "grad_norm": 1.1284777951725222,
      "learning_rate": 2.72864398332142e-05,
      "loss": 0.0878,
      "step": 663
    },
    {
      "epoch": 0.47108903866619367,
      "grad_norm": 0.9075783941081457,
      "learning_rate": 2.7276340710138018e-05,
      "loss": 0.1004,
      "step": 664
    },
    {
      "epoch": 0.4717985101099681,
      "grad_norm": 1.6346176075968935,
      "learning_rate": 2.7266224705088818e-05,
      "loss": 0.1208,
      "step": 665
    },
    {
      "epoch": 0.47250798155374246,
      "grad_norm": 0.8893006366287542,
      "learning_rate": 2.7256091831977777e-05,
      "loss": 0.108,
      "step": 666
    },
    {
      "epoch": 0.47321745299751683,
      "grad_norm": 0.6066635635070098,
      "learning_rate": 2.724594210473925e-05,
      "loss": 0.1193,
      "step": 667
    },
    {
      "epoch": 0.47392692444129125,
      "grad_norm": 2.2617699651468803,
      "learning_rate": 2.7235775537330762e-05,
      "loss": 0.0949,
      "step": 668
    },
    {
      "epoch": 0.4746363958850656,
      "grad_norm": 0.9911915355086106,
      "learning_rate": 2.7225592143733e-05,
      "loss": 0.107,
      "step": 669
    },
    {
      "epoch": 0.47534586732884,
      "grad_norm": 0.8654961257820052,
      "learning_rate": 2.7215391937949804e-05,
      "loss": 0.0871,
      "step": 670
    },
    {
      "epoch": 0.4760553387726144,
      "grad_norm": 1.5292807969294366,
      "learning_rate": 2.7205174934008114e-05,
      "loss": 0.0963,
      "step": 671
    },
    {
      "epoch": 0.4767648102163888,
      "grad_norm": 1.8107797685487508,
      "learning_rate": 2.7194941145957978e-05,
      "loss": 0.0896,
      "step": 672
    },
    {
      "epoch": 0.4774742816601632,
      "grad_norm": 1.347356915687135,
      "learning_rate": 2.718469058787253e-05,
      "loss": 0.0909,
      "step": 673
    },
    {
      "epoch": 0.47818375310393757,
      "grad_norm": 0.49881022334598185,
      "learning_rate": 2.7174423273847966e-05,
      "loss": 0.0837,
      "step": 674
    },
    {
      "epoch": 0.47889322454771194,
      "grad_norm": 0.5119640903633077,
      "learning_rate": 2.7164139218003512e-05,
      "loss": 0.0829,
      "step": 675
    },
    {
      "epoch": 0.47960269599148636,
      "grad_norm": 0.9152905776537597,
      "learning_rate": 2.7153838434481425e-05,
      "loss": 0.1013,
      "step": 676
    },
    {
      "epoch": 0.48031216743526073,
      "grad_norm": 0.7024837026839604,
      "learning_rate": 2.7143520937446962e-05,
      "loss": 0.0985,
      "step": 677
    },
    {
      "epoch": 0.4810216388790351,
      "grad_norm": 0.7029537730285206,
      "learning_rate": 2.7133186741088368e-05,
      "loss": 0.0948,
      "step": 678
    },
    {
      "epoch": 0.4817311103228095,
      "grad_norm": 1.8432972902959943,
      "learning_rate": 2.7122835859616858e-05,
      "loss": 0.1077,
      "step": 679
    },
    {
      "epoch": 0.4824405817665839,
      "grad_norm": 2.078876528673767,
      "learning_rate": 2.7112468307266573e-05,
      "loss": 0.0837,
      "step": 680
    },
    {
      "epoch": 0.4831500532103583,
      "grad_norm": 1.8178603886160074,
      "learning_rate": 2.7102084098294594e-05,
      "loss": 0.1022,
      "step": 681
    },
    {
      "epoch": 0.4838595246541327,
      "grad_norm": 1.2206480779701385,
      "learning_rate": 2.7091683246980906e-05,
      "loss": 0.0851,
      "step": 682
    },
    {
      "epoch": 0.48456899609790705,
      "grad_norm": 2.373198620025064,
      "learning_rate": 2.7081265767628374e-05,
      "loss": 0.1096,
      "step": 683
    },
    {
      "epoch": 0.48527846754168147,
      "grad_norm": 1.476341729093348,
      "learning_rate": 2.707083167456274e-05,
      "loss": 0.0937,
      "step": 684
    },
    {
      "epoch": 0.48598793898545584,
      "grad_norm": 2.0123887574107115,
      "learning_rate": 2.706038098213258e-05,
      "loss": 0.0926,
      "step": 685
    },
    {
      "epoch": 0.4866974104292302,
      "grad_norm": 1.9712652442438652,
      "learning_rate": 2.70499137047093e-05,
      "loss": 0.1288,
      "step": 686
    },
    {
      "epoch": 0.4874068818730046,
      "grad_norm": 1.3035829499584661,
      "learning_rate": 2.7039429856687127e-05,
      "loss": 0.1084,
      "step": 687
    },
    {
      "epoch": 0.488116353316779,
      "grad_norm": 0.8915820939259244,
      "learning_rate": 2.7028929452483046e-05,
      "loss": 0.085,
      "step": 688
    },
    {
      "epoch": 0.48882582476055336,
      "grad_norm": 0.8003348125003388,
      "learning_rate": 2.7018412506536844e-05,
      "loss": 0.0896,
      "step": 689
    },
    {
      "epoch": 0.4895352962043278,
      "grad_norm": 3.1378260825988677,
      "learning_rate": 2.7007879033311033e-05,
      "loss": 0.1144,
      "step": 690
    },
    {
      "epoch": 0.49024476764810215,
      "grad_norm": 2.2627703714920706,
      "learning_rate": 2.6997329047290858e-05,
      "loss": 0.1454,
      "step": 691
    },
    {
      "epoch": 0.4909542390918766,
      "grad_norm": 2.0697201901878897,
      "learning_rate": 2.698676256298427e-05,
      "loss": 0.13,
      "step": 692
    },
    {
      "epoch": 0.49166371053565094,
      "grad_norm": 1.006225479132585,
      "learning_rate": 2.6976179594921915e-05,
      "loss": 0.0926,
      "step": 693
    },
    {
      "epoch": 0.4923731819794253,
      "grad_norm": 0.7339127732550675,
      "learning_rate": 2.69655801576571e-05,
      "loss": 0.0884,
      "step": 694
    },
    {
      "epoch": 0.49308265342319973,
      "grad_norm": 2.353956710221174,
      "learning_rate": 2.6954964265765783e-05,
      "loss": 0.0903,
      "step": 695
    },
    {
      "epoch": 0.4937921248669741,
      "grad_norm": 3.3233971275306624,
      "learning_rate": 2.6944331933846544e-05,
      "loss": 0.1096,
      "step": 696
    },
    {
      "epoch": 0.49450159631074847,
      "grad_norm": 3.404905547734324,
      "learning_rate": 2.693368317652058e-05,
      "loss": 0.1156,
      "step": 697
    },
    {
      "epoch": 0.4952110677545229,
      "grad_norm": 2.4248121315905093,
      "learning_rate": 2.6923018008431673e-05,
      "loss": 0.0739,
      "step": 698
    },
    {
      "epoch": 0.49592053919829726,
      "grad_norm": 1.7005592891486174,
      "learning_rate": 2.6912336444246163e-05,
      "loss": 0.1147,
      "step": 699
    },
    {
      "epoch": 0.4966300106420717,
      "grad_norm": 1.0154861830206332,
      "learning_rate": 2.690163849865296e-05,
      "loss": 0.0706,
      "step": 700
    },
    {
      "epoch": 0.49733948208584605,
      "grad_norm": 2.098368571271643,
      "learning_rate": 2.6890924186363463e-05,
      "loss": 0.0871,
      "step": 701
    },
    {
      "epoch": 0.4980489535296204,
      "grad_norm": 1.5229138474279507,
      "learning_rate": 2.6880193522111622e-05,
      "loss": 0.1033,
      "step": 702
    },
    {
      "epoch": 0.49875842497339484,
      "grad_norm": 2.530414208783178,
      "learning_rate": 2.6869446520653842e-05,
      "loss": 0.1041,
      "step": 703
    },
    {
      "epoch": 0.4994678964171692,
      "grad_norm": 1.489347416927812,
      "learning_rate": 2.6858683196769004e-05,
      "loss": 0.0967,
      "step": 704
    },
    {
      "epoch": 0.5001773678609436,
      "grad_norm": 1.04396886001101,
      "learning_rate": 2.684790356525844e-05,
      "loss": 0.118,
      "step": 705
    },
    {
      "epoch": 0.500886839304718,
      "grad_norm": 2.593904907774733,
      "learning_rate": 2.6837107640945904e-05,
      "loss": 0.0879,
      "step": 706
    },
    {
      "epoch": 0.5015963107484924,
      "grad_norm": 1.866562579139544,
      "learning_rate": 2.6826295438677554e-05,
      "loss": 0.1339,
      "step": 707
    },
    {
      "epoch": 0.5023057821922667,
      "grad_norm": 1.5369853576336128,
      "learning_rate": 2.681546697332193e-05,
      "loss": 0.098,
      "step": 708
    },
    {
      "epoch": 0.5030152536360412,
      "grad_norm": 1.4868926078314282,
      "learning_rate": 2.6804622259769942e-05,
      "loss": 0.1104,
      "step": 709
    },
    {
      "epoch": 0.5037247250798156,
      "grad_norm": 0.8181632030731728,
      "learning_rate": 2.6793761312934843e-05,
      "loss": 0.0992,
      "step": 710
    },
    {
      "epoch": 0.5044341965235899,
      "grad_norm": 0.43046680958879047,
      "learning_rate": 2.6782884147752206e-05,
      "loss": 0.0998,
      "step": 711
    },
    {
      "epoch": 0.5051436679673643,
      "grad_norm": 0.9321564299679312,
      "learning_rate": 2.6771990779179916e-05,
      "loss": 0.0927,
      "step": 712
    },
    {
      "epoch": 0.5058531394111387,
      "grad_norm": 0.8541355314610722,
      "learning_rate": 2.6761081222198126e-05,
      "loss": 0.0797,
      "step": 713
    },
    {
      "epoch": 0.506562610854913,
      "grad_norm": 1.0433472198749454,
      "learning_rate": 2.675015549180926e-05,
      "loss": 0.072,
      "step": 714
    },
    {
      "epoch": 0.5072720822986875,
      "grad_norm": 0.5203971876763894,
      "learning_rate": 2.673921360303798e-05,
      "loss": 0.0695,
      "step": 715
    },
    {
      "epoch": 0.5079815537424619,
      "grad_norm": 1.6724417277230796,
      "learning_rate": 2.6728255570931174e-05,
      "loss": 0.0919,
      "step": 716
    },
    {
      "epoch": 0.5086910251862362,
      "grad_norm": 0.9265002936651544,
      "learning_rate": 2.6717281410557924e-05,
      "loss": 0.0806,
      "step": 717
    },
    {
      "epoch": 0.5094004966300106,
      "grad_norm": 1.123291846804393,
      "learning_rate": 2.6706291137009488e-05,
      "loss": 0.1025,
      "step": 718
    },
    {
      "epoch": 0.5101099680737851,
      "grad_norm": 1.7832198327424,
      "learning_rate": 2.669528476539929e-05,
      "loss": 0.101,
      "step": 719
    },
    {
      "epoch": 0.5108194395175594,
      "grad_norm": 0.9549248781537912,
      "learning_rate": 2.668426231086289e-05,
      "loss": 0.108,
      "step": 720
    },
    {
      "epoch": 0.5115289109613338,
      "grad_norm": 0.8002956332109035,
      "learning_rate": 2.667322378855796e-05,
      "loss": 0.0744,
      "step": 721
    },
    {
      "epoch": 0.5122383824051082,
      "grad_norm": 1.411775937448295,
      "learning_rate": 2.6662169213664273e-05,
      "loss": 0.1029,
      "step": 722
    },
    {
      "epoch": 0.5129478538488825,
      "grad_norm": 3.0358142810774416,
      "learning_rate": 2.665109860138367e-05,
      "loss": 0.0904,
      "step": 723
    },
    {
      "epoch": 0.513657325292657,
      "grad_norm": 1.4618254132636626,
      "learning_rate": 2.6640011966940055e-05,
      "loss": 0.1209,
      "step": 724
    },
    {
      "epoch": 0.5143667967364314,
      "grad_norm": 1.9699019369734254,
      "learning_rate": 2.6628909325579363e-05,
      "loss": 0.1032,
      "step": 725
    },
    {
      "epoch": 0.5150762681802058,
      "grad_norm": 1.1561104662443713,
      "learning_rate": 2.6617790692569534e-05,
      "loss": 0.126,
      "step": 726
    },
    {
      "epoch": 0.5157857396239801,
      "grad_norm": 1.226127283924977,
      "learning_rate": 2.6606656083200503e-05,
      "loss": 0.0709,
      "step": 727
    },
    {
      "epoch": 0.5164952110677545,
      "grad_norm": 1.0454229251540932,
      "learning_rate": 2.6595505512784175e-05,
      "loss": 0.0877,
      "step": 728
    },
    {
      "epoch": 0.517204682511529,
      "grad_norm": 0.7152619511168274,
      "learning_rate": 2.6584338996654412e-05,
      "loss": 0.0914,
      "step": 729
    },
    {
      "epoch": 0.5179141539553033,
      "grad_norm": 0.7678687910268428,
      "learning_rate": 2.657315655016699e-05,
      "loss": 0.1048,
      "step": 730
    },
    {
      "epoch": 0.5186236253990777,
      "grad_norm": 0.5302101019819314,
      "learning_rate": 2.6561958188699604e-05,
      "loss": 0.0959,
      "step": 731
    },
    {
      "epoch": 0.5193330968428521,
      "grad_norm": 0.6005287656785708,
      "learning_rate": 2.6550743927651826e-05,
      "loss": 0.0745,
      "step": 732
    },
    {
      "epoch": 0.5200425682866264,
      "grad_norm": 1.0619878502071893,
      "learning_rate": 2.6539513782445097e-05,
      "loss": 0.0771,
      "step": 733
    },
    {
      "epoch": 0.5207520397304009,
      "grad_norm": 0.6249804070305147,
      "learning_rate": 2.6528267768522698e-05,
      "loss": 0.1064,
      "step": 734
    },
    {
      "epoch": 0.5214615111741753,
      "grad_norm": 1.4627916616815637,
      "learning_rate": 2.651700590134974e-05,
      "loss": 0.0856,
      "step": 735
    },
    {
      "epoch": 0.5221709826179496,
      "grad_norm": 0.814142555962611,
      "learning_rate": 2.650572819641313e-05,
      "loss": 0.1258,
      "step": 736
    },
    {
      "epoch": 0.522880454061724,
      "grad_norm": 0.9653069902169457,
      "learning_rate": 2.6494434669221543e-05,
      "loss": 0.104,
      "step": 737
    },
    {
      "epoch": 0.5235899255054984,
      "grad_norm": 0.7096130309147914,
      "learning_rate": 2.6483125335305435e-05,
      "loss": 0.0879,
      "step": 738
    },
    {
      "epoch": 0.5242993969492727,
      "grad_norm": 1.0493588114817471,
      "learning_rate": 2.647180021021698e-05,
      "loss": 0.0855,
      "step": 739
    },
    {
      "epoch": 0.5250088683930472,
      "grad_norm": 1.605499799477196,
      "learning_rate": 2.6460459309530077e-05,
      "loss": 0.0797,
      "step": 740
    },
    {
      "epoch": 0.5257183398368216,
      "grad_norm": 1.4475212015020267,
      "learning_rate": 2.6449102648840312e-05,
      "loss": 0.1189,
      "step": 741
    },
    {
      "epoch": 0.5264278112805959,
      "grad_norm": 2.6659082823578704,
      "learning_rate": 2.6437730243764957e-05,
      "loss": 0.1087,
      "step": 742
    },
    {
      "epoch": 0.5271372827243703,
      "grad_norm": 2.5652982985052617,
      "learning_rate": 2.6426342109942914e-05,
      "loss": 0.0966,
      "step": 743
    },
    {
      "epoch": 0.5278467541681447,
      "grad_norm": 2.5228507629099397,
      "learning_rate": 2.6414938263034736e-05,
      "loss": 0.1214,
      "step": 744
    },
    {
      "epoch": 0.5285562256119192,
      "grad_norm": 2.1205431984293095,
      "learning_rate": 2.6403518718722568e-05,
      "loss": 0.0888,
      "step": 745
    },
    {
      "epoch": 0.5292656970556935,
      "grad_norm": 2.2997587183258075,
      "learning_rate": 2.639208349271015e-05,
      "loss": 0.1089,
      "step": 746
    },
    {
      "epoch": 0.5299751684994679,
      "grad_norm": 3.4591558890395544,
      "learning_rate": 2.638063260072279e-05,
      "loss": 0.1135,
      "step": 747
    },
    {
      "epoch": 0.5306846399432423,
      "grad_norm": 2.5281951408322207,
      "learning_rate": 2.6369166058507332e-05,
      "loss": 0.0764,
      "step": 748
    },
    {
      "epoch": 0.5313941113870166,
      "grad_norm": 1.208412137677219,
      "learning_rate": 2.635768388183214e-05,
      "loss": 0.0942,
      "step": 749
    },
    {
      "epoch": 0.5321035828307911,
      "grad_norm": 2.9435897965503712,
      "learning_rate": 2.6346186086487082e-05,
      "loss": 0.1212,
      "step": 750
    },
    {
      "epoch": 0.5328130542745655,
      "grad_norm": 0.8127274545273222,
      "learning_rate": 2.6334672688283507e-05,
      "loss": 0.1066,
      "step": 751
    },
    {
      "epoch": 0.5335225257183398,
      "grad_norm": 0.8038233278889797,
      "learning_rate": 2.6323143703054215e-05,
      "loss": 0.0862,
      "step": 752
    },
    {
      "epoch": 0.5342319971621142,
      "grad_norm": 2.125129565609008,
      "learning_rate": 2.6311599146653446e-05,
      "loss": 0.0842,
      "step": 753
    },
    {
      "epoch": 0.5349414686058886,
      "grad_norm": 1.617967494045547,
      "learning_rate": 2.630003903495685e-05,
      "loss": 0.1042,
      "step": 754
    },
    {
      "epoch": 0.535650940049663,
      "grad_norm": 3.08363377551552,
      "learning_rate": 2.628846338386146e-05,
      "loss": 0.1326,
      "step": 755
    },
    {
      "epoch": 0.5363604114934374,
      "grad_norm": 1.1215425613584107,
      "learning_rate": 2.62768722092857e-05,
      "loss": 0.0926,
      "step": 756
    },
    {
      "epoch": 0.5370698829372118,
      "grad_norm": 1.607383340627383,
      "learning_rate": 2.6265265527169317e-05,
      "loss": 0.1229,
      "step": 757
    },
    {
      "epoch": 0.5377793543809861,
      "grad_norm": 0.9534396676675384,
      "learning_rate": 2.62536433534734e-05,
      "loss": 0.0814,
      "step": 758
    },
    {
      "epoch": 0.5384888258247605,
      "grad_norm": 1.120557646141017,
      "learning_rate": 2.6242005704180337e-05,
      "loss": 0.1125,
      "step": 759
    },
    {
      "epoch": 0.539198297268535,
      "grad_norm": 2.962251586928889,
      "learning_rate": 2.6230352595293784e-05,
      "loss": 0.1377,
      "step": 760
    },
    {
      "epoch": 0.5399077687123093,
      "grad_norm": 2.5098803048761957,
      "learning_rate": 2.6218684042838684e-05,
      "loss": 0.1168,
      "step": 761
    },
    {
      "epoch": 0.5406172401560837,
      "grad_norm": 0.795818025200409,
      "learning_rate": 2.6207000062861197e-05,
      "loss": 0.0875,
      "step": 762
    },
    {
      "epoch": 0.5413267115998581,
      "grad_norm": 0.7532520448233155,
      "learning_rate": 2.6195300671428695e-05,
      "loss": 0.0908,
      "step": 763
    },
    {
      "epoch": 0.5420361830436325,
      "grad_norm": 0.583994788802027,
      "learning_rate": 2.618358588462976e-05,
      "loss": 0.0876,
      "step": 764
    },
    {
      "epoch": 0.5427456544874069,
      "grad_norm": 1.3750277611547006,
      "learning_rate": 2.6171855718574132e-05,
      "loss": 0.0913,
      "step": 765
    },
    {
      "epoch": 0.5434551259311813,
      "grad_norm": 2.246543050993341,
      "learning_rate": 2.616011018939271e-05,
      "loss": 0.0932,
      "step": 766
    },
    {
      "epoch": 0.5441645973749557,
      "grad_norm": 1.0520067729285154,
      "learning_rate": 2.6148349313237506e-05,
      "loss": 0.1041,
      "step": 767
    },
    {
      "epoch": 0.54487406881873,
      "grad_norm": 0.5749659299419089,
      "learning_rate": 2.6136573106281658e-05,
      "loss": 0.0756,
      "step": 768
    },
    {
      "epoch": 0.5455835402625044,
      "grad_norm": 0.460291107139318,
      "learning_rate": 2.6124781584719365e-05,
      "loss": 0.0925,
      "step": 769
    },
    {
      "epoch": 0.5462930117062789,
      "grad_norm": 1.4084500922827743,
      "learning_rate": 2.6112974764765893e-05,
      "loss": 0.121,
      "step": 770
    },
    {
      "epoch": 0.5470024831500532,
      "grad_norm": 0.4872463366732842,
      "learning_rate": 2.6101152662657553e-05,
      "loss": 0.0742,
      "step": 771
    },
    {
      "epoch": 0.5477119545938276,
      "grad_norm": 1.7127983342142834,
      "learning_rate": 2.6089315294651658e-05,
      "loss": 0.0774,
      "step": 772
    },
    {
      "epoch": 0.548421426037602,
      "grad_norm": 1.2275087447801774,
      "learning_rate": 2.6077462677026528e-05,
      "loss": 0.1115,
      "step": 773
    },
    {
      "epoch": 0.5491308974813763,
      "grad_norm": 0.783829340606162,
      "learning_rate": 2.606559482608145e-05,
      "loss": 0.0999,
      "step": 774
    },
    {
      "epoch": 0.5498403689251508,
      "grad_norm": 0.7591657809330782,
      "learning_rate": 2.6053711758136647e-05,
      "loss": 0.0945,
      "step": 775
    },
    {
      "epoch": 0.5505498403689252,
      "grad_norm": 1.9996854299067182,
      "learning_rate": 2.6041813489533287e-05,
      "loss": 0.1193,
      "step": 776
    },
    {
      "epoch": 0.5512593118126995,
      "grad_norm": 1.3441520369060471,
      "learning_rate": 2.6029900036633426e-05,
      "loss": 0.0705,
      "step": 777
    },
    {
      "epoch": 0.5519687832564739,
      "grad_norm": 0.9731978063293658,
      "learning_rate": 2.6017971415820015e-05,
      "loss": 0.1088,
      "step": 778
    },
    {
      "epoch": 0.5526782547002483,
      "grad_norm": 0.47074387264321105,
      "learning_rate": 2.6006027643496844e-05,
      "loss": 0.1115,
      "step": 779
    },
    {
      "epoch": 0.5533877261440227,
      "grad_norm": 0.4459623208027827,
      "learning_rate": 2.5994068736088563e-05,
      "loss": 0.0652,
      "step": 780
    },
    {
      "epoch": 0.5540971975877971,
      "grad_norm": 0.7766313994127577,
      "learning_rate": 2.5982094710040615e-05,
      "loss": 0.1114,
      "step": 781
    },
    {
      "epoch": 0.5548066690315715,
      "grad_norm": 0.7215001802725537,
      "learning_rate": 2.5970105581819235e-05,
      "loss": 0.0995,
      "step": 782
    },
    {
      "epoch": 0.5555161404753459,
      "grad_norm": 1.6858093716183007,
      "learning_rate": 2.5958101367911445e-05,
      "loss": 0.1151,
      "step": 783
    },
    {
      "epoch": 0.5562256119191202,
      "grad_norm": 0.6330021713809463,
      "learning_rate": 2.5946082084824993e-05,
      "loss": 0.092,
      "step": 784
    },
    {
      "epoch": 0.5569350833628947,
      "grad_norm": 0.4366898773404169,
      "learning_rate": 2.5934047749088354e-05,
      "loss": 0.0993,
      "step": 785
    },
    {
      "epoch": 0.5576445548066691,
      "grad_norm": 0.6386817625242636,
      "learning_rate": 2.5921998377250702e-05,
      "loss": 0.1023,
      "step": 786
    },
    {
      "epoch": 0.5583540262504434,
      "grad_norm": 1.4308772296433327,
      "learning_rate": 2.59099339858819e-05,
      "loss": 0.0915,
      "step": 787
    },
    {
      "epoch": 0.5590634976942178,
      "grad_norm": 0.8785230584609734,
      "learning_rate": 2.5897854591572448e-05,
      "loss": 0.0971,
      "step": 788
    },
    {
      "epoch": 0.5597729691379922,
      "grad_norm": 0.5417002656527303,
      "learning_rate": 2.588576021093349e-05,
      "loss": 0.0741,
      "step": 789
    },
    {
      "epoch": 0.5604824405817666,
      "grad_norm": 0.9114214709812948,
      "learning_rate": 2.5873650860596773e-05,
      "loss": 0.0965,
      "step": 790
    },
    {
      "epoch": 0.561191912025541,
      "grad_norm": 1.1694912799997828,
      "learning_rate": 2.5861526557214635e-05,
      "loss": 0.0918,
      "step": 791
    },
    {
      "epoch": 0.5619013834693154,
      "grad_norm": 1.6671866970608116,
      "learning_rate": 2.5849387317459964e-05,
      "loss": 0.1061,
      "step": 792
    },
    {
      "epoch": 0.5626108549130897,
      "grad_norm": 1.108277047799898,
      "learning_rate": 2.5837233158026198e-05,
      "loss": 0.1001,
      "step": 793
    },
    {
      "epoch": 0.5633203263568641,
      "grad_norm": 1.603574267091752,
      "learning_rate": 2.58250640956273e-05,
      "loss": 0.0577,
      "step": 794
    },
    {
      "epoch": 0.5640297978006386,
      "grad_norm": 0.6668103256442313,
      "learning_rate": 2.5812880146997713e-05,
      "loss": 0.0813,
      "step": 795
    },
    {
      "epoch": 0.5647392692444129,
      "grad_norm": 0.710584709802994,
      "learning_rate": 2.5800681328892356e-05,
      "loss": 0.0696,
      "step": 796
    },
    {
      "epoch": 0.5654487406881873,
      "grad_norm": 1.7588513930534522,
      "learning_rate": 2.5788467658086587e-05,
      "loss": 0.1204,
      "step": 797
    },
    {
      "epoch": 0.5661582121319617,
      "grad_norm": 0.855372758724027,
      "learning_rate": 2.5776239151376208e-05,
      "loss": 0.1056,
      "step": 798
    },
    {
      "epoch": 0.566867683575736,
      "grad_norm": 1.8712457235962936,
      "learning_rate": 2.5763995825577412e-05,
      "loss": 0.1199,
      "step": 799
    },
    {
      "epoch": 0.5675771550195104,
      "grad_norm": 2.072265552146076,
      "learning_rate": 2.575173769752677e-05,
      "loss": 0.084,
      "step": 800
    },
    {
      "epoch": 0.5682866264632849,
      "grad_norm": 0.7135020706378001,
      "learning_rate": 2.5739464784081206e-05,
      "loss": 0.0856,
      "step": 801
    },
    {
      "epoch": 0.5689960979070593,
      "grad_norm": 0.7898046734727573,
      "learning_rate": 2.5727177102117984e-05,
      "loss": 0.0756,
      "step": 802
    },
    {
      "epoch": 0.5697055693508336,
      "grad_norm": 0.8115510714828558,
      "learning_rate": 2.5714874668534674e-05,
      "loss": 0.0669,
      "step": 803
    },
    {
      "epoch": 0.570415040794608,
      "grad_norm": 0.5626285798981052,
      "learning_rate": 2.570255750024912e-05,
      "loss": 0.0781,
      "step": 804
    },
    {
      "epoch": 0.5711245122383825,
      "grad_norm": 0.6806073847222428,
      "learning_rate": 2.5690225614199455e-05,
      "loss": 0.0933,
      "step": 805
    },
    {
      "epoch": 0.5718339836821568,
      "grad_norm": 0.5217530079880197,
      "learning_rate": 2.5677879027344032e-05,
      "loss": 0.0791,
      "step": 806
    },
    {
      "epoch": 0.5725434551259312,
      "grad_norm": 0.5116552244445357,
      "learning_rate": 2.5665517756661413e-05,
      "loss": 0.0776,
      "step": 807
    },
    {
      "epoch": 0.5732529265697056,
      "grad_norm": 0.6350678456881726,
      "learning_rate": 2.5653141819150377e-05,
      "loss": 0.082,
      "step": 808
    },
    {
      "epoch": 0.5739623980134799,
      "grad_norm": 1.3910880529044372,
      "learning_rate": 2.564075123182985e-05,
      "loss": 0.0959,
      "step": 809
    },
    {
      "epoch": 0.5746718694572543,
      "grad_norm": 2.5936570580785974,
      "learning_rate": 2.5628346011738917e-05,
      "loss": 0.1284,
      "step": 810
    },
    {
      "epoch": 0.5753813409010288,
      "grad_norm": 1.7079327940679043,
      "learning_rate": 2.561592617593678e-05,
      "loss": 0.1455,
      "step": 811
    },
    {
      "epoch": 0.5760908123448031,
      "grad_norm": 0.8076755669055823,
      "learning_rate": 2.5603491741502744e-05,
      "loss": 0.0883,
      "step": 812
    },
    {
      "epoch": 0.5768002837885775,
      "grad_norm": 0.7843673465546542,
      "learning_rate": 2.5591042725536182e-05,
      "loss": 0.0965,
      "step": 813
    },
    {
      "epoch": 0.5775097552323519,
      "grad_norm": 1.096149422678392,
      "learning_rate": 2.5578579145156526e-05,
      "loss": 0.076,
      "step": 814
    },
    {
      "epoch": 0.5782192266761262,
      "grad_norm": 2.3191828243442822,
      "learning_rate": 2.5566101017503236e-05,
      "loss": 0.0737,
      "step": 815
    },
    {
      "epoch": 0.5789286981199007,
      "grad_norm": 1.9482811673768063,
      "learning_rate": 2.555360835973577e-05,
      "loss": 0.0809,
      "step": 816
    },
    {
      "epoch": 0.5796381695636751,
      "grad_norm": 0.8178379334584898,
      "learning_rate": 2.5541101189033577e-05,
      "loss": 0.06,
      "step": 817
    },
    {
      "epoch": 0.5803476410074494,
      "grad_norm": 1.907998112121273,
      "learning_rate": 2.552857952259606e-05,
      "loss": 0.0936,
      "step": 818
    },
    {
      "epoch": 0.5810571124512238,
      "grad_norm": 1.6474874572664588,
      "learning_rate": 2.5516043377642547e-05,
      "loss": 0.0857,
      "step": 819
    },
    {
      "epoch": 0.5817665838949982,
      "grad_norm": 0.3583242515218562,
      "learning_rate": 2.550349277141229e-05,
      "loss": 0.077,
      "step": 820
    },
    {
      "epoch": 0.5824760553387727,
      "grad_norm": 0.6262164393628536,
      "learning_rate": 2.5490927721164425e-05,
      "loss": 0.0885,
      "step": 821
    },
    {
      "epoch": 0.583185526782547,
      "grad_norm": 0.9977730868441356,
      "learning_rate": 2.547834824417794e-05,
      "loss": 0.1034,
      "step": 822
    },
    {
      "epoch": 0.5838949982263214,
      "grad_norm": 0.9680949153414267,
      "learning_rate": 2.5465754357751677e-05,
      "loss": 0.0827,
      "step": 823
    },
    {
      "epoch": 0.5846044696700958,
      "grad_norm": 2.1428455277941305,
      "learning_rate": 2.545314607920428e-05,
      "loss": 0.0935,
      "step": 824
    },
    {
      "epoch": 0.5853139411138701,
      "grad_norm": 2.031360969713229,
      "learning_rate": 2.54405234258742e-05,
      "loss": 0.1056,
      "step": 825
    },
    {
      "epoch": 0.5860234125576446,
      "grad_norm": 1.1558471430345059,
      "learning_rate": 2.5427886415119635e-05,
      "loss": 0.1053,
      "step": 826
    },
    {
      "epoch": 0.586732884001419,
      "grad_norm": 0.7941591508708179,
      "learning_rate": 2.5415235064318545e-05,
      "loss": 0.0925,
      "step": 827
    },
    {
      "epoch": 0.5874423554451933,
      "grad_norm": 2.1195007389131826,
      "learning_rate": 2.5402569390868603e-05,
      "loss": 0.1014,
      "step": 828
    },
    {
      "epoch": 0.5881518268889677,
      "grad_norm": 3.244665311141249,
      "learning_rate": 2.5389889412187176e-05,
      "loss": 0.1197,
      "step": 829
    },
    {
      "epoch": 0.5888612983327421,
      "grad_norm": 2.122774717720425,
      "learning_rate": 2.53771951457113e-05,
      "loss": 0.091,
      "step": 830
    },
    {
      "epoch": 0.5895707697765165,
      "grad_norm": 1.8047505984883907,
      "learning_rate": 2.5364486608897673e-05,
      "loss": 0.099,
      "step": 831
    },
    {
      "epoch": 0.5902802412202909,
      "grad_norm": 1.6139869401204507,
      "learning_rate": 2.535176381922261e-05,
      "loss": 0.0689,
      "step": 832
    },
    {
      "epoch": 0.5909897126640653,
      "grad_norm": 2.1292515617181764,
      "learning_rate": 2.5339026794182022e-05,
      "loss": 0.1209,
      "step": 833
    },
    {
      "epoch": 0.5916991841078396,
      "grad_norm": 1.9061668229606223,
      "learning_rate": 2.5326275551291387e-05,
      "loss": 0.1089,
      "step": 834
    },
    {
      "epoch": 0.592408655551614,
      "grad_norm": 0.9570073185751499,
      "learning_rate": 2.5313510108085756e-05,
      "loss": 0.0868,
      "step": 835
    },
    {
      "epoch": 0.5931181269953885,
      "grad_norm": 2.0237900698570166,
      "learning_rate": 2.53007304821197e-05,
      "loss": 0.1094,
      "step": 836
    },
    {
      "epoch": 0.5938275984391628,
      "grad_norm": 0.8985982637187208,
      "learning_rate": 2.528793669096729e-05,
      "loss": 0.0719,
      "step": 837
    },
    {
      "epoch": 0.5945370698829372,
      "grad_norm": 0.5753978078219799,
      "learning_rate": 2.527512875222207e-05,
      "loss": 0.0782,
      "step": 838
    },
    {
      "epoch": 0.5952465413267116,
      "grad_norm": 0.7716820927544801,
      "learning_rate": 2.5262306683497055e-05,
      "loss": 0.0634,
      "step": 839
    },
    {
      "epoch": 0.595956012770486,
      "grad_norm": 0.9632918532408972,
      "learning_rate": 2.524947050242468e-05,
      "loss": 0.096,
      "step": 840
    },
    {
      "epoch": 0.5966654842142604,
      "grad_norm": 0.6355599788606805,
      "learning_rate": 2.5236620226656794e-05,
      "loss": 0.0817,
      "step": 841
    },
    {
      "epoch": 0.5973749556580348,
      "grad_norm": 1.3229548173888461,
      "learning_rate": 2.522375587386462e-05,
      "loss": 0.1006,
      "step": 842
    },
    {
      "epoch": 0.5980844271018092,
      "grad_norm": 1.127138809306561,
      "learning_rate": 2.5210877461738744e-05,
      "loss": 0.0874,
      "step": 843
    },
    {
      "epoch": 0.5987938985455835,
      "grad_norm": 0.8935293476001288,
      "learning_rate": 2.5197985007989083e-05,
      "loss": 0.0791,
      "step": 844
    },
    {
      "epoch": 0.5995033699893579,
      "grad_norm": 1.5266498759539535,
      "learning_rate": 2.518507853034488e-05,
      "loss": 0.0829,
      "step": 845
    },
    {
      "epoch": 0.6002128414331324,
      "grad_norm": 2.0989014614444628,
      "learning_rate": 2.517215804655463e-05,
      "loss": 0.0921,
      "step": 846
    },
    {
      "epoch": 0.6009223128769067,
      "grad_norm": 2.0605155760579645,
      "learning_rate": 2.5159223574386117e-05,
      "loss": 0.093,
      "step": 847
    },
    {
      "epoch": 0.6016317843206811,
      "grad_norm": 1.5818159439068615,
      "learning_rate": 2.5146275131626354e-05,
      "loss": 0.0796,
      "step": 848
    },
    {
      "epoch": 0.6023412557644555,
      "grad_norm": 0.7611136827290826,
      "learning_rate": 2.5133312736081563e-05,
      "loss": 0.0931,
      "step": 849
    },
    {
      "epoch": 0.6030507272082298,
      "grad_norm": 0.5726332337113191,
      "learning_rate": 2.5120336405577152e-05,
      "loss": 0.0839,
      "step": 850
    },
    {
      "epoch": 0.6037601986520043,
      "grad_norm": 1.0624547704910516,
      "learning_rate": 2.510734615795769e-05,
      "loss": 0.1098,
      "step": 851
    },
    {
      "epoch": 0.6044696700957787,
      "grad_norm": 1.7695216750776055,
      "learning_rate": 2.5094342011086902e-05,
      "loss": 0.0917,
      "step": 852
    },
    {
      "epoch": 0.605179141539553,
      "grad_norm": 2.0173813017738267,
      "learning_rate": 2.5081323982847602e-05,
      "loss": 0.1065,
      "step": 853
    },
    {
      "epoch": 0.6058886129833274,
      "grad_norm": 1.452417196311744,
      "learning_rate": 2.506829209114171e-05,
      "loss": 0.0627,
      "step": 854
    },
    {
      "epoch": 0.6065980844271018,
      "grad_norm": 0.8805815316368366,
      "learning_rate": 2.50552463538902e-05,
      "loss": 0.0644,
      "step": 855
    },
    {
      "epoch": 0.6073075558708761,
      "grad_norm": 1.987018699939349,
      "learning_rate": 2.5042186789033095e-05,
      "loss": 0.1173,
      "step": 856
    },
    {
      "epoch": 0.6080170273146506,
      "grad_norm": 1.3761985983204825,
      "learning_rate": 2.502911341452943e-05,
      "loss": 0.0886,
      "step": 857
    },
    {
      "epoch": 0.608726498758425,
      "grad_norm": 0.8421769441751846,
      "learning_rate": 2.5016026248357223e-05,
      "loss": 0.0609,
      "step": 858
    },
    {
      "epoch": 0.6094359702021993,
      "grad_norm": 0.787038329096824,
      "learning_rate": 2.5002925308513475e-05,
      "loss": 0.1069,
      "step": 859
    },
    {
      "epoch": 0.6101454416459737,
      "grad_norm": 1.5520753898161679,
      "learning_rate": 2.498981061301411e-05,
      "loss": 0.0954,
      "step": 860
    },
    {
      "epoch": 0.6108549130897482,
      "grad_norm": 0.4868249284046237,
      "learning_rate": 2.4976682179893974e-05,
      "loss": 0.0844,
      "step": 861
    },
    {
      "epoch": 0.6115643845335226,
      "grad_norm": 0.5289063830611047,
      "learning_rate": 2.4963540027206812e-05,
      "loss": 0.1219,
      "step": 862
    },
    {
      "epoch": 0.6122738559772969,
      "grad_norm": 1.396962891016018,
      "learning_rate": 2.4950384173025226e-05,
      "loss": 0.0716,
      "step": 863
    },
    {
      "epoch": 0.6129833274210713,
      "grad_norm": 1.2577692839362375,
      "learning_rate": 2.4937214635440665e-05,
      "loss": 0.0946,
      "step": 864
    },
    {
      "epoch": 0.6136927988648457,
      "grad_norm": 0.6086722079399003,
      "learning_rate": 2.492403143256339e-05,
      "loss": 0.0869,
      "step": 865
    },
    {
      "epoch": 0.61440227030862,
      "grad_norm": 0.9960610420622015,
      "learning_rate": 2.4910834582522466e-05,
      "loss": 0.1012,
      "step": 866
    },
    {
      "epoch": 0.6151117417523945,
      "grad_norm": 1.1862531122799593,
      "learning_rate": 2.489762410346571e-05,
      "loss": 0.0631,
      "step": 867
    },
    {
      "epoch": 0.6158212131961689,
      "grad_norm": 1.7734611678957337,
      "learning_rate": 2.4884400013559688e-05,
      "loss": 0.0984,
      "step": 868
    },
    {
      "epoch": 0.6165306846399432,
      "grad_norm": 3.96825830012531,
      "learning_rate": 2.487116233098969e-05,
      "loss": 0.1226,
      "step": 869
    },
    {
      "epoch": 0.6172401560837176,
      "grad_norm": 2.5935895958927286,
      "learning_rate": 2.4857911073959676e-05,
      "loss": 0.1,
      "step": 870
    },
    {
      "epoch": 0.617949627527492,
      "grad_norm": 0.64578715947487,
      "learning_rate": 2.4844646260692304e-05,
      "loss": 0.064,
      "step": 871
    },
    {
      "epoch": 0.6186590989712664,
      "grad_norm": 0.7064371198186336,
      "learning_rate": 2.483136790942885e-05,
      "loss": 0.0867,
      "step": 872
    },
    {
      "epoch": 0.6193685704150408,
      "grad_norm": 0.6456220234365463,
      "learning_rate": 2.4818076038429214e-05,
      "loss": 0.0626,
      "step": 873
    },
    {
      "epoch": 0.6200780418588152,
      "grad_norm": 1.5889644535933756,
      "learning_rate": 2.4804770665971894e-05,
      "loss": 0.1122,
      "step": 874
    },
    {
      "epoch": 0.6207875133025895,
      "grad_norm": 2.36229311624623,
      "learning_rate": 2.4791451810353948e-05,
      "loss": 0.1218,
      "step": 875
    },
    {
      "epoch": 0.621496984746364,
      "grad_norm": 1.1435066597963754,
      "learning_rate": 2.4778119489890976e-05,
      "loss": 0.0536,
      "step": 876
    },
    {
      "epoch": 0.6222064561901384,
      "grad_norm": 0.6421460597478851,
      "learning_rate": 2.4764773722917094e-05,
      "loss": 0.0822,
      "step": 877
    },
    {
      "epoch": 0.6229159276339127,
      "grad_norm": 1.462402586369017,
      "learning_rate": 2.4751414527784918e-05,
      "loss": 0.1013,
      "step": 878
    },
    {
      "epoch": 0.6236253990776871,
      "grad_norm": 1.2871417789368593,
      "learning_rate": 2.473804192286551e-05,
      "loss": 0.0978,
      "step": 879
    },
    {
      "epoch": 0.6243348705214615,
      "grad_norm": 0.7449031483151409,
      "learning_rate": 2.47246559265484e-05,
      "loss": 0.0737,
      "step": 880
    },
    {
      "epoch": 0.625044341965236,
      "grad_norm": 1.7189189405930545,
      "learning_rate": 2.4711256557241507e-05,
      "loss": 0.073,
      "step": 881
    },
    {
      "epoch": 0.6257538134090103,
      "grad_norm": 1.114608101781195,
      "learning_rate": 2.4697843833371156e-05,
      "loss": 0.1015,
      "step": 882
    },
    {
      "epoch": 0.6264632848527847,
      "grad_norm": 1.9049891229967852,
      "learning_rate": 2.4684417773382037e-05,
      "loss": 0.1025,
      "step": 883
    },
    {
      "epoch": 0.6271727562965591,
      "grad_norm": 1.238574438484414,
      "learning_rate": 2.467097839573717e-05,
      "loss": 0.0802,
      "step": 884
    },
    {
      "epoch": 0.6278822277403334,
      "grad_norm": 1.0114207756336435,
      "learning_rate": 2.4657525718917887e-05,
      "loss": 0.093,
      "step": 885
    },
    {
      "epoch": 0.6285916991841078,
      "grad_norm": 1.8354185266302383,
      "learning_rate": 2.4644059761423827e-05,
      "loss": 0.0949,
      "step": 886
    },
    {
      "epoch": 0.6293011706278823,
      "grad_norm": 1.0836390869477819,
      "learning_rate": 2.463058054177288e-05,
      "loss": 0.0755,
      "step": 887
    },
    {
      "epoch": 0.6300106420716566,
      "grad_norm": 1.066045965499429,
      "learning_rate": 2.4617088078501163e-05,
      "loss": 0.1139,
      "step": 888
    },
    {
      "epoch": 0.630720113515431,
      "grad_norm": 0.828729187302372,
      "learning_rate": 2.4603582390163023e-05,
      "loss": 0.1,
      "step": 889
    },
    {
      "epoch": 0.6314295849592054,
      "grad_norm": 1.4089547827862339,
      "learning_rate": 2.459006349533099e-05,
      "loss": 0.1028,
      "step": 890
    },
    {
      "epoch": 0.6321390564029797,
      "grad_norm": 0.5858434642074677,
      "learning_rate": 2.4576531412595745e-05,
      "loss": 0.0591,
      "step": 891
    },
    {
      "epoch": 0.6328485278467542,
      "grad_norm": 0.6798521381418388,
      "learning_rate": 2.4562986160566117e-05,
      "loss": 0.1005,
      "step": 892
    },
    {
      "epoch": 0.6335579992905286,
      "grad_norm": 0.6506463919191321,
      "learning_rate": 2.4549427757869038e-05,
      "loss": 0.0821,
      "step": 893
    },
    {
      "epoch": 0.6342674707343029,
      "grad_norm": 1.882694465032466,
      "learning_rate": 2.4535856223149525e-05,
      "loss": 0.0846,
      "step": 894
    },
    {
      "epoch": 0.6349769421780773,
      "grad_norm": 0.6150434826675685,
      "learning_rate": 2.452227157507065e-05,
      "loss": 0.0866,
      "step": 895
    },
    {
      "epoch": 0.6356864136218517,
      "grad_norm": 1.688939520741653,
      "learning_rate": 2.4508673832313525e-05,
      "loss": 0.0699,
      "step": 896
    },
    {
      "epoch": 0.636395885065626,
      "grad_norm": 0.7121104706825553,
      "learning_rate": 2.449506301357727e-05,
      "loss": 0.0952,
      "step": 897
    },
    {
      "epoch": 0.6371053565094005,
      "grad_norm": 1.1626310666894542,
      "learning_rate": 2.4481439137578977e-05,
      "loss": 0.0882,
      "step": 898
    },
    {
      "epoch": 0.6378148279531749,
      "grad_norm": 0.9853285020494227,
      "learning_rate": 2.446780222305371e-05,
      "loss": 0.1004,
      "step": 899
    },
    {
      "epoch": 0.6385242993969493,
      "grad_norm": 0.5977362751205011,
      "learning_rate": 2.445415228875444e-05,
      "loss": 0.0926,
      "step": 900
    }
  ],
  "logging_steps": 1,
  "max_steps": 2820,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 50,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 581717677735936.0,
  "train_batch_size": 10,
  "trial_name": null,
  "trial_params": null
}