{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.666556455612498,
  "eval_steps": 300,
  "global_step": 3024,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00022042210833746626,
      "grad_norm": 1.0902734994888306,
      "learning_rate": 0.0,
      "loss": 2.8631,
      "step": 1
    },
    {
      "epoch": 0.0004408442166749325,
      "grad_norm": 1.0188628435134888,
      "learning_rate": 8.810572687224671e-07,
      "loss": 2.8811,
      "step": 2
    },
    {
      "epoch": 0.0006612663250123987,
      "grad_norm": 1.1590542793273926,
      "learning_rate": 1.7621145374449343e-06,
      "loss": 2.9364,
      "step": 3
    },
    {
      "epoch": 0.000881688433349865,
      "grad_norm": 1.0604320764541626,
      "learning_rate": 2.643171806167401e-06,
      "loss": 2.951,
      "step": 4
    },
    {
      "epoch": 0.0011021105416873312,
      "grad_norm": 1.1602253913879395,
      "learning_rate": 3.5242290748898685e-06,
      "loss": 2.8145,
      "step": 5
    },
    {
      "epoch": 0.0013225326500247974,
      "grad_norm": 1.0046895742416382,
      "learning_rate": 4.4052863436123355e-06,
      "loss": 2.9343,
      "step": 6
    },
    {
      "epoch": 0.0015429547583622636,
      "grad_norm": 1.148353099822998,
      "learning_rate": 5.286343612334802e-06,
      "loss": 3.0591,
      "step": 7
    },
    {
      "epoch": 0.00176337686669973,
      "grad_norm": 1.019972324371338,
      "learning_rate": 6.167400881057269e-06,
      "loss": 2.8394,
      "step": 8
    },
    {
      "epoch": 0.001983798975037196,
      "grad_norm": 1.0576002597808838,
      "learning_rate": 7.048458149779737e-06,
      "loss": 2.9543,
      "step": 9
    },
    {
      "epoch": 0.0022042210833746623,
      "grad_norm": 1.0417252779006958,
      "learning_rate": 7.929515418502203e-06,
      "loss": 2.9992,
      "step": 10
    },
    {
      "epoch": 0.0024246431917121286,
      "grad_norm": 1.0588741302490234,
      "learning_rate": 8.810572687224671e-06,
      "loss": 2.943,
      "step": 11
    },
    {
      "epoch": 0.002645065300049595,
      "grad_norm": 1.0592830181121826,
      "learning_rate": 9.691629955947137e-06,
      "loss": 3.085,
      "step": 12
    },
    {
      "epoch": 0.002865487408387061,
      "grad_norm": 1.0683331489562988,
      "learning_rate": 1.0572687224669605e-05,
      "loss": 3.0791,
      "step": 13
    },
    {
      "epoch": 0.0030859095167245273,
      "grad_norm": 1.0180429220199585,
      "learning_rate": 1.1453744493392071e-05,
      "loss": 2.8024,
      "step": 14
    },
    {
      "epoch": 0.0033063316250619935,
      "grad_norm": 1.0357601642608643,
      "learning_rate": 1.2334801762114539e-05,
      "loss": 2.9536,
      "step": 15
    },
    {
      "epoch": 0.00352675373339946,
      "grad_norm": 0.9907547831535339,
      "learning_rate": 1.3215859030837005e-05,
      "loss": 2.9941,
      "step": 16
    },
    {
      "epoch": 0.0037471758417369264,
      "grad_norm": 1.0317749977111816,
      "learning_rate": 1.4096916299559474e-05,
      "loss": 2.9267,
      "step": 17
    },
    {
      "epoch": 0.003967597950074392,
      "grad_norm": 0.977344810962677,
      "learning_rate": 1.4977973568281939e-05,
      "loss": 2.9133,
      "step": 18
    },
    {
      "epoch": 0.0041880200584118584,
      "grad_norm": 0.9885391592979431,
      "learning_rate": 1.5859030837004406e-05,
      "loss": 2.911,
      "step": 19
    },
    {
      "epoch": 0.004408442166749325,
      "grad_norm": 1.039089560508728,
      "learning_rate": 1.6740088105726872e-05,
      "loss": 2.9327,
      "step": 20
    },
    {
      "epoch": 0.004628864275086791,
      "grad_norm": 1.0893826484680176,
      "learning_rate": 1.7621145374449342e-05,
      "loss": 2.9438,
      "step": 21
    },
    {
      "epoch": 0.004849286383424257,
      "grad_norm": 1.0010955333709717,
      "learning_rate": 1.8502202643171808e-05,
      "loss": 2.9752,
      "step": 22
    },
    {
      "epoch": 0.005069708491761723,
      "grad_norm": 1.0444799661636353,
      "learning_rate": 1.9383259911894274e-05,
      "loss": 2.9847,
      "step": 23
    },
    {
      "epoch": 0.00529013060009919,
      "grad_norm": 0.9992923736572266,
      "learning_rate": 2.026431718061674e-05,
      "loss": 2.9711,
      "step": 24
    },
    {
      "epoch": 0.005510552708436656,
      "grad_norm": 0.9513102173805237,
      "learning_rate": 2.114537444933921e-05,
      "loss": 2.9208,
      "step": 25
    },
    {
      "epoch": 0.005730974816774122,
      "grad_norm": 0.9202016592025757,
      "learning_rate": 2.2026431718061676e-05,
      "loss": 2.8817,
      "step": 26
    },
    {
      "epoch": 0.005951396925111588,
      "grad_norm": 1.0238031148910522,
      "learning_rate": 2.2907488986784142e-05,
      "loss": 2.8754,
      "step": 27
    },
    {
      "epoch": 0.0061718190334490546,
      "grad_norm": 0.8789689540863037,
      "learning_rate": 2.3788546255506608e-05,
      "loss": 2.834,
      "step": 28
    },
    {
      "epoch": 0.006392241141786521,
      "grad_norm": 0.8910471200942993,
      "learning_rate": 2.4669603524229077e-05,
      "loss": 2.901,
      "step": 29
    },
    {
      "epoch": 0.006612663250123987,
      "grad_norm": 0.9833038449287415,
      "learning_rate": 2.5550660792951543e-05,
      "loss": 2.9124,
      "step": 30
    },
    {
      "epoch": 0.006833085358461453,
      "grad_norm": 0.9674044847488403,
      "learning_rate": 2.643171806167401e-05,
      "loss": 2.8915,
      "step": 31
    },
    {
      "epoch": 0.00705350746679892,
      "grad_norm": 0.9992914795875549,
      "learning_rate": 2.731277533039648e-05,
      "loss": 2.8713,
      "step": 32
    },
    {
      "epoch": 0.007273929575136387,
      "grad_norm": 0.9932569265365601,
      "learning_rate": 2.819383259911895e-05,
      "loss": 2.7879,
      "step": 33
    },
    {
      "epoch": 0.007494351683473853,
      "grad_norm": 0.9038912653923035,
      "learning_rate": 2.907488986784141e-05,
      "loss": 2.8429,
      "step": 34
    },
    {
      "epoch": 0.007714773791811319,
      "grad_norm": 1.0180914402008057,
      "learning_rate": 2.9955947136563877e-05,
      "loss": 2.8419,
      "step": 35
    },
    {
      "epoch": 0.007935195900148784,
      "grad_norm": 0.9740180969238281,
      "learning_rate": 3.0837004405286347e-05,
      "loss": 2.8557,
      "step": 36
    },
    {
      "epoch": 0.008155618008486252,
      "grad_norm": 0.9407585859298706,
      "learning_rate": 3.171806167400881e-05,
      "loss": 2.9522,
      "step": 37
    },
    {
      "epoch": 0.008376040116823717,
      "grad_norm": 1.0118271112442017,
      "learning_rate": 3.259911894273128e-05,
      "loss": 3.0238,
      "step": 38
    },
    {
      "epoch": 0.008596462225161184,
      "grad_norm": 0.8947811722755432,
      "learning_rate": 3.3480176211453745e-05,
      "loss": 2.9336,
      "step": 39
    },
    {
      "epoch": 0.00881688433349865,
      "grad_norm": 0.8921943306922913,
      "learning_rate": 3.436123348017621e-05,
      "loss": 2.8537,
      "step": 40
    },
    {
      "epoch": 0.009037306441836116,
      "grad_norm": 0.9475488662719727,
      "learning_rate": 3.5242290748898684e-05,
      "loss": 2.8783,
      "step": 41
    },
    {
      "epoch": 0.009257728550173582,
      "grad_norm": 0.9774518609046936,
      "learning_rate": 3.612334801762114e-05,
      "loss": 3.0864,
      "step": 42
    },
    {
      "epoch": 0.009478150658511049,
      "grad_norm": 0.9830888509750366,
      "learning_rate": 3.7004405286343616e-05,
      "loss": 2.925,
      "step": 43
    },
    {
      "epoch": 0.009698572766848514,
      "grad_norm": 0.9802113175392151,
      "learning_rate": 3.788546255506608e-05,
      "loss": 2.8993,
      "step": 44
    },
    {
      "epoch": 0.009918994875185981,
      "grad_norm": 1.0162456035614014,
      "learning_rate": 3.876651982378855e-05,
      "loss": 2.9194,
      "step": 45
    },
    {
      "epoch": 0.010139416983523447,
      "grad_norm": 0.9545007944107056,
      "learning_rate": 3.9647577092511014e-05,
      "loss": 2.8747,
      "step": 46
    },
    {
      "epoch": 0.010359839091860914,
      "grad_norm": 0.9669350385665894,
      "learning_rate": 4.052863436123348e-05,
      "loss": 3.0318,
      "step": 47
    },
    {
      "epoch": 0.01058026120019838,
      "grad_norm": 1.0896497964859009,
      "learning_rate": 4.1409691629955946e-05,
      "loss": 2.8929,
      "step": 48
    },
    {
      "epoch": 0.010800683308535846,
      "grad_norm": 0.9810357689857483,
      "learning_rate": 4.229074889867842e-05,
      "loss": 2.9166,
      "step": 49
    },
    {
      "epoch": 0.011021105416873312,
      "grad_norm": 1.0470596551895142,
      "learning_rate": 4.3171806167400885e-05,
      "loss": 2.8509,
      "step": 50
    },
    {
      "epoch": 0.011241527525210779,
      "grad_norm": 0.920775830745697,
      "learning_rate": 4.405286343612335e-05,
      "loss": 2.8554,
      "step": 51
    },
    {
      "epoch": 0.011461949633548244,
      "grad_norm": 0.9224264621734619,
      "learning_rate": 4.493392070484582e-05,
      "loss": 2.9209,
      "step": 52
    },
    {
      "epoch": 0.011682371741885711,
      "grad_norm": 0.9094575643539429,
      "learning_rate": 4.5814977973568283e-05,
      "loss": 2.7529,
      "step": 53
    },
    {
      "epoch": 0.011902793850223177,
      "grad_norm": 0.9546058177947998,
      "learning_rate": 4.6696035242290756e-05,
      "loss": 2.9499,
      "step": 54
    },
    {
      "epoch": 0.012123215958560644,
      "grad_norm": 0.9186086654663086,
      "learning_rate": 4.7577092511013216e-05,
      "loss": 2.8111,
      "step": 55
    },
    {
      "epoch": 0.012343638066898109,
      "grad_norm": 1.0403311252593994,
      "learning_rate": 4.845814977973568e-05,
      "loss": 2.8112,
      "step": 56
    },
    {
      "epoch": 0.012564060175235576,
      "grad_norm": 1.0234464406967163,
      "learning_rate": 4.9339207048458155e-05,
      "loss": 2.9278,
      "step": 57
    },
    {
      "epoch": 0.012784482283573042,
      "grad_norm": 0.9609548449516296,
      "learning_rate": 5.022026431718062e-05,
      "loss": 2.8863,
      "step": 58
    },
    {
      "epoch": 0.013004904391910509,
      "grad_norm": 0.9902917742729187,
      "learning_rate": 5.110132158590309e-05,
      "loss": 2.9219,
      "step": 59
    },
    {
      "epoch": 0.013225326500247974,
      "grad_norm": 1.004833698272705,
      "learning_rate": 5.198237885462556e-05,
      "loss": 2.9008,
      "step": 60
    },
    {
      "epoch": 0.013445748608585441,
      "grad_norm": 1.04329514503479,
      "learning_rate": 5.286343612334802e-05,
      "loss": 2.8337,
      "step": 61
    },
    {
      "epoch": 0.013666170716922907,
      "grad_norm": 0.9978932738304138,
      "learning_rate": 5.3744493392070485e-05,
      "loss": 2.8229,
      "step": 62
    },
    {
      "epoch": 0.013886592825260374,
      "grad_norm": 0.9282989501953125,
      "learning_rate": 5.462555066079296e-05,
      "loss": 2.9685,
      "step": 63
    },
    {
      "epoch": 0.01410701493359784,
      "grad_norm": 0.9025146961212158,
      "learning_rate": 5.550660792951542e-05,
      "loss": 2.7947,
      "step": 64
    },
    {
      "epoch": 0.014327437041935306,
      "grad_norm": 1.0075218677520752,
      "learning_rate": 5.63876651982379e-05,
      "loss": 2.7726,
      "step": 65
    },
    {
      "epoch": 0.014547859150272773,
      "grad_norm": 0.9327206611633301,
      "learning_rate": 5.7268722466960356e-05,
      "loss": 2.8897,
      "step": 66
    },
    {
      "epoch": 0.014768281258610239,
      "grad_norm": 0.9513474702835083,
      "learning_rate": 5.814977973568282e-05,
      "loss": 2.8438,
      "step": 67
    },
    {
      "epoch": 0.014988703366947706,
      "grad_norm": 1.053801417350769,
      "learning_rate": 5.9030837004405295e-05,
      "loss": 2.8045,
      "step": 68
    },
    {
      "epoch": 0.015209125475285171,
      "grad_norm": 1.0158675909042358,
      "learning_rate": 5.9911894273127754e-05,
      "loss": 2.8478,
      "step": 69
    },
    {
      "epoch": 0.015429547583622638,
      "grad_norm": 0.9726863503456116,
      "learning_rate": 6.079295154185022e-05,
      "loss": 2.8792,
      "step": 70
    },
    {
      "epoch": 0.015649969691960103,
      "grad_norm": 0.9474403262138367,
      "learning_rate": 6.167400881057269e-05,
      "loss": 2.9084,
      "step": 71
    },
    {
      "epoch": 0.01587039180029757,
      "grad_norm": 1.0420753955841064,
      "learning_rate": 6.255506607929515e-05,
      "loss": 2.8498,
      "step": 72
    },
    {
      "epoch": 0.016090813908635038,
      "grad_norm": 0.9667215943336487,
      "learning_rate": 6.343612334801763e-05,
      "loss": 2.7516,
      "step": 73
    },
    {
      "epoch": 0.016311236016972503,
      "grad_norm": 1.0524910688400269,
      "learning_rate": 6.43171806167401e-05,
      "loss": 2.8404,
      "step": 74
    },
    {
      "epoch": 0.01653165812530997,
      "grad_norm": 1.0096701383590698,
      "learning_rate": 6.519823788546256e-05,
      "loss": 2.7015,
      "step": 75
    },
    {
      "epoch": 0.016752080233647434,
      "grad_norm": 0.9700260758399963,
      "learning_rate": 6.607929515418503e-05,
      "loss": 2.8014,
      "step": 76
    },
    {
      "epoch": 0.016972502341984903,
      "grad_norm": 1.0620126724243164,
      "learning_rate": 6.696035242290749e-05,
      "loss": 2.8354,
      "step": 77
    },
    {
      "epoch": 0.017192924450322368,
      "grad_norm": 1.0552520751953125,
      "learning_rate": 6.784140969162996e-05,
      "loss": 2.8699,
      "step": 78
    },
    {
      "epoch": 0.017413346558659833,
      "grad_norm": 1.016897439956665,
      "learning_rate": 6.872246696035242e-05,
      "loss": 2.8191,
      "step": 79
    },
    {
      "epoch": 0.0176337686669973,
      "grad_norm": 1.0052814483642578,
      "learning_rate": 6.96035242290749e-05,
      "loss": 2.8606,
      "step": 80
    },
    {
      "epoch": 0.017854190775334768,
      "grad_norm": 0.9919174909591675,
      "learning_rate": 7.048458149779737e-05,
      "loss": 2.7872,
      "step": 81
    },
    {
      "epoch": 0.018074612883672233,
      "grad_norm": 1.0091278553009033,
      "learning_rate": 7.136563876651983e-05,
      "loss": 2.7858,
      "step": 82
    },
    {
      "epoch": 0.0182950349920097,
      "grad_norm": 1.0801664590835571,
      "learning_rate": 7.224669603524229e-05,
      "loss": 2.8844,
      "step": 83
    },
    {
      "epoch": 0.018515457100347164,
      "grad_norm": 1.0763691663742065,
      "learning_rate": 7.312775330396476e-05,
      "loss": 2.8109,
      "step": 84
    },
    {
      "epoch": 0.018735879208684632,
      "grad_norm": 1.0106059312820435,
      "learning_rate": 7.400881057268723e-05,
      "loss": 2.8356,
      "step": 85
    },
    {
      "epoch": 0.018956301317022098,
      "grad_norm": 1.115525722503662,
      "learning_rate": 7.48898678414097e-05,
      "loss": 2.8525,
      "step": 86
    },
    {
      "epoch": 0.019176723425359563,
      "grad_norm": 1.1630377769470215,
      "learning_rate": 7.577092511013216e-05,
      "loss": 2.8956,
      "step": 87
    },
    {
      "epoch": 0.01939714553369703,
      "grad_norm": 1.085883378982544,
      "learning_rate": 7.665198237885462e-05,
      "loss": 2.8353,
      "step": 88
    },
    {
      "epoch": 0.019617567642034497,
      "grad_norm": 1.0363401174545288,
      "learning_rate": 7.75330396475771e-05,
      "loss": 2.7295,
      "step": 89
    },
    {
      "epoch": 0.019837989750371963,
      "grad_norm": 1.1129146814346313,
      "learning_rate": 7.841409691629956e-05,
      "loss": 2.9385,
      "step": 90
    },
    {
      "epoch": 0.020058411858709428,
      "grad_norm": 1.209983468055725,
      "learning_rate": 7.929515418502203e-05,
      "loss": 2.8222,
      "step": 91
    },
    {
      "epoch": 0.020278833967046894,
      "grad_norm": 1.1120164394378662,
      "learning_rate": 8.01762114537445e-05,
      "loss": 2.7613,
      "step": 92
    },
    {
      "epoch": 0.020499256075384362,
      "grad_norm": 1.026802897453308,
      "learning_rate": 8.105726872246696e-05,
      "loss": 2.8145,
      "step": 93
    },
    {
      "epoch": 0.020719678183721828,
      "grad_norm": 1.1136006116867065,
      "learning_rate": 8.193832599118943e-05,
      "loss": 2.7817,
      "step": 94
    },
    {
      "epoch": 0.020940100292059293,
      "grad_norm": 1.1209471225738525,
      "learning_rate": 8.281938325991189e-05,
      "loss": 2.8607,
      "step": 95
    },
    {
      "epoch": 0.02116052240039676,
      "grad_norm": 1.1269468069076538,
      "learning_rate": 8.370044052863437e-05,
      "loss": 2.7962,
      "step": 96
    },
    {
      "epoch": 0.021380944508734227,
      "grad_norm": 0.986998438835144,
      "learning_rate": 8.458149779735684e-05,
      "loss": 2.8029,
      "step": 97
    },
    {
      "epoch": 0.021601366617071693,
      "grad_norm": 1.082733392715454,
      "learning_rate": 8.54625550660793e-05,
      "loss": 2.8204,
      "step": 98
    },
    {
      "epoch": 0.021821788725409158,
      "grad_norm": 1.0857740640640259,
      "learning_rate": 8.634361233480177e-05,
      "loss": 2.8793,
      "step": 99
    },
    {
      "epoch": 0.022042210833746623,
      "grad_norm": 1.1036165952682495,
      "learning_rate": 8.722466960352423e-05,
      "loss": 2.7183,
      "step": 100
    },
    {
      "epoch": 0.022262632942084092,
      "grad_norm": 1.0742735862731934,
      "learning_rate": 8.81057268722467e-05,
      "loss": 2.7941,
      "step": 101
    },
    {
      "epoch": 0.022483055050421558,
      "grad_norm": 1.054746389389038,
      "learning_rate": 8.898678414096918e-05,
      "loss": 2.6803,
      "step": 102
    },
    {
      "epoch": 0.022703477158759023,
      "grad_norm": 1.1802846193313599,
      "learning_rate": 8.986784140969163e-05,
      "loss": 2.7274,
      "step": 103
    },
    {
      "epoch": 0.02292389926709649,
      "grad_norm": 1.1999224424362183,
      "learning_rate": 9.07488986784141e-05,
      "loss": 2.7644,
      "step": 104
    },
    {
      "epoch": 0.023144321375433957,
      "grad_norm": 1.1602308750152588,
      "learning_rate": 9.162995594713657e-05,
      "loss": 2.7682,
      "step": 105
    },
    {
      "epoch": 0.023364743483771423,
      "grad_norm": 1.1744916439056396,
      "learning_rate": 9.251101321585903e-05,
      "loss": 2.8672,
      "step": 106
    },
    {
      "epoch": 0.023585165592108888,
      "grad_norm": 1.2088391780853271,
      "learning_rate": 9.339207048458151e-05,
      "loss": 2.9259,
      "step": 107
    },
    {
      "epoch": 0.023805587700446353,
      "grad_norm": 1.3755017518997192,
      "learning_rate": 9.427312775330397e-05,
      "loss": 2.8693,
      "step": 108
    },
    {
      "epoch": 0.024026009808783822,
      "grad_norm": 1.1486746072769165,
      "learning_rate": 9.515418502202643e-05,
      "loss": 2.8097,
      "step": 109
    },
    {
      "epoch": 0.024246431917121287,
      "grad_norm": 1.1463654041290283,
      "learning_rate": 9.60352422907489e-05,
      "loss": 2.8188,
      "step": 110
    },
    {
      "epoch": 0.024466854025458753,
      "grad_norm": 1.1618080139160156,
      "learning_rate": 9.691629955947136e-05,
      "loss": 2.7484,
      "step": 111
    },
    {
      "epoch": 0.024687276133796218,
      "grad_norm": 1.2124923467636108,
      "learning_rate": 9.779735682819384e-05,
      "loss": 2.6552,
      "step": 112
    },
    {
      "epoch": 0.024907698242133687,
      "grad_norm": 1.0731984376907349,
      "learning_rate": 9.867841409691631e-05,
      "loss": 2.7287,
      "step": 113
    },
    {
      "epoch": 0.025128120350471152,
      "grad_norm": 1.2142680883407593,
      "learning_rate": 9.955947136563877e-05,
      "loss": 2.8168,
      "step": 114
    },
    {
      "epoch": 0.025348542458808618,
      "grad_norm": 1.0104131698608398,
      "learning_rate": 0.00010044052863436124,
      "loss": 2.6983,
      "step": 115
    },
    {
      "epoch": 0.025568964567146083,
      "grad_norm": 1.133567452430725,
      "learning_rate": 0.00010132158590308371,
      "loss": 2.7352,
      "step": 116
    },
    {
      "epoch": 0.025789386675483552,
      "grad_norm": 1.1962589025497437,
      "learning_rate": 0.00010220264317180617,
      "loss": 2.7111,
      "step": 117
    },
    {
      "epoch": 0.026009808783821017,
      "grad_norm": 1.1725496053695679,
      "learning_rate": 0.00010308370044052865,
      "loss": 2.8093,
      "step": 118
    },
    {
      "epoch": 0.026230230892158483,
      "grad_norm": 1.1837313175201416,
      "learning_rate": 0.00010396475770925112,
      "loss": 2.7728,
      "step": 119
    },
    {
      "epoch": 0.026450653000495948,
      "grad_norm": 1.159765362739563,
      "learning_rate": 0.00010484581497797356,
      "loss": 2.7307,
      "step": 120
    },
    {
      "epoch": 0.026671075108833417,
      "grad_norm": 1.146296501159668,
      "learning_rate": 0.00010572687224669604,
      "loss": 2.7501,
      "step": 121
    },
    {
      "epoch": 0.026891497217170882,
      "grad_norm": 1.1889545917510986,
      "learning_rate": 0.00010660792951541851,
      "loss": 2.7121,
      "step": 122
    },
    {
      "epoch": 0.027111919325508348,
      "grad_norm": 1.2013332843780518,
      "learning_rate": 0.00010748898678414097,
      "loss": 2.7321,
      "step": 123
    },
    {
      "epoch": 0.027332341433845813,
      "grad_norm": 1.1040483713150024,
      "learning_rate": 0.00010837004405286344,
      "loss": 2.6157,
      "step": 124
    },
    {
      "epoch": 0.027552763542183282,
      "grad_norm": 1.1754599809646606,
      "learning_rate": 0.00010925110132158592,
      "loss": 2.7483,
      "step": 125
    },
    {
      "epoch": 0.027773185650520747,
      "grad_norm": 1.076499104499817,
      "learning_rate": 0.00011013215859030836,
      "loss": 2.7957,
      "step": 126
    },
    {
      "epoch": 0.027993607758858213,
      "grad_norm": 1.0961052179336548,
      "learning_rate": 0.00011101321585903083,
      "loss": 2.6854,
      "step": 127
    },
    {
      "epoch": 0.02821402986719568,
      "grad_norm": 1.3141061067581177,
      "learning_rate": 0.00011189427312775331,
      "loss": 2.7701,
      "step": 128
    },
    {
      "epoch": 0.028434451975533147,
      "grad_norm": 1.1996828317642212,
      "learning_rate": 0.0001127753303964758,
      "loss": 2.7153,
      "step": 129
    },
    {
      "epoch": 0.028654874083870612,
      "grad_norm": 1.1670674085617065,
      "learning_rate": 0.00011365638766519824,
      "loss": 2.7157,
      "step": 130
    },
    {
      "epoch": 0.028875296192208078,
      "grad_norm": 1.2428879737854004,
      "learning_rate": 0.00011453744493392071,
      "loss": 2.899,
      "step": 131
    },
    {
      "epoch": 0.029095718300545546,
      "grad_norm": 1.3322181701660156,
      "learning_rate": 0.00011541850220264318,
      "loss": 2.7512,
      "step": 132
    },
    {
      "epoch": 0.02931614040888301,
      "grad_norm": 1.2957340478897095,
      "learning_rate": 0.00011629955947136564,
      "loss": 2.7296,
      "step": 133
    },
    {
      "epoch": 0.029536562517220477,
      "grad_norm": 1.2379623651504517,
      "learning_rate": 0.00011718061674008812,
      "loss": 2.821,
      "step": 134
    },
    {
      "epoch": 0.029756984625557942,
      "grad_norm": 1.267970085144043,
      "learning_rate": 0.00011806167400881059,
      "loss": 2.6735,
      "step": 135
    },
    {
      "epoch": 0.02997740673389541,
      "grad_norm": 1.2754652500152588,
      "learning_rate": 0.00011894273127753304,
      "loss": 2.758,
      "step": 136
    },
    {
      "epoch": 0.030197828842232877,
      "grad_norm": 1.124410629272461,
      "learning_rate": 0.00011982378854625551,
      "loss": 2.6013,
      "step": 137
    },
    {
      "epoch": 0.030418250950570342,
      "grad_norm": 1.307695746421814,
      "learning_rate": 0.00012070484581497798,
      "loss": 2.7469,
      "step": 138
    },
    {
      "epoch": 0.030638673058907807,
      "grad_norm": 1.1287624835968018,
      "learning_rate": 0.00012158590308370044,
      "loss": 2.787,
      "step": 139
    },
    {
      "epoch": 0.030859095167245276,
      "grad_norm": 1.2877527475357056,
      "learning_rate": 0.0001224669603524229,
      "loss": 2.6635,
      "step": 140
    },
    {
      "epoch": 0.03107951727558274,
      "grad_norm": 1.2947174310684204,
      "learning_rate": 0.00012334801762114539,
      "loss": 2.7466,
      "step": 141
    },
    {
      "epoch": 0.03129993938392021,
      "grad_norm": 1.1997637748718262,
      "learning_rate": 0.00012422907488986786,
      "loss": 2.7669,
      "step": 142
    },
    {
      "epoch": 0.031520361492257676,
      "grad_norm": 1.166050910949707,
      "learning_rate": 0.0001251101321585903,
      "loss": 2.7069,
      "step": 143
    },
    {
      "epoch": 0.03174078360059514,
      "grad_norm": 1.1992591619491577,
      "learning_rate": 0.00012599118942731278,
      "loss": 2.6699,
      "step": 144
    },
    {
      "epoch": 0.03196120570893261,
      "grad_norm": 1.3016748428344727,
      "learning_rate": 0.00012687224669603525,
      "loss": 2.7157,
      "step": 145
    },
    {
      "epoch": 0.032181627817270075,
      "grad_norm": 1.2239021062850952,
      "learning_rate": 0.0001277533039647577,
      "loss": 2.6428,
      "step": 146
    },
    {
      "epoch": 0.03240204992560754,
      "grad_norm": 1.128831148147583,
      "learning_rate": 0.0001286343612334802,
      "loss": 2.7575,
      "step": 147
    },
    {
      "epoch": 0.032622472033945006,
      "grad_norm": 1.1445996761322021,
      "learning_rate": 0.00012951541850220267,
      "loss": 2.7332,
      "step": 148
    },
    {
      "epoch": 0.03284289414228247,
      "grad_norm": 1.2626391649246216,
      "learning_rate": 0.00013039647577092511,
      "loss": 2.7013,
      "step": 149
    },
    {
      "epoch": 0.03306331625061994,
      "grad_norm": 1.2260299921035767,
      "learning_rate": 0.0001312775330396476,
      "loss": 2.6778,
      "step": 150
    },
    {
      "epoch": 0.033283738358957406,
      "grad_norm": 1.1503586769104004,
      "learning_rate": 0.00013215859030837006,
      "loss": 2.7366,
      "step": 151
    },
    {
      "epoch": 0.03350416046729487,
      "grad_norm": 1.1261030435562134,
      "learning_rate": 0.0001330396475770925,
      "loss": 2.7239,
      "step": 152
    },
    {
      "epoch": 0.033724582575632336,
      "grad_norm": 1.185463786125183,
      "learning_rate": 0.00013392070484581498,
      "loss": 2.7341,
      "step": 153
    },
    {
      "epoch": 0.033945004683969805,
      "grad_norm": 1.3002212047576904,
      "learning_rate": 0.00013480176211453745,
      "loss": 2.6794,
      "step": 154
    },
    {
      "epoch": 0.03416542679230727,
      "grad_norm": 1.280072569847107,
      "learning_rate": 0.00013568281938325992,
      "loss": 2.7224,
      "step": 155
    },
    {
      "epoch": 0.034385848900644736,
      "grad_norm": 1.2194390296936035,
      "learning_rate": 0.00013656387665198237,
      "loss": 2.7972,
      "step": 156
    },
    {
      "epoch": 0.0346062710089822,
      "grad_norm": 1.1741853952407837,
      "learning_rate": 0.00013744493392070484,
      "loss": 2.8229,
      "step": 157
    },
    {
      "epoch": 0.03482669311731967,
      "grad_norm": 1.2798399925231934,
      "learning_rate": 0.00013832599118942732,
      "loss": 2.8484,
      "step": 158
    },
    {
      "epoch": 0.035047115225657136,
      "grad_norm": 1.103670597076416,
      "learning_rate": 0.0001392070484581498,
      "loss": 2.6493,
      "step": 159
    },
    {
      "epoch": 0.0352675373339946,
      "grad_norm": 1.1544209718704224,
      "learning_rate": 0.00014008810572687226,
      "loss": 2.6755,
      "step": 160
    },
    {
      "epoch": 0.035487959442332066,
      "grad_norm": 1.2688531875610352,
      "learning_rate": 0.00014096916299559473,
      "loss": 2.6409,
      "step": 161
    },
    {
      "epoch": 0.035708381550669535,
      "grad_norm": 1.202978491783142,
      "learning_rate": 0.00014185022026431718,
      "loss": 2.7178,
      "step": 162
    },
    {
      "epoch": 0.035928803659007,
      "grad_norm": 1.1194953918457031,
      "learning_rate": 0.00014273127753303965,
      "loss": 2.6311,
      "step": 163
    },
    {
      "epoch": 0.036149225767344466,
      "grad_norm": 1.2073768377304077,
      "learning_rate": 0.00014361233480176213,
      "loss": 2.7232,
      "step": 164
    },
    {
      "epoch": 0.03636964787568193,
      "grad_norm": 1.2793333530426025,
      "learning_rate": 0.00014449339207048457,
      "loss": 2.8138,
      "step": 165
    },
    {
      "epoch": 0.0365900699840194,
      "grad_norm": 1.2258241176605225,
      "learning_rate": 0.00014537444933920705,
      "loss": 2.7766,
      "step": 166
    },
    {
      "epoch": 0.036810492092356865,
      "grad_norm": 1.2007017135620117,
      "learning_rate": 0.00014625550660792952,
      "loss": 2.795,
      "step": 167
    },
    {
      "epoch": 0.03703091420069433,
      "grad_norm": 1.2315441370010376,
      "learning_rate": 0.000147136563876652,
      "loss": 2.7523,
      "step": 168
    },
    {
      "epoch": 0.037251336309031796,
      "grad_norm": 1.1540018320083618,
      "learning_rate": 0.00014801762114537446,
      "loss": 2.6695,
      "step": 169
    },
    {
      "epoch": 0.037471758417369265,
      "grad_norm": 1.3196309804916382,
      "learning_rate": 0.00014889867841409694,
      "loss": 2.6495,
      "step": 170
    },
    {
      "epoch": 0.03769218052570673,
      "grad_norm": 1.3437321186065674,
      "learning_rate": 0.0001497797356828194,
      "loss": 2.5975,
      "step": 171
    },
    {
      "epoch": 0.037912602634044196,
      "grad_norm": 1.2361513376235962,
      "learning_rate": 0.00015066079295154186,
      "loss": 2.7234,
      "step": 172
    },
    {
      "epoch": 0.03813302474238166,
      "grad_norm": 1.2328113317489624,
      "learning_rate": 0.00015154185022026433,
      "loss": 2.758,
      "step": 173
    },
    {
      "epoch": 0.038353446850719126,
      "grad_norm": 1.252739667892456,
      "learning_rate": 0.0001524229074889868,
      "loss": 2.7166,
      "step": 174
    },
    {
      "epoch": 0.038573868959056595,
      "grad_norm": 1.1520521640777588,
      "learning_rate": 0.00015330396475770925,
      "loss": 2.6392,
      "step": 175
    },
    {
      "epoch": 0.03879429106739406,
      "grad_norm": 1.1344709396362305,
      "learning_rate": 0.00015418502202643172,
      "loss": 2.7585,
      "step": 176
    },
    {
      "epoch": 0.039014713175731526,
      "grad_norm": 1.2142118215560913,
      "learning_rate": 0.0001550660792951542,
      "loss": 2.7146,
      "step": 177
    },
    {
      "epoch": 0.039235135284068995,
      "grad_norm": 1.1355364322662354,
      "learning_rate": 0.00015594713656387664,
      "loss": 2.8583,
      "step": 178
    },
    {
      "epoch": 0.03945555739240646,
      "grad_norm": 1.1016793251037598,
      "learning_rate": 0.0001568281938325991,
      "loss": 2.6303,
      "step": 179
    },
    {
      "epoch": 0.039675979500743926,
      "grad_norm": 1.1493167877197266,
      "learning_rate": 0.0001577092511013216,
      "loss": 2.6638,
      "step": 180
    },
    {
      "epoch": 0.03989640160908139,
      "grad_norm": 1.2781190872192383,
      "learning_rate": 0.00015859030837004406,
      "loss": 2.6742,
      "step": 181
    },
    {
      "epoch": 0.040116823717418856,
      "grad_norm": 1.1129626035690308,
      "learning_rate": 0.00015947136563876653,
      "loss": 2.5078,
      "step": 182
    },
    {
      "epoch": 0.040337245825756325,
      "grad_norm": 1.082762598991394,
      "learning_rate": 0.000160352422907489,
      "loss": 2.7336,
      "step": 183
    },
    {
      "epoch": 0.04055766793409379,
      "grad_norm": 1.3079731464385986,
      "learning_rate": 0.00016123348017621148,
      "loss": 2.8446,
      "step": 184
    },
    {
      "epoch": 0.040778090042431256,
      "grad_norm": 1.0784714221954346,
      "learning_rate": 0.00016211453744493392,
      "loss": 2.6121,
      "step": 185
    },
    {
      "epoch": 0.040998512150768725,
      "grad_norm": 1.2676712274551392,
      "learning_rate": 0.0001629955947136564,
      "loss": 2.7004,
      "step": 186
    },
    {
      "epoch": 0.04121893425910619,
      "grad_norm": 1.1900205612182617,
      "learning_rate": 0.00016387665198237887,
      "loss": 2.6003,
      "step": 187
    },
    {
      "epoch": 0.041439356367443655,
      "grad_norm": 1.1665024757385254,
      "learning_rate": 0.0001647577092511013,
      "loss": 2.6853,
      "step": 188
    },
    {
      "epoch": 0.041659778475781124,
      "grad_norm": 1.3548634052276611,
      "learning_rate": 0.00016563876651982379,
      "loss": 2.792,
      "step": 189
    },
    {
      "epoch": 0.041880200584118586,
      "grad_norm": 1.311342716217041,
      "learning_rate": 0.00016651982378854626,
      "loss": 2.6698,
      "step": 190
    },
    {
      "epoch": 0.042100622692456055,
      "grad_norm": 1.1759463548660278,
      "learning_rate": 0.00016740088105726873,
      "loss": 2.6868,
      "step": 191
    },
    {
      "epoch": 0.04232104480079352,
      "grad_norm": 1.1529422998428345,
      "learning_rate": 0.0001682819383259912,
      "loss": 2.6682,
      "step": 192
    },
    {
      "epoch": 0.042541466909130986,
      "grad_norm": 1.1619828939437866,
      "learning_rate": 0.00016916299559471368,
      "loss": 2.662,
      "step": 193
    },
    {
      "epoch": 0.042761889017468455,
      "grad_norm": 1.1582976579666138,
      "learning_rate": 0.00017004405286343612,
      "loss": 2.6026,
      "step": 194
    },
    {
      "epoch": 0.042982311125805917,
      "grad_norm": 1.3071789741516113,
      "learning_rate": 0.0001709251101321586,
      "loss": 2.7033,
      "step": 195
    },
    {
      "epoch": 0.043202733234143385,
      "grad_norm": 1.2555397748947144,
      "learning_rate": 0.00017180616740088107,
      "loss": 2.7493,
      "step": 196
    },
    {
      "epoch": 0.043423155342480854,
      "grad_norm": 1.1479676961898804,
      "learning_rate": 0.00017268722466960354,
      "loss": 2.7575,
      "step": 197
    },
    {
      "epoch": 0.043643577450818316,
      "grad_norm": 1.2268725633621216,
      "learning_rate": 0.000173568281938326,
      "loss": 2.6144,
      "step": 198
    },
    {
      "epoch": 0.043863999559155785,
      "grad_norm": 1.0594466924667358,
      "learning_rate": 0.00017444933920704846,
      "loss": 2.6492,
      "step": 199
    },
    {
      "epoch": 0.04408442166749325,
      "grad_norm": 1.2911630868911743,
      "learning_rate": 0.00017533039647577093,
      "loss": 2.6443,
      "step": 200
    },
    {
      "epoch": 0.044304843775830716,
      "grad_norm": 1.043413758277893,
      "learning_rate": 0.0001762114537444934,
      "loss": 2.6823,
      "step": 201
    },
    {
      "epoch": 0.044525265884168184,
      "grad_norm": 1.0424656867980957,
      "learning_rate": 0.00017709251101321588,
      "loss": 2.6377,
      "step": 202
    },
    {
      "epoch": 0.044745687992505646,
      "grad_norm": 1.0896776914596558,
      "learning_rate": 0.00017797356828193835,
      "loss": 2.7139,
      "step": 203
    },
    {
      "epoch": 0.044966110100843115,
      "grad_norm": 1.1621649265289307,
      "learning_rate": 0.0001788546255506608,
      "loss": 2.4821,
      "step": 204
    },
    {
      "epoch": 0.045186532209180584,
      "grad_norm": 1.19429612159729,
      "learning_rate": 0.00017973568281938327,
      "loss": 2.726,
      "step": 205
    },
    {
      "epoch": 0.045406954317518046,
      "grad_norm": 1.1593459844589233,
      "learning_rate": 0.00018061674008810574,
      "loss": 2.6589,
      "step": 206
    },
    {
      "epoch": 0.045627376425855515,
      "grad_norm": 1.2253782749176025,
      "learning_rate": 0.0001814977973568282,
      "loss": 2.7442,
      "step": 207
    },
    {
      "epoch": 0.04584779853419298,
      "grad_norm": 1.1185951232910156,
      "learning_rate": 0.00018237885462555066,
      "loss": 2.6464,
      "step": 208
    },
    {
      "epoch": 0.046068220642530446,
      "grad_norm": 4.0011796951293945,
      "learning_rate": 0.00018325991189427313,
      "loss": 2.699,
      "step": 209
    },
    {
      "epoch": 0.046288642750867914,
      "grad_norm": 1.2698900699615479,
      "learning_rate": 0.0001841409691629956,
      "loss": 2.6563,
      "step": 210
    },
    {
      "epoch": 0.046509064859205376,
      "grad_norm": 1.2476387023925781,
      "learning_rate": 0.00018502202643171805,
      "loss": 2.6132,
      "step": 211
    },
    {
      "epoch": 0.046729486967542845,
      "grad_norm": 1.0719910860061646,
      "learning_rate": 0.00018590308370044055,
      "loss": 2.6265,
      "step": 212
    },
    {
      "epoch": 0.046949909075880314,
      "grad_norm": 1.2692852020263672,
      "learning_rate": 0.00018678414096916303,
      "loss": 2.7656,
      "step": 213
    },
    {
      "epoch": 0.047170331184217776,
      "grad_norm": 1.3062279224395752,
      "learning_rate": 0.00018766519823788547,
      "loss": 2.7643,
      "step": 214
    },
    {
      "epoch": 0.047390753292555245,
      "grad_norm": 1.1375887393951416,
      "learning_rate": 0.00018854625550660794,
      "loss": 2.6984,
      "step": 215
    },
    {
      "epoch": 0.04761117540089271,
      "grad_norm": 1.3240320682525635,
      "learning_rate": 0.00018942731277533042,
      "loss": 2.6187,
      "step": 216
    },
    {
      "epoch": 0.047831597509230175,
      "grad_norm": 1.4199762344360352,
      "learning_rate": 0.00019030837004405286,
      "loss": 2.6808,
      "step": 217
    },
    {
      "epoch": 0.048052019617567644,
      "grad_norm": 1.154629111289978,
      "learning_rate": 0.00019118942731277534,
      "loss": 2.77,
      "step": 218
    },
    {
      "epoch": 0.048272441725905106,
      "grad_norm": 1.1730583906173706,
      "learning_rate": 0.0001920704845814978,
      "loss": 2.5491,
      "step": 219
    },
    {
      "epoch": 0.048492863834242575,
      "grad_norm": 1.207818865776062,
      "learning_rate": 0.00019295154185022025,
      "loss": 2.5568,
      "step": 220
    },
    {
      "epoch": 0.048713285942580044,
      "grad_norm": 1.2437340021133423,
      "learning_rate": 0.00019383259911894273,
      "loss": 2.7564,
      "step": 221
    },
    {
      "epoch": 0.048933708050917506,
      "grad_norm": 1.1629035472869873,
      "learning_rate": 0.0001947136563876652,
      "loss": 2.6209,
      "step": 222
    },
    {
      "epoch": 0.049154130159254975,
      "grad_norm": 1.292086124420166,
      "learning_rate": 0.00019559471365638767,
      "loss": 2.7145,
      "step": 223
    },
    {
      "epoch": 0.049374552267592436,
      "grad_norm": 1.250493049621582,
      "learning_rate": 0.00019647577092511015,
      "loss": 2.7919,
      "step": 224
    },
    {
      "epoch": 0.049594974375929905,
      "grad_norm": 1.1603630781173706,
      "learning_rate": 0.00019735682819383262,
      "loss": 2.6696,
      "step": 225
    },
    {
      "epoch": 0.049815396484267374,
      "grad_norm": 1.223188877105713,
      "learning_rate": 0.0001982378854625551,
      "loss": 2.6402,
      "step": 226
    },
    {
      "epoch": 0.050035818592604836,
      "grad_norm": 1.1897919178009033,
      "learning_rate": 0.00019911894273127754,
      "loss": 2.6532,
      "step": 227
    },
    {
      "epoch": 0.050256240700942305,
      "grad_norm": 1.19523286819458,
      "learning_rate": 0.0002,
      "loss": 2.5438,
      "step": 228
    },
    {
      "epoch": 0.050476662809279774,
      "grad_norm": 1.252063512802124,
      "learning_rate": 0.00019999997343467154,
      "loss": 2.5172,
      "step": 229
    },
    {
      "epoch": 0.050697084917617236,
      "grad_norm": 1.0902361869812012,
      "learning_rate": 0.00019999989373870026,
      "loss": 2.7124,
      "step": 230
    },
    {
      "epoch": 0.050917507025954704,
      "grad_norm": 1.0788612365722656,
      "learning_rate": 0.00019999976091212847,
      "loss": 2.5067,
      "step": 231
    },
    {
      "epoch": 0.051137929134292166,
      "grad_norm": 1.104329228401184,
      "learning_rate": 0.00019999957495502678,
      "loss": 2.5282,
      "step": 232
    },
    {
      "epoch": 0.051358351242629635,
      "grad_norm": 1.0818891525268555,
      "learning_rate": 0.00019999933586749394,
      "loss": 2.6125,
      "step": 233
    },
    {
      "epoch": 0.051578773350967104,
      "grad_norm": 1.2198753356933594,
      "learning_rate": 0.00019999904364965707,
      "loss": 2.7671,
      "step": 234
    },
    {
      "epoch": 0.051799195459304566,
      "grad_norm": 1.0849870443344116,
      "learning_rate": 0.00019999869830167135,
      "loss": 2.6643,
      "step": 235
    },
    {
      "epoch": 0.052019617567642035,
      "grad_norm": 1.1677992343902588,
      "learning_rate": 0.0001999982998237203,
      "loss": 2.6917,
      "step": 236
    },
    {
      "epoch": 0.052240039675979504,
      "grad_norm": 1.0969672203063965,
      "learning_rate": 0.0001999978482160156,
      "loss": 2.5347,
      "step": 237
    },
    {
      "epoch": 0.052460461784316965,
      "grad_norm": 1.174895167350769,
      "learning_rate": 0.00019999734347879723,
      "loss": 2.6167,
      "step": 238
    },
    {
      "epoch": 0.052680883892654434,
      "grad_norm": 1.1888599395751953,
      "learning_rate": 0.00019999678561233335,
      "loss": 2.6404,
      "step": 239
    },
    {
      "epoch": 0.052901306000991896,
      "grad_norm": 1.175835371017456,
      "learning_rate": 0.00019999617461692037,
      "loss": 2.5611,
      "step": 240
    },
    {
      "epoch": 0.053121728109329365,
      "grad_norm": 1.0939195156097412,
      "learning_rate": 0.00019999551049288287,
      "loss": 2.6313,
      "step": 241
    },
    {
      "epoch": 0.053342150217666834,
      "grad_norm": 0.9940528869628906,
      "learning_rate": 0.00019999479324057373,
      "loss": 2.6747,
      "step": 242
    },
    {
      "epoch": 0.053562572326004296,
      "grad_norm": 1.0665769577026367,
      "learning_rate": 0.00019999402286037404,
      "loss": 2.6753,
      "step": 243
    },
    {
      "epoch": 0.053782994434341765,
      "grad_norm": 1.1430526971817017,
      "learning_rate": 0.0001999931993526931,
      "loss": 2.7127,
      "step": 244
    },
    {
      "epoch": 0.05400341654267923,
      "grad_norm": 1.0763020515441895,
      "learning_rate": 0.00019999232271796847,
      "loss": 2.5846,
      "step": 245
    },
    {
      "epoch": 0.054223838651016695,
      "grad_norm": 1.3430004119873047,
      "learning_rate": 0.00019999139295666588,
      "loss": 2.6567,
      "step": 246
    },
    {
      "epoch": 0.054444260759354164,
      "grad_norm": 1.0010523796081543,
      "learning_rate": 0.00019999041006927933,
      "loss": 2.6219,
      "step": 247
    },
    {
      "epoch": 0.054664682867691626,
      "grad_norm": 1.119281530380249,
      "learning_rate": 0.00019998937405633105,
      "loss": 2.6944,
      "step": 248
    },
    {
      "epoch": 0.054885104976029095,
      "grad_norm": 1.130104899406433,
      "learning_rate": 0.0001999882849183714,
      "loss": 2.6659,
      "step": 249
    },
    {
      "epoch": 0.055105527084366564,
      "grad_norm": 1.0189566612243652,
      "learning_rate": 0.00019998714265597917,
      "loss": 2.5891,
      "step": 250
    },
    {
      "epoch": 0.055325949192704026,
      "grad_norm": 1.0385394096374512,
      "learning_rate": 0.00019998594726976116,
      "loss": 2.6022,
      "step": 251
    },
    {
      "epoch": 0.055546371301041494,
      "grad_norm": 1.1705793142318726,
      "learning_rate": 0.0001999846987603525,
      "loss": 2.6534,
      "step": 252
    },
    {
      "epoch": 0.05576679340937896,
      "grad_norm": 1.2315279245376587,
      "learning_rate": 0.0001999833971284166,
      "loss": 2.7018,
      "step": 253
    },
    {
      "epoch": 0.055987215517716425,
      "grad_norm": 1.059363842010498,
      "learning_rate": 0.00019998204237464494,
      "loss": 2.5506,
      "step": 254
    },
    {
      "epoch": 0.056207637626053894,
      "grad_norm": 1.0944340229034424,
      "learning_rate": 0.00019998063449975733,
      "loss": 2.6011,
      "step": 255
    },
    {
      "epoch": 0.05642805973439136,
      "grad_norm": 1.1723021268844604,
      "learning_rate": 0.0001999791735045018,
      "loss": 2.476,
      "step": 256
    },
    {
      "epoch": 0.056648481842728825,
      "grad_norm": 1.185259461402893,
      "learning_rate": 0.0001999776593896546,
      "loss": 2.6986,
      "step": 257
    },
    {
      "epoch": 0.056868903951066294,
      "grad_norm": 1.1675103902816772,
      "learning_rate": 0.00019997609215602019,
      "loss": 2.7242,
      "step": 258
    },
    {
      "epoch": 0.057089326059403755,
      "grad_norm": 1.2605112791061401,
      "learning_rate": 0.0001999744718044312,
      "loss": 2.5872,
      "step": 259
    },
    {
      "epoch": 0.057309748167741224,
      "grad_norm": 1.1281145811080933,
      "learning_rate": 0.00019997279833574859,
      "loss": 2.6156,
      "step": 260
    },
    {
      "epoch": 0.05753017027607869,
      "grad_norm": 0.9942386746406555,
      "learning_rate": 0.00019997107175086142,
      "loss": 2.5428,
      "step": 261
    },
    {
      "epoch": 0.057750592384416155,
      "grad_norm": 1.1306432485580444,
      "learning_rate": 0.0001999692920506871,
      "loss": 2.5919,
      "step": 262
    },
    {
      "epoch": 0.057971014492753624,
      "grad_norm": 1.1059356927871704,
      "learning_rate": 0.0001999674592361712,
      "loss": 2.6328,
      "step": 263
    },
    {
      "epoch": 0.05819143660109109,
      "grad_norm": 1.2119783163070679,
      "learning_rate": 0.00019996557330828745,
      "loss": 2.6673,
      "step": 264
    },
    {
      "epoch": 0.058411858709428555,
      "grad_norm": 1.2209452390670776,
      "learning_rate": 0.0001999636342680379,
      "loss": 2.656,
      "step": 265
    },
    {
      "epoch": 0.05863228081776602,
      "grad_norm": 1.0552315711975098,
      "learning_rate": 0.00019996164211645275,
      "loss": 2.5778,
      "step": 266
    },
    {
      "epoch": 0.058852702926103485,
      "grad_norm": 1.0153791904449463,
      "learning_rate": 0.00019995959685459048,
      "loss": 2.5308,
      "step": 267
    },
    {
      "epoch": 0.059073125034440954,
      "grad_norm": 1.0505188703536987,
      "learning_rate": 0.0001999574984835377,
      "loss": 2.5901,
      "step": 268
    },
    {
      "epoch": 0.05929354714277842,
      "grad_norm": 1.0802797079086304,
      "learning_rate": 0.0001999553470044093,
      "loss": 2.6682,
      "step": 269
    },
    {
      "epoch": 0.059513969251115885,
      "grad_norm": 1.04459547996521,
      "learning_rate": 0.00019995314241834838,
      "loss": 2.5955,
      "step": 270
    },
    {
      "epoch": 0.059734391359453354,
      "grad_norm": 1.1181187629699707,
      "learning_rate": 0.0001999508847265263,
      "loss": 2.5759,
      "step": 271
    },
    {
      "epoch": 0.05995481346779082,
      "grad_norm": 1.0479682683944702,
      "learning_rate": 0.00019994857393014254,
      "loss": 2.4875,
      "step": 272
    },
    {
      "epoch": 0.060175235576128284,
      "grad_norm": 1.216740369796753,
      "learning_rate": 0.00019994621003042483,
      "loss": 2.4871,
      "step": 273
    },
    {
      "epoch": 0.06039565768446575,
      "grad_norm": 1.1287986040115356,
      "learning_rate": 0.00019994379302862915,
      "loss": 2.647,
      "step": 274
    },
    {
      "epoch": 0.060616079792803215,
      "grad_norm": 1.0109150409698486,
      "learning_rate": 0.00019994132292603967,
      "loss": 2.5068,
      "step": 275
    },
    {
      "epoch": 0.060836501901140684,
      "grad_norm": 1.0635604858398438,
      "learning_rate": 0.00019993879972396875,
      "loss": 2.5414,
      "step": 276
    },
    {
      "epoch": 0.06105692400947815,
      "grad_norm": 1.320532202720642,
      "learning_rate": 0.00019993622342375698,
      "loss": 2.5429,
      "step": 277
    },
    {
      "epoch": 0.061277346117815615,
      "grad_norm": 1.0382379293441772,
      "learning_rate": 0.00019993359402677323,
      "loss": 2.5078,
      "step": 278
    },
    {
      "epoch": 0.061497768226153084,
      "grad_norm": 1.015177845954895,
      "learning_rate": 0.00019993091153441443,
      "loss": 2.6989,
      "step": 279
    },
    {
      "epoch": 0.06171819033449055,
      "grad_norm": 1.0025787353515625,
      "learning_rate": 0.00019992817594810587,
      "loss": 2.5354,
      "step": 280
    },
    {
      "epoch": 0.061938612442828014,
      "grad_norm": 1.073781132698059,
      "learning_rate": 0.00019992538726930096,
      "loss": 2.5468,
      "step": 281
    },
    {
      "epoch": 0.06215903455116548,
      "grad_norm": 1.0404961109161377,
      "learning_rate": 0.0001999225454994813,
      "loss": 2.5974,
      "step": 282
    },
    {
      "epoch": 0.062379456659502945,
      "grad_norm": 1.0296337604522705,
      "learning_rate": 0.00019991965064015683,
      "loss": 2.4955,
      "step": 283
    },
    {
      "epoch": 0.06259987876784041,
      "grad_norm": 0.9854639768600464,
      "learning_rate": 0.00019991670269286553,
      "loss": 2.5476,
      "step": 284
    },
    {
      "epoch": 0.06282030087617788,
      "grad_norm": 1.0850048065185547,
      "learning_rate": 0.0001999137016591737,
      "loss": 2.5925,
      "step": 285
    },
    {
      "epoch": 0.06304072298451535,
      "grad_norm": 1.0606012344360352,
      "learning_rate": 0.00019991064754067584,
      "loss": 2.5838,
      "step": 286
    },
    {
      "epoch": 0.06326114509285281,
      "grad_norm": 0.954569935798645,
      "learning_rate": 0.00019990754033899456,
      "loss": 2.6847,
      "step": 287
    },
    {
      "epoch": 0.06348156720119028,
      "grad_norm": 1.0242857933044434,
      "learning_rate": 0.00019990438005578075,
      "loss": 2.5434,
      "step": 288
    },
    {
      "epoch": 0.06370198930952775,
      "grad_norm": 0.99779212474823,
      "learning_rate": 0.00019990116669271353,
      "loss": 2.6176,
      "step": 289
    },
    {
      "epoch": 0.06392241141786521,
      "grad_norm": 1.0726990699768066,
      "learning_rate": 0.00019989790025150015,
      "loss": 2.627,
      "step": 290
    },
    {
      "epoch": 0.06414283352620267,
      "grad_norm": 1.1521965265274048,
      "learning_rate": 0.00019989458073387608,
      "loss": 2.569,
      "step": 291
    },
    {
      "epoch": 0.06436325563454015,
      "grad_norm": 1.0927982330322266,
      "learning_rate": 0.00019989120814160503,
      "loss": 2.5932,
      "step": 292
    },
    {
      "epoch": 0.06458367774287761,
      "grad_norm": 0.9862792491912842,
      "learning_rate": 0.00019988778247647887,
      "loss": 2.5475,
      "step": 293
    },
    {
      "epoch": 0.06480409985121507,
      "grad_norm": 1.0542311668395996,
      "learning_rate": 0.00019988430374031765,
      "loss": 2.5474,
      "step": 294
    },
    {
      "epoch": 0.06502452195955254,
      "grad_norm": 0.9558664560317993,
      "learning_rate": 0.0001998807719349697,
      "loss": 2.5697,
      "step": 295
    },
    {
      "epoch": 0.06524494406789001,
      "grad_norm": 1.0041104555130005,
      "learning_rate": 0.00019987718706231144,
      "loss": 2.484,
      "step": 296
    },
    {
      "epoch": 0.06546536617622747,
      "grad_norm": 1.0605469942092896,
      "learning_rate": 0.00019987354912424755,
      "loss": 2.4055,
      "step": 297
    },
    {
      "epoch": 0.06568578828456494,
      "grad_norm": 0.9525481462478638,
      "learning_rate": 0.00019986985812271092,
      "loss": 2.6133,
      "step": 298
    },
    {
      "epoch": 0.06590621039290241,
      "grad_norm": 1.0610707998275757,
      "learning_rate": 0.0001998661140596626,
      "loss": 2.604,
      "step": 299
    },
    {
      "epoch": 0.06612663250123987,
      "grad_norm": 0.9677058458328247,
      "learning_rate": 0.0001998623169370918,
      "loss": 2.5478,
      "step": 300
    },
    {
      "epoch": 0.06612663250123987,
      "eval_bpb": 1.3343920602548716,
      "eval_loss": 2.570512533187866,
      "eval_runtime": 184.4149,
      "eval_samples_per_second": 62.142,
      "eval_steps_per_second": 7.771,
      "step": 300
    },
    {
      "epoch": 0.06634705460957734,
      "grad_norm": 1.0058132410049438,
      "learning_rate": 0.00019985846675701598,
      "loss": 2.607,
      "step": 301
    },
    {
      "epoch": 0.06656747671791481,
      "grad_norm": 0.9704034328460693,
      "learning_rate": 0.00019985456352148074,
      "loss": 2.5769,
      "step": 302
    },
    {
      "epoch": 0.06678789882625227,
      "grad_norm": 0.9321823120117188,
      "learning_rate": 0.00019985060723255995,
      "loss": 2.5252,
      "step": 303
    },
    {
      "epoch": 0.06700832093458974,
      "grad_norm": 1.0902271270751953,
      "learning_rate": 0.00019984659789235555,
      "loss": 2.5014,
      "step": 304
    },
    {
      "epoch": 0.06722874304292721,
      "grad_norm": 1.1287496089935303,
      "learning_rate": 0.00019984253550299775,
      "loss": 2.6157,
      "step": 305
    },
    {
      "epoch": 0.06744916515126467,
      "grad_norm": 1.072405457496643,
      "learning_rate": 0.00019983842006664494,
      "loss": 2.6272,
      "step": 306
    },
    {
      "epoch": 0.06766958725960213,
      "grad_norm": 0.9595354795455933,
      "learning_rate": 0.0001998342515854837,
      "loss": 2.4719,
      "step": 307
    },
    {
      "epoch": 0.06789000936793961,
      "grad_norm": 1.0227371454238892,
      "learning_rate": 0.0001998300300617287,
      "loss": 2.5033,
      "step": 308
    },
    {
      "epoch": 0.06811043147627707,
      "grad_norm": 1.0306305885314941,
      "learning_rate": 0.00019982575549762292,
      "loss": 2.5043,
      "step": 309
    },
    {
      "epoch": 0.06833085358461453,
      "grad_norm": 0.9609490633010864,
      "learning_rate": 0.00019982142789543746,
      "loss": 2.5456,
      "step": 310
    },
    {
      "epoch": 0.068551275692952,
      "grad_norm": 1.0611956119537354,
      "learning_rate": 0.00019981704725747156,
      "loss": 2.4204,
      "step": 311
    },
    {
      "epoch": 0.06877169780128947,
      "grad_norm": 1.056251049041748,
      "learning_rate": 0.00019981261358605274,
      "loss": 2.5607,
      "step": 312
    },
    {
      "epoch": 0.06899211990962693,
      "grad_norm": 0.981169581413269,
      "learning_rate": 0.00019980812688353658,
      "loss": 2.5149,
      "step": 313
    },
    {
      "epoch": 0.0692125420179644,
      "grad_norm": 0.9895249009132385,
      "learning_rate": 0.00019980358715230697,
      "loss": 2.6091,
      "step": 314
    },
    {
      "epoch": 0.06943296412630187,
      "grad_norm": 1.0336586236953735,
      "learning_rate": 0.0001997989943947758,
      "loss": 2.4531,
      "step": 315
    },
    {
      "epoch": 0.06965338623463933,
      "grad_norm": 1.1045012474060059,
      "learning_rate": 0.00019979434861338332,
      "loss": 2.5853,
      "step": 316
    },
    {
      "epoch": 0.0698738083429768,
      "grad_norm": 1.0102009773254395,
      "learning_rate": 0.00019978964981059782,
      "loss": 2.51,
      "step": 317
    },
    {
      "epoch": 0.07009423045131427,
      "grad_norm": 0.941723644733429,
      "learning_rate": 0.00019978489798891584,
      "loss": 2.614,
      "step": 318
    },
    {
      "epoch": 0.07031465255965173,
      "grad_norm": 0.9607812166213989,
      "learning_rate": 0.000199780093150862,
      "loss": 2.7059,
      "step": 319
    },
    {
      "epoch": 0.0705350746679892,
      "grad_norm": 1.0180697441101074,
      "learning_rate": 0.0001997752352989892,
      "loss": 2.5576,
      "step": 320
    },
    {
      "epoch": 0.07075549677632667,
      "grad_norm": 0.9359689950942993,
      "learning_rate": 0.0001997703244358784,
      "loss": 2.4893,
      "step": 321
    },
    {
      "epoch": 0.07097591888466413,
      "grad_norm": 0.9270784258842468,
      "learning_rate": 0.0001997653605641388,
      "loss": 2.5419,
      "step": 322
    },
    {
      "epoch": 0.0711963409930016,
      "grad_norm": 0.9588329195976257,
      "learning_rate": 0.00019976034368640773,
      "loss": 2.6452,
      "step": 323
    },
    {
      "epoch": 0.07141676310133907,
      "grad_norm": 0.9756755232810974,
      "learning_rate": 0.0001997552738053507,
      "loss": 2.623,
      "step": 324
    },
    {
      "epoch": 0.07163718520967653,
      "grad_norm": 0.9358682632446289,
      "learning_rate": 0.00019975015092366135,
      "loss": 2.4797,
      "step": 325
    },
    {
      "epoch": 0.071857607318014,
      "grad_norm": 1.0220993757247925,
      "learning_rate": 0.00019974497504406154,
      "loss": 2.6261,
      "step": 326
    },
    {
      "epoch": 0.07207802942635147,
      "grad_norm": 1.017448902130127,
      "learning_rate": 0.0001997397461693012,
      "loss": 2.4732,
      "step": 327
    },
    {
      "epoch": 0.07229845153468893,
      "grad_norm": 0.8911523818969727,
      "learning_rate": 0.0001997344643021585,
      "loss": 2.5708,
      "step": 328
    },
    {
      "epoch": 0.0725188736430264,
      "grad_norm": 0.9566131234169006,
      "learning_rate": 0.00019972912944543972,
      "loss": 2.476,
      "step": 329
    },
    {
      "epoch": 0.07273929575136386,
      "grad_norm": 1.1435223817825317,
      "learning_rate": 0.00019972374160197927,
      "loss": 2.6076,
      "step": 330
    },
    {
      "epoch": 0.07295971785970133,
      "grad_norm": 1.0611422061920166,
      "learning_rate": 0.00019971830077463983,
      "loss": 2.5051,
      "step": 331
    },
    {
      "epoch": 0.0731801399680388,
      "grad_norm": 0.9957279562950134,
      "learning_rate": 0.00019971280696631208,
      "loss": 2.464,
      "step": 332
    },
    {
      "epoch": 0.07340056207637626,
      "grad_norm": 1.1713024377822876,
      "learning_rate": 0.0001997072601799149,
      "loss": 2.5517,
      "step": 333
    },
    {
      "epoch": 0.07362098418471373,
      "grad_norm": 1.060089349746704,
      "learning_rate": 0.0001997016604183954,
      "loss": 2.5313,
      "step": 334
    },
    {
      "epoch": 0.07384140629305119,
      "grad_norm": 1.0913212299346924,
      "learning_rate": 0.00019969600768472874,
      "loss": 2.4531,
      "step": 335
    },
    {
      "epoch": 0.07406182840138865,
      "grad_norm": 1.0202929973602295,
      "learning_rate": 0.00019969030198191824,
      "loss": 2.5567,
      "step": 336
    },
    {
      "epoch": 0.07428225050972613,
      "grad_norm": 1.0500272512435913,
      "learning_rate": 0.00019968454331299542,
      "loss": 2.4986,
      "step": 337
    },
    {
      "epoch": 0.07450267261806359,
      "grad_norm": 0.9147382378578186,
      "learning_rate": 0.00019967873168101984,
      "loss": 2.5614,
      "step": 338
    },
    {
      "epoch": 0.07472309472640105,
      "grad_norm": 1.0528326034545898,
      "learning_rate": 0.0001996728670890793,
      "loss": 2.5063,
      "step": 339
    },
    {
      "epoch": 0.07494351683473853,
      "grad_norm": 0.9187337160110474,
      "learning_rate": 0.0001996669495402897,
      "loss": 2.4702,
      "step": 340
    },
    {
      "epoch": 0.07516393894307599,
      "grad_norm": 0.9364808201789856,
      "learning_rate": 0.000199660979037795,
      "loss": 2.5439,
      "step": 341
    },
    {
      "epoch": 0.07538436105141345,
      "grad_norm": 0.8891624808311462,
      "learning_rate": 0.00019965495558476747,
      "loss": 2.5229,
      "step": 342
    },
    {
      "epoch": 0.07560478315975093,
      "grad_norm": 0.9202736020088196,
      "learning_rate": 0.00019964887918440732,
      "loss": 2.5506,
      "step": 343
    },
    {
      "epoch": 0.07582520526808839,
      "grad_norm": 0.8543094992637634,
      "learning_rate": 0.00019964274983994308,
      "loss": 2.5203,
      "step": 344
    },
    {
      "epoch": 0.07604562737642585,
      "grad_norm": 0.9874389171600342,
      "learning_rate": 0.00019963656755463122,
      "loss": 2.5391,
      "step": 345
    },
    {
      "epoch": 0.07626604948476332,
      "grad_norm": 0.9775609374046326,
      "learning_rate": 0.00019963033233175645,
      "loss": 2.5101,
      "step": 346
    },
    {
      "epoch": 0.07648647159310079,
      "grad_norm": 1.0834159851074219,
      "learning_rate": 0.00019962404417463164,
      "loss": 2.5539,
      "step": 347
    },
    {
      "epoch": 0.07670689370143825,
      "grad_norm": 1.0348787307739258,
      "learning_rate": 0.00019961770308659767,
      "loss": 2.4839,
      "step": 348
    },
    {
      "epoch": 0.07692731580977571,
      "grad_norm": 0.9452861547470093,
      "learning_rate": 0.0001996113090710236,
      "loss": 2.4587,
      "step": 349
    },
    {
      "epoch": 0.07714773791811319,
      "grad_norm": 1.0046755075454712,
      "learning_rate": 0.00019960486213130667,
      "loss": 2.4463,
      "step": 350
    },
    {
      "epoch": 0.07736816002645065,
      "grad_norm": 0.9473358988761902,
      "learning_rate": 0.00019959836227087215,
      "loss": 2.4686,
      "step": 351
    },
    {
      "epoch": 0.07758858213478811,
      "grad_norm": 1.0065165758132935,
      "learning_rate": 0.00019959180949317341,
      "loss": 2.5795,
      "step": 352
    },
    {
      "epoch": 0.07780900424312559,
      "grad_norm": 0.8850712180137634,
      "learning_rate": 0.00019958520380169206,
      "loss": 2.5506,
      "step": 353
    },
    {
      "epoch": 0.07802942635146305,
      "grad_norm": 0.9753082990646362,
      "learning_rate": 0.0001995785451999377,
      "loss": 2.5102,
      "step": 354
    },
    {
      "epoch": 0.07824984845980051,
      "grad_norm": 1.0976569652557373,
      "learning_rate": 0.0001995718336914481,
      "loss": 2.6491,
      "step": 355
    },
    {
      "epoch": 0.07847027056813799,
      "grad_norm": 0.9241803884506226,
      "learning_rate": 0.00019956506927978912,
      "loss": 2.5097,
      "step": 356
    },
    {
      "epoch": 0.07869069267647545,
      "grad_norm": 0.9074403643608093,
      "learning_rate": 0.00019955825196855476,
      "loss": 2.5131,
      "step": 357
    },
    {
      "epoch": 0.07891111478481291,
      "grad_norm": 0.94773268699646,
      "learning_rate": 0.0001995513817613671,
      "loss": 2.556,
      "step": 358
    },
    {
      "epoch": 0.07913153689315039,
      "grad_norm": 0.9532046318054199,
      "learning_rate": 0.0001995444586618763,
      "loss": 2.6529,
      "step": 359
    },
    {
      "epoch": 0.07935195900148785,
      "grad_norm": 0.9709388613700867,
      "learning_rate": 0.00019953748267376064,
      "loss": 2.4793,
      "step": 360
    },
    {
      "epoch": 0.07957238110982531,
      "grad_norm": 1.0065760612487793,
      "learning_rate": 0.00019953045380072658,
      "loss": 2.5533,
      "step": 361
    },
    {
      "epoch": 0.07979280321816277,
      "grad_norm": 1.0102945566177368,
      "learning_rate": 0.00019952337204650852,
      "loss": 2.5337,
      "step": 362
    },
    {
      "epoch": 0.08001322532650025,
      "grad_norm": 1.0382335186004639,
      "learning_rate": 0.00019951623741486913,
      "loss": 2.4898,
      "step": 363
    },
    {
      "epoch": 0.08023364743483771,
      "grad_norm": 0.9390854835510254,
      "learning_rate": 0.000199509049909599,
      "loss": 2.5256,
      "step": 364
    },
    {
      "epoch": 0.08045406954317517,
      "grad_norm": 0.934540867805481,
      "learning_rate": 0.00019950180953451696,
      "loss": 2.5221,
      "step": 365
    },
    {
      "epoch": 0.08067449165151265,
      "grad_norm": 0.9361838698387146,
      "learning_rate": 0.00019949451629346982,
      "loss": 2.5743,
      "step": 366
    },
    {
      "epoch": 0.08089491375985011,
      "grad_norm": 0.9200062155723572,
      "learning_rate": 0.00019948717019033257,
      "loss": 2.4862,
      "step": 367
    },
    {
      "epoch": 0.08111533586818757,
      "grad_norm": 0.9814015030860901,
      "learning_rate": 0.00019947977122900822,
      "loss": 2.4724,
      "step": 368
    },
    {
      "epoch": 0.08133575797652505,
      "grad_norm": 0.99007648229599,
      "learning_rate": 0.0001994723194134279,
      "loss": 2.7644,
      "step": 369
    },
    {
      "epoch": 0.08155618008486251,
      "grad_norm": 1.005972146987915,
      "learning_rate": 0.00019946481474755078,
      "loss": 2.4657,
      "step": 370
    },
    {
      "epoch": 0.08177660219319997,
      "grad_norm": 0.944976806640625,
      "learning_rate": 0.00019945725723536417,
      "loss": 2.6123,
      "step": 371
    },
    {
      "epoch": 0.08199702430153745,
      "grad_norm": 0.9215418100357056,
      "learning_rate": 0.00019944964688088342,
      "loss": 2.4531,
      "step": 372
    },
    {
      "epoch": 0.08221744640987491,
      "grad_norm": 0.9512657523155212,
      "learning_rate": 0.00019944198368815195,
      "loss": 2.5679,
      "step": 373
    },
    {
      "epoch": 0.08243786851821237,
      "grad_norm": 0.9463131427764893,
      "learning_rate": 0.0001994342676612413,
      "loss": 2.4607,
      "step": 374
    },
    {
      "epoch": 0.08265829062654985,
      "grad_norm": 0.887722373008728,
      "learning_rate": 0.00019942649880425097,
      "loss": 2.5111,
      "step": 375
    },
    {
      "epoch": 0.08287871273488731,
      "grad_norm": 0.9085755944252014,
      "learning_rate": 0.00019941867712130868,
      "loss": 2.6053,
      "step": 376
    },
    {
      "epoch": 0.08309913484322477,
      "grad_norm": 0.9946001768112183,
      "learning_rate": 0.0001994108026165701,
      "loss": 2.5034,
      "step": 377
    },
    {
      "epoch": 0.08331955695156225,
      "grad_norm": 0.9627758860588074,
      "learning_rate": 0.00019940287529421902,
      "loss": 2.4026,
      "step": 378
    },
    {
      "epoch": 0.08353997905989971,
      "grad_norm": 1.0222094058990479,
      "learning_rate": 0.0001993948951584673,
      "loss": 2.5467,
      "step": 379
    },
    {
      "epoch": 0.08376040116823717,
      "grad_norm": 0.9453365802764893,
      "learning_rate": 0.0001993868622135548,
      "loss": 2.516,
      "step": 380
    },
    {
      "epoch": 0.08398082327657463,
      "grad_norm": 0.9918437004089355,
      "learning_rate": 0.00019937877646374947,
      "loss": 2.5218,
      "step": 381
    },
    {
      "epoch": 0.08420124538491211,
      "grad_norm": 1.0235252380371094,
      "learning_rate": 0.00019937063791334735,
      "loss": 2.5604,
      "step": 382
    },
    {
      "epoch": 0.08442166749324957,
      "grad_norm": 0.9259534478187561,
      "learning_rate": 0.00019936244656667252,
      "loss": 2.5545,
      "step": 383
    },
    {
      "epoch": 0.08464208960158703,
      "grad_norm": 0.9255642890930176,
      "learning_rate": 0.0001993542024280771,
      "loss": 2.4851,
      "step": 384
    },
    {
      "epoch": 0.08486251170992451,
      "grad_norm": 1.0333445072174072,
      "learning_rate": 0.0001993459055019412,
      "loss": 2.4634,
      "step": 385
    },
    {
      "epoch": 0.08508293381826197,
      "grad_norm": 0.9594116806983948,
      "learning_rate": 0.00019933755579267307,
      "loss": 2.4288,
      "step": 386
    },
    {
      "epoch": 0.08530335592659943,
      "grad_norm": 0.8851143717765808,
      "learning_rate": 0.00019932915330470895,
      "loss": 2.4069,
      "step": 387
    },
    {
      "epoch": 0.08552377803493691,
      "grad_norm": 0.8739263415336609,
      "learning_rate": 0.00019932069804251312,
      "loss": 2.4217,
      "step": 388
    },
    {
      "epoch": 0.08574420014327437,
      "grad_norm": 1.1335114240646362,
      "learning_rate": 0.00019931219001057798,
      "loss": 2.4369,
      "step": 389
    },
    {
      "epoch": 0.08596462225161183,
      "grad_norm": 0.9614477753639221,
      "learning_rate": 0.00019930362921342384,
      "loss": 2.5127,
      "step": 390
    },
    {
      "epoch": 0.08618504435994931,
      "grad_norm": 0.9562881588935852,
      "learning_rate": 0.00019929501565559913,
      "loss": 2.5261,
      "step": 391
    },
    {
      "epoch": 0.08640546646828677,
      "grad_norm": 1.008569359779358,
      "learning_rate": 0.00019928634934168033,
      "loss": 2.4217,
      "step": 392
    },
    {
      "epoch": 0.08662588857662423,
      "grad_norm": 1.0147508382797241,
      "learning_rate": 0.00019927763027627183,
      "loss": 2.5248,
      "step": 393
    },
    {
      "epoch": 0.08684631068496171,
      "grad_norm": 0.9009134769439697,
      "learning_rate": 0.0001992688584640062,
      "loss": 2.4557,
      "step": 394
    },
    {
      "epoch": 0.08706673279329917,
      "grad_norm": 0.8987658619880676,
      "learning_rate": 0.00019926003390954394,
      "loss": 2.3934,
      "step": 395
    },
    {
      "epoch": 0.08728715490163663,
      "grad_norm": 0.8982332348823547,
      "learning_rate": 0.00019925115661757352,
      "loss": 2.4488,
      "step": 396
    },
    {
      "epoch": 0.0875075770099741,
      "grad_norm": 0.8786985874176025,
      "learning_rate": 0.00019924222659281163,
      "loss": 2.3843,
      "step": 397
    },
    {
      "epoch": 0.08772799911831157,
      "grad_norm": 0.9246736764907837,
      "learning_rate": 0.00019923324384000276,
      "loss": 2.4293,
      "step": 398
    },
    {
      "epoch": 0.08794842122664903,
      "grad_norm": 1.0569993257522583,
      "learning_rate": 0.00019922420836391953,
      "loss": 2.4942,
      "step": 399
    },
    {
      "epoch": 0.0881688433349865,
      "grad_norm": 0.9720404148101807,
      "learning_rate": 0.00019921512016936256,
      "loss": 2.4714,
      "step": 400
    },
    {
      "epoch": 0.08838926544332397,
      "grad_norm": 0.8854937553405762,
      "learning_rate": 0.00019920597926116046,
      "loss": 2.4273,
      "step": 401
    },
    {
      "epoch": 0.08860968755166143,
      "grad_norm": 0.9463419318199158,
      "learning_rate": 0.00019919678564416985,
      "loss": 2.4793,
      "step": 402
    },
    {
      "epoch": 0.0888301096599989,
      "grad_norm": 1.0318056344985962,
      "learning_rate": 0.00019918753932327536,
      "loss": 2.6117,
      "step": 403
    },
    {
      "epoch": 0.08905053176833637,
      "grad_norm": 0.9207020998001099,
      "learning_rate": 0.00019917824030338962,
      "loss": 2.5118,
      "step": 404
    },
    {
      "epoch": 0.08927095387667383,
      "grad_norm": 0.9606727957725525,
      "learning_rate": 0.0001991688885894533,
      "loss": 2.3812,
      "step": 405
    },
    {
      "epoch": 0.08949137598501129,
      "grad_norm": 0.8585346341133118,
      "learning_rate": 0.00019915948418643495,
      "loss": 2.4597,
      "step": 406
    },
    {
      "epoch": 0.08971179809334877,
      "grad_norm": 1.050809383392334,
      "learning_rate": 0.00019915002709933123,
      "loss": 2.4818,
      "step": 407
    },
    {
      "epoch": 0.08993222020168623,
      "grad_norm": 0.9610028266906738,
      "learning_rate": 0.00019914051733316678,
      "loss": 2.5756,
      "step": 408
    },
    {
      "epoch": 0.09015264231002369,
      "grad_norm": 0.9628870487213135,
      "learning_rate": 0.00019913095489299416,
      "loss": 2.4974,
      "step": 409
    },
    {
      "epoch": 0.09037306441836117,
      "grad_norm": 0.9134366512298584,
      "learning_rate": 0.00019912133978389397,
      "loss": 2.5054,
      "step": 410
    },
    {
      "epoch": 0.09059348652669863,
      "grad_norm": 0.9124761819839478,
      "learning_rate": 0.00019911167201097478,
      "loss": 2.5169,
      "step": 411
    },
    {
      "epoch": 0.09081390863503609,
      "grad_norm": 1.0526609420776367,
      "learning_rate": 0.00019910195157937316,
      "loss": 2.4767,
      "step": 412
    },
    {
      "epoch": 0.09103433074337355,
      "grad_norm": 0.9776419997215271,
      "learning_rate": 0.0001990921784942536,
      "loss": 2.44,
      "step": 413
    },
    {
      "epoch": 0.09125475285171103,
      "grad_norm": 0.9121394753456116,
      "learning_rate": 0.00019908235276080868,
      "loss": 2.535,
      "step": 414
    },
    {
      "epoch": 0.09147517496004849,
      "grad_norm": 0.9045057892799377,
      "learning_rate": 0.00019907247438425877,
      "loss": 2.418,
      "step": 415
    },
    {
      "epoch": 0.09169559706838595,
      "grad_norm": 0.9509839415550232,
      "learning_rate": 0.0001990625433698524,
      "loss": 2.5261,
      "step": 416
    },
    {
      "epoch": 0.09191601917672343,
      "grad_norm": 0.9806126952171326,
      "learning_rate": 0.00019905255972286597,
      "loss": 2.4228,
      "step": 417
    },
    {
      "epoch": 0.09213644128506089,
      "grad_norm": 0.9475283622741699,
      "learning_rate": 0.00019904252344860382,
      "loss": 2.5327,
      "step": 418
    },
    {
      "epoch": 0.09235686339339835,
      "grad_norm": 0.9911667108535767,
      "learning_rate": 0.0001990324345523983,
      "loss": 2.4966,
      "step": 419
    },
    {
      "epoch": 0.09257728550173583,
      "grad_norm": 1.023637056350708,
      "learning_rate": 0.00019902229303960973,
      "loss": 2.409,
      "step": 420
    },
    {
      "epoch": 0.09279770761007329,
      "grad_norm": 1.0310337543487549,
      "learning_rate": 0.00019901209891562633,
      "loss": 2.4271,
      "step": 421
    },
    {
      "epoch": 0.09301812971841075,
      "grad_norm": 0.9007945656776428,
      "learning_rate": 0.00019900185218586436,
      "loss": 2.4942,
      "step": 422
    },
    {
      "epoch": 0.09323855182674823,
      "grad_norm": 0.9427668452262878,
      "learning_rate": 0.00019899155285576793,
      "loss": 2.5288,
      "step": 423
    },
    {
      "epoch": 0.09345897393508569,
      "grad_norm": 0.8711615800857544,
      "learning_rate": 0.00019898120093080914,
      "loss": 2.444,
      "step": 424
    },
    {
      "epoch": 0.09367939604342315,
      "grad_norm": 0.9296178221702576,
      "learning_rate": 0.00019897079641648804,
      "loss": 2.6144,
      "step": 425
    },
    {
      "epoch": 0.09389981815176063,
      "grad_norm": 0.9377577304840088,
      "learning_rate": 0.00019896033931833263,
      "loss": 2.3652,
      "step": 426
    },
    {
      "epoch": 0.09412024026009809,
      "grad_norm": 0.913809061050415,
      "learning_rate": 0.00019894982964189882,
      "loss": 2.5276,
      "step": 427
    },
    {
      "epoch": 0.09434066236843555,
      "grad_norm": 0.8244354724884033,
      "learning_rate": 0.0001989392673927705,
      "loss": 2.453,
      "step": 428
    },
    {
      "epoch": 0.09456108447677301,
      "grad_norm": 1.0042369365692139,
      "learning_rate": 0.00019892865257655943,
      "loss": 2.4314,
      "step": 429
    },
    {
      "epoch": 0.09478150658511049,
      "grad_norm": 0.9384554028511047,
      "learning_rate": 0.00019891798519890537,
      "loss": 2.5096,
      "step": 430
    },
    {
      "epoch": 0.09500192869344795,
      "grad_norm": 0.8491098880767822,
      "learning_rate": 0.00019890726526547588,
      "loss": 2.4726,
      "step": 431
    },
    {
      "epoch": 0.09522235080178541,
      "grad_norm": 0.8897377252578735,
      "learning_rate": 0.00019889649278196665,
      "loss": 2.4726,
      "step": 432
    },
    {
      "epoch": 0.09544277291012289,
      "grad_norm": 0.8562506437301636,
      "learning_rate": 0.00019888566775410105,
      "loss": 2.4173,
      "step": 433
    },
    {
      "epoch": 0.09566319501846035,
      "grad_norm": 0.9565961956977844,
      "learning_rate": 0.0001988747901876306,
      "loss": 2.4803,
      "step": 434
    },
    {
      "epoch": 0.09588361712679781,
      "grad_norm": 0.8897228240966797,
      "learning_rate": 0.00019886386008833456,
      "loss": 2.6022,
      "step": 435
    },
    {
      "epoch": 0.09610403923513529,
      "grad_norm": 0.9127510190010071,
      "learning_rate": 0.00019885287746202018,
      "loss": 2.4575,
      "step": 436
    },
    {
      "epoch": 0.09632446134347275,
      "grad_norm": 0.855552077293396,
      "learning_rate": 0.00019884184231452263,
      "loss": 2.4731,
      "step": 437
    },
    {
      "epoch": 0.09654488345181021,
      "grad_norm": 1.0619487762451172,
      "learning_rate": 0.0001988307546517049,
      "loss": 2.5555,
      "step": 438
    },
    {
      "epoch": 0.09676530556014769,
      "grad_norm": 0.9608181715011597,
      "learning_rate": 0.00019881961447945794,
      "loss": 2.4719,
      "step": 439
    },
    {
      "epoch": 0.09698572766848515,
      "grad_norm": 0.8941650986671448,
      "learning_rate": 0.00019880842180370066,
      "loss": 2.4814,
      "step": 440
    },
    {
      "epoch": 0.09720614977682261,
      "grad_norm": 0.8752286434173584,
      "learning_rate": 0.00019879717663037974,
      "loss": 2.365,
      "step": 441
    },
    {
      "epoch": 0.09742657188516009,
      "grad_norm": 0.8613802194595337,
      "learning_rate": 0.00019878587896546984,
      "loss": 2.4584,
      "step": 442
    },
    {
      "epoch": 0.09764699399349755,
      "grad_norm": 0.9235289096832275,
      "learning_rate": 0.00019877452881497348,
      "loss": 2.4407,
      "step": 443
    },
    {
      "epoch": 0.09786741610183501,
      "grad_norm": 0.8464532494544983,
      "learning_rate": 0.00019876312618492107,
      "loss": 2.583,
      "step": 444
    },
    {
      "epoch": 0.09808783821017249,
      "grad_norm": 0.8383076190948486,
      "learning_rate": 0.0001987516710813709,
      "loss": 2.3862,
      "step": 445
    },
    {
      "epoch": 0.09830826031850995,
      "grad_norm": 0.8581581115722656,
      "learning_rate": 0.00019874016351040914,
      "loss": 2.3272,
      "step": 446
    },
    {
      "epoch": 0.09852868242684741,
      "grad_norm": 0.8450644016265869,
      "learning_rate": 0.00019872860347814984,
      "loss": 2.5125,
      "step": 447
    },
    {
      "epoch": 0.09874910453518487,
      "grad_norm": 0.9529692530632019,
      "learning_rate": 0.00019871699099073493,
      "loss": 2.5053,
      "step": 448
    },
    {
      "epoch": 0.09896952664352235,
      "grad_norm": 0.8587654829025269,
      "learning_rate": 0.0001987053260543342,
      "loss": 2.4321,
      "step": 449
    },
    {
      "epoch": 0.09918994875185981,
      "grad_norm": 0.9468286633491516,
      "learning_rate": 0.00019869360867514528,
      "loss": 2.5023,
      "step": 450
    },
    {
      "epoch": 0.09941037086019727,
      "grad_norm": 0.9098283052444458,
      "learning_rate": 0.00019868183885939374,
      "loss": 2.4781,
      "step": 451
    },
    {
      "epoch": 0.09963079296853475,
      "grad_norm": 0.9070652723312378,
      "learning_rate": 0.0001986700166133329,
      "loss": 2.4231,
      "step": 452
    },
    {
      "epoch": 0.09985121507687221,
      "grad_norm": 0.9446213841438293,
      "learning_rate": 0.00019865814194324405,
      "loss": 2.482,
      "step": 453
    },
    {
      "epoch": 0.10007163718520967,
      "grad_norm": 0.8510197401046753,
      "learning_rate": 0.00019864621485543625,
      "loss": 2.568,
      "step": 454
    },
    {
      "epoch": 0.10029205929354715,
      "grad_norm": 0.883448600769043,
      "learning_rate": 0.00019863423535624648,
      "loss": 2.4848,
      "step": 455
    },
    {
      "epoch": 0.10051248140188461,
      "grad_norm": 0.8431950211524963,
      "learning_rate": 0.00019862220345203945,
      "loss": 2.4715,
      "step": 456
    },
    {
      "epoch": 0.10073290351022207,
      "grad_norm": 0.8632293343544006,
      "learning_rate": 0.00019861011914920786,
      "loss": 2.4506,
      "step": 457
    },
    {
      "epoch": 0.10095332561855955,
      "grad_norm": 0.9084724187850952,
      "learning_rate": 0.00019859798245417217,
      "loss": 2.3485,
      "step": 458
    },
    {
      "epoch": 0.10117374772689701,
      "grad_norm": 0.8660531044006348,
      "learning_rate": 0.00019858579337338065,
      "loss": 2.4454,
      "step": 459
    },
    {
      "epoch": 0.10139416983523447,
      "grad_norm": 0.88209068775177,
      "learning_rate": 0.00019857355191330945,
      "loss": 2.502,
      "step": 460
    },
    {
      "epoch": 0.10161459194357195,
      "grad_norm": 0.9294498562812805,
      "learning_rate": 0.00019856125808046256,
      "loss": 2.4261,
      "step": 461
    },
    {
      "epoch": 0.10183501405190941,
      "grad_norm": 0.9578845500946045,
      "learning_rate": 0.00019854891188137173,
      "loss": 2.4625,
      "step": 462
    },
    {
      "epoch": 0.10205543616024687,
      "grad_norm": 0.8328554034233093,
      "learning_rate": 0.00019853651332259667,
      "loss": 2.4723,
      "step": 463
    },
    {
      "epoch": 0.10227585826858433,
      "grad_norm": 0.8760340809822083,
      "learning_rate": 0.00019852406241072475,
      "loss": 2.3569,
      "step": 464
    },
    {
      "epoch": 0.10249628037692181,
      "grad_norm": 0.9791350960731506,
      "learning_rate": 0.0001985115591523712,
      "loss": 2.5306,
      "step": 465
    },
    {
      "epoch": 0.10271670248525927,
      "grad_norm": 0.8385408520698547,
      "learning_rate": 0.00019849900355417907,
      "loss": 2.4061,
      "step": 466
    },
    {
      "epoch": 0.10293712459359673,
      "grad_norm": 0.8370196223258972,
      "learning_rate": 0.00019848639562281933,
      "loss": 2.4028,
      "step": 467
    },
    {
      "epoch": 0.10315754670193421,
      "grad_norm": 0.8802523612976074,
      "learning_rate": 0.0001984737353649906,
      "loss": 2.4398,
      "step": 468
    },
    {
      "epoch": 0.10337796881027167,
      "grad_norm": 0.8279548287391663,
      "learning_rate": 0.0001984610227874193,
      "loss": 2.3364,
      "step": 469
    },
    {
      "epoch": 0.10359839091860913,
      "grad_norm": 0.8453421592712402,
      "learning_rate": 0.00019844825789685981,
      "loss": 2.4406,
      "step": 470
    },
    {
      "epoch": 0.10381881302694661,
      "grad_norm": 0.8373908996582031,
      "learning_rate": 0.00019843544070009415,
      "loss": 2.3578,
      "step": 471
    },
    {
      "epoch": 0.10403923513528407,
      "grad_norm": 0.837346613407135,
      "learning_rate": 0.00019842257120393218,
      "loss": 2.4727,
      "step": 472
    },
    {
      "epoch": 0.10425965724362153,
      "grad_norm": 0.8748987913131714,
      "learning_rate": 0.00019840964941521155,
      "loss": 2.4342,
      "step": 473
    },
    {
      "epoch": 0.10448007935195901,
      "grad_norm": 0.8476139307022095,
      "learning_rate": 0.00019839667534079768,
      "loss": 2.4026,
      "step": 474
    },
    {
      "epoch": 0.10470050146029647,
      "grad_norm": 0.8471121191978455,
      "learning_rate": 0.00019838364898758378,
      "loss": 2.4069,
      "step": 475
    },
    {
      "epoch": 0.10492092356863393,
      "grad_norm": 0.8639768958091736,
      "learning_rate": 0.00019837057036249087,
      "loss": 2.56,
      "step": 476
    },
    {
      "epoch": 0.1051413456769714,
      "grad_norm": 0.9214815497398376,
      "learning_rate": 0.00019835743947246767,
      "loss": 2.4863,
      "step": 477
    },
    {
      "epoch": 0.10536176778530887,
      "grad_norm": 0.9511749148368835,
      "learning_rate": 0.00019834425632449075,
      "loss": 2.5774,
      "step": 478
    },
    {
      "epoch": 0.10558218989364633,
      "grad_norm": 0.898415744304657,
      "learning_rate": 0.00019833102092556436,
      "loss": 2.4855,
      "step": 479
    },
    {
      "epoch": 0.10580261200198379,
      "grad_norm": 0.9238340258598328,
      "learning_rate": 0.00019831773328272061,
      "loss": 2.4493,
      "step": 480
    },
    {
      "epoch": 0.10602303411032127,
      "grad_norm": 0.8870402574539185,
      "learning_rate": 0.00019830439340301926,
      "loss": 2.3677,
      "step": 481
    },
    {
      "epoch": 0.10624345621865873,
      "grad_norm": 0.8639870882034302,
      "learning_rate": 0.00019829100129354789,
      "loss": 2.3846,
      "step": 482
    },
    {
      "epoch": 0.10646387832699619,
      "grad_norm": 1.0021145343780518,
      "learning_rate": 0.00019827755696142182,
      "loss": 2.422,
      "step": 483
    },
    {
      "epoch": 0.10668430043533367,
      "grad_norm": 0.8831726908683777,
      "learning_rate": 0.00019826406041378408,
      "loss": 2.5148,
      "step": 484
    },
    {
      "epoch": 0.10690472254367113,
      "grad_norm": 1.0083831548690796,
      "learning_rate": 0.00019825051165780556,
      "loss": 2.4185,
      "step": 485
    },
    {
      "epoch": 0.10712514465200859,
      "grad_norm": 0.876901388168335,
      "learning_rate": 0.0001982369107006847,
      "loss": 2.551,
      "step": 486
    },
    {
      "epoch": 0.10734556676034607,
      "grad_norm": 1.0126320123672485,
      "learning_rate": 0.00019822325754964784,
      "loss": 2.5691,
      "step": 487
    },
    {
      "epoch": 0.10756598886868353,
      "grad_norm": 0.9211132526397705,
      "learning_rate": 0.000198209552211949,
      "loss": 2.4536,
      "step": 488
    },
    {
      "epoch": 0.10778641097702099,
      "grad_norm": 1.0158780813217163,
      "learning_rate": 0.00019819579469486986,
      "loss": 2.4785,
      "step": 489
    },
    {
      "epoch": 0.10800683308535847,
      "grad_norm": 0.8850935697555542,
      "learning_rate": 0.0001981819850057199,
      "loss": 2.4978,
      "step": 490
    },
    {
      "epoch": 0.10822725519369593,
      "grad_norm": 0.971473753452301,
      "learning_rate": 0.00019816812315183636,
      "loss": 2.3871,
      "step": 491
    },
    {
      "epoch": 0.10844767730203339,
      "grad_norm": 0.9635391235351562,
      "learning_rate": 0.00019815420914058406,
      "loss": 2.4745,
      "step": 492
    },
    {
      "epoch": 0.10866809941037087,
      "grad_norm": 0.8572251796722412,
      "learning_rate": 0.0001981402429793556,
      "loss": 2.4598,
      "step": 493
    },
    {
      "epoch": 0.10888852151870833,
      "grad_norm": 0.9158925414085388,
      "learning_rate": 0.00019812622467557133,
      "loss": 2.4172,
      "step": 494
    },
    {
      "epoch": 0.10910894362704579,
      "grad_norm": 0.8697640895843506,
      "learning_rate": 0.00019811215423667927,
      "loss": 2.3102,
      "step": 495
    },
    {
      "epoch": 0.10932936573538325,
      "grad_norm": 1.0629687309265137,
      "learning_rate": 0.00019809803167015512,
      "loss": 2.4374,
      "step": 496
    },
    {
      "epoch": 0.10954978784372073,
      "grad_norm": 0.8372770547866821,
      "learning_rate": 0.0001980838569835023,
      "loss": 2.4868,
      "step": 497
    },
    {
      "epoch": 0.10977020995205819,
      "grad_norm": 0.8978103995323181,
      "learning_rate": 0.0001980696301842519,
      "loss": 2.5612,
      "step": 498
    },
    {
      "epoch": 0.10999063206039565,
      "grad_norm": 0.9411421418190002,
      "learning_rate": 0.00019805535127996276,
      "loss": 2.4318,
      "step": 499
    },
    {
      "epoch": 0.11021105416873313,
      "grad_norm": 0.8958708047866821,
      "learning_rate": 0.00019804102027822128,
      "loss": 2.4879,
      "step": 500
    },
    {
      "epoch": 0.11043147627707059,
      "grad_norm": 0.931808352470398,
      "learning_rate": 0.00019802663718664168,
      "loss": 2.4576,
      "step": 501
    },
    {
      "epoch": 0.11065189838540805,
      "grad_norm": 0.9320315718650818,
      "learning_rate": 0.00019801220201286573,
      "loss": 2.3402,
      "step": 502
    },
    {
      "epoch": 0.11087232049374553,
      "grad_norm": 0.963492214679718,
      "learning_rate": 0.000197997714764563,
      "loss": 2.4612,
      "step": 503
    },
    {
      "epoch": 0.11109274260208299,
      "grad_norm": 0.8989193439483643,
      "learning_rate": 0.0001979831754494306,
      "loss": 2.457,
      "step": 504
    },
    {
      "epoch": 0.11131316471042045,
      "grad_norm": 0.9210177063941956,
      "learning_rate": 0.00019796858407519343,
      "loss": 2.463,
      "step": 505
    },
    {
      "epoch": 0.11153358681875793,
      "grad_norm": 0.977287769317627,
      "learning_rate": 0.0001979539406496039,
      "loss": 2.4622,
      "step": 506
    },
    {
      "epoch": 0.11175400892709539,
      "grad_norm": 0.9263628125190735,
      "learning_rate": 0.00019793924518044225,
      "loss": 2.4455,
      "step": 507
    },
    {
      "epoch": 0.11197443103543285,
      "grad_norm": 0.8218191862106323,
      "learning_rate": 0.0001979244976755162,
      "loss": 2.371,
      "step": 508
    },
    {
      "epoch": 0.11219485314377033,
      "grad_norm": 0.9691293239593506,
      "learning_rate": 0.00019790969814266121,
      "loss": 2.4406,
      "step": 509
    },
    {
      "epoch": 0.11241527525210779,
      "grad_norm": 0.8875851035118103,
      "learning_rate": 0.00019789484658974045,
      "loss": 2.4554,
      "step": 510
    },
    {
      "epoch": 0.11263569736044525,
      "grad_norm": 0.8086339235305786,
      "learning_rate": 0.00019787994302464455,
      "loss": 2.5517,
      "step": 511
    },
    {
      "epoch": 0.11285611946878273,
      "grad_norm": 0.9300101399421692,
      "learning_rate": 0.0001978649874552919,
      "loss": 2.4344,
      "step": 512
    },
    {
      "epoch": 0.11307654157712019,
      "grad_norm": 0.8738496899604797,
      "learning_rate": 0.0001978499798896285,
      "loss": 2.3522,
      "step": 513
    },
    {
      "epoch": 0.11329696368545765,
      "grad_norm": 0.8192237019538879,
      "learning_rate": 0.00019783492033562797,
      "loss": 2.4756,
      "step": 514
    },
    {
      "epoch": 0.11351738579379511,
      "grad_norm": 0.8979835510253906,
      "learning_rate": 0.00019781980880129155,
      "loss": 2.4212,
      "step": 515
    },
    {
      "epoch": 0.11373780790213259,
      "grad_norm": 0.9438988566398621,
      "learning_rate": 0.0001978046452946481,
      "loss": 2.393,
      "step": 516
    },
    {
      "epoch": 0.11395823001047005,
      "grad_norm": 0.9210966229438782,
      "learning_rate": 0.0001977894298237541,
      "loss": 2.3917,
      "step": 517
    },
    {
      "epoch": 0.11417865211880751,
      "grad_norm": 1.0489916801452637,
      "learning_rate": 0.0001977741623966936,
      "loss": 2.4195,
      "step": 518
    },
    {
      "epoch": 0.11439907422714499,
      "grad_norm": 0.95262610912323,
      "learning_rate": 0.00019775884302157827,
      "loss": 2.5422,
      "step": 519
    },
    {
      "epoch": 0.11461949633548245,
      "grad_norm": 0.9434237480163574,
      "learning_rate": 0.00019774347170654745,
      "loss": 2.3437,
      "step": 520
    },
    {
      "epoch": 0.11483991844381991,
      "grad_norm": 0.9688859581947327,
      "learning_rate": 0.000197728048459768,
      "loss": 2.2932,
      "step": 521
    },
    {
      "epoch": 0.11506034055215739,
      "grad_norm": 0.9786443710327148,
      "learning_rate": 0.0001977125732894344,
      "loss": 2.4509,
      "step": 522
    },
    {
      "epoch": 0.11528076266049485,
      "grad_norm": 0.8669262528419495,
      "learning_rate": 0.00019769704620376866,
      "loss": 2.4153,
      "step": 523
    },
    {
      "epoch": 0.11550118476883231,
      "grad_norm": 0.9269571304321289,
      "learning_rate": 0.00019768146721102048,
      "loss": 2.2801,
      "step": 524
    },
    {
      "epoch": 0.11572160687716979,
      "grad_norm": 0.9928492903709412,
      "learning_rate": 0.00019766583631946705,
      "loss": 2.323,
      "step": 525
    },
    {
      "epoch": 0.11594202898550725,
      "grad_norm": 0.8634804487228394,
      "learning_rate": 0.0001976501535374132,
      "loss": 2.4961,
      "step": 526
    },
    {
      "epoch": 0.11616245109384471,
      "grad_norm": 0.8569937348365784,
      "learning_rate": 0.00019763441887319126,
      "loss": 2.5007,
      "step": 527
    },
    {
      "epoch": 0.11638287320218219,
      "grad_norm": 0.9240749478340149,
      "learning_rate": 0.00019761863233516117,
      "loss": 2.3732,
      "step": 528
    },
    {
      "epoch": 0.11660329531051965,
      "grad_norm": 0.876115083694458,
      "learning_rate": 0.00019760279393171042,
      "loss": 2.3252,
      "step": 529
    },
    {
      "epoch": 0.11682371741885711,
      "grad_norm": 0.9264326691627502,
      "learning_rate": 0.00019758690367125407,
      "loss": 2.5123,
      "step": 530
    },
    {
      "epoch": 0.11704413952719457,
      "grad_norm": 0.9125890135765076,
      "learning_rate": 0.0001975709615622347,
      "loss": 2.4093,
      "step": 531
    },
    {
      "epoch": 0.11726456163553205,
      "grad_norm": 0.8883729577064514,
      "learning_rate": 0.00019755496761312245,
      "loss": 2.4854,
      "step": 532
    },
    {
      "epoch": 0.11748498374386951,
      "grad_norm": 0.9333995580673218,
      "learning_rate": 0.00019753892183241507,
      "loss": 2.3915,
      "step": 533
    },
    {
      "epoch": 0.11770540585220697,
      "grad_norm": 0.9708004593849182,
      "learning_rate": 0.00019752282422863773,
      "loss": 2.4182,
      "step": 534
    },
    {
      "epoch": 0.11792582796054445,
      "grad_norm": 0.8696708083152771,
      "learning_rate": 0.00019750667481034317,
      "loss": 2.462,
      "step": 535
    },
    {
      "epoch": 0.11814625006888191,
      "grad_norm": 0.9083542823791504,
      "learning_rate": 0.00019749047358611177,
      "loss": 2.3821,
      "step": 536
    },
    {
      "epoch": 0.11836667217721937,
      "grad_norm": 0.8722495436668396,
      "learning_rate": 0.0001974742205645513,
      "loss": 2.4268,
      "step": 537
    },
    {
      "epoch": 0.11858709428555685,
      "grad_norm": 0.8861777782440186,
      "learning_rate": 0.00019745791575429705,
      "loss": 2.4852,
      "step": 538
    },
    {
      "epoch": 0.11880751639389431,
      "grad_norm": 0.9034215211868286,
      "learning_rate": 0.00019744155916401196,
      "loss": 2.4554,
      "step": 539
    },
    {
      "epoch": 0.11902793850223177,
      "grad_norm": 0.8417815566062927,
      "learning_rate": 0.00019742515080238635,
      "loss": 2.3419,
      "step": 540
    },
    {
      "epoch": 0.11924836061056925,
      "grad_norm": 0.8288694620132446,
      "learning_rate": 0.0001974086906781381,
      "loss": 2.3654,
      "step": 541
    },
    {
      "epoch": 0.11946878271890671,
      "grad_norm": 0.8954201936721802,
      "learning_rate": 0.00019739217880001254,
      "loss": 2.2975,
      "step": 542
    },
    {
      "epoch": 0.11968920482724417,
      "grad_norm": 0.9126654267311096,
      "learning_rate": 0.00019737561517678258,
      "loss": 2.4459,
      "step": 543
    },
    {
      "epoch": 0.11990962693558165,
      "grad_norm": 0.842075765132904,
      "learning_rate": 0.00019735899981724862,
      "loss": 2.3324,
      "step": 544
    },
    {
      "epoch": 0.12013004904391911,
      "grad_norm": 0.910011887550354,
      "learning_rate": 0.00019734233273023842,
      "loss": 2.3049,
      "step": 545
    },
    {
      "epoch": 0.12035047115225657,
      "grad_norm": 0.8967334628105164,
      "learning_rate": 0.00019732561392460734,
      "loss": 2.4204,
      "step": 546
    },
    {
      "epoch": 0.12057089326059403,
      "grad_norm": 0.9131148457527161,
      "learning_rate": 0.00019730884340923827,
      "loss": 2.4588,
      "step": 547
    },
    {
      "epoch": 0.1207913153689315,
      "grad_norm": 0.8855314254760742,
      "learning_rate": 0.0001972920211930414,
      "loss": 2.4634,
      "step": 548
    },
    {
      "epoch": 0.12101173747726897,
      "grad_norm": 0.8510825634002686,
      "learning_rate": 0.0001972751472849545,
      "loss": 2.4309,
      "step": 549
    },
    {
      "epoch": 0.12123215958560643,
      "grad_norm": 0.8599485754966736,
      "learning_rate": 0.00019725822169394284,
      "loss": 2.4095,
      "step": 550
    },
    {
      "epoch": 0.1214525816939439,
      "grad_norm": 0.8969146609306335,
      "learning_rate": 0.00019724124442899906,
      "loss": 2.4247,
      "step": 551
    },
    {
      "epoch": 0.12167300380228137,
      "grad_norm": 0.8816750645637512,
      "learning_rate": 0.0001972242154991433,
      "loss": 2.3948,
      "step": 552
    },
    {
      "epoch": 0.12189342591061883,
      "grad_norm": 0.8850644826889038,
      "learning_rate": 0.00019720713491342312,
      "loss": 2.4766,
      "step": 553
    },
    {
      "epoch": 0.1221138480189563,
      "grad_norm": 0.9006227850914001,
      "learning_rate": 0.0001971900026809136,
      "loss": 2.4771,
      "step": 554
    },
    {
      "epoch": 0.12233427012729377,
      "grad_norm": 0.8950319886207581,
      "learning_rate": 0.00019717281881071713,
      "loss": 2.4085,
      "step": 555
    },
    {
      "epoch": 0.12255469223563123,
      "grad_norm": 0.8512940406799316,
      "learning_rate": 0.0001971555833119637,
      "loss": 2.3406,
      "step": 556
    },
    {
      "epoch": 0.1227751143439687,
      "grad_norm": 0.8547590970993042,
      "learning_rate": 0.00019713829619381058,
      "loss": 2.4099,
      "step": 557
    },
    {
      "epoch": 0.12299553645230617,
      "grad_norm": 0.8668827414512634,
      "learning_rate": 0.00019712095746544255,
      "loss": 2.4097,
      "step": 558
    },
    {
      "epoch": 0.12321595856064363,
      "grad_norm": 0.9190183877944946,
      "learning_rate": 0.0001971035671360718,
      "loss": 2.404,
      "step": 559
    },
    {
      "epoch": 0.1234363806689811,
      "grad_norm": 0.9003810286521912,
      "learning_rate": 0.00019708612521493788,
      "loss": 2.3288,
      "step": 560
    },
    {
      "epoch": 0.12365680277731857,
      "grad_norm": 0.9198606610298157,
      "learning_rate": 0.00019706863171130786,
      "loss": 2.3918,
      "step": 561
    },
    {
      "epoch": 0.12387722488565603,
      "grad_norm": 0.8573213219642639,
      "learning_rate": 0.00019705108663447613,
      "loss": 2.4864,
      "step": 562
    },
    {
      "epoch": 0.12409764699399349,
      "grad_norm": 0.8480684757232666,
      "learning_rate": 0.00019703348999376453,
      "loss": 2.3066,
      "step": 563
    },
    {
      "epoch": 0.12431806910233097,
      "grad_norm": 0.9620939493179321,
      "learning_rate": 0.0001970158417985222,
      "loss": 2.4952,
      "step": 564
    },
    {
      "epoch": 0.12453849121066843,
      "grad_norm": 0.8678823113441467,
      "learning_rate": 0.0001969981420581258,
      "loss": 2.3492,
      "step": 565
    },
    {
      "epoch": 0.12475891331900589,
      "grad_norm": 0.8374990224838257,
      "learning_rate": 0.0001969803907819793,
      "loss": 2.4,
      "step": 566
    },
    {
      "epoch": 0.12497933542734337,
      "grad_norm": 0.83769690990448,
      "learning_rate": 0.00019696258797951405,
      "loss": 2.3846,
      "step": 567
    },
    {
      "epoch": 0.12519975753568083,
      "grad_norm": 0.8264140486717224,
      "learning_rate": 0.00019694473366018887,
      "loss": 2.4146,
      "step": 568
    },
    {
      "epoch": 0.1254201796440183,
      "grad_norm": 0.797687292098999,
      "learning_rate": 0.0001969268278334898,
      "loss": 2.3555,
      "step": 569
    },
    {
      "epoch": 0.12564060175235575,
      "grad_norm": 0.86750727891922,
      "learning_rate": 0.00019690887050893035,
      "loss": 2.4338,
      "step": 570
    },
    {
      "epoch": 0.12586102386069323,
      "grad_norm": 0.8704907298088074,
      "learning_rate": 0.00019689086169605137,
      "loss": 2.337,
      "step": 571
    },
    {
      "epoch": 0.1260814459690307,
      "grad_norm": 0.7821561694145203,
      "learning_rate": 0.00019687280140442102,
      "loss": 2.4047,
      "step": 572
    },
    {
      "epoch": 0.12630186807736815,
      "grad_norm": 0.8935369253158569,
      "learning_rate": 0.00019685468964363495,
      "loss": 2.3393,
      "step": 573
    },
    {
      "epoch": 0.12652229018570563,
      "grad_norm": 0.9224193692207336,
      "learning_rate": 0.00019683652642331595,
      "loss": 2.4327,
      "step": 574
    },
    {
      "epoch": 0.1267427122940431,
      "grad_norm": 0.955625593662262,
      "learning_rate": 0.00019681831175311432,
      "loss": 2.3724,
      "step": 575
    },
    {
      "epoch": 0.12696313440238055,
      "grad_norm": 0.9032315015792847,
      "learning_rate": 0.00019680004564270762,
      "loss": 2.4158,
      "step": 576
    },
    {
      "epoch": 0.12718355651071803,
      "grad_norm": 0.8479869961738586,
      "learning_rate": 0.00019678172810180073,
      "loss": 2.3895,
      "step": 577
    },
    {
      "epoch": 0.1274039786190555,
      "grad_norm": 0.8186668157577515,
      "learning_rate": 0.0001967633591401259,
      "loss": 2.4053,
      "step": 578
    },
    {
      "epoch": 0.12762440072739295,
      "grad_norm": 0.9172797203063965,
      "learning_rate": 0.0001967449387674427,
      "loss": 2.4341,
      "step": 579
    },
    {
      "epoch": 0.12784482283573043,
      "grad_norm": 0.9016140103340149,
      "learning_rate": 0.00019672646699353793,
      "loss": 2.4239,
      "step": 580
    },
    {
      "epoch": 0.1280652449440679,
      "grad_norm": 0.9162954688072205,
      "learning_rate": 0.00019670794382822587,
      "loss": 2.4656,
      "step": 581
    },
    {
      "epoch": 0.12828566705240535,
      "grad_norm": 0.8306110501289368,
      "learning_rate": 0.0001966893692813479,
      "loss": 2.3063,
      "step": 582
    },
    {
      "epoch": 0.12850608916074283,
      "grad_norm": 0.8596447110176086,
      "learning_rate": 0.00019667074336277284,
      "loss": 2.4086,
      "step": 583
    },
    {
      "epoch": 0.1287265112690803,
      "grad_norm": 0.9531026482582092,
      "learning_rate": 0.00019665206608239678,
      "loss": 2.4181,
      "step": 584
    },
    {
      "epoch": 0.12894693337741775,
      "grad_norm": 0.8444340229034424,
      "learning_rate": 0.00019663333745014306,
      "loss": 2.3932,
      "step": 585
    },
    {
      "epoch": 0.12916735548575523,
      "grad_norm": 0.8822381496429443,
      "learning_rate": 0.00019661455747596232,
      "loss": 2.5171,
      "step": 586
    },
    {
      "epoch": 0.1293877775940927,
      "grad_norm": 0.9273701906204224,
      "learning_rate": 0.0001965957261698325,
      "loss": 2.4179,
      "step": 587
    },
    {
      "epoch": 0.12960819970243015,
      "grad_norm": 0.9508959054946899,
      "learning_rate": 0.0001965768435417588,
      "loss": 2.4237,
      "step": 588
    },
    {
      "epoch": 0.12982862181076762,
      "grad_norm": 0.9460341334342957,
      "learning_rate": 0.00019655790960177366,
      "loss": 2.3397,
      "step": 589
    },
    {
      "epoch": 0.13004904391910507,
      "grad_norm": 0.9277721643447876,
      "learning_rate": 0.0001965389243599368,
      "loss": 2.4301,
      "step": 590
    },
    {
      "epoch": 0.13026946602744255,
      "grad_norm": 0.8583454489707947,
      "learning_rate": 0.00019651988782633524,
      "loss": 2.4996,
      "step": 591
    },
    {
      "epoch": 0.13048988813578002,
      "grad_norm": 0.8946618437767029,
      "learning_rate": 0.0001965008000110832,
      "loss": 2.3794,
      "step": 592
    },
    {
      "epoch": 0.13071031024411747,
      "grad_norm": 0.8247033357620239,
      "learning_rate": 0.00019648166092432217,
      "loss": 2.527,
      "step": 593
    },
    {
      "epoch": 0.13093073235245495,
      "grad_norm": 0.8655588626861572,
      "learning_rate": 0.00019646247057622084,
      "loss": 2.4269,
      "step": 594
    },
    {
      "epoch": 0.13115115446079242,
      "grad_norm": 0.827886700630188,
      "learning_rate": 0.00019644322897697521,
      "loss": 2.3273,
      "step": 595
    },
    {
      "epoch": 0.13137157656912987,
      "grad_norm": 0.8388336896896362,
      "learning_rate": 0.00019642393613680844,
      "loss": 2.3426,
      "step": 596
    },
    {
      "epoch": 0.13159199867746735,
      "grad_norm": 0.7935357689857483,
      "learning_rate": 0.00019640459206597094,
      "loss": 2.3522,
      "step": 597
    },
    {
      "epoch": 0.13181242078580482,
      "grad_norm": 0.8987339735031128,
      "learning_rate": 0.0001963851967747404,
      "loss": 2.4594,
      "step": 598
    },
    {
      "epoch": 0.13203284289414227,
      "grad_norm": 0.8827887177467346,
      "learning_rate": 0.00019636575027342156,
      "loss": 2.3329,
      "step": 599
    },
    {
      "epoch": 0.13225326500247975,
      "grad_norm": 0.870638370513916,
      "learning_rate": 0.00019634625257234655,
      "loss": 2.3498,
      "step": 600
    },
    {
      "epoch": 0.13225326500247975,
      "eval_bpb": 1.2384879814456038,
      "eval_loss": 2.38625431060791,
      "eval_runtime": 184.3703,
      "eval_samples_per_second": 62.158,
      "eval_steps_per_second": 7.772,
      "step": 600
    },
    {
      "epoch": 0.13247368711081722,
      "grad_norm": 0.8607369065284729,
      "learning_rate": 0.00019632670368187466,
      "loss": 2.3902,
      "step": 601
    },
    {
      "epoch": 0.13269410921915467,
      "grad_norm": 0.8496939539909363,
      "learning_rate": 0.00019630710361239224,
      "loss": 2.2775,
      "step": 602
    },
    {
      "epoch": 0.13291453132749215,
      "grad_norm": 0.8246957063674927,
      "learning_rate": 0.00019628745237431302,
      "loss": 2.4513,
      "step": 603
    },
    {
      "epoch": 0.13313495343582962,
      "grad_norm": 0.7840679883956909,
      "learning_rate": 0.00019626774997807777,
      "loss": 2.4557,
      "step": 604
    },
    {
      "epoch": 0.13335537554416707,
      "grad_norm": 0.8735194206237793,
      "learning_rate": 0.00019624799643415454,
      "loss": 2.411,
      "step": 605
    },
    {
      "epoch": 0.13357579765250455,
      "grad_norm": 0.7976757287979126,
      "learning_rate": 0.0001962281917530385,
      "loss": 2.3164,
      "step": 606
    },
    {
      "epoch": 0.13379621976084202,
      "grad_norm": 0.7971235513687134,
      "learning_rate": 0.00019620833594525205,
      "loss": 2.4333,
      "step": 607
    },
    {
      "epoch": 0.13401664186917947,
      "grad_norm": 0.8541024327278137,
      "learning_rate": 0.00019618842902134465,
      "loss": 2.4151,
      "step": 608
    },
    {
      "epoch": 0.13423706397751695,
      "grad_norm": 0.9011279344558716,
      "learning_rate": 0.00019616847099189304,
      "loss": 2.4725,
      "step": 609
    },
    {
      "epoch": 0.13445748608585442,
      "grad_norm": 0.8120802044868469,
      "learning_rate": 0.00019614846186750102,
      "loss": 2.3855,
      "step": 610
    },
    {
      "epoch": 0.13467790819419187,
      "grad_norm": 0.7903881669044495,
      "learning_rate": 0.00019612840165879957,
      "loss": 2.4274,
      "step": 611
    },
    {
      "epoch": 0.13489833030252935,
      "grad_norm": 0.862602174282074,
      "learning_rate": 0.0001961082903764468,
      "loss": 2.2542,
      "step": 612
    },
    {
      "epoch": 0.13511875241086682,
      "grad_norm": 0.8453419208526611,
      "learning_rate": 0.00019608812803112796,
      "loss": 2.4495,
      "step": 613
    },
    {
      "epoch": 0.13533917451920427,
      "grad_norm": 0.8418912291526794,
      "learning_rate": 0.00019606791463355546,
      "loss": 2.3996,
      "step": 614
    },
    {
      "epoch": 0.13555959662754175,
      "grad_norm": 0.8727667331695557,
      "learning_rate": 0.00019604765019446885,
      "loss": 2.4788,
      "step": 615
    },
    {
      "epoch": 0.13578001873587922,
      "grad_norm": 0.8681816458702087,
      "learning_rate": 0.00019602733472463469,
      "loss": 2.3527,
      "step": 616
    },
    {
      "epoch": 0.13600044084421667,
      "grad_norm": 0.8631755113601685,
      "learning_rate": 0.00019600696823484672,
      "loss": 2.4381,
      "step": 617
    },
    {
      "epoch": 0.13622086295255414,
      "grad_norm": 0.8595993518829346,
      "learning_rate": 0.00019598655073592585,
      "loss": 2.3189,
      "step": 618
    },
    {
      "epoch": 0.13644128506089162,
      "grad_norm": 0.823720395565033,
      "learning_rate": 0.00019596608223871998,
      "loss": 2.4361,
      "step": 619
    },
    {
      "epoch": 0.13666170716922907,
      "grad_norm": 0.9071880578994751,
      "learning_rate": 0.00019594556275410416,
      "loss": 2.3861,
      "step": 620
    },
    {
      "epoch": 0.13688212927756654,
      "grad_norm": 0.9035383462905884,
      "learning_rate": 0.00019592499229298058,
      "loss": 2.3502,
      "step": 621
    },
    {
      "epoch": 0.137102551385904,
      "grad_norm": 0.8292778134346008,
      "learning_rate": 0.0001959043708662784,
      "loss": 2.2872,
      "step": 622
    },
    {
      "epoch": 0.13732297349424147,
      "grad_norm": 0.969109058380127,
      "learning_rate": 0.00019588369848495395,
      "loss": 2.4673,
      "step": 623
    },
    {
      "epoch": 0.13754339560257894,
      "grad_norm": 0.8532236218452454,
      "learning_rate": 0.00019586297515999062,
      "loss": 2.2476,
      "step": 624
    },
    {
      "epoch": 0.1377638177109164,
      "grad_norm": 0.8120128512382507,
      "learning_rate": 0.0001958422009023988,
      "loss": 2.3663,
      "step": 625
    },
    {
      "epoch": 0.13798423981925387,
      "grad_norm": 0.9549170136451721,
      "learning_rate": 0.000195821375723216,
      "loss": 2.3549,
      "step": 626
    },
    {
      "epoch": 0.13820466192759134,
      "grad_norm": 0.8904028534889221,
      "learning_rate": 0.00019580049963350678,
      "loss": 2.2823,
      "step": 627
    },
    {
      "epoch": 0.1384250840359288,
      "grad_norm": 0.8398650288581848,
      "learning_rate": 0.0001957795726443628,
      "loss": 2.3272,
      "step": 628
    },
    {
      "epoch": 0.13864550614426627,
      "grad_norm": 0.8714176416397095,
      "learning_rate": 0.00019575859476690262,
      "loss": 2.4472,
      "step": 629
    },
    {
      "epoch": 0.13886592825260374,
      "grad_norm": 0.8230476379394531,
      "learning_rate": 0.00019573756601227196,
      "loss": 2.3518,
      "step": 630
    },
    {
      "epoch": 0.1390863503609412,
      "grad_norm": 0.8205571174621582,
      "learning_rate": 0.00019571648639164356,
      "loss": 2.3992,
      "step": 631
    },
    {
      "epoch": 0.13930677246927867,
      "grad_norm": 0.8590588569641113,
      "learning_rate": 0.00019569535591621711,
      "loss": 2.4043,
      "step": 632
    },
    {
      "epoch": 0.13952719457761614,
      "grad_norm": 0.8276355862617493,
      "learning_rate": 0.00019567417459721945,
      "loss": 2.3137,
      "step": 633
    },
    {
      "epoch": 0.1397476166859536,
      "grad_norm": 0.8279332518577576,
      "learning_rate": 0.00019565294244590427,
      "loss": 2.2845,
      "step": 634
    },
    {
      "epoch": 0.13996803879429107,
      "grad_norm": 0.7957282066345215,
      "learning_rate": 0.00019563165947355238,
      "loss": 2.322,
      "step": 635
    },
    {
      "epoch": 0.14018846090262854,
      "grad_norm": 0.8372554779052734,
      "learning_rate": 0.00019561032569147157,
      "loss": 2.4157,
      "step": 636
    },
    {
      "epoch": 0.140408883010966,
      "grad_norm": 0.8898109793663025,
      "learning_rate": 0.00019558894111099663,
      "loss": 2.4915,
      "step": 637
    },
    {
      "epoch": 0.14062930511930347,
      "grad_norm": 0.8251047134399414,
      "learning_rate": 0.0001955675057434893,
      "loss": 2.2732,
      "step": 638
    },
    {
      "epoch": 0.14084972722764094,
      "grad_norm": 0.9171207547187805,
      "learning_rate": 0.00019554601960033833,
      "loss": 2.3614,
      "step": 639
    },
    {
      "epoch": 0.1410701493359784,
      "grad_norm": 0.809741199016571,
      "learning_rate": 0.0001955244826929595,
      "loss": 2.2807,
      "step": 640
    },
    {
      "epoch": 0.14129057144431587,
      "grad_norm": 0.8434454202651978,
      "learning_rate": 0.00019550289503279545,
      "loss": 2.3416,
      "step": 641
    },
    {
      "epoch": 0.14151099355265334,
      "grad_norm": 0.8153087496757507,
      "learning_rate": 0.00019548125663131592,
      "loss": 2.3133,
      "step": 642
    },
    {
      "epoch": 0.1417314156609908,
      "grad_norm": 0.8099733591079712,
      "learning_rate": 0.00019545956750001742,
      "loss": 2.3935,
      "step": 643
    },
    {
      "epoch": 0.14195183776932827,
      "grad_norm": 0.9326735734939575,
      "learning_rate": 0.0001954378276504236,
      "loss": 2.2981,
      "step": 644
    },
    {
      "epoch": 0.14217225987766574,
      "grad_norm": 0.8654516339302063,
      "learning_rate": 0.00019541603709408506,
      "loss": 2.3919,
      "step": 645
    },
    {
      "epoch": 0.1423926819860032,
      "grad_norm": 0.7792201042175293,
      "learning_rate": 0.0001953941958425791,
      "loss": 2.3159,
      "step": 646
    },
    {
      "epoch": 0.14261310409434066,
      "grad_norm": 0.8007216453552246,
      "learning_rate": 0.0001953723039075102,
      "loss": 2.2073,
      "step": 647
    },
    {
      "epoch": 0.14283352620267814,
      "grad_norm": 0.8130624294281006,
      "learning_rate": 0.00019535036130050975,
      "loss": 2.3948,
      "step": 648
    },
    {
      "epoch": 0.1430539483110156,
      "grad_norm": 0.8614869713783264,
      "learning_rate": 0.00019532836803323592,
      "loss": 2.453,
      "step": 649
    },
    {
      "epoch": 0.14327437041935306,
      "grad_norm": 0.7928333878517151,
      "learning_rate": 0.0001953063241173739,
      "loss": 2.4241,
      "step": 650
    },
    {
      "epoch": 0.14349479252769054,
      "grad_norm": 0.8143974542617798,
      "learning_rate": 0.00019528422956463577,
      "loss": 2.3367,
      "step": 651
    },
    {
      "epoch": 0.143715214636028,
      "grad_norm": 0.9282017946243286,
      "learning_rate": 0.0001952620843867605,
      "loss": 2.2924,
      "step": 652
    },
    {
      "epoch": 0.14393563674436546,
      "grad_norm": 0.8618532419204712,
      "learning_rate": 0.000195239888595514,
      "loss": 2.3698,
      "step": 653
    },
    {
      "epoch": 0.14415605885270294,
      "grad_norm": 0.9235208630561829,
      "learning_rate": 0.000195217642202689,
      "loss": 2.3773,
      "step": 654
    },
    {
      "epoch": 0.1443764809610404,
      "grad_norm": 0.8765738606452942,
      "learning_rate": 0.0001951953452201052,
      "loss": 2.3504,
      "step": 655
    },
    {
      "epoch": 0.14459690306937786,
      "grad_norm": 0.8894654512405396,
      "learning_rate": 0.00019517299765960908,
      "loss": 2.3311,
      "step": 656
    },
    {
      "epoch": 0.1448173251777153,
      "grad_norm": 0.826357364654541,
      "learning_rate": 0.00019515059953307407,
      "loss": 2.3092,
      "step": 657
    },
    {
      "epoch": 0.1450377472860528,
      "grad_norm": 0.9697829484939575,
      "learning_rate": 0.00019512815085240046,
      "loss": 2.4684,
      "step": 658
    },
    {
      "epoch": 0.14525816939439026,
      "grad_norm": 0.895580530166626,
      "learning_rate": 0.00019510565162951537,
      "loss": 2.3853,
      "step": 659
    },
    {
      "epoch": 0.1454785915027277,
      "grad_norm": 0.9855853915214539,
      "learning_rate": 0.00019508310187637276,
      "loss": 2.4275,
      "step": 660
    },
    {
      "epoch": 0.1456990136110652,
      "grad_norm": 0.8341454863548279,
      "learning_rate": 0.0001950605016049535,
      "loss": 2.4865,
      "step": 661
    },
    {
      "epoch": 0.14591943571940266,
      "grad_norm": 0.8472642302513123,
      "learning_rate": 0.00019503785082726523,
      "loss": 2.4796,
      "step": 662
    },
    {
      "epoch": 0.1461398578277401,
      "grad_norm": 0.8735401034355164,
      "learning_rate": 0.0001950151495553425,
      "loss": 2.3654,
      "step": 663
    },
    {
      "epoch": 0.1463602799360776,
      "grad_norm": 0.8771145343780518,
      "learning_rate": 0.0001949923978012466,
      "loss": 2.3447,
      "step": 664
    },
    {
      "epoch": 0.14658070204441506,
      "grad_norm": 0.8112457394599915,
      "learning_rate": 0.00019496959557706572,
      "loss": 2.3218,
      "step": 665
    },
    {
      "epoch": 0.1468011241527525,
      "grad_norm": 0.8239163756370544,
      "learning_rate": 0.00019494674289491476,
      "loss": 2.3652,
      "step": 666
    },
    {
      "epoch": 0.14702154626108999,
      "grad_norm": 0.8809289932250977,
      "learning_rate": 0.00019492383976693563,
      "loss": 2.3587,
      "step": 667
    },
    {
      "epoch": 0.14724196836942746,
      "grad_norm": 0.9105719923973083,
      "learning_rate": 0.00019490088620529678,
      "loss": 2.3815,
      "step": 668
    },
    {
      "epoch": 0.1474623904777649,
      "grad_norm": 0.8243231177330017,
      "learning_rate": 0.00019487788222219367,
      "loss": 2.4275,
      "step": 669
    },
    {
      "epoch": 0.14768281258610239,
      "grad_norm": 0.7805619835853577,
      "learning_rate": 0.00019485482782984846,
      "loss": 2.2634,
      "step": 670
    },
    {
      "epoch": 0.14790323469443986,
      "grad_norm": 0.8525617122650146,
      "learning_rate": 0.00019483172304051008,
      "loss": 2.369,
      "step": 671
    },
    {
      "epoch": 0.1481236568027773,
      "grad_norm": 0.7677012085914612,
      "learning_rate": 0.00019480856786645423,
      "loss": 2.4406,
      "step": 672
    },
    {
      "epoch": 0.14834407891111479,
      "grad_norm": 0.8337851762771606,
      "learning_rate": 0.00019478536231998342,
      "loss": 2.3942,
      "step": 673
    },
    {
      "epoch": 0.14856450101945226,
      "grad_norm": 0.8950392007827759,
      "learning_rate": 0.00019476210641342697,
      "loss": 2.4125,
      "step": 674
    },
    {
      "epoch": 0.1487849231277897,
      "grad_norm": 0.8016397356987,
      "learning_rate": 0.00019473880015914087,
      "loss": 2.27,
      "step": 675
    },
    {
      "epoch": 0.14900534523612718,
      "grad_norm": 0.764148473739624,
      "learning_rate": 0.00019471544356950782,
      "loss": 2.1923,
      "step": 676
    },
    {
      "epoch": 0.14922576734446466,
      "grad_norm": 0.8675015568733215,
      "learning_rate": 0.00019469203665693736,
      "loss": 2.3989,
      "step": 677
    },
    {
      "epoch": 0.1494461894528021,
      "grad_norm": 0.8498058319091797,
      "learning_rate": 0.0001946685794338658,
      "loss": 2.4334,
      "step": 678
    },
    {
      "epoch": 0.14966661156113958,
      "grad_norm": 0.785318911075592,
      "learning_rate": 0.00019464507191275604,
      "loss": 2.3836,
      "step": 679
    },
    {
      "epoch": 0.14988703366947706,
      "grad_norm": 0.8675945997238159,
      "learning_rate": 0.00019462151410609783,
      "loss": 2.4265,
      "step": 680
    },
    {
      "epoch": 0.1501074557778145,
      "grad_norm": 0.8440405130386353,
      "learning_rate": 0.00019459790602640754,
      "loss": 2.3836,
      "step": 681
    },
    {
      "epoch": 0.15032787788615198,
      "grad_norm": 0.8514573574066162,
      "learning_rate": 0.00019457424768622836,
      "loss": 2.4278,
      "step": 682
    },
    {
      "epoch": 0.15054829999448946,
      "grad_norm": 0.8564261198043823,
      "learning_rate": 0.00019455053909813008,
      "loss": 2.3488,
      "step": 683
    },
    {
      "epoch": 0.1507687221028269,
      "grad_norm": 0.8667369484901428,
      "learning_rate": 0.00019452678027470923,
      "loss": 2.3297,
      "step": 684
    },
    {
      "epoch": 0.15098914421116438,
      "grad_norm": 0.7938596606254578,
      "learning_rate": 0.00019450297122858902,
      "loss": 2.3252,
      "step": 685
    },
    {
      "epoch": 0.15120956631950186,
      "grad_norm": 0.7393359541893005,
      "learning_rate": 0.00019447911197241938,
      "loss": 2.1877,
      "step": 686
    },
    {
      "epoch": 0.1514299884278393,
      "grad_norm": 0.9528443813323975,
      "learning_rate": 0.00019445520251887687,
      "loss": 2.3051,
      "step": 687
    },
    {
      "epoch": 0.15165041053617678,
      "grad_norm": 0.8548237681388855,
      "learning_rate": 0.00019443124288066475,
      "loss": 2.3309,
      "step": 688
    },
    {
      "epoch": 0.15187083264451423,
      "grad_norm": 0.8516532778739929,
      "learning_rate": 0.00019440723307051293,
      "loss": 2.2434,
      "step": 689
    },
    {
      "epoch": 0.1520912547528517,
      "grad_norm": 0.8477699756622314,
      "learning_rate": 0.000194383173101178,
      "loss": 2.3313,
      "step": 690
    },
    {
      "epoch": 0.15231167686118918,
      "grad_norm": 0.8487361073493958,
      "learning_rate": 0.00019435906298544313,
      "loss": 2.3165,
      "step": 691
    },
    {
      "epoch": 0.15253209896952663,
      "grad_norm": 0.7854897975921631,
      "learning_rate": 0.00019433490273611823,
      "loss": 2.3242,
      "step": 692
    },
    {
      "epoch": 0.1527525210778641,
      "grad_norm": 0.8711138963699341,
      "learning_rate": 0.00019431069236603977,
      "loss": 2.3005,
      "step": 693
    },
    {
      "epoch": 0.15297294318620158,
      "grad_norm": 0.8421791791915894,
      "learning_rate": 0.00019428643188807088,
      "loss": 2.4097,
      "step": 694
    },
    {
      "epoch": 0.15319336529453903,
      "grad_norm": 0.8071432113647461,
      "learning_rate": 0.00019426212131510133,
      "loss": 2.3521,
      "step": 695
    },
    {
      "epoch": 0.1534137874028765,
      "grad_norm": 0.9302831888198853,
      "learning_rate": 0.00019423776066004747,
      "loss": 2.3452,
      "step": 696
    },
    {
      "epoch": 0.15363420951121398,
      "grad_norm": 0.8103287220001221,
      "learning_rate": 0.00019421334993585231,
      "loss": 2.3483,
      "step": 697
    },
    {
      "epoch": 0.15385463161955143,
      "grad_norm": 0.7609269022941589,
      "learning_rate": 0.0001941888891554854,
      "loss": 2.4062,
      "step": 698
    },
    {
      "epoch": 0.1540750537278889,
      "grad_norm": 0.7956115007400513,
      "learning_rate": 0.0001941643783319429,
      "loss": 2.3151,
      "step": 699
    },
    {
      "epoch": 0.15429547583622638,
      "grad_norm": 0.8774739503860474,
      "learning_rate": 0.0001941398174782476,
      "loss": 2.3707,
      "step": 700
    },
    {
      "epoch": 0.15451589794456383,
      "grad_norm": 0.9990003108978271,
      "learning_rate": 0.0001941152066074488,
      "loss": 2.3712,
      "step": 701
    },
    {
      "epoch": 0.1547363200529013,
      "grad_norm": 0.8081811666488647,
      "learning_rate": 0.00019409054573262248,
      "loss": 2.3437,
      "step": 702
    },
    {
      "epoch": 0.15495674216123878,
      "grad_norm": 0.8063721656799316,
      "learning_rate": 0.00019406583486687108,
      "loss": 2.2001,
      "step": 703
    },
    {
      "epoch": 0.15517716426957623,
      "grad_norm": 0.8059102892875671,
      "learning_rate": 0.00019404107402332368,
      "loss": 2.3131,
      "step": 704
    },
    {
      "epoch": 0.1553975863779137,
      "grad_norm": 0.9136509895324707,
      "learning_rate": 0.00019401626321513584,
      "loss": 2.3728,
      "step": 705
    },
    {
      "epoch": 0.15561800848625118,
      "grad_norm": 0.8856276273727417,
      "learning_rate": 0.00019399140245548972,
      "loss": 2.517,
      "step": 706
    },
    {
      "epoch": 0.15583843059458863,
      "grad_norm": 0.8584710359573364,
      "learning_rate": 0.000193966491757594,
      "loss": 2.2749,
      "step": 707
    },
    {
      "epoch": 0.1560588527029261,
      "grad_norm": 0.8318008184432983,
      "learning_rate": 0.0001939415311346839,
      "loss": 2.3542,
      "step": 708
    },
    {
      "epoch": 0.15627927481126358,
      "grad_norm": 0.858058750629425,
      "learning_rate": 0.00019391652060002116,
      "loss": 2.4127,
      "step": 709
    },
    {
      "epoch": 0.15649969691960103,
      "grad_norm": 0.8169947266578674,
      "learning_rate": 0.0001938914601668941,
      "loss": 2.2572,
      "step": 710
    },
    {
      "epoch": 0.1567201190279385,
      "grad_norm": 0.7780962586402893,
      "learning_rate": 0.0001938663498486174,
      "loss": 2.3755,
      "step": 711
    },
    {
      "epoch": 0.15694054113627598,
      "grad_norm": 0.8056047558784485,
      "learning_rate": 0.00019384118965853239,
      "loss": 2.3158,
      "step": 712
    },
    {
      "epoch": 0.15716096324461343,
      "grad_norm": 0.7953647375106812,
      "learning_rate": 0.00019381597961000681,
      "loss": 2.2682,
      "step": 713
    },
    {
      "epoch": 0.1573813853529509,
      "grad_norm": 0.7993754744529724,
      "learning_rate": 0.000193790719716435,
      "loss": 2.3406,
      "step": 714
    },
    {
      "epoch": 0.15760180746128838,
      "grad_norm": 0.8089137673377991,
      "learning_rate": 0.00019376540999123761,
      "loss": 2.2932,
      "step": 715
    },
    {
      "epoch": 0.15782222956962583,
      "grad_norm": 0.8295260071754456,
      "learning_rate": 0.0001937400504478619,
      "loss": 2.3535,
      "step": 716
    },
    {
      "epoch": 0.1580426516779633,
      "grad_norm": 0.8526468873023987,
      "learning_rate": 0.0001937146410997816,
      "loss": 2.4022,
      "step": 717
    },
    {
      "epoch": 0.15826307378630078,
      "grad_norm": 0.8820013999938965,
      "learning_rate": 0.0001936891819604968,
      "loss": 2.3563,
      "step": 718
    },
    {
      "epoch": 0.15848349589463823,
      "grad_norm": 0.8639278411865234,
      "learning_rate": 0.00019366367304353416,
      "loss": 2.3981,
      "step": 719
    },
    {
      "epoch": 0.1587039180029757,
      "grad_norm": 0.8538240194320679,
      "learning_rate": 0.00019363811436244675,
      "loss": 2.2617,
      "step": 720
    },
    {
      "epoch": 0.15892434011131318,
      "grad_norm": 0.8420394659042358,
      "learning_rate": 0.000193612505930814,
      "loss": 2.4381,
      "step": 721
    },
    {
      "epoch": 0.15914476221965063,
      "grad_norm": 0.8430236577987671,
      "learning_rate": 0.00019358684776224185,
      "loss": 2.3651,
      "step": 722
    },
    {
      "epoch": 0.1593651843279881,
      "grad_norm": 0.8562780022621155,
      "learning_rate": 0.0001935611398703627,
      "loss": 2.3283,
      "step": 723
    },
    {
      "epoch": 0.15958560643632555,
      "grad_norm": 0.9056751132011414,
      "learning_rate": 0.00019353538226883526,
      "loss": 2.3352,
      "step": 724
    },
    {
      "epoch": 0.15980602854466303,
      "grad_norm": 0.8337851762771606,
      "learning_rate": 0.00019350957497134478,
      "loss": 2.238,
      "step": 725
    },
    {
      "epoch": 0.1600264506530005,
      "grad_norm": 0.8333474397659302,
      "learning_rate": 0.00019348371799160282,
      "loss": 2.2724,
      "step": 726
    },
    {
      "epoch": 0.16024687276133795,
      "grad_norm": 0.8491765260696411,
      "learning_rate": 0.00019345781134334734,
      "loss": 2.4154,
      "step": 727
    },
    {
      "epoch": 0.16046729486967543,
      "grad_norm": 0.8710476756095886,
      "learning_rate": 0.00019343185504034277,
      "loss": 2.4053,
      "step": 728
    },
    {
      "epoch": 0.1606877169780129,
      "grad_norm": 0.8709803819656372,
      "learning_rate": 0.00019340584909637978,
      "loss": 2.4196,
      "step": 729
    },
    {
      "epoch": 0.16090813908635035,
      "grad_norm": 0.8141766786575317,
      "learning_rate": 0.00019337979352527555,
      "loss": 2.3171,
      "step": 730
    },
    {
      "epoch": 0.16112856119468782,
      "grad_norm": 0.7565165162086487,
      "learning_rate": 0.00019335368834087356,
      "loss": 2.2094,
      "step": 731
    },
    {
      "epoch": 0.1613489833030253,
      "grad_norm": 0.809673011302948,
      "learning_rate": 0.00019332753355704368,
      "loss": 2.3047,
      "step": 732
    },
    {
      "epoch": 0.16156940541136275,
      "grad_norm": 0.8433878421783447,
      "learning_rate": 0.00019330132918768216,
      "loss": 2.3446,
      "step": 733
    },
    {
      "epoch": 0.16178982751970022,
      "grad_norm": 0.8634533882141113,
      "learning_rate": 0.00019327507524671147,
      "loss": 2.3385,
      "step": 734
    },
    {
      "epoch": 0.1620102496280377,
      "grad_norm": 0.8720917105674744,
      "learning_rate": 0.00019324877174808054,
      "loss": 2.2881,
      "step": 735
    },
    {
      "epoch": 0.16223067173637515,
      "grad_norm": 0.8049135208129883,
      "learning_rate": 0.0001932224187057646,
      "loss": 2.3864,
      "step": 736
    },
    {
      "epoch": 0.16245109384471262,
      "grad_norm": 0.8659010529518127,
      "learning_rate": 0.00019319601613376514,
      "loss": 2.296,
      "step": 737
    },
    {
      "epoch": 0.1626715159530501,
      "grad_norm": 0.8483356833457947,
      "learning_rate": 0.00019316956404611012,
      "loss": 2.3094,
      "step": 738
    },
    {
      "epoch": 0.16289193806138755,
      "grad_norm": 0.8280044198036194,
      "learning_rate": 0.00019314306245685363,
      "loss": 2.2589,
      "step": 739
    },
    {
      "epoch": 0.16311236016972502,
      "grad_norm": 0.8388136029243469,
      "learning_rate": 0.00019311651138007618,
      "loss": 2.429,
      "step": 740
    },
    {
      "epoch": 0.1633327822780625,
      "grad_norm": 0.7954942584037781,
      "learning_rate": 0.0001930899108298845,
      "loss": 2.2437,
      "step": 741
    },
    {
      "epoch": 0.16355320438639995,
      "grad_norm": 0.7747330069541931,
      "learning_rate": 0.00019306326082041165,
      "loss": 2.3683,
      "step": 742
    },
    {
      "epoch": 0.16377362649473742,
      "grad_norm": 0.8457129001617432,
      "learning_rate": 0.00019303656136581695,
      "loss": 2.3067,
      "step": 743
    },
    {
      "epoch": 0.1639940486030749,
      "grad_norm": 0.8742474913597107,
      "learning_rate": 0.00019300981248028597,
      "loss": 2.3346,
      "step": 744
    },
    {
      "epoch": 0.16421447071141235,
      "grad_norm": 0.9028410315513611,
      "learning_rate": 0.00019298301417803066,
      "loss": 2.3207,
      "step": 745
    },
    {
      "epoch": 0.16443489281974982,
      "grad_norm": 0.8183161020278931,
      "learning_rate": 0.00019295616647328903,
      "loss": 2.2925,
      "step": 746
    },
    {
      "epoch": 0.1646553149280873,
      "grad_norm": 0.7908958196640015,
      "learning_rate": 0.00019292926938032552,
      "loss": 2.461,
      "step": 747
    },
    {
      "epoch": 0.16487573703642475,
      "grad_norm": 0.8609653115272522,
      "learning_rate": 0.00019290232291343067,
      "loss": 2.348,
      "step": 748
    },
    {
      "epoch": 0.16509615914476222,
      "grad_norm": 0.7725762724876404,
      "learning_rate": 0.00019287532708692134,
      "loss": 2.3751,
      "step": 749
    },
    {
      "epoch": 0.1653165812530997,
      "grad_norm": 0.8004394769668579,
      "learning_rate": 0.0001928482819151406,
      "loss": 2.466,
      "step": 750
    },
    {
      "epoch": 0.16553700336143715,
      "grad_norm": 0.7967116236686707,
      "learning_rate": 0.00019282118741245773,
      "loss": 2.459,
      "step": 751
    },
    {
      "epoch": 0.16575742546977462,
      "grad_norm": 0.775549590587616,
      "learning_rate": 0.0001927940435932682,
      "loss": 2.3904,
      "step": 752
    },
    {
      "epoch": 0.1659778475781121,
      "grad_norm": 0.7814269065856934,
      "learning_rate": 0.0001927668504719937,
      "loss": 2.3059,
      "step": 753
    },
    {
      "epoch": 0.16619826968644955,
      "grad_norm": 0.8013400435447693,
      "learning_rate": 0.00019273960806308213,
      "loss": 2.3636,
      "step": 754
    },
    {
      "epoch": 0.16641869179478702,
      "grad_norm": 0.7999504208564758,
      "learning_rate": 0.00019271231638100754,
      "loss": 2.407,
      "step": 755
    },
    {
      "epoch": 0.1666391139031245,
      "grad_norm": 0.8220574855804443,
      "learning_rate": 0.0001926849754402702,
      "loss": 2.2821,
      "step": 756
    },
    {
      "epoch": 0.16685953601146195,
      "grad_norm": 0.8551945090293884,
      "learning_rate": 0.00019265758525539652,
      "loss": 2.3535,
      "step": 757
    },
    {
      "epoch": 0.16707995811979942,
      "grad_norm": 0.7888833284378052,
      "learning_rate": 0.0001926301458409391,
      "loss": 2.2612,
      "step": 758
    },
    {
      "epoch": 0.16730038022813687,
      "grad_norm": 0.9234763979911804,
      "learning_rate": 0.00019260265721147664,
      "loss": 2.4261,
      "step": 759
    },
    {
      "epoch": 0.16752080233647434,
      "grad_norm": 0.8712771534919739,
      "learning_rate": 0.00019257511938161408,
      "loss": 2.2661,
      "step": 760
    },
    {
      "epoch": 0.16774122444481182,
      "grad_norm": 0.8544451594352722,
      "learning_rate": 0.00019254753236598243,
      "loss": 2.3091,
      "step": 761
    },
    {
      "epoch": 0.16796164655314927,
      "grad_norm": 0.8262011408805847,
      "learning_rate": 0.00019251989617923886,
      "loss": 2.3397,
      "step": 762
    },
    {
      "epoch": 0.16818206866148674,
      "grad_norm": 0.7879481315612793,
      "learning_rate": 0.00019249221083606662,
      "loss": 2.2715,
      "step": 763
    },
    {
      "epoch": 0.16840249076982422,
      "grad_norm": 0.8525984287261963,
      "learning_rate": 0.00019246447635117516,
      "loss": 2.3177,
      "step": 764
    },
    {
      "epoch": 0.16862291287816167,
      "grad_norm": 0.8159623742103577,
      "learning_rate": 0.00019243669273929996,
      "loss": 2.241,
      "step": 765
    },
    {
      "epoch": 0.16884333498649914,
      "grad_norm": 0.8777662515640259,
      "learning_rate": 0.00019240886001520267,
      "loss": 2.2473,
      "step": 766
    },
    {
      "epoch": 0.16906375709483662,
      "grad_norm": 0.8362585306167603,
      "learning_rate": 0.00019238097819367095,
      "loss": 2.2989,
      "step": 767
    },
    {
      "epoch": 0.16928417920317407,
      "grad_norm": 0.7509730458259583,
      "learning_rate": 0.00019235304728951866,
      "loss": 2.2742,
      "step": 768
    },
    {
      "epoch": 0.16950460131151154,
      "grad_norm": 0.8441339731216431,
      "learning_rate": 0.00019232506731758564,
      "loss": 2.3404,
      "step": 769
    },
    {
      "epoch": 0.16972502341984902,
      "grad_norm": 0.8148443698883057,
      "learning_rate": 0.00019229703829273781,
      "loss": 2.2353,
      "step": 770
    },
    {
      "epoch": 0.16994544552818647,
      "grad_norm": 0.8158761262893677,
      "learning_rate": 0.00019226896022986724,
      "loss": 2.3022,
      "step": 771
    },
    {
      "epoch": 0.17016586763652394,
      "grad_norm": 0.7531272768974304,
      "learning_rate": 0.00019224083314389192,
      "loss": 2.4138,
      "step": 772
    },
    {
      "epoch": 0.17038628974486142,
      "grad_norm": 0.8751167058944702,
      "learning_rate": 0.000192212657049756,
      "loss": 2.3149,
      "step": 773
    },
    {
      "epoch": 0.17060671185319887,
      "grad_norm": 0.8692744970321655,
      "learning_rate": 0.00019218443196242956,
      "loss": 2.3177,
      "step": 774
    },
    {
      "epoch": 0.17082713396153634,
      "grad_norm": 0.8779696226119995,
      "learning_rate": 0.00019215615789690887,
      "loss": 2.3094,
      "step": 775
    },
    {
      "epoch": 0.17104755606987382,
      "grad_norm": 0.8745964765548706,
      "learning_rate": 0.00019212783486821606,
      "loss": 2.259,
      "step": 776
    },
    {
      "epoch": 0.17126797817821127,
      "grad_norm": 0.8801835775375366,
      "learning_rate": 0.00019209946289139933,
      "loss": 2.3595,
      "step": 777
    },
    {
      "epoch": 0.17148840028654874,
      "grad_norm": 0.8479698300361633,
      "learning_rate": 0.00019207104198153295,
      "loss": 2.3773,
      "step": 778
    },
    {
      "epoch": 0.17170882239488622,
      "grad_norm": 0.7830620408058167,
      "learning_rate": 0.00019204257215371713,
      "loss": 2.3348,
      "step": 779
    },
    {
      "epoch": 0.17192924450322367,
      "grad_norm": 0.845029890537262,
      "learning_rate": 0.00019201405342307804,
      "loss": 2.3201,
      "step": 780
    },
    {
      "epoch": 0.17214966661156114,
      "grad_norm": 0.830081045627594,
      "learning_rate": 0.00019198548580476787,
      "loss": 2.3604,
      "step": 781
    },
    {
      "epoch": 0.17237008871989862,
      "grad_norm": 0.8488532304763794,
      "learning_rate": 0.00019195686931396485,
      "loss": 2.3065,
      "step": 782
    },
    {
      "epoch": 0.17259051082823607,
      "grad_norm": 0.8133013844490051,
      "learning_rate": 0.000191928203965873,
      "loss": 2.2839,
      "step": 783
    },
    {
      "epoch": 0.17281093293657354,
      "grad_norm": 0.7860105633735657,
      "learning_rate": 0.00019189948977572252,
      "loss": 2.2341,
      "step": 784
    },
    {
      "epoch": 0.17303135504491102,
      "grad_norm": 0.8637593388557434,
      "learning_rate": 0.0001918707267587694,
      "loss": 2.2886,
      "step": 785
    },
    {
      "epoch": 0.17325177715324847,
      "grad_norm": 0.7853116393089294,
      "learning_rate": 0.00019184191493029558,
      "loss": 2.3824,
      "step": 786
    },
    {
      "epoch": 0.17347219926158594,
      "grad_norm": 0.9432803392410278,
      "learning_rate": 0.00019181305430560902,
      "loss": 2.3719,
      "step": 787
    },
    {
      "epoch": 0.17369262136992342,
      "grad_norm": 0.874030351638794,
      "learning_rate": 0.00019178414490004356,
      "loss": 2.3882,
      "step": 788
    },
    {
      "epoch": 0.17391304347826086,
      "grad_norm": 0.8431480526924133,
      "learning_rate": 0.00019175518672895892,
      "loss": 2.3338,
      "step": 789
    },
    {
      "epoch": 0.17413346558659834,
      "grad_norm": 0.7761268615722656,
      "learning_rate": 0.0001917261798077408,
      "loss": 2.404,
      "step": 790
    },
    {
      "epoch": 0.1743538876949358,
      "grad_norm": 0.8312525153160095,
      "learning_rate": 0.0001916971241518008,
      "loss": 2.3407,
      "step": 791
    },
    {
      "epoch": 0.17457430980327326,
      "grad_norm": 0.8206736445426941,
      "learning_rate": 0.0001916680197765763,
      "loss": 2.3416,
      "step": 792
    },
    {
      "epoch": 0.17479473191161074,
      "grad_norm": 0.819446861743927,
      "learning_rate": 0.00019163886669753072,
      "loss": 2.3031,
      "step": 793
    },
    {
      "epoch": 0.1750151540199482,
      "grad_norm": 0.7587359547615051,
      "learning_rate": 0.00019160966493015323,
      "loss": 2.2409,
      "step": 794
    },
    {
      "epoch": 0.17523557612828566,
      "grad_norm": 0.800542414188385,
      "learning_rate": 0.00019158041448995894,
      "loss": 2.2921,
      "step": 795
    },
    {
      "epoch": 0.17545599823662314,
      "grad_norm": 0.8142752051353455,
      "learning_rate": 0.0001915511153924888,
      "loss": 2.2313,
      "step": 796
    },
    {
      "epoch": 0.1756764203449606,
      "grad_norm": 0.8178637027740479,
      "learning_rate": 0.0001915217676533096,
      "loss": 2.4049,
      "step": 797
    },
    {
      "epoch": 0.17589684245329806,
      "grad_norm": 0.8038167357444763,
      "learning_rate": 0.00019149237128801404,
      "loss": 2.2783,
      "step": 798
    },
    {
      "epoch": 0.17611726456163554,
      "grad_norm": 0.8934348821640015,
      "learning_rate": 0.00019146292631222055,
      "loss": 2.2502,
      "step": 799
    },
    {
      "epoch": 0.176337686669973,
      "grad_norm": 0.7946847677230835,
      "learning_rate": 0.00019143343274157345,
      "loss": 2.3777,
      "step": 800
    },
    {
      "epoch": 0.17655810877831046,
      "grad_norm": 0.9755481481552124,
      "learning_rate": 0.00019140389059174283,
      "loss": 2.3834,
      "step": 801
    },
    {
      "epoch": 0.17677853088664794,
      "grad_norm": 0.7911919951438904,
      "learning_rate": 0.00019137429987842473,
      "loss": 2.3946,
      "step": 802
    },
    {
      "epoch": 0.1769989529949854,
      "grad_norm": 0.8035269379615784,
      "learning_rate": 0.0001913446606173408,
      "loss": 2.2894,
      "step": 803
    },
    {
      "epoch": 0.17721937510332286,
      "grad_norm": 0.74834144115448,
      "learning_rate": 0.0001913149728242386,
      "loss": 2.2001,
      "step": 804
    },
    {
      "epoch": 0.17743979721166034,
      "grad_norm": 0.8397321105003357,
      "learning_rate": 0.00019128523651489146,
      "loss": 2.2035,
      "step": 805
    },
    {
      "epoch": 0.1776602193199978,
      "grad_norm": 0.7935678958892822,
      "learning_rate": 0.00019125545170509847,
      "loss": 2.2567,
      "step": 806
    },
    {
      "epoch": 0.17788064142833526,
      "grad_norm": 0.8398521542549133,
      "learning_rate": 0.0001912256184106845,
      "loss": 2.3047,
      "step": 807
    },
    {
      "epoch": 0.17810106353667274,
      "grad_norm": 0.873275637626648,
      "learning_rate": 0.00019119573664750018,
      "loss": 2.2223,
      "step": 808
    },
    {
      "epoch": 0.17832148564501019,
      "grad_norm": 0.7939439415931702,
      "learning_rate": 0.00019116580643142183,
      "loss": 2.3624,
      "step": 809
    },
    {
      "epoch": 0.17854190775334766,
      "grad_norm": 0.8182405829429626,
      "learning_rate": 0.00019113582777835166,
      "loss": 2.3823,
      "step": 810
    },
    {
      "epoch": 0.17876232986168514,
      "grad_norm": 0.8572868704795837,
      "learning_rate": 0.00019110580070421746,
      "loss": 2.412,
      "step": 811
    },
    {
      "epoch": 0.17898275197002259,
      "grad_norm": 0.8120009303092957,
      "learning_rate": 0.00019107572522497284,
      "loss": 2.3182,
      "step": 812
    },
    {
      "epoch": 0.17920317407836006,
      "grad_norm": 0.8325198292732239,
      "learning_rate": 0.0001910456013565971,
      "loss": 2.295,
      "step": 813
    },
    {
      "epoch": 0.17942359618669754,
      "grad_norm": 0.8716862201690674,
      "learning_rate": 0.00019101542911509524,
      "loss": 2.3022,
      "step": 814
    },
    {
      "epoch": 0.17964401829503499,
      "grad_norm": 0.8214452266693115,
      "learning_rate": 0.00019098520851649795,
      "loss": 2.2099,
      "step": 815
    },
    {
      "epoch": 0.17986444040337246,
      "grad_norm": 0.799541175365448,
      "learning_rate": 0.00019095493957686167,
      "loss": 2.3943,
      "step": 816
    },
    {
      "epoch": 0.18008486251170994,
      "grad_norm": 0.8183647990226746,
      "learning_rate": 0.00019092462231226848,
      "loss": 2.4086,
      "step": 817
    },
    {
      "epoch": 0.18030528462004738,
      "grad_norm": 0.8423382043838501,
      "learning_rate": 0.00019089425673882615,
      "loss": 2.4599,
      "step": 818
    },
    {
      "epoch": 0.18052570672838486,
      "grad_norm": 0.8709390759468079,
      "learning_rate": 0.00019086384287266808,
      "loss": 2.4651,
      "step": 819
    },
    {
      "epoch": 0.18074612883672234,
      "grad_norm": 0.8037587404251099,
      "learning_rate": 0.00019083338072995335,
      "loss": 2.3133,
      "step": 820
    },
    {
      "epoch": 0.18096655094505978,
      "grad_norm": 0.7483827471733093,
      "learning_rate": 0.00019080287032686673,
      "loss": 2.2269,
      "step": 821
    },
    {
      "epoch": 0.18118697305339726,
      "grad_norm": 0.8425301909446716,
      "learning_rate": 0.0001907723116796186,
      "loss": 2.3009,
      "step": 822
    },
    {
      "epoch": 0.18140739516173474,
      "grad_norm": 0.773988664150238,
      "learning_rate": 0.00019074170480444495,
      "loss": 2.2388,
      "step": 823
    },
    {
      "epoch": 0.18162781727007218,
      "grad_norm": 0.8526201248168945,
      "learning_rate": 0.0001907110497176074,
      "loss": 2.217,
      "step": 824
    },
    {
      "epoch": 0.18184823937840966,
      "grad_norm": 0.7970326542854309,
      "learning_rate": 0.00019068034643539324,
      "loss": 2.2387,
      "step": 825
    },
    {
      "epoch": 0.1820686614867471,
      "grad_norm": 0.7904958128929138,
      "learning_rate": 0.00019064959497411526,
      "loss": 2.2695,
      "step": 826
    },
    {
      "epoch": 0.18228908359508458,
      "grad_norm": 0.7954844236373901,
      "learning_rate": 0.00019061879535011198,
      "loss": 2.2528,
      "step": 827
    },
    {
      "epoch": 0.18250950570342206,
      "grad_norm": 0.7589882612228394,
      "learning_rate": 0.0001905879475797474,
      "loss": 2.3842,
      "step": 828
    },
    {
      "epoch": 0.1827299278117595,
      "grad_norm": 0.8610112071037292,
      "learning_rate": 0.0001905570516794112,
      "loss": 2.3223,
      "step": 829
    },
    {
      "epoch": 0.18295034992009698,
      "grad_norm": 0.8207979202270508,
      "learning_rate": 0.0001905261076655185,
      "loss": 2.5263,
      "step": 830
    },
    {
      "epoch": 0.18317077202843446,
      "grad_norm": 0.7460384964942932,
      "learning_rate": 0.0001904951155545101,
      "loss": 2.2545,
      "step": 831
    },
    {
      "epoch": 0.1833911941367719,
      "grad_norm": 0.8191743493080139,
      "learning_rate": 0.00019046407536285232,
      "loss": 2.3682,
      "step": 832
    },
    {
      "epoch": 0.18361161624510938,
      "grad_norm": 0.7235543727874756,
      "learning_rate": 0.00019043298710703701,
      "loss": 2.2548,
      "step": 833
    },
    {
      "epoch": 0.18383203835344686,
      "grad_norm": 0.8223721981048584,
      "learning_rate": 0.00019040185080358154,
      "loss": 2.4037,
      "step": 834
    },
    {
      "epoch": 0.1840524604617843,
      "grad_norm": 0.7922571897506714,
      "learning_rate": 0.00019037066646902887,
      "loss": 2.2866,
      "step": 835
    },
    {
      "epoch": 0.18427288257012178,
      "grad_norm": 0.8124774098396301,
      "learning_rate": 0.0001903394341199474,
      "loss": 2.2756,
      "step": 836
    },
    {
      "epoch": 0.18449330467845926,
      "grad_norm": 0.7676643133163452,
      "learning_rate": 0.0001903081537729311,
      "loss": 2.2692,
      "step": 837
    },
    {
      "epoch": 0.1847137267867967,
      "grad_norm": 0.8624024391174316,
      "learning_rate": 0.00019027682544459947,
      "loss": 2.3731,
      "step": 838
    },
    {
      "epoch": 0.18493414889513418,
      "grad_norm": 0.8140825033187866,
      "learning_rate": 0.0001902454491515974,
      "loss": 2.3336,
      "step": 839
    },
    {
      "epoch": 0.18515457100347166,
      "grad_norm": 0.8307098746299744,
      "learning_rate": 0.0001902140249105953,
      "loss": 2.3935,
      "step": 840
    },
    {
      "epoch": 0.1853749931118091,
      "grad_norm": 0.7764741778373718,
      "learning_rate": 0.00019018255273828914,
      "loss": 2.2595,
      "step": 841
    },
    {
      "epoch": 0.18559541522014658,
      "grad_norm": 0.843216598033905,
      "learning_rate": 0.00019015103265140025,
      "loss": 2.3666,
      "step": 842
    },
    {
      "epoch": 0.18581583732848406,
      "grad_norm": 0.8216284513473511,
      "learning_rate": 0.0001901194646666755,
      "loss": 2.2572,
      "step": 843
    },
    {
      "epoch": 0.1860362594368215,
      "grad_norm": 0.8269878625869751,
      "learning_rate": 0.00019008784880088713,
      "loss": 2.3362,
      "step": 844
    },
    {
      "epoch": 0.18625668154515898,
      "grad_norm": 0.8070132732391357,
      "learning_rate": 0.00019005618507083284,
      "loss": 2.4217,
      "step": 845
    },
    {
      "epoch": 0.18647710365349646,
      "grad_norm": 0.7605726718902588,
      "learning_rate": 0.00019002447349333586,
      "loss": 2.1693,
      "step": 846
    },
    {
      "epoch": 0.1866975257618339,
      "grad_norm": 0.7682884931564331,
      "learning_rate": 0.00018999271408524465,
      "loss": 2.266,
      "step": 847
    },
    {
      "epoch": 0.18691794787017138,
      "grad_norm": 0.8491914868354797,
      "learning_rate": 0.00018996090686343328,
      "loss": 2.353,
      "step": 848
    },
    {
      "epoch": 0.18713836997850886,
      "grad_norm": 0.7628803849220276,
      "learning_rate": 0.00018992905184480108,
      "loss": 2.251,
      "step": 849
    },
    {
      "epoch": 0.1873587920868463,
      "grad_norm": 0.7754936218261719,
      "learning_rate": 0.00018989714904627287,
      "loss": 2.2813,
      "step": 850
    },
    {
      "epoch": 0.18757921419518378,
      "grad_norm": 0.7823554873466492,
      "learning_rate": 0.00018986519848479877,
      "loss": 2.3317,
      "step": 851
    },
    {
      "epoch": 0.18779963630352126,
      "grad_norm": 0.853978157043457,
      "learning_rate": 0.00018983320017735438,
      "loss": 2.2676,
      "step": 852
    },
    {
      "epoch": 0.1880200584118587,
      "grad_norm": 0.7791696190834045,
      "learning_rate": 0.0001898011541409406,
      "loss": 2.3234,
      "step": 853
    },
    {
      "epoch": 0.18824048052019618,
      "grad_norm": 0.8516479134559631,
      "learning_rate": 0.00018976906039258365,
      "loss": 2.3393,
      "step": 854
    },
    {
      "epoch": 0.18846090262853366,
      "grad_norm": 0.7646697759628296,
      "learning_rate": 0.00018973691894933522,
      "loss": 2.2691,
      "step": 855
    },
    {
      "epoch": 0.1886813247368711,
      "grad_norm": 0.7396589517593384,
      "learning_rate": 0.0001897047298282722,
      "loss": 2.2969,
      "step": 856
    },
    {
      "epoch": 0.18890174684520858,
      "grad_norm": 0.8285578489303589,
      "learning_rate": 0.00018967249304649694,
      "loss": 2.29,
      "step": 857
    },
    {
      "epoch": 0.18912216895354603,
      "grad_norm": 0.849145233631134,
      "learning_rate": 0.000189640208621137,
      "loss": 2.2528,
      "step": 858
    },
    {
      "epoch": 0.1893425910618835,
      "grad_norm": 0.7979539036750793,
      "learning_rate": 0.00018960787656934538,
      "loss": 2.3365,
      "step": 859
    },
    {
      "epoch": 0.18956301317022098,
      "grad_norm": 0.7297232151031494,
      "learning_rate": 0.00018957549690830022,
      "loss": 2.2997,
      "step": 860
    },
    {
      "epoch": 0.18978343527855843,
      "grad_norm": 0.8183102011680603,
      "learning_rate": 0.00018954306965520513,
      "loss": 2.3243,
      "step": 861
    },
    {
      "epoch": 0.1900038573868959,
      "grad_norm": 0.8105300664901733,
      "learning_rate": 0.00018951059482728884,
      "loss": 2.2573,
      "step": 862
    },
    {
      "epoch": 0.19022427949523338,
      "grad_norm": 0.7869380116462708,
      "learning_rate": 0.00018947807244180554,
      "loss": 2.3205,
      "step": 863
    },
    {
      "epoch": 0.19044470160357083,
      "grad_norm": 0.8360190391540527,
      "learning_rate": 0.0001894455025160345,
      "loss": 2.2892,
      "step": 864
    },
    {
      "epoch": 0.1906651237119083,
      "grad_norm": 0.7873448729515076,
      "learning_rate": 0.00018941288506728037,
      "loss": 2.2769,
      "step": 865
    },
    {
      "epoch": 0.19088554582024578,
      "grad_norm": 0.7672109603881836,
      "learning_rate": 0.00018938022011287303,
      "loss": 2.2896,
      "step": 866
    },
    {
      "epoch": 0.19110596792858323,
      "grad_norm": 0.7979982495307922,
      "learning_rate": 0.00018934750767016754,
      "loss": 2.3412,
      "step": 867
    },
    {
      "epoch": 0.1913263900369207,
      "grad_norm": 0.8764357566833496,
      "learning_rate": 0.0001893147477565443,
      "loss": 2.1611,
      "step": 868
    },
    {
      "epoch": 0.19154681214525818,
      "grad_norm": 0.7849046587944031,
      "learning_rate": 0.00018928194038940883,
      "loss": 2.2158,
      "step": 869
    },
    {
      "epoch": 0.19176723425359563,
      "grad_norm": 0.7107301950454712,
      "learning_rate": 0.00018924908558619186,
      "loss": 2.2763,
      "step": 870
    },
    {
      "epoch": 0.1919876563619331,
      "grad_norm": 0.8673571348190308,
      "learning_rate": 0.00018921618336434945,
      "loss": 2.4213,
      "step": 871
    },
    {
      "epoch": 0.19220807847027058,
      "grad_norm": 0.789853572845459,
      "learning_rate": 0.00018918323374136269,
      "loss": 2.2956,
      "step": 872
    },
    {
      "epoch": 0.19242850057860802,
      "grad_norm": 0.8626810908317566,
      "learning_rate": 0.00018915023673473798,
      "loss": 2.3492,
      "step": 873
    },
    {
      "epoch": 0.1926489226869455,
      "grad_norm": 0.843966007232666,
      "learning_rate": 0.0001891171923620068,
      "loss": 2.2976,
      "step": 874
    },
    {
      "epoch": 0.19286934479528298,
      "grad_norm": 0.8234602212905884,
      "learning_rate": 0.0001890841006407259,
      "loss": 2.3567,
      "step": 875
    },
    {
      "epoch": 0.19308976690362042,
      "grad_norm": 0.7797042727470398,
      "learning_rate": 0.00018905096158847707,
      "loss": 2.2247,
      "step": 876
    },
    {
      "epoch": 0.1933101890119579,
      "grad_norm": 0.7938992381095886,
      "learning_rate": 0.00018901777522286735,
      "loss": 2.3042,
      "step": 877
    },
    {
      "epoch": 0.19353061112029538,
      "grad_norm": 0.8011882901191711,
      "learning_rate": 0.00018898454156152886,
      "loss": 2.2033,
      "step": 878
    },
    {
      "epoch": 0.19375103322863282,
      "grad_norm": 0.7793745994567871,
      "learning_rate": 0.00018895126062211888,
      "loss": 2.3498,
      "step": 879
    },
    {
      "epoch": 0.1939714553369703,
      "grad_norm": 0.8767029047012329,
      "learning_rate": 0.00018891793242231976,
      "loss": 2.1941,
      "step": 880
    },
    {
      "epoch": 0.19419187744530778,
      "grad_norm": 0.7984660267829895,
      "learning_rate": 0.00018888455697983902,
      "loss": 2.3806,
      "step": 881
    },
    {
      "epoch": 0.19441229955364522,
      "grad_norm": 0.8204585313796997,
      "learning_rate": 0.00018885113431240923,
      "loss": 2.2936,
      "step": 882
    },
    {
      "epoch": 0.1946327216619827,
      "grad_norm": 0.8303782939910889,
      "learning_rate": 0.00018881766443778805,
      "loss": 2.308,
      "step": 883
    },
    {
      "epoch": 0.19485314377032018,
      "grad_norm": 0.7725619673728943,
      "learning_rate": 0.0001887841473737583,
      "loss": 2.2228,
      "step": 884
    },
    {
      "epoch": 0.19507356587865762,
      "grad_norm": 0.7900967001914978,
      "learning_rate": 0.00018875058313812777,
      "loss": 2.2457,
      "step": 885
    },
    {
      "epoch": 0.1952939879869951,
      "grad_norm": 0.8020488023757935,
      "learning_rate": 0.00018871697174872943,
      "loss": 2.2846,
      "step": 886
    },
    {
      "epoch": 0.19551441009533257,
      "grad_norm": 0.8442305326461792,
      "learning_rate": 0.00018868331322342114,
      "loss": 2.3174,
      "step": 887
    },
    {
      "epoch": 0.19573483220367002,
      "grad_norm": 0.827603816986084,
      "learning_rate": 0.00018864960758008592,
      "loss": 2.3095,
      "step": 888
    },
    {
      "epoch": 0.1959552543120075,
      "grad_norm": 0.7261413931846619,
      "learning_rate": 0.00018861585483663187,
      "loss": 2.2728,
      "step": 889
    },
    {
      "epoch": 0.19617567642034497,
      "grad_norm": 0.8733895421028137,
      "learning_rate": 0.00018858205501099193,
      "loss": 2.3223,
      "step": 890
    },
    {
      "epoch": 0.19639609852868242,
      "grad_norm": 0.8804321885108948,
      "learning_rate": 0.00018854820812112428,
      "loss": 2.3096,
      "step": 891
    },
    {
      "epoch": 0.1966165206370199,
      "grad_norm": 0.8966788649559021,
      "learning_rate": 0.00018851431418501191,
      "loss": 2.2747,
      "step": 892
    },
    {
      "epoch": 0.19683694274535735,
      "grad_norm": 0.8415718078613281,
      "learning_rate": 0.00018848037322066294,
      "loss": 2.2898,
      "step": 893
    },
    {
      "epoch": 0.19705736485369482,
      "grad_norm": 0.8438727259635925,
      "learning_rate": 0.00018844638524611042,
      "loss": 2.2855,
      "step": 894
    },
    {
      "epoch": 0.1972777869620323,
      "grad_norm": 0.8360188603401184,
      "learning_rate": 0.0001884123502794124,
      "loss": 2.3387,
      "step": 895
    },
    {
      "epoch": 0.19749820907036975,
      "grad_norm": 0.8791637420654297,
      "learning_rate": 0.0001883782683386518,
      "loss": 2.2719,
      "step": 896
    },
    {
      "epoch": 0.19771863117870722,
      "grad_norm": 0.911509096622467,
      "learning_rate": 0.0001883441394419367,
      "loss": 2.3326,
      "step": 897
    },
    {
      "epoch": 0.1979390532870447,
      "grad_norm": 0.8057364821434021,
      "learning_rate": 0.0001883099636073999,
      "loss": 2.3631,
      "step": 898
    },
    {
      "epoch": 0.19815947539538215,
      "grad_norm": 0.7934995293617249,
      "learning_rate": 0.0001882757408531993,
      "loss": 2.2726,
      "step": 899
    },
    {
      "epoch": 0.19837989750371962,
      "grad_norm": 0.8284590840339661,
      "learning_rate": 0.0001882414711975177,
      "loss": 2.299,
      "step": 900
    },
    {
      "epoch": 0.19837989750371962,
      "eval_bpb": 1.1919259744439494,
      "eval_loss": 2.2966935634613037,
      "eval_runtime": 184.3018,
      "eval_samples_per_second": 62.181,
      "eval_steps_per_second": 7.775,
      "step": 900
    },
    {
      "epoch": 0.1986003196120571,
      "grad_norm": 0.8964036107063293,
      "learning_rate": 0.0001882071546585627,
      "loss": 2.3859,
      "step": 901
    },
    {
      "epoch": 0.19882074172039454,
      "grad_norm": 0.8807835578918457,
      "learning_rate": 0.00018817279125456698,
      "loss": 2.278,
      "step": 902
    },
    {
      "epoch": 0.19904116382873202,
      "grad_norm": 0.8288506865501404,
      "learning_rate": 0.00018813838100378806,
      "loss": 2.2565,
      "step": 903
    },
    {
      "epoch": 0.1992615859370695,
      "grad_norm": 0.8622190952301025,
      "learning_rate": 0.00018810392392450826,
      "loss": 2.2836,
      "step": 904
    },
    {
      "epoch": 0.19948200804540694,
      "grad_norm": 0.7835306525230408,
      "learning_rate": 0.0001880694200350349,
      "loss": 2.4251,
      "step": 905
    },
    {
      "epoch": 0.19970243015374442,
      "grad_norm": 0.8808237314224243,
      "learning_rate": 0.0001880348693537001,
      "loss": 2.4866,
      "step": 906
    },
    {
      "epoch": 0.1999228522620819,
      "grad_norm": 0.7805156111717224,
      "learning_rate": 0.0001880002718988609,
      "loss": 2.2724,
      "step": 907
    },
    {
      "epoch": 0.20014327437041934,
      "grad_norm": 0.8006852865219116,
      "learning_rate": 0.00018796562768889913,
      "loss": 2.276,
      "step": 908
    },
    {
      "epoch": 0.20036369647875682,
      "grad_norm": 0.7996565699577332,
      "learning_rate": 0.00018793093674222144,
      "loss": 2.2788,
      "step": 909
    },
    {
      "epoch": 0.2005841185870943,
      "grad_norm": 0.7668651342391968,
      "learning_rate": 0.00018789619907725943,
      "loss": 2.2819,
      "step": 910
    },
    {
      "epoch": 0.20080454069543174,
      "grad_norm": 0.8454254865646362,
      "learning_rate": 0.00018786141471246944,
      "loss": 2.3467,
      "step": 911
    },
    {
      "epoch": 0.20102496280376922,
      "grad_norm": 0.8126739263534546,
      "learning_rate": 0.0001878265836663326,
      "loss": 2.2591,
      "step": 912
    },
    {
      "epoch": 0.2012453849121067,
      "grad_norm": 0.8330004811286926,
      "learning_rate": 0.0001877917059573549,
      "loss": 2.3688,
      "step": 913
    },
    {
      "epoch": 0.20146580702044414,
      "grad_norm": 0.7879951000213623,
      "learning_rate": 0.00018775678160406705,
      "loss": 2.3432,
      "step": 914
    },
    {
      "epoch": 0.20168622912878162,
      "grad_norm": 0.735474169254303,
      "learning_rate": 0.00018772181062502464,
      "loss": 2.2852,
      "step": 915
    },
    {
      "epoch": 0.2019066512371191,
      "grad_norm": 0.7818914651870728,
      "learning_rate": 0.000187686793038808,
      "loss": 2.2973,
      "step": 916
    },
    {
      "epoch": 0.20212707334545654,
      "grad_norm": 0.7712001800537109,
      "learning_rate": 0.0001876517288640221,
      "loss": 2.1616,
      "step": 917
    },
    {
      "epoch": 0.20234749545379402,
      "grad_norm": 0.8388001322746277,
      "learning_rate": 0.00018761661811929686,
      "loss": 2.3489,
      "step": 918
    },
    {
      "epoch": 0.2025679175621315,
      "grad_norm": 0.858984112739563,
      "learning_rate": 0.00018758146082328687,
      "loss": 2.2773,
      "step": 919
    },
    {
      "epoch": 0.20278833967046894,
      "grad_norm": 0.8120707869529724,
      "learning_rate": 0.00018754625699467136,
      "loss": 2.3109,
      "step": 920
    },
    {
      "epoch": 0.20300876177880642,
      "grad_norm": 0.7937056422233582,
      "learning_rate": 0.00018751100665215437,
      "loss": 2.3226,
      "step": 921
    },
    {
      "epoch": 0.2032291838871439,
      "grad_norm": 0.7561761736869812,
      "learning_rate": 0.00018747570981446466,
      "loss": 2.399,
      "step": 922
    },
    {
      "epoch": 0.20344960599548134,
      "grad_norm": 0.7496448755264282,
      "learning_rate": 0.00018744036650035563,
      "loss": 2.346,
      "step": 923
    },
    {
      "epoch": 0.20367002810381882,
      "grad_norm": 0.8264040946960449,
      "learning_rate": 0.00018740497672860548,
      "loss": 2.222,
      "step": 924
    },
    {
      "epoch": 0.20389045021215627,
      "grad_norm": 0.769943118095398,
      "learning_rate": 0.00018736954051801698,
      "loss": 2.2315,
      "step": 925
    },
    {
      "epoch": 0.20411087232049374,
      "grad_norm": 0.7803930640220642,
      "learning_rate": 0.00018733405788741764,
      "loss": 2.257,
      "step": 926
    },
    {
      "epoch": 0.20433129442883122,
      "grad_norm": 0.7363802790641785,
      "learning_rate": 0.00018729852885565962,
      "loss": 2.3228,
      "step": 927
    },
    {
      "epoch": 0.20455171653716867,
      "grad_norm": 0.7507681250572205,
      "learning_rate": 0.0001872629534416197,
      "loss": 2.2299,
      "step": 928
    },
    {
      "epoch": 0.20477213864550614,
      "grad_norm": 0.7167561054229736,
      "learning_rate": 0.00018722733166419937,
      "loss": 2.2205,
      "step": 929
    },
    {
      "epoch": 0.20499256075384362,
      "grad_norm": 0.7825846076011658,
      "learning_rate": 0.00018719166354232467,
      "loss": 2.2808,
      "step": 930
    },
    {
      "epoch": 0.20521298286218106,
      "grad_norm": 0.8400839567184448,
      "learning_rate": 0.00018715594909494634,
      "loss": 2.3528,
      "step": 931
    },
    {
      "epoch": 0.20543340497051854,
      "grad_norm": 0.845869243144989,
      "learning_rate": 0.00018712018834103972,
      "loss": 2.1727,
      "step": 932
    },
    {
      "epoch": 0.20565382707885602,
      "grad_norm": 0.8175389766693115,
      "learning_rate": 0.00018708438129960467,
      "loss": 2.3137,
      "step": 933
    },
    {
      "epoch": 0.20587424918719346,
      "grad_norm": 0.7702092528343201,
      "learning_rate": 0.00018704852798966576,
      "loss": 2.323,
      "step": 934
    },
    {
      "epoch": 0.20609467129553094,
      "grad_norm": 0.7667554616928101,
      "learning_rate": 0.00018701262843027209,
      "loss": 2.1894,
      "step": 935
    },
    {
      "epoch": 0.20631509340386842,
      "grad_norm": 0.8353850841522217,
      "learning_rate": 0.0001869766826404973,
      "loss": 2.3326,
      "step": 936
    },
    {
      "epoch": 0.20653551551220586,
      "grad_norm": 0.7595258355140686,
      "learning_rate": 0.0001869406906394396,
      "loss": 2.2425,
      "step": 937
    },
    {
      "epoch": 0.20675593762054334,
      "grad_norm": 0.8109163641929626,
      "learning_rate": 0.00018690465244622183,
      "loss": 2.3114,
      "step": 938
    },
    {
      "epoch": 0.20697635972888082,
      "grad_norm": 0.8757756352424622,
      "learning_rate": 0.0001868685680799913,
      "loss": 2.3025,
      "step": 939
    },
    {
      "epoch": 0.20719678183721826,
      "grad_norm": 0.806124210357666,
      "learning_rate": 0.0001868324375599199,
      "loss": 2.2566,
      "step": 940
    },
    {
      "epoch": 0.20741720394555574,
      "grad_norm": 0.820923924446106,
      "learning_rate": 0.000186796260905204,
      "loss": 2.2311,
      "step": 941
    },
    {
      "epoch": 0.20763762605389322,
      "grad_norm": 0.7700275778770447,
      "learning_rate": 0.00018676003813506445,
      "loss": 2.4598,
      "step": 942
    },
    {
      "epoch": 0.20785804816223066,
      "grad_norm": 0.7977682948112488,
      "learning_rate": 0.00018672376926874668,
      "loss": 2.3088,
      "step": 943
    },
    {
      "epoch": 0.20807847027056814,
      "grad_norm": 0.8273078203201294,
      "learning_rate": 0.00018668745432552056,
      "loss": 2.2111,
      "step": 944
    },
    {
      "epoch": 0.20829889237890561,
      "grad_norm": 0.7400541305541992,
      "learning_rate": 0.0001866510933246805,
      "loss": 2.3631,
      "step": 945
    },
    {
      "epoch": 0.20851931448724306,
      "grad_norm": 0.7759760618209839,
      "learning_rate": 0.00018661468628554528,
      "loss": 2.2292,
      "step": 946
    },
    {
      "epoch": 0.20873973659558054,
      "grad_norm": 0.8207471370697021,
      "learning_rate": 0.00018657823322745825,
      "loss": 2.2371,
      "step": 947
    },
    {
      "epoch": 0.20896015870391801,
      "grad_norm": 0.7618723511695862,
      "learning_rate": 0.00018654173416978714,
      "loss": 2.2491,
      "step": 948
    },
    {
      "epoch": 0.20918058081225546,
      "grad_norm": 0.7443767786026001,
      "learning_rate": 0.0001865051891319241,
      "loss": 2.3405,
      "step": 949
    },
    {
      "epoch": 0.20940100292059294,
      "grad_norm": 0.8049965500831604,
      "learning_rate": 0.00018646859813328584,
      "loss": 2.2261,
      "step": 950
    },
    {
      "epoch": 0.2096214250289304,
      "grad_norm": 0.7891556620597839,
      "learning_rate": 0.0001864319611933133,
      "loss": 2.2075,
      "step": 951
    },
    {
      "epoch": 0.20984184713726786,
      "grad_norm": 0.7497437596321106,
      "learning_rate": 0.000186395278331472,
      "loss": 2.2222,
      "step": 952
    },
    {
      "epoch": 0.21006226924560534,
      "grad_norm": 0.7408785223960876,
      "learning_rate": 0.00018635854956725173,
      "loss": 2.2409,
      "step": 953
    },
    {
      "epoch": 0.2102826913539428,
      "grad_norm": 0.7750483155250549,
      "learning_rate": 0.00018632177492016675,
      "loss": 2.2812,
      "step": 954
    },
    {
      "epoch": 0.21050311346228026,
      "grad_norm": 0.7849144339561462,
      "learning_rate": 0.0001862849544097557,
      "loss": 2.3125,
      "step": 955
    },
    {
      "epoch": 0.21072353557061774,
      "grad_norm": 0.7638430595397949,
      "learning_rate": 0.0001862480880555815,
      "loss": 2.2008,
      "step": 956
    },
    {
      "epoch": 0.2109439576789552,
      "grad_norm": 0.7880803942680359,
      "learning_rate": 0.00018621117587723154,
      "loss": 2.3281,
      "step": 957
    },
    {
      "epoch": 0.21116437978729266,
      "grad_norm": 0.7419774532318115,
      "learning_rate": 0.00018617421789431747,
      "loss": 2.2351,
      "step": 958
    },
    {
      "epoch": 0.21138480189563014,
      "grad_norm": 0.8148466348648071,
      "learning_rate": 0.00018613721412647532,
      "loss": 2.1873,
      "step": 959
    },
    {
      "epoch": 0.21160522400396758,
      "grad_norm": 0.7992485165596008,
      "learning_rate": 0.00018610016459336544,
      "loss": 2.2217,
      "step": 960
    },
    {
      "epoch": 0.21182564611230506,
      "grad_norm": 0.7942092418670654,
      "learning_rate": 0.00018606306931467247,
      "loss": 2.303,
      "step": 961
    },
    {
      "epoch": 0.21204606822064254,
      "grad_norm": 0.7458913922309875,
      "learning_rate": 0.0001860259283101054,
      "loss": 2.1866,
      "step": 962
    },
    {
      "epoch": 0.21226649032897998,
      "grad_norm": 0.8077496886253357,
      "learning_rate": 0.00018598874159939747,
      "loss": 2.2566,
      "step": 963
    },
    {
      "epoch": 0.21248691243731746,
      "grad_norm": 0.7780013084411621,
      "learning_rate": 0.00018595150920230625,
      "loss": 2.2529,
      "step": 964
    },
    {
      "epoch": 0.21270733454565494,
      "grad_norm": 0.867590606212616,
      "learning_rate": 0.0001859142311386135,
      "loss": 2.35,
      "step": 965
    },
    {
      "epoch": 0.21292775665399238,
      "grad_norm": 0.8436008095741272,
      "learning_rate": 0.0001858769074281254,
      "loss": 2.173,
      "step": 966
    },
    {
      "epoch": 0.21314817876232986,
      "grad_norm": 0.9141267538070679,
      "learning_rate": 0.00018583953809067218,
      "loss": 2.1887,
      "step": 967
    },
    {
      "epoch": 0.21336860087066734,
      "grad_norm": 0.8681933879852295,
      "learning_rate": 0.00018580212314610846,
      "loss": 2.2596,
      "step": 968
    },
    {
      "epoch": 0.21358902297900478,
      "grad_norm": 0.9017385840415955,
      "learning_rate": 0.00018576466261431307,
      "loss": 2.4138,
      "step": 969
    },
    {
      "epoch": 0.21380944508734226,
      "grad_norm": 0.9393825531005859,
      "learning_rate": 0.00018572715651518897,
      "loss": 2.3856,
      "step": 970
    },
    {
      "epoch": 0.21402986719567973,
      "grad_norm": 0.792549192905426,
      "learning_rate": 0.00018568960486866343,
      "loss": 2.2931,
      "step": 971
    },
    {
      "epoch": 0.21425028930401718,
      "grad_norm": 0.8979101777076721,
      "learning_rate": 0.00018565200769468792,
      "loss": 2.2509,
      "step": 972
    },
    {
      "epoch": 0.21447071141235466,
      "grad_norm": 0.8961436748504639,
      "learning_rate": 0.00018561436501323805,
      "loss": 2.3015,
      "step": 973
    },
    {
      "epoch": 0.21469113352069213,
      "grad_norm": 0.9042580723762512,
      "learning_rate": 0.0001855766768443136,
      "loss": 2.3762,
      "step": 974
    },
    {
      "epoch": 0.21491155562902958,
      "grad_norm": 0.8965139389038086,
      "learning_rate": 0.00018553894320793858,
      "loss": 2.2861,
      "step": 975
    },
    {
      "epoch": 0.21513197773736706,
      "grad_norm": 0.8597142100334167,
      "learning_rate": 0.00018550116412416104,
      "loss": 2.195,
      "step": 976
    },
    {
      "epoch": 0.21535239984570453,
      "grad_norm": 0.800984263420105,
      "learning_rate": 0.00018546333961305333,
      "loss": 2.3491,
      "step": 977
    },
    {
      "epoch": 0.21557282195404198,
      "grad_norm": 0.8369321227073669,
      "learning_rate": 0.00018542546969471183,
      "loss": 2.2939,
      "step": 978
    },
    {
      "epoch": 0.21579324406237946,
      "grad_norm": 0.9097290635108948,
      "learning_rate": 0.0001853875543892571,
      "loss": 2.298,
      "step": 979
    },
    {
      "epoch": 0.21601366617071693,
      "grad_norm": 0.8758633732795715,
      "learning_rate": 0.00018534959371683377,
      "loss": 2.3401,
      "step": 980
    },
    {
      "epoch": 0.21623408827905438,
      "grad_norm": 0.7626383900642395,
      "learning_rate": 0.00018531158769761064,
      "loss": 2.2801,
      "step": 981
    },
    {
      "epoch": 0.21645451038739186,
      "grad_norm": 0.7398207187652588,
      "learning_rate": 0.0001852735363517805,
      "loss": 2.3974,
      "step": 982
    },
    {
      "epoch": 0.21667493249572933,
      "grad_norm": 0.8146309852600098,
      "learning_rate": 0.00018523543969956025,
      "loss": 2.2661,
      "step": 983
    },
    {
      "epoch": 0.21689535460406678,
      "grad_norm": 0.8620419502258301,
      "learning_rate": 0.00018519729776119096,
      "loss": 2.4034,
      "step": 984
    },
    {
      "epoch": 0.21711577671240426,
      "grad_norm": 0.7637450098991394,
      "learning_rate": 0.00018515911055693772,
      "loss": 2.2641,
      "step": 985
    },
    {
      "epoch": 0.21733619882074173,
      "grad_norm": 0.7800227403640747,
      "learning_rate": 0.00018512087810708955,
      "loss": 2.3006,
      "step": 986
    },
    {
      "epoch": 0.21755662092907918,
      "grad_norm": 0.810896635055542,
      "learning_rate": 0.00018508260043195967,
      "loss": 2.3514,
      "step": 987
    },
    {
      "epoch": 0.21777704303741666,
      "grad_norm": 0.9432726502418518,
      "learning_rate": 0.00018504427755188521,
      "loss": 2.1436,
      "step": 988
    },
    {
      "epoch": 0.21799746514575413,
      "grad_norm": 0.8213148713111877,
      "learning_rate": 0.0001850059094872274,
      "loss": 2.3022,
      "step": 989
    },
    {
      "epoch": 0.21821788725409158,
      "grad_norm": 0.7722258567810059,
      "learning_rate": 0.0001849674962583715,
      "loss": 2.2701,
      "step": 990
    },
    {
      "epoch": 0.21843830936242906,
      "grad_norm": 0.8392160534858704,
      "learning_rate": 0.0001849290378857266,
      "loss": 2.231,
      "step": 991
    },
    {
      "epoch": 0.2186587314707665,
      "grad_norm": 0.7816172242164612,
      "learning_rate": 0.0001848905343897259,
      "loss": 2.3206,
      "step": 992
    },
    {
      "epoch": 0.21887915357910398,
      "grad_norm": 0.8324223756790161,
      "learning_rate": 0.00018485198579082666,
      "loss": 2.2559,
      "step": 993
    },
    {
      "epoch": 0.21909957568744146,
      "grad_norm": 0.7789288759231567,
      "learning_rate": 0.0001848133921095099,
      "loss": 2.3013,
      "step": 994
    },
    {
      "epoch": 0.2193199977957789,
      "grad_norm": 0.817395806312561,
      "learning_rate": 0.00018477475336628078,
      "loss": 2.3448,
      "step": 995
    },
    {
      "epoch": 0.21954041990411638,
      "grad_norm": 0.7743339538574219,
      "learning_rate": 0.00018473606958166827,
      "loss": 2.1651,
      "step": 996
    },
    {
      "epoch": 0.21976084201245386,
      "grad_norm": 0.8676672577857971,
      "learning_rate": 0.00018469734077622532,
      "loss": 2.1935,
      "step": 997
    },
    {
      "epoch": 0.2199812641207913,
      "grad_norm": 0.7683441638946533,
      "learning_rate": 0.0001846585669705288,
      "loss": 2.2333,
      "step": 998
    },
    {
      "epoch": 0.22020168622912878,
      "grad_norm": 0.7916432619094849,
      "learning_rate": 0.00018461974818517946,
      "loss": 2.3397,
      "step": 999
    },
    {
      "epoch": 0.22042210833746625,
      "grad_norm": 0.8173052668571472,
      "learning_rate": 0.00018458088444080203,
      "loss": 2.2488,
      "step": 1000
    },
    {
      "epoch": 0.2206425304458037,
      "grad_norm": 0.7672227025032043,
      "learning_rate": 0.00018454197575804506,
      "loss": 2.3537,
      "step": 1001
    },
    {
      "epoch": 0.22086295255414118,
      "grad_norm": 0.8164885640144348,
      "learning_rate": 0.00018450302215758098,
      "loss": 2.2785,
      "step": 1002
    },
    {
      "epoch": 0.22108337466247865,
      "grad_norm": 0.7681257724761963,
      "learning_rate": 0.00018446402366010606,
      "loss": 2.151,
      "step": 1003
    },
    {
      "epoch": 0.2213037967708161,
      "grad_norm": 0.8099995851516724,
      "learning_rate": 0.00018442498028634053,
      "loss": 2.2094,
      "step": 1004
    },
    {
      "epoch": 0.22152421887915358,
      "grad_norm": 0.7719568014144897,
      "learning_rate": 0.0001843858920570283,
      "loss": 2.2668,
      "step": 1005
    },
    {
      "epoch": 0.22174464098749105,
      "grad_norm": 0.7770114541053772,
      "learning_rate": 0.00018434675899293732,
      "loss": 2.4026,
      "step": 1006
    },
    {
      "epoch": 0.2219650630958285,
      "grad_norm": 0.7813636064529419,
      "learning_rate": 0.00018430758111485915,
      "loss": 2.2608,
      "step": 1007
    },
    {
      "epoch": 0.22218548520416598,
      "grad_norm": 0.7801280617713928,
      "learning_rate": 0.00018426835844360929,
      "loss": 2.2522,
      "step": 1008
    },
    {
      "epoch": 0.22240590731250345,
      "grad_norm": 0.7802456617355347,
      "learning_rate": 0.00018422909100002696,
      "loss": 2.263,
      "step": 1009
    },
    {
      "epoch": 0.2226263294208409,
      "grad_norm": 0.770309567451477,
      "learning_rate": 0.00018418977880497528,
      "loss": 2.2345,
      "step": 1010
    },
    {
      "epoch": 0.22284675152917838,
      "grad_norm": 0.7488002777099609,
      "learning_rate": 0.000184150421879341,
      "loss": 2.2997,
      "step": 1011
    },
    {
      "epoch": 0.22306717363751585,
      "grad_norm": 0.8127683401107788,
      "learning_rate": 0.00018411102024403481,
      "loss": 2.3005,
      "step": 1012
    },
    {
      "epoch": 0.2232875957458533,
      "grad_norm": 0.8222635984420776,
      "learning_rate": 0.000184071573919991,
      "loss": 2.1503,
      "step": 1013
    },
    {
      "epoch": 0.22350801785419078,
      "grad_norm": 0.8823630809783936,
      "learning_rate": 0.00018403208292816766,
      "loss": 2.3232,
      "step": 1014
    },
    {
      "epoch": 0.22372843996252825,
      "grad_norm": 0.743665874004364,
      "learning_rate": 0.0001839925472895466,
      "loss": 2.1881,
      "step": 1015
    },
    {
      "epoch": 0.2239488620708657,
      "grad_norm": 0.7798638939857483,
      "learning_rate": 0.0001839529670251334,
      "loss": 2.291,
      "step": 1016
    },
    {
      "epoch": 0.22416928417920318,
      "grad_norm": 0.7869151830673218,
      "learning_rate": 0.00018391334215595725,
      "loss": 2.3823,
      "step": 1017
    },
    {
      "epoch": 0.22438970628754065,
      "grad_norm": 0.8263714909553528,
      "learning_rate": 0.0001838736727030712,
      "loss": 2.2246,
      "step": 1018
    },
    {
      "epoch": 0.2246101283958781,
      "grad_norm": 0.7697818279266357,
      "learning_rate": 0.00018383395868755184,
      "loss": 2.2319,
      "step": 1019
    },
    {
      "epoch": 0.22483055050421558,
      "grad_norm": 0.7541553974151611,
      "learning_rate": 0.00018379420013049948,
      "loss": 2.2503,
      "step": 1020
    },
    {
      "epoch": 0.22505097261255305,
      "grad_norm": 0.8404936194419861,
      "learning_rate": 0.00018375439705303813,
      "loss": 2.2284,
      "step": 1021
    },
    {
      "epoch": 0.2252713947208905,
      "grad_norm": 0.8484320044517517,
      "learning_rate": 0.00018371454947631536,
      "loss": 2.2389,
      "step": 1022
    },
    {
      "epoch": 0.22549181682922798,
      "grad_norm": 0.8099342584609985,
      "learning_rate": 0.00018367465742150253,
      "loss": 2.3368,
      "step": 1023
    },
    {
      "epoch": 0.22571223893756545,
      "grad_norm": 0.7582220435142517,
      "learning_rate": 0.00018363472090979452,
      "loss": 2.1727,
      "step": 1024
    },
    {
      "epoch": 0.2259326610459029,
      "grad_norm": 0.7743244767189026,
      "learning_rate": 0.00018359473996240985,
      "loss": 2.3503,
      "step": 1025
    },
    {
      "epoch": 0.22615308315424038,
      "grad_norm": 0.79170161485672,
      "learning_rate": 0.0001835547146005907,
      "loss": 2.2992,
      "step": 1026
    },
    {
      "epoch": 0.22637350526257782,
      "grad_norm": 0.9316833019256592,
      "learning_rate": 0.00018351464484560274,
      "loss": 2.2991,
      "step": 1027
    },
    {
      "epoch": 0.2265939273709153,
      "grad_norm": 0.7939894199371338,
      "learning_rate": 0.00018347453071873536,
      "loss": 2.2385,
      "step": 1028
    },
    {
      "epoch": 0.22681434947925277,
      "grad_norm": 0.793380856513977,
      "learning_rate": 0.0001834343722413014,
      "loss": 2.3193,
      "step": 1029
    },
    {
      "epoch": 0.22703477158759022,
      "grad_norm": 0.7591364979743958,
      "learning_rate": 0.00018339416943463735,
      "loss": 2.204,
      "step": 1030
    },
    {
      "epoch": 0.2272551936959277,
      "grad_norm": 0.7855841517448425,
      "learning_rate": 0.00018335392232010327,
      "loss": 2.3081,
      "step": 1031
    },
    {
      "epoch": 0.22747561580426517,
      "grad_norm": 0.8415141701698303,
      "learning_rate": 0.00018331363091908266,
      "loss": 2.4076,
      "step": 1032
    },
    {
      "epoch": 0.22769603791260262,
      "grad_norm": 0.8164039254188538,
      "learning_rate": 0.0001832732952529826,
      "loss": 2.2934,
      "step": 1033
    },
    {
      "epoch": 0.2279164600209401,
      "grad_norm": 0.7584627866744995,
      "learning_rate": 0.00018323291534323371,
      "loss": 2.2842,
      "step": 1034
    },
    {
      "epoch": 0.22813688212927757,
      "grad_norm": 0.755256175994873,
      "learning_rate": 0.00018319249121129014,
      "loss": 2.2056,
      "step": 1035
    },
    {
      "epoch": 0.22835730423761502,
      "grad_norm": 0.7633174061775208,
      "learning_rate": 0.00018315202287862945,
      "loss": 2.2232,
      "step": 1036
    },
    {
      "epoch": 0.2285777263459525,
      "grad_norm": 0.7563494443893433,
      "learning_rate": 0.00018311151036675273,
      "loss": 2.2454,
      "step": 1037
    },
    {
      "epoch": 0.22879814845428997,
      "grad_norm": 0.7570247054100037,
      "learning_rate": 0.00018307095369718456,
      "loss": 2.2423,
      "step": 1038
    },
    {
      "epoch": 0.22901857056262742,
      "grad_norm": 0.7824550867080688,
      "learning_rate": 0.000183030352891473,
      "loss": 2.2978,
      "step": 1039
    },
    {
      "epoch": 0.2292389926709649,
      "grad_norm": 0.7699748873710632,
      "learning_rate": 0.00018298970797118945,
      "loss": 2.2083,
      "step": 1040
    },
    {
      "epoch": 0.22945941477930237,
      "grad_norm": 0.7217649817466736,
      "learning_rate": 0.00018294901895792885,
      "loss": 2.3387,
      "step": 1041
    },
    {
      "epoch": 0.22967983688763982,
      "grad_norm": 0.7403888702392578,
      "learning_rate": 0.00018290828587330958,
      "loss": 2.2873,
      "step": 1042
    },
    {
      "epoch": 0.2299002589959773,
      "grad_norm": 0.7493407130241394,
      "learning_rate": 0.0001828675087389734,
      "loss": 2.2787,
      "step": 1043
    },
    {
      "epoch": 0.23012068110431477,
      "grad_norm": 0.8172875642776489,
      "learning_rate": 0.00018282668757658538,
      "loss": 2.427,
      "step": 1044
    },
    {
      "epoch": 0.23034110321265222,
      "grad_norm": 0.7300758957862854,
      "learning_rate": 0.00018278582240783414,
      "loss": 2.2555,
      "step": 1045
    },
    {
      "epoch": 0.2305615253209897,
      "grad_norm": 0.7548211812973022,
      "learning_rate": 0.0001827449132544316,
      "loss": 2.289,
      "step": 1046
    },
    {
      "epoch": 0.23078194742932717,
      "grad_norm": 0.7575384378433228,
      "learning_rate": 0.00018270396013811307,
      "loss": 2.1802,
      "step": 1047
    },
    {
      "epoch": 0.23100236953766462,
      "grad_norm": 0.740778923034668,
      "learning_rate": 0.00018266296308063718,
      "loss": 2.2069,
      "step": 1048
    },
    {
      "epoch": 0.2312227916460021,
      "grad_norm": 0.8235701322555542,
      "learning_rate": 0.00018262192210378596,
      "loss": 2.3309,
      "step": 1049
    },
    {
      "epoch": 0.23144321375433957,
      "grad_norm": 0.8264084458351135,
      "learning_rate": 0.00018258083722936478,
      "loss": 2.3386,
      "step": 1050
    },
    {
      "epoch": 0.23166363586267702,
      "grad_norm": 0.7833823561668396,
      "learning_rate": 0.00018253970847920226,
      "loss": 2.2715,
      "step": 1051
    },
    {
      "epoch": 0.2318840579710145,
      "grad_norm": 0.8685325384140015,
      "learning_rate": 0.00018249853587515037,
      "loss": 2.156,
      "step": 1052
    },
    {
      "epoch": 0.23210448007935197,
      "grad_norm": 0.745798647403717,
      "learning_rate": 0.00018245731943908444,
      "loss": 2.1415,
      "step": 1053
    },
    {
      "epoch": 0.23232490218768942,
      "grad_norm": 0.8023180365562439,
      "learning_rate": 0.00018241605919290296,
      "loss": 2.2938,
      "step": 1054
    },
    {
      "epoch": 0.2325453242960269,
      "grad_norm": 0.8139461874961853,
      "learning_rate": 0.00018237475515852782,
      "loss": 2.3204,
      "step": 1055
    },
    {
      "epoch": 0.23276574640436437,
      "grad_norm": 0.8700809478759766,
      "learning_rate": 0.00018233340735790413,
      "loss": 2.1617,
      "step": 1056
    },
    {
      "epoch": 0.23298616851270182,
      "grad_norm": 0.8505682945251465,
      "learning_rate": 0.00018229201581300018,
      "loss": 2.2838,
      "step": 1057
    },
    {
      "epoch": 0.2332065906210393,
      "grad_norm": 0.767876923084259,
      "learning_rate": 0.00018225058054580765,
      "loss": 2.1932,
      "step": 1058
    },
    {
      "epoch": 0.23342701272937674,
      "grad_norm": 0.8229421973228455,
      "learning_rate": 0.00018220910157834132,
      "loss": 2.2859,
      "step": 1059
    },
    {
      "epoch": 0.23364743483771422,
      "grad_norm": 0.8934770226478577,
      "learning_rate": 0.00018216757893263927,
      "loss": 2.2346,
      "step": 1060
    },
    {
      "epoch": 0.2338678569460517,
      "grad_norm": 0.8439505696296692,
      "learning_rate": 0.00018212601263076277,
      "loss": 2.2052,
      "step": 1061
    },
    {
      "epoch": 0.23408827905438914,
      "grad_norm": 0.7795543670654297,
      "learning_rate": 0.00018208440269479618,
      "loss": 2.2555,
      "step": 1062
    },
    {
      "epoch": 0.23430870116272662,
      "grad_norm": 0.7115225195884705,
      "learning_rate": 0.00018204274914684722,
      "loss": 2.2827,
      "step": 1063
    },
    {
      "epoch": 0.2345291232710641,
      "grad_norm": 0.8144369721412659,
      "learning_rate": 0.00018200105200904667,
      "loss": 2.1905,
      "step": 1064
    },
    {
      "epoch": 0.23474954537940154,
      "grad_norm": 0.8396324515342712,
      "learning_rate": 0.0001819593113035485,
      "loss": 2.1788,
      "step": 1065
    },
    {
      "epoch": 0.23496996748773902,
      "grad_norm": 0.7406641244888306,
      "learning_rate": 0.00018191752705252977,
      "loss": 2.3124,
      "step": 1066
    },
    {
      "epoch": 0.2351903895960765,
      "grad_norm": 0.8337415456771851,
      "learning_rate": 0.0001818756992781908,
      "loss": 2.285,
      "step": 1067
    },
    {
      "epoch": 0.23541081170441394,
      "grad_norm": 0.6862708926200867,
      "learning_rate": 0.0001818338280027549,
      "loss": 2.1953,
      "step": 1068
    },
    {
      "epoch": 0.23563123381275142,
      "grad_norm": 0.7343995571136475,
      "learning_rate": 0.00018179191324846862,
      "loss": 2.2832,
      "step": 1069
    },
    {
      "epoch": 0.2358516559210889,
      "grad_norm": 0.8205724358558655,
      "learning_rate": 0.00018174995503760148,
      "loss": 2.1541,
      "step": 1070
    },
    {
      "epoch": 0.23607207802942634,
      "grad_norm": 0.8589067459106445,
      "learning_rate": 0.00018170795339244618,
      "loss": 2.231,
      "step": 1071
    },
    {
      "epoch": 0.23629250013776382,
      "grad_norm": 0.8235065340995789,
      "learning_rate": 0.0001816659083353185,
      "loss": 2.1639,
      "step": 1072
    },
    {
      "epoch": 0.2365129222461013,
      "grad_norm": 0.8446851372718811,
      "learning_rate": 0.0001816238198885572,
      "loss": 2.1992,
      "step": 1073
    },
    {
      "epoch": 0.23673334435443874,
      "grad_norm": 0.8148869276046753,
      "learning_rate": 0.00018158168807452416,
      "loss": 2.1625,
      "step": 1074
    },
    {
      "epoch": 0.23695376646277622,
      "grad_norm": 0.7790500521659851,
      "learning_rate": 0.0001815395129156043,
      "loss": 2.3757,
      "step": 1075
    },
    {
      "epoch": 0.2371741885711137,
      "grad_norm": 0.789910614490509,
      "learning_rate": 0.0001814972944342056,
      "loss": 2.2046,
      "step": 1076
    },
    {
      "epoch": 0.23739461067945114,
      "grad_norm": 0.8793061971664429,
      "learning_rate": 0.0001814550326527589,
      "loss": 2.3194,
      "step": 1077
    },
    {
      "epoch": 0.23761503278778862,
      "grad_norm": 0.820549488067627,
      "learning_rate": 0.0001814127275937183,
      "loss": 2.2685,
      "step": 1078
    },
    {
      "epoch": 0.2378354548961261,
      "grad_norm": 0.8006362318992615,
      "learning_rate": 0.0001813703792795607,
      "loss": 2.3241,
      "step": 1079
    },
    {
      "epoch": 0.23805587700446354,
      "grad_norm": 0.7934444546699524,
      "learning_rate": 0.00018132798773278603,
      "loss": 2.2698,
      "step": 1080
    },
    {
      "epoch": 0.23827629911280102,
      "grad_norm": 0.7564600706100464,
      "learning_rate": 0.0001812855529759172,
      "loss": 2.0908,
      "step": 1081
    },
    {
      "epoch": 0.2384967212211385,
      "grad_norm": 0.8287239670753479,
      "learning_rate": 0.00018124307503150004,
      "loss": 2.2265,
      "step": 1082
    },
    {
      "epoch": 0.23871714332947594,
      "grad_norm": 0.8079926371574402,
      "learning_rate": 0.00018120055392210343,
      "loss": 2.3263,
      "step": 1083
    },
    {
      "epoch": 0.23893756543781342,
      "grad_norm": 0.827910840511322,
      "learning_rate": 0.00018115798967031907,
      "loss": 2.3119,
      "step": 1084
    },
    {
      "epoch": 0.2391579875461509,
      "grad_norm": 0.8228992819786072,
      "learning_rate": 0.00018111538229876163,
      "loss": 2.1814,
      "step": 1085
    },
    {
      "epoch": 0.23937840965448834,
      "grad_norm": 0.7576643228530884,
      "learning_rate": 0.00018107273183006872,
      "loss": 2.4035,
      "step": 1086
    },
    {
      "epoch": 0.23959883176282581,
      "grad_norm": 0.7380769848823547,
      "learning_rate": 0.00018103003828690077,
      "loss": 2.2226,
      "step": 1087
    },
    {
      "epoch": 0.2398192538711633,
      "grad_norm": 0.8089278936386108,
      "learning_rate": 0.00018098730169194117,
      "loss": 2.2578,
      "step": 1088
    },
    {
      "epoch": 0.24003967597950074,
      "grad_norm": 0.7937178015708923,
      "learning_rate": 0.00018094452206789612,
      "loss": 2.4655,
      "step": 1089
    },
    {
      "epoch": 0.24026009808783821,
      "grad_norm": 0.7379800081253052,
      "learning_rate": 0.00018090169943749476,
      "loss": 2.1528,
      "step": 1090
    },
    {
      "epoch": 0.2404805201961757,
      "grad_norm": 0.7448075413703918,
      "learning_rate": 0.00018085883382348898,
      "loss": 2.5696,
      "step": 1091
    },
    {
      "epoch": 0.24070094230451314,
      "grad_norm": 0.757524311542511,
      "learning_rate": 0.00018081592524865362,
      "loss": 2.3222,
      "step": 1092
    },
    {
      "epoch": 0.2409213644128506,
      "grad_norm": 0.7744084596633911,
      "learning_rate": 0.00018077297373578625,
      "loss": 2.2326,
      "step": 1093
    },
    {
      "epoch": 0.24114178652118806,
      "grad_norm": 0.7720345258712769,
      "learning_rate": 0.00018072997930770735,
      "loss": 2.2956,
      "step": 1094
    },
    {
      "epoch": 0.24136220862952554,
      "grad_norm": 0.7412413358688354,
      "learning_rate": 0.00018068694198726003,
      "loss": 2.2223,
      "step": 1095
    },
    {
      "epoch": 0.241582630737863,
      "grad_norm": 0.8095024824142456,
      "learning_rate": 0.00018064386179731038,
      "loss": 2.2284,
      "step": 1096
    },
    {
      "epoch": 0.24180305284620046,
      "grad_norm": 0.7581536173820496,
      "learning_rate": 0.0001806007387607472,
      "loss": 2.2834,
      "step": 1097
    },
    {
      "epoch": 0.24202347495453794,
      "grad_norm": 0.7913536429405212,
      "learning_rate": 0.00018055757290048202,
      "loss": 2.2437,
      "step": 1098
    },
    {
      "epoch": 0.2422438970628754,
      "grad_norm": 0.7251597046852112,
      "learning_rate": 0.00018051436423944912,
      "loss": 2.2727,
      "step": 1099
    },
    {
      "epoch": 0.24246431917121286,
      "grad_norm": 0.7658928632736206,
      "learning_rate": 0.00018047111280060558,
      "loss": 2.3461,
      "step": 1100
    },
    {
      "epoch": 0.24268474127955034,
      "grad_norm": 0.793036162853241,
      "learning_rate": 0.00018042781860693114,
      "loss": 2.2934,
      "step": 1101
    },
    {
      "epoch": 0.2429051633878878,
      "grad_norm": 0.7389605641365051,
      "learning_rate": 0.00018038448168142832,
      "loss": 2.3245,
      "step": 1102
    },
    {
      "epoch": 0.24312558549622526,
      "grad_norm": 0.7472773194313049,
      "learning_rate": 0.0001803411020471223,
      "loss": 2.2454,
      "step": 1103
    },
    {
      "epoch": 0.24334600760456274,
      "grad_norm": 0.7701893448829651,
      "learning_rate": 0.00018029767972706099,
      "loss": 2.3047,
      "step": 1104
    },
    {
      "epoch": 0.2435664297129002,
      "grad_norm": 0.8578325510025024,
      "learning_rate": 0.0001802542147443149,
      "loss": 2.089,
      "step": 1105
    },
    {
      "epoch": 0.24378685182123766,
      "grad_norm": 0.8135702610015869,
      "learning_rate": 0.00018021070712197733,
      "loss": 2.25,
      "step": 1106
    },
    {
      "epoch": 0.24400727392957514,
      "grad_norm": 0.7723056077957153,
      "learning_rate": 0.00018016715688316413,
      "loss": 2.3692,
      "step": 1107
    },
    {
      "epoch": 0.2442276960379126,
      "grad_norm": 0.8015069961547852,
      "learning_rate": 0.0001801235640510138,
      "loss": 2.3804,
      "step": 1108
    },
    {
      "epoch": 0.24444811814625006,
      "grad_norm": 0.8061476349830627,
      "learning_rate": 0.00018007992864868752,
      "loss": 2.2759,
      "step": 1109
    },
    {
      "epoch": 0.24466854025458754,
      "grad_norm": 0.8964433073997498,
      "learning_rate": 0.0001800362506993691,
      "loss": 2.3868,
      "step": 1110
    },
    {
      "epoch": 0.244888962362925,
      "grad_norm": 0.8317962288856506,
      "learning_rate": 0.00017999253022626482,
      "loss": 2.2922,
      "step": 1111
    },
    {
      "epoch": 0.24510938447126246,
      "grad_norm": 0.7780240774154663,
      "learning_rate": 0.00017994876725260382,
      "loss": 2.2423,
      "step": 1112
    },
    {
      "epoch": 0.24532980657959993,
      "grad_norm": 0.7902697920799255,
      "learning_rate": 0.0001799049618016375,
      "loss": 2.1691,
      "step": 1113
    },
    {
      "epoch": 0.2455502286879374,
      "grad_norm": 0.8700764775276184,
      "learning_rate": 0.00017986111389664008,
      "loss": 2.2048,
      "step": 1114
    },
    {
      "epoch": 0.24577065079627486,
      "grad_norm": 0.7950437664985657,
      "learning_rate": 0.0001798172235609082,
      "loss": 2.3268,
      "step": 1115
    },
    {
      "epoch": 0.24599107290461233,
      "grad_norm": 0.7321479320526123,
      "learning_rate": 0.00017977329081776108,
      "loss": 2.2936,
      "step": 1116
    },
    {
      "epoch": 0.2462114950129498,
      "grad_norm": 0.7673276662826538,
      "learning_rate": 0.0001797293156905405,
      "loss": 2.2917,
      "step": 1117
    },
    {
      "epoch": 0.24643191712128726,
      "grad_norm": 0.7858921885490417,
      "learning_rate": 0.0001796852982026107,
      "loss": 2.1989,
      "step": 1118
    },
    {
      "epoch": 0.24665233922962473,
      "grad_norm": 0.7160308361053467,
      "learning_rate": 0.00017964123837735848,
      "loss": 2.3384,
      "step": 1119
    },
    {
      "epoch": 0.2468727613379622,
      "grad_norm": 0.7792894244194031,
      "learning_rate": 0.00017959713623819315,
      "loss": 2.2075,
      "step": 1120
    },
    {
      "epoch": 0.24709318344629966,
      "grad_norm": 0.7811226844787598,
      "learning_rate": 0.0001795529918085464,
      "loss": 2.3548,
      "step": 1121
    },
    {
      "epoch": 0.24731360555463713,
      "grad_norm": 0.7646430134773254,
      "learning_rate": 0.00017950880511187248,
      "loss": 2.2019,
      "step": 1122
    },
    {
      "epoch": 0.2475340276629746,
      "grad_norm": 0.762254536151886,
      "learning_rate": 0.00017946457617164807,
      "loss": 2.1925,
      "step": 1123
    },
    {
      "epoch": 0.24775444977131206,
      "grad_norm": 0.7943596243858337,
      "learning_rate": 0.0001794203050113723,
      "loss": 2.1842,
      "step": 1124
    },
    {
      "epoch": 0.24797487187964953,
      "grad_norm": 0.7508061528205872,
      "learning_rate": 0.00017937599165456677,
      "loss": 2.2987,
      "step": 1125
    },
    {
      "epoch": 0.24819529398798698,
      "grad_norm": 0.7786548137664795,
      "learning_rate": 0.00017933163612477538,
      "loss": 2.4021,
      "step": 1126
    },
    {
      "epoch": 0.24841571609632446,
      "grad_norm": 0.7466885447502136,
      "learning_rate": 0.0001792872384455646,
      "loss": 2.2203,
      "step": 1127
    },
    {
      "epoch": 0.24863613820466193,
      "grad_norm": 0.7724654674530029,
      "learning_rate": 0.00017924279864052313,
      "loss": 2.256,
      "step": 1128
    },
    {
      "epoch": 0.24885656031299938,
      "grad_norm": 0.8125854730606079,
      "learning_rate": 0.00017919831673326216,
      "loss": 2.2474,
      "step": 1129
    },
    {
      "epoch": 0.24907698242133686,
      "grad_norm": 0.8088080883026123,
      "learning_rate": 0.00017915379274741525,
      "loss": 2.258,
      "step": 1130
    },
    {
      "epoch": 0.24929740452967433,
      "grad_norm": 0.7342892289161682,
      "learning_rate": 0.00017910922670663824,
      "loss": 2.2464,
      "step": 1131
    },
    {
      "epoch": 0.24951782663801178,
      "grad_norm": 0.769363522529602,
      "learning_rate": 0.0001790646186346094,
      "loss": 2.2791,
      "step": 1132
    },
    {
      "epoch": 0.24973824874634926,
      "grad_norm": 0.7705099582672119,
      "learning_rate": 0.00017901996855502924,
      "loss": 2.2672,
      "step": 1133
    },
    {
      "epoch": 0.24995867085468673,
      "grad_norm": 0.7565601468086243,
      "learning_rate": 0.0001789752764916207,
      "loss": 2.2603,
      "step": 1134
    },
    {
      "epoch": 0.2501790929630242,
      "grad_norm": 0.8287437558174133,
      "learning_rate": 0.0001789305424681289,
      "loss": 2.3134,
      "step": 1135
    },
    {
      "epoch": 0.25039951507136166,
      "grad_norm": 0.7118788361549377,
      "learning_rate": 0.0001788857665083214,
      "loss": 2.2246,
      "step": 1136
    },
    {
      "epoch": 0.2506199371796991,
      "grad_norm": 0.7979238629341125,
      "learning_rate": 0.0001788409486359879,
      "loss": 2.2012,
      "step": 1137
    },
    {
      "epoch": 0.2508403592880366,
      "grad_norm": 0.819365382194519,
      "learning_rate": 0.00017879608887494045,
      "loss": 2.3005,
      "step": 1138
    },
    {
      "epoch": 0.25106078139637406,
      "grad_norm": 0.7899625301361084,
      "learning_rate": 0.00017875118724901333,
      "loss": 2.1958,
      "step": 1139
    },
    {
      "epoch": 0.2512812035047115,
      "grad_norm": 0.7965990900993347,
      "learning_rate": 0.0001787062437820631,
      "loss": 2.1725,
      "step": 1140
    },
    {
      "epoch": 0.251501625613049,
      "grad_norm": 0.8428856134414673,
      "learning_rate": 0.00017866125849796852,
      "loss": 2.2447,
      "step": 1141
    },
    {
      "epoch": 0.25172204772138645,
      "grad_norm": 0.8208176493644714,
      "learning_rate": 0.00017861623142063047,
      "loss": 2.2462,
      "step": 1142
    },
    {
      "epoch": 0.2519424698297239,
      "grad_norm": 0.7162795066833496,
      "learning_rate": 0.00017857116257397225,
      "loss": 2.2783,
      "step": 1143
    },
    {
      "epoch": 0.2521628919380614,
      "grad_norm": 0.7613261938095093,
      "learning_rate": 0.00017852605198193916,
      "loss": 2.2041,
      "step": 1144
    },
    {
      "epoch": 0.25238331404639885,
      "grad_norm": 0.8035195469856262,
      "learning_rate": 0.0001784808996684988,
      "loss": 2.2691,
      "step": 1145
    },
    {
      "epoch": 0.2526037361547363,
      "grad_norm": 0.8068618774414062,
      "learning_rate": 0.00017843570565764086,
      "loss": 2.3416,
      "step": 1146
    },
    {
      "epoch": 0.2528241582630738,
      "grad_norm": 0.7737892866134644,
      "learning_rate": 0.00017839046997337725,
      "loss": 2.2631,
      "step": 1147
    },
    {
      "epoch": 0.25304458037141125,
      "grad_norm": 0.7640506625175476,
      "learning_rate": 0.00017834519263974197,
      "loss": 2.2275,
      "step": 1148
    },
    {
      "epoch": 0.2532650024797487,
      "grad_norm": 0.7581484913825989,
      "learning_rate": 0.00017829987368079113,
      "loss": 2.2723,
      "step": 1149
    },
    {
      "epoch": 0.2534854245880862,
      "grad_norm": 0.762021541595459,
      "learning_rate": 0.00017825451312060303,
      "loss": 2.2376,
      "step": 1150
    },
    {
      "epoch": 0.25370584669642365,
      "grad_norm": 0.7851666212081909,
      "learning_rate": 0.00017820911098327802,
      "loss": 2.2049,
      "step": 1151
    },
    {
      "epoch": 0.2539262688047611,
      "grad_norm": 0.7283372282981873,
      "learning_rate": 0.00017816366729293855,
      "loss": 2.2649,
      "step": 1152
    },
    {
      "epoch": 0.2541466909130986,
      "grad_norm": 0.7682250738143921,
      "learning_rate": 0.00017811818207372917,
      "loss": 2.2671,
      "step": 1153
    },
    {
      "epoch": 0.25436711302143605,
      "grad_norm": 0.795171320438385,
      "learning_rate": 0.00017807265534981645,
      "loss": 2.3939,
      "step": 1154
    },
    {
      "epoch": 0.2545875351297735,
      "grad_norm": 0.7334957122802734,
      "learning_rate": 0.00017802708714538906,
      "loss": 2.2859,
      "step": 1155
    },
    {
      "epoch": 0.254807957238111,
      "grad_norm": 0.771388053894043,
      "learning_rate": 0.00017798147748465768,
      "loss": 2.2653,
      "step": 1156
    },
    {
      "epoch": 0.25502837934644845,
      "grad_norm": 0.7203769683837891,
      "learning_rate": 0.000177935826391855,
      "loss": 2.3413,
      "step": 1157
    },
    {
      "epoch": 0.2552488014547859,
      "grad_norm": 0.7557588815689087,
      "learning_rate": 0.00017789013389123582,
      "loss": 2.2197,
      "step": 1158
    },
    {
      "epoch": 0.2554692235631234,
      "grad_norm": 0.768765389919281,
      "learning_rate": 0.00017784440000707675,
      "loss": 2.2835,
      "step": 1159
    },
    {
      "epoch": 0.25568964567146085,
      "grad_norm": 0.7449793219566345,
      "learning_rate": 0.00017779862476367662,
      "loss": 2.2533,
      "step": 1160
    },
    {
      "epoch": 0.2559100677797983,
      "grad_norm": 0.8096796870231628,
      "learning_rate": 0.000177752808185356,
      "loss": 2.2731,
      "step": 1161
    },
    {
      "epoch": 0.2561304898881358,
      "grad_norm": 0.7691149711608887,
      "learning_rate": 0.00017770695029645762,
      "loss": 2.2718,
      "step": 1162
    },
    {
      "epoch": 0.25635091199647325,
      "grad_norm": 0.7652122974395752,
      "learning_rate": 0.00017766105112134608,
      "loss": 2.2511,
      "step": 1163
    },
    {
      "epoch": 0.2565713341048107,
      "grad_norm": 0.739314615726471,
      "learning_rate": 0.00017761511068440787,
      "loss": 2.2791,
      "step": 1164
    },
    {
      "epoch": 0.2567917562131482,
      "grad_norm": 0.82001131772995,
      "learning_rate": 0.00017756912901005148,
      "loss": 2.2756,
      "step": 1165
    },
    {
      "epoch": 0.25701217832148565,
      "grad_norm": 0.7397631406784058,
      "learning_rate": 0.00017752310612270726,
      "loss": 2.2954,
      "step": 1166
    },
    {
      "epoch": 0.2572326004298231,
      "grad_norm": 0.7123408317565918,
      "learning_rate": 0.00017747704204682747,
      "loss": 2.3134,
      "step": 1167
    },
    {
      "epoch": 0.2574530225381606,
      "grad_norm": 0.7590401768684387,
      "learning_rate": 0.00017743093680688628,
      "loss": 2.2819,
      "step": 1168
    },
    {
      "epoch": 0.25767344464649805,
      "grad_norm": 0.8649609684944153,
      "learning_rate": 0.00017738479042737968,
      "loss": 2.3249,
      "step": 1169
    },
    {
      "epoch": 0.2578938667548355,
      "grad_norm": 0.7520982623100281,
      "learning_rate": 0.00017733860293282555,
      "loss": 2.1853,
      "step": 1170
    },
    {
      "epoch": 0.258114288863173,
      "grad_norm": 0.7400807738304138,
      "learning_rate": 0.0001772923743477636,
      "loss": 2.1167,
      "step": 1171
    },
    {
      "epoch": 0.25833471097151045,
      "grad_norm": 0.8288618922233582,
      "learning_rate": 0.00017724610469675543,
      "loss": 2.192,
      "step": 1172
    },
    {
      "epoch": 0.2585551330798479,
      "grad_norm": 0.8398124575614929,
      "learning_rate": 0.00017719979400438433,
      "loss": 2.2396,
      "step": 1173
    },
    {
      "epoch": 0.2587755551881854,
      "grad_norm": 0.7916671633720398,
      "learning_rate": 0.00017715344229525553,
      "loss": 2.2209,
      "step": 1174
    },
    {
      "epoch": 0.25899597729652285,
      "grad_norm": 0.8531112670898438,
      "learning_rate": 0.000177107049593996,
      "loss": 2.2583,
      "step": 1175
    },
    {
      "epoch": 0.2592163994048603,
      "grad_norm": 0.8505764007568359,
      "learning_rate": 0.00017706061592525445,
      "loss": 2.2281,
      "step": 1176
    },
    {
      "epoch": 0.25943682151319775,
      "grad_norm": 0.722248911857605,
      "learning_rate": 0.00017701414131370143,
      "loss": 2.3282,
      "step": 1177
    },
    {
      "epoch": 0.25965724362153525,
      "grad_norm": 0.76747727394104,
      "learning_rate": 0.00017696762578402918,
      "loss": 2.0709,
      "step": 1178
    },
    {
      "epoch": 0.2598776657298727,
      "grad_norm": 0.8048473000526428,
      "learning_rate": 0.00017692106936095177,
      "loss": 2.1533,
      "step": 1179
    },
    {
      "epoch": 0.26009808783821015,
      "grad_norm": 0.8929221034049988,
      "learning_rate": 0.0001768744720692048,
      "loss": 2.2331,
      "step": 1180
    },
    {
      "epoch": 0.26031850994654765,
      "grad_norm": 0.7724414467811584,
      "learning_rate": 0.00017682783393354584,
      "loss": 2.2575,
      "step": 1181
    },
    {
      "epoch": 0.2605389320548851,
      "grad_norm": 0.7092270255088806,
      "learning_rate": 0.00017678115497875402,
      "loss": 2.1811,
      "step": 1182
    },
    {
      "epoch": 0.26075935416322255,
      "grad_norm": 0.8068305253982544,
      "learning_rate": 0.00017673443522963012,
      "loss": 2.3148,
      "step": 1183
    },
    {
      "epoch": 0.26097977627156005,
      "grad_norm": 0.8523135781288147,
      "learning_rate": 0.00017668767471099673,
      "loss": 2.2975,
      "step": 1184
    },
    {
      "epoch": 0.2612001983798975,
      "grad_norm": 0.7929922342300415,
      "learning_rate": 0.00017664087344769792,
      "loss": 2.1388,
      "step": 1185
    },
    {
      "epoch": 0.26142062048823494,
      "grad_norm": 0.7902235388755798,
      "learning_rate": 0.00017659403146459956,
      "loss": 2.2232,
      "step": 1186
    },
    {
      "epoch": 0.26164104259657245,
      "grad_norm": 0.801682710647583,
      "learning_rate": 0.00017654714878658913,
      "loss": 2.1823,
      "step": 1187
    },
    {
      "epoch": 0.2618614647049099,
      "grad_norm": 0.7665947675704956,
      "learning_rate": 0.0001765002254385757,
      "loss": 2.1584,
      "step": 1188
    },
    {
      "epoch": 0.26208188681324734,
      "grad_norm": 0.811362087726593,
      "learning_rate": 0.00017645326144548988,
      "loss": 2.2869,
      "step": 1189
    },
    {
      "epoch": 0.26230230892158485,
      "grad_norm": 0.7159063816070557,
      "learning_rate": 0.00017640625683228403,
      "loss": 2.2271,
      "step": 1190
    },
    {
      "epoch": 0.2625227310299223,
      "grad_norm": 0.7492620348930359,
      "learning_rate": 0.00017635921162393196,
      "loss": 2.2321,
      "step": 1191
    },
    {
      "epoch": 0.26274315313825974,
      "grad_norm": 0.7522966265678406,
      "learning_rate": 0.00017631212584542916,
      "loss": 2.2739,
      "step": 1192
    },
    {
      "epoch": 0.26296357524659725,
      "grad_norm": 0.7496805191040039,
      "learning_rate": 0.00017626499952179254,
      "loss": 2.2967,
      "step": 1193
    },
    {
      "epoch": 0.2631839973549347,
      "grad_norm": 0.7681370973587036,
      "learning_rate": 0.00017621783267806067,
      "loss": 2.2779,
      "step": 1194
    },
    {
      "epoch": 0.26340441946327214,
      "grad_norm": 0.7621352672576904,
      "learning_rate": 0.00017617062533929357,
      "loss": 2.2038,
      "step": 1195
    },
    {
      "epoch": 0.26362484157160965,
      "grad_norm": 0.8017644882202148,
      "learning_rate": 0.00017612337753057284,
      "loss": 2.2037,
      "step": 1196
    },
    {
      "epoch": 0.2638452636799471,
      "grad_norm": 0.7210404872894287,
      "learning_rate": 0.00017607608927700156,
      "loss": 2.1176,
      "step": 1197
    },
    {
      "epoch": 0.26406568578828454,
      "grad_norm": 0.7688873410224915,
      "learning_rate": 0.0001760287606037043,
      "loss": 2.3307,
      "step": 1198
    },
    {
      "epoch": 0.26428610789662205,
      "grad_norm": 0.7414920330047607,
      "learning_rate": 0.000175981391535827,
      "loss": 2.2218,
      "step": 1199
    },
    {
      "epoch": 0.2645065300049595,
      "grad_norm": 0.7469229102134705,
      "learning_rate": 0.00017593398209853723,
      "loss": 2.2563,
      "step": 1200
    },
    {
      "epoch": 0.2645065300049595,
      "eval_bpb": 1.160369363889425,
      "eval_loss": 2.2362303733825684,
      "eval_runtime": 184.3551,
      "eval_samples_per_second": 62.163,
      "eval_steps_per_second": 7.773,
      "step": 1200
    },
    {
      "epoch": 0.26472695211329694,
      "grad_norm": 0.8251612186431885,
      "learning_rate": 0.00017588653231702397,
      "loss": 2.3331,
      "step": 1201
    },
    {
      "epoch": 0.26494737422163445,
      "grad_norm": 0.7838966846466064,
      "learning_rate": 0.00017583904221649753,
      "loss": 2.2536,
      "step": 1202
    },
    {
      "epoch": 0.2651677963299719,
      "grad_norm": 0.753894567489624,
      "learning_rate": 0.00017579151182218979,
      "loss": 2.2172,
      "step": 1203
    },
    {
      "epoch": 0.26538821843830934,
      "grad_norm": 0.8166255950927734,
      "learning_rate": 0.00017574394115935386,
      "loss": 2.0968,
      "step": 1204
    },
    {
      "epoch": 0.26560864054664685,
      "grad_norm": 0.7846564054489136,
      "learning_rate": 0.0001756963302532644,
      "loss": 2.1709,
      "step": 1205
    },
    {
      "epoch": 0.2658290626549843,
      "grad_norm": 0.829788863658905,
      "learning_rate": 0.00017564867912921742,
      "loss": 2.1438,
      "step": 1206
    },
    {
      "epoch": 0.26604948476332174,
      "grad_norm": 0.7755435705184937,
      "learning_rate": 0.00017560098781253025,
      "loss": 2.2795,
      "step": 1207
    },
    {
      "epoch": 0.26626990687165925,
      "grad_norm": 0.7415637969970703,
      "learning_rate": 0.0001755532563285416,
      "loss": 2.2505,
      "step": 1208
    },
    {
      "epoch": 0.2664903289799967,
      "grad_norm": 0.761690616607666,
      "learning_rate": 0.0001755054847026115,
      "loss": 2.4052,
      "step": 1209
    },
    {
      "epoch": 0.26671075108833414,
      "grad_norm": 0.7691196203231812,
      "learning_rate": 0.00017545767296012137,
      "loss": 2.1923,
      "step": 1210
    },
    {
      "epoch": 0.26693117319667164,
      "grad_norm": 0.8170738816261292,
      "learning_rate": 0.00017540982112647388,
      "loss": 2.2757,
      "step": 1211
    },
    {
      "epoch": 0.2671515953050091,
      "grad_norm": 0.7558908462524414,
      "learning_rate": 0.00017536192922709302,
      "loss": 2.184,
      "step": 1212
    },
    {
      "epoch": 0.26737201741334654,
      "grad_norm": 0.733386218547821,
      "learning_rate": 0.00017531399728742408,
      "loss": 2.1401,
      "step": 1213
    },
    {
      "epoch": 0.26759243952168404,
      "grad_norm": 0.8102444410324097,
      "learning_rate": 0.00017526602533293363,
      "loss": 2.2988,
      "step": 1214
    },
    {
      "epoch": 0.2678128616300215,
      "grad_norm": 0.8099070191383362,
      "learning_rate": 0.00017521801338910944,
      "loss": 2.2224,
      "step": 1215
    },
    {
      "epoch": 0.26803328373835894,
      "grad_norm": 0.7494538426399231,
      "learning_rate": 0.00017516996148146064,
      "loss": 2.2948,
      "step": 1216
    },
    {
      "epoch": 0.26825370584669644,
      "grad_norm": 0.8341994881629944,
      "learning_rate": 0.0001751218696355174,
      "loss": 2.1415,
      "step": 1217
    },
    {
      "epoch": 0.2684741279550339,
      "grad_norm": 0.7866456508636475,
      "learning_rate": 0.00017507373787683142,
      "loss": 2.2318,
      "step": 1218
    },
    {
      "epoch": 0.26869455006337134,
      "grad_norm": 0.722846508026123,
      "learning_rate": 0.00017502556623097526,
      "loss": 2.2171,
      "step": 1219
    },
    {
      "epoch": 0.26891497217170884,
      "grad_norm": 0.795566737651825,
      "learning_rate": 0.00017497735472354287,
      "loss": 2.2594,
      "step": 1220
    },
    {
      "epoch": 0.2691353942800463,
      "grad_norm": 0.8013664484024048,
      "learning_rate": 0.0001749291033801494,
      "loss": 2.224,
      "step": 1221
    },
    {
      "epoch": 0.26935581638838374,
      "grad_norm": 0.7825328707695007,
      "learning_rate": 0.00017488081222643106,
      "loss": 2.3593,
      "step": 1222
    },
    {
      "epoch": 0.26957623849672124,
      "grad_norm": 0.7791579961776733,
      "learning_rate": 0.0001748324812880452,
      "loss": 2.2815,
      "step": 1223
    },
    {
      "epoch": 0.2697966606050587,
      "grad_norm": 0.8268334865570068,
      "learning_rate": 0.0001747841105906705,
      "loss": 2.2298,
      "step": 1224
    },
    {
      "epoch": 0.27001708271339614,
      "grad_norm": 0.8455089926719666,
      "learning_rate": 0.0001747357001600065,
      "loss": 2.3018,
      "step": 1225
    },
    {
      "epoch": 0.27023750482173364,
      "grad_norm": 0.7790243625640869,
      "learning_rate": 0.0001746872500217741,
      "loss": 2.2054,
      "step": 1226
    },
    {
      "epoch": 0.2704579269300711,
      "grad_norm": 0.7765048146247864,
      "learning_rate": 0.00017463876020171504,
      "loss": 2.2819,
      "step": 1227
    },
    {
      "epoch": 0.27067834903840854,
      "grad_norm": 0.7852209806442261,
      "learning_rate": 0.0001745902307255924,
      "loss": 2.1704,
      "step": 1228
    },
    {
      "epoch": 0.27089877114674604,
      "grad_norm": 0.7832857966423035,
      "learning_rate": 0.00017454166161919013,
      "loss": 2.2334,
      "step": 1229
    },
    {
      "epoch": 0.2711191932550835,
      "grad_norm": 0.780556321144104,
      "learning_rate": 0.0001744930529083134,
      "loss": 2.179,
      "step": 1230
    },
    {
      "epoch": 0.27133961536342094,
      "grad_norm": 0.772138774394989,
      "learning_rate": 0.0001744444046187882,
      "loss": 2.1966,
      "step": 1231
    },
    {
      "epoch": 0.27156003747175844,
      "grad_norm": 0.7777549624443054,
      "learning_rate": 0.00017439571677646185,
      "loss": 2.2646,
      "step": 1232
    },
    {
      "epoch": 0.2717804595800959,
      "grad_norm": 0.7746325135231018,
      "learning_rate": 0.0001743469894072024,
      "loss": 2.1641,
      "step": 1233
    },
    {
      "epoch": 0.27200088168843334,
      "grad_norm": 0.7858633995056152,
      "learning_rate": 0.00017429822253689907,
      "loss": 2.2713,
      "step": 1234
    },
    {
      "epoch": 0.27222130379677084,
      "grad_norm": 0.749160647392273,
      "learning_rate": 0.00017424941619146204,
      "loss": 2.2256,
      "step": 1235
    },
    {
      "epoch": 0.2724417259051083,
      "grad_norm": 0.7558864951133728,
      "learning_rate": 0.00017420057039682236,
      "loss": 2.236,
      "step": 1236
    },
    {
      "epoch": 0.27266214801344574,
      "grad_norm": 0.7917948961257935,
      "learning_rate": 0.0001741516851789322,
      "loss": 2.2762,
      "step": 1237
    },
    {
      "epoch": 0.27288257012178324,
      "grad_norm": 0.7605097889900208,
      "learning_rate": 0.00017410276056376456,
      "loss": 2.2586,
      "step": 1238
    },
    {
      "epoch": 0.2731029922301207,
      "grad_norm": 0.7575761675834656,
      "learning_rate": 0.0001740537965773134,
      "loss": 2.2686,
      "step": 1239
    },
    {
      "epoch": 0.27332341433845814,
      "grad_norm": 0.7631157040596008,
      "learning_rate": 0.00017400479324559368,
      "loss": 2.1882,
      "step": 1240
    },
    {
      "epoch": 0.27354383644679564,
      "grad_norm": 0.7289288640022278,
      "learning_rate": 0.0001739557505946411,
      "loss": 2.1446,
      "step": 1241
    },
    {
      "epoch": 0.2737642585551331,
      "grad_norm": 0.712980329990387,
      "learning_rate": 0.0001739066686505124,
      "loss": 2.3693,
      "step": 1242
    },
    {
      "epoch": 0.27398468066347054,
      "grad_norm": 0.7260845899581909,
      "learning_rate": 0.00017385754743928512,
      "loss": 2.2867,
      "step": 1243
    },
    {
      "epoch": 0.274205102771808,
      "grad_norm": 0.7633198499679565,
      "learning_rate": 0.0001738083869870577,
      "loss": 2.2114,
      "step": 1244
    },
    {
      "epoch": 0.2744255248801455,
      "grad_norm": 0.720999538898468,
      "learning_rate": 0.00017375918731994937,
      "loss": 2.2611,
      "step": 1245
    },
    {
      "epoch": 0.27464594698848294,
      "grad_norm": 0.7855420112609863,
      "learning_rate": 0.0001737099484641003,
      "loss": 2.2509,
      "step": 1246
    },
    {
      "epoch": 0.2748663690968204,
      "grad_norm": 0.7828165888786316,
      "learning_rate": 0.00017366067044567135,
      "loss": 2.1914,
      "step": 1247
    },
    {
      "epoch": 0.2750867912051579,
      "grad_norm": 0.7687867879867554,
      "learning_rate": 0.00017361135329084428,
      "loss": 2.1641,
      "step": 1248
    },
    {
      "epoch": 0.27530721331349534,
      "grad_norm": 0.7461012601852417,
      "learning_rate": 0.00017356199702582163,
      "loss": 2.136,
      "step": 1249
    },
    {
      "epoch": 0.2755276354218328,
      "grad_norm": 0.7446545362472534,
      "learning_rate": 0.0001735126016768267,
      "loss": 2.2456,
      "step": 1250
    },
    {
      "epoch": 0.2757480575301703,
      "grad_norm": 0.8081080317497253,
      "learning_rate": 0.0001734631672701036,
      "loss": 2.3746,
      "step": 1251
    },
    {
      "epoch": 0.27596847963850774,
      "grad_norm": 0.785469651222229,
      "learning_rate": 0.00017341369383191708,
      "loss": 2.2542,
      "step": 1252
    },
    {
      "epoch": 0.2761889017468452,
      "grad_norm": 0.7757176160812378,
      "learning_rate": 0.00017336418138855274,
      "loss": 2.2043,
      "step": 1253
    },
    {
      "epoch": 0.2764093238551827,
      "grad_norm": 0.8267678618431091,
      "learning_rate": 0.00017331462996631687,
      "loss": 2.2365,
      "step": 1254
    },
    {
      "epoch": 0.27662974596352014,
      "grad_norm": 0.7604681849479675,
      "learning_rate": 0.00017326503959153649,
      "loss": 2.1411,
      "step": 1255
    },
    {
      "epoch": 0.2768501680718576,
      "grad_norm": 0.7717955112457275,
      "learning_rate": 0.00017321541029055926,
      "loss": 2.1272,
      "step": 1256
    },
    {
      "epoch": 0.2770705901801951,
      "grad_norm": 0.799120306968689,
      "learning_rate": 0.00017316574208975352,
      "loss": 2.204,
      "step": 1257
    },
    {
      "epoch": 0.27729101228853253,
      "grad_norm": 0.8386200070381165,
      "learning_rate": 0.00017311603501550838,
      "loss": 2.276,
      "step": 1258
    },
    {
      "epoch": 0.27751143439687,
      "grad_norm": 0.7919439673423767,
      "learning_rate": 0.00017306628909423354,
      "loss": 2.1286,
      "step": 1259
    },
    {
      "epoch": 0.2777318565052075,
      "grad_norm": 0.7885940074920654,
      "learning_rate": 0.00017301650435235923,
      "loss": 2.3643,
      "step": 1260
    },
    {
      "epoch": 0.27795227861354493,
      "grad_norm": 0.7869325876235962,
      "learning_rate": 0.00017296668081633653,
      "loss": 2.179,
      "step": 1261
    },
    {
      "epoch": 0.2781727007218824,
      "grad_norm": 0.8283299803733826,
      "learning_rate": 0.00017291681851263696,
      "loss": 2.1743,
      "step": 1262
    },
    {
      "epoch": 0.2783931228302199,
      "grad_norm": 0.7671277523040771,
      "learning_rate": 0.00017286691746775268,
      "loss": 2.3122,
      "step": 1263
    },
    {
      "epoch": 0.27861354493855733,
      "grad_norm": 0.784153938293457,
      "learning_rate": 0.00017281697770819647,
      "loss": 2.3251,
      "step": 1264
    },
    {
      "epoch": 0.2788339670468948,
      "grad_norm": 0.7816244959831238,
      "learning_rate": 0.00017276699926050163,
      "loss": 2.197,
      "step": 1265
    },
    {
      "epoch": 0.2790543891552323,
      "grad_norm": 0.7952396273612976,
      "learning_rate": 0.00017271698215122206,
      "loss": 2.2329,
      "step": 1266
    },
    {
      "epoch": 0.27927481126356973,
      "grad_norm": 0.7601649165153503,
      "learning_rate": 0.00017266692640693213,
      "loss": 2.2126,
      "step": 1267
    },
    {
      "epoch": 0.2794952333719072,
      "grad_norm": 0.7981962561607361,
      "learning_rate": 0.00017261683205422687,
      "loss": 2.1715,
      "step": 1268
    },
    {
      "epoch": 0.2797156554802447,
      "grad_norm": 0.7731349468231201,
      "learning_rate": 0.00017256669911972164,
      "loss": 2.1659,
      "step": 1269
    },
    {
      "epoch": 0.27993607758858213,
      "grad_norm": 0.8225795030593872,
      "learning_rate": 0.00017251652763005247,
      "loss": 2.3833,
      "step": 1270
    },
    {
      "epoch": 0.2801564996969196,
      "grad_norm": 0.7319188714027405,
      "learning_rate": 0.0001724663176118758,
      "loss": 2.1875,
      "step": 1271
    },
    {
      "epoch": 0.2803769218052571,
      "grad_norm": 0.7665533423423767,
      "learning_rate": 0.00017241606909186848,
      "loss": 2.1883,
      "step": 1272
    },
    {
      "epoch": 0.28059734391359453,
      "grad_norm": 0.7382875084877014,
      "learning_rate": 0.000172365782096728,
      "loss": 2.1502,
      "step": 1273
    },
    {
      "epoch": 0.280817766021932,
      "grad_norm": 0.7826829552650452,
      "learning_rate": 0.000172315456653172,
      "loss": 2.2735,
      "step": 1274
    },
    {
      "epoch": 0.2810381881302695,
      "grad_norm": 0.749201774597168,
      "learning_rate": 0.00017226509278793885,
      "loss": 2.2516,
      "step": 1275
    },
    {
      "epoch": 0.28125861023860693,
      "grad_norm": 0.7362197041511536,
      "learning_rate": 0.00017221469052778715,
      "loss": 2.2064,
      "step": 1276
    },
    {
      "epoch": 0.2814790323469444,
      "grad_norm": 0.7749683260917664,
      "learning_rate": 0.00017216424989949597,
      "loss": 2.336,
      "step": 1277
    },
    {
      "epoch": 0.2816994544552819,
      "grad_norm": 0.8076945543289185,
      "learning_rate": 0.00017211377092986476,
      "loss": 2.3004,
      "step": 1278
    },
    {
      "epoch": 0.28191987656361933,
      "grad_norm": 0.7306873202323914,
      "learning_rate": 0.00017206325364571326,
      "loss": 2.2022,
      "step": 1279
    },
    {
      "epoch": 0.2821402986719568,
      "grad_norm": 0.8039346933364868,
      "learning_rate": 0.00017201269807388175,
      "loss": 2.0911,
      "step": 1280
    },
    {
      "epoch": 0.2823607207802943,
      "grad_norm": 0.7406665086746216,
      "learning_rate": 0.00017196210424123062,
      "loss": 2.2695,
      "step": 1281
    },
    {
      "epoch": 0.28258114288863173,
      "grad_norm": 0.7329961657524109,
      "learning_rate": 0.0001719114721746408,
      "loss": 2.1675,
      "step": 1282
    },
    {
      "epoch": 0.2828015649969692,
      "grad_norm": 0.8032838106155396,
      "learning_rate": 0.00017186080190101334,
      "loss": 2.2252,
      "step": 1283
    },
    {
      "epoch": 0.2830219871053067,
      "grad_norm": 0.8144814372062683,
      "learning_rate": 0.00017181009344726978,
      "loss": 2.0996,
      "step": 1284
    },
    {
      "epoch": 0.28324240921364413,
      "grad_norm": 0.7788400650024414,
      "learning_rate": 0.00017175934684035183,
      "loss": 2.2982,
      "step": 1285
    },
    {
      "epoch": 0.2834628313219816,
      "grad_norm": 0.7443048357963562,
      "learning_rate": 0.0001717085621072215,
      "loss": 2.2752,
      "step": 1286
    },
    {
      "epoch": 0.2836832534303191,
      "grad_norm": 0.7056987285614014,
      "learning_rate": 0.000171657739274861,
      "loss": 2.112,
      "step": 1287
    },
    {
      "epoch": 0.28390367553865653,
      "grad_norm": 0.7627144455909729,
      "learning_rate": 0.0001716068783702729,
      "loss": 2.0977,
      "step": 1288
    },
    {
      "epoch": 0.284124097646994,
      "grad_norm": 0.7449780106544495,
      "learning_rate": 0.00017155597942047988,
      "loss": 2.1318,
      "step": 1289
    },
    {
      "epoch": 0.2843445197553315,
      "grad_norm": 0.7299350500106812,
      "learning_rate": 0.00017150504245252494,
      "loss": 2.3149,
      "step": 1290
    },
    {
      "epoch": 0.28456494186366893,
      "grad_norm": 0.7681246995925903,
      "learning_rate": 0.00017145406749347118,
      "loss": 2.1683,
      "step": 1291
    },
    {
      "epoch": 0.2847853639720064,
      "grad_norm": 0.710610032081604,
      "learning_rate": 0.00017140305457040195,
      "loss": 2.4105,
      "step": 1292
    },
    {
      "epoch": 0.2850057860803439,
      "grad_norm": 0.8703294396400452,
      "learning_rate": 0.00017135200371042077,
      "loss": 2.1349,
      "step": 1293
    },
    {
      "epoch": 0.28522620818868133,
      "grad_norm": 0.7634097337722778,
      "learning_rate": 0.00017130091494065125,
      "loss": 2.1122,
      "step": 1294
    },
    {
      "epoch": 0.2854466302970188,
      "grad_norm": 0.8051940202713013,
      "learning_rate": 0.00017124978828823726,
      "loss": 2.2079,
      "step": 1295
    },
    {
      "epoch": 0.2856670524053563,
      "grad_norm": 0.7098733186721802,
      "learning_rate": 0.00017119862378034263,
      "loss": 2.1184,
      "step": 1296
    },
    {
      "epoch": 0.28588747451369373,
      "grad_norm": 0.7350472807884216,
      "learning_rate": 0.0001711474214441515,
      "loss": 2.2722,
      "step": 1297
    },
    {
      "epoch": 0.2861078966220312,
      "grad_norm": 0.7440918684005737,
      "learning_rate": 0.00017109618130686793,
      "loss": 2.2145,
      "step": 1298
    },
    {
      "epoch": 0.2863283187303687,
      "grad_norm": 0.823554515838623,
      "learning_rate": 0.00017104490339571617,
      "loss": 2.2351,
      "step": 1299
    },
    {
      "epoch": 0.28654874083870613,
      "grad_norm": 0.7253199815750122,
      "learning_rate": 0.00017099358773794054,
      "loss": 2.2308,
      "step": 1300
    },
    {
      "epoch": 0.2867691629470436,
      "grad_norm": 0.7917382121086121,
      "learning_rate": 0.00017094223436080533,
      "loss": 2.2419,
      "step": 1301
    },
    {
      "epoch": 0.2869895850553811,
      "grad_norm": 0.8260071277618408,
      "learning_rate": 0.00017089084329159496,
      "loss": 2.2246,
      "step": 1302
    },
    {
      "epoch": 0.28721000716371853,
      "grad_norm": 0.7765074968338013,
      "learning_rate": 0.00017083941455761385,
      "loss": 2.2823,
      "step": 1303
    },
    {
      "epoch": 0.287430429272056,
      "grad_norm": 0.8356334567070007,
      "learning_rate": 0.0001707879481861864,
      "loss": 2.327,
      "step": 1304
    },
    {
      "epoch": 0.2876508513803935,
      "grad_norm": 0.7431723475456238,
      "learning_rate": 0.00017073644420465705,
      "loss": 2.1823,
      "step": 1305
    },
    {
      "epoch": 0.2878712734887309,
      "grad_norm": 0.7840165495872498,
      "learning_rate": 0.0001706849026403902,
      "loss": 2.1742,
      "step": 1306
    },
    {
      "epoch": 0.2880916955970684,
      "grad_norm": 0.745380699634552,
      "learning_rate": 0.0001706333235207702,
      "loss": 2.2173,
      "step": 1307
    },
    {
      "epoch": 0.2883121177054059,
      "grad_norm": 0.7910583019256592,
      "learning_rate": 0.00017058170687320144,
      "loss": 2.1932,
      "step": 1308
    },
    {
      "epoch": 0.2885325398137433,
      "grad_norm": 0.7655038237571716,
      "learning_rate": 0.0001705300527251081,
      "loss": 2.2,
      "step": 1309
    },
    {
      "epoch": 0.2887529619220808,
      "grad_norm": 0.76920086145401,
      "learning_rate": 0.0001704783611039344,
      "loss": 2.2193,
      "step": 1310
    },
    {
      "epoch": 0.2889733840304182,
      "grad_norm": 0.7292153239250183,
      "learning_rate": 0.00017042663203714446,
      "loss": 2.2773,
      "step": 1311
    },
    {
      "epoch": 0.2891938061387557,
      "grad_norm": 0.7255344390869141,
      "learning_rate": 0.00017037486555222225,
      "loss": 2.2684,
      "step": 1312
    },
    {
      "epoch": 0.2894142282470932,
      "grad_norm": 0.7909936308860779,
      "learning_rate": 0.00017032306167667164,
      "loss": 2.2231,
      "step": 1313
    },
    {
      "epoch": 0.2896346503554306,
      "grad_norm": 0.7148476839065552,
      "learning_rate": 0.0001702712204380164,
      "loss": 2.1157,
      "step": 1314
    },
    {
      "epoch": 0.2898550724637681,
      "grad_norm": 0.7406220436096191,
      "learning_rate": 0.00017021934186380012,
      "loss": 2.1401,
      "step": 1315
    },
    {
      "epoch": 0.2900754945721056,
      "grad_norm": 0.7505598664283752,
      "learning_rate": 0.00017016742598158615,
      "loss": 2.1095,
      "step": 1316
    },
    {
      "epoch": 0.290295916680443,
      "grad_norm": 0.771493673324585,
      "learning_rate": 0.00017011547281895782,
      "loss": 2.2446,
      "step": 1317
    },
    {
      "epoch": 0.2905163387887805,
      "grad_norm": 0.7649971842765808,
      "learning_rate": 0.0001700634824035182,
      "loss": 2.1406,
      "step": 1318
    },
    {
      "epoch": 0.290736760897118,
      "grad_norm": 0.7823048233985901,
      "learning_rate": 0.00017001145476289007,
      "loss": 2.2277,
      "step": 1319
    },
    {
      "epoch": 0.2909571830054554,
      "grad_norm": 0.7321711778640747,
      "learning_rate": 0.0001699593899247161,
      "loss": 2.2193,
      "step": 1320
    },
    {
      "epoch": 0.2911776051137929,
      "grad_norm": 0.8028239607810974,
      "learning_rate": 0.00016990728791665865,
      "loss": 2.2621,
      "step": 1321
    },
    {
      "epoch": 0.2913980272221304,
      "grad_norm": 0.7878668904304504,
      "learning_rate": 0.00016985514876639992,
      "loss": 2.1286,
      "step": 1322
    },
    {
      "epoch": 0.2916184493304678,
      "grad_norm": 0.7442121505737305,
      "learning_rate": 0.00016980297250164173,
      "loss": 2.1746,
      "step": 1323
    },
    {
      "epoch": 0.2918388714388053,
      "grad_norm": 0.760478138923645,
      "learning_rate": 0.00016975075915010566,
      "loss": 2.2902,
      "step": 1324
    },
    {
      "epoch": 0.2920592935471428,
      "grad_norm": 0.8058551549911499,
      "learning_rate": 0.00016969850873953303,
      "loss": 2.0822,
      "step": 1325
    },
    {
      "epoch": 0.2922797156554802,
      "grad_norm": 0.7996037006378174,
      "learning_rate": 0.00016964622129768487,
      "loss": 2.1685,
      "step": 1326
    },
    {
      "epoch": 0.2925001377638177,
      "grad_norm": 0.7167731523513794,
      "learning_rate": 0.00016959389685234174,
      "loss": 2.1068,
      "step": 1327
    },
    {
      "epoch": 0.2927205598721552,
      "grad_norm": 0.7742071747779846,
      "learning_rate": 0.00016954153543130405,
      "loss": 2.2076,
      "step": 1328
    },
    {
      "epoch": 0.2929409819804926,
      "grad_norm": 0.7153973579406738,
      "learning_rate": 0.00016948913706239175,
      "loss": 2.1974,
      "step": 1329
    },
    {
      "epoch": 0.2931614040888301,
      "grad_norm": 0.7211584448814392,
      "learning_rate": 0.00016943670177344445,
      "loss": 2.2222,
      "step": 1330
    },
    {
      "epoch": 0.2933818261971676,
      "grad_norm": 0.7888138890266418,
      "learning_rate": 0.0001693842295923213,
      "loss": 2.1708,
      "step": 1331
    },
    {
      "epoch": 0.293602248305505,
      "grad_norm": 0.7466800808906555,
      "learning_rate": 0.00016933172054690117,
      "loss": 2.3648,
      "step": 1332
    },
    {
      "epoch": 0.2938226704138425,
      "grad_norm": 0.7609893083572388,
      "learning_rate": 0.0001692791746650824,
      "loss": 2.1658,
      "step": 1333
    },
    {
      "epoch": 0.29404309252217997,
      "grad_norm": 0.7258310914039612,
      "learning_rate": 0.00016922659197478307,
      "loss": 2.0353,
      "step": 1334
    },
    {
      "epoch": 0.2942635146305174,
      "grad_norm": 0.8021120429039001,
      "learning_rate": 0.0001691739725039406,
      "loss": 2.1926,
      "step": 1335
    },
    {
      "epoch": 0.2944839367388549,
      "grad_norm": 0.726929783821106,
      "learning_rate": 0.00016912131628051214,
      "loss": 2.2789,
      "step": 1336
    },
    {
      "epoch": 0.29470435884719237,
      "grad_norm": 0.7556727528572083,
      "learning_rate": 0.00016906862333247423,
      "loss": 2.2068,
      "step": 1337
    },
    {
      "epoch": 0.2949247809555298,
      "grad_norm": 0.7518954873085022,
      "learning_rate": 0.00016901589368782303,
      "loss": 2.2541,
      "step": 1338
    },
    {
      "epoch": 0.2951452030638673,
      "grad_norm": 0.8447859287261963,
      "learning_rate": 0.00016896312737457408,
      "loss": 2.2507,
      "step": 1339
    },
    {
      "epoch": 0.29536562517220477,
      "grad_norm": 1.1242687702178955,
      "learning_rate": 0.00016891032442076253,
      "loss": 2.2894,
      "step": 1340
    },
    {
      "epoch": 0.2955860472805422,
      "grad_norm": 0.7969946265220642,
      "learning_rate": 0.0001688574848544429,
      "loss": 2.287,
      "step": 1341
    },
    {
      "epoch": 0.2958064693888797,
      "grad_norm": 0.758046567440033,
      "learning_rate": 0.00016880460870368922,
      "loss": 2.1963,
      "step": 1342
    },
    {
      "epoch": 0.29602689149721717,
      "grad_norm": 0.8002006411552429,
      "learning_rate": 0.00016875169599659496,
      "loss": 2.3307,
      "step": 1343
    },
    {
      "epoch": 0.2962473136055546,
      "grad_norm": 0.7745321989059448,
      "learning_rate": 0.00016869874676127294,
      "loss": 2.1107,
      "step": 1344
    },
    {
      "epoch": 0.2964677357138921,
      "grad_norm": 0.7248173952102661,
      "learning_rate": 0.00016864576102585546,
      "loss": 2.2054,
      "step": 1345
    },
    {
      "epoch": 0.29668815782222957,
      "grad_norm": 0.7786215543746948,
      "learning_rate": 0.00016859273881849416,
      "loss": 2.1615,
      "step": 1346
    },
    {
      "epoch": 0.296908579930567,
      "grad_norm": 0.7825655937194824,
      "learning_rate": 0.00016853968016736012,
      "loss": 2.1958,
      "step": 1347
    },
    {
      "epoch": 0.2971290020389045,
      "grad_norm": 0.7332908511161804,
      "learning_rate": 0.00016848658510064377,
      "loss": 2.1952,
      "step": 1348
    },
    {
      "epoch": 0.29734942414724197,
      "grad_norm": 0.7981249690055847,
      "learning_rate": 0.00016843345364655482,
      "loss": 2.1869,
      "step": 1349
    },
    {
      "epoch": 0.2975698462555794,
      "grad_norm": 0.7685749530792236,
      "learning_rate": 0.00016838028583332238,
      "loss": 2.2178,
      "step": 1350
    },
    {
      "epoch": 0.2977902683639169,
      "grad_norm": 0.7474701404571533,
      "learning_rate": 0.00016832708168919487,
      "loss": 2.1828,
      "step": 1351
    },
    {
      "epoch": 0.29801069047225437,
      "grad_norm": 0.8272140622138977,
      "learning_rate": 0.00016827384124243997,
      "loss": 2.2645,
      "step": 1352
    },
    {
      "epoch": 0.2982311125805918,
      "grad_norm": 0.7531329393386841,
      "learning_rate": 0.0001682205645213447,
      "loss": 2.1499,
      "step": 1353
    },
    {
      "epoch": 0.2984515346889293,
      "grad_norm": 0.7509371042251587,
      "learning_rate": 0.00016816725155421533,
      "loss": 2.1212,
      "step": 1354
    },
    {
      "epoch": 0.29867195679726677,
      "grad_norm": 0.7285223007202148,
      "learning_rate": 0.00016811390236937742,
      "loss": 2.2107,
      "step": 1355
    },
    {
      "epoch": 0.2988923789056042,
      "grad_norm": 0.7343676686286926,
      "learning_rate": 0.00016806051699517566,
      "loss": 2.2265,
      "step": 1356
    },
    {
      "epoch": 0.2991128010139417,
      "grad_norm": 0.7556804418563843,
      "learning_rate": 0.00016800709545997412,
      "loss": 2.1835,
      "step": 1357
    },
    {
      "epoch": 0.29933322312227917,
      "grad_norm": 0.7752113938331604,
      "learning_rate": 0.000167953637792156,
      "loss": 2.2281,
      "step": 1358
    },
    {
      "epoch": 0.2995536452306166,
      "grad_norm": 0.8000509142875671,
      "learning_rate": 0.00016790014402012372,
      "loss": 2.2829,
      "step": 1359
    },
    {
      "epoch": 0.2997740673389541,
      "grad_norm": 0.7730531096458435,
      "learning_rate": 0.00016784661417229885,
      "loss": 2.2826,
      "step": 1360
    },
    {
      "epoch": 0.29999448944729157,
      "grad_norm": 0.73183673620224,
      "learning_rate": 0.00016779304827712217,
      "loss": 2.0977,
      "step": 1361
    },
    {
      "epoch": 0.300214911555629,
      "grad_norm": 0.738059937953949,
      "learning_rate": 0.00016773944636305356,
      "loss": 2.1585,
      "step": 1362
    },
    {
      "epoch": 0.3004353336639665,
      "grad_norm": 0.7233401536941528,
      "learning_rate": 0.00016768580845857208,
      "loss": 2.2233,
      "step": 1363
    },
    {
      "epoch": 0.30065575577230397,
      "grad_norm": 0.7403899431228638,
      "learning_rate": 0.00016763213459217594,
      "loss": 2.1944,
      "step": 1364
    },
    {
      "epoch": 0.3008761778806414,
      "grad_norm": 0.7406932711601257,
      "learning_rate": 0.00016757842479238238,
      "loss": 2.1976,
      "step": 1365
    },
    {
      "epoch": 0.3010965999889789,
      "grad_norm": 0.7360256910324097,
      "learning_rate": 0.00016752467908772776,
      "loss": 2.1359,
      "step": 1366
    },
    {
      "epoch": 0.30131702209731637,
      "grad_norm": 0.7741120457649231,
      "learning_rate": 0.00016747089750676755,
      "loss": 2.1544,
      "step": 1367
    },
    {
      "epoch": 0.3015374442056538,
      "grad_norm": 0.7715880274772644,
      "learning_rate": 0.00016741708007807625,
      "loss": 2.3567,
      "step": 1368
    },
    {
      "epoch": 0.3017578663139913,
      "grad_norm": 0.8402259349822998,
      "learning_rate": 0.00016736322683024745,
      "loss": 2.13,
      "step": 1369
    },
    {
      "epoch": 0.30197828842232877,
      "grad_norm": 0.7581931352615356,
      "learning_rate": 0.00016730933779189364,
      "loss": 2.1747,
      "step": 1370
    },
    {
      "epoch": 0.3021987105306662,
      "grad_norm": 0.7869582772254944,
      "learning_rate": 0.00016725541299164652,
      "loss": 2.1287,
      "step": 1371
    },
    {
      "epoch": 0.3024191326390037,
      "grad_norm": 0.8170201778411865,
      "learning_rate": 0.00016720145245815662,
      "loss": 2.171,
      "step": 1372
    },
    {
      "epoch": 0.30263955474734117,
      "grad_norm": 0.7717443704605103,
      "learning_rate": 0.00016714745622009355,
      "loss": 2.3506,
      "step": 1373
    },
    {
      "epoch": 0.3028599768556786,
      "grad_norm": 0.7505819797515869,
      "learning_rate": 0.0001670934243061459,
      "loss": 2.2559,
      "step": 1374
    },
    {
      "epoch": 0.3030803989640161,
      "grad_norm": 0.7750145792961121,
      "learning_rate": 0.00016703935674502115,
      "loss": 2.1196,
      "step": 1375
    },
    {
      "epoch": 0.30330082107235357,
      "grad_norm": 0.8530693650245667,
      "learning_rate": 0.00016698525356544574,
      "loss": 2.2001,
      "step": 1376
    },
    {
      "epoch": 0.303521243180691,
      "grad_norm": 0.7216318845748901,
      "learning_rate": 0.00016693111479616505,
      "loss": 2.2052,
      "step": 1377
    },
    {
      "epoch": 0.30374166528902846,
      "grad_norm": 0.821911096572876,
      "learning_rate": 0.0001668769404659434,
      "loss": 2.2347,
      "step": 1378
    },
    {
      "epoch": 0.30396208739736597,
      "grad_norm": 0.7303586006164551,
      "learning_rate": 0.0001668227306035639,
      "loss": 2.3472,
      "step": 1379
    },
    {
      "epoch": 0.3041825095057034,
      "grad_norm": 0.751124918460846,
      "learning_rate": 0.00016676848523782868,
      "loss": 2.2703,
      "step": 1380
    },
    {
      "epoch": 0.30440293161404086,
      "grad_norm": 0.7501354217529297,
      "learning_rate": 0.00016671420439755857,
      "loss": 2.2338,
      "step": 1381
    },
    {
      "epoch": 0.30462335372237836,
      "grad_norm": 0.8355147838592529,
      "learning_rate": 0.00016665988811159345,
      "loss": 2.2313,
      "step": 1382
    },
    {
      "epoch": 0.3048437758307158,
      "grad_norm": 0.8050654530525208,
      "learning_rate": 0.00016660553640879182,
      "loss": 2.232,
      "step": 1383
    },
    {
      "epoch": 0.30506419793905326,
      "grad_norm": 0.732281506061554,
      "learning_rate": 0.00016655114931803116,
      "loss": 2.1346,
      "step": 1384
    },
    {
      "epoch": 0.30528462004739076,
      "grad_norm": 0.8058892488479614,
      "learning_rate": 0.00016649672686820766,
      "loss": 2.3091,
      "step": 1385
    },
    {
      "epoch": 0.3055050421557282,
      "grad_norm": 0.7754947543144226,
      "learning_rate": 0.00016644226908823633,
      "loss": 2.237,
      "step": 1386
    },
    {
      "epoch": 0.30572546426406566,
      "grad_norm": 0.7642831802368164,
      "learning_rate": 0.00016638777600705093,
      "loss": 2.1295,
      "step": 1387
    },
    {
      "epoch": 0.30594588637240316,
      "grad_norm": 0.7168982625007629,
      "learning_rate": 0.00016633324765360404,
      "loss": 2.2233,
      "step": 1388
    },
    {
      "epoch": 0.3061663084807406,
      "grad_norm": 0.7286316156387329,
      "learning_rate": 0.00016627868405686685,
      "loss": 2.1822,
      "step": 1389
    },
    {
      "epoch": 0.30638673058907806,
      "grad_norm": 0.8448231220245361,
      "learning_rate": 0.00016622408524582947,
      "loss": 2.1413,
      "step": 1390
    },
    {
      "epoch": 0.30660715269741556,
      "grad_norm": 0.7416625618934631,
      "learning_rate": 0.00016616945124950052,
      "loss": 2.1957,
      "step": 1391
    },
    {
      "epoch": 0.306827574805753,
      "grad_norm": 0.8372024297714233,
      "learning_rate": 0.0001661147820969074,
      "loss": 2.2813,
      "step": 1392
    },
    {
      "epoch": 0.30704799691409046,
      "grad_norm": 0.7228584885597229,
      "learning_rate": 0.00016606007781709624,
      "loss": 2.2801,
      "step": 1393
    },
    {
      "epoch": 0.30726841902242796,
      "grad_norm": 0.7378079891204834,
      "learning_rate": 0.00016600533843913175,
      "loss": 2.2618,
      "step": 1394
    },
    {
      "epoch": 0.3074888411307654,
      "grad_norm": 0.7407245635986328,
      "learning_rate": 0.00016595056399209735,
      "loss": 2.1034,
      "step": 1395
    },
    {
      "epoch": 0.30770926323910286,
      "grad_norm": 0.719724714756012,
      "learning_rate": 0.00016589575450509504,
      "loss": 2.1846,
      "step": 1396
    },
    {
      "epoch": 0.30792968534744036,
      "grad_norm": 0.8015205264091492,
      "learning_rate": 0.00016584091000724547,
      "loss": 2.2641,
      "step": 1397
    },
    {
      "epoch": 0.3081501074557778,
      "grad_norm": 0.7663782238960266,
      "learning_rate": 0.00016578603052768787,
      "loss": 2.3642,
      "step": 1398
    },
    {
      "epoch": 0.30837052956411526,
      "grad_norm": 0.7513239979743958,
      "learning_rate": 0.00016573111609558008,
      "loss": 2.0912,
      "step": 1399
    },
    {
      "epoch": 0.30859095167245276,
      "grad_norm": 0.7337515354156494,
      "learning_rate": 0.00016567616674009847,
      "loss": 2.2312,
      "step": 1400
    },
    {
      "epoch": 0.3088113737807902,
      "grad_norm": 0.7333741784095764,
      "learning_rate": 0.00016562118249043804,
      "loss": 2.2509,
      "step": 1401
    },
    {
      "epoch": 0.30903179588912766,
      "grad_norm": 0.8001041412353516,
      "learning_rate": 0.00016556616337581225,
      "loss": 2.2471,
      "step": 1402
    },
    {
      "epoch": 0.30925221799746516,
      "grad_norm": 0.7846009731292725,
      "learning_rate": 0.00016551110942545312,
      "loss": 2.1356,
      "step": 1403
    },
    {
      "epoch": 0.3094726401058026,
      "grad_norm": 0.7363162040710449,
      "learning_rate": 0.0001654560206686112,
      "loss": 2.1642,
      "step": 1404
    },
    {
      "epoch": 0.30969306221414006,
      "grad_norm": 0.7500562071800232,
      "learning_rate": 0.00016540089713455548,
      "loss": 2.1883,
      "step": 1405
    },
    {
      "epoch": 0.30991348432247756,
      "grad_norm": 0.7399133443832397,
      "learning_rate": 0.00016534573885257348,
      "loss": 2.2326,
      "step": 1406
    },
    {
      "epoch": 0.310133906430815,
      "grad_norm": 0.8109293580055237,
      "learning_rate": 0.0001652905458519711,
      "loss": 2.2469,
      "step": 1407
    },
    {
      "epoch": 0.31035432853915246,
      "grad_norm": 0.7497938871383667,
      "learning_rate": 0.00016523531816207285,
      "loss": 2.1969,
      "step": 1408
    },
    {
      "epoch": 0.31057475064748996,
      "grad_norm": 0.7552210092544556,
      "learning_rate": 0.00016518005581222148,
      "loss": 2.2249,
      "step": 1409
    },
    {
      "epoch": 0.3107951727558274,
      "grad_norm": 0.6931548118591309,
      "learning_rate": 0.00016512475883177826,
      "loss": 2.2557,
      "step": 1410
    },
    {
      "epoch": 0.31101559486416486,
      "grad_norm": 0.7809411287307739,
      "learning_rate": 0.00016506942725012281,
      "loss": 2.1998,
      "step": 1411
    },
    {
      "epoch": 0.31123601697250236,
      "grad_norm": 0.7492542862892151,
      "learning_rate": 0.00016501406109665324,
      "loss": 2.0384,
      "step": 1412
    },
    {
      "epoch": 0.3114564390808398,
      "grad_norm": 0.7817258238792419,
      "learning_rate": 0.00016495866040078586,
      "loss": 2.1613,
      "step": 1413
    },
    {
      "epoch": 0.31167686118917726,
      "grad_norm": 0.7840137481689453,
      "learning_rate": 0.0001649032251919555,
      "loss": 2.1394,
      "step": 1414
    },
    {
      "epoch": 0.31189728329751476,
      "grad_norm": 0.8244825601577759,
      "learning_rate": 0.00016484775549961524,
      "loss": 2.2096,
      "step": 1415
    },
    {
      "epoch": 0.3121177054058522,
      "grad_norm": 0.766228973865509,
      "learning_rate": 0.00016479225135323643,
      "loss": 2.2137,
      "step": 1416
    },
    {
      "epoch": 0.31233812751418966,
      "grad_norm": 0.7980702519416809,
      "learning_rate": 0.00016473671278230886,
      "loss": 2.1309,
      "step": 1417
    },
    {
      "epoch": 0.31255854962252716,
      "grad_norm": 0.7728880047798157,
      "learning_rate": 0.0001646811398163405,
      "loss": 2.2048,
      "step": 1418
    },
    {
      "epoch": 0.3127789717308646,
      "grad_norm": 0.7497773766517639,
      "learning_rate": 0.00016462553248485762,
      "loss": 2.2132,
      "step": 1419
    },
    {
      "epoch": 0.31299939383920206,
      "grad_norm": 0.8991342186927795,
      "learning_rate": 0.0001645698908174048,
      "loss": 2.3148,
      "step": 1420
    },
    {
      "epoch": 0.31321981594753956,
      "grad_norm": 0.7158945798873901,
      "learning_rate": 0.00016451421484354477,
      "loss": 2.189,
      "step": 1421
    },
    {
      "epoch": 0.313440238055877,
      "grad_norm": 0.7353668212890625,
      "learning_rate": 0.0001644585045928586,
      "loss": 2.1305,
      "step": 1422
    },
    {
      "epoch": 0.31366066016421446,
      "grad_norm": 0.8375792503356934,
      "learning_rate": 0.00016440276009494547,
      "loss": 2.3217,
      "step": 1423
    },
    {
      "epoch": 0.31388108227255196,
      "grad_norm": 0.8508047461509705,
      "learning_rate": 0.00016434698137942282,
      "loss": 2.2116,
      "step": 1424
    },
    {
      "epoch": 0.3141015043808894,
      "grad_norm": 0.7933481931686401,
      "learning_rate": 0.00016429116847592625,
      "loss": 2.1541,
      "step": 1425
    },
    {
      "epoch": 0.31432192648922685,
      "grad_norm": 0.8110189437866211,
      "learning_rate": 0.0001642353214141095,
      "loss": 2.2363,
      "step": 1426
    },
    {
      "epoch": 0.31454234859756436,
      "grad_norm": 0.8223790526390076,
      "learning_rate": 0.0001641794402236445,
      "loss": 2.2408,
      "step": 1427
    },
    {
      "epoch": 0.3147627707059018,
      "grad_norm": 0.7742643356323242,
      "learning_rate": 0.00016412352493422132,
      "loss": 2.222,
      "step": 1428
    },
    {
      "epoch": 0.31498319281423925,
      "grad_norm": 0.7825959920883179,
      "learning_rate": 0.00016406757557554802,
      "loss": 2.1374,
      "step": 1429
    },
    {
      "epoch": 0.31520361492257676,
      "grad_norm": 0.8011865019798279,
      "learning_rate": 0.00016401159217735098,
      "loss": 2.1977,
      "step": 1430
    },
    {
      "epoch": 0.3154240370309142,
      "grad_norm": 0.7719798684120178,
      "learning_rate": 0.00016395557476937448,
      "loss": 2.2113,
      "step": 1431
    },
    {
      "epoch": 0.31564445913925165,
      "grad_norm": 0.7831125855445862,
      "learning_rate": 0.00016389952338138097,
      "loss": 2.3064,
      "step": 1432
    },
    {
      "epoch": 0.31586488124758916,
      "grad_norm": 0.7797256708145142,
      "learning_rate": 0.00016384343804315088,
      "loss": 2.2048,
      "step": 1433
    },
    {
      "epoch": 0.3160853033559266,
      "grad_norm": 0.790550172328949,
      "learning_rate": 0.00016378731878448276,
      "loss": 2.3377,
      "step": 1434
    },
    {
      "epoch": 0.31630572546426405,
      "grad_norm": 0.7469916343688965,
      "learning_rate": 0.00016373116563519314,
      "loss": 2.2409,
      "step": 1435
    },
    {
      "epoch": 0.31652614757260156,
      "grad_norm": 0.7746044993400574,
      "learning_rate": 0.0001636749786251165,
      "loss": 2.2322,
      "step": 1436
    },
    {
      "epoch": 0.316746569680939,
      "grad_norm": 0.748622477054596,
      "learning_rate": 0.00016361875778410543,
      "loss": 2.0579,
      "step": 1437
    },
    {
      "epoch": 0.31696699178927645,
      "grad_norm": 0.7819235920906067,
      "learning_rate": 0.00016356250314203044,
      "loss": 2.2052,
      "step": 1438
    },
    {
      "epoch": 0.31718741389761396,
      "grad_norm": 0.7809705138206482,
      "learning_rate": 0.00016350621472877993,
      "loss": 2.2054,
      "step": 1439
    },
    {
      "epoch": 0.3174078360059514,
      "grad_norm": 0.8232284784317017,
      "learning_rate": 0.00016344989257426033,
      "loss": 2.2361,
      "step": 1440
    },
    {
      "epoch": 0.31762825811428885,
      "grad_norm": 0.7503793239593506,
      "learning_rate": 0.000163393536708396,
      "loss": 2.1401,
      "step": 1441
    },
    {
      "epoch": 0.31784868022262636,
      "grad_norm": 0.7937236428260803,
      "learning_rate": 0.00016333714716112915,
      "loss": 2.0548,
      "step": 1442
    },
    {
      "epoch": 0.3180691023309638,
      "grad_norm": 0.8357060551643372,
      "learning_rate": 0.00016328072396241993,
      "loss": 2.4029,
      "step": 1443
    },
    {
      "epoch": 0.31828952443930125,
      "grad_norm": 0.7146298289299011,
      "learning_rate": 0.00016322426714224632,
      "loss": 2.2743,
      "step": 1444
    },
    {
      "epoch": 0.31850994654763876,
      "grad_norm": 0.7832850813865662,
      "learning_rate": 0.00016316777673060426,
      "loss": 2.1769,
      "step": 1445
    },
    {
      "epoch": 0.3187303686559762,
      "grad_norm": 0.7765694856643677,
      "learning_rate": 0.00016311125275750746,
      "loss": 2.2619,
      "step": 1446
    },
    {
      "epoch": 0.31895079076431365,
      "grad_norm": 0.8856155276298523,
      "learning_rate": 0.00016305469525298744,
      "loss": 2.3,
      "step": 1447
    },
    {
      "epoch": 0.3191712128726511,
      "grad_norm": 0.7651874423027039,
      "learning_rate": 0.0001629981042470936,
      "loss": 2.1635,
      "step": 1448
    },
    {
      "epoch": 0.3193916349809886,
      "grad_norm": 0.7492339015007019,
      "learning_rate": 0.0001629414797698931,
      "loss": 2.2147,
      "step": 1449
    },
    {
      "epoch": 0.31961205708932605,
      "grad_norm": 0.8455783128738403,
      "learning_rate": 0.00016288482185147096,
      "loss": 2.1922,
      "step": 1450
    },
    {
      "epoch": 0.3198324791976635,
      "grad_norm": 0.8730701804161072,
      "learning_rate": 0.00016282813052192983,
      "loss": 2.1427,
      "step": 1451
    },
    {
      "epoch": 0.320052901306001,
      "grad_norm": 0.7711134552955627,
      "learning_rate": 0.0001627714058113902,
      "loss": 2.2935,
      "step": 1452
    },
    {
      "epoch": 0.32027332341433845,
      "grad_norm": 0.6922009587287903,
      "learning_rate": 0.0001627146477499903,
      "loss": 2.1869,
      "step": 1453
    },
    {
      "epoch": 0.3204937455226759,
      "grad_norm": 0.8023621439933777,
      "learning_rate": 0.00016265785636788605,
      "loss": 2.2151,
      "step": 1454
    },
    {
      "epoch": 0.3207141676310134,
      "grad_norm": 0.7575031518936157,
      "learning_rate": 0.0001626010316952511,
      "loss": 2.126,
      "step": 1455
    },
    {
      "epoch": 0.32093458973935085,
      "grad_norm": 0.8092314004898071,
      "learning_rate": 0.00016254417376227674,
      "loss": 2.1763,
      "step": 1456
    },
    {
      "epoch": 0.3211550118476883,
      "grad_norm": 0.7591700553894043,
      "learning_rate": 0.00016248728259917196,
      "loss": 2.3187,
      "step": 1457
    },
    {
      "epoch": 0.3213754339560258,
      "grad_norm": 0.7855573296546936,
      "learning_rate": 0.00016243035823616347,
      "loss": 2.1667,
      "step": 1458
    },
    {
      "epoch": 0.32159585606436325,
      "grad_norm": 0.7513642907142639,
      "learning_rate": 0.00016237340070349548,
      "loss": 2.1833,
      "step": 1459
    },
    {
      "epoch": 0.3218162781727007,
      "grad_norm": 0.8085818290710449,
      "learning_rate": 0.00016231641003142997,
      "loss": 2.2943,
      "step": 1460
    },
    {
      "epoch": 0.3220367002810382,
      "grad_norm": 0.7721956968307495,
      "learning_rate": 0.00016225938625024638,
      "loss": 2.1689,
      "step": 1461
    },
    {
      "epoch": 0.32225712238937565,
      "grad_norm": 0.7419075965881348,
      "learning_rate": 0.00016220232939024193,
      "loss": 2.1791,
      "step": 1462
    },
    {
      "epoch": 0.3224775444977131,
      "grad_norm": 0.7640455365180969,
      "learning_rate": 0.00016214523948173123,
      "loss": 2.2023,
      "step": 1463
    },
    {
      "epoch": 0.3226979666060506,
      "grad_norm": 0.7739687561988831,
      "learning_rate": 0.00016208811655504652,
      "loss": 2.1988,
      "step": 1464
    },
    {
      "epoch": 0.32291838871438805,
      "grad_norm": 0.816199779510498,
      "learning_rate": 0.0001620309606405376,
      "loss": 2.23,
      "step": 1465
    },
    {
      "epoch": 0.3231388108227255,
      "grad_norm": 0.790183424949646,
      "learning_rate": 0.00016197377176857178,
      "loss": 2.2376,
      "step": 1466
    },
    {
      "epoch": 0.323359232931063,
      "grad_norm": 0.8260217905044556,
      "learning_rate": 0.0001619165499695339,
      "loss": 2.2365,
      "step": 1467
    },
    {
      "epoch": 0.32357965503940045,
      "grad_norm": 0.8381696343421936,
      "learning_rate": 0.0001618592952738263,
      "loss": 2.2766,
      "step": 1468
    },
    {
      "epoch": 0.3238000771477379,
      "grad_norm": 0.7686355710029602,
      "learning_rate": 0.0001618020077118687,
      "loss": 2.1388,
      "step": 1469
    },
    {
      "epoch": 0.3240204992560754,
      "grad_norm": 0.8632002472877502,
      "learning_rate": 0.00016174468731409836,
      "loss": 2.2509,
      "step": 1470
    },
    {
      "epoch": 0.32424092136441285,
      "grad_norm": 0.7807295322418213,
      "learning_rate": 0.0001616873341109701,
      "loss": 2.2612,
      "step": 1471
    },
    {
      "epoch": 0.3244613434727503,
      "grad_norm": 0.8067259788513184,
      "learning_rate": 0.00016162994813295593,
      "loss": 2.1595,
      "step": 1472
    },
    {
      "epoch": 0.3246817655810878,
      "grad_norm": 0.8152427673339844,
      "learning_rate": 0.00016157252941054545,
      "loss": 2.1497,
      "step": 1473
    },
    {
      "epoch": 0.32490218768942525,
      "grad_norm": 0.7492986917495728,
      "learning_rate": 0.00016151507797424562,
      "loss": 2.1506,
      "step": 1474
    },
    {
      "epoch": 0.3251226097977627,
      "grad_norm": 0.8071069121360779,
      "learning_rate": 0.00016145759385458071,
      "loss": 2.1895,
      "step": 1475
    },
    {
      "epoch": 0.3253430319061002,
      "grad_norm": 0.7536904811859131,
      "learning_rate": 0.0001614000770820925,
      "loss": 2.3289,
      "step": 1476
    },
    {
      "epoch": 0.32556345401443765,
      "grad_norm": 0.7741612792015076,
      "learning_rate": 0.00016134252768733992,
      "loss": 2.1805,
      "step": 1477
    },
    {
      "epoch": 0.3257838761227751,
      "grad_norm": 0.7815119028091431,
      "learning_rate": 0.00016128494570089944,
      "loss": 2.251,
      "step": 1478
    },
    {
      "epoch": 0.3260042982311126,
      "grad_norm": 0.7355189919471741,
      "learning_rate": 0.0001612273311533647,
      "loss": 2.1921,
      "step": 1479
    },
    {
      "epoch": 0.32622472033945005,
      "grad_norm": 0.7572214007377625,
      "learning_rate": 0.00016116968407534665,
      "loss": 2.2178,
      "step": 1480
    },
    {
      "epoch": 0.3264451424477875,
      "grad_norm": 0.7917924523353577,
      "learning_rate": 0.0001611120044974736,
      "loss": 2.1547,
      "step": 1481
    },
    {
      "epoch": 0.326665564556125,
      "grad_norm": 0.7619233131408691,
      "learning_rate": 0.00016105429245039115,
      "loss": 2.2133,
      "step": 1482
    },
    {
      "epoch": 0.32688598666446245,
      "grad_norm": 0.8287803530693054,
      "learning_rate": 0.00016099654796476197,
      "loss": 2.298,
      "step": 1483
    },
    {
      "epoch": 0.3271064087727999,
      "grad_norm": 0.779815137386322,
      "learning_rate": 0.00016093877107126615,
      "loss": 2.2306,
      "step": 1484
    },
    {
      "epoch": 0.3273268308811374,
      "grad_norm": 0.7546185851097107,
      "learning_rate": 0.00016088096180060096,
      "loss": 2.2316,
      "step": 1485
    },
    {
      "epoch": 0.32754725298947485,
      "grad_norm": 0.8377430438995361,
      "learning_rate": 0.0001608231201834808,
      "loss": 2.2347,
      "step": 1486
    },
    {
      "epoch": 0.3277676750978123,
      "grad_norm": 0.7637773752212524,
      "learning_rate": 0.0001607652462506373,
      "loss": 2.1951,
      "step": 1487
    },
    {
      "epoch": 0.3279880972061498,
      "grad_norm": 0.7494173049926758,
      "learning_rate": 0.0001607073400328193,
      "loss": 2.1758,
      "step": 1488
    },
    {
      "epoch": 0.32820851931448725,
      "grad_norm": 0.7730987668037415,
      "learning_rate": 0.00016064940156079271,
      "loss": 2.2343,
      "step": 1489
    },
    {
      "epoch": 0.3284289414228247,
      "grad_norm": 0.7987918853759766,
      "learning_rate": 0.00016059143086534063,
      "loss": 2.0845,
      "step": 1490
    },
    {
      "epoch": 0.3286493635311622,
      "grad_norm": 0.7362668514251709,
      "learning_rate": 0.0001605334279772633,
      "loss": 2.1691,
      "step": 1491
    },
    {
      "epoch": 0.32886978563949965,
      "grad_norm": 0.7510612607002258,
      "learning_rate": 0.00016047539292737804,
      "loss": 2.2708,
      "step": 1492
    },
    {
      "epoch": 0.3290902077478371,
      "grad_norm": 0.7529436349868774,
      "learning_rate": 0.0001604173257465192,
      "loss": 2.1966,
      "step": 1493
    },
    {
      "epoch": 0.3293106298561746,
      "grad_norm": 0.7356380224227905,
      "learning_rate": 0.0001603592264655383,
      "loss": 2.2416,
      "step": 1494
    },
    {
      "epoch": 0.32953105196451205,
      "grad_norm": 0.72124183177948,
      "learning_rate": 0.00016030109511530383,
      "loss": 2.1453,
      "step": 1495
    },
    {
      "epoch": 0.3297514740728495,
      "grad_norm": 0.7277483344078064,
      "learning_rate": 0.00016024293172670138,
      "loss": 2.1706,
      "step": 1496
    },
    {
      "epoch": 0.329971896181187,
      "grad_norm": 0.7867790460586548,
      "learning_rate": 0.00016018473633063355,
      "loss": 2.3274,
      "step": 1497
    },
    {
      "epoch": 0.33019231828952444,
      "grad_norm": 0.741937518119812,
      "learning_rate": 0.00016012650895801995,
      "loss": 2.2181,
      "step": 1498
    },
    {
      "epoch": 0.3304127403978619,
      "grad_norm": 0.8306846618652344,
      "learning_rate": 0.00016006824963979711,
      "loss": 2.2136,
      "step": 1499
    },
    {
      "epoch": 0.3306331625061994,
      "grad_norm": 0.7927491068840027,
      "learning_rate": 0.00016000995840691865,
      "loss": 2.2182,
      "step": 1500
    },
    {
      "epoch": 0.3306331625061994,
      "eval_bpb": 1.137543681247871,
      "eval_loss": 2.192302703857422,
      "eval_runtime": 184.2393,
      "eval_samples_per_second": 62.202,
      "eval_steps_per_second": 7.778,
      "step": 1500
    },
    {
      "epoch": 0.33085358461453684,
      "grad_norm": 0.7413404583930969,
      "learning_rate": 0.00015995163529035504,
      "loss": 2.2239,
      "step": 1501
    },
    {
      "epoch": 0.3310740067228743,
      "grad_norm": 0.7550901174545288,
      "learning_rate": 0.0001598932803210938,
      "loss": 2.4256,
      "step": 1502
    },
    {
      "epoch": 0.3312944288312118,
      "grad_norm": 0.7998407483100891,
      "learning_rate": 0.0001598348935301392,
      "loss": 2.0832,
      "step": 1503
    },
    {
      "epoch": 0.33151485093954924,
      "grad_norm": 0.7943581342697144,
      "learning_rate": 0.00015977647494851263,
      "loss": 2.1966,
      "step": 1504
    },
    {
      "epoch": 0.3317352730478867,
      "grad_norm": 0.7476860880851746,
      "learning_rate": 0.00015971802460725218,
      "loss": 2.2208,
      "step": 1505
    },
    {
      "epoch": 0.3319556951562242,
      "grad_norm": 0.7920387387275696,
      "learning_rate": 0.000159659542537413,
      "loss": 2.1932,
      "step": 1506
    },
    {
      "epoch": 0.33217611726456164,
      "grad_norm": 0.7473288178443909,
      "learning_rate": 0.0001596010287700669,
      "loss": 2.1886,
      "step": 1507
    },
    {
      "epoch": 0.3323965393728991,
      "grad_norm": 0.7604488134384155,
      "learning_rate": 0.00015954248333630266,
      "loss": 2.0854,
      "step": 1508
    },
    {
      "epoch": 0.3326169614812366,
      "grad_norm": 0.7953976988792419,
      "learning_rate": 0.0001594839062672259,
      "loss": 2.2359,
      "step": 1509
    },
    {
      "epoch": 0.33283738358957404,
      "grad_norm": 0.7268176078796387,
      "learning_rate": 0.00015942529759395896,
      "loss": 2.1622,
      "step": 1510
    },
    {
      "epoch": 0.3330578056979115,
      "grad_norm": 0.7924153208732605,
      "learning_rate": 0.000159366657347641,
      "loss": 2.2657,
      "step": 1511
    },
    {
      "epoch": 0.333278227806249,
      "grad_norm": 0.7428545355796814,
      "learning_rate": 0.00015930798555942798,
      "loss": 2.0549,
      "step": 1512
    },
    {
      "epoch": 0.33349864991458644,
      "grad_norm": 0.768286406993866,
      "learning_rate": 0.00015924928226049265,
      "loss": 2.1973,
      "step": 1513
    },
    {
      "epoch": 0.3337190720229239,
      "grad_norm": 0.7660419344902039,
      "learning_rate": 0.0001591905474820244,
      "loss": 2.2837,
      "step": 1514
    },
    {
      "epoch": 0.33393949413126134,
      "grad_norm": 0.8032327890396118,
      "learning_rate": 0.00015913178125522942,
      "loss": 2.2646,
      "step": 1515
    },
    {
      "epoch": 0.33415991623959884,
      "grad_norm": 0.775571346282959,
      "learning_rate": 0.00015907298361133063,
      "loss": 2.2424,
      "step": 1516
    },
    {
      "epoch": 0.3343803383479363,
      "grad_norm": 0.7613062858581543,
      "learning_rate": 0.00015901415458156754,
      "loss": 2.2368,
      "step": 1517
    },
    {
      "epoch": 0.33460076045627374,
      "grad_norm": 0.7357932329177856,
      "learning_rate": 0.00015895529419719643,
      "loss": 2.1813,
      "step": 1518
    },
    {
      "epoch": 0.33482118256461124,
      "grad_norm": 0.7090514898300171,
      "learning_rate": 0.00015889640248949025,
      "loss": 2.2191,
      "step": 1519
    },
    {
      "epoch": 0.3350416046729487,
      "grad_norm": 0.6923114657402039,
      "learning_rate": 0.0001588374794897385,
      "loss": 2.2341,
      "step": 1520
    },
    {
      "epoch": 0.33526202678128614,
      "grad_norm": 0.7307350039482117,
      "learning_rate": 0.00015877852522924732,
      "loss": 2.2264,
      "step": 1521
    },
    {
      "epoch": 0.33548244888962364,
      "grad_norm": 0.7955405116081238,
      "learning_rate": 0.0001587195397393396,
      "loss": 2.1738,
      "step": 1522
    },
    {
      "epoch": 0.3357028709979611,
      "grad_norm": 0.8326416015625,
      "learning_rate": 0.00015866052305135462,
      "loss": 2.1491,
      "step": 1523
    },
    {
      "epoch": 0.33592329310629854,
      "grad_norm": 0.7896016240119934,
      "learning_rate": 0.0001586014751966484,
      "loss": 2.2215,
      "step": 1524
    },
    {
      "epoch": 0.33614371521463604,
      "grad_norm": 0.7470634579658508,
      "learning_rate": 0.0001585423962065934,
      "loss": 2.1597,
      "step": 1525
    },
    {
      "epoch": 0.3363641373229735,
      "grad_norm": 0.7455403804779053,
      "learning_rate": 0.00015848328611257876,
      "loss": 2.2065,
      "step": 1526
    },
    {
      "epoch": 0.33658455943131094,
      "grad_norm": 0.7747969031333923,
      "learning_rate": 0.00015842414494601,
      "loss": 2.1769,
      "step": 1527
    },
    {
      "epoch": 0.33680498153964844,
      "grad_norm": 0.7751805782318115,
      "learning_rate": 0.0001583649727383092,
      "loss": 2.2055,
      "step": 1528
    },
    {
      "epoch": 0.3370254036479859,
      "grad_norm": 0.7543962597846985,
      "learning_rate": 0.00015830576952091495,
      "loss": 2.2922,
      "step": 1529
    },
    {
      "epoch": 0.33724582575632334,
      "grad_norm": 0.7450518012046814,
      "learning_rate": 0.00015824653532528236,
      "loss": 2.2203,
      "step": 1530
    },
    {
      "epoch": 0.33746624786466084,
      "grad_norm": 0.736697793006897,
      "learning_rate": 0.00015818727018288286,
      "loss": 2.1914,
      "step": 1531
    },
    {
      "epoch": 0.3376866699729983,
      "grad_norm": 0.7560701966285706,
      "learning_rate": 0.00015812797412520445,
      "loss": 2.2982,
      "step": 1532
    },
    {
      "epoch": 0.33790709208133574,
      "grad_norm": 0.736017107963562,
      "learning_rate": 0.00015806864718375158,
      "loss": 2.1076,
      "step": 1533
    },
    {
      "epoch": 0.33812751418967324,
      "grad_norm": 0.7118814587593079,
      "learning_rate": 0.00015800928939004494,
      "loss": 2.1976,
      "step": 1534
    },
    {
      "epoch": 0.3383479362980107,
      "grad_norm": 0.764025092124939,
      "learning_rate": 0.00015794990077562176,
      "loss": 2.189,
      "step": 1535
    },
    {
      "epoch": 0.33856835840634814,
      "grad_norm": 0.7253943681716919,
      "learning_rate": 0.00015789048137203562,
      "loss": 2.3437,
      "step": 1536
    },
    {
      "epoch": 0.33878878051468564,
      "grad_norm": 0.7427663207054138,
      "learning_rate": 0.0001578310312108564,
      "loss": 2.2179,
      "step": 1537
    },
    {
      "epoch": 0.3390092026230231,
      "grad_norm": 0.7878320217132568,
      "learning_rate": 0.0001577715503236704,
      "loss": 2.1847,
      "step": 1538
    },
    {
      "epoch": 0.33922962473136054,
      "grad_norm": 0.7854688167572021,
      "learning_rate": 0.00015771203874208017,
      "loss": 2.1872,
      "step": 1539
    },
    {
      "epoch": 0.33945004683969804,
      "grad_norm": 0.8263827562332153,
      "learning_rate": 0.00015765249649770464,
      "loss": 2.2014,
      "step": 1540
    },
    {
      "epoch": 0.3396704689480355,
      "grad_norm": 0.8038614988327026,
      "learning_rate": 0.00015759292362217897,
      "loss": 2.1451,
      "step": 1541
    },
    {
      "epoch": 0.33989089105637293,
      "grad_norm": 0.7991663217544556,
      "learning_rate": 0.0001575333201471546,
      "loss": 2.2033,
      "step": 1542
    },
    {
      "epoch": 0.34011131316471044,
      "grad_norm": 0.843662440776825,
      "learning_rate": 0.0001574736861042993,
      "loss": 2.2797,
      "step": 1543
    },
    {
      "epoch": 0.3403317352730479,
      "grad_norm": 0.841734766960144,
      "learning_rate": 0.00015741402152529704,
      "loss": 2.1378,
      "step": 1544
    },
    {
      "epoch": 0.34055215738138533,
      "grad_norm": 0.807227611541748,
      "learning_rate": 0.0001573543264418479,
      "loss": 2.2053,
      "step": 1545
    },
    {
      "epoch": 0.34077257948972284,
      "grad_norm": 0.8120733499526978,
      "learning_rate": 0.0001572946008856684,
      "loss": 2.0658,
      "step": 1546
    },
    {
      "epoch": 0.3409930015980603,
      "grad_norm": 0.7973092198371887,
      "learning_rate": 0.000157234844888491,
      "loss": 2.1838,
      "step": 1547
    },
    {
      "epoch": 0.34121342370639773,
      "grad_norm": 0.7541339993476868,
      "learning_rate": 0.00015717505848206455,
      "loss": 2.1625,
      "step": 1548
    },
    {
      "epoch": 0.34143384581473524,
      "grad_norm": 0.7628739476203918,
      "learning_rate": 0.00015711524169815394,
      "loss": 2.1822,
      "step": 1549
    },
    {
      "epoch": 0.3416542679230727,
      "grad_norm": 0.8073936700820923,
      "learning_rate": 0.00015705539456854016,
      "loss": 2.1726,
      "step": 1550
    },
    {
      "epoch": 0.34187469003141013,
      "grad_norm": 0.7674597501754761,
      "learning_rate": 0.00015699551712502043,
      "loss": 2.2265,
      "step": 1551
    },
    {
      "epoch": 0.34209511213974764,
      "grad_norm": 0.7606397271156311,
      "learning_rate": 0.00015693560939940803,
      "loss": 2.1934,
      "step": 1552
    },
    {
      "epoch": 0.3423155342480851,
      "grad_norm": 0.7480971217155457,
      "learning_rate": 0.0001568756714235323,
      "loss": 2.0554,
      "step": 1553
    },
    {
      "epoch": 0.34253595635642253,
      "grad_norm": 0.8327158689498901,
      "learning_rate": 0.00015681570322923874,
      "loss": 2.2332,
      "step": 1554
    },
    {
      "epoch": 0.34275637846476004,
      "grad_norm": 0.7172732949256897,
      "learning_rate": 0.0001567557048483888,
      "loss": 2.1312,
      "step": 1555
    },
    {
      "epoch": 0.3429768005730975,
      "grad_norm": 0.72996985912323,
      "learning_rate": 0.00015669567631286,
      "loss": 2.1475,
      "step": 1556
    },
    {
      "epoch": 0.34319722268143493,
      "grad_norm": 0.7929143309593201,
      "learning_rate": 0.00015663561765454597,
      "loss": 2.1875,
      "step": 1557
    },
    {
      "epoch": 0.34341764478977244,
      "grad_norm": 0.8069830536842346,
      "learning_rate": 0.0001565755289053562,
      "loss": 2.299,
      "step": 1558
    },
    {
      "epoch": 0.3436380668981099,
      "grad_norm": 0.883747398853302,
      "learning_rate": 0.0001565154100972162,
      "loss": 2.154,
      "step": 1559
    },
    {
      "epoch": 0.34385848900644733,
      "grad_norm": 0.7093352675437927,
      "learning_rate": 0.0001564552612620676,
      "loss": 2.0216,
      "step": 1560
    },
    {
      "epoch": 0.34407891111478484,
      "grad_norm": 0.8060243725776672,
      "learning_rate": 0.0001563950824318678,
      "loss": 2.0867,
      "step": 1561
    },
    {
      "epoch": 0.3442993332231223,
      "grad_norm": 0.8463153839111328,
      "learning_rate": 0.0001563348736385902,
      "loss": 2.2689,
      "step": 1562
    },
    {
      "epoch": 0.34451975533145973,
      "grad_norm": 0.9291605949401855,
      "learning_rate": 0.00015627463491422417,
      "loss": 2.1808,
      "step": 1563
    },
    {
      "epoch": 0.34474017743979724,
      "grad_norm": 0.7537291646003723,
      "learning_rate": 0.0001562143662907749,
      "loss": 2.2237,
      "step": 1564
    },
    {
      "epoch": 0.3449605995481347,
      "grad_norm": 0.701511800289154,
      "learning_rate": 0.00015615406780026354,
      "loss": 2.1741,
      "step": 1565
    },
    {
      "epoch": 0.34518102165647213,
      "grad_norm": 0.7842611074447632,
      "learning_rate": 0.00015609373947472705,
      "loss": 2.2113,
      "step": 1566
    },
    {
      "epoch": 0.34540144376480963,
      "grad_norm": 0.7826152443885803,
      "learning_rate": 0.00015603338134621828,
      "loss": 2.3292,
      "step": 1567
    },
    {
      "epoch": 0.3456218658731471,
      "grad_norm": 0.8300393223762512,
      "learning_rate": 0.0001559729934468059,
      "loss": 2.1961,
      "step": 1568
    },
    {
      "epoch": 0.34584228798148453,
      "grad_norm": 0.8215047717094421,
      "learning_rate": 0.00015591257580857437,
      "loss": 2.2381,
      "step": 1569
    },
    {
      "epoch": 0.34606271008982203,
      "grad_norm": 0.7488734722137451,
      "learning_rate": 0.00015585212846362401,
      "loss": 2.2308,
      "step": 1570
    },
    {
      "epoch": 0.3462831321981595,
      "grad_norm": 0.7769961953163147,
      "learning_rate": 0.00015579165144407087,
      "loss": 2.2531,
      "step": 1571
    },
    {
      "epoch": 0.34650355430649693,
      "grad_norm": 0.7778148651123047,
      "learning_rate": 0.00015573114478204683,
      "loss": 2.177,
      "step": 1572
    },
    {
      "epoch": 0.34672397641483443,
      "grad_norm": 0.7712897062301636,
      "learning_rate": 0.00015567060850969945,
      "loss": 2.1973,
      "step": 1573
    },
    {
      "epoch": 0.3469443985231719,
      "grad_norm": 0.799004316329956,
      "learning_rate": 0.00015561004265919202,
      "loss": 2.1111,
      "step": 1574
    },
    {
      "epoch": 0.34716482063150933,
      "grad_norm": 0.7687596082687378,
      "learning_rate": 0.0001555494472627036,
      "loss": 2.1888,
      "step": 1575
    },
    {
      "epoch": 0.34738524273984683,
      "grad_norm": 0.7000755071640015,
      "learning_rate": 0.00015548882235242895,
      "loss": 2.1223,
      "step": 1576
    },
    {
      "epoch": 0.3476056648481843,
      "grad_norm": 0.7892736792564392,
      "learning_rate": 0.00015542816796057843,
      "loss": 2.1704,
      "step": 1577
    },
    {
      "epoch": 0.34782608695652173,
      "grad_norm": 0.7471149563789368,
      "learning_rate": 0.00015536748411937814,
      "loss": 2.2761,
      "step": 1578
    },
    {
      "epoch": 0.34804650906485923,
      "grad_norm": 0.7847639918327332,
      "learning_rate": 0.0001553067708610698,
      "loss": 2.1286,
      "step": 1579
    },
    {
      "epoch": 0.3482669311731967,
      "grad_norm": 0.7326849699020386,
      "learning_rate": 0.00015524602821791075,
      "loss": 2.2141,
      "step": 1580
    },
    {
      "epoch": 0.34848735328153413,
      "grad_norm": 0.7386582493782043,
      "learning_rate": 0.00015518525622217402,
      "loss": 2.1862,
      "step": 1581
    },
    {
      "epoch": 0.3487077753898716,
      "grad_norm": 0.75742107629776,
      "learning_rate": 0.0001551244549061481,
      "loss": 2.2623,
      "step": 1582
    },
    {
      "epoch": 0.3489281974982091,
      "grad_norm": 0.7756809592247009,
      "learning_rate": 0.00015506362430213718,
      "loss": 2.2892,
      "step": 1583
    },
    {
      "epoch": 0.34914861960654653,
      "grad_norm": 0.7680872678756714,
      "learning_rate": 0.00015500276444246092,
      "loss": 2.1142,
      "step": 1584
    },
    {
      "epoch": 0.349369041714884,
      "grad_norm": 0.7100783586502075,
      "learning_rate": 0.00015494187535945457,
      "loss": 2.2498,
      "step": 1585
    },
    {
      "epoch": 0.3495894638232215,
      "grad_norm": 0.780300498008728,
      "learning_rate": 0.0001548809570854689,
      "loss": 2.2168,
      "step": 1586
    },
    {
      "epoch": 0.34980988593155893,
      "grad_norm": 0.748386561870575,
      "learning_rate": 0.0001548200096528702,
      "loss": 2.1408,
      "step": 1587
    },
    {
      "epoch": 0.3500303080398964,
      "grad_norm": 0.7577478289604187,
      "learning_rate": 0.00015475903309404023,
      "loss": 2.2191,
      "step": 1588
    },
    {
      "epoch": 0.3502507301482339,
      "grad_norm": 0.7430839538574219,
      "learning_rate": 0.00015469802744137625,
      "loss": 2.1625,
      "step": 1589
    },
    {
      "epoch": 0.3504711522565713,
      "grad_norm": 0.7901488542556763,
      "learning_rate": 0.00015463699272729096,
      "loss": 2.2161,
      "step": 1590
    },
    {
      "epoch": 0.3506915743649088,
      "grad_norm": 0.7919915318489075,
      "learning_rate": 0.0001545759289842125,
      "loss": 2.169,
      "step": 1591
    },
    {
      "epoch": 0.3509119964732463,
      "grad_norm": 0.8144817352294922,
      "learning_rate": 0.0001545148362445844,
      "loss": 2.1496,
      "step": 1592
    },
    {
      "epoch": 0.3511324185815837,
      "grad_norm": 0.7012849450111389,
      "learning_rate": 0.00015445371454086572,
      "loss": 2.3247,
      "step": 1593
    },
    {
      "epoch": 0.3513528406899212,
      "grad_norm": 0.7154977321624756,
      "learning_rate": 0.00015439256390553077,
      "loss": 2.1242,
      "step": 1594
    },
    {
      "epoch": 0.3515732627982587,
      "grad_norm": 0.7194611430168152,
      "learning_rate": 0.00015433138437106927,
      "loss": 2.204,
      "step": 1595
    },
    {
      "epoch": 0.3517936849065961,
      "grad_norm": 0.6987180709838867,
      "learning_rate": 0.00015427017596998633,
      "loss": 2.1304,
      "step": 1596
    },
    {
      "epoch": 0.3520141070149336,
      "grad_norm": 0.74656081199646,
      "learning_rate": 0.0001542089387348024,
      "loss": 2.2921,
      "step": 1597
    },
    {
      "epoch": 0.3522345291232711,
      "grad_norm": 0.7372604608535767,
      "learning_rate": 0.00015414767269805317,
      "loss": 2.1771,
      "step": 1598
    },
    {
      "epoch": 0.3524549512316085,
      "grad_norm": 0.7552307844161987,
      "learning_rate": 0.00015408637789228973,
      "loss": 2.1646,
      "step": 1599
    },
    {
      "epoch": 0.352675373339946,
      "grad_norm": 0.801944375038147,
      "learning_rate": 0.0001540250543500784,
      "loss": 2.2511,
      "step": 1600
    },
    {
      "epoch": 0.3528957954482835,
      "grad_norm": 0.7501528263092041,
      "learning_rate": 0.0001539637021040008,
      "loss": 2.1626,
      "step": 1601
    },
    {
      "epoch": 0.3531162175566209,
      "grad_norm": 0.7769619822502136,
      "learning_rate": 0.00015390232118665372,
      "loss": 2.2741,
      "step": 1602
    },
    {
      "epoch": 0.3533366396649584,
      "grad_norm": 0.7637968063354492,
      "learning_rate": 0.0001538409116306493,
      "loss": 2.0642,
      "step": 1603
    },
    {
      "epoch": 0.3535570617732959,
      "grad_norm": 0.7950186133384705,
      "learning_rate": 0.00015377947346861487,
      "loss": 2.2273,
      "step": 1604
    },
    {
      "epoch": 0.3537774838816333,
      "grad_norm": 0.7634771466255188,
      "learning_rate": 0.00015371800673319284,
      "loss": 2.1536,
      "step": 1605
    },
    {
      "epoch": 0.3539979059899708,
      "grad_norm": 0.7191499471664429,
      "learning_rate": 0.00015365651145704092,
      "loss": 2.1274,
      "step": 1606
    },
    {
      "epoch": 0.3542183280983083,
      "grad_norm": 0.7392830848693848,
      "learning_rate": 0.00015359498767283204,
      "loss": 2.3086,
      "step": 1607
    },
    {
      "epoch": 0.3544387502066457,
      "grad_norm": 0.751330554485321,
      "learning_rate": 0.00015353343541325406,
      "loss": 2.173,
      "step": 1608
    },
    {
      "epoch": 0.3546591723149832,
      "grad_norm": 0.7705582976341248,
      "learning_rate": 0.0001534718547110102,
      "loss": 2.2288,
      "step": 1609
    },
    {
      "epoch": 0.3548795944233207,
      "grad_norm": 0.7641940712928772,
      "learning_rate": 0.00015341024559881864,
      "loss": 2.1606,
      "step": 1610
    },
    {
      "epoch": 0.3551000165316581,
      "grad_norm": 0.7602839469909668,
      "learning_rate": 0.0001533486081094127,
      "loss": 2.2509,
      "step": 1611
    },
    {
      "epoch": 0.3553204386399956,
      "grad_norm": 0.7365640997886658,
      "learning_rate": 0.00015328694227554083,
      "loss": 2.2353,
      "step": 1612
    },
    {
      "epoch": 0.3555408607483331,
      "grad_norm": 0.7408837676048279,
      "learning_rate": 0.00015322524812996644,
      "loss": 2.0667,
      "step": 1613
    },
    {
      "epoch": 0.3557612828566705,
      "grad_norm": 0.7007336616516113,
      "learning_rate": 0.0001531635257054681,
      "loss": 2.1292,
      "step": 1614
    },
    {
      "epoch": 0.355981704965008,
      "grad_norm": 0.7536913156509399,
      "learning_rate": 0.00015310177503483925,
      "loss": 2.1636,
      "step": 1615
    },
    {
      "epoch": 0.3562021270733455,
      "grad_norm": 0.8112906217575073,
      "learning_rate": 0.0001530399961508885,
      "loss": 2.0687,
      "step": 1616
    },
    {
      "epoch": 0.3564225491816829,
      "grad_norm": 0.7317469716072083,
      "learning_rate": 0.00015297818908643936,
      "loss": 2.1995,
      "step": 1617
    },
    {
      "epoch": 0.35664297129002037,
      "grad_norm": 0.7309946417808533,
      "learning_rate": 0.0001529163538743303,
      "loss": 2.0338,
      "step": 1618
    },
    {
      "epoch": 0.3568633933983579,
      "grad_norm": 0.7615324258804321,
      "learning_rate": 0.00015285449054741482,
      "loss": 2.1804,
      "step": 1619
    },
    {
      "epoch": 0.3570838155066953,
      "grad_norm": 0.7380123734474182,
      "learning_rate": 0.00015279259913856128,
      "loss": 2.086,
      "step": 1620
    },
    {
      "epoch": 0.35730423761503277,
      "grad_norm": 0.7511727213859558,
      "learning_rate": 0.000152730679680653,
      "loss": 2.2824,
      "step": 1621
    },
    {
      "epoch": 0.3575246597233703,
      "grad_norm": 0.7294765114784241,
      "learning_rate": 0.0001526687322065882,
      "loss": 2.1983,
      "step": 1622
    },
    {
      "epoch": 0.3577450818317077,
      "grad_norm": 0.7465571165084839,
      "learning_rate": 0.00015260675674927998,
      "loss": 2.142,
      "step": 1623
    },
    {
      "epoch": 0.35796550394004517,
      "grad_norm": 0.7645461559295654,
      "learning_rate": 0.00015254475334165626,
      "loss": 2.193,
      "step": 1624
    },
    {
      "epoch": 0.3581859260483827,
      "grad_norm": 0.8520223498344421,
      "learning_rate": 0.00015248272201665998,
      "loss": 2.1525,
      "step": 1625
    },
    {
      "epoch": 0.3584063481567201,
      "grad_norm": 0.7243084907531738,
      "learning_rate": 0.00015242066280724868,
      "loss": 2.157,
      "step": 1626
    },
    {
      "epoch": 0.35862677026505757,
      "grad_norm": 0.7566795349121094,
      "learning_rate": 0.00015235857574639482,
      "loss": 2.2116,
      "step": 1627
    },
    {
      "epoch": 0.3588471923733951,
      "grad_norm": 0.7924532890319824,
      "learning_rate": 0.00015229646086708574,
      "loss": 2.1602,
      "step": 1628
    },
    {
      "epoch": 0.3590676144817325,
      "grad_norm": 0.8169381022453308,
      "learning_rate": 0.00015223431820232344,
      "loss": 2.3864,
      "step": 1629
    },
    {
      "epoch": 0.35928803659006997,
      "grad_norm": 0.7269949316978455,
      "learning_rate": 0.00015217214778512472,
      "loss": 2.0556,
      "step": 1630
    },
    {
      "epoch": 0.3595084586984075,
      "grad_norm": 0.7651529312133789,
      "learning_rate": 0.00015210994964852116,
      "loss": 2.1706,
      "step": 1631
    },
    {
      "epoch": 0.3597288808067449,
      "grad_norm": 0.7160031199455261,
      "learning_rate": 0.000152047723825559,
      "loss": 2.2027,
      "step": 1632
    },
    {
      "epoch": 0.35994930291508237,
      "grad_norm": 0.764735221862793,
      "learning_rate": 0.00015198547034929924,
      "loss": 2.2337,
      "step": 1633
    },
    {
      "epoch": 0.3601697250234199,
      "grad_norm": 0.801038920879364,
      "learning_rate": 0.0001519231892528176,
      "loss": 2.2829,
      "step": 1634
    },
    {
      "epoch": 0.3603901471317573,
      "grad_norm": 0.8050826787948608,
      "learning_rate": 0.00015186088056920439,
      "loss": 2.1114,
      "step": 1635
    },
    {
      "epoch": 0.36061056924009477,
      "grad_norm": 0.7475934624671936,
      "learning_rate": 0.00015179854433156463,
      "loss": 2.1994,
      "step": 1636
    },
    {
      "epoch": 0.3608309913484323,
      "grad_norm": 0.7905291318893433,
      "learning_rate": 0.00015173618057301796,
      "loss": 2.3045,
      "step": 1637
    },
    {
      "epoch": 0.3610514134567697,
      "grad_norm": 0.785793125629425,
      "learning_rate": 0.0001516737893266987,
      "loss": 2.249,
      "step": 1638
    },
    {
      "epoch": 0.36127183556510717,
      "grad_norm": 0.7263822555541992,
      "learning_rate": 0.00015161137062575574,
      "loss": 2.2123,
      "step": 1639
    },
    {
      "epoch": 0.3614922576734447,
      "grad_norm": 0.807851254940033,
      "learning_rate": 0.00015154892450335245,
      "loss": 2.1959,
      "step": 1640
    },
    {
      "epoch": 0.3617126797817821,
      "grad_norm": 0.7545026540756226,
      "learning_rate": 0.00015148645099266695,
      "loss": 2.1968,
      "step": 1641
    },
    {
      "epoch": 0.36193310189011957,
      "grad_norm": 0.7531028985977173,
      "learning_rate": 0.0001514239501268918,
      "loss": 2.1183,
      "step": 1642
    },
    {
      "epoch": 0.36215352399845707,
      "grad_norm": 0.7845927476882935,
      "learning_rate": 0.00015136142193923412,
      "loss": 2.1316,
      "step": 1643
    },
    {
      "epoch": 0.3623739461067945,
      "grad_norm": 0.76968914270401,
      "learning_rate": 0.0001512988664629156,
      "loss": 2.144,
      "step": 1644
    },
    {
      "epoch": 0.36259436821513197,
      "grad_norm": 0.7606697678565979,
      "learning_rate": 0.00015123628373117222,
      "loss": 2.2539,
      "step": 1645
    },
    {
      "epoch": 0.36281479032346947,
      "grad_norm": 0.8052710890769958,
      "learning_rate": 0.0001511736737772548,
      "loss": 2.2336,
      "step": 1646
    },
    {
      "epoch": 0.3630352124318069,
      "grad_norm": 0.7553932070732117,
      "learning_rate": 0.00015111103663442826,
      "loss": 2.1611,
      "step": 1647
    },
    {
      "epoch": 0.36325563454014437,
      "grad_norm": 0.7814359068870544,
      "learning_rate": 0.00015104837233597223,
      "loss": 2.2121,
      "step": 1648
    },
    {
      "epoch": 0.3634760566484818,
      "grad_norm": 0.7969289422035217,
      "learning_rate": 0.00015098568091518057,
      "loss": 2.1023,
      "step": 1649
    },
    {
      "epoch": 0.3636964787568193,
      "grad_norm": 0.7863852381706238,
      "learning_rate": 0.00015092296240536173,
      "loss": 2.2173,
      "step": 1650
    },
    {
      "epoch": 0.36391690086515677,
      "grad_norm": 0.73615562915802,
      "learning_rate": 0.0001508602168398384,
      "loss": 2.2535,
      "step": 1651
    },
    {
      "epoch": 0.3641373229734942,
      "grad_norm": 0.7132788300514221,
      "learning_rate": 0.0001507974442519478,
      "loss": 2.3299,
      "step": 1652
    },
    {
      "epoch": 0.3643577450818317,
      "grad_norm": 0.7713133692741394,
      "learning_rate": 0.0001507346446750413,
      "loss": 2.1642,
      "step": 1653
    },
    {
      "epoch": 0.36457816719016917,
      "grad_norm": 0.8035727143287659,
      "learning_rate": 0.0001506718181424848,
      "loss": 2.1378,
      "step": 1654
    },
    {
      "epoch": 0.3647985892985066,
      "grad_norm": 0.7977191805839539,
      "learning_rate": 0.00015060896468765847,
      "loss": 2.1653,
      "step": 1655
    },
    {
      "epoch": 0.3650190114068441,
      "grad_norm": 0.7864581942558289,
      "learning_rate": 0.0001505460843439567,
      "loss": 2.2436,
      "step": 1656
    },
    {
      "epoch": 0.36523943351518157,
      "grad_norm": 0.8324298858642578,
      "learning_rate": 0.00015048317714478826,
      "loss": 2.2038,
      "step": 1657
    },
    {
      "epoch": 0.365459855623519,
      "grad_norm": 0.726242184638977,
      "learning_rate": 0.00015042024312357616,
      "loss": 2.1917,
      "step": 1658
    },
    {
      "epoch": 0.3656802777318565,
      "grad_norm": 0.8299428820610046,
      "learning_rate": 0.00015035728231375765,
      "loss": 2.1663,
      "step": 1659
    },
    {
      "epoch": 0.36590069984019397,
      "grad_norm": 0.7172455787658691,
      "learning_rate": 0.00015029429474878425,
      "loss": 2.0182,
      "step": 1660
    },
    {
      "epoch": 0.3661211219485314,
      "grad_norm": 0.754937469959259,
      "learning_rate": 0.00015023128046212162,
      "loss": 2.2727,
      "step": 1661
    },
    {
      "epoch": 0.3663415440568689,
      "grad_norm": 0.7750821113586426,
      "learning_rate": 0.00015016823948724972,
      "loss": 2.242,
      "step": 1662
    },
    {
      "epoch": 0.36656196616520637,
      "grad_norm": 0.7088861465454102,
      "learning_rate": 0.00015010517185766258,
      "loss": 2.1483,
      "step": 1663
    },
    {
      "epoch": 0.3667823882735438,
      "grad_norm": 0.8502269983291626,
      "learning_rate": 0.0001500420776068685,
      "loss": 2.2238,
      "step": 1664
    },
    {
      "epoch": 0.3670028103818813,
      "grad_norm": 0.7441015243530273,
      "learning_rate": 0.0001499789567683898,
      "loss": 2.1767,
      "step": 1665
    },
    {
      "epoch": 0.36722323249021877,
      "grad_norm": 0.7452082633972168,
      "learning_rate": 0.00014991580937576306,
      "loss": 2.0039,
      "step": 1666
    },
    {
      "epoch": 0.3674436545985562,
      "grad_norm": 0.7913992404937744,
      "learning_rate": 0.00014985263546253886,
      "loss": 2.2931,
      "step": 1667
    },
    {
      "epoch": 0.3676640767068937,
      "grad_norm": 0.8035774230957031,
      "learning_rate": 0.000149789435062282,
      "loss": 2.1302,
      "step": 1668
    },
    {
      "epoch": 0.36788449881523116,
      "grad_norm": 0.7503395676612854,
      "learning_rate": 0.00014972620820857116,
      "loss": 2.158,
      "step": 1669
    },
    {
      "epoch": 0.3681049209235686,
      "grad_norm": 0.7788349390029907,
      "learning_rate": 0.0001496629549349992,
      "loss": 2.0768,
      "step": 1670
    },
    {
      "epoch": 0.3683253430319061,
      "grad_norm": 0.7529222965240479,
      "learning_rate": 0.00014959967527517306,
      "loss": 2.1997,
      "step": 1671
    },
    {
      "epoch": 0.36854576514024356,
      "grad_norm": 0.7325438261032104,
      "learning_rate": 0.00014953636926271356,
      "loss": 2.1505,
      "step": 1672
    },
    {
      "epoch": 0.368766187248581,
      "grad_norm": 0.7590321898460388,
      "learning_rate": 0.00014947303693125573,
      "loss": 2.1624,
      "step": 1673
    },
    {
      "epoch": 0.3689866093569185,
      "grad_norm": 0.6775720119476318,
      "learning_rate": 0.0001494096783144483,
      "loss": 2.0263,
      "step": 1674
    },
    {
      "epoch": 0.36920703146525596,
      "grad_norm": 0.7550601959228516,
      "learning_rate": 0.0001493462934459542,
      "loss": 2.0623,
      "step": 1675
    },
    {
      "epoch": 0.3694274535735934,
      "grad_norm": 0.7302157282829285,
      "learning_rate": 0.00014928288235945022,
      "loss": 2.1774,
      "step": 1676
    },
    {
      "epoch": 0.3696478756819309,
      "grad_norm": 0.7762846946716309,
      "learning_rate": 0.0001492194450886271,
      "loss": 2.1502,
      "step": 1677
    },
    {
      "epoch": 0.36986829779026836,
      "grad_norm": 0.759445071220398,
      "learning_rate": 0.00014915598166718945,
      "loss": 2.1824,
      "step": 1678
    },
    {
      "epoch": 0.3700887198986058,
      "grad_norm": 0.7697039246559143,
      "learning_rate": 0.00014909249212885584,
      "loss": 2.1889,
      "step": 1679
    },
    {
      "epoch": 0.3703091420069433,
      "grad_norm": 0.765840470790863,
      "learning_rate": 0.00014902897650735861,
      "loss": 2.152,
      "step": 1680
    },
    {
      "epoch": 0.37052956411528076,
      "grad_norm": 0.7353255152702332,
      "learning_rate": 0.0001489654348364441,
      "loss": 2.1149,
      "step": 1681
    },
    {
      "epoch": 0.3707499862236182,
      "grad_norm": 0.7466758489608765,
      "learning_rate": 0.00014890186714987236,
      "loss": 2.2679,
      "step": 1682
    },
    {
      "epoch": 0.3709704083319557,
      "grad_norm": 0.7484522461891174,
      "learning_rate": 0.00014883827348141738,
      "loss": 2.1502,
      "step": 1683
    },
    {
      "epoch": 0.37119083044029316,
      "grad_norm": 0.6904578804969788,
      "learning_rate": 0.00014877465386486686,
      "loss": 2.2467,
      "step": 1684
    },
    {
      "epoch": 0.3714112525486306,
      "grad_norm": 0.7274008989334106,
      "learning_rate": 0.00014871100833402227,
      "loss": 2.2084,
      "step": 1685
    },
    {
      "epoch": 0.3716316746569681,
      "grad_norm": 0.7931236028671265,
      "learning_rate": 0.00014864733692269896,
      "loss": 2.1215,
      "step": 1686
    },
    {
      "epoch": 0.37185209676530556,
      "grad_norm": 0.7659665942192078,
      "learning_rate": 0.00014858363966472598,
      "loss": 2.2145,
      "step": 1687
    },
    {
      "epoch": 0.372072518873643,
      "grad_norm": 0.6841546893119812,
      "learning_rate": 0.0001485199165939461,
      "loss": 2.0379,
      "step": 1688
    },
    {
      "epoch": 0.3722929409819805,
      "grad_norm": 0.7753267288208008,
      "learning_rate": 0.00014845616774421578,
      "loss": 2.0345,
      "step": 1689
    },
    {
      "epoch": 0.37251336309031796,
      "grad_norm": 0.722675085067749,
      "learning_rate": 0.0001483923931494052,
      "loss": 2.123,
      "step": 1690
    },
    {
      "epoch": 0.3727337851986554,
      "grad_norm": 0.7144778370857239,
      "learning_rate": 0.0001483285928433982,
      "loss": 2.131,
      "step": 1691
    },
    {
      "epoch": 0.3729542073069929,
      "grad_norm": 0.768337070941925,
      "learning_rate": 0.00014826476686009236,
      "loss": 2.3464,
      "step": 1692
    },
    {
      "epoch": 0.37317462941533036,
      "grad_norm": 0.7677947282791138,
      "learning_rate": 0.00014820091523339883,
      "loss": 2.0736,
      "step": 1693
    },
    {
      "epoch": 0.3733950515236678,
      "grad_norm": 0.7503165602684021,
      "learning_rate": 0.00014813703799724235,
      "loss": 2.2615,
      "step": 1694
    },
    {
      "epoch": 0.3736154736320053,
      "grad_norm": 0.7588987350463867,
      "learning_rate": 0.0001480731351855614,
      "loss": 2.1922,
      "step": 1695
    },
    {
      "epoch": 0.37383589574034276,
      "grad_norm": 0.6898938417434692,
      "learning_rate": 0.00014800920683230787,
      "loss": 2.1438,
      "step": 1696
    },
    {
      "epoch": 0.3740563178486802,
      "grad_norm": 0.7822191119194031,
      "learning_rate": 0.00014794525297144736,
      "loss": 2.1438,
      "step": 1697
    },
    {
      "epoch": 0.3742767399570177,
      "grad_norm": 0.7533077001571655,
      "learning_rate": 0.00014788127363695897,
      "loss": 2.2261,
      "step": 1698
    },
    {
      "epoch": 0.37449716206535516,
      "grad_norm": 0.7279696464538574,
      "learning_rate": 0.00014781726886283535,
      "loss": 2.1888,
      "step": 1699
    },
    {
      "epoch": 0.3747175841736926,
      "grad_norm": 0.724980890750885,
      "learning_rate": 0.00014775323868308266,
      "loss": 2.1923,
      "step": 1700
    },
    {
      "epoch": 0.3749380062820301,
      "grad_norm": 0.7906318306922913,
      "learning_rate": 0.00014768918313172056,
      "loss": 2.1661,
      "step": 1701
    },
    {
      "epoch": 0.37515842839036756,
      "grad_norm": 0.7190761566162109,
      "learning_rate": 0.00014762510224278214,
      "loss": 2.0792,
      "step": 1702
    },
    {
      "epoch": 0.375378850498705,
      "grad_norm": 0.8134549260139465,
      "learning_rate": 0.00014756099605031406,
      "loss": 2.155,
      "step": 1703
    },
    {
      "epoch": 0.3755992726070425,
      "grad_norm": 0.7394132614135742,
      "learning_rate": 0.00014749686458837631,
      "loss": 2.1327,
      "step": 1704
    },
    {
      "epoch": 0.37581969471537996,
      "grad_norm": 0.7816113829612732,
      "learning_rate": 0.0001474327078910424,
      "loss": 2.2181,
      "step": 1705
    },
    {
      "epoch": 0.3760401168237174,
      "grad_norm": 0.7499991059303284,
      "learning_rate": 0.00014736852599239917,
      "loss": 2.1284,
      "step": 1706
    },
    {
      "epoch": 0.3762605389320549,
      "grad_norm": 0.7417294979095459,
      "learning_rate": 0.0001473043189265469,
      "loss": 2.1182,
      "step": 1707
    },
    {
      "epoch": 0.37648096104039236,
      "grad_norm": 0.726040244102478,
      "learning_rate": 0.0001472400867275992,
      "loss": 2.0128,
      "step": 1708
    },
    {
      "epoch": 0.3767013831487298,
      "grad_norm": 0.8408939838409424,
      "learning_rate": 0.0001471758294296831,
      "loss": 2.2519,
      "step": 1709
    },
    {
      "epoch": 0.3769218052570673,
      "grad_norm": 0.6871330142021179,
      "learning_rate": 0.0001471115470669389,
      "loss": 2.1579,
      "step": 1710
    },
    {
      "epoch": 0.37714222736540476,
      "grad_norm": 0.7216675281524658,
      "learning_rate": 0.0001470472396735203,
      "loss": 2.189,
      "step": 1711
    },
    {
      "epoch": 0.3773626494737422,
      "grad_norm": 0.7643664479255676,
      "learning_rate": 0.00014698290728359413,
      "loss": 2.1577,
      "step": 1712
    },
    {
      "epoch": 0.3775830715820797,
      "grad_norm": 0.8339837193489075,
      "learning_rate": 0.00014691854993134068,
      "loss": 2.1998,
      "step": 1713
    },
    {
      "epoch": 0.37780349369041716,
      "grad_norm": 0.7167848348617554,
      "learning_rate": 0.0001468541676509534,
      "loss": 2.0671,
      "step": 1714
    },
    {
      "epoch": 0.3780239157987546,
      "grad_norm": 0.7136709094047546,
      "learning_rate": 0.00014678976047663905,
      "loss": 2.252,
      "step": 1715
    },
    {
      "epoch": 0.37824433790709205,
      "grad_norm": 0.8154271841049194,
      "learning_rate": 0.00014672532844261764,
      "loss": 2.1236,
      "step": 1716
    },
    {
      "epoch": 0.37846476001542956,
      "grad_norm": 0.7786619663238525,
      "learning_rate": 0.00014666087158312223,
      "loss": 2.0966,
      "step": 1717
    },
    {
      "epoch": 0.378685182123767,
      "grad_norm": 0.7883463501930237,
      "learning_rate": 0.0001465963899323992,
      "loss": 2.1251,
      "step": 1718
    },
    {
      "epoch": 0.37890560423210445,
      "grad_norm": 0.6976659893989563,
      "learning_rate": 0.00014653188352470808,
      "loss": 2.1473,
      "step": 1719
    },
    {
      "epoch": 0.37912602634044196,
      "grad_norm": 0.7714041471481323,
      "learning_rate": 0.0001464673523943216,
      "loss": 2.1598,
      "step": 1720
    },
    {
      "epoch": 0.3793464484487794,
      "grad_norm": 0.7802526354789734,
      "learning_rate": 0.0001464027965755255,
      "loss": 2.1159,
      "step": 1721
    },
    {
      "epoch": 0.37956687055711685,
      "grad_norm": 0.7290459275245667,
      "learning_rate": 0.00014633821610261874,
      "loss": 2.2289,
      "step": 1722
    },
    {
      "epoch": 0.37978729266545436,
      "grad_norm": 0.7309211492538452,
      "learning_rate": 0.00014627361100991336,
      "loss": 2.0677,
      "step": 1723
    },
    {
      "epoch": 0.3800077147737918,
      "grad_norm": 0.7955397367477417,
      "learning_rate": 0.00014620898133173445,
      "loss": 2.3169,
      "step": 1724
    },
    {
      "epoch": 0.38022813688212925,
      "grad_norm": 0.8237335681915283,
      "learning_rate": 0.0001461443271024202,
      "loss": 2.3139,
      "step": 1725
    },
    {
      "epoch": 0.38044855899046676,
      "grad_norm": 0.7050057649612427,
      "learning_rate": 0.00014607964835632183,
      "loss": 2.1647,
      "step": 1726
    },
    {
      "epoch": 0.3806689810988042,
      "grad_norm": 0.7740164399147034,
      "learning_rate": 0.00014601494512780355,
      "loss": 2.0492,
      "step": 1727
    },
    {
      "epoch": 0.38088940320714165,
      "grad_norm": 0.7608056664466858,
      "learning_rate": 0.0001459502174512426,
      "loss": 2.1503,
      "step": 1728
    },
    {
      "epoch": 0.38110982531547916,
      "grad_norm": 0.7651968598365784,
      "learning_rate": 0.0001458854653610293,
      "loss": 2.1691,
      "step": 1729
    },
    {
      "epoch": 0.3813302474238166,
      "grad_norm": 0.6962910890579224,
      "learning_rate": 0.00014582068889156682,
      "loss": 2.1646,
      "step": 1730
    },
    {
      "epoch": 0.38155066953215405,
      "grad_norm": 0.8001052141189575,
      "learning_rate": 0.0001457558880772713,
      "loss": 2.1661,
      "step": 1731
    },
    {
      "epoch": 0.38177109164049156,
      "grad_norm": 0.7888841032981873,
      "learning_rate": 0.00014569106295257188,
      "loss": 2.2228,
      "step": 1732
    },
    {
      "epoch": 0.381991513748829,
      "grad_norm": 0.7404809594154358,
      "learning_rate": 0.0001456262135519105,
      "loss": 2.2157,
      "step": 1733
    },
    {
      "epoch": 0.38221193585716645,
      "grad_norm": 0.7719851732254028,
      "learning_rate": 0.00014556133990974214,
      "loss": 2.0717,
      "step": 1734
    },
    {
      "epoch": 0.38243235796550396,
      "grad_norm": 0.7675809264183044,
      "learning_rate": 0.00014549644206053459,
      "loss": 2.0581,
      "step": 1735
    },
    {
      "epoch": 0.3826527800738414,
      "grad_norm": 0.7293387651443481,
      "learning_rate": 0.0001454315200387685,
      "loss": 2.1224,
      "step": 1736
    },
    {
      "epoch": 0.38287320218217885,
      "grad_norm": 0.7652633190155029,
      "learning_rate": 0.00014536657387893734,
      "loss": 2.2558,
      "step": 1737
    },
    {
      "epoch": 0.38309362429051635,
      "grad_norm": 0.7151110172271729,
      "learning_rate": 0.0001453016036155474,
      "loss": 2.1903,
      "step": 1738
    },
    {
      "epoch": 0.3833140463988538,
      "grad_norm": 0.7137725949287415,
      "learning_rate": 0.0001452366092831179,
      "loss": 2.1517,
      "step": 1739
    },
    {
      "epoch": 0.38353446850719125,
      "grad_norm": 0.7266117930412292,
      "learning_rate": 0.00014517159091618074,
      "loss": 2.1162,
      "step": 1740
    },
    {
      "epoch": 0.38375489061552875,
      "grad_norm": 0.7862868905067444,
      "learning_rate": 0.0001451065485492805,
      "loss": 2.14,
      "step": 1741
    },
    {
      "epoch": 0.3839753127238662,
      "grad_norm": 0.7520882487297058,
      "learning_rate": 0.00014504148221697476,
      "loss": 2.1483,
      "step": 1742
    },
    {
      "epoch": 0.38419573483220365,
      "grad_norm": 0.7251361012458801,
      "learning_rate": 0.00014497639195383362,
      "loss": 2.0687,
      "step": 1743
    },
    {
      "epoch": 0.38441615694054115,
      "grad_norm": 0.8137205243110657,
      "learning_rate": 0.0001449112777944399,
      "loss": 2.242,
      "step": 1744
    },
    {
      "epoch": 0.3846365790488786,
      "grad_norm": 0.7788509130477905,
      "learning_rate": 0.0001448461397733893,
      "loss": 2.1705,
      "step": 1745
    },
    {
      "epoch": 0.38485700115721605,
      "grad_norm": 0.7557998895645142,
      "learning_rate": 0.00014478097792529002,
      "loss": 2.2271,
      "step": 1746
    },
    {
      "epoch": 0.38507742326555355,
      "grad_norm": 0.7329299449920654,
      "learning_rate": 0.00014471579228476298,
      "loss": 2.2261,
      "step": 1747
    },
    {
      "epoch": 0.385297845373891,
      "grad_norm": 0.7327920794487,
      "learning_rate": 0.00014465058288644174,
      "loss": 2.3119,
      "step": 1748
    },
    {
      "epoch": 0.38551826748222845,
      "grad_norm": 0.7725517749786377,
      "learning_rate": 0.00014458534976497252,
      "loss": 2.1992,
      "step": 1749
    },
    {
      "epoch": 0.38573868959056595,
      "grad_norm": 0.7303383946418762,
      "learning_rate": 0.000144520092955014,
      "loss": 2.1963,
      "step": 1750
    },
    {
      "epoch": 0.3859591116989034,
      "grad_norm": 0.7677925229072571,
      "learning_rate": 0.00014445481249123767,
      "loss": 2.2414,
      "step": 1751
    },
    {
      "epoch": 0.38617953380724085,
      "grad_norm": 0.7369000315666199,
      "learning_rate": 0.00014438950840832738,
      "loss": 2.1137,
      "step": 1752
    },
    {
      "epoch": 0.38639995591557835,
      "grad_norm": 2.1524970531463623,
      "learning_rate": 0.0001443241807409797,
      "loss": 1.9864,
      "step": 1753
    },
    {
      "epoch": 0.3866203780239158,
      "grad_norm": 0.7527486085891724,
      "learning_rate": 0.00014425882952390361,
      "loss": 2.2446,
      "step": 1754
    },
    {
      "epoch": 0.38684080013225325,
      "grad_norm": 0.791573166847229,
      "learning_rate": 0.00014419345479182064,
      "loss": 2.3064,
      "step": 1755
    },
    {
      "epoch": 0.38706122224059075,
      "grad_norm": 0.8285712599754333,
      "learning_rate": 0.0001441280565794648,
      "loss": 2.2121,
      "step": 1756
    },
    {
      "epoch": 0.3872816443489282,
      "grad_norm": 0.7692622542381287,
      "learning_rate": 0.00014406263492158257,
      "loss": 2.249,
      "step": 1757
    },
    {
      "epoch": 0.38750206645726565,
      "grad_norm": 0.7104260921478271,
      "learning_rate": 0.00014399718985293297,
      "loss": 2.1429,
      "step": 1758
    },
    {
      "epoch": 0.38772248856560315,
      "grad_norm": 0.739372193813324,
      "learning_rate": 0.0001439317214082874,
      "loss": 2.1482,
      "step": 1759
    },
    {
      "epoch": 0.3879429106739406,
      "grad_norm": 0.6737844944000244,
      "learning_rate": 0.0001438662296224296,
      "loss": 2.1246,
      "step": 1760
    },
    {
      "epoch": 0.38816333278227805,
      "grad_norm": 0.7288551330566406,
      "learning_rate": 0.00014380071453015585,
      "loss": 2.1937,
      "step": 1761
    },
    {
      "epoch": 0.38838375489061555,
      "grad_norm": 0.7136619687080383,
      "learning_rate": 0.0001437351761662747,
      "loss": 2.0453,
      "step": 1762
    },
    {
      "epoch": 0.388604176998953,
      "grad_norm": 0.7284583449363708,
      "learning_rate": 0.00014366961456560714,
      "loss": 2.205,
      "step": 1763
    },
    {
      "epoch": 0.38882459910729045,
      "grad_norm": 0.7327271699905396,
      "learning_rate": 0.0001436040297629865,
      "loss": 2.0305,
      "step": 1764
    },
    {
      "epoch": 0.38904502121562795,
      "grad_norm": 0.715551495552063,
      "learning_rate": 0.00014353842179325835,
      "loss": 2.118,
      "step": 1765
    },
    {
      "epoch": 0.3892654433239654,
      "grad_norm": 0.767373263835907,
      "learning_rate": 0.00014347279069128072,
      "loss": 2.1838,
      "step": 1766
    },
    {
      "epoch": 0.38948586543230285,
      "grad_norm": 0.7620293498039246,
      "learning_rate": 0.0001434071364919238,
      "loss": 2.0999,
      "step": 1767
    },
    {
      "epoch": 0.38970628754064035,
      "grad_norm": 0.7197187542915344,
      "learning_rate": 0.0001433414592300701,
      "loss": 2.1315,
      "step": 1768
    },
    {
      "epoch": 0.3899267096489778,
      "grad_norm": 0.7252985835075378,
      "learning_rate": 0.00014327575894061435,
      "loss": 2.088,
      "step": 1769
    },
    {
      "epoch": 0.39014713175731525,
      "grad_norm": 0.795889139175415,
      "learning_rate": 0.00014321003565846366,
      "loss": 2.2206,
      "step": 1770
    },
    {
      "epoch": 0.39036755386565275,
      "grad_norm": 0.7353513836860657,
      "learning_rate": 0.0001431442894185371,
      "loss": 2.1662,
      "step": 1771
    },
    {
      "epoch": 0.3905879759739902,
      "grad_norm": 0.7359495759010315,
      "learning_rate": 0.00014307852025576614,
      "loss": 2.044,
      "step": 1772
    },
    {
      "epoch": 0.39080839808232765,
      "grad_norm": 0.7053994536399841,
      "learning_rate": 0.00014301272820509437,
      "loss": 2.0764,
      "step": 1773
    },
    {
      "epoch": 0.39102882019066515,
      "grad_norm": 0.7268823981285095,
      "learning_rate": 0.00014294691330147755,
      "loss": 2.1704,
      "step": 1774
    },
    {
      "epoch": 0.3912492422990026,
      "grad_norm": 0.7452391386032104,
      "learning_rate": 0.00014288107557988355,
      "loss": 2.1965,
      "step": 1775
    },
    {
      "epoch": 0.39146966440734005,
      "grad_norm": 0.7420617938041687,
      "learning_rate": 0.0001428152150752924,
      "loss": 2.2879,
      "step": 1776
    },
    {
      "epoch": 0.39169008651567755,
      "grad_norm": 0.771698534488678,
      "learning_rate": 0.00014274933182269618,
      "loss": 2.1655,
      "step": 1777
    },
    {
      "epoch": 0.391910508624015,
      "grad_norm": 0.7692624926567078,
      "learning_rate": 0.00014268342585709913,
      "loss": 2.1472,
      "step": 1778
    },
    {
      "epoch": 0.39213093073235245,
      "grad_norm": 0.7454381585121155,
      "learning_rate": 0.00014261749721351753,
      "loss": 2.1571,
      "step": 1779
    },
    {
      "epoch": 0.39235135284068995,
      "grad_norm": 0.7070075869560242,
      "learning_rate": 0.0001425515459269797,
      "loss": 2.1256,
      "step": 1780
    },
    {
      "epoch": 0.3925717749490274,
      "grad_norm": 0.7762736678123474,
      "learning_rate": 0.00014248557203252598,
      "loss": 2.1969,
      "step": 1781
    },
    {
      "epoch": 0.39279219705736484,
      "grad_norm": 0.7099081873893738,
      "learning_rate": 0.0001424195755652087,
      "loss": 2.2676,
      "step": 1782
    },
    {
      "epoch": 0.3930126191657023,
      "grad_norm": 0.7106368541717529,
      "learning_rate": 0.0001423535565600923,
      "loss": 2.305,
      "step": 1783
    },
    {
      "epoch": 0.3932330412740398,
      "grad_norm": 0.8141341209411621,
      "learning_rate": 0.00014228751505225302,
      "loss": 2.2198,
      "step": 1784
    },
    {
      "epoch": 0.39345346338237724,
      "grad_norm": 0.7553448677062988,
      "learning_rate": 0.0001422214510767792,
      "loss": 2.1646,
      "step": 1785
    },
    {
      "epoch": 0.3936738854907147,
      "grad_norm": 0.6921107172966003,
      "learning_rate": 0.00014215536466877107,
      "loss": 2.13,
      "step": 1786
    },
    {
      "epoch": 0.3938943075990522,
      "grad_norm": 0.716844916343689,
      "learning_rate": 0.00014208925586334074,
      "loss": 2.2426,
      "step": 1787
    },
    {
      "epoch": 0.39411472970738964,
      "grad_norm": 0.7484984397888184,
      "learning_rate": 0.00014202312469561228,
      "loss": 2.1571,
      "step": 1788
    },
    {
      "epoch": 0.3943351518157271,
      "grad_norm": 0.7787557244300842,
      "learning_rate": 0.0001419569712007216,
      "loss": 2.1282,
      "step": 1789
    },
    {
      "epoch": 0.3945555739240646,
      "grad_norm": 0.7344565391540527,
      "learning_rate": 0.00014189079541381645,
      "loss": 2.1229,
      "step": 1790
    },
    {
      "epoch": 0.39477599603240204,
      "grad_norm": 0.7014215588569641,
      "learning_rate": 0.00014182459737005653,
      "loss": 2.1522,
      "step": 1791
    },
    {
      "epoch": 0.3949964181407395,
      "grad_norm": 0.7292867302894592,
      "learning_rate": 0.00014175837710461326,
      "loss": 2.113,
      "step": 1792
    },
    {
      "epoch": 0.395216840249077,
      "grad_norm": 0.6991387605667114,
      "learning_rate": 0.0001416921346526699,
      "loss": 2.2298,
      "step": 1793
    },
    {
      "epoch": 0.39543726235741444,
      "grad_norm": 0.7190842032432556,
      "learning_rate": 0.00014162587004942152,
      "loss": 2.0985,
      "step": 1794
    },
    {
      "epoch": 0.3956576844657519,
      "grad_norm": 0.7260407209396362,
      "learning_rate": 0.0001415595833300749,
      "loss": 2.0449,
      "step": 1795
    },
    {
      "epoch": 0.3958781065740894,
      "grad_norm": 0.7450221180915833,
      "learning_rate": 0.0001414932745298487,
      "loss": 2.1097,
      "step": 1796
    },
    {
      "epoch": 0.39609852868242684,
      "grad_norm": 0.7041311264038086,
      "learning_rate": 0.00014142694368397308,
      "loss": 2.1708,
      "step": 1797
    },
    {
      "epoch": 0.3963189507907643,
      "grad_norm": 0.7584436535835266,
      "learning_rate": 0.00014136059082769017,
      "loss": 2.1594,
      "step": 1798
    },
    {
      "epoch": 0.3965393728991018,
      "grad_norm": 0.8013955354690552,
      "learning_rate": 0.00014129421599625363,
      "loss": 2.0942,
      "step": 1799
    },
    {
      "epoch": 0.39675979500743924,
      "grad_norm": 0.7169339656829834,
      "learning_rate": 0.00014122781922492884,
      "loss": 2.0049,
      "step": 1800
    },
    {
      "epoch": 0.39675979500743924,
      "eval_bpb": 1.1214255902721026,
      "eval_loss": 2.1613826751708984,
      "eval_runtime": 184.2698,
      "eval_samples_per_second": 62.191,
      "eval_steps_per_second": 7.777,
      "step": 1800
    },
    {
      "epoch": 0.3969802171157767,
      "grad_norm": 0.6889115571975708,
      "learning_rate": 0.00014116140054899287,
      "loss": 2.1439,
      "step": 1801
    },
    {
      "epoch": 0.3972006392241142,
      "grad_norm": 0.7608737945556641,
      "learning_rate": 0.00014109496000373439,
      "loss": 2.0638,
      "step": 1802
    },
    {
      "epoch": 0.39742106133245164,
      "grad_norm": 0.7601191401481628,
      "learning_rate": 0.00014102849762445368,
      "loss": 2.0302,
      "step": 1803
    },
    {
      "epoch": 0.3976414834407891,
      "grad_norm": 0.7324056029319763,
      "learning_rate": 0.00014096201344646265,
      "loss": 2.1515,
      "step": 1804
    },
    {
      "epoch": 0.3978619055491266,
      "grad_norm": 0.7224534153938293,
      "learning_rate": 0.00014089550750508474,
      "loss": 2.1014,
      "step": 1805
    },
    {
      "epoch": 0.39808232765746404,
      "grad_norm": 0.7573065757751465,
      "learning_rate": 0.00014082897983565506,
      "loss": 2.1226,
      "step": 1806
    },
    {
      "epoch": 0.3983027497658015,
      "grad_norm": 0.762500524520874,
      "learning_rate": 0.0001407624304735202,
      "loss": 2.1671,
      "step": 1807
    },
    {
      "epoch": 0.398523171874139,
      "grad_norm": 0.7368777394294739,
      "learning_rate": 0.00014069585945403822,
      "loss": 2.1348,
      "step": 1808
    },
    {
      "epoch": 0.39874359398247644,
      "grad_norm": 0.6987581253051758,
      "learning_rate": 0.00014062926681257878,
      "loss": 2.1057,
      "step": 1809
    },
    {
      "epoch": 0.3989640160908139,
      "grad_norm": 0.8186967968940735,
      "learning_rate": 0.00014056265258452296,
      "loss": 2.1928,
      "step": 1810
    },
    {
      "epoch": 0.3991844381991514,
      "grad_norm": 0.7304983139038086,
      "learning_rate": 0.0001404960168052633,
      "loss": 2.194,
      "step": 1811
    },
    {
      "epoch": 0.39940486030748884,
      "grad_norm": 0.7157537341117859,
      "learning_rate": 0.00014042935951020396,
      "loss": 2.1432,
      "step": 1812
    },
    {
      "epoch": 0.3996252824158263,
      "grad_norm": 0.7384117245674133,
      "learning_rate": 0.00014036268073476028,
      "loss": 1.9702,
      "step": 1813
    },
    {
      "epoch": 0.3998457045241638,
      "grad_norm": 0.827658474445343,
      "learning_rate": 0.00014029598051435918,
      "loss": 2.1273,
      "step": 1814
    },
    {
      "epoch": 0.40006612663250124,
      "grad_norm": 0.7454954385757446,
      "learning_rate": 0.00014022925888443889,
      "loss": 2.0809,
      "step": 1815
    },
    {
      "epoch": 0.4002865487408387,
      "grad_norm": 0.7312313914299011,
      "learning_rate": 0.00014016251588044905,
      "loss": 2.2064,
      "step": 1816
    },
    {
      "epoch": 0.4005069708491762,
      "grad_norm": 0.7326645255088806,
      "learning_rate": 0.00014009575153785072,
      "loss": 2.1134,
      "step": 1817
    },
    {
      "epoch": 0.40072739295751364,
      "grad_norm": 0.7677224278450012,
      "learning_rate": 0.00014002896589211618,
      "loss": 2.1883,
      "step": 1818
    },
    {
      "epoch": 0.4009478150658511,
      "grad_norm": 0.8730478882789612,
      "learning_rate": 0.0001399621589787291,
      "loss": 2.1336,
      "step": 1819
    },
    {
      "epoch": 0.4011682371741886,
      "grad_norm": 0.7417235970497131,
      "learning_rate": 0.0001398953308331844,
      "loss": 2.1551,
      "step": 1820
    },
    {
      "epoch": 0.40138865928252604,
      "grad_norm": 0.7469131350517273,
      "learning_rate": 0.00013982848149098837,
      "loss": 2.0867,
      "step": 1821
    },
    {
      "epoch": 0.4016090813908635,
      "grad_norm": 0.7583255171775818,
      "learning_rate": 0.00013976161098765846,
      "loss": 2.1259,
      "step": 1822
    },
    {
      "epoch": 0.401829503499201,
      "grad_norm": 0.8445776104927063,
      "learning_rate": 0.00013969471935872344,
      "loss": 2.1483,
      "step": 1823
    },
    {
      "epoch": 0.40204992560753844,
      "grad_norm": 0.7677097916603088,
      "learning_rate": 0.00013962780663972325,
      "loss": 2.1337,
      "step": 1824
    },
    {
      "epoch": 0.4022703477158759,
      "grad_norm": 0.7140787243843079,
      "learning_rate": 0.00013956087286620905,
      "loss": 2.1716,
      "step": 1825
    },
    {
      "epoch": 0.4024907698242134,
      "grad_norm": 0.7463569641113281,
      "learning_rate": 0.0001394939180737432,
      "loss": 2.3078,
      "step": 1826
    },
    {
      "epoch": 0.40271119193255084,
      "grad_norm": 0.7097780108451843,
      "learning_rate": 0.00013942694229789922,
      "loss": 2.1132,
      "step": 1827
    },
    {
      "epoch": 0.4029316140408883,
      "grad_norm": 0.714766800403595,
      "learning_rate": 0.0001393599455742618,
      "loss": 2.197,
      "step": 1828
    },
    {
      "epoch": 0.4031520361492258,
      "grad_norm": 0.6811312437057495,
      "learning_rate": 0.0001392929279384267,
      "loss": 2.1243,
      "step": 1829
    },
    {
      "epoch": 0.40337245825756324,
      "grad_norm": 0.7522096633911133,
      "learning_rate": 0.0001392258894260009,
      "loss": 2.1284,
      "step": 1830
    },
    {
      "epoch": 0.4035928803659007,
      "grad_norm": 0.69526207447052,
      "learning_rate": 0.0001391588300726023,
      "loss": 2.1352,
      "step": 1831
    },
    {
      "epoch": 0.4038133024742382,
      "grad_norm": 0.7274652719497681,
      "learning_rate": 0.00013909174991386008,
      "loss": 2.2601,
      "step": 1832
    },
    {
      "epoch": 0.40403372458257564,
      "grad_norm": 0.7499006986618042,
      "learning_rate": 0.0001390246489854143,
      "loss": 2.1839,
      "step": 1833
    },
    {
      "epoch": 0.4042541466909131,
      "grad_norm": 0.6934055685997009,
      "learning_rate": 0.00013895752732291617,
      "loss": 2.0999,
      "step": 1834
    },
    {
      "epoch": 0.4044745687992506,
      "grad_norm": 0.7465680837631226,
      "learning_rate": 0.00013889038496202782,
      "loss": 2.0586,
      "step": 1835
    },
    {
      "epoch": 0.40469499090758804,
      "grad_norm": 0.7383532524108887,
      "learning_rate": 0.00013882322193842244,
      "loss": 2.1739,
      "step": 1836
    },
    {
      "epoch": 0.4049154130159255,
      "grad_norm": 0.7292195558547974,
      "learning_rate": 0.0001387560382877842,
      "loss": 2.126,
      "step": 1837
    },
    {
      "epoch": 0.405135835124263,
      "grad_norm": 0.7494935393333435,
      "learning_rate": 0.00013868883404580823,
      "loss": 2.1221,
      "step": 1838
    },
    {
      "epoch": 0.40535625723260044,
      "grad_norm": 0.6731420159339905,
      "learning_rate": 0.00013862160924820057,
      "loss": 2.0935,
      "step": 1839
    },
    {
      "epoch": 0.4055766793409379,
      "grad_norm": 0.7691079378128052,
      "learning_rate": 0.00013855436393067822,
      "loss": 2.097,
      "step": 1840
    },
    {
      "epoch": 0.4057971014492754,
      "grad_norm": 0.8253318667411804,
      "learning_rate": 0.000138487098128969,
      "loss": 2.0655,
      "step": 1841
    },
    {
      "epoch": 0.40601752355761284,
      "grad_norm": 0.7114134430885315,
      "learning_rate": 0.00013841981187881172,
      "loss": 2.2654,
      "step": 1842
    },
    {
      "epoch": 0.4062379456659503,
      "grad_norm": 0.7421424388885498,
      "learning_rate": 0.00013835250521595596,
      "loss": 2.1716,
      "step": 1843
    },
    {
      "epoch": 0.4064583677742878,
      "grad_norm": 0.7248939871788025,
      "learning_rate": 0.00013828517817616226,
      "loss": 2.2398,
      "step": 1844
    },
    {
      "epoch": 0.40667878988262524,
      "grad_norm": 0.7202637791633606,
      "learning_rate": 0.00013821783079520186,
      "loss": 2.2341,
      "step": 1845
    },
    {
      "epoch": 0.4068992119909627,
      "grad_norm": 0.7517955899238586,
      "learning_rate": 0.0001381504631088569,
      "loss": 2.1891,
      "step": 1846
    },
    {
      "epoch": 0.4071196340993002,
      "grad_norm": 0.7334721684455872,
      "learning_rate": 0.0001380830751529203,
      "loss": 2.143,
      "step": 1847
    },
    {
      "epoch": 0.40734005620763764,
      "grad_norm": 0.7389910221099854,
      "learning_rate": 0.00013801566696319562,
      "loss": 2.2282,
      "step": 1848
    },
    {
      "epoch": 0.4075604783159751,
      "grad_norm": 0.7700773477554321,
      "learning_rate": 0.0001379482385754974,
      "loss": 2.1978,
      "step": 1849
    },
    {
      "epoch": 0.40778090042431253,
      "grad_norm": 0.7703351378440857,
      "learning_rate": 0.00013788079002565072,
      "loss": 2.1324,
      "step": 1850
    },
    {
      "epoch": 0.40800132253265003,
      "grad_norm": 0.72658371925354,
      "learning_rate": 0.00013781332134949143,
      "loss": 2.25,
      "step": 1851
    },
    {
      "epoch": 0.4082217446409875,
      "grad_norm": 0.7649495601654053,
      "learning_rate": 0.00013774583258286608,
      "loss": 2.1018,
      "step": 1852
    },
    {
      "epoch": 0.40844216674932493,
      "grad_norm": 0.7844642400741577,
      "learning_rate": 0.00013767832376163192,
      "loss": 2.2594,
      "step": 1853
    },
    {
      "epoch": 0.40866258885766243,
      "grad_norm": 0.7527579069137573,
      "learning_rate": 0.00013761079492165683,
      "loss": 2.2497,
      "step": 1854
    },
    {
      "epoch": 0.4088830109659999,
      "grad_norm": 0.7703769207000732,
      "learning_rate": 0.0001375432460988193,
      "loss": 2.1512,
      "step": 1855
    },
    {
      "epoch": 0.40910343307433733,
      "grad_norm": 0.8460434079170227,
      "learning_rate": 0.00013747567732900852,
      "loss": 2.197,
      "step": 1856
    },
    {
      "epoch": 0.40932385518267483,
      "grad_norm": 0.7334707379341125,
      "learning_rate": 0.00013740808864812416,
      "loss": 2.2492,
      "step": 1857
    },
    {
      "epoch": 0.4095442772910123,
      "grad_norm": 0.7772679924964905,
      "learning_rate": 0.0001373404800920765,
      "loss": 2.1247,
      "step": 1858
    },
    {
      "epoch": 0.40976469939934973,
      "grad_norm": 0.7642465233802795,
      "learning_rate": 0.00013727285169678655,
      "loss": 2.1213,
      "step": 1859
    },
    {
      "epoch": 0.40998512150768723,
      "grad_norm": 0.7491151690483093,
      "learning_rate": 0.00013720520349818559,
      "loss": 2.051,
      "step": 1860
    },
    {
      "epoch": 0.4102055436160247,
      "grad_norm": 0.7408165335655212,
      "learning_rate": 0.0001371375355322156,
      "loss": 2.161,
      "step": 1861
    },
    {
      "epoch": 0.41042596572436213,
      "grad_norm": 0.7814253568649292,
      "learning_rate": 0.000137069847834829,
      "loss": 2.259,
      "step": 1862
    },
    {
      "epoch": 0.41064638783269963,
      "grad_norm": 0.7223923206329346,
      "learning_rate": 0.00013700214044198874,
      "loss": 2.2027,
      "step": 1863
    },
    {
      "epoch": 0.4108668099410371,
      "grad_norm": 0.776635468006134,
      "learning_rate": 0.00013693441338966816,
      "loss": 2.254,
      "step": 1864
    },
    {
      "epoch": 0.41108723204937453,
      "grad_norm": 0.8195370435714722,
      "learning_rate": 0.00013686666671385112,
      "loss": 2.1084,
      "step": 1865
    },
    {
      "epoch": 0.41130765415771203,
      "grad_norm": 0.7610825896263123,
      "learning_rate": 0.00013679890045053186,
      "loss": 2.093,
      "step": 1866
    },
    {
      "epoch": 0.4115280762660495,
      "grad_norm": 0.6869077086448669,
      "learning_rate": 0.00013673111463571505,
      "loss": 2.1419,
      "step": 1867
    },
    {
      "epoch": 0.41174849837438693,
      "grad_norm": 0.7754110097885132,
      "learning_rate": 0.0001366633093054157,
      "loss": 2.1999,
      "step": 1868
    },
    {
      "epoch": 0.41196892048272443,
      "grad_norm": 0.7701969146728516,
      "learning_rate": 0.0001365954844956593,
      "loss": 2.1404,
      "step": 1869
    },
    {
      "epoch": 0.4121893425910619,
      "grad_norm": 0.8170925378799438,
      "learning_rate": 0.00013652764024248154,
      "loss": 2.1146,
      "step": 1870
    },
    {
      "epoch": 0.41240976469939933,
      "grad_norm": 0.7050665616989136,
      "learning_rate": 0.00013645977658192858,
      "loss": 2.1038,
      "step": 1871
    },
    {
      "epoch": 0.41263018680773683,
      "grad_norm": 0.7185353636741638,
      "learning_rate": 0.00013639189355005678,
      "loss": 2.1547,
      "step": 1872
    },
    {
      "epoch": 0.4128506089160743,
      "grad_norm": 0.7926629185676575,
      "learning_rate": 0.0001363239911829329,
      "loss": 2.2064,
      "step": 1873
    },
    {
      "epoch": 0.41307103102441173,
      "grad_norm": 0.8253393173217773,
      "learning_rate": 0.00013625606951663387,
      "loss": 2.1316,
      "step": 1874
    },
    {
      "epoch": 0.41329145313274923,
      "grad_norm": 0.7314622402191162,
      "learning_rate": 0.00013618812858724687,
      "loss": 2.0209,
      "step": 1875
    },
    {
      "epoch": 0.4135118752410867,
      "grad_norm": 0.7326939702033997,
      "learning_rate": 0.00013612016843086948,
      "loss": 2.0738,
      "step": 1876
    },
    {
      "epoch": 0.4137322973494241,
      "grad_norm": 0.7850172519683838,
      "learning_rate": 0.00013605218908360932,
      "loss": 2.1062,
      "step": 1877
    },
    {
      "epoch": 0.41395271945776163,
      "grad_norm": 0.7641253471374512,
      "learning_rate": 0.0001359841905815842,
      "loss": 2.1028,
      "step": 1878
    },
    {
      "epoch": 0.4141731415660991,
      "grad_norm": 0.8041000366210938,
      "learning_rate": 0.00013591617296092228,
      "loss": 2.0662,
      "step": 1879
    },
    {
      "epoch": 0.4143935636744365,
      "grad_norm": 0.7376822829246521,
      "learning_rate": 0.00013584813625776168,
      "loss": 2.1584,
      "step": 1880
    },
    {
      "epoch": 0.41461398578277403,
      "grad_norm": 0.7875715494155884,
      "learning_rate": 0.0001357800805082508,
      "loss": 2.1926,
      "step": 1881
    },
    {
      "epoch": 0.4148344078911115,
      "grad_norm": 0.7706781625747681,
      "learning_rate": 0.00013571200574854809,
      "loss": 2.1863,
      "step": 1882
    },
    {
      "epoch": 0.4150548299994489,
      "grad_norm": 0.6930294036865234,
      "learning_rate": 0.00013564391201482213,
      "loss": 2.2033,
      "step": 1883
    },
    {
      "epoch": 0.41527525210778643,
      "grad_norm": 0.7882454991340637,
      "learning_rate": 0.00013557579934325156,
      "loss": 2.2028,
      "step": 1884
    },
    {
      "epoch": 0.4154956742161239,
      "grad_norm": 0.7351268529891968,
      "learning_rate": 0.00013550766777002506,
      "loss": 2.1536,
      "step": 1885
    },
    {
      "epoch": 0.4157160963244613,
      "grad_norm": 0.8328057527542114,
      "learning_rate": 0.00013543951733134142,
      "loss": 2.1078,
      "step": 1886
    },
    {
      "epoch": 0.41593651843279883,
      "grad_norm": 0.7742167115211487,
      "learning_rate": 0.00013537134806340944,
      "loss": 2.1239,
      "step": 1887
    },
    {
      "epoch": 0.4161569405411363,
      "grad_norm": 0.7191008925437927,
      "learning_rate": 0.00013530316000244782,
      "loss": 1.9939,
      "step": 1888
    },
    {
      "epoch": 0.4163773626494737,
      "grad_norm": 0.7160856127738953,
      "learning_rate": 0.00013523495318468537,
      "loss": 2.0884,
      "step": 1889
    },
    {
      "epoch": 0.41659778475781123,
      "grad_norm": 0.80043625831604,
      "learning_rate": 0.00013516672764636085,
      "loss": 2.118,
      "step": 1890
    },
    {
      "epoch": 0.4168182068661487,
      "grad_norm": 0.7432541847229004,
      "learning_rate": 0.00013509848342372287,
      "loss": 2.166,
      "step": 1891
    },
    {
      "epoch": 0.4170386289744861,
      "grad_norm": 0.7285271883010864,
      "learning_rate": 0.0001350302205530301,
      "loss": 2.1439,
      "step": 1892
    },
    {
      "epoch": 0.41725905108282363,
      "grad_norm": 0.8171418905258179,
      "learning_rate": 0.00013496193907055104,
      "loss": 2.112,
      "step": 1893
    },
    {
      "epoch": 0.4174794731911611,
      "grad_norm": 0.7740259170532227,
      "learning_rate": 0.00013489363901256405,
      "loss": 2.3244,
      "step": 1894
    },
    {
      "epoch": 0.4176998952994985,
      "grad_norm": 0.8001946806907654,
      "learning_rate": 0.0001348253204153574,
      "loss": 2.0571,
      "step": 1895
    },
    {
      "epoch": 0.41792031740783603,
      "grad_norm": 0.7954029440879822,
      "learning_rate": 0.00013475698331522924,
      "loss": 2.1691,
      "step": 1896
    },
    {
      "epoch": 0.4181407395161735,
      "grad_norm": 0.7826483845710754,
      "learning_rate": 0.00013468862774848752,
      "loss": 2.1871,
      "step": 1897
    },
    {
      "epoch": 0.4183611616245109,
      "grad_norm": 0.8088982105255127,
      "learning_rate": 0.00013462025375145,
      "loss": 2.2293,
      "step": 1898
    },
    {
      "epoch": 0.41858158373284843,
      "grad_norm": 0.8062360286712646,
      "learning_rate": 0.0001345518613604442,
      "loss": 2.1429,
      "step": 1899
    },
    {
      "epoch": 0.4188020058411859,
      "grad_norm": 0.7310142517089844,
      "learning_rate": 0.00013448345061180752,
      "loss": 2.0812,
      "step": 1900
    },
    {
      "epoch": 0.4190224279495233,
      "grad_norm": 0.7631806135177612,
      "learning_rate": 0.00013441502154188693,
      "loss": 2.1684,
      "step": 1901
    },
    {
      "epoch": 0.4192428500578608,
      "grad_norm": 0.7512317299842834,
      "learning_rate": 0.00013434657418703936,
      "loss": 2.0893,
      "step": 1902
    },
    {
      "epoch": 0.4194632721661983,
      "grad_norm": 0.6649432182312012,
      "learning_rate": 0.00013427810858363128,
      "loss": 2.1462,
      "step": 1903
    },
    {
      "epoch": 0.4196836942745357,
      "grad_norm": 0.768694281578064,
      "learning_rate": 0.00013420962476803894,
      "loss": 2.1717,
      "step": 1904
    },
    {
      "epoch": 0.4199041163828732,
      "grad_norm": 0.7335543632507324,
      "learning_rate": 0.0001341411227766482,
      "loss": 2.0946,
      "step": 1905
    },
    {
      "epoch": 0.4201245384912107,
      "grad_norm": 0.7287137508392334,
      "learning_rate": 0.00013407260264585463,
      "loss": 2.0911,
      "step": 1906
    },
    {
      "epoch": 0.4203449605995481,
      "grad_norm": 0.7602452635765076,
      "learning_rate": 0.00013400406441206345,
      "loss": 2.188,
      "step": 1907
    },
    {
      "epoch": 0.4205653827078856,
      "grad_norm": 0.7017931342124939,
      "learning_rate": 0.00013393550811168948,
      "loss": 2.1458,
      "step": 1908
    },
    {
      "epoch": 0.4207858048162231,
      "grad_norm": 0.7828822731971741,
      "learning_rate": 0.00013386693378115706,
      "loss": 1.9779,
      "step": 1909
    },
    {
      "epoch": 0.4210062269245605,
      "grad_norm": 0.8326809406280518,
      "learning_rate": 0.00013379834145690032,
      "loss": 2.1168,
      "step": 1910
    },
    {
      "epoch": 0.421226649032898,
      "grad_norm": 0.7755996584892273,
      "learning_rate": 0.00013372973117536265,
      "loss": 2.2207,
      "step": 1911
    },
    {
      "epoch": 0.4214470711412355,
      "grad_norm": 0.7590673565864563,
      "learning_rate": 0.00013366110297299724,
      "loss": 2.1379,
      "step": 1912
    },
    {
      "epoch": 0.4216674932495729,
      "grad_norm": 0.7288358807563782,
      "learning_rate": 0.0001335924568862667,
      "loss": 2.1333,
      "step": 1913
    },
    {
      "epoch": 0.4218879153579104,
      "grad_norm": 0.7217111587524414,
      "learning_rate": 0.00013352379295164315,
      "loss": 2.0779,
      "step": 1914
    },
    {
      "epoch": 0.4221083374662479,
      "grad_norm": 0.7622907757759094,
      "learning_rate": 0.00013345511120560813,
      "loss": 2.2662,
      "step": 1915
    },
    {
      "epoch": 0.4223287595745853,
      "grad_norm": 0.7750532031059265,
      "learning_rate": 0.00013338641168465277,
      "loss": 2.1354,
      "step": 1916
    },
    {
      "epoch": 0.42254918168292277,
      "grad_norm": 0.8343643546104431,
      "learning_rate": 0.00013331769442527755,
      "loss": 2.1004,
      "step": 1917
    },
    {
      "epoch": 0.4227696037912603,
      "grad_norm": 0.7843561768531799,
      "learning_rate": 0.0001332489594639924,
      "loss": 2.0599,
      "step": 1918
    },
    {
      "epoch": 0.4229900258995977,
      "grad_norm": 0.8090461492538452,
      "learning_rate": 0.0001331802068373167,
      "loss": 2.2074,
      "step": 1919
    },
    {
      "epoch": 0.42321044800793517,
      "grad_norm": 0.7820500135421753,
      "learning_rate": 0.0001331114365817791,
      "loss": 2.1469,
      "step": 1920
    },
    {
      "epoch": 0.4234308701162727,
      "grad_norm": 0.8311426639556885,
      "learning_rate": 0.00013304264873391773,
      "loss": 2.2832,
      "step": 1921
    },
    {
      "epoch": 0.4236512922246101,
      "grad_norm": 0.734589695930481,
      "learning_rate": 0.00013297384333028,
      "loss": 2.137,
      "step": 1922
    },
    {
      "epoch": 0.42387171433294757,
      "grad_norm": 0.7362533211708069,
      "learning_rate": 0.0001329050204074227,
      "loss": 2.0262,
      "step": 1923
    },
    {
      "epoch": 0.4240921364412851,
      "grad_norm": 0.7070260047912598,
      "learning_rate": 0.00013283618000191187,
      "loss": 2.1043,
      "step": 1924
    },
    {
      "epoch": 0.4243125585496225,
      "grad_norm": 0.7380604147911072,
      "learning_rate": 0.0001327673221503229,
      "loss": 2.1535,
      "step": 1925
    },
    {
      "epoch": 0.42453298065795997,
      "grad_norm": 0.7468699216842651,
      "learning_rate": 0.00013269844688924043,
      "loss": 2.1404,
      "step": 1926
    },
    {
      "epoch": 0.42475340276629747,
      "grad_norm": 0.7948705554008484,
      "learning_rate": 0.0001326295542552583,
      "loss": 2.2042,
      "step": 1927
    },
    {
      "epoch": 0.4249738248746349,
      "grad_norm": 0.7763567566871643,
      "learning_rate": 0.00013256064428497966,
      "loss": 2.0991,
      "step": 1928
    },
    {
      "epoch": 0.42519424698297237,
      "grad_norm": 0.7509999871253967,
      "learning_rate": 0.00013249171701501677,
      "loss": 2.1056,
      "step": 1929
    },
    {
      "epoch": 0.42541466909130987,
      "grad_norm": 0.7326681017875671,
      "learning_rate": 0.00013242277248199123,
      "loss": 2.1258,
      "step": 1930
    },
    {
      "epoch": 0.4256350911996473,
      "grad_norm": 0.7363303899765015,
      "learning_rate": 0.00013235381072253363,
      "loss": 2.1532,
      "step": 1931
    },
    {
      "epoch": 0.42585551330798477,
      "grad_norm": 0.7516984343528748,
      "learning_rate": 0.00013228483177328386,
      "loss": 2.2205,
      "step": 1932
    },
    {
      "epoch": 0.42607593541632227,
      "grad_norm": 0.7701885104179382,
      "learning_rate": 0.00013221583567089086,
      "loss": 2.1714,
      "step": 1933
    },
    {
      "epoch": 0.4262963575246597,
      "grad_norm": 0.7550076246261597,
      "learning_rate": 0.00013214682245201275,
      "loss": 2.1856,
      "step": 1934
    },
    {
      "epoch": 0.42651677963299717,
      "grad_norm": 0.7052896618843079,
      "learning_rate": 0.00013207779215331669,
      "loss": 2.2122,
      "step": 1935
    },
    {
      "epoch": 0.42673720174133467,
      "grad_norm": 0.711096465587616,
      "learning_rate": 0.00013200874481147893,
      "loss": 2.0983,
      "step": 1936
    },
    {
      "epoch": 0.4269576238496721,
      "grad_norm": 0.6758118271827698,
      "learning_rate": 0.00013193968046318474,
      "loss": 2.2863,
      "step": 1937
    },
    {
      "epoch": 0.42717804595800957,
      "grad_norm": 0.752456545829773,
      "learning_rate": 0.0001318705991451285,
      "loss": 2.2002,
      "step": 1938
    },
    {
      "epoch": 0.42739846806634707,
      "grad_norm": 0.70738285779953,
      "learning_rate": 0.00013180150089401357,
      "loss": 2.1831,
      "step": 1939
    },
    {
      "epoch": 0.4276188901746845,
      "grad_norm": 0.7642098665237427,
      "learning_rate": 0.00013173238574655231,
      "loss": 2.1266,
      "step": 1940
    },
    {
      "epoch": 0.42783931228302197,
      "grad_norm": 0.7049610614776611,
      "learning_rate": 0.00013166325373946606,
      "loss": 2.2478,
      "step": 1941
    },
    {
      "epoch": 0.42805973439135947,
      "grad_norm": 0.7409012913703918,
      "learning_rate": 0.00013159410490948506,
      "loss": 2.1838,
      "step": 1942
    },
    {
      "epoch": 0.4282801564996969,
      "grad_norm": 0.7849397659301758,
      "learning_rate": 0.00013152493929334856,
      "loss": 2.1053,
      "step": 1943
    },
    {
      "epoch": 0.42850057860803437,
      "grad_norm": 0.8339770436286926,
      "learning_rate": 0.00013145575692780472,
      "loss": 2.3745,
      "step": 1944
    },
    {
      "epoch": 0.42872100071637187,
      "grad_norm": 0.7362892627716064,
      "learning_rate": 0.00013138655784961057,
      "loss": 2.1492,
      "step": 1945
    },
    {
      "epoch": 0.4289414228247093,
      "grad_norm": 0.7500270009040833,
      "learning_rate": 0.00013131734209553207,
      "loss": 2.2304,
      "step": 1946
    },
    {
      "epoch": 0.42916184493304677,
      "grad_norm": 0.7168473601341248,
      "learning_rate": 0.00013124810970234398,
      "loss": 2.1874,
      "step": 1947
    },
    {
      "epoch": 0.42938226704138427,
      "grad_norm": 0.7436546087265015,
      "learning_rate": 0.0001311788607068299,
      "loss": 2.2984,
      "step": 1948
    },
    {
      "epoch": 0.4296026891497217,
      "grad_norm": 0.7248890995979309,
      "learning_rate": 0.0001311095951457823,
      "loss": 2.1226,
      "step": 1949
    },
    {
      "epoch": 0.42982311125805917,
      "grad_norm": 0.7059470415115356,
      "learning_rate": 0.00013104031305600242,
      "loss": 2.1147,
      "step": 1950
    },
    {
      "epoch": 0.43004353336639667,
      "grad_norm": 0.7105802297592163,
      "learning_rate": 0.0001309710144743003,
      "loss": 2.1076,
      "step": 1951
    },
    {
      "epoch": 0.4302639554747341,
      "grad_norm": 0.7560177445411682,
      "learning_rate": 0.00013090169943749476,
      "loss": 2.1211,
      "step": 1952
    },
    {
      "epoch": 0.43048437758307156,
      "grad_norm": 0.7431502342224121,
      "learning_rate": 0.00013083236798241328,
      "loss": 2.1203,
      "step": 1953
    },
    {
      "epoch": 0.43070479969140907,
      "grad_norm": 0.7105681896209717,
      "learning_rate": 0.0001307630201458921,
      "loss": 2.3169,
      "step": 1954
    },
    {
      "epoch": 0.4309252217997465,
      "grad_norm": 0.7803578972816467,
      "learning_rate": 0.0001306936559647763,
      "loss": 2.1035,
      "step": 1955
    },
    {
      "epoch": 0.43114564390808396,
      "grad_norm": 0.6951037049293518,
      "learning_rate": 0.0001306242754759194,
      "loss": 2.1769,
      "step": 1956
    },
    {
      "epoch": 0.43136606601642147,
      "grad_norm": 0.7846428751945496,
      "learning_rate": 0.00013055487871618378,
      "loss": 2.1737,
      "step": 1957
    },
    {
      "epoch": 0.4315864881247589,
      "grad_norm": 0.7548218965530396,
      "learning_rate": 0.00013048546572244036,
      "loss": 2.2129,
      "step": 1958
    },
    {
      "epoch": 0.43180691023309636,
      "grad_norm": 0.7029872536659241,
      "learning_rate": 0.0001304160365315687,
      "loss": 2.1791,
      "step": 1959
    },
    {
      "epoch": 0.43202733234143387,
      "grad_norm": 0.7244204878807068,
      "learning_rate": 0.00013034659118045706,
      "loss": 2.1335,
      "step": 1960
    },
    {
      "epoch": 0.4322477544497713,
      "grad_norm": 0.7269174456596375,
      "learning_rate": 0.00013027712970600214,
      "loss": 2.0632,
      "step": 1961
    },
    {
      "epoch": 0.43246817655810876,
      "grad_norm": 0.7231120467185974,
      "learning_rate": 0.0001302076521451093,
      "loss": 2.2514,
      "step": 1962
    },
    {
      "epoch": 0.43268859866644627,
      "grad_norm": 0.7326658368110657,
      "learning_rate": 0.00013013815853469247,
      "loss": 2.1652,
      "step": 1963
    },
    {
      "epoch": 0.4329090207747837,
      "grad_norm": 0.7170719504356384,
      "learning_rate": 0.00013006864891167396,
      "loss": 2.1847,
      "step": 1964
    },
    {
      "epoch": 0.43312944288312116,
      "grad_norm": 0.7418602705001831,
      "learning_rate": 0.0001299991233129848,
      "loss": 2.2128,
      "step": 1965
    },
    {
      "epoch": 0.43334986499145867,
      "grad_norm": 0.7569207549095154,
      "learning_rate": 0.0001299295817755643,
      "loss": 2.1186,
      "step": 1966
    },
    {
      "epoch": 0.4335702870997961,
      "grad_norm": 0.7975344657897949,
      "learning_rate": 0.0001298600243363604,
      "loss": 2.1082,
      "step": 1967
    },
    {
      "epoch": 0.43379070920813356,
      "grad_norm": 0.74268639087677,
      "learning_rate": 0.00012979045103232945,
      "loss": 2.1668,
      "step": 1968
    },
    {
      "epoch": 0.43401113131647107,
      "grad_norm": 0.7399777173995972,
      "learning_rate": 0.00012972086190043614,
      "loss": 2.1633,
      "step": 1969
    },
    {
      "epoch": 0.4342315534248085,
      "grad_norm": 0.8036502599716187,
      "learning_rate": 0.00012965125697765364,
      "loss": 2.1802,
      "step": 1970
    },
    {
      "epoch": 0.43445197553314596,
      "grad_norm": 0.782170295715332,
      "learning_rate": 0.0001295816363009635,
      "loss": 2.1297,
      "step": 1971
    },
    {
      "epoch": 0.43467239764148347,
      "grad_norm": 0.7180203795433044,
      "learning_rate": 0.00012951199990735569,
      "loss": 2.2621,
      "step": 1972
    },
    {
      "epoch": 0.4348928197498209,
      "grad_norm": 0.7247685194015503,
      "learning_rate": 0.00012944234783382843,
      "loss": 2.1618,
      "step": 1973
    },
    {
      "epoch": 0.43511324185815836,
      "grad_norm": 0.7231366038322449,
      "learning_rate": 0.00012937268011738837,
      "loss": 2.1216,
      "step": 1974
    },
    {
      "epoch": 0.43533366396649587,
      "grad_norm": 0.7611396908760071,
      "learning_rate": 0.0001293029967950504,
      "loss": 2.1008,
      "step": 1975
    },
    {
      "epoch": 0.4355540860748333,
      "grad_norm": 0.8033261895179749,
      "learning_rate": 0.00012923329790383772,
      "loss": 2.1103,
      "step": 1976
    },
    {
      "epoch": 0.43577450818317076,
      "grad_norm": 0.7023736834526062,
      "learning_rate": 0.0001291635834807818,
      "loss": 2.1109,
      "step": 1977
    },
    {
      "epoch": 0.43599493029150826,
      "grad_norm": 0.7346162796020508,
      "learning_rate": 0.0001290938535629224,
      "loss": 2.0073,
      "step": 1978
    },
    {
      "epoch": 0.4362153523998457,
      "grad_norm": 0.7361973524093628,
      "learning_rate": 0.0001290241081873075,
      "loss": 2.147,
      "step": 1979
    },
    {
      "epoch": 0.43643577450818316,
      "grad_norm": 0.7692379355430603,
      "learning_rate": 0.0001289543473909932,
      "loss": 2.1205,
      "step": 1980
    },
    {
      "epoch": 0.43665619661652066,
      "grad_norm": 0.756182849407196,
      "learning_rate": 0.00012888457121104394,
      "loss": 2.0966,
      "step": 1981
    },
    {
      "epoch": 0.4368766187248581,
      "grad_norm": 0.6930826902389526,
      "learning_rate": 0.00012881477968453226,
      "loss": 2.1979,
      "step": 1982
    },
    {
      "epoch": 0.43709704083319556,
      "grad_norm": 0.7021186351776123,
      "learning_rate": 0.00012874497284853881,
      "loss": 2.1281,
      "step": 1983
    },
    {
      "epoch": 0.437317462941533,
      "grad_norm": 0.7813003659248352,
      "learning_rate": 0.00012867515074015248,
      "loss": 2.1729,
      "step": 1984
    },
    {
      "epoch": 0.4375378850498705,
      "grad_norm": 0.7753021717071533,
      "learning_rate": 0.00012860531339647016,
      "loss": 2.0494,
      "step": 1985
    },
    {
      "epoch": 0.43775830715820796,
      "grad_norm": 0.7863062024116516,
      "learning_rate": 0.0001285354608545969,
      "loss": 2.1332,
      "step": 1986
    },
    {
      "epoch": 0.4379787292665454,
      "grad_norm": 0.7499800324440002,
      "learning_rate": 0.00012846559315164585,
      "loss": 2.0864,
      "step": 1987
    },
    {
      "epoch": 0.4381991513748829,
      "grad_norm": 0.799106776714325,
      "learning_rate": 0.00012839571032473814,
      "loss": 2.0843,
      "step": 1988
    },
    {
      "epoch": 0.43841957348322036,
      "grad_norm": 0.7594006061553955,
      "learning_rate": 0.00012832581241100302,
      "loss": 2.1579,
      "step": 1989
    },
    {
      "epoch": 0.4386399955915578,
      "grad_norm": 0.6702348589897156,
      "learning_rate": 0.00012825589944757767,
      "loss": 2.0675,
      "step": 1990
    },
    {
      "epoch": 0.4388604176998953,
      "grad_norm": 0.7613632678985596,
      "learning_rate": 0.0001281859714716073,
      "loss": 2.1565,
      "step": 1991
    },
    {
      "epoch": 0.43908083980823276,
      "grad_norm": 0.7174550890922546,
      "learning_rate": 0.00012811602852024512,
      "loss": 2.1407,
      "step": 1992
    },
    {
      "epoch": 0.4393012619165702,
      "grad_norm": 0.7647293210029602,
      "learning_rate": 0.00012804607063065227,
      "loss": 2.1797,
      "step": 1993
    },
    {
      "epoch": 0.4395216840249077,
      "grad_norm": 0.6767035722732544,
      "learning_rate": 0.0001279760978399979,
      "loss": 2.1008,
      "step": 1994
    },
    {
      "epoch": 0.43974210613324516,
      "grad_norm": 0.7468788027763367,
      "learning_rate": 0.00012790611018545893,
      "loss": 2.103,
      "step": 1995
    },
    {
      "epoch": 0.4399625282415826,
      "grad_norm": 0.7784439325332642,
      "learning_rate": 0.00012783610770422029,
      "loss": 2.0638,
      "step": 1996
    },
    {
      "epoch": 0.4401829503499201,
      "grad_norm": 0.7071070671081543,
      "learning_rate": 0.00012776609043347475,
      "loss": 2.0708,
      "step": 1997
    },
    {
      "epoch": 0.44040337245825756,
      "grad_norm": 0.7477219700813293,
      "learning_rate": 0.000127696058410423,
      "loss": 2.1647,
      "step": 1998
    },
    {
      "epoch": 0.440623794566595,
      "grad_norm": 0.771935224533081,
      "learning_rate": 0.00012762601167227343,
      "loss": 2.2038,
      "step": 1999
    },
    {
      "epoch": 0.4408442166749325,
      "grad_norm": 0.6906226873397827,
      "learning_rate": 0.0001275559502562424,
      "loss": 2.2016,
      "step": 2000
    },
    {
      "epoch": 0.44106463878326996,
      "grad_norm": 0.7634009718894958,
      "learning_rate": 0.00012748587419955396,
      "loss": 2.1646,
      "step": 2001
    },
    {
      "epoch": 0.4412850608916074,
      "grad_norm": 0.7529551982879639,
      "learning_rate": 0.00012741578353943998,
      "loss": 2.2307,
      "step": 2002
    },
    {
      "epoch": 0.4415054829999449,
      "grad_norm": 0.7626884579658508,
      "learning_rate": 0.0001273456783131401,
      "loss": 2.1168,
      "step": 2003
    },
    {
      "epoch": 0.44172590510828236,
      "grad_norm": 0.6805245280265808,
      "learning_rate": 0.00012727555855790166,
      "loss": 2.1108,
      "step": 2004
    },
    {
      "epoch": 0.4419463272166198,
      "grad_norm": 0.7127764225006104,
      "learning_rate": 0.00012720542431097982,
      "loss": 2.1422,
      "step": 2005
    },
    {
      "epoch": 0.4421667493249573,
      "grad_norm": 0.7429713606834412,
      "learning_rate": 0.0001271352756096373,
      "loss": 2.2031,
      "step": 2006
    },
    {
      "epoch": 0.44238717143329476,
      "grad_norm": 0.742557168006897,
      "learning_rate": 0.00012706511249114462,
      "loss": 2.1706,
      "step": 2007
    },
    {
      "epoch": 0.4426075935416322,
      "grad_norm": 0.7075592279434204,
      "learning_rate": 0.00012699493499277983,
      "loss": 2.2209,
      "step": 2008
    },
    {
      "epoch": 0.4428280156499697,
      "grad_norm": 0.7912879586219788,
      "learning_rate": 0.00012692474315182875,
      "loss": 2.3263,
      "step": 2009
    },
    {
      "epoch": 0.44304843775830716,
      "grad_norm": 0.7327765822410583,
      "learning_rate": 0.00012685453700558475,
      "loss": 2.0387,
      "step": 2010
    },
    {
      "epoch": 0.4432688598666446,
      "grad_norm": 0.7346118092536926,
      "learning_rate": 0.00012678431659134885,
      "loss": 2.1235,
      "step": 2011
    },
    {
      "epoch": 0.4434892819749821,
      "grad_norm": 0.7420346736907959,
      "learning_rate": 0.00012671408194642956,
      "loss": 2.1393,
      "step": 2012
    },
    {
      "epoch": 0.44370970408331956,
      "grad_norm": 0.7412707805633545,
      "learning_rate": 0.00012664383310814303,
      "loss": 2.1812,
      "step": 2013
    },
    {
      "epoch": 0.443930126191657,
      "grad_norm": 0.7446384429931641,
      "learning_rate": 0.00012657357011381293,
      "loss": 2.0798,
      "step": 2014
    },
    {
      "epoch": 0.4441505482999945,
      "grad_norm": 0.7095444202423096,
      "learning_rate": 0.00012650329300077046,
      "loss": 2.1117,
      "step": 2015
    },
    {
      "epoch": 0.44437097040833196,
      "grad_norm": 0.7023048400878906,
      "learning_rate": 0.00012643300180635434,
      "loss": 2.0492,
      "step": 2016
    },
    {
      "epoch": 0.4445913925166694,
      "grad_norm": 0.721912682056427,
      "learning_rate": 0.0001263626965679107,
      "loss": 1.9539,
      "step": 2017
    },
    {
      "epoch": 0.4448118146250069,
      "grad_norm": 0.7520800828933716,
      "learning_rate": 0.00012629237732279314,
      "loss": 2.1516,
      "step": 2018
    },
    {
      "epoch": 0.44503223673334436,
      "grad_norm": 0.7262508273124695,
      "learning_rate": 0.0001262220441083628,
      "loss": 2.0326,
      "step": 2019
    },
    {
      "epoch": 0.4452526588416818,
      "grad_norm": 0.7140916585922241,
      "learning_rate": 0.0001261516969619882,
      "loss": 2.1952,
      "step": 2020
    },
    {
      "epoch": 0.4454730809500193,
      "grad_norm": 0.7059095501899719,
      "learning_rate": 0.00012608133592104519,
      "loss": 2.249,
      "step": 2021
    },
    {
      "epoch": 0.44569350305835675,
      "grad_norm": 0.7781206965446472,
      "learning_rate": 0.00012601096102291706,
      "loss": 2.2063,
      "step": 2022
    },
    {
      "epoch": 0.4459139251666942,
      "grad_norm": 0.7289319038391113,
      "learning_rate": 0.00012594057230499444,
      "loss": 2.1244,
      "step": 2023
    },
    {
      "epoch": 0.4461343472750317,
      "grad_norm": 0.7769238948822021,
      "learning_rate": 0.00012587016980467535,
      "loss": 2.3549,
      "step": 2024
    },
    {
      "epoch": 0.44635476938336915,
      "grad_norm": 0.7663074135780334,
      "learning_rate": 0.00012579975355936511,
      "loss": 2.2729,
      "step": 2025
    },
    {
      "epoch": 0.4465751914917066,
      "grad_norm": 0.7959907054901123,
      "learning_rate": 0.00012572932360647632,
      "loss": 2.0949,
      "step": 2026
    },
    {
      "epoch": 0.4467956136000441,
      "grad_norm": 0.7511812448501587,
      "learning_rate": 0.00012565887998342888,
      "loss": 2.2423,
      "step": 2027
    },
    {
      "epoch": 0.44701603570838155,
      "grad_norm": 0.7291713953018188,
      "learning_rate": 0.0001255884227276499,
      "loss": 2.0437,
      "step": 2028
    },
    {
      "epoch": 0.447236457816719,
      "grad_norm": 0.7504295706748962,
      "learning_rate": 0.00012551795187657383,
      "loss": 2.2448,
      "step": 2029
    },
    {
      "epoch": 0.4474568799250565,
      "grad_norm": 0.7342820763587952,
      "learning_rate": 0.00012544746746764228,
      "loss": 2.0976,
      "step": 2030
    },
    {
      "epoch": 0.44767730203339395,
      "grad_norm": 0.7276750802993774,
      "learning_rate": 0.00012537696953830412,
      "loss": 2.1147,
      "step": 2031
    },
    {
      "epoch": 0.4478977241417314,
      "grad_norm": 0.7386437654495239,
      "learning_rate": 0.0001253064581260153,
      "loss": 2.1017,
      "step": 2032
    },
    {
      "epoch": 0.4481181462500689,
      "grad_norm": 0.7956511974334717,
      "learning_rate": 0.00012523593326823907,
      "loss": 2.0616,
      "step": 2033
    },
    {
      "epoch": 0.44833856835840635,
      "grad_norm": 0.7176355123519897,
      "learning_rate": 0.0001251653950024457,
      "loss": 2.2097,
      "step": 2034
    },
    {
      "epoch": 0.4485589904667438,
      "grad_norm": 0.7612098455429077,
      "learning_rate": 0.00012509484336611264,
      "loss": 2.1969,
      "step": 2035
    },
    {
      "epoch": 0.4487794125750813,
      "grad_norm": 0.6905826330184937,
      "learning_rate": 0.0001250242783967244,
      "loss": 2.0881,
      "step": 2036
    },
    {
      "epoch": 0.44899983468341875,
      "grad_norm": 0.7713532447814941,
      "learning_rate": 0.0001249537001317727,
      "loss": 2.1988,
      "step": 2037
    },
    {
      "epoch": 0.4492202567917562,
      "grad_norm": 0.7739441990852356,
      "learning_rate": 0.00012488310860875622,
      "loss": 2.1235,
      "step": 2038
    },
    {
      "epoch": 0.4494406789000937,
      "grad_norm": 0.7787931561470032,
      "learning_rate": 0.0001248125038651806,
      "loss": 2.0953,
      "step": 2039
    },
    {
      "epoch": 0.44966110100843115,
      "grad_norm": 0.7591634392738342,
      "learning_rate": 0.0001247418859385587,
      "loss": 2.1444,
      "step": 2040
    },
    {
      "epoch": 0.4498815231167686,
      "grad_norm": 0.7531824707984924,
      "learning_rate": 0.00012467125486641026,
      "loss": 2.1341,
      "step": 2041
    },
    {
      "epoch": 0.4501019452251061,
      "grad_norm": 0.7513538599014282,
      "learning_rate": 0.00012460061068626202,
      "loss": 2.1785,
      "step": 2042
    },
    {
      "epoch": 0.45032236733344355,
      "grad_norm": 0.7482624650001526,
      "learning_rate": 0.00012452995343564772,
      "loss": 2.042,
      "step": 2043
    },
    {
      "epoch": 0.450542789441781,
      "grad_norm": 0.7148988842964172,
      "learning_rate": 0.000124459283152108,
      "loss": 2.1314,
      "step": 2044
    },
    {
      "epoch": 0.4507632115501185,
      "grad_norm": 0.8215563297271729,
      "learning_rate": 0.0001243885998731905,
      "loss": 2.1172,
      "step": 2045
    },
    {
      "epoch": 0.45098363365845595,
      "grad_norm": 0.7962205410003662,
      "learning_rate": 0.00012431790363644964,
      "loss": 2.0812,
      "step": 2046
    },
    {
      "epoch": 0.4512040557667934,
      "grad_norm": 0.6927281618118286,
      "learning_rate": 0.0001242471944794468,
      "loss": 2.1467,
      "step": 2047
    },
    {
      "epoch": 0.4514244778751309,
      "grad_norm": 0.7194108366966248,
      "learning_rate": 0.0001241764724397503,
      "loss": 2.1026,
      "step": 2048
    },
    {
      "epoch": 0.45164489998346835,
      "grad_norm": 0.7728255391120911,
      "learning_rate": 0.00012410573755493513,
      "loss": 2.2155,
      "step": 2049
    },
    {
      "epoch": 0.4518653220918058,
      "grad_norm": 0.7110655307769775,
      "learning_rate": 0.00012403498986258326,
      "loss": 2.0611,
      "step": 2050
    },
    {
      "epoch": 0.45208574420014325,
      "grad_norm": 0.7718685865402222,
      "learning_rate": 0.00012396422940028336,
      "loss": 2.097,
      "step": 2051
    },
    {
      "epoch": 0.45230616630848075,
      "grad_norm": 0.7339054942131042,
      "learning_rate": 0.00012389345620563094,
      "loss": 2.2517,
      "step": 2052
    },
    {
      "epoch": 0.4525265884168182,
      "grad_norm": 0.7200445532798767,
      "learning_rate": 0.00012382267031622827,
      "loss": 2.2145,
      "step": 2053
    },
    {
      "epoch": 0.45274701052515565,
      "grad_norm": 0.7359113693237305,
      "learning_rate": 0.00012375187176968442,
      "loss": 2.1823,
      "step": 2054
    },
    {
      "epoch": 0.45296743263349315,
      "grad_norm": 0.6854445338249207,
      "learning_rate": 0.000123681060603615,
      "loss": 2.0185,
      "step": 2055
    },
    {
      "epoch": 0.4531878547418306,
      "grad_norm": 0.7387434840202332,
      "learning_rate": 0.0001236102368556425,
      "loss": 2.2118,
      "step": 2056
    },
    {
      "epoch": 0.45340827685016805,
      "grad_norm": 0.7578263878822327,
      "learning_rate": 0.0001235394005633961,
      "loss": 2.1293,
      "step": 2057
    },
    {
      "epoch": 0.45362869895850555,
      "grad_norm": 0.719235897064209,
      "learning_rate": 0.0001234685517645115,
      "loss": 2.2491,
      "step": 2058
    },
    {
      "epoch": 0.453849121066843,
      "grad_norm": 0.7150795459747314,
      "learning_rate": 0.00012339769049663122,
      "loss": 2.1426,
      "step": 2059
    },
    {
      "epoch": 0.45406954317518045,
      "grad_norm": 0.761154294013977,
      "learning_rate": 0.00012332681679740423,
      "loss": 2.1721,
      "step": 2060
    },
    {
      "epoch": 0.45428996528351795,
      "grad_norm": 0.7550941109657288,
      "learning_rate": 0.00012325593070448626,
      "loss": 2.1696,
      "step": 2061
    },
    {
      "epoch": 0.4545103873918554,
      "grad_norm": 0.7408329248428345,
      "learning_rate": 0.0001231850322555395,
      "loss": 2.0806,
      "step": 2062
    },
    {
      "epoch": 0.45473080950019285,
      "grad_norm": 0.7362509965896606,
      "learning_rate": 0.0001231141214882328,
      "loss": 2.0786,
      "step": 2063
    },
    {
      "epoch": 0.45495123160853035,
      "grad_norm": 0.7687767744064331,
      "learning_rate": 0.00012304319844024151,
      "loss": 2.1119,
      "step": 2064
    },
    {
      "epoch": 0.4551716537168678,
      "grad_norm": 0.7135959267616272,
      "learning_rate": 0.00012297226314924754,
      "loss": 2.1799,
      "step": 2065
    },
    {
      "epoch": 0.45539207582520524,
      "grad_norm": 0.7590441703796387,
      "learning_rate": 0.00012290131565293922,
      "loss": 2.1888,
      "step": 2066
    },
    {
      "epoch": 0.45561249793354275,
      "grad_norm": 0.775193989276886,
      "learning_rate": 0.00012283035598901142,
      "loss": 2.265,
      "step": 2067
    },
    {
      "epoch": 0.4558329200418802,
      "grad_norm": 0.7218160033226013,
      "learning_rate": 0.00012275938419516552,
      "loss": 2.1226,
      "step": 2068
    },
    {
      "epoch": 0.45605334215021764,
      "grad_norm": 0.6814563870429993,
      "learning_rate": 0.0001226884003091093,
      "loss": 2.1449,
      "step": 2069
    },
    {
      "epoch": 0.45627376425855515,
      "grad_norm": 0.6946084499359131,
      "learning_rate": 0.00012261740436855694,
      "loss": 2.0925,
      "step": 2070
    },
    {
      "epoch": 0.4564941863668926,
      "grad_norm": 0.7398958802223206,
      "learning_rate": 0.00012254639641122903,
      "loss": 2.0746,
      "step": 2071
    },
    {
      "epoch": 0.45671460847523004,
      "grad_norm": 0.7802919149398804,
      "learning_rate": 0.00012247537647485258,
      "loss": 2.1981,
      "step": 2072
    },
    {
      "epoch": 0.45693503058356755,
      "grad_norm": 0.7398055791854858,
      "learning_rate": 0.000122404344597161,
      "loss": 2.1474,
      "step": 2073
    },
    {
      "epoch": 0.457155452691905,
      "grad_norm": 0.6779043674468994,
      "learning_rate": 0.00012233330081589395,
      "loss": 2.1597,
      "step": 2074
    },
    {
      "epoch": 0.45737587480024244,
      "grad_norm": 0.7335158586502075,
      "learning_rate": 0.00012226224516879744,
      "loss": 2.1339,
      "step": 2075
    },
    {
      "epoch": 0.45759629690857995,
      "grad_norm": 0.6953923106193542,
      "learning_rate": 0.00012219117769362382,
      "loss": 2.1343,
      "step": 2076
    },
    {
      "epoch": 0.4578167190169174,
      "grad_norm": 0.7590250968933105,
      "learning_rate": 0.00012212009842813172,
      "loss": 2.0355,
      "step": 2077
    },
    {
      "epoch": 0.45803714112525484,
      "grad_norm": 0.7068051695823669,
      "learning_rate": 0.000122049007410086,
      "loss": 2.0343,
      "step": 2078
    },
    {
      "epoch": 0.45825756323359235,
      "grad_norm": 0.7807449102401733,
      "learning_rate": 0.0001219779046772578,
      "loss": 2.1927,
      "step": 2079
    },
    {
      "epoch": 0.4584779853419298,
      "grad_norm": 0.7421174049377441,
      "learning_rate": 0.00012190679026742447,
      "loss": 2.1855,
      "step": 2080
    },
    {
      "epoch": 0.45869840745026724,
      "grad_norm": 0.699884831905365,
      "learning_rate": 0.00012183566421836956,
      "loss": 2.1907,
      "step": 2081
    },
    {
      "epoch": 0.45891882955860475,
      "grad_norm": 0.7483757734298706,
      "learning_rate": 0.0001217645265678828,
      "loss": 2.132,
      "step": 2082
    },
    {
      "epoch": 0.4591392516669422,
      "grad_norm": 0.7277604937553406,
      "learning_rate": 0.00012169337735376007,
      "loss": 2.1488,
      "step": 2083
    },
    {
      "epoch": 0.45935967377527964,
      "grad_norm": 0.6952170133590698,
      "learning_rate": 0.00012162221661380344,
      "loss": 2.2124,
      "step": 2084
    },
    {
      "epoch": 0.45958009588361715,
      "grad_norm": 0.7431680560112,
      "learning_rate": 0.00012155104438582108,
      "loss": 2.1971,
      "step": 2085
    },
    {
      "epoch": 0.4598005179919546,
      "grad_norm": 0.717024028301239,
      "learning_rate": 0.00012147986070762729,
      "loss": 2.1072,
      "step": 2086
    },
    {
      "epoch": 0.46002094010029204,
      "grad_norm": 0.713097333908081,
      "learning_rate": 0.00012140866561704239,
      "loss": 2.0784,
      "step": 2087
    },
    {
      "epoch": 0.46024136220862955,
      "grad_norm": 0.7400550842285156,
      "learning_rate": 0.00012133745915189278,
      "loss": 2.039,
      "step": 2088
    },
    {
      "epoch": 0.460461784316967,
      "grad_norm": 0.7631825804710388,
      "learning_rate": 0.00012126624135001097,
      "loss": 2.1564,
      "step": 2089
    },
    {
      "epoch": 0.46068220642530444,
      "grad_norm": 0.7432628273963928,
      "learning_rate": 0.00012119501224923542,
      "loss": 2.1604,
      "step": 2090
    },
    {
      "epoch": 0.46090262853364194,
      "grad_norm": 0.6737289428710938,
      "learning_rate": 0.00012112377188741062,
      "loss": 2.1329,
      "step": 2091
    },
    {
      "epoch": 0.4611230506419794,
      "grad_norm": 0.6865572929382324,
      "learning_rate": 0.00012105252030238705,
      "loss": 2.1423,
      "step": 2092
    },
    {
      "epoch": 0.46134347275031684,
      "grad_norm": 0.7198622822761536,
      "learning_rate": 0.00012098125753202114,
      "loss": 2.0708,
      "step": 2093
    },
    {
      "epoch": 0.46156389485865434,
      "grad_norm": 0.7435035109519958,
      "learning_rate": 0.00012090998361417525,
      "loss": 2.0288,
      "step": 2094
    },
    {
      "epoch": 0.4617843169669918,
      "grad_norm": 0.7253754734992981,
      "learning_rate": 0.00012083869858671773,
      "loss": 2.1285,
      "step": 2095
    },
    {
      "epoch": 0.46200473907532924,
      "grad_norm": 0.6749172210693359,
      "learning_rate": 0.00012076740248752271,
      "loss": 2.1373,
      "step": 2096
    },
    {
      "epoch": 0.46222516118366674,
      "grad_norm": 0.6663712859153748,
      "learning_rate": 0.00012069609535447036,
      "loss": 2.1143,
      "step": 2097
    },
    {
      "epoch": 0.4624455832920042,
      "grad_norm": 0.7608535289764404,
      "learning_rate": 0.00012062477722544656,
      "loss": 2.1327,
      "step": 2098
    },
    {
      "epoch": 0.46266600540034164,
      "grad_norm": 0.7111762762069702,
      "learning_rate": 0.00012055344813834312,
      "loss": 2.0887,
      "step": 2099
    },
    {
      "epoch": 0.46288642750867914,
      "grad_norm": 0.7298000454902649,
      "learning_rate": 0.00012048210813105769,
      "loss": 2.1086,
      "step": 2100
    },
    {
      "epoch": 0.46288642750867914,
      "eval_bpb": 1.107709667847284,
      "eval_loss": 2.1349997520446777,
      "eval_runtime": 184.2515,
      "eval_samples_per_second": 62.198,
      "eval_steps_per_second": 7.777,
      "step": 2100
    },
    {
      "epoch": 0.4631068496170166,
      "grad_norm": 0.7360554337501526,
      "learning_rate": 0.00012041075724149364,
      "loss": 2.14,
      "step": 2101
    },
    {
      "epoch": 0.46332727172535404,
      "grad_norm": 0.7007378935813904,
      "learning_rate": 0.00012033939550756022,
      "loss": 2.2031,
      "step": 2102
    },
    {
      "epoch": 0.46354769383369154,
      "grad_norm": 0.7007319927215576,
      "learning_rate": 0.0001202680229671723,
      "loss": 2.1215,
      "step": 2103
    },
    {
      "epoch": 0.463768115942029,
      "grad_norm": 0.7597978711128235,
      "learning_rate": 0.00012019663965825063,
      "loss": 2.0732,
      "step": 2104
    },
    {
      "epoch": 0.46398853805036644,
      "grad_norm": 0.7188252210617065,
      "learning_rate": 0.00012012524561872164,
      "loss": 2.2241,
      "step": 2105
    },
    {
      "epoch": 0.46420896015870394,
      "grad_norm": 0.7616143226623535,
      "learning_rate": 0.00012005384088651745,
      "loss": 2.0942,
      "step": 2106
    },
    {
      "epoch": 0.4644293822670414,
      "grad_norm": 0.752273440361023,
      "learning_rate": 0.00011998242549957583,
      "loss": 2.2016,
      "step": 2107
    },
    {
      "epoch": 0.46464980437537884,
      "grad_norm": 0.7024292349815369,
      "learning_rate": 0.00011991099949584032,
      "loss": 2.215,
      "step": 2108
    },
    {
      "epoch": 0.46487022648371634,
      "grad_norm": 0.7120001912117004,
      "learning_rate": 0.00011983956291325995,
      "loss": 1.9992,
      "step": 2109
    },
    {
      "epoch": 0.4650906485920538,
      "grad_norm": 0.7334387898445129,
      "learning_rate": 0.00011976811578978945,
      "loss": 2.1426,
      "step": 2110
    },
    {
      "epoch": 0.46531107070039124,
      "grad_norm": 0.7318382263183594,
      "learning_rate": 0.00011969665816338915,
      "loss": 2.2842,
      "step": 2111
    },
    {
      "epoch": 0.46553149280872874,
      "grad_norm": 4.146671295166016,
      "learning_rate": 0.00011962519007202502,
      "loss": 2.0867,
      "step": 2112
    },
    {
      "epoch": 0.4657519149170662,
      "grad_norm": 0.7452883720397949,
      "learning_rate": 0.00011955371155366846,
      "loss": 2.0452,
      "step": 2113
    },
    {
      "epoch": 0.46597233702540364,
      "grad_norm": 0.8541472554206848,
      "learning_rate": 0.00011948222264629649,
      "loss": 2.231,
      "step": 2114
    },
    {
      "epoch": 0.46619275913374114,
      "grad_norm": 0.6908783912658691,
      "learning_rate": 0.00011941072338789163,
      "loss": 2.0982,
      "step": 2115
    },
    {
      "epoch": 0.4664131812420786,
      "grad_norm": 0.7137777805328369,
      "learning_rate": 0.00011933921381644193,
      "loss": 2.1167,
      "step": 2116
    },
    {
      "epoch": 0.46663360335041604,
      "grad_norm": 0.7703360319137573,
      "learning_rate": 0.00011926769396994089,
      "loss": 2.2474,
      "step": 2117
    },
    {
      "epoch": 0.4668540254587535,
      "grad_norm": 0.796297013759613,
      "learning_rate": 0.00011919616388638748,
      "loss": 2.068,
      "step": 2118
    },
    {
      "epoch": 0.467074447567091,
      "grad_norm": 0.7730244994163513,
      "learning_rate": 0.00011912462360378606,
      "loss": 2.1134,
      "step": 2119
    },
    {
      "epoch": 0.46729486967542844,
      "grad_norm": 0.7715543508529663,
      "learning_rate": 0.00011905307316014649,
      "loss": 2.0894,
      "step": 2120
    },
    {
      "epoch": 0.4675152917837659,
      "grad_norm": 0.7846410274505615,
      "learning_rate": 0.00011898151259348398,
      "loss": 2.101,
      "step": 2121
    },
    {
      "epoch": 0.4677357138921034,
      "grad_norm": 0.744963526725769,
      "learning_rate": 0.00011890994194181915,
      "loss": 2.1774,
      "step": 2122
    },
    {
      "epoch": 0.46795613600044084,
      "grad_norm": 0.7751252055168152,
      "learning_rate": 0.00011883836124317791,
      "loss": 2.1161,
      "step": 2123
    },
    {
      "epoch": 0.4681765581087783,
      "grad_norm": 0.7955819368362427,
      "learning_rate": 0.0001187667705355916,
      "loss": 2.3257,
      "step": 2124
    },
    {
      "epoch": 0.4683969802171158,
      "grad_norm": 0.7849298715591431,
      "learning_rate": 0.00011869516985709681,
      "loss": 2.0067,
      "step": 2125
    },
    {
      "epoch": 0.46861740232545324,
      "grad_norm": 0.8336432576179504,
      "learning_rate": 0.00011862355924573547,
      "loss": 2.0956,
      "step": 2126
    },
    {
      "epoch": 0.4688378244337907,
      "grad_norm": 0.7471098303794861,
      "learning_rate": 0.00011855193873955474,
      "loss": 2.1182,
      "step": 2127
    },
    {
      "epoch": 0.4690582465421282,
      "grad_norm": 0.6919596195220947,
      "learning_rate": 0.00011848030837660709,
      "loss": 2.0215,
      "step": 2128
    },
    {
      "epoch": 0.46927866865046564,
      "grad_norm": 0.718055248260498,
      "learning_rate": 0.00011840866819495023,
      "loss": 2.1756,
      "step": 2129
    },
    {
      "epoch": 0.4694990907588031,
      "grad_norm": 0.8368803858757019,
      "learning_rate": 0.00011833701823264698,
      "loss": 2.0557,
      "step": 2130
    },
    {
      "epoch": 0.4697195128671406,
      "grad_norm": 0.7911282777786255,
      "learning_rate": 0.00011826535852776551,
      "loss": 2.226,
      "step": 2131
    },
    {
      "epoch": 0.46993993497547804,
      "grad_norm": 0.794008731842041,
      "learning_rate": 0.00011819368911837903,
      "loss": 2.1643,
      "step": 2132
    },
    {
      "epoch": 0.4701603570838155,
      "grad_norm": 0.7203131318092346,
      "learning_rate": 0.000118122010042566,
      "loss": 2.1001,
      "step": 2133
    },
    {
      "epoch": 0.470380779192153,
      "grad_norm": 0.7338320016860962,
      "learning_rate": 0.00011805032133841,
      "loss": 2.1165,
      "step": 2134
    },
    {
      "epoch": 0.47060120130049043,
      "grad_norm": 0.7342604994773865,
      "learning_rate": 0.00011797862304399971,
      "loss": 2.1426,
      "step": 2135
    },
    {
      "epoch": 0.4708216234088279,
      "grad_norm": 0.7228399515151978,
      "learning_rate": 0.00011790691519742887,
      "loss": 2.2444,
      "step": 2136
    },
    {
      "epoch": 0.4710420455171654,
      "grad_norm": 0.7702513933181763,
      "learning_rate": 0.0001178351978367963,
      "loss": 2.1138,
      "step": 2137
    },
    {
      "epoch": 0.47126246762550283,
      "grad_norm": 0.7022801041603088,
      "learning_rate": 0.00011776347100020602,
      "loss": 2.1435,
      "step": 2138
    },
    {
      "epoch": 0.4714828897338403,
      "grad_norm": 0.8309962153434753,
      "learning_rate": 0.00011769173472576689,
      "loss": 2.0374,
      "step": 2139
    },
    {
      "epoch": 0.4717033118421778,
      "grad_norm": 0.7213171720504761,
      "learning_rate": 0.00011761998905159286,
      "loss": 2.1439,
      "step": 2140
    },
    {
      "epoch": 0.47192373395051523,
      "grad_norm": 0.8065428733825684,
      "learning_rate": 0.0001175482340158029,
      "loss": 2.1233,
      "step": 2141
    },
    {
      "epoch": 0.4721441560588527,
      "grad_norm": 0.6911716461181641,
      "learning_rate": 0.00011747646965652092,
      "loss": 2.0781,
      "step": 2142
    },
    {
      "epoch": 0.4723645781671902,
      "grad_norm": 0.7340414524078369,
      "learning_rate": 0.00011740469601187581,
      "loss": 2.0745,
      "step": 2143
    },
    {
      "epoch": 0.47258500027552763,
      "grad_norm": 0.7015679478645325,
      "learning_rate": 0.00011733291312000136,
      "loss": 2.0935,
      "step": 2144
    },
    {
      "epoch": 0.4728054223838651,
      "grad_norm": 0.6985213756561279,
      "learning_rate": 0.00011726112101903632,
      "loss": 2.1958,
      "step": 2145
    },
    {
      "epoch": 0.4730258444922026,
      "grad_norm": 0.782618522644043,
      "learning_rate": 0.00011718931974712428,
      "loss": 2.0976,
      "step": 2146
    },
    {
      "epoch": 0.47324626660054003,
      "grad_norm": 0.7215837836265564,
      "learning_rate": 0.00011711750934241373,
      "loss": 2.1112,
      "step": 2147
    },
    {
      "epoch": 0.4734666887088775,
      "grad_norm": 0.7655563950538635,
      "learning_rate": 0.00011704568984305802,
      "loss": 2.1619,
      "step": 2148
    },
    {
      "epoch": 0.473687110817215,
      "grad_norm": 0.7296313643455505,
      "learning_rate": 0.0001169738612872153,
      "loss": 2.0786,
      "step": 2149
    },
    {
      "epoch": 0.47390753292555243,
      "grad_norm": 0.730797290802002,
      "learning_rate": 0.00011690202371304857,
      "loss": 2.1273,
      "step": 2150
    },
    {
      "epoch": 0.4741279550338899,
      "grad_norm": 0.7522875070571899,
      "learning_rate": 0.00011683017715872561,
      "loss": 2.1709,
      "step": 2151
    },
    {
      "epoch": 0.4743483771422274,
      "grad_norm": 0.7360877394676208,
      "learning_rate": 0.00011675832166241897,
      "loss": 2.1669,
      "step": 2152
    },
    {
      "epoch": 0.47456879925056483,
      "grad_norm": 0.6977972388267517,
      "learning_rate": 0.00011668645726230594,
      "loss": 2.089,
      "step": 2153
    },
    {
      "epoch": 0.4747892213589023,
      "grad_norm": 0.7941269278526306,
      "learning_rate": 0.00011661458399656853,
      "loss": 2.2189,
      "step": 2154
    },
    {
      "epoch": 0.4750096434672398,
      "grad_norm": 0.7417099475860596,
      "learning_rate": 0.00011654270190339352,
      "loss": 2.1828,
      "step": 2155
    },
    {
      "epoch": 0.47523006557557723,
      "grad_norm": 0.7257744669914246,
      "learning_rate": 0.0001164708110209723,
      "loss": 2.126,
      "step": 2156
    },
    {
      "epoch": 0.4754504876839147,
      "grad_norm": 0.7313467264175415,
      "learning_rate": 0.00011639891138750098,
      "loss": 2.1472,
      "step": 2157
    },
    {
      "epoch": 0.4756709097922522,
      "grad_norm": 0.7980832457542419,
      "learning_rate": 0.00011632700304118032,
      "loss": 2.0481,
      "step": 2158
    },
    {
      "epoch": 0.47589133190058963,
      "grad_norm": 0.6923190355300903,
      "learning_rate": 0.00011625508602021567,
      "loss": 2.1869,
      "step": 2159
    },
    {
      "epoch": 0.4761117540089271,
      "grad_norm": 0.7212961316108704,
      "learning_rate": 0.00011618316036281703,
      "loss": 2.238,
      "step": 2160
    },
    {
      "epoch": 0.4763321761172646,
      "grad_norm": 0.7680858373641968,
      "learning_rate": 0.000116111226107199,
      "loss": 2.0846,
      "step": 2161
    },
    {
      "epoch": 0.47655259822560203,
      "grad_norm": 0.7469529509544373,
      "learning_rate": 0.00011603928329158067,
      "loss": 2.0557,
      "step": 2162
    },
    {
      "epoch": 0.4767730203339395,
      "grad_norm": 0.7201281189918518,
      "learning_rate": 0.00011596733195418576,
      "loss": 2.0737,
      "step": 2163
    },
    {
      "epoch": 0.476993442442277,
      "grad_norm": 0.74009108543396,
      "learning_rate": 0.00011589537213324246,
      "loss": 2.1433,
      "step": 2164
    },
    {
      "epoch": 0.47721386455061443,
      "grad_norm": 0.7365696430206299,
      "learning_rate": 0.00011582340386698356,
      "loss": 2.1218,
      "step": 2165
    },
    {
      "epoch": 0.4774342866589519,
      "grad_norm": 0.7733455300331116,
      "learning_rate": 0.00011575142719364624,
      "loss": 2.1026,
      "step": 2166
    },
    {
      "epoch": 0.4776547087672894,
      "grad_norm": 0.6984227895736694,
      "learning_rate": 0.00011567944215147218,
      "loss": 2.2299,
      "step": 2167
    },
    {
      "epoch": 0.47787513087562683,
      "grad_norm": 0.7184359431266785,
      "learning_rate": 0.00011560744877870748,
      "loss": 2.0858,
      "step": 2168
    },
    {
      "epoch": 0.4780955529839643,
      "grad_norm": 0.7634490728378296,
      "learning_rate": 0.00011553544711360272,
      "loss": 2.1578,
      "step": 2169
    },
    {
      "epoch": 0.4783159750923018,
      "grad_norm": 0.7594035863876343,
      "learning_rate": 0.00011546343719441284,
      "loss": 2.2584,
      "step": 2170
    },
    {
      "epoch": 0.47853639720063923,
      "grad_norm": 0.7782122492790222,
      "learning_rate": 0.00011539141905939719,
      "loss": 2.1148,
      "step": 2171
    },
    {
      "epoch": 0.4787568193089767,
      "grad_norm": 0.7872025370597839,
      "learning_rate": 0.00011531939274681951,
      "loss": 2.2431,
      "step": 2172
    },
    {
      "epoch": 0.4789772414173142,
      "grad_norm": 0.7323572635650635,
      "learning_rate": 0.00011524735829494782,
      "loss": 2.1078,
      "step": 2173
    },
    {
      "epoch": 0.47919766352565163,
      "grad_norm": 0.724209189414978,
      "learning_rate": 0.00011517531574205447,
      "loss": 2.0647,
      "step": 2174
    },
    {
      "epoch": 0.4794180856339891,
      "grad_norm": 0.708552896976471,
      "learning_rate": 0.00011510326512641618,
      "loss": 2.1257,
      "step": 2175
    },
    {
      "epoch": 0.4796385077423266,
      "grad_norm": 0.7659894824028015,
      "learning_rate": 0.00011503120648631393,
      "loss": 2.1603,
      "step": 2176
    },
    {
      "epoch": 0.47985892985066403,
      "grad_norm": 0.7503497004508972,
      "learning_rate": 0.0001149591398600329,
      "loss": 2.1555,
      "step": 2177
    },
    {
      "epoch": 0.4800793519590015,
      "grad_norm": 0.7375928163528442,
      "learning_rate": 0.00011488706528586261,
      "loss": 2.3012,
      "step": 2178
    },
    {
      "epoch": 0.480299774067339,
      "grad_norm": 0.6804953217506409,
      "learning_rate": 0.00011481498280209673,
      "loss": 2.065,
      "step": 2179
    },
    {
      "epoch": 0.48052019617567643,
      "grad_norm": 0.7386634349822998,
      "learning_rate": 0.00011474289244703315,
      "loss": 2.1612,
      "step": 2180
    },
    {
      "epoch": 0.4807406182840139,
      "grad_norm": 0.7534124851226807,
      "learning_rate": 0.00011467079425897398,
      "loss": 2.0954,
      "step": 2181
    },
    {
      "epoch": 0.4809610403923514,
      "grad_norm": 0.7145940065383911,
      "learning_rate": 0.00011459868827622546,
      "loss": 2.0174,
      "step": 2182
    },
    {
      "epoch": 0.48118146250068883,
      "grad_norm": 0.7267360091209412,
      "learning_rate": 0.00011452657453709792,
      "loss": 2.0731,
      "step": 2183
    },
    {
      "epoch": 0.4814018846090263,
      "grad_norm": 0.7257755994796753,
      "learning_rate": 0.0001144544530799059,
      "loss": 2.0906,
      "step": 2184
    },
    {
      "epoch": 0.4816223067173637,
      "grad_norm": 0.7680655717849731,
      "learning_rate": 0.000114382323942968,
      "loss": 2.0314,
      "step": 2185
    },
    {
      "epoch": 0.4818427288257012,
      "grad_norm": 0.6998295783996582,
      "learning_rate": 0.00011431018716460691,
      "loss": 2.0154,
      "step": 2186
    },
    {
      "epoch": 0.4820631509340387,
      "grad_norm": 0.7526646852493286,
      "learning_rate": 0.00011423804278314938,
      "loss": 2.1797,
      "step": 2187
    },
    {
      "epoch": 0.4822835730423761,
      "grad_norm": 0.6984138488769531,
      "learning_rate": 0.00011416589083692619,
      "loss": 2.1822,
      "step": 2188
    },
    {
      "epoch": 0.4825039951507136,
      "grad_norm": 0.6860188841819763,
      "learning_rate": 0.00011409373136427213,
      "loss": 2.1318,
      "step": 2189
    },
    {
      "epoch": 0.4827244172590511,
      "grad_norm": 0.7493700385093689,
      "learning_rate": 0.00011402156440352599,
      "loss": 2.0846,
      "step": 2190
    },
    {
      "epoch": 0.4829448393673885,
      "grad_norm": 0.7893133759498596,
      "learning_rate": 0.00011394938999303055,
      "loss": 2.0786,
      "step": 2191
    },
    {
      "epoch": 0.483165261475726,
      "grad_norm": 0.707939088344574,
      "learning_rate": 0.00011387720817113262,
      "loss": 2.1771,
      "step": 2192
    },
    {
      "epoch": 0.4833856835840635,
      "grad_norm": 0.6779071092605591,
      "learning_rate": 0.00011380501897618279,
      "loss": 2.0564,
      "step": 2193
    },
    {
      "epoch": 0.4836061056924009,
      "grad_norm": 0.6828545331954956,
      "learning_rate": 0.00011373282244653571,
      "loss": 2.1195,
      "step": 2194
    },
    {
      "epoch": 0.4838265278007384,
      "grad_norm": 0.7574862241744995,
      "learning_rate": 0.00011366061862054983,
      "loss": 2.0656,
      "step": 2195
    },
    {
      "epoch": 0.4840469499090759,
      "grad_norm": 0.7548485994338989,
      "learning_rate": 0.00011358840753658752,
      "loss": 2.1351,
      "step": 2196
    },
    {
      "epoch": 0.4842673720174133,
      "grad_norm": 0.6783545613288879,
      "learning_rate": 0.00011351618923301506,
      "loss": 2.1838,
      "step": 2197
    },
    {
      "epoch": 0.4844877941257508,
      "grad_norm": 0.6986470222473145,
      "learning_rate": 0.00011344396374820244,
      "loss": 2.1048,
      "step": 2198
    },
    {
      "epoch": 0.4847082162340883,
      "grad_norm": 0.7281811833381653,
      "learning_rate": 0.00011337173112052357,
      "loss": 2.2575,
      "step": 2199
    },
    {
      "epoch": 0.4849286383424257,
      "grad_norm": 0.7745989561080933,
      "learning_rate": 0.00011329949138835611,
      "loss": 2.2179,
      "step": 2200
    },
    {
      "epoch": 0.4851490604507632,
      "grad_norm": 0.7232148051261902,
      "learning_rate": 0.00011322724459008151,
      "loss": 2.1118,
      "step": 2201
    },
    {
      "epoch": 0.4853694825591007,
      "grad_norm": 0.7727935910224915,
      "learning_rate": 0.00011315499076408496,
      "loss": 2.1568,
      "step": 2202
    },
    {
      "epoch": 0.4855899046674381,
      "grad_norm": 0.7579684853553772,
      "learning_rate": 0.0001130827299487554,
      "loss": 2.211,
      "step": 2203
    },
    {
      "epoch": 0.4858103267757756,
      "grad_norm": 0.7257877588272095,
      "learning_rate": 0.00011301046218248546,
      "loss": 2.2733,
      "step": 2204
    },
    {
      "epoch": 0.4860307488841131,
      "grad_norm": 0.7889353036880493,
      "learning_rate": 0.00011293818750367149,
      "loss": 2.0124,
      "step": 2205
    },
    {
      "epoch": 0.4862511709924505,
      "grad_norm": 0.762830376625061,
      "learning_rate": 0.00011286590595071351,
      "loss": 2.2875,
      "step": 2206
    },
    {
      "epoch": 0.486471593100788,
      "grad_norm": 0.6978264451026917,
      "learning_rate": 0.00011279361756201516,
      "loss": 2.119,
      "step": 2207
    },
    {
      "epoch": 0.4866920152091255,
      "grad_norm": 0.7260254621505737,
      "learning_rate": 0.00011272132237598376,
      "loss": 2.1182,
      "step": 2208
    },
    {
      "epoch": 0.4869124373174629,
      "grad_norm": 0.759587824344635,
      "learning_rate": 0.00011264902043103025,
      "loss": 2.1387,
      "step": 2209
    },
    {
      "epoch": 0.4871328594258004,
      "grad_norm": 0.7517433166503906,
      "learning_rate": 0.00011257671176556905,
      "loss": 2.2336,
      "step": 2210
    },
    {
      "epoch": 0.48735328153413787,
      "grad_norm": 0.79505854845047,
      "learning_rate": 0.00011250439641801829,
      "loss": 2.0733,
      "step": 2211
    },
    {
      "epoch": 0.4875737036424753,
      "grad_norm": 0.7517215609550476,
      "learning_rate": 0.00011243207442679955,
      "loss": 2.1289,
      "step": 2212
    },
    {
      "epoch": 0.4877941257508128,
      "grad_norm": 0.7425985932350159,
      "learning_rate": 0.00011235974583033799,
      "loss": 2.0773,
      "step": 2213
    },
    {
      "epoch": 0.48801454785915027,
      "grad_norm": 0.7477490901947021,
      "learning_rate": 0.00011228741066706228,
      "loss": 2.095,
      "step": 2214
    },
    {
      "epoch": 0.4882349699674877,
      "grad_norm": 0.783370316028595,
      "learning_rate": 0.0001122150689754046,
      "loss": 2.1714,
      "step": 2215
    },
    {
      "epoch": 0.4884553920758252,
      "grad_norm": 0.7841491103172302,
      "learning_rate": 0.00011214272079380048,
      "loss": 2.0632,
      "step": 2216
    },
    {
      "epoch": 0.48867581418416267,
      "grad_norm": 0.7677288055419922,
      "learning_rate": 0.000112070366160689,
      "loss": 2.0448,
      "step": 2217
    },
    {
      "epoch": 0.4888962362925001,
      "grad_norm": 0.7191498279571533,
      "learning_rate": 0.00011199800511451273,
      "loss": 2.0481,
      "step": 2218
    },
    {
      "epoch": 0.4891166584008376,
      "grad_norm": 0.7551862597465515,
      "learning_rate": 0.00011192563769371752,
      "loss": 2.045,
      "step": 2219
    },
    {
      "epoch": 0.48933708050917507,
      "grad_norm": 0.7334035038948059,
      "learning_rate": 0.00011185326393675269,
      "loss": 2.2923,
      "step": 2220
    },
    {
      "epoch": 0.4895575026175125,
      "grad_norm": 0.7416696548461914,
      "learning_rate": 0.00011178088388207082,
      "loss": 2.1392,
      "step": 2221
    },
    {
      "epoch": 0.48977792472585,
      "grad_norm": 0.8072440028190613,
      "learning_rate": 0.00011170849756812796,
      "loss": 2.2041,
      "step": 2222
    },
    {
      "epoch": 0.48999834683418747,
      "grad_norm": 0.7119025588035583,
      "learning_rate": 0.00011163610503338343,
      "loss": 2.0979,
      "step": 2223
    },
    {
      "epoch": 0.4902187689425249,
      "grad_norm": 0.7384172677993774,
      "learning_rate": 0.00011156370631629982,
      "loss": 2.0786,
      "step": 2224
    },
    {
      "epoch": 0.4904391910508624,
      "grad_norm": 0.7105064392089844,
      "learning_rate": 0.0001114913014553431,
      "loss": 2.146,
      "step": 2225
    },
    {
      "epoch": 0.49065961315919987,
      "grad_norm": 0.7530391812324524,
      "learning_rate": 0.00011141889048898246,
      "loss": 2.066,
      "step": 2226
    },
    {
      "epoch": 0.4908800352675373,
      "grad_norm": 0.7428449988365173,
      "learning_rate": 0.00011134647345569025,
      "loss": 2.1716,
      "step": 2227
    },
    {
      "epoch": 0.4911004573758748,
      "grad_norm": 0.7560111880302429,
      "learning_rate": 0.00011127405039394216,
      "loss": 2.112,
      "step": 2228
    },
    {
      "epoch": 0.49132087948421227,
      "grad_norm": 0.8205085396766663,
      "learning_rate": 0.00011120162134221704,
      "loss": 2.1777,
      "step": 2229
    },
    {
      "epoch": 0.4915413015925497,
      "grad_norm": 0.6896772384643555,
      "learning_rate": 0.0001111291863389969,
      "loss": 2.2326,
      "step": 2230
    },
    {
      "epoch": 0.4917617237008872,
      "grad_norm": 0.7244004011154175,
      "learning_rate": 0.00011105674542276694,
      "loss": 2.2293,
      "step": 2231
    },
    {
      "epoch": 0.49198214580922467,
      "grad_norm": 0.7901533246040344,
      "learning_rate": 0.00011098429863201551,
      "loss": 2.1759,
      "step": 2232
    },
    {
      "epoch": 0.4922025679175621,
      "grad_norm": 0.7375038266181946,
      "learning_rate": 0.00011091184600523406,
      "loss": 2.13,
      "step": 2233
    },
    {
      "epoch": 0.4924229900258996,
      "grad_norm": 0.748469352722168,
      "learning_rate": 0.00011083938758091714,
      "loss": 2.1217,
      "step": 2234
    },
    {
      "epoch": 0.49264341213423707,
      "grad_norm": 0.7392828464508057,
      "learning_rate": 0.00011076692339756238,
      "loss": 2.0339,
      "step": 2235
    },
    {
      "epoch": 0.4928638342425745,
      "grad_norm": 0.7181299924850464,
      "learning_rate": 0.00011069445349367051,
      "loss": 2.0531,
      "step": 2236
    },
    {
      "epoch": 0.493084256350912,
      "grad_norm": 0.7241842150688171,
      "learning_rate": 0.00011062197790774524,
      "loss": 2.1092,
      "step": 2237
    },
    {
      "epoch": 0.49330467845924947,
      "grad_norm": 0.7071158289909363,
      "learning_rate": 0.0001105494966782933,
      "loss": 2.0804,
      "step": 2238
    },
    {
      "epoch": 0.4935251005675869,
      "grad_norm": 0.7230584621429443,
      "learning_rate": 0.00011047700984382449,
      "loss": 2.1369,
      "step": 2239
    },
    {
      "epoch": 0.4937455226759244,
      "grad_norm": 0.7029907703399658,
      "learning_rate": 0.00011040451744285154,
      "loss": 2.1874,
      "step": 2240
    },
    {
      "epoch": 0.49396594478426187,
      "grad_norm": 0.7267805337905884,
      "learning_rate": 0.0001103320195138901,
      "loss": 2.1569,
      "step": 2241
    },
    {
      "epoch": 0.4941863668925993,
      "grad_norm": 0.776170015335083,
      "learning_rate": 0.00011025951609545884,
      "loss": 2.2203,
      "step": 2242
    },
    {
      "epoch": 0.4944067890009368,
      "grad_norm": 0.7368004322052002,
      "learning_rate": 0.00011018700722607924,
      "loss": 2.1532,
      "step": 2243
    },
    {
      "epoch": 0.49462721110927427,
      "grad_norm": 0.7149397730827332,
      "learning_rate": 0.00011011449294427578,
      "loss": 2.0991,
      "step": 2244
    },
    {
      "epoch": 0.4948476332176117,
      "grad_norm": 0.7853299379348755,
      "learning_rate": 0.00011004197328857578,
      "loss": 2.2221,
      "step": 2245
    },
    {
      "epoch": 0.4950680553259492,
      "grad_norm": 0.7368461489677429,
      "learning_rate": 0.00010996944829750944,
      "loss": 2.1794,
      "step": 2246
    },
    {
      "epoch": 0.49528847743428667,
      "grad_norm": 0.7351876497268677,
      "learning_rate": 0.00010989691800960969,
      "loss": 2.1097,
      "step": 2247
    },
    {
      "epoch": 0.4955088995426241,
      "grad_norm": 0.791110634803772,
      "learning_rate": 0.00010982438246341238,
      "loss": 2.0861,
      "step": 2248
    },
    {
      "epoch": 0.4957293216509616,
      "grad_norm": 0.7824254631996155,
      "learning_rate": 0.00010975184169745613,
      "loss": 2.0762,
      "step": 2249
    },
    {
      "epoch": 0.49594974375929907,
      "grad_norm": 0.7440174221992493,
      "learning_rate": 0.00010967929575028233,
      "loss": 2.0659,
      "step": 2250
    },
    {
      "epoch": 0.4961701658676365,
      "grad_norm": 0.6962351202964783,
      "learning_rate": 0.00010960674466043513,
      "loss": 2.177,
      "step": 2251
    },
    {
      "epoch": 0.49639058797597396,
      "grad_norm": 0.7629963755607605,
      "learning_rate": 0.00010953418846646138,
      "loss": 2.1953,
      "step": 2252
    },
    {
      "epoch": 0.49661101008431147,
      "grad_norm": 0.7953025698661804,
      "learning_rate": 0.00010946162720691062,
      "loss": 2.11,
      "step": 2253
    },
    {
      "epoch": 0.4968314321926489,
      "grad_norm": 0.7949841618537903,
      "learning_rate": 0.0001093890609203352,
      "loss": 2.1877,
      "step": 2254
    },
    {
      "epoch": 0.49705185430098636,
      "grad_norm": 0.7407306432723999,
      "learning_rate": 0.00010931648964529002,
      "loss": 2.2287,
      "step": 2255
    },
    {
      "epoch": 0.49727227640932387,
      "grad_norm": 0.7858365774154663,
      "learning_rate": 0.0001092439134203327,
      "loss": 2.0826,
      "step": 2256
    },
    {
      "epoch": 0.4974926985176613,
      "grad_norm": 0.6945883631706238,
      "learning_rate": 0.00010917133228402343,
      "loss": 2.1265,
      "step": 2257
    },
    {
      "epoch": 0.49771312062599876,
      "grad_norm": 0.7256321310997009,
      "learning_rate": 0.0001090987462749251,
      "loss": 2.1975,
      "step": 2258
    },
    {
      "epoch": 0.49793354273433627,
      "grad_norm": 0.7836722135543823,
      "learning_rate": 0.00010902615543160306,
      "loss": 2.0965,
      "step": 2259
    },
    {
      "epoch": 0.4981539648426737,
      "grad_norm": 0.7689793109893799,
      "learning_rate": 0.00010895355979262537,
      "loss": 2.1622,
      "step": 2260
    },
    {
      "epoch": 0.49837438695101116,
      "grad_norm": 0.7154636383056641,
      "learning_rate": 0.00010888095939656253,
      "loss": 2.11,
      "step": 2261
    },
    {
      "epoch": 0.49859480905934866,
      "grad_norm": 0.6939550638198853,
      "learning_rate": 0.00010880835428198764,
      "loss": 2.0747,
      "step": 2262
    },
    {
      "epoch": 0.4988152311676861,
      "grad_norm": 0.7907360792160034,
      "learning_rate": 0.00010873574448747626,
      "loss": 2.2224,
      "step": 2263
    },
    {
      "epoch": 0.49903565327602356,
      "grad_norm": 0.754021167755127,
      "learning_rate": 0.00010866313005160642,
      "loss": 2.1627,
      "step": 2264
    },
    {
      "epoch": 0.49925607538436106,
      "grad_norm": 0.7287217974662781,
      "learning_rate": 0.00010859051101295868,
      "loss": 2.1405,
      "step": 2265
    },
    {
      "epoch": 0.4994764974926985,
      "grad_norm": 0.7428033947944641,
      "learning_rate": 0.00010851788741011599,
      "loss": 2.0418,
      "step": 2266
    },
    {
      "epoch": 0.49969691960103596,
      "grad_norm": 0.7200166583061218,
      "learning_rate": 0.00010844525928166378,
      "loss": 2.1454,
      "step": 2267
    },
    {
      "epoch": 0.49991734170937346,
      "grad_norm": 0.7365553975105286,
      "learning_rate": 0.00010837262666618983,
      "loss": 2.1826,
      "step": 2268
    },
    {
      "epoch": 0.5001377638177109,
      "grad_norm": 0.738065779209137,
      "learning_rate": 0.00010829998960228436,
      "loss": 2.1641,
      "step": 2269
    },
    {
      "epoch": 0.5003581859260484,
      "grad_norm": 0.7608245611190796,
      "learning_rate": 0.00010822734812853981,
      "loss": 2.2381,
      "step": 2270
    },
    {
      "epoch": 0.5005786080343858,
      "grad_norm": 0.7546910643577576,
      "learning_rate": 0.0001081547022835512,
      "loss": 2.1639,
      "step": 2271
    },
    {
      "epoch": 0.5007990301427233,
      "grad_norm": 0.7248235940933228,
      "learning_rate": 0.00010808205210591569,
      "loss": 2.1611,
      "step": 2272
    },
    {
      "epoch": 0.5010194522510608,
      "grad_norm": 0.7164865732192993,
      "learning_rate": 0.00010800939763423281,
      "loss": 2.0979,
      "step": 2273
    },
    {
      "epoch": 0.5012398743593982,
      "grad_norm": 0.7351658940315247,
      "learning_rate": 0.00010793673890710436,
      "loss": 2.1538,
      "step": 2274
    },
    {
      "epoch": 0.5014602964677357,
      "grad_norm": 0.7341381311416626,
      "learning_rate": 0.00010786407596313437,
      "loss": 2.0873,
      "step": 2275
    },
    {
      "epoch": 0.5016807185760732,
      "grad_norm": 0.7923981547355652,
      "learning_rate": 0.00010779140884092915,
      "loss": 2.042,
      "step": 2276
    },
    {
      "epoch": 0.5019011406844106,
      "grad_norm": 0.7908720374107361,
      "learning_rate": 0.00010771873757909723,
      "loss": 2.0901,
      "step": 2277
    },
    {
      "epoch": 0.5021215627927481,
      "grad_norm": 0.7684547305107117,
      "learning_rate": 0.00010764606221624933,
      "loss": 2.0337,
      "step": 2278
    },
    {
      "epoch": 0.5023419849010856,
      "grad_norm": 0.697100818157196,
      "learning_rate": 0.00010757338279099838,
      "loss": 2.0977,
      "step": 2279
    },
    {
      "epoch": 0.502562407009423,
      "grad_norm": 0.7589430809020996,
      "learning_rate": 0.00010750069934195937,
      "loss": 2.0988,
      "step": 2280
    },
    {
      "epoch": 0.5027828291177605,
      "grad_norm": 0.8291736841201782,
      "learning_rate": 0.00010742801190774951,
      "loss": 2.0578,
      "step": 2281
    },
    {
      "epoch": 0.503003251226098,
      "grad_norm": 0.7273815274238586,
      "learning_rate": 0.00010735532052698816,
      "loss": 2.1709,
      "step": 2282
    },
    {
      "epoch": 0.5032236733344354,
      "grad_norm": 0.7001538872718811,
      "learning_rate": 0.00010728262523829666,
      "loss": 2.0671,
      "step": 2283
    },
    {
      "epoch": 0.5034440954427729,
      "grad_norm": 0.7181422114372253,
      "learning_rate": 0.00010720992608029854,
      "loss": 2.0956,
      "step": 2284
    },
    {
      "epoch": 0.5036645175511104,
      "grad_norm": 0.7898855209350586,
      "learning_rate": 0.00010713722309161933,
      "loss": 2.1099,
      "step": 2285
    },
    {
      "epoch": 0.5038849396594478,
      "grad_norm": 0.719743549823761,
      "learning_rate": 0.00010706451631088659,
      "loss": 2.0654,
      "step": 2286
    },
    {
      "epoch": 0.5041053617677853,
      "grad_norm": 0.7248046398162842,
      "learning_rate": 0.00010699180577672993,
      "loss": 2.0973,
      "step": 2287
    },
    {
      "epoch": 0.5043257838761228,
      "grad_norm": 0.7269353270530701,
      "learning_rate": 0.00010691909152778094,
      "loss": 2.0616,
      "step": 2288
    },
    {
      "epoch": 0.5045462059844602,
      "grad_norm": 0.764214277267456,
      "learning_rate": 0.00010684637360267317,
      "loss": 2.1107,
      "step": 2289
    },
    {
      "epoch": 0.5047666280927977,
      "grad_norm": 0.6994331479072571,
      "learning_rate": 0.00010677365204004215,
      "loss": 2.137,
      "step": 2290
    },
    {
      "epoch": 0.5049870502011352,
      "grad_norm": 0.7181599736213684,
      "learning_rate": 0.00010670092687852527,
      "loss": 2.0207,
      "step": 2291
    },
    {
      "epoch": 0.5052074723094726,
      "grad_norm": 0.6994780898094177,
      "learning_rate": 0.00010662819815676192,
      "loss": 2.0978,
      "step": 2292
    },
    {
      "epoch": 0.5054278944178101,
      "grad_norm": 0.7156521677970886,
      "learning_rate": 0.00010655546591339334,
      "loss": 2.1238,
      "step": 2293
    },
    {
      "epoch": 0.5056483165261476,
      "grad_norm": 0.718743622303009,
      "learning_rate": 0.00010648273018706267,
      "loss": 2.1641,
      "step": 2294
    },
    {
      "epoch": 0.505868738634485,
      "grad_norm": 0.7338970899581909,
      "learning_rate": 0.00010640999101641486,
      "loss": 2.1966,
      "step": 2295
    },
    {
      "epoch": 0.5060891607428225,
      "grad_norm": 0.751520037651062,
      "learning_rate": 0.00010633724844009671,
      "loss": 2.1508,
      "step": 2296
    },
    {
      "epoch": 0.50630958285116,
      "grad_norm": 0.7206107974052429,
      "learning_rate": 0.00010626450249675681,
      "loss": 2.1353,
      "step": 2297
    },
    {
      "epoch": 0.5065300049594974,
      "grad_norm": 0.6990411281585693,
      "learning_rate": 0.0001061917532250456,
      "loss": 2.2518,
      "step": 2298
    },
    {
      "epoch": 0.5067504270678349,
      "grad_norm": 0.7750140428543091,
      "learning_rate": 0.00010611900066361525,
      "loss": 1.9783,
      "step": 2299
    },
    {
      "epoch": 0.5069708491761724,
      "grad_norm": 0.6840697526931763,
      "learning_rate": 0.00010604624485111963,
      "loss": 2.2036,
      "step": 2300
    },
    {
      "epoch": 0.5071912712845098,
      "grad_norm": 0.6723833084106445,
      "learning_rate": 0.0001059734858262144,
      "loss": 2.2435,
      "step": 2301
    },
    {
      "epoch": 0.5074116933928473,
      "grad_norm": 0.7244005799293518,
      "learning_rate": 0.0001059007236275569,
      "loss": 2.0863,
      "step": 2302
    },
    {
      "epoch": 0.5076321155011848,
      "grad_norm": 0.7472475171089172,
      "learning_rate": 0.00010582795829380617,
      "loss": 2.1535,
      "step": 2303
    },
    {
      "epoch": 0.5078525376095222,
      "grad_norm": 0.7104523181915283,
      "learning_rate": 0.00010575518986362292,
      "loss": 2.2027,
      "step": 2304
    },
    {
      "epoch": 0.5080729597178597,
      "grad_norm": 0.816002607345581,
      "learning_rate": 0.00010568241837566947,
      "loss": 2.2729,
      "step": 2305
    },
    {
      "epoch": 0.5082933818261972,
      "grad_norm": 0.7171566486358643,
      "learning_rate": 0.00010560964386860985,
      "loss": 2.0659,
      "step": 2306
    },
    {
      "epoch": 0.5085138039345346,
      "grad_norm": 0.7246453762054443,
      "learning_rate": 0.00010553686638110958,
      "loss": 2.0402,
      "step": 2307
    },
    {
      "epoch": 0.5087342260428721,
      "grad_norm": 0.7671604752540588,
      "learning_rate": 0.00010546408595183578,
      "loss": 2.1553,
      "step": 2308
    },
    {
      "epoch": 0.5089546481512096,
      "grad_norm": 0.7255107164382935,
      "learning_rate": 0.00010539130261945727,
      "loss": 2.117,
      "step": 2309
    },
    {
      "epoch": 0.509175070259547,
      "grad_norm": 0.7393897175788879,
      "learning_rate": 0.00010531851642264421,
      "loss": 2.1317,
      "step": 2310
    },
    {
      "epoch": 0.5093954923678845,
      "grad_norm": 0.7659652233123779,
      "learning_rate": 0.00010524572740006848,
      "loss": 2.2729,
      "step": 2311
    },
    {
      "epoch": 0.509615914476222,
      "grad_norm": 0.7430046796798706,
      "learning_rate": 0.00010517293559040332,
      "loss": 2.2388,
      "step": 2312
    },
    {
      "epoch": 0.5098363365845594,
      "grad_norm": 0.7999524474143982,
      "learning_rate": 0.00010510014103232347,
      "loss": 2.2142,
      "step": 2313
    },
    {
      "epoch": 0.5100567586928969,
      "grad_norm": 0.6972869038581848,
      "learning_rate": 0.0001050273437645052,
      "loss": 2.0845,
      "step": 2314
    },
    {
      "epoch": 0.5102771808012344,
      "grad_norm": 0.7284692525863647,
      "learning_rate": 0.00010495454382562614,
      "loss": 2.1592,
      "step": 2315
    },
    {
      "epoch": 0.5104976029095718,
      "grad_norm": 0.7461291551589966,
      "learning_rate": 0.00010488174125436545,
      "loss": 2.2007,
      "step": 2316
    },
    {
      "epoch": 0.5107180250179093,
      "grad_norm": 0.7367551922798157,
      "learning_rate": 0.00010480893608940352,
      "loss": 2.0446,
      "step": 2317
    },
    {
      "epoch": 0.5109384471262468,
      "grad_norm": 0.7795435190200806,
      "learning_rate": 0.00010473612836942226,
      "loss": 2.1235,
      "step": 2318
    },
    {
      "epoch": 0.5111588692345842,
      "grad_norm": 0.6959381699562073,
      "learning_rate": 0.00010466331813310489,
      "loss": 2.1375,
      "step": 2319
    },
    {
      "epoch": 0.5113792913429217,
      "grad_norm": 0.7892638444900513,
      "learning_rate": 0.00010459050541913596,
      "loss": 2.1691,
      "step": 2320
    },
    {
      "epoch": 0.5115997134512592,
      "grad_norm": 0.745792031288147,
      "learning_rate": 0.00010451769026620132,
      "loss": 2.0906,
      "step": 2321
    },
    {
      "epoch": 0.5118201355595966,
      "grad_norm": 0.6911309361457825,
      "learning_rate": 0.00010444487271298821,
      "loss": 2.0818,
      "step": 2322
    },
    {
      "epoch": 0.5120405576679341,
      "grad_norm": 0.6992109417915344,
      "learning_rate": 0.00010437205279818501,
      "loss": 2.0301,
      "step": 2323
    },
    {
      "epoch": 0.5122609797762716,
      "grad_norm": 0.6829534769058228,
      "learning_rate": 0.00010429923056048139,
      "loss": 2.1355,
      "step": 2324
    },
    {
      "epoch": 0.512481401884609,
      "grad_norm": 0.7427393794059753,
      "learning_rate": 0.00010422640603856837,
      "loss": 2.1777,
      "step": 2325
    },
    {
      "epoch": 0.5127018239929465,
      "grad_norm": 0.7605338096618652,
      "learning_rate": 0.00010415357927113805,
      "loss": 2.1085,
      "step": 2326
    },
    {
      "epoch": 0.512922246101284,
      "grad_norm": 0.7413177490234375,
      "learning_rate": 0.0001040807502968838,
      "loss": 2.2125,
      "step": 2327
    },
    {
      "epoch": 0.5131426682096214,
      "grad_norm": 0.7050037980079651,
      "learning_rate": 0.00010400791915450009,
      "loss": 2.0707,
      "step": 2328
    },
    {
      "epoch": 0.5133630903179589,
      "grad_norm": 0.7424399852752686,
      "learning_rate": 0.0001039350858826826,
      "loss": 2.2153,
      "step": 2329
    },
    {
      "epoch": 0.5135835124262964,
      "grad_norm": 0.7386239171028137,
      "learning_rate": 0.00010386225052012812,
      "loss": 2.0725,
      "step": 2330
    },
    {
      "epoch": 0.5138039345346338,
      "grad_norm": 0.7047468423843384,
      "learning_rate": 0.00010378941310553458,
      "loss": 2.1127,
      "step": 2331
    },
    {
      "epoch": 0.5140243566429713,
      "grad_norm": 0.734835684299469,
      "learning_rate": 0.00010371657367760097,
      "loss": 2.1143,
      "step": 2332
    },
    {
      "epoch": 0.5142447787513088,
      "grad_norm": 0.7444323897361755,
      "learning_rate": 0.00010364373227502736,
      "loss": 2.2712,
      "step": 2333
    },
    {
      "epoch": 0.5144652008596462,
      "grad_norm": 0.7184443473815918,
      "learning_rate": 0.00010357088893651484,
      "loss": 2.099,
      "step": 2334
    },
    {
      "epoch": 0.5146856229679837,
      "grad_norm": 0.7584931254386902,
      "learning_rate": 0.00010349804370076554,
      "loss": 2.0676,
      "step": 2335
    },
    {
      "epoch": 0.5149060450763212,
      "grad_norm": 0.6984606385231018,
      "learning_rate": 0.00010342519660648266,
      "loss": 2.0988,
      "step": 2336
    },
    {
      "epoch": 0.5151264671846586,
      "grad_norm": 0.765191376209259,
      "learning_rate": 0.00010335234769237035,
      "loss": 2.1588,
      "step": 2337
    },
    {
      "epoch": 0.5153468892929961,
      "grad_norm": 0.6860733032226562,
      "learning_rate": 0.00010327949699713366,
      "loss": 2.035,
      "step": 2338
    },
    {
      "epoch": 0.5155673114013336,
      "grad_norm": 0.7537800669670105,
      "learning_rate": 0.00010320664455947865,
      "loss": 2.0633,
      "step": 2339
    },
    {
      "epoch": 0.515787733509671,
      "grad_norm": 0.7827544808387756,
      "learning_rate": 0.00010313379041811235,
      "loss": 2.1436,
      "step": 2340
    },
    {
      "epoch": 0.5160081556180085,
      "grad_norm": 0.7431343197822571,
      "learning_rate": 0.00010306093461174262,
      "loss": 2.1355,
      "step": 2341
    },
    {
      "epoch": 0.516228577726346,
      "grad_norm": 0.7279850840568542,
      "learning_rate": 0.00010298807717907822,
      "loss": 2.0545,
      "step": 2342
    },
    {
      "epoch": 0.5164489998346834,
      "grad_norm": 0.6864837408065796,
      "learning_rate": 0.00010291521815882878,
      "loss": 2.1083,
      "step": 2343
    },
    {
      "epoch": 0.5166694219430209,
      "grad_norm": 0.7535674571990967,
      "learning_rate": 0.0001028423575897048,
      "loss": 2.0281,
      "step": 2344
    },
    {
      "epoch": 0.5168898440513584,
      "grad_norm": 0.7894843816757202,
      "learning_rate": 0.00010276949551041752,
      "loss": 2.103,
      "step": 2345
    },
    {
      "epoch": 0.5171102661596958,
      "grad_norm": 0.8042173981666565,
      "learning_rate": 0.0001026966319596791,
      "loss": 2.0909,
      "step": 2346
    },
    {
      "epoch": 0.5173306882680333,
      "grad_norm": 0.7270336747169495,
      "learning_rate": 0.0001026237669762024,
      "loss": 2.202,
      "step": 2347
    },
    {
      "epoch": 0.5175511103763708,
      "grad_norm": 0.7120144367218018,
      "learning_rate": 0.00010255090059870107,
      "loss": 2.1016,
      "step": 2348
    },
    {
      "epoch": 0.5177715324847082,
      "grad_norm": 0.7502390742301941,
      "learning_rate": 0.00010247803286588952,
      "loss": 2.1774,
      "step": 2349
    },
    {
      "epoch": 0.5179919545930457,
      "grad_norm": 0.7358745336532593,
      "learning_rate": 0.00010240516381648278,
      "loss": 2.0764,
      "step": 2350
    },
    {
      "epoch": 0.5182123767013832,
      "grad_norm": 0.7425426840782166,
      "learning_rate": 0.00010233229348919673,
      "loss": 2.0518,
      "step": 2351
    },
    {
      "epoch": 0.5184327988097206,
      "grad_norm": 0.7377017140388489,
      "learning_rate": 0.00010225942192274782,
      "loss": 2.1712,
      "step": 2352
    },
    {
      "epoch": 0.5186532209180581,
      "grad_norm": 0.7634795904159546,
      "learning_rate": 0.00010218654915585321,
      "loss": 2.0454,
      "step": 2353
    },
    {
      "epoch": 0.5188736430263955,
      "grad_norm": 0.7245787978172302,
      "learning_rate": 0.00010211367522723067,
      "loss": 2.1787,
      "step": 2354
    },
    {
      "epoch": 0.519094065134733,
      "grad_norm": 0.725459098815918,
      "learning_rate": 0.00010204080017559861,
      "loss": 2.0646,
      "step": 2355
    },
    {
      "epoch": 0.5193144872430705,
      "grad_norm": 0.7333802580833435,
      "learning_rate": 0.00010196792403967599,
      "loss": 2.0887,
      "step": 2356
    },
    {
      "epoch": 0.5195349093514079,
      "grad_norm": 0.7809171676635742,
      "learning_rate": 0.00010189504685818239,
      "loss": 2.112,
      "step": 2357
    },
    {
      "epoch": 0.5197553314597454,
      "grad_norm": 0.7910590171813965,
      "learning_rate": 0.00010182216866983796,
      "loss": 1.9916,
      "step": 2358
    },
    {
      "epoch": 0.5199757535680829,
      "grad_norm": 0.748784601688385,
      "learning_rate": 0.00010174928951336333,
      "loss": 2.0928,
      "step": 2359
    },
    {
      "epoch": 0.5201961756764203,
      "grad_norm": 0.7520399689674377,
      "learning_rate": 0.00010167640942747973,
      "loss": 2.1091,
      "step": 2360
    },
    {
      "epoch": 0.5204165977847578,
      "grad_norm": 0.7261412739753723,
      "learning_rate": 0.00010160352845090875,
      "loss": 2.1346,
      "step": 2361
    },
    {
      "epoch": 0.5206370198930953,
      "grad_norm": 0.8056336641311646,
      "learning_rate": 0.00010153064662237259,
      "loss": 2.0735,
      "step": 2362
    },
    {
      "epoch": 0.5208574420014327,
      "grad_norm": 0.7445294857025146,
      "learning_rate": 0.00010145776398059381,
      "loss": 2.1952,
      "step": 2363
    },
    {
      "epoch": 0.5210778641097702,
      "grad_norm": 0.7453138828277588,
      "learning_rate": 0.00010138488056429545,
      "loss": 2.0509,
      "step": 2364
    },
    {
      "epoch": 0.5212982862181077,
      "grad_norm": 0.7898041605949402,
      "learning_rate": 0.00010131199641220095,
      "loss": 2.181,
      "step": 2365
    },
    {
      "epoch": 0.5215187083264451,
      "grad_norm": 0.6948133707046509,
      "learning_rate": 0.00010123911156303414,
      "loss": 2.1029,
      "step": 2366
    },
    {
      "epoch": 0.5217391304347826,
      "grad_norm": 0.7568143010139465,
      "learning_rate": 0.00010116622605551919,
      "loss": 2.0902,
      "step": 2367
    },
    {
      "epoch": 0.5219595525431201,
      "grad_norm": 0.7933834195137024,
      "learning_rate": 0.00010109333992838072,
      "loss": 2.1809,
      "step": 2368
    },
    {
      "epoch": 0.5221799746514575,
      "grad_norm": 0.7274755835533142,
      "learning_rate": 0.00010102045322034354,
      "loss": 2.063,
      "step": 2369
    },
    {
      "epoch": 0.522400396759795,
      "grad_norm": 0.7387127876281738,
      "learning_rate": 0.00010094756597013288,
      "loss": 2.1083,
      "step": 2370
    },
    {
      "epoch": 0.5226208188681325,
      "grad_norm": 0.7596807479858398,
      "learning_rate": 0.00010087467821647419,
      "loss": 2.0649,
      "step": 2371
    },
    {
      "epoch": 0.5228412409764699,
      "grad_norm": 0.8398569226264954,
      "learning_rate": 0.0001008017899980932,
      "loss": 2.0662,
      "step": 2372
    },
    {
      "epoch": 0.5230616630848074,
      "grad_norm": 0.718865692615509,
      "learning_rate": 0.00010072890135371591,
      "loss": 2.0369,
      "step": 2373
    },
    {
      "epoch": 0.5232820851931449,
      "grad_norm": 0.7491081953048706,
      "learning_rate": 0.00010065601232206856,
      "loss": 2.0984,
      "step": 2374
    },
    {
      "epoch": 0.5235025073014823,
      "grad_norm": 0.761888325214386,
      "learning_rate": 0.00010058312294187755,
      "loss": 2.0662,
      "step": 2375
    },
    {
      "epoch": 0.5237229294098198,
      "grad_norm": 0.7707873582839966,
      "learning_rate": 0.0001005102332518695,
      "loss": 2.0041,
      "step": 2376
    },
    {
      "epoch": 0.5239433515181573,
      "grad_norm": 0.7662652730941772,
      "learning_rate": 0.00010043734329077115,
      "loss": 2.0761,
      "step": 2377
    },
    {
      "epoch": 0.5241637736264947,
      "grad_norm": 0.7301992774009705,
      "learning_rate": 0.00010036445309730944,
      "loss": 2.2095,
      "step": 2378
    },
    {
      "epoch": 0.5243841957348322,
      "grad_norm": 0.7849206328392029,
      "learning_rate": 0.0001002915627102114,
      "loss": 2.0973,
      "step": 2379
    },
    {
      "epoch": 0.5246046178431697,
      "grad_norm": 0.7418490052223206,
      "learning_rate": 0.0001002186721682042,
      "loss": 2.0851,
      "step": 2380
    },
    {
      "epoch": 0.5248250399515071,
      "grad_norm": 0.7440804839134216,
      "learning_rate": 0.00010014578151001505,
      "loss": 1.9984,
      "step": 2381
    },
    {
      "epoch": 0.5250454620598446,
      "grad_norm": 0.7701501250267029,
      "learning_rate": 0.0001000728907743712,
      "loss": 2.0452,
      "step": 2382
    },
    {
      "epoch": 0.5252658841681821,
      "grad_norm": 0.7155986428260803,
      "learning_rate": 0.0001,
      "loss": 2.1206,
      "step": 2383
    },
    {
      "epoch": 0.5254863062765195,
      "grad_norm": 0.7232067584991455,
      "learning_rate": 9.992710922562882e-05,
      "loss": 2.0279,
      "step": 2384
    },
    {
      "epoch": 0.525706728384857,
      "grad_norm": 0.7734766602516174,
      "learning_rate": 9.985421848998497e-05,
      "loss": 2.1295,
      "step": 2385
    },
    {
      "epoch": 0.5259271504931945,
      "grad_norm": 0.7615062594413757,
      "learning_rate": 9.978132783179582e-05,
      "loss": 2.1009,
      "step": 2386
    },
    {
      "epoch": 0.5261475726015319,
      "grad_norm": 0.6790528893470764,
      "learning_rate": 9.970843728978861e-05,
      "loss": 2.2253,
      "step": 2387
    },
    {
      "epoch": 0.5263679947098694,
      "grad_norm": 0.7891032695770264,
      "learning_rate": 9.963554690269058e-05,
      "loss": 2.1234,
      "step": 2388
    },
    {
      "epoch": 0.5265884168182069,
      "grad_norm": 0.6787043213844299,
      "learning_rate": 9.956265670922887e-05,
      "loss": 1.9926,
      "step": 2389
    },
    {
      "epoch": 0.5268088389265443,
      "grad_norm": 0.7175703644752502,
      "learning_rate": 9.948976674813053e-05,
      "loss": 2.1432,
      "step": 2390
    },
    {
      "epoch": 0.5270292610348818,
      "grad_norm": 0.7546351552009583,
      "learning_rate": 9.941687705812247e-05,
      "loss": 2.0741,
      "step": 2391
    },
    {
      "epoch": 0.5272496831432193,
      "grad_norm": 0.7595307230949402,
      "learning_rate": 9.934398767793148e-05,
      "loss": 2.2466,
      "step": 2392
    },
    {
      "epoch": 0.5274701052515567,
      "grad_norm": 0.7148977518081665,
      "learning_rate": 9.927109864628413e-05,
      "loss": 2.0635,
      "step": 2393
    },
    {
      "epoch": 0.5276905273598942,
      "grad_norm": 0.7231335639953613,
      "learning_rate": 9.919821000190685e-05,
      "loss": 2.057,
      "step": 2394
    },
    {
      "epoch": 0.5279109494682317,
      "grad_norm": 0.8479407429695129,
      "learning_rate": 9.912532178352586e-05,
      "loss": 2.1954,
      "step": 2395
    },
    {
      "epoch": 0.5281313715765691,
      "grad_norm": 0.7424615621566772,
      "learning_rate": 9.905243402986715e-05,
      "loss": 1.989,
      "step": 2396
    },
    {
      "epoch": 0.5283517936849066,
      "grad_norm": 0.7145938873291016,
      "learning_rate": 9.897954677965647e-05,
      "loss": 2.0835,
      "step": 2397
    },
    {
      "epoch": 0.5285722157932441,
      "grad_norm": 0.7530732154846191,
      "learning_rate": 9.890666007161929e-05,
      "loss": 2.1158,
      "step": 2398
    },
    {
      "epoch": 0.5287926379015815,
      "grad_norm": 0.7778010964393616,
      "learning_rate": 9.883377394448079e-05,
      "loss": 2.0616,
      "step": 2399
    },
    {
      "epoch": 0.529013060009919,
      "grad_norm": 0.745461642742157,
      "learning_rate": 9.876088843696586e-05,
      "loss": 2.1374,
      "step": 2400
    },
    {
      "epoch": 0.529013060009919,
      "eval_bpb": 1.097041453312076,
      "eval_loss": 2.114431381225586,
      "eval_runtime": 184.3451,
      "eval_samples_per_second": 62.166,
      "eval_steps_per_second": 7.773,
      "step": 2400
    },
    {
      "epoch": 0.5292334821182565,
      "grad_norm": 0.7492656111717224,
      "learning_rate": 9.868800358779904e-05,
      "loss": 2.0349,
      "step": 2401
    },
    {
      "epoch": 0.5294539042265939,
      "grad_norm": 0.7588576078414917,
      "learning_rate": 9.861511943570458e-05,
      "loss": 2.0736,
      "step": 2402
    },
    {
      "epoch": 0.5296743263349314,
      "grad_norm": 0.7466787695884705,
      "learning_rate": 9.854223601940621e-05,
      "loss": 2.0129,
      "step": 2403
    },
    {
      "epoch": 0.5298947484432689,
      "grad_norm": 0.7237175107002258,
      "learning_rate": 9.846935337762743e-05,
      "loss": 2.1246,
      "step": 2404
    },
    {
      "epoch": 0.5301151705516063,
      "grad_norm": 0.700493335723877,
      "learning_rate": 9.839647154909126e-05,
      "loss": 1.9581,
      "step": 2405
    },
    {
      "epoch": 0.5303355926599438,
      "grad_norm": 0.7349368333816528,
      "learning_rate": 9.832359057252029e-05,
      "loss": 2.128,
      "step": 2406
    },
    {
      "epoch": 0.5305560147682813,
      "grad_norm": 0.691068708896637,
      "learning_rate": 9.825071048663669e-05,
      "loss": 1.9246,
      "step": 2407
    },
    {
      "epoch": 0.5307764368766187,
      "grad_norm": 0.755749523639679,
      "learning_rate": 9.817783133016206e-05,
      "loss": 2.1919,
      "step": 2408
    },
    {
      "epoch": 0.5309968589849562,
      "grad_norm": 0.7726244330406189,
      "learning_rate": 9.810495314181762e-05,
      "loss": 2.1939,
      "step": 2409
    },
    {
      "epoch": 0.5312172810932937,
      "grad_norm": 0.7548342347145081,
      "learning_rate": 9.803207596032404e-05,
      "loss": 2.1274,
      "step": 2410
    },
    {
      "epoch": 0.5314377032016311,
      "grad_norm": 0.7109224796295166,
      "learning_rate": 9.795919982440141e-05,
      "loss": 2.1243,
      "step": 2411
    },
    {
      "epoch": 0.5316581253099686,
      "grad_norm": 0.7492320537567139,
      "learning_rate": 9.788632477276934e-05,
      "loss": 2.0721,
      "step": 2412
    },
    {
      "epoch": 0.5318785474183061,
      "grad_norm": 0.7388356328010559,
      "learning_rate": 9.781345084414681e-05,
      "loss": 2.1682,
      "step": 2413
    },
    {
      "epoch": 0.5320989695266435,
      "grad_norm": 0.6332778334617615,
      "learning_rate": 9.77405780772522e-05,
      "loss": 2.0629,
      "step": 2414
    },
    {
      "epoch": 0.532319391634981,
      "grad_norm": 0.7349172234535217,
      "learning_rate": 9.76677065108033e-05,
      "loss": 2.0478,
      "step": 2415
    },
    {
      "epoch": 0.5325398137433185,
      "grad_norm": 0.7684071660041809,
      "learning_rate": 9.759483618351723e-05,
      "loss": 2.1329,
      "step": 2416
    },
    {
      "epoch": 0.5327602358516559,
      "grad_norm": 0.7337966561317444,
      "learning_rate": 9.752196713411053e-05,
      "loss": 2.1733,
      "step": 2417
    },
    {
      "epoch": 0.5329806579599934,
      "grad_norm": 0.7086362242698669,
      "learning_rate": 9.744909940129895e-05,
      "loss": 2.168,
      "step": 2418
    },
    {
      "epoch": 0.5332010800683309,
      "grad_norm": 0.640836775302887,
      "learning_rate": 9.737623302379764e-05,
      "loss": 2.1283,
      "step": 2419
    },
    {
      "epoch": 0.5334215021766683,
      "grad_norm": 0.7106740474700928,
      "learning_rate": 9.730336804032094e-05,
      "loss": 2.0656,
      "step": 2420
    },
    {
      "epoch": 0.5336419242850058,
      "grad_norm": 0.7549802660942078,
      "learning_rate": 9.723050448958252e-05,
      "loss": 2.1248,
      "step": 2421
    },
    {
      "epoch": 0.5338623463933433,
      "grad_norm": 0.7230193614959717,
      "learning_rate": 9.715764241029525e-05,
      "loss": 2.113,
      "step": 2422
    },
    {
      "epoch": 0.5340827685016807,
      "grad_norm": 0.7354143857955933,
      "learning_rate": 9.708478184117124e-05,
      "loss": 2.1964,
      "step": 2423
    },
    {
      "epoch": 0.5343031906100182,
      "grad_norm": 0.7463932633399963,
      "learning_rate": 9.701192282092178e-05,
      "loss": 2.1239,
      "step": 2424
    },
    {
      "epoch": 0.5345236127183557,
      "grad_norm": 0.7280251979827881,
      "learning_rate": 9.693906538825738e-05,
      "loss": 2.1175,
      "step": 2425
    },
    {
      "epoch": 0.5347440348266931,
      "grad_norm": 0.7743918299674988,
      "learning_rate": 9.686620958188763e-05,
      "loss": 2.0425,
      "step": 2426
    },
    {
      "epoch": 0.5349644569350306,
      "grad_norm": 0.6969741582870483,
      "learning_rate": 9.679335544052133e-05,
      "loss": 2.1197,
      "step": 2427
    },
    {
      "epoch": 0.5351848790433681,
      "grad_norm": 0.7510476112365723,
      "learning_rate": 9.672050300286636e-05,
      "loss": 2.0573,
      "step": 2428
    },
    {
      "epoch": 0.5354053011517055,
      "grad_norm": 0.7155231237411499,
      "learning_rate": 9.664765230762968e-05,
      "loss": 2.1111,
      "step": 2429
    },
    {
      "epoch": 0.535625723260043,
      "grad_norm": 0.7504848837852478,
      "learning_rate": 9.657480339351735e-05,
      "loss": 2.1488,
      "step": 2430
    },
    {
      "epoch": 0.5358461453683805,
      "grad_norm": 0.7709487676620483,
      "learning_rate": 9.650195629923448e-05,
      "loss": 2.154,
      "step": 2431
    },
    {
      "epoch": 0.5360665674767179,
      "grad_norm": 0.7641658782958984,
      "learning_rate": 9.642911106348519e-05,
      "loss": 2.1278,
      "step": 2432
    },
    {
      "epoch": 0.5362869895850554,
      "grad_norm": 0.7346528768539429,
      "learning_rate": 9.635626772497267e-05,
      "loss": 2.1184,
      "step": 2433
    },
    {
      "epoch": 0.5365074116933929,
      "grad_norm": 0.7252072691917419,
      "learning_rate": 9.628342632239907e-05,
      "loss": 2.1134,
      "step": 2434
    },
    {
      "epoch": 0.5367278338017303,
      "grad_norm": 0.7970731258392334,
      "learning_rate": 9.621058689446544e-05,
      "loss": 2.2419,
      "step": 2435
    },
    {
      "epoch": 0.5369482559100678,
      "grad_norm": 0.7952181696891785,
      "learning_rate": 9.613774947987189e-05,
      "loss": 2.0737,
      "step": 2436
    },
    {
      "epoch": 0.5371686780184053,
      "grad_norm": 0.878703773021698,
      "learning_rate": 9.606491411731741e-05,
      "loss": 2.1486,
      "step": 2437
    },
    {
      "epoch": 0.5373891001267427,
      "grad_norm": 0.7779567837715149,
      "learning_rate": 9.599208084549993e-05,
      "loss": 2.0509,
      "step": 2438
    },
    {
      "epoch": 0.5376095222350802,
      "grad_norm": 0.7076535224914551,
      "learning_rate": 9.591924970311622e-05,
      "loss": 2.0001,
      "step": 2439
    },
    {
      "epoch": 0.5378299443434177,
      "grad_norm": 0.7524128556251526,
      "learning_rate": 9.584642072886196e-05,
      "loss": 2.1408,
      "step": 2440
    },
    {
      "epoch": 0.5380503664517551,
      "grad_norm": 0.7417291402816772,
      "learning_rate": 9.577359396143165e-05,
      "loss": 2.1224,
      "step": 2441
    },
    {
      "epoch": 0.5382707885600926,
      "grad_norm": 0.7389922738075256,
      "learning_rate": 9.570076943951862e-05,
      "loss": 2.0254,
      "step": 2442
    },
    {
      "epoch": 0.5384912106684301,
      "grad_norm": 0.7647872567176819,
      "learning_rate": 9.562794720181504e-05,
      "loss": 2.0922,
      "step": 2443
    },
    {
      "epoch": 0.5387116327767675,
      "grad_norm": 0.741159975528717,
      "learning_rate": 9.555512728701183e-05,
      "loss": 2.0759,
      "step": 2444
    },
    {
      "epoch": 0.538932054885105,
      "grad_norm": 0.6893795728683472,
      "learning_rate": 9.548230973379869e-05,
      "loss": 2.1217,
      "step": 2445
    },
    {
      "epoch": 0.5391524769934425,
      "grad_norm": 0.7472835183143616,
      "learning_rate": 9.540949458086409e-05,
      "loss": 2.0422,
      "step": 2446
    },
    {
      "epoch": 0.5393728991017799,
      "grad_norm": 0.7019094228744507,
      "learning_rate": 9.533668186689515e-05,
      "loss": 2.171,
      "step": 2447
    },
    {
      "epoch": 0.5395933212101174,
      "grad_norm": 0.7536553144454956,
      "learning_rate": 9.526387163057777e-05,
      "loss": 2.117,
      "step": 2448
    },
    {
      "epoch": 0.5398137433184549,
      "grad_norm": 0.7248945832252502,
      "learning_rate": 9.519106391059652e-05,
      "loss": 1.8944,
      "step": 2449
    },
    {
      "epoch": 0.5400341654267923,
      "grad_norm": 0.7478938698768616,
      "learning_rate": 9.511825874563457e-05,
      "loss": 1.9637,
      "step": 2450
    },
    {
      "epoch": 0.5402545875351298,
      "grad_norm": 0.7088088393211365,
      "learning_rate": 9.504545617437385e-05,
      "loss": 2.0988,
      "step": 2451
    },
    {
      "epoch": 0.5404750096434673,
      "grad_norm": 0.7106070518493652,
      "learning_rate": 9.49726562354948e-05,
      "loss": 2.0049,
      "step": 2452
    },
    {
      "epoch": 0.5406954317518047,
      "grad_norm": 0.768764853477478,
      "learning_rate": 9.489985896767653e-05,
      "loss": 2.2683,
      "step": 2453
    },
    {
      "epoch": 0.5409158538601422,
      "grad_norm": 0.7273726463317871,
      "learning_rate": 9.48270644095967e-05,
      "loss": 2.0138,
      "step": 2454
    },
    {
      "epoch": 0.5411362759684797,
      "grad_norm": 0.7017658948898315,
      "learning_rate": 9.475427259993154e-05,
      "loss": 2.1689,
      "step": 2455
    },
    {
      "epoch": 0.5413566980768171,
      "grad_norm": 0.7411829233169556,
      "learning_rate": 9.46814835773558e-05,
      "loss": 2.1072,
      "step": 2456
    },
    {
      "epoch": 0.5415771201851546,
      "grad_norm": 0.7827083468437195,
      "learning_rate": 9.460869738054276e-05,
      "loss": 2.1273,
      "step": 2457
    },
    {
      "epoch": 0.5417975422934921,
      "grad_norm": 0.713304877281189,
      "learning_rate": 9.453591404816423e-05,
      "loss": 2.0323,
      "step": 2458
    },
    {
      "epoch": 0.5420179644018295,
      "grad_norm": 0.7656096816062927,
      "learning_rate": 9.446313361889045e-05,
      "loss": 2.0936,
      "step": 2459
    },
    {
      "epoch": 0.542238386510167,
      "grad_norm": 0.7407435178756714,
      "learning_rate": 9.439035613139016e-05,
      "loss": 2.1789,
      "step": 2460
    },
    {
      "epoch": 0.5424588086185045,
      "grad_norm": 0.7380139231681824,
      "learning_rate": 9.431758162433054e-05,
      "loss": 2.0934,
      "step": 2461
    },
    {
      "epoch": 0.5426792307268419,
      "grad_norm": 0.7368414998054504,
      "learning_rate": 9.42448101363771e-05,
      "loss": 2.0941,
      "step": 2462
    },
    {
      "epoch": 0.5428996528351794,
      "grad_norm": 0.8122841715812683,
      "learning_rate": 9.417204170619384e-05,
      "loss": 2.0587,
      "step": 2463
    },
    {
      "epoch": 0.5431200749435169,
      "grad_norm": 0.6961782574653625,
      "learning_rate": 9.409927637244313e-05,
      "loss": 1.9902,
      "step": 2464
    },
    {
      "epoch": 0.5433404970518543,
      "grad_norm": 0.6991739273071289,
      "learning_rate": 9.402651417378563e-05,
      "loss": 2.1754,
      "step": 2465
    },
    {
      "epoch": 0.5435609191601918,
      "grad_norm": 0.7423298954963684,
      "learning_rate": 9.39537551488804e-05,
      "loss": 2.153,
      "step": 2466
    },
    {
      "epoch": 0.5437813412685293,
      "grad_norm": 0.7656598091125488,
      "learning_rate": 9.388099933638479e-05,
      "loss": 2.2254,
      "step": 2467
    },
    {
      "epoch": 0.5440017633768667,
      "grad_norm": 0.7766461968421936,
      "learning_rate": 9.380824677495441e-05,
      "loss": 2.2017,
      "step": 2468
    },
    {
      "epoch": 0.5442221854852042,
      "grad_norm": 0.7797102928161621,
      "learning_rate": 9.37354975032432e-05,
      "loss": 2.026,
      "step": 2469
    },
    {
      "epoch": 0.5444426075935417,
      "grad_norm": 0.7748450636863708,
      "learning_rate": 9.366275155990331e-05,
      "loss": 2.0993,
      "step": 2470
    },
    {
      "epoch": 0.5446630297018791,
      "grad_norm": 0.7835527658462524,
      "learning_rate": 9.359000898358517e-05,
      "loss": 2.2588,
      "step": 2471
    },
    {
      "epoch": 0.5448834518102166,
      "grad_norm": 0.7929511070251465,
      "learning_rate": 9.351726981293735e-05,
      "loss": 2.1169,
      "step": 2472
    },
    {
      "epoch": 0.5451038739185541,
      "grad_norm": 0.7144936323165894,
      "learning_rate": 9.34445340866067e-05,
      "loss": 2.0949,
      "step": 2473
    },
    {
      "epoch": 0.5453242960268915,
      "grad_norm": 0.729508101940155,
      "learning_rate": 9.337180184323813e-05,
      "loss": 2.1528,
      "step": 2474
    },
    {
      "epoch": 0.545544718135229,
      "grad_norm": 0.6976228952407837,
      "learning_rate": 9.329907312147478e-05,
      "loss": 2.1122,
      "step": 2475
    },
    {
      "epoch": 0.5457651402435665,
      "grad_norm": 0.749882161617279,
      "learning_rate": 9.32263479599579e-05,
      "loss": 2.0516,
      "step": 2476
    },
    {
      "epoch": 0.5459855623519039,
      "grad_norm": 0.7147037982940674,
      "learning_rate": 9.315362639732684e-05,
      "loss": 2.1114,
      "step": 2477
    },
    {
      "epoch": 0.5462059844602414,
      "grad_norm": 0.6739283800125122,
      "learning_rate": 9.308090847221905e-05,
      "loss": 2.0303,
      "step": 2478
    },
    {
      "epoch": 0.5464264065685789,
      "grad_norm": 0.7590286731719971,
      "learning_rate": 9.300819422327005e-05,
      "loss": 2.0982,
      "step": 2479
    },
    {
      "epoch": 0.5466468286769163,
      "grad_norm": 0.7487385272979736,
      "learning_rate": 9.29354836891134e-05,
      "loss": 2.1748,
      "step": 2480
    },
    {
      "epoch": 0.5468672507852538,
      "grad_norm": 0.7135498523712158,
      "learning_rate": 9.286277690838068e-05,
      "loss": 2.1085,
      "step": 2481
    },
    {
      "epoch": 0.5470876728935913,
      "grad_norm": 0.765966534614563,
      "learning_rate": 9.279007391970148e-05,
      "loss": 2.044,
      "step": 2482
    },
    {
      "epoch": 0.5473080950019287,
      "grad_norm": 0.7785701751708984,
      "learning_rate": 9.271737476170337e-05,
      "loss": 2.073,
      "step": 2483
    },
    {
      "epoch": 0.5475285171102662,
      "grad_norm": 0.706071674823761,
      "learning_rate": 9.264467947301187e-05,
      "loss": 2.2237,
      "step": 2484
    },
    {
      "epoch": 0.5477489392186037,
      "grad_norm": 0.7369053363800049,
      "learning_rate": 9.257198809225051e-05,
      "loss": 2.0919,
      "step": 2485
    },
    {
      "epoch": 0.5479693613269411,
      "grad_norm": 0.7222207188606262,
      "learning_rate": 9.249930065804066e-05,
      "loss": 2.1467,
      "step": 2486
    },
    {
      "epoch": 0.5481897834352786,
      "grad_norm": 0.6981198787689209,
      "learning_rate": 9.242661720900164e-05,
      "loss": 2.0617,
      "step": 2487
    },
    {
      "epoch": 0.548410205543616,
      "grad_norm": 0.7341172695159912,
      "learning_rate": 9.235393778375068e-05,
      "loss": 1.9353,
      "step": 2488
    },
    {
      "epoch": 0.5486306276519535,
      "grad_norm": 0.726080596446991,
      "learning_rate": 9.228126242090278e-05,
      "loss": 2.1252,
      "step": 2489
    },
    {
      "epoch": 0.548851049760291,
      "grad_norm": 0.7581496238708496,
      "learning_rate": 9.220859115907087e-05,
      "loss": 2.2197,
      "step": 2490
    },
    {
      "epoch": 0.5490714718686284,
      "grad_norm": 0.7511199116706848,
      "learning_rate": 9.213592403686566e-05,
      "loss": 2.095,
      "step": 2491
    },
    {
      "epoch": 0.5492918939769659,
      "grad_norm": 0.7089020609855652,
      "learning_rate": 9.206326109289567e-05,
      "loss": 2.1794,
      "step": 2492
    },
    {
      "epoch": 0.5495123160853034,
      "grad_norm": 0.7732926607131958,
      "learning_rate": 9.19906023657672e-05,
      "loss": 2.2316,
      "step": 2493
    },
    {
      "epoch": 0.5497327381936408,
      "grad_norm": 0.7595569491386414,
      "learning_rate": 9.191794789408433e-05,
      "loss": 2.0908,
      "step": 2494
    },
    {
      "epoch": 0.5499531603019783,
      "grad_norm": 0.6933780908584595,
      "learning_rate": 9.184529771644882e-05,
      "loss": 2.1544,
      "step": 2495
    },
    {
      "epoch": 0.5501735824103158,
      "grad_norm": 0.7363986968994141,
      "learning_rate": 9.17726518714602e-05,
      "loss": 2.19,
      "step": 2496
    },
    {
      "epoch": 0.5503940045186532,
      "grad_norm": 0.727321982383728,
      "learning_rate": 9.17000103977157e-05,
      "loss": 2.2105,
      "step": 2497
    },
    {
      "epoch": 0.5506144266269907,
      "grad_norm": 0.7555034160614014,
      "learning_rate": 9.162737333381019e-05,
      "loss": 2.0309,
      "step": 2498
    },
    {
      "epoch": 0.5508348487353282,
      "grad_norm": 0.7890347242355347,
      "learning_rate": 9.155474071833623e-05,
      "loss": 2.1822,
      "step": 2499
    },
    {
      "epoch": 0.5510552708436656,
      "grad_norm": 0.7221497297286987,
      "learning_rate": 9.148211258988403e-05,
      "loss": 2.0952,
      "step": 2500
    },
    {
      "epoch": 0.5512756929520031,
      "grad_norm": 0.7150618433952332,
      "learning_rate": 9.140948898704137e-05,
      "loss": 2.1527,
      "step": 2501
    },
    {
      "epoch": 0.5514961150603406,
      "grad_norm": 0.7122634649276733,
      "learning_rate": 9.133686994839362e-05,
      "loss": 2.2042,
      "step": 2502
    },
    {
      "epoch": 0.551716537168678,
      "grad_norm": 0.7129616737365723,
      "learning_rate": 9.126425551252376e-05,
      "loss": 2.1693,
      "step": 2503
    },
    {
      "epoch": 0.5519369592770155,
      "grad_norm": 0.8008817434310913,
      "learning_rate": 9.119164571801236e-05,
      "loss": 2.123,
      "step": 2504
    },
    {
      "epoch": 0.552157381385353,
      "grad_norm": 0.7142249941825867,
      "learning_rate": 9.111904060343746e-05,
      "loss": 2.2407,
      "step": 2505
    },
    {
      "epoch": 0.5523778034936904,
      "grad_norm": 0.6844157576560974,
      "learning_rate": 9.104644020737463e-05,
      "loss": 2.0831,
      "step": 2506
    },
    {
      "epoch": 0.5525982256020279,
      "grad_norm": 0.7909813523292542,
      "learning_rate": 9.097384456839693e-05,
      "loss": 2.0165,
      "step": 2507
    },
    {
      "epoch": 0.5528186477103654,
      "grad_norm": 0.7435783743858337,
      "learning_rate": 9.090125372507492e-05,
      "loss": 2.1805,
      "step": 2508
    },
    {
      "epoch": 0.5530390698187028,
      "grad_norm": 0.7177037596702576,
      "learning_rate": 9.082866771597659e-05,
      "loss": 2.0417,
      "step": 2509
    },
    {
      "epoch": 0.5532594919270403,
      "grad_norm": 0.7721659541130066,
      "learning_rate": 9.075608657966733e-05,
      "loss": 2.139,
      "step": 2510
    },
    {
      "epoch": 0.5534799140353778,
      "grad_norm": 0.7061505913734436,
      "learning_rate": 9.068351035471e-05,
      "loss": 1.9975,
      "step": 2511
    },
    {
      "epoch": 0.5537003361437152,
      "grad_norm": 0.7638243436813354,
      "learning_rate": 9.061093907966482e-05,
      "loss": 2.147,
      "step": 2512
    },
    {
      "epoch": 0.5539207582520527,
      "grad_norm": 0.7193301916122437,
      "learning_rate": 9.05383727930894e-05,
      "loss": 2.0824,
      "step": 2513
    },
    {
      "epoch": 0.5541411803603902,
      "grad_norm": 0.7116189002990723,
      "learning_rate": 9.046581153353866e-05,
      "loss": 1.969,
      "step": 2514
    },
    {
      "epoch": 0.5543616024687276,
      "grad_norm": 0.7310522794723511,
      "learning_rate": 9.039325533956491e-05,
      "loss": 1.9733,
      "step": 2515
    },
    {
      "epoch": 0.5545820245770651,
      "grad_norm": 0.7326749563217163,
      "learning_rate": 9.03207042497177e-05,
      "loss": 2.2003,
      "step": 2516
    },
    {
      "epoch": 0.5548024466854026,
      "grad_norm": 0.6912440061569214,
      "learning_rate": 9.024815830254388e-05,
      "loss": 2.136,
      "step": 2517
    },
    {
      "epoch": 0.55502286879374,
      "grad_norm": 0.7075268030166626,
      "learning_rate": 9.017561753658764e-05,
      "loss": 2.1263,
      "step": 2518
    },
    {
      "epoch": 0.5552432909020775,
      "grad_norm": 0.7587385773658752,
      "learning_rate": 9.010308199039034e-05,
      "loss": 2.1838,
      "step": 2519
    },
    {
      "epoch": 0.555463713010415,
      "grad_norm": 0.7844798564910889,
      "learning_rate": 9.00305517024906e-05,
      "loss": 2.16,
      "step": 2520
    },
    {
      "epoch": 0.5556841351187524,
      "grad_norm": 0.7841442823410034,
      "learning_rate": 8.995802671142423e-05,
      "loss": 2.1409,
      "step": 2521
    },
    {
      "epoch": 0.5559045572270899,
      "grad_norm": 0.7968512773513794,
      "learning_rate": 8.988550705572426e-05,
      "loss": 2.0172,
      "step": 2522
    },
    {
      "epoch": 0.5561249793354274,
      "grad_norm": 0.7370349764823914,
      "learning_rate": 8.98129927739208e-05,
      "loss": 2.0736,
      "step": 2523
    },
    {
      "epoch": 0.5563454014437648,
      "grad_norm": 0.747785210609436,
      "learning_rate": 8.974048390454121e-05,
      "loss": 2.0134,
      "step": 2524
    },
    {
      "epoch": 0.5565658235521023,
      "grad_norm": 0.8101722002029419,
      "learning_rate": 8.966798048610994e-05,
      "loss": 2.0564,
      "step": 2525
    },
    {
      "epoch": 0.5567862456604398,
      "grad_norm": 0.7874711751937866,
      "learning_rate": 8.959548255714851e-05,
      "loss": 2.0385,
      "step": 2526
    },
    {
      "epoch": 0.5570066677687772,
      "grad_norm": 0.7428283095359802,
      "learning_rate": 8.952299015617554e-05,
      "loss": 2.133,
      "step": 2527
    },
    {
      "epoch": 0.5572270898771147,
      "grad_norm": 0.6968439817428589,
      "learning_rate": 8.945050332170672e-05,
      "loss": 2.0545,
      "step": 2528
    },
    {
      "epoch": 0.5574475119854522,
      "grad_norm": 0.7829073071479797,
      "learning_rate": 8.93780220922548e-05,
      "loss": 2.1362,
      "step": 2529
    },
    {
      "epoch": 0.5576679340937896,
      "grad_norm": 0.7195518016815186,
      "learning_rate": 8.93055465063295e-05,
      "loss": 2.1711,
      "step": 2530
    },
    {
      "epoch": 0.5578883562021271,
      "grad_norm": 0.8129485845565796,
      "learning_rate": 8.923307660243761e-05,
      "loss": 2.0578,
      "step": 2531
    },
    {
      "epoch": 0.5581087783104646,
      "grad_norm": 0.7411863803863525,
      "learning_rate": 8.916061241908287e-05,
      "loss": 2.1039,
      "step": 2532
    },
    {
      "epoch": 0.558329200418802,
      "grad_norm": 0.7586485147476196,
      "learning_rate": 8.908815399476594e-05,
      "loss": 2.0642,
      "step": 2533
    },
    {
      "epoch": 0.5585496225271395,
      "grad_norm": 0.744561493396759,
      "learning_rate": 8.901570136798448e-05,
      "loss": 2.0735,
      "step": 2534
    },
    {
      "epoch": 0.558770044635477,
      "grad_norm": 0.7438688278198242,
      "learning_rate": 8.894325457723307e-05,
      "loss": 2.0485,
      "step": 2535
    },
    {
      "epoch": 0.5589904667438144,
      "grad_norm": 0.7886449098587036,
      "learning_rate": 8.887081366100312e-05,
      "loss": 2.1552,
      "step": 2536
    },
    {
      "epoch": 0.5592108888521519,
      "grad_norm": 0.7179954648017883,
      "learning_rate": 8.879837865778298e-05,
      "loss": 2.1946,
      "step": 2537
    },
    {
      "epoch": 0.5594313109604894,
      "grad_norm": 0.7174047231674194,
      "learning_rate": 8.872594960605785e-05,
      "loss": 2.1696,
      "step": 2538
    },
    {
      "epoch": 0.5596517330688268,
      "grad_norm": 0.7249564528465271,
      "learning_rate": 8.865352654430977e-05,
      "loss": 2.115,
      "step": 2539
    },
    {
      "epoch": 0.5598721551771643,
      "grad_norm": 0.7426702380180359,
      "learning_rate": 8.858110951101756e-05,
      "loss": 2.0595,
      "step": 2540
    },
    {
      "epoch": 0.5600925772855018,
      "grad_norm": 0.7878005504608154,
      "learning_rate": 8.85086985446569e-05,
      "loss": 2.1308,
      "step": 2541
    },
    {
      "epoch": 0.5603129993938392,
      "grad_norm": 0.752798855304718,
      "learning_rate": 8.84362936837002e-05,
      "loss": 2.0729,
      "step": 2542
    },
    {
      "epoch": 0.5605334215021767,
      "grad_norm": 0.7631442546844482,
      "learning_rate": 8.836389496661661e-05,
      "loss": 2.0498,
      "step": 2543
    },
    {
      "epoch": 0.5607538436105142,
      "grad_norm": 0.7307388186454773,
      "learning_rate": 8.829150243187208e-05,
      "loss": 2.0786,
      "step": 2544
    },
    {
      "epoch": 0.5609742657188516,
      "grad_norm": 0.7418521046638489,
      "learning_rate": 8.821911611792921e-05,
      "loss": 2.22,
      "step": 2545
    },
    {
      "epoch": 0.5611946878271891,
      "grad_norm": 0.7297216653823853,
      "learning_rate": 8.814673606324735e-05,
      "loss": 2.1756,
      "step": 2546
    },
    {
      "epoch": 0.5614151099355266,
      "grad_norm": 0.6803293824195862,
      "learning_rate": 8.80743623062825e-05,
      "loss": 2.1683,
      "step": 2547
    },
    {
      "epoch": 0.561635532043864,
      "grad_norm": 0.7323028445243835,
      "learning_rate": 8.80019948854873e-05,
      "loss": 2.0645,
      "step": 2548
    },
    {
      "epoch": 0.5618559541522015,
      "grad_norm": 0.7214212417602539,
      "learning_rate": 8.792963383931101e-05,
      "loss": 2.2275,
      "step": 2549
    },
    {
      "epoch": 0.562076376260539,
      "grad_norm": 0.7199147939682007,
      "learning_rate": 8.785727920619957e-05,
      "loss": 2.2226,
      "step": 2550
    },
    {
      "epoch": 0.5622967983688764,
      "grad_norm": 0.7055409550666809,
      "learning_rate": 8.778493102459546e-05,
      "loss": 2.1144,
      "step": 2551
    },
    {
      "epoch": 0.5625172204772139,
      "grad_norm": 0.7668905854225159,
      "learning_rate": 8.771258933293775e-05,
      "loss": 2.1631,
      "step": 2552
    },
    {
      "epoch": 0.5627376425855514,
      "grad_norm": 0.7345185875892639,
      "learning_rate": 8.764025416966205e-05,
      "loss": 2.1107,
      "step": 2553
    },
    {
      "epoch": 0.5629580646938888,
      "grad_norm": 0.7094461917877197,
      "learning_rate": 8.75679255732005e-05,
      "loss": 2.0107,
      "step": 2554
    },
    {
      "epoch": 0.5631784868022263,
      "grad_norm": 0.707360565662384,
      "learning_rate": 8.749560358198176e-05,
      "loss": 2.2158,
      "step": 2555
    },
    {
      "epoch": 0.5633989089105638,
      "grad_norm": 0.7104003429412842,
      "learning_rate": 8.742328823443093e-05,
      "loss": 2.1316,
      "step": 2556
    },
    {
      "epoch": 0.5636193310189012,
      "grad_norm": 0.7723707556724548,
      "learning_rate": 8.735097956896977e-05,
      "loss": 2.0465,
      "step": 2557
    },
    {
      "epoch": 0.5638397531272387,
      "grad_norm": 0.7682229280471802,
      "learning_rate": 8.727867762401623e-05,
      "loss": 2.1876,
      "step": 2558
    },
    {
      "epoch": 0.5640601752355762,
      "grad_norm": 0.7033430933952332,
      "learning_rate": 8.720638243798484e-05,
      "loss": 2.0855,
      "step": 2559
    },
    {
      "epoch": 0.5642805973439136,
      "grad_norm": 0.7451592087745667,
      "learning_rate": 8.71340940492865e-05,
      "loss": 2.1778,
      "step": 2560
    },
    {
      "epoch": 0.5645010194522511,
      "grad_norm": 0.7498093843460083,
      "learning_rate": 8.706181249632852e-05,
      "loss": 2.1997,
      "step": 2561
    },
    {
      "epoch": 0.5647214415605886,
      "grad_norm": 0.7052710056304932,
      "learning_rate": 8.698953781751458e-05,
      "loss": 2.1792,
      "step": 2562
    },
    {
      "epoch": 0.564941863668926,
      "grad_norm": 0.7348800897598267,
      "learning_rate": 8.691727005124464e-05,
      "loss": 2.1745,
      "step": 2563
    },
    {
      "epoch": 0.5651622857772635,
      "grad_norm": 0.701416552066803,
      "learning_rate": 8.684500923591508e-05,
      "loss": 2.0479,
      "step": 2564
    },
    {
      "epoch": 0.565382707885601,
      "grad_norm": 0.715344250202179,
      "learning_rate": 8.677275540991851e-05,
      "loss": 2.0568,
      "step": 2565
    },
    {
      "epoch": 0.5656031299939384,
      "grad_norm": 0.7342721819877625,
      "learning_rate": 8.670050861164391e-05,
      "loss": 2.0171,
      "step": 2566
    },
    {
      "epoch": 0.5658235521022759,
      "grad_norm": 0.7492346167564392,
      "learning_rate": 8.662826887947643e-05,
      "loss": 2.0693,
      "step": 2567
    },
    {
      "epoch": 0.5660439742106134,
      "grad_norm": 0.7454676628112793,
      "learning_rate": 8.655603625179759e-05,
      "loss": 2.2054,
      "step": 2568
    },
    {
      "epoch": 0.5662643963189508,
      "grad_norm": 0.74738609790802,
      "learning_rate": 8.648381076698498e-05,
      "loss": 2.1522,
      "step": 2569
    },
    {
      "epoch": 0.5664848184272883,
      "grad_norm": 0.7795509099960327,
      "learning_rate": 8.641159246341249e-05,
      "loss": 2.0977,
      "step": 2570
    },
    {
      "epoch": 0.5667052405356258,
      "grad_norm": 0.797263503074646,
      "learning_rate": 8.633938137945018e-05,
      "loss": 2.1128,
      "step": 2571
    },
    {
      "epoch": 0.5669256626439632,
      "grad_norm": 0.7227104902267456,
      "learning_rate": 8.626717755346431e-05,
      "loss": 2.2695,
      "step": 2572
    },
    {
      "epoch": 0.5671460847523007,
      "grad_norm": 0.716749906539917,
      "learning_rate": 8.619498102381722e-05,
      "loss": 2.1236,
      "step": 2573
    },
    {
      "epoch": 0.5673665068606382,
      "grad_norm": 0.7079432010650635,
      "learning_rate": 8.612279182886742e-05,
      "loss": 2.2032,
      "step": 2574
    },
    {
      "epoch": 0.5675869289689756,
      "grad_norm": 0.7057801485061646,
      "learning_rate": 8.605061000696946e-05,
      "loss": 2.0308,
      "step": 2575
    },
    {
      "epoch": 0.5678073510773131,
      "grad_norm": 0.7675472497940063,
      "learning_rate": 8.597843559647405e-05,
      "loss": 2.3242,
      "step": 2576
    },
    {
      "epoch": 0.5680277731856506,
      "grad_norm": 0.8048058748245239,
      "learning_rate": 8.590626863572792e-05,
      "loss": 2.1649,
      "step": 2577
    },
    {
      "epoch": 0.568248195293988,
      "grad_norm": 0.7290545701980591,
      "learning_rate": 8.583410916307386e-05,
      "loss": 2.091,
      "step": 2578
    },
    {
      "epoch": 0.5684686174023255,
      "grad_norm": 0.732778787612915,
      "learning_rate": 8.576195721685066e-05,
      "loss": 2.2247,
      "step": 2579
    },
    {
      "epoch": 0.568689039510663,
      "grad_norm": 0.762588381767273,
      "learning_rate": 8.568981283539312e-05,
      "loss": 2.3379,
      "step": 2580
    },
    {
      "epoch": 0.5689094616190004,
      "grad_norm": 0.7231842875480652,
      "learning_rate": 8.561767605703204e-05,
      "loss": 2.0053,
      "step": 2581
    },
    {
      "epoch": 0.5691298837273379,
      "grad_norm": 0.737369179725647,
      "learning_rate": 8.554554692009414e-05,
      "loss": 2.1387,
      "step": 2582
    },
    {
      "epoch": 0.5693503058356754,
      "grad_norm": 0.7455813884735107,
      "learning_rate": 8.54734254629021e-05,
      "loss": 2.1182,
      "step": 2583
    },
    {
      "epoch": 0.5695707279440128,
      "grad_norm": 0.7371118068695068,
      "learning_rate": 8.540131172377456e-05,
      "loss": 2.175,
      "step": 2584
    },
    {
      "epoch": 0.5697911500523503,
      "grad_norm": 0.7807168960571289,
      "learning_rate": 8.532920574102602e-05,
      "loss": 2.1193,
      "step": 2585
    },
    {
      "epoch": 0.5700115721606878,
      "grad_norm": 0.6769904494285583,
      "learning_rate": 8.525710755296684e-05,
      "loss": 2.1077,
      "step": 2586
    },
    {
      "epoch": 0.5702319942690252,
      "grad_norm": 0.7558202743530273,
      "learning_rate": 8.518501719790328e-05,
      "loss": 2.2223,
      "step": 2587
    },
    {
      "epoch": 0.5704524163773627,
      "grad_norm": 0.7260102033615112,
      "learning_rate": 8.51129347141374e-05,
      "loss": 2.0678,
      "step": 2588
    },
    {
      "epoch": 0.5706728384857002,
      "grad_norm": 0.6764949560165405,
      "learning_rate": 8.504086013996712e-05,
      "loss": 2.0011,
      "step": 2589
    },
    {
      "epoch": 0.5708932605940376,
      "grad_norm": 0.7435042262077332,
      "learning_rate": 8.496879351368609e-05,
      "loss": 2.1015,
      "step": 2590
    },
    {
      "epoch": 0.5711136827023751,
      "grad_norm": 0.7744899392127991,
      "learning_rate": 8.489673487358383e-05,
      "loss": 2.0454,
      "step": 2591
    },
    {
      "epoch": 0.5713341048107126,
      "grad_norm": 0.7426540851593018,
      "learning_rate": 8.482468425794555e-05,
      "loss": 2.2159,
      "step": 2592
    },
    {
      "epoch": 0.57155452691905,
      "grad_norm": 0.6824113130569458,
      "learning_rate": 8.475264170505222e-05,
      "loss": 2.0842,
      "step": 2593
    },
    {
      "epoch": 0.5717749490273875,
      "grad_norm": 0.7572205066680908,
      "learning_rate": 8.468060725318052e-05,
      "loss": 2.1331,
      "step": 2594
    },
    {
      "epoch": 0.571995371135725,
      "grad_norm": 0.7153971195220947,
      "learning_rate": 8.460858094060282e-05,
      "loss": 2.0581,
      "step": 2595
    },
    {
      "epoch": 0.5722157932440624,
      "grad_norm": 0.7130942344665527,
      "learning_rate": 8.453656280558719e-05,
      "loss": 2.0433,
      "step": 2596
    },
    {
      "epoch": 0.5724362153523999,
      "grad_norm": 0.7097151875495911,
      "learning_rate": 8.44645528863973e-05,
      "loss": 2.1628,
      "step": 2597
    },
    {
      "epoch": 0.5726566374607374,
      "grad_norm": 0.6633721590042114,
      "learning_rate": 8.439255122129254e-05,
      "loss": 1.9627,
      "step": 2598
    },
    {
      "epoch": 0.5728770595690748,
      "grad_norm": 0.6983052492141724,
      "learning_rate": 8.432055784852785e-05,
      "loss": 2.0515,
      "step": 2599
    },
    {
      "epoch": 0.5730974816774123,
      "grad_norm": 0.7122889161109924,
      "learning_rate": 8.424857280635376e-05,
      "loss": 2.2102,
      "step": 2600
    },
    {
      "epoch": 0.5733179037857498,
      "grad_norm": 0.6825004816055298,
      "learning_rate": 8.417659613301646e-05,
      "loss": 2.0006,
      "step": 2601
    },
    {
      "epoch": 0.5735383258940872,
      "grad_norm": 0.7185415029525757,
      "learning_rate": 8.410462786675755e-05,
      "loss": 2.1845,
      "step": 2602
    },
    {
      "epoch": 0.5737587480024247,
      "grad_norm": 0.6819283366203308,
      "learning_rate": 8.403266804581429e-05,
      "loss": 2.1952,
      "step": 2603
    },
    {
      "epoch": 0.5739791701107622,
      "grad_norm": 0.7213916182518005,
      "learning_rate": 8.396071670841937e-05,
      "loss": 2.0788,
      "step": 2604
    },
    {
      "epoch": 0.5741995922190996,
      "grad_norm": 0.7240177989006042,
      "learning_rate": 8.388877389280106e-05,
      "loss": 2.0442,
      "step": 2605
    },
    {
      "epoch": 0.5744200143274371,
      "grad_norm": 0.7735614776611328,
      "learning_rate": 8.3816839637183e-05,
      "loss": 2.1148,
      "step": 2606
    },
    {
      "epoch": 0.5746404364357746,
      "grad_norm": 0.7171289324760437,
      "learning_rate": 8.374491397978438e-05,
      "loss": 2.0422,
      "step": 2607
    },
    {
      "epoch": 0.574860858544112,
      "grad_norm": 0.7875083684921265,
      "learning_rate": 8.367299695881973e-05,
      "loss": 2.1969,
      "step": 2608
    },
    {
      "epoch": 0.5750812806524495,
      "grad_norm": 0.7620639204978943,
      "learning_rate": 8.360108861249905e-05,
      "loss": 2.0881,
      "step": 2609
    },
    {
      "epoch": 0.575301702760787,
      "grad_norm": 0.7313951849937439,
      "learning_rate": 8.352918897902772e-05,
      "loss": 2.0615,
      "step": 2610
    },
    {
      "epoch": 0.5755221248691244,
      "grad_norm": 0.6929225921630859,
      "learning_rate": 8.345729809660649e-05,
      "loss": 2.1104,
      "step": 2611
    },
    {
      "epoch": 0.5757425469774619,
      "grad_norm": 0.7808670401573181,
      "learning_rate": 8.338541600343147e-05,
      "loss": 2.2671,
      "step": 2612
    },
    {
      "epoch": 0.5759629690857994,
      "grad_norm": 0.7745012640953064,
      "learning_rate": 8.331354273769407e-05,
      "loss": 2.2114,
      "step": 2613
    },
    {
      "epoch": 0.5761833911941368,
      "grad_norm": 0.7358115315437317,
      "learning_rate": 8.324167833758104e-05,
      "loss": 2.0153,
      "step": 2614
    },
    {
      "epoch": 0.5764038133024743,
      "grad_norm": 0.7375444769859314,
      "learning_rate": 8.316982284127438e-05,
      "loss": 2.2355,
      "step": 2615
    },
    {
      "epoch": 0.5766242354108118,
      "grad_norm": 0.7345470786094666,
      "learning_rate": 8.309797628695144e-05,
      "loss": 2.025,
      "step": 2616
    },
    {
      "epoch": 0.5768446575191492,
      "grad_norm": 0.7567511200904846,
      "learning_rate": 8.302613871278474e-05,
      "loss": 2.1208,
      "step": 2617
    },
    {
      "epoch": 0.5770650796274867,
      "grad_norm": 0.7271445989608765,
      "learning_rate": 8.295431015694202e-05,
      "loss": 2.0525,
      "step": 2618
    },
    {
      "epoch": 0.5772855017358242,
      "grad_norm": 0.6797577738761902,
      "learning_rate": 8.28824906575863e-05,
      "loss": 2.0027,
      "step": 2619
    },
    {
      "epoch": 0.5775059238441616,
      "grad_norm": 0.761390209197998,
      "learning_rate": 8.281068025287575e-05,
      "loss": 2.2242,
      "step": 2620
    },
    {
      "epoch": 0.577726345952499,
      "grad_norm": 0.7398227453231812,
      "learning_rate": 8.273887898096369e-05,
      "loss": 2.0515,
      "step": 2621
    },
    {
      "epoch": 0.5779467680608364,
      "grad_norm": 0.700165331363678,
      "learning_rate": 8.266708687999866e-05,
      "loss": 2.11,
      "step": 2622
    },
    {
      "epoch": 0.578167190169174,
      "grad_norm": 0.7190209031105042,
      "learning_rate": 8.259530398812421e-05,
      "loss": 2.1392,
      "step": 2623
    },
    {
      "epoch": 0.5783876122775115,
      "grad_norm": 0.7021013498306274,
      "learning_rate": 8.25235303434791e-05,
      "loss": 2.118,
      "step": 2624
    },
    {
      "epoch": 0.5786080343858488,
      "grad_norm": 0.727543294429779,
      "learning_rate": 8.245176598419713e-05,
      "loss": 2.1396,
      "step": 2625
    },
    {
      "epoch": 0.5788284564941863,
      "grad_norm": 0.7494224905967712,
      "learning_rate": 8.238001094840716e-05,
      "loss": 2.1022,
      "step": 2626
    },
    {
      "epoch": 0.5790488786025239,
      "grad_norm": 0.7438943386077881,
      "learning_rate": 8.230826527423313e-05,
      "loss": 2.1947,
      "step": 2627
    },
    {
      "epoch": 0.5792693007108612,
      "grad_norm": 0.6862063407897949,
      "learning_rate": 8.223652899979402e-05,
      "loss": 2.0313,
      "step": 2628
    },
    {
      "epoch": 0.5794897228191987,
      "grad_norm": 0.7102648019790649,
      "learning_rate": 8.216480216320371e-05,
      "loss": 2.0807,
      "step": 2629
    },
    {
      "epoch": 0.5797101449275363,
      "grad_norm": 0.7810218334197998,
      "learning_rate": 8.209308480257119e-05,
      "loss": 2.1251,
      "step": 2630
    },
    {
      "epoch": 0.5799305670358736,
      "grad_norm": 0.7512586712837219,
      "learning_rate": 8.202137695600034e-05,
      "loss": 2.2243,
      "step": 2631
    },
    {
      "epoch": 0.5801509891442111,
      "grad_norm": 0.7023181319236755,
      "learning_rate": 8.194967866159002e-05,
      "loss": 2.0131,
      "step": 2632
    },
    {
      "epoch": 0.5803714112525487,
      "grad_norm": 0.6931320428848267,
      "learning_rate": 8.187798995743402e-05,
      "loss": 2.0536,
      "step": 2633
    },
    {
      "epoch": 0.580591833360886,
      "grad_norm": 0.7285056114196777,
      "learning_rate": 8.180631088162102e-05,
      "loss": 2.0488,
      "step": 2634
    },
    {
      "epoch": 0.5808122554692235,
      "grad_norm": 0.7166817784309387,
      "learning_rate": 8.173464147223456e-05,
      "loss": 2.0041,
      "step": 2635
    },
    {
      "epoch": 0.581032677577561,
      "grad_norm": 0.8001969456672668,
      "learning_rate": 8.166298176735302e-05,
      "loss": 2.0566,
      "step": 2636
    },
    {
      "epoch": 0.5812530996858984,
      "grad_norm": 0.7055504322052002,
      "learning_rate": 8.15913318050498e-05,
      "loss": 2.0459,
      "step": 2637
    },
    {
      "epoch": 0.581473521794236,
      "grad_norm": 0.7528848648071289,
      "learning_rate": 8.15196916233929e-05,
      "loss": 2.0485,
      "step": 2638
    },
    {
      "epoch": 0.5816939439025735,
      "grad_norm": 0.7095149755477905,
      "learning_rate": 8.144806126044527e-05,
      "loss": 2.2716,
      "step": 2639
    },
    {
      "epoch": 0.5819143660109108,
      "grad_norm": 0.6842925548553467,
      "learning_rate": 8.137644075426454e-05,
      "loss": 2.1402,
      "step": 2640
    },
    {
      "epoch": 0.5821347881192483,
      "grad_norm": 0.7400439977645874,
      "learning_rate": 8.130483014290318e-05,
      "loss": 2.1002,
      "step": 2641
    },
    {
      "epoch": 0.5823552102275859,
      "grad_norm": 0.671662449836731,
      "learning_rate": 8.12332294644084e-05,
      "loss": 2.2139,
      "step": 2642
    },
    {
      "epoch": 0.5825756323359232,
      "grad_norm": 0.7912471294403076,
      "learning_rate": 8.116163875682211e-05,
      "loss": 2.2171,
      "step": 2643
    },
    {
      "epoch": 0.5827960544442607,
      "grad_norm": 0.7683003544807434,
      "learning_rate": 8.109005805818089e-05,
      "loss": 2.1217,
      "step": 2644
    },
    {
      "epoch": 0.5830164765525983,
      "grad_norm": 0.7864294052124023,
      "learning_rate": 8.101848740651603e-05,
      "loss": 2.1658,
      "step": 2645
    },
    {
      "epoch": 0.5832368986609356,
      "grad_norm": 0.7607876062393188,
      "learning_rate": 8.094692683985353e-05,
      "loss": 2.0156,
      "step": 2646
    },
    {
      "epoch": 0.5834573207692731,
      "grad_norm": 0.7581495642662048,
      "learning_rate": 8.087537639621395e-05,
      "loss": 2.0206,
      "step": 2647
    },
    {
      "epoch": 0.5836777428776107,
      "grad_norm": 0.7397662997245789,
      "learning_rate": 8.080383611361254e-05,
      "loss": 2.1292,
      "step": 2648
    },
    {
      "epoch": 0.583898164985948,
      "grad_norm": 0.7391577363014221,
      "learning_rate": 8.073230603005914e-05,
      "loss": 2.0279,
      "step": 2649
    },
    {
      "epoch": 0.5841185870942855,
      "grad_norm": 0.7051839232444763,
      "learning_rate": 8.066078618355808e-05,
      "loss": 1.9918,
      "step": 2650
    },
    {
      "epoch": 0.584339009202623,
      "grad_norm": 0.7147255539894104,
      "learning_rate": 8.058927661210838e-05,
      "loss": 2.0476,
      "step": 2651
    },
    {
      "epoch": 0.5845594313109604,
      "grad_norm": 0.8380058407783508,
      "learning_rate": 8.051777735370354e-05,
      "loss": 2.058,
      "step": 2652
    },
    {
      "epoch": 0.584779853419298,
      "grad_norm": 0.735348105430603,
      "learning_rate": 8.044628844633157e-05,
      "loss": 2.045,
      "step": 2653
    },
    {
      "epoch": 0.5850002755276354,
      "grad_norm": 0.7296150922775269,
      "learning_rate": 8.0374809927975e-05,
      "loss": 2.2092,
      "step": 2654
    },
    {
      "epoch": 0.5852206976359728,
      "grad_norm": 0.7700657248497009,
      "learning_rate": 8.030334183661086e-05,
      "loss": 2.1671,
      "step": 2655
    },
    {
      "epoch": 0.5854411197443103,
      "grad_norm": 0.7571229338645935,
      "learning_rate": 8.02318842102106e-05,
      "loss": 2.1388,
      "step": 2656
    },
    {
      "epoch": 0.5856615418526478,
      "grad_norm": 0.7969363331794739,
      "learning_rate": 8.01604370867401e-05,
      "loss": 2.151,
      "step": 2657
    },
    {
      "epoch": 0.5858819639609852,
      "grad_norm": 0.7473337650299072,
      "learning_rate": 8.008900050415973e-05,
      "loss": 2.1569,
      "step": 2658
    },
    {
      "epoch": 0.5861023860693227,
      "grad_norm": 0.7453312277793884,
      "learning_rate": 8.001757450042418e-05,
      "loss": 2.2713,
      "step": 2659
    },
    {
      "epoch": 0.5863228081776602,
      "grad_norm": 0.7751100659370422,
      "learning_rate": 7.994615911348257e-05,
      "loss": 2.1452,
      "step": 2660
    },
    {
      "epoch": 0.5865432302859976,
      "grad_norm": 0.6771991848945618,
      "learning_rate": 7.98747543812784e-05,
      "loss": 2.1456,
      "step": 2661
    },
    {
      "epoch": 0.5867636523943351,
      "grad_norm": 0.7564780712127686,
      "learning_rate": 7.98033603417494e-05,
      "loss": 2.0393,
      "step": 2662
    },
    {
      "epoch": 0.5869840745026726,
      "grad_norm": 0.7548890709877014,
      "learning_rate": 7.97319770328277e-05,
      "loss": 2.0592,
      "step": 2663
    },
    {
      "epoch": 0.58720449661101,
      "grad_norm": 0.7311673164367676,
      "learning_rate": 7.966060449243982e-05,
      "loss": 2.0645,
      "step": 2664
    },
    {
      "epoch": 0.5874249187193475,
      "grad_norm": 0.7643336057662964,
      "learning_rate": 7.958924275850636e-05,
      "loss": 2.1224,
      "step": 2665
    },
    {
      "epoch": 0.587645340827685,
      "grad_norm": 0.7388177514076233,
      "learning_rate": 7.951789186894231e-05,
      "loss": 2.0323,
      "step": 2666
    },
    {
      "epoch": 0.5878657629360224,
      "grad_norm": 0.7242656350135803,
      "learning_rate": 7.944655186165686e-05,
      "loss": 2.0062,
      "step": 2667
    },
    {
      "epoch": 0.5880861850443599,
      "grad_norm": 0.7420862913131714,
      "learning_rate": 7.937522277455343e-05,
      "loss": 2.0592,
      "step": 2668
    },
    {
      "epoch": 0.5883066071526974,
      "grad_norm": 0.7476639747619629,
      "learning_rate": 7.930390464552965e-05,
      "loss": 2.173,
      "step": 2669
    },
    {
      "epoch": 0.5885270292610348,
      "grad_norm": 0.7474034428596497,
      "learning_rate": 7.92325975124773e-05,
      "loss": 2.1659,
      "step": 2670
    },
    {
      "epoch": 0.5887474513693723,
      "grad_norm": 0.7865075469017029,
      "learning_rate": 7.916130141328231e-05,
      "loss": 2.0774,
      "step": 2671
    },
    {
      "epoch": 0.5889678734777098,
      "grad_norm": 0.7650787830352783,
      "learning_rate": 7.909001638582478e-05,
      "loss": 2.0434,
      "step": 2672
    },
    {
      "epoch": 0.5891882955860472,
      "grad_norm": 0.7125250697135925,
      "learning_rate": 7.901874246797888e-05,
      "loss": 2.0789,
      "step": 2673
    },
    {
      "epoch": 0.5894087176943847,
      "grad_norm": 0.7083979249000549,
      "learning_rate": 7.894747969761297e-05,
      "loss": 2.2476,
      "step": 2674
    },
    {
      "epoch": 0.5896291398027222,
      "grad_norm": 0.7141408920288086,
      "learning_rate": 7.887622811258939e-05,
      "loss": 2.0775,
      "step": 2675
    },
    {
      "epoch": 0.5898495619110596,
      "grad_norm": 0.7696911692619324,
      "learning_rate": 7.880498775076462e-05,
      "loss": 2.1862,
      "step": 2676
    },
    {
      "epoch": 0.5900699840193971,
      "grad_norm": 0.7391520738601685,
      "learning_rate": 7.873375864998906e-05,
      "loss": 2.0865,
      "step": 2677
    },
    {
      "epoch": 0.5902904061277346,
      "grad_norm": 0.6940497756004333,
      "learning_rate": 7.866254084810724e-05,
      "loss": 2.1002,
      "step": 2678
    },
    {
      "epoch": 0.590510828236072,
      "grad_norm": 0.7086051106452942,
      "learning_rate": 7.859133438295764e-05,
      "loss": 2.0306,
      "step": 2679
    },
    {
      "epoch": 0.5907312503444095,
      "grad_norm": 0.7082808613777161,
      "learning_rate": 7.852013929237272e-05,
      "loss": 2.1087,
      "step": 2680
    },
    {
      "epoch": 0.590951672452747,
      "grad_norm": 0.7266216278076172,
      "learning_rate": 7.844895561417893e-05,
      "loss": 2.0582,
      "step": 2681
    },
    {
      "epoch": 0.5911720945610844,
      "grad_norm": 0.7384577393531799,
      "learning_rate": 7.83777833861966e-05,
      "loss": 2.2582,
      "step": 2682
    },
    {
      "epoch": 0.5913925166694219,
      "grad_norm": 0.706230640411377,
      "learning_rate": 7.830662264623997e-05,
      "loss": 2.0106,
      "step": 2683
    },
    {
      "epoch": 0.5916129387777594,
      "grad_norm": 0.7162230014801025,
      "learning_rate": 7.823547343211725e-05,
      "loss": 2.0438,
      "step": 2684
    },
    {
      "epoch": 0.5918333608860968,
      "grad_norm": 0.743025004863739,
      "learning_rate": 7.816433578163048e-05,
      "loss": 2.1321,
      "step": 2685
    },
    {
      "epoch": 0.5920537829944343,
      "grad_norm": 0.793809175491333,
      "learning_rate": 7.809320973257556e-05,
      "loss": 2.1769,
      "step": 2686
    },
    {
      "epoch": 0.5922742051027718,
      "grad_norm": 0.7207586765289307,
      "learning_rate": 7.802209532274223e-05,
      "loss": 2.0845,
      "step": 2687
    },
    {
      "epoch": 0.5924946272111092,
      "grad_norm": 0.7435044646263123,
      "learning_rate": 7.795099258991404e-05,
      "loss": 1.9301,
      "step": 2688
    },
    {
      "epoch": 0.5927150493194467,
      "grad_norm": 0.7569065690040588,
      "learning_rate": 7.787990157186827e-05,
      "loss": 2.0365,
      "step": 2689
    },
    {
      "epoch": 0.5929354714277842,
      "grad_norm": 0.7191382646560669,
      "learning_rate": 7.780882230637617e-05,
      "loss": 2.212,
      "step": 2690
    },
    {
      "epoch": 0.5931558935361216,
      "grad_norm": 0.796032190322876,
      "learning_rate": 7.773775483120258e-05,
      "loss": 1.9993,
      "step": 2691
    },
    {
      "epoch": 0.5933763156444591,
      "grad_norm": 0.6840097904205322,
      "learning_rate": 7.766669918410608e-05,
      "loss": 2.1941,
      "step": 2692
    },
    {
      "epoch": 0.5935967377527966,
      "grad_norm": 0.7837522625923157,
      "learning_rate": 7.759565540283901e-05,
      "loss": 2.1539,
      "step": 2693
    },
    {
      "epoch": 0.593817159861134,
      "grad_norm": 0.7307417392730713,
      "learning_rate": 7.75246235251474e-05,
      "loss": 2.0807,
      "step": 2694
    },
    {
      "epoch": 0.5940375819694715,
      "grad_norm": 0.707915723323822,
      "learning_rate": 7.745360358877098e-05,
      "loss": 2.0617,
      "step": 2695
    },
    {
      "epoch": 0.594258004077809,
      "grad_norm": 0.7777189016342163,
      "learning_rate": 7.738259563144308e-05,
      "loss": 2.1484,
      "step": 2696
    },
    {
      "epoch": 0.5944784261861464,
      "grad_norm": 0.6708951592445374,
      "learning_rate": 7.731159969089073e-05,
      "loss": 2.012,
      "step": 2697
    },
    {
      "epoch": 0.5946988482944839,
      "grad_norm": 0.729523777961731,
      "learning_rate": 7.724061580483449e-05,
      "loss": 2.0668,
      "step": 2698
    },
    {
      "epoch": 0.5949192704028214,
      "grad_norm": 0.7528461813926697,
      "learning_rate": 7.716964401098859e-05,
      "loss": 2.1193,
      "step": 2699
    },
    {
      "epoch": 0.5951396925111588,
      "grad_norm": 0.7361257672309875,
      "learning_rate": 7.709868434706082e-05,
      "loss": 2.0451,
      "step": 2700
    },
    {
      "epoch": 0.5951396925111588,
      "eval_bpb": 1.0889559310165162,
      "eval_loss": 2.098937749862671,
      "eval_runtime": 184.3291,
      "eval_samples_per_second": 62.171,
      "eval_steps_per_second": 7.774,
      "step": 2700
    },
    {
      "epoch": 0.5953601146194963,
      "grad_norm": 0.7330508232116699,
      "learning_rate": 7.702773685075249e-05,
      "loss": 2.102,
      "step": 2701
    },
    {
      "epoch": 0.5955805367278338,
      "grad_norm": 0.7347526550292969,
      "learning_rate": 7.695680155975851e-05,
      "loss": 2.1805,
      "step": 2702
    },
    {
      "epoch": 0.5958009588361712,
      "grad_norm": 0.7290721535682678,
      "learning_rate": 7.68858785117672e-05,
      "loss": 2.0885,
      "step": 2703
    },
    {
      "epoch": 0.5960213809445087,
      "grad_norm": 0.7439634799957275,
      "learning_rate": 7.681496774446052e-05,
      "loss": 2.1089,
      "step": 2704
    },
    {
      "epoch": 0.5962418030528462,
      "grad_norm": 0.7170200943946838,
      "learning_rate": 7.674406929551377e-05,
      "loss": 1.9535,
      "step": 2705
    },
    {
      "epoch": 0.5964622251611836,
      "grad_norm": 0.7437044382095337,
      "learning_rate": 7.667318320259578e-05,
      "loss": 2.1477,
      "step": 2706
    },
    {
      "epoch": 0.5966826472695211,
      "grad_norm": 0.710358202457428,
      "learning_rate": 7.660230950336881e-05,
      "loss": 2.168,
      "step": 2707
    },
    {
      "epoch": 0.5969030693778586,
      "grad_norm": 0.7261993885040283,
      "learning_rate": 7.653144823548852e-05,
      "loss": 2.1884,
      "step": 2708
    },
    {
      "epoch": 0.597123491486196,
      "grad_norm": 0.756950855255127,
      "learning_rate": 7.646059943660393e-05,
      "loss": 2.0833,
      "step": 2709
    },
    {
      "epoch": 0.5973439135945335,
      "grad_norm": 0.7370526790618896,
      "learning_rate": 7.638976314435751e-05,
      "loss": 2.0395,
      "step": 2710
    },
    {
      "epoch": 0.597564335702871,
      "grad_norm": 0.7254035472869873,
      "learning_rate": 7.631893939638503e-05,
      "loss": 2.1282,
      "step": 2711
    },
    {
      "epoch": 0.5977847578112084,
      "grad_norm": 0.800471842288971,
      "learning_rate": 7.624812823031564e-05,
      "loss": 2.0334,
      "step": 2712
    },
    {
      "epoch": 0.5980051799195459,
      "grad_norm": 0.7469788789749146,
      "learning_rate": 7.617732968377174e-05,
      "loss": 2.0415,
      "step": 2713
    },
    {
      "epoch": 0.5982256020278834,
      "grad_norm": 0.7510771155357361,
      "learning_rate": 7.61065437943691e-05,
      "loss": 2.1486,
      "step": 2714
    },
    {
      "epoch": 0.5984460241362208,
      "grad_norm": 0.7523826956748962,
      "learning_rate": 7.60357705997167e-05,
      "loss": 2.1898,
      "step": 2715
    },
    {
      "epoch": 0.5986664462445583,
      "grad_norm": 0.7260956168174744,
      "learning_rate": 7.596501013741675e-05,
      "loss": 2.061,
      "step": 2716
    },
    {
      "epoch": 0.5988868683528958,
      "grad_norm": 0.781171977519989,
      "learning_rate": 7.589426244506486e-05,
      "loss": 2.1429,
      "step": 2717
    },
    {
      "epoch": 0.5991072904612332,
      "grad_norm": 0.6899266242980957,
      "learning_rate": 7.582352756024971e-05,
      "loss": 2.0968,
      "step": 2718
    },
    {
      "epoch": 0.5993277125695707,
      "grad_norm": 0.746330201625824,
      "learning_rate": 7.575280552055319e-05,
      "loss": 2.0869,
      "step": 2719
    },
    {
      "epoch": 0.5995481346779082,
      "grad_norm": 0.7655307650566101,
      "learning_rate": 7.568209636355037e-05,
      "loss": 2.1922,
      "step": 2720
    },
    {
      "epoch": 0.5997685567862456,
      "grad_norm": 0.7307795882225037,
      "learning_rate": 7.561140012680951e-05,
      "loss": 2.0397,
      "step": 2721
    },
    {
      "epoch": 0.5999889788945831,
      "grad_norm": 0.7907556295394897,
      "learning_rate": 7.554071684789198e-05,
      "loss": 2.135,
      "step": 2722
    },
    {
      "epoch": 0.6002094010029206,
      "grad_norm": 0.7930383682250977,
      "learning_rate": 7.547004656435229e-05,
      "loss": 2.1342,
      "step": 2723
    },
    {
      "epoch": 0.600429823111258,
      "grad_norm": 0.8798115849494934,
      "learning_rate": 7.5399389313738e-05,
      "loss": 2.0967,
      "step": 2724
    },
    {
      "epoch": 0.6006502452195955,
      "grad_norm": 0.7643640041351318,
      "learning_rate": 7.532874513358977e-05,
      "loss": 2.0944,
      "step": 2725
    },
    {
      "epoch": 0.600870667327933,
      "grad_norm": 0.751966655254364,
      "learning_rate": 7.525811406144133e-05,
      "loss": 2.0646,
      "step": 2726
    },
    {
      "epoch": 0.6010910894362704,
      "grad_norm": 0.7091186046600342,
      "learning_rate": 7.518749613481941e-05,
      "loss": 2.0524,
      "step": 2727
    },
    {
      "epoch": 0.6013115115446079,
      "grad_norm": 0.774990439414978,
      "learning_rate": 7.511689139124382e-05,
      "loss": 2.2815,
      "step": 2728
    },
    {
      "epoch": 0.6015319336529454,
      "grad_norm": 0.7981061339378357,
      "learning_rate": 7.504629986822732e-05,
      "loss": 2.1391,
      "step": 2729
    },
    {
      "epoch": 0.6017523557612828,
      "grad_norm": 0.7406912446022034,
      "learning_rate": 7.49757216032756e-05,
      "loss": 2.0451,
      "step": 2730
    },
    {
      "epoch": 0.6019727778696203,
      "grad_norm": 0.763405978679657,
      "learning_rate": 7.49051566338874e-05,
      "loss": 2.0671,
      "step": 2731
    },
    {
      "epoch": 0.6021931999779578,
      "grad_norm": 0.7163340449333191,
      "learning_rate": 7.483460499755433e-05,
      "loss": 2.1067,
      "step": 2732
    },
    {
      "epoch": 0.6024136220862952,
      "grad_norm": 0.7582015991210938,
      "learning_rate": 7.476406673176095e-05,
      "loss": 2.14,
      "step": 2733
    },
    {
      "epoch": 0.6026340441946327,
      "grad_norm": 0.7495436668395996,
      "learning_rate": 7.46935418739847e-05,
      "loss": 2.0139,
      "step": 2734
    },
    {
      "epoch": 0.6028544663029702,
      "grad_norm": 0.7356929183006287,
      "learning_rate": 7.462303046169592e-05,
      "loss": 2.0833,
      "step": 2735
    },
    {
      "epoch": 0.6030748884113076,
      "grad_norm": 0.7437918782234192,
      "learning_rate": 7.455253253235774e-05,
      "loss": 2.0278,
      "step": 2736
    },
    {
      "epoch": 0.6032953105196451,
      "grad_norm": 0.7280755639076233,
      "learning_rate": 7.44820481234262e-05,
      "loss": 2.0801,
      "step": 2737
    },
    {
      "epoch": 0.6035157326279826,
      "grad_norm": 0.7572881579399109,
      "learning_rate": 7.441157727235015e-05,
      "loss": 2.0614,
      "step": 2738
    },
    {
      "epoch": 0.60373615473632,
      "grad_norm": 0.7079784870147705,
      "learning_rate": 7.434112001657118e-05,
      "loss": 2.1317,
      "step": 2739
    },
    {
      "epoch": 0.6039565768446575,
      "grad_norm": 0.7590844035148621,
      "learning_rate": 7.427067639352372e-05,
      "loss": 2.1282,
      "step": 2740
    },
    {
      "epoch": 0.604176998952995,
      "grad_norm": 0.7550207376480103,
      "learning_rate": 7.420024644063492e-05,
      "loss": 2.1149,
      "step": 2741
    },
    {
      "epoch": 0.6043974210613324,
      "grad_norm": 0.7127838730812073,
      "learning_rate": 7.412983019532468e-05,
      "loss": 2.198,
      "step": 2742
    },
    {
      "epoch": 0.6046178431696699,
      "grad_norm": 0.738637387752533,
      "learning_rate": 7.405942769500555e-05,
      "loss": 2.1876,
      "step": 2743
    },
    {
      "epoch": 0.6048382652780074,
      "grad_norm": 0.7345795631408691,
      "learning_rate": 7.398903897708297e-05,
      "loss": 2.1588,
      "step": 2744
    },
    {
      "epoch": 0.6050586873863448,
      "grad_norm": 0.7326908111572266,
      "learning_rate": 7.391866407895482e-05,
      "loss": 2.0499,
      "step": 2745
    },
    {
      "epoch": 0.6052791094946823,
      "grad_norm": 0.7031310200691223,
      "learning_rate": 7.384830303801181e-05,
      "loss": 2.0247,
      "step": 2746
    },
    {
      "epoch": 0.6054995316030198,
      "grad_norm": 0.7494580745697021,
      "learning_rate": 7.377795589163719e-05,
      "loss": 2.1327,
      "step": 2747
    },
    {
      "epoch": 0.6057199537113572,
      "grad_norm": 0.7185010313987732,
      "learning_rate": 7.370762267720685e-05,
      "loss": 1.9736,
      "step": 2748
    },
    {
      "epoch": 0.6059403758196947,
      "grad_norm": 0.7364353537559509,
      "learning_rate": 7.363730343208934e-05,
      "loss": 2.0373,
      "step": 2749
    },
    {
      "epoch": 0.6061607979280322,
      "grad_norm": 0.6954378485679626,
      "learning_rate": 7.35669981936457e-05,
      "loss": 2.0227,
      "step": 2750
    },
    {
      "epoch": 0.6063812200363696,
      "grad_norm": 0.7266446352005005,
      "learning_rate": 7.349670699922955e-05,
      "loss": 2.1922,
      "step": 2751
    },
    {
      "epoch": 0.6066016421447071,
      "grad_norm": 0.7645584344863892,
      "learning_rate": 7.342642988618708e-05,
      "loss": 2.1525,
      "step": 2752
    },
    {
      "epoch": 0.6068220642530446,
      "grad_norm": 0.7032352089881897,
      "learning_rate": 7.3356166891857e-05,
      "loss": 2.1294,
      "step": 2753
    },
    {
      "epoch": 0.607042486361382,
      "grad_norm": 0.7121439576148987,
      "learning_rate": 7.328591805357048e-05,
      "loss": 2.0433,
      "step": 2754
    },
    {
      "epoch": 0.6072629084697195,
      "grad_norm": 0.7339364886283875,
      "learning_rate": 7.321568340865119e-05,
      "loss": 2.0966,
      "step": 2755
    },
    {
      "epoch": 0.6074833305780569,
      "grad_norm": 0.7135085463523865,
      "learning_rate": 7.314546299441529e-05,
      "loss": 1.9835,
      "step": 2756
    },
    {
      "epoch": 0.6077037526863944,
      "grad_norm": 0.7258579730987549,
      "learning_rate": 7.307525684817129e-05,
      "loss": 2.0399,
      "step": 2757
    },
    {
      "epoch": 0.6079241747947319,
      "grad_norm": 0.7161139845848083,
      "learning_rate": 7.30050650072202e-05,
      "loss": 2.0774,
      "step": 2758
    },
    {
      "epoch": 0.6081445969030693,
      "grad_norm": 0.7751717567443848,
      "learning_rate": 7.293488750885542e-05,
      "loss": 2.0665,
      "step": 2759
    },
    {
      "epoch": 0.6083650190114068,
      "grad_norm": 0.7212087512016296,
      "learning_rate": 7.286472439036272e-05,
      "loss": 2.1587,
      "step": 2760
    },
    {
      "epoch": 0.6085854411197443,
      "grad_norm": 0.7590039372444153,
      "learning_rate": 7.27945756890202e-05,
      "loss": 2.0265,
      "step": 2761
    },
    {
      "epoch": 0.6088058632280817,
      "grad_norm": 0.7360100150108337,
      "learning_rate": 7.272444144209836e-05,
      "loss": 2.1842,
      "step": 2762
    },
    {
      "epoch": 0.6090262853364192,
      "grad_norm": 0.7158344984054565,
      "learning_rate": 7.265432168685996e-05,
      "loss": 2.0967,
      "step": 2763
    },
    {
      "epoch": 0.6092467074447567,
      "grad_norm": 0.7693300843238831,
      "learning_rate": 7.258421646056007e-05,
      "loss": 2.0416,
      "step": 2764
    },
    {
      "epoch": 0.6094671295530941,
      "grad_norm": 0.7025851011276245,
      "learning_rate": 7.25141258004461e-05,
      "loss": 1.9658,
      "step": 2765
    },
    {
      "epoch": 0.6096875516614316,
      "grad_norm": 0.6819362044334412,
      "learning_rate": 7.244404974375764e-05,
      "loss": 2.2908,
      "step": 2766
    },
    {
      "epoch": 0.6099079737697691,
      "grad_norm": 0.7687162756919861,
      "learning_rate": 7.237398832772659e-05,
      "loss": 1.9637,
      "step": 2767
    },
    {
      "epoch": 0.6101283958781065,
      "grad_norm": 0.7330003380775452,
      "learning_rate": 7.230394158957705e-05,
      "loss": 2.1167,
      "step": 2768
    },
    {
      "epoch": 0.610348817986444,
      "grad_norm": 0.7117829918861389,
      "learning_rate": 7.223390956652523e-05,
      "loss": 1.9852,
      "step": 2769
    },
    {
      "epoch": 0.6105692400947815,
      "grad_norm": 0.7314755320549011,
      "learning_rate": 7.21638922957797e-05,
      "loss": 2.1395,
      "step": 2770
    },
    {
      "epoch": 0.6107896622031189,
      "grad_norm": 0.7234629392623901,
      "learning_rate": 7.209388981454108e-05,
      "loss": 2.0331,
      "step": 2771
    },
    {
      "epoch": 0.6110100843114564,
      "grad_norm": 0.708051860332489,
      "learning_rate": 7.202390216000212e-05,
      "loss": 2.0573,
      "step": 2772
    },
    {
      "epoch": 0.6112305064197939,
      "grad_norm": 0.7324710488319397,
      "learning_rate": 7.195392936934771e-05,
      "loss": 2.1078,
      "step": 2773
    },
    {
      "epoch": 0.6114509285281313,
      "grad_norm": 0.6876938343048096,
      "learning_rate": 7.188397147975488e-05,
      "loss": 2.0168,
      "step": 2774
    },
    {
      "epoch": 0.6116713506364688,
      "grad_norm": 0.7154436111450195,
      "learning_rate": 7.181402852839272e-05,
      "loss": 1.9858,
      "step": 2775
    },
    {
      "epoch": 0.6118917727448063,
      "grad_norm": 0.7233720421791077,
      "learning_rate": 7.174410055242236e-05,
      "loss": 2.0199,
      "step": 2776
    },
    {
      "epoch": 0.6121121948531437,
      "grad_norm": 0.7399248480796814,
      "learning_rate": 7.167418758899702e-05,
      "loss": 2.1234,
      "step": 2777
    },
    {
      "epoch": 0.6123326169614812,
      "grad_norm": 0.7848982214927673,
      "learning_rate": 7.160428967526187e-05,
      "loss": 2.0666,
      "step": 2778
    },
    {
      "epoch": 0.6125530390698187,
      "grad_norm": 0.7393653392791748,
      "learning_rate": 7.153440684835417e-05,
      "loss": 2.0927,
      "step": 2779
    },
    {
      "epoch": 0.6127734611781561,
      "grad_norm": 0.777467668056488,
      "learning_rate": 7.14645391454031e-05,
      "loss": 2.0484,
      "step": 2780
    },
    {
      "epoch": 0.6129938832864936,
      "grad_norm": 0.764718770980835,
      "learning_rate": 7.139468660352986e-05,
      "loss": 2.1124,
      "step": 2781
    },
    {
      "epoch": 0.6132143053948311,
      "grad_norm": 0.7803124785423279,
      "learning_rate": 7.132484925984754e-05,
      "loss": 2.3194,
      "step": 2782
    },
    {
      "epoch": 0.6134347275031685,
      "grad_norm": 0.7015304565429688,
      "learning_rate": 7.125502715146122e-05,
      "loss": 2.078,
      "step": 2783
    },
    {
      "epoch": 0.613655149611506,
      "grad_norm": 0.7579970955848694,
      "learning_rate": 7.118522031546778e-05,
      "loss": 2.1551,
      "step": 2784
    },
    {
      "epoch": 0.6138755717198435,
      "grad_norm": 0.7808073163032532,
      "learning_rate": 7.111542878895607e-05,
      "loss": 2.0683,
      "step": 2785
    },
    {
      "epoch": 0.6140959938281809,
      "grad_norm": 0.7627034187316895,
      "learning_rate": 7.104565260900681e-05,
      "loss": 1.9833,
      "step": 2786
    },
    {
      "epoch": 0.6143164159365184,
      "grad_norm": 0.6829303503036499,
      "learning_rate": 7.097589181269255e-05,
      "loss": 2.0872,
      "step": 2787
    },
    {
      "epoch": 0.6145368380448559,
      "grad_norm": 0.7341774702072144,
      "learning_rate": 7.090614643707762e-05,
      "loss": 2.1746,
      "step": 2788
    },
    {
      "epoch": 0.6147572601531933,
      "grad_norm": 0.7393391728401184,
      "learning_rate": 7.083641651921824e-05,
      "loss": 2.06,
      "step": 2789
    },
    {
      "epoch": 0.6149776822615308,
      "grad_norm": 0.7622285485267639,
      "learning_rate": 7.076670209616234e-05,
      "loss": 2.1322,
      "step": 2790
    },
    {
      "epoch": 0.6151981043698683,
      "grad_norm": 0.7606104016304016,
      "learning_rate": 7.069700320494966e-05,
      "loss": 2.0038,
      "step": 2791
    },
    {
      "epoch": 0.6154185264782057,
      "grad_norm": 0.8114116191864014,
      "learning_rate": 7.062731988261166e-05,
      "loss": 2.0321,
      "step": 2792
    },
    {
      "epoch": 0.6156389485865432,
      "grad_norm": 0.6752614378929138,
      "learning_rate": 7.05576521661716e-05,
      "loss": 2.0087,
      "step": 2793
    },
    {
      "epoch": 0.6158593706948807,
      "grad_norm": 0.7434911131858826,
      "learning_rate": 7.048800009264434e-05,
      "loss": 2.1609,
      "step": 2794
    },
    {
      "epoch": 0.6160797928032181,
      "grad_norm": 0.7156621813774109,
      "learning_rate": 7.041836369903653e-05,
      "loss": 2.0175,
      "step": 2795
    },
    {
      "epoch": 0.6163002149115556,
      "grad_norm": 0.7479173541069031,
      "learning_rate": 7.034874302234638e-05,
      "loss": 2.0566,
      "step": 2796
    },
    {
      "epoch": 0.6165206370198931,
      "grad_norm": 0.7910670638084412,
      "learning_rate": 7.027913809956388e-05,
      "loss": 2.0832,
      "step": 2797
    },
    {
      "epoch": 0.6167410591282305,
      "grad_norm": 0.7796617746353149,
      "learning_rate": 7.020954896767058e-05,
      "loss": 2.0377,
      "step": 2798
    },
    {
      "epoch": 0.616961481236568,
      "grad_norm": 0.7695031762123108,
      "learning_rate": 7.013997566363959e-05,
      "loss": 2.1471,
      "step": 2799
    },
    {
      "epoch": 0.6171819033449055,
      "grad_norm": 0.7549078464508057,
      "learning_rate": 7.00704182244357e-05,
      "loss": 2.1181,
      "step": 2800
    },
    {
      "epoch": 0.6174023254532429,
      "grad_norm": 0.7182407379150391,
      "learning_rate": 7.000087668701521e-05,
      "loss": 2.0052,
      "step": 2801
    },
    {
      "epoch": 0.6176227475615804,
      "grad_norm": 0.7079635262489319,
      "learning_rate": 6.993135108832603e-05,
      "loss": 2.0871,
      "step": 2802
    },
    {
      "epoch": 0.6178431696699179,
      "grad_norm": 0.7191575169563293,
      "learning_rate": 6.986184146530755e-05,
      "loss": 2.0253,
      "step": 2803
    },
    {
      "epoch": 0.6180635917782553,
      "grad_norm": 0.7645883560180664,
      "learning_rate": 6.979234785489071e-05,
      "loss": 2.1112,
      "step": 2804
    },
    {
      "epoch": 0.6182840138865928,
      "grad_norm": 0.7503339648246765,
      "learning_rate": 6.972287029399789e-05,
      "loss": 2.031,
      "step": 2805
    },
    {
      "epoch": 0.6185044359949303,
      "grad_norm": 0.8079873323440552,
      "learning_rate": 6.965340881954296e-05,
      "loss": 2.1414,
      "step": 2806
    },
    {
      "epoch": 0.6187248581032677,
      "grad_norm": 0.7931023240089417,
      "learning_rate": 6.95839634684313e-05,
      "loss": 2.1381,
      "step": 2807
    },
    {
      "epoch": 0.6189452802116052,
      "grad_norm": 0.712077796459198,
      "learning_rate": 6.951453427755968e-05,
      "loss": 2.0108,
      "step": 2808
    },
    {
      "epoch": 0.6191657023199427,
      "grad_norm": 0.7282228469848633,
      "learning_rate": 6.944512128381624e-05,
      "loss": 2.1553,
      "step": 2809
    },
    {
      "epoch": 0.6193861244282801,
      "grad_norm": 0.827318549156189,
      "learning_rate": 6.937572452408063e-05,
      "loss": 2.1864,
      "step": 2810
    },
    {
      "epoch": 0.6196065465366176,
      "grad_norm": 0.7493376731872559,
      "learning_rate": 6.930634403522373e-05,
      "loss": 2.0991,
      "step": 2811
    },
    {
      "epoch": 0.6198269686449551,
      "grad_norm": 0.7572352886199951,
      "learning_rate": 6.923697985410789e-05,
      "loss": 2.0094,
      "step": 2812
    },
    {
      "epoch": 0.6200473907532925,
      "grad_norm": 0.7347143292427063,
      "learning_rate": 6.916763201758674e-05,
      "loss": 2.0681,
      "step": 2813
    },
    {
      "epoch": 0.62026781286163,
      "grad_norm": 0.7059430480003357,
      "learning_rate": 6.909830056250527e-05,
      "loss": 2.0675,
      "step": 2814
    },
    {
      "epoch": 0.6204882349699675,
      "grad_norm": 0.7574236392974854,
      "learning_rate": 6.90289855256997e-05,
      "loss": 2.122,
      "step": 2815
    },
    {
      "epoch": 0.6207086570783049,
      "grad_norm": 0.6864249110221863,
      "learning_rate": 6.895968694399762e-05,
      "loss": 2.0833,
      "step": 2816
    },
    {
      "epoch": 0.6209290791866424,
      "grad_norm": 0.744102418422699,
      "learning_rate": 6.889040485421774e-05,
      "loss": 2.2059,
      "step": 2817
    },
    {
      "epoch": 0.6211495012949799,
      "grad_norm": 0.6920540928840637,
      "learning_rate": 6.882113929317015e-05,
      "loss": 2.0486,
      "step": 2818
    },
    {
      "epoch": 0.6213699234033173,
      "grad_norm": 0.6970950365066528,
      "learning_rate": 6.875189029765608e-05,
      "loss": 1.978,
      "step": 2819
    },
    {
      "epoch": 0.6215903455116548,
      "grad_norm": 0.7010766863822937,
      "learning_rate": 6.868265790446796e-05,
      "loss": 2.2172,
      "step": 2820
    },
    {
      "epoch": 0.6218107676199923,
      "grad_norm": 0.7584542632102966,
      "learning_rate": 6.861344215038946e-05,
      "loss": 2.0618,
      "step": 2821
    },
    {
      "epoch": 0.6220311897283297,
      "grad_norm": 0.6899442672729492,
      "learning_rate": 6.854424307219528e-05,
      "loss": 2.1388,
      "step": 2822
    },
    {
      "epoch": 0.6222516118366672,
      "grad_norm": 0.7364000082015991,
      "learning_rate": 6.847506070665144e-05,
      "loss": 2.0347,
      "step": 2823
    },
    {
      "epoch": 0.6224720339450047,
      "grad_norm": 0.7737076878547668,
      "learning_rate": 6.840589509051495e-05,
      "loss": 2.1431,
      "step": 2824
    },
    {
      "epoch": 0.6226924560533421,
      "grad_norm": 0.7030546069145203,
      "learning_rate": 6.833674626053396e-05,
      "loss": 1.9907,
      "step": 2825
    },
    {
      "epoch": 0.6229128781616796,
      "grad_norm": 0.6712526679039001,
      "learning_rate": 6.82676142534477e-05,
      "loss": 2.0704,
      "step": 2826
    },
    {
      "epoch": 0.6231333002700171,
      "grad_norm": 0.7731640934944153,
      "learning_rate": 6.819849910598643e-05,
      "loss": 2.098,
      "step": 2827
    },
    {
      "epoch": 0.6233537223783545,
      "grad_norm": 0.7477912902832031,
      "learning_rate": 6.81294008548715e-05,
      "loss": 2.1202,
      "step": 2828
    },
    {
      "epoch": 0.623574144486692,
      "grad_norm": 0.7735987901687622,
      "learning_rate": 6.806031953681527e-05,
      "loss": 2.1141,
      "step": 2829
    },
    {
      "epoch": 0.6237945665950295,
      "grad_norm": 0.6560885310173035,
      "learning_rate": 6.799125518852109e-05,
      "loss": 2.051,
      "step": 2830
    },
    {
      "epoch": 0.6240149887033669,
      "grad_norm": 0.732367217540741,
      "learning_rate": 6.792220784668335e-05,
      "loss": 2.0855,
      "step": 2831
    },
    {
      "epoch": 0.6242354108117044,
      "grad_norm": 0.7268504500389099,
      "learning_rate": 6.785317754798727e-05,
      "loss": 2.0873,
      "step": 2832
    },
    {
      "epoch": 0.6244558329200419,
      "grad_norm": 0.7238990068435669,
      "learning_rate": 6.778416432910914e-05,
      "loss": 2.1807,
      "step": 2833
    },
    {
      "epoch": 0.6246762550283793,
      "grad_norm": 0.7326024770736694,
      "learning_rate": 6.771516822671617e-05,
      "loss": 2.0197,
      "step": 2834
    },
    {
      "epoch": 0.6248966771367168,
      "grad_norm": 0.7739957571029663,
      "learning_rate": 6.764618927746639e-05,
      "loss": 2.1359,
      "step": 2835
    },
    {
      "epoch": 0.6251170992450543,
      "grad_norm": 0.7399500012397766,
      "learning_rate": 6.757722751800881e-05,
      "loss": 2.0449,
      "step": 2836
    },
    {
      "epoch": 0.6253375213533917,
      "grad_norm": 0.7018681168556213,
      "learning_rate": 6.750828298498324e-05,
      "loss": 2.018,
      "step": 2837
    },
    {
      "epoch": 0.6255579434617292,
      "grad_norm": 0.7527201771736145,
      "learning_rate": 6.743935571502038e-05,
      "loss": 2.0827,
      "step": 2838
    },
    {
      "epoch": 0.6257783655700667,
      "grad_norm": 0.709136426448822,
      "learning_rate": 6.737044574474172e-05,
      "loss": 2.1458,
      "step": 2839
    },
    {
      "epoch": 0.6259987876784041,
      "grad_norm": 0.6900165677070618,
      "learning_rate": 6.730155311075958e-05,
      "loss": 1.9629,
      "step": 2840
    },
    {
      "epoch": 0.6262192097867416,
      "grad_norm": 0.6699858903884888,
      "learning_rate": 6.72326778496771e-05,
      "loss": 2.0109,
      "step": 2841
    },
    {
      "epoch": 0.6264396318950791,
      "grad_norm": 0.7272409796714783,
      "learning_rate": 6.716381999808816e-05,
      "loss": 2.1203,
      "step": 2842
    },
    {
      "epoch": 0.6266600540034165,
      "grad_norm": 0.767312228679657,
      "learning_rate": 6.709497959257736e-05,
      "loss": 2.0131,
      "step": 2843
    },
    {
      "epoch": 0.626880476111754,
      "grad_norm": 0.7840553522109985,
      "learning_rate": 6.702615666972005e-05,
      "loss": 2.0784,
      "step": 2844
    },
    {
      "epoch": 0.6271008982200915,
      "grad_norm": 0.7371819019317627,
      "learning_rate": 6.695735126608232e-05,
      "loss": 2.0831,
      "step": 2845
    },
    {
      "epoch": 0.6273213203284289,
      "grad_norm": 0.7085704207420349,
      "learning_rate": 6.688856341822095e-05,
      "loss": 2.0489,
      "step": 2846
    },
    {
      "epoch": 0.6275417424367664,
      "grad_norm": 0.7498862743377686,
      "learning_rate": 6.681979316268334e-05,
      "loss": 2.0606,
      "step": 2847
    },
    {
      "epoch": 0.6277621645451039,
      "grad_norm": 0.7472095489501953,
      "learning_rate": 6.675104053600763e-05,
      "loss": 2.101,
      "step": 2848
    },
    {
      "epoch": 0.6279825866534413,
      "grad_norm": 0.7332403659820557,
      "learning_rate": 6.668230557472244e-05,
      "loss": 2.1272,
      "step": 2849
    },
    {
      "epoch": 0.6282030087617788,
      "grad_norm": 0.7318052649497986,
      "learning_rate": 6.661358831534722e-05,
      "loss": 2.1375,
      "step": 2850
    },
    {
      "epoch": 0.6284234308701163,
      "grad_norm": 0.7408273816108704,
      "learning_rate": 6.654488879439186e-05,
      "loss": 2.143,
      "step": 2851
    },
    {
      "epoch": 0.6286438529784537,
      "grad_norm": 0.8091197609901428,
      "learning_rate": 6.647620704835689e-05,
      "loss": 2.2642,
      "step": 2852
    },
    {
      "epoch": 0.6288642750867912,
      "grad_norm": 0.7354791164398193,
      "learning_rate": 6.640754311373331e-05,
      "loss": 2.0863,
      "step": 2853
    },
    {
      "epoch": 0.6290846971951287,
      "grad_norm": 0.7170647382736206,
      "learning_rate": 6.633889702700277e-05,
      "loss": 2.0302,
      "step": 2854
    },
    {
      "epoch": 0.6293051193034661,
      "grad_norm": 0.7310125827789307,
      "learning_rate": 6.627026882463736e-05,
      "loss": 2.0112,
      "step": 2855
    },
    {
      "epoch": 0.6295255414118036,
      "grad_norm": 0.7428621053695679,
      "learning_rate": 6.620165854309972e-05,
      "loss": 2.1,
      "step": 2856
    },
    {
      "epoch": 0.6297459635201411,
      "grad_norm": 0.7111678123474121,
      "learning_rate": 6.613306621884295e-05,
      "loss": 2.0999,
      "step": 2857
    },
    {
      "epoch": 0.6299663856284785,
      "grad_norm": 0.7336297631263733,
      "learning_rate": 6.606449188831057e-05,
      "loss": 2.295,
      "step": 2858
    },
    {
      "epoch": 0.630186807736816,
      "grad_norm": 0.692692756652832,
      "learning_rate": 6.599593558793657e-05,
      "loss": 2.1894,
      "step": 2859
    },
    {
      "epoch": 0.6304072298451535,
      "grad_norm": 0.7032929062843323,
      "learning_rate": 6.592739735414539e-05,
      "loss": 2.129,
      "step": 2860
    },
    {
      "epoch": 0.6306276519534909,
      "grad_norm": 0.6886752247810364,
      "learning_rate": 6.585887722335183e-05,
      "loss": 2.1166,
      "step": 2861
    },
    {
      "epoch": 0.6308480740618284,
      "grad_norm": 0.7885454893112183,
      "learning_rate": 6.579037523196108e-05,
      "loss": 2.0616,
      "step": 2862
    },
    {
      "epoch": 0.6310684961701659,
      "grad_norm": 0.7414758801460266,
      "learning_rate": 6.572189141636874e-05,
      "loss": 2.1443,
      "step": 2863
    },
    {
      "epoch": 0.6312889182785033,
      "grad_norm": 0.7647479772567749,
      "learning_rate": 6.565342581296066e-05,
      "loss": 2.1764,
      "step": 2864
    },
    {
      "epoch": 0.6315093403868408,
      "grad_norm": 0.7073794603347778,
      "learning_rate": 6.558497845811306e-05,
      "loss": 2.1581,
      "step": 2865
    },
    {
      "epoch": 0.6317297624951783,
      "grad_norm": 0.6897346377372742,
      "learning_rate": 6.551654938819252e-05,
      "loss": 2.1702,
      "step": 2866
    },
    {
      "epoch": 0.6319501846035157,
      "grad_norm": 0.7756693959236145,
      "learning_rate": 6.544813863955581e-05,
      "loss": 2.0684,
      "step": 2867
    },
    {
      "epoch": 0.6321706067118532,
      "grad_norm": 0.7122083902359009,
      "learning_rate": 6.537974624855003e-05,
      "loss": 2.0858,
      "step": 2868
    },
    {
      "epoch": 0.6323910288201907,
      "grad_norm": 0.7361372709274292,
      "learning_rate": 6.531137225151252e-05,
      "loss": 2.1482,
      "step": 2869
    },
    {
      "epoch": 0.6326114509285281,
      "grad_norm": 0.7003417015075684,
      "learning_rate": 6.524301668477079e-05,
      "loss": 2.0857,
      "step": 2870
    },
    {
      "epoch": 0.6328318730368656,
      "grad_norm": 0.7269735932350159,
      "learning_rate": 6.517467958464264e-05,
      "loss": 2.1505,
      "step": 2871
    },
    {
      "epoch": 0.6330522951452031,
      "grad_norm": 0.7193023562431335,
      "learning_rate": 6.510636098743601e-05,
      "loss": 2.1274,
      "step": 2872
    },
    {
      "epoch": 0.6332727172535405,
      "grad_norm": 0.778118908405304,
      "learning_rate": 6.503806092944902e-05,
      "loss": 2.1311,
      "step": 2873
    },
    {
      "epoch": 0.633493139361878,
      "grad_norm": 0.7600733041763306,
      "learning_rate": 6.496977944696992e-05,
      "loss": 2.0698,
      "step": 2874
    },
    {
      "epoch": 0.6337135614702155,
      "grad_norm": 0.7130941152572632,
      "learning_rate": 6.490151657627716e-05,
      "loss": 2.1302,
      "step": 2875
    },
    {
      "epoch": 0.6339339835785529,
      "grad_norm": 0.7717383503913879,
      "learning_rate": 6.483327235363916e-05,
      "loss": 2.0334,
      "step": 2876
    },
    {
      "epoch": 0.6341544056868904,
      "grad_norm": 0.7721238732337952,
      "learning_rate": 6.476504681531463e-05,
      "loss": 2.0761,
      "step": 2877
    },
    {
      "epoch": 0.6343748277952279,
      "grad_norm": 0.7250845432281494,
      "learning_rate": 6.46968399975522e-05,
      "loss": 2.1473,
      "step": 2878
    },
    {
      "epoch": 0.6345952499035653,
      "grad_norm": 0.7835193276405334,
      "learning_rate": 6.46286519365906e-05,
      "loss": 2.1627,
      "step": 2879
    },
    {
      "epoch": 0.6348156720119028,
      "grad_norm": 0.7827721238136292,
      "learning_rate": 6.456048266865859e-05,
      "loss": 2.181,
      "step": 2880
    },
    {
      "epoch": 0.6350360941202403,
      "grad_norm": 0.7645891308784485,
      "learning_rate": 6.449233222997495e-05,
      "loss": 2.0866,
      "step": 2881
    },
    {
      "epoch": 0.6352565162285777,
      "grad_norm": 0.7321974635124207,
      "learning_rate": 6.442420065674845e-05,
      "loss": 2.0791,
      "step": 2882
    },
    {
      "epoch": 0.6354769383369152,
      "grad_norm": 0.7592393755912781,
      "learning_rate": 6.435608798517788e-05,
      "loss": 2.1078,
      "step": 2883
    },
    {
      "epoch": 0.6356973604452527,
      "grad_norm": 0.7080832123756409,
      "learning_rate": 6.428799425145194e-05,
      "loss": 2.0044,
      "step": 2884
    },
    {
      "epoch": 0.6359177825535901,
      "grad_norm": 0.741446852684021,
      "learning_rate": 6.421991949174921e-05,
      "loss": 2.1043,
      "step": 2885
    },
    {
      "epoch": 0.6361382046619276,
      "grad_norm": 0.735375702381134,
      "learning_rate": 6.415186374223834e-05,
      "loss": 2.0884,
      "step": 2886
    },
    {
      "epoch": 0.6363586267702651,
      "grad_norm": 0.728032112121582,
      "learning_rate": 6.408382703907776e-05,
      "loss": 2.054,
      "step": 2887
    },
    {
      "epoch": 0.6365790488786025,
      "grad_norm": 0.7075155973434448,
      "learning_rate": 6.40158094184158e-05,
      "loss": 2.1185,
      "step": 2888
    },
    {
      "epoch": 0.63679947098694,
      "grad_norm": 0.7338078618049622,
      "learning_rate": 6.394781091639072e-05,
      "loss": 2.0331,
      "step": 2889
    },
    {
      "epoch": 0.6370198930952775,
      "grad_norm": 0.703484296798706,
      "learning_rate": 6.387983156913056e-05,
      "loss": 2.0876,
      "step": 2890
    },
    {
      "epoch": 0.6372403152036149,
      "grad_norm": 0.7787843942642212,
      "learning_rate": 6.381187141275313e-05,
      "loss": 2.0908,
      "step": 2891
    },
    {
      "epoch": 0.6374607373119524,
      "grad_norm": 0.746830940246582,
      "learning_rate": 6.374393048336618e-05,
      "loss": 2.1165,
      "step": 2892
    },
    {
      "epoch": 0.6376811594202898,
      "grad_norm": 0.6967924237251282,
      "learning_rate": 6.367600881706712e-05,
      "loss": 2.0149,
      "step": 2893
    },
    {
      "epoch": 0.6379015815286273,
      "grad_norm": 0.7473634481430054,
      "learning_rate": 6.360810644994323e-05,
      "loss": 2.0632,
      "step": 2894
    },
    {
      "epoch": 0.6381220036369648,
      "grad_norm": 0.7151963114738464,
      "learning_rate": 6.354022341807144e-05,
      "loss": 2.0497,
      "step": 2895
    },
    {
      "epoch": 0.6383424257453022,
      "grad_norm": 0.6950897574424744,
      "learning_rate": 6.347235975751849e-05,
      "loss": 2.2002,
      "step": 2896
    },
    {
      "epoch": 0.6385628478536397,
      "grad_norm": 0.7444342970848083,
      "learning_rate": 6.340451550434074e-05,
      "loss": 2.0805,
      "step": 2897
    },
    {
      "epoch": 0.6387832699619772,
      "grad_norm": 0.7178138494491577,
      "learning_rate": 6.333669069458432e-05,
      "loss": 2.0337,
      "step": 2898
    },
    {
      "epoch": 0.6390036920703146,
      "grad_norm": 0.6804183125495911,
      "learning_rate": 6.3268885364285e-05,
      "loss": 2.0914,
      "step": 2899
    },
    {
      "epoch": 0.6392241141786521,
      "grad_norm": 0.7130212783813477,
      "learning_rate": 6.320109954946818e-05,
      "loss": 2.1284,
      "step": 2900
    },
    {
      "epoch": 0.6394445362869896,
      "grad_norm": 0.6899566054344177,
      "learning_rate": 6.313333328614892e-05,
      "loss": 2.0324,
      "step": 2901
    },
    {
      "epoch": 0.639664958395327,
      "grad_norm": 0.7451682090759277,
      "learning_rate": 6.306558661033184e-05,
      "loss": 2.0139,
      "step": 2902
    },
    {
      "epoch": 0.6398853805036645,
      "grad_norm": 0.7107532620429993,
      "learning_rate": 6.299785955801127e-05,
      "loss": 2.1105,
      "step": 2903
    },
    {
      "epoch": 0.640105802612002,
      "grad_norm": 0.7066952586174011,
      "learning_rate": 6.2930152165171e-05,
      "loss": 2.238,
      "step": 2904
    },
    {
      "epoch": 0.6403262247203394,
      "grad_norm": 0.7357462644577026,
      "learning_rate": 6.286246446778442e-05,
      "loss": 2.1168,
      "step": 2905
    },
    {
      "epoch": 0.6405466468286769,
      "grad_norm": 0.7621426582336426,
      "learning_rate": 6.279479650181442e-05,
      "loss": 2.1255,
      "step": 2906
    },
    {
      "epoch": 0.6407670689370144,
      "grad_norm": 0.7244865298271179,
      "learning_rate": 6.272714830321347e-05,
      "loss": 2.1372,
      "step": 2907
    },
    {
      "epoch": 0.6409874910453518,
      "grad_norm": 0.7078894972801208,
      "learning_rate": 6.265951990792347e-05,
      "loss": 1.9504,
      "step": 2908
    },
    {
      "epoch": 0.6412079131536893,
      "grad_norm": 0.77734375,
      "learning_rate": 6.259191135187585e-05,
      "loss": 1.9795,
      "step": 2909
    },
    {
      "epoch": 0.6414283352620268,
      "grad_norm": 0.7206412553787231,
      "learning_rate": 6.252432267099149e-05,
      "loss": 2.0562,
      "step": 2910
    },
    {
      "epoch": 0.6416487573703642,
      "grad_norm": 0.7021327018737793,
      "learning_rate": 6.245675390118071e-05,
      "loss": 2.0686,
      "step": 2911
    },
    {
      "epoch": 0.6418691794787017,
      "grad_norm": 0.7300772666931152,
      "learning_rate": 6.238920507834318e-05,
      "loss": 1.9931,
      "step": 2912
    },
    {
      "epoch": 0.6420896015870392,
      "grad_norm": 0.7183913588523865,
      "learning_rate": 6.232167623836809e-05,
      "loss": 2.1745,
      "step": 2913
    },
    {
      "epoch": 0.6423100236953766,
      "grad_norm": 0.7198596000671387,
      "learning_rate": 6.225416741713394e-05,
      "loss": 2.141,
      "step": 2914
    },
    {
      "epoch": 0.6425304458037141,
      "grad_norm": 0.721310555934906,
      "learning_rate": 6.21866786505086e-05,
      "loss": 2.2365,
      "step": 2915
    },
    {
      "epoch": 0.6427508679120516,
      "grad_norm": 0.6810656785964966,
      "learning_rate": 6.21192099743493e-05,
      "loss": 2.1059,
      "step": 2916
    },
    {
      "epoch": 0.642971290020389,
      "grad_norm": 0.7496649622917175,
      "learning_rate": 6.205176142450262e-05,
      "loss": 1.9943,
      "step": 2917
    },
    {
      "epoch": 0.6431917121287265,
      "grad_norm": 0.7024210095405579,
      "learning_rate": 6.198433303680439e-05,
      "loss": 1.9997,
      "step": 2918
    },
    {
      "epoch": 0.643412134237064,
      "grad_norm": 0.7470824718475342,
      "learning_rate": 6.191692484707973e-05,
      "loss": 2.1553,
      "step": 2919
    },
    {
      "epoch": 0.6436325563454014,
      "grad_norm": 0.682499349117279,
      "learning_rate": 6.18495368911431e-05,
      "loss": 2.1014,
      "step": 2920
    },
    {
      "epoch": 0.6438529784537389,
      "grad_norm": 0.780802309513092,
      "learning_rate": 6.178216920479815e-05,
      "loss": 2.141,
      "step": 2921
    },
    {
      "epoch": 0.6440734005620764,
      "grad_norm": 0.7788317203521729,
      "learning_rate": 6.171482182383777e-05,
      "loss": 2.1136,
      "step": 2922
    },
    {
      "epoch": 0.6442938226704138,
      "grad_norm": 0.7297016382217407,
      "learning_rate": 6.164749478404407e-05,
      "loss": 2.0537,
      "step": 2923
    },
    {
      "epoch": 0.6445142447787513,
      "grad_norm": 0.7366299033164978,
      "learning_rate": 6.158018812118834e-05,
      "loss": 2.0412,
      "step": 2924
    },
    {
      "epoch": 0.6447346668870888,
      "grad_norm": 0.7722187638282776,
      "learning_rate": 6.151290187103104e-05,
      "loss": 2.1227,
      "step": 2925
    },
    {
      "epoch": 0.6449550889954262,
      "grad_norm": 0.7546471357345581,
      "learning_rate": 6.144563606932183e-05,
      "loss": 2.0801,
      "step": 2926
    },
    {
      "epoch": 0.6451755111037637,
      "grad_norm": 0.7677512764930725,
      "learning_rate": 6.137839075179944e-05,
      "loss": 2.027,
      "step": 2927
    },
    {
      "epoch": 0.6453959332121012,
      "grad_norm": 0.7305790781974792,
      "learning_rate": 6.131116595419178e-05,
      "loss": 2.0213,
      "step": 2928
    },
    {
      "epoch": 0.6456163553204386,
      "grad_norm": 0.6968057155609131,
      "learning_rate": 6.124396171221579e-05,
      "loss": 2.0017,
      "step": 2929
    },
    {
      "epoch": 0.6458367774287761,
      "grad_norm": 0.6797564625740051,
      "learning_rate": 6.117677806157755e-05,
      "loss": 2.146,
      "step": 2930
    },
    {
      "epoch": 0.6460571995371136,
      "grad_norm": 0.7572391629219055,
      "learning_rate": 6.110961503797219e-05,
      "loss": 2.0413,
      "step": 2931
    },
    {
      "epoch": 0.646277621645451,
      "grad_norm": 0.7696342468261719,
      "learning_rate": 6.104247267708386e-05,
      "loss": 2.0822,
      "step": 2932
    },
    {
      "epoch": 0.6464980437537885,
      "grad_norm": 0.7011362910270691,
      "learning_rate": 6.0975351014585716e-05,
      "loss": 2.0056,
      "step": 2933
    },
    {
      "epoch": 0.646718465862126,
      "grad_norm": 0.699828028678894,
      "learning_rate": 6.090825008613993e-05,
      "loss": 2.087,
      "step": 2934
    },
    {
      "epoch": 0.6469388879704634,
      "grad_norm": 0.748678982257843,
      "learning_rate": 6.084116992739769e-05,
      "loss": 2.1416,
      "step": 2935
    },
    {
      "epoch": 0.6471593100788009,
      "grad_norm": 0.6666243672370911,
      "learning_rate": 6.077411057399912e-05,
      "loss": 2.0836,
      "step": 2936
    },
    {
      "epoch": 0.6473797321871384,
      "grad_norm": 0.6745938658714294,
      "learning_rate": 6.070707206157329e-05,
      "loss": 2.128,
      "step": 2937
    },
    {
      "epoch": 0.6476001542954758,
      "grad_norm": 0.7773693799972534,
      "learning_rate": 6.064005442573824e-05,
      "loss": 2.109,
      "step": 2938
    },
    {
      "epoch": 0.6478205764038133,
      "grad_norm": 0.7589658498764038,
      "learning_rate": 6.05730577021008e-05,
      "loss": 2.0394,
      "step": 2939
    },
    {
      "epoch": 0.6480409985121508,
      "grad_norm": 0.7654257416725159,
      "learning_rate": 6.050608192625683e-05,
      "loss": 2.1764,
      "step": 2940
    },
    {
      "epoch": 0.6482614206204882,
      "grad_norm": 0.7441624402999878,
      "learning_rate": 6.0439127133790984e-05,
      "loss": 2.0706,
      "step": 2941
    },
    {
      "epoch": 0.6484818427288257,
      "grad_norm": 0.7324864864349365,
      "learning_rate": 6.0372193360276776e-05,
      "loss": 2.1275,
      "step": 2942
    },
    {
      "epoch": 0.6487022648371632,
      "grad_norm": 0.7897171378135681,
      "learning_rate": 6.030528064127659e-05,
      "loss": 2.2662,
      "step": 2943
    },
    {
      "epoch": 0.6489226869455006,
      "grad_norm": 0.7587531805038452,
      "learning_rate": 6.0238389012341554e-05,
      "loss": 2.0397,
      "step": 2944
    },
    {
      "epoch": 0.6491431090538381,
      "grad_norm": 0.6990668773651123,
      "learning_rate": 6.017151850901166e-05,
      "loss": 2.066,
      "step": 2945
    },
    {
      "epoch": 0.6493635311621756,
      "grad_norm": 0.7191402912139893,
      "learning_rate": 6.0104669166815606e-05,
      "loss": 1.9778,
      "step": 2946
    },
    {
      "epoch": 0.649583953270513,
      "grad_norm": 0.720093309879303,
      "learning_rate": 6.0037841021270924e-05,
      "loss": 2.0368,
      "step": 2947
    },
    {
      "epoch": 0.6498043753788505,
      "grad_norm": 0.8040246367454529,
      "learning_rate": 5.997103410788385e-05,
      "loss": 2.1776,
      "step": 2948
    },
    {
      "epoch": 0.650024797487188,
      "grad_norm": 0.6664162874221802,
      "learning_rate": 5.990424846214931e-05,
      "loss": 2.0627,
      "step": 2949
    },
    {
      "epoch": 0.6502452195955254,
      "grad_norm": 0.6916616559028625,
      "learning_rate": 5.983748411955097e-05,
      "loss": 2.0742,
      "step": 2950
    },
    {
      "epoch": 0.6504656417038629,
      "grad_norm": 0.7645388841629028,
      "learning_rate": 5.977074111556116e-05,
      "loss": 2.156,
      "step": 2951
    },
    {
      "epoch": 0.6506860638122004,
      "grad_norm": 0.7310187816619873,
      "learning_rate": 5.970401948564088e-05,
      "loss": 2.0877,
      "step": 2952
    },
    {
      "epoch": 0.6509064859205378,
      "grad_norm": 0.7291997671127319,
      "learning_rate": 5.963731926523975e-05,
      "loss": 2.0983,
      "step": 2953
    },
    {
      "epoch": 0.6511269080288753,
      "grad_norm": 0.7307660579681396,
      "learning_rate": 5.957064048979608e-05,
      "loss": 2.095,
      "step": 2954
    },
    {
      "epoch": 0.6513473301372128,
      "grad_norm": 0.7694621086120605,
      "learning_rate": 5.950398319473669e-05,
      "loss": 2.1962,
      "step": 2955
    },
    {
      "epoch": 0.6515677522455502,
      "grad_norm": 0.708839476108551,
      "learning_rate": 5.943734741547706e-05,
      "loss": 2.0532,
      "step": 2956
    },
    {
      "epoch": 0.6517881743538877,
      "grad_norm": 0.7092377543449402,
      "learning_rate": 5.937073318742123e-05,
      "loss": 2.0697,
      "step": 2957
    },
    {
      "epoch": 0.6520085964622252,
      "grad_norm": 0.7080748081207275,
      "learning_rate": 5.9304140545961784e-05,
      "loss": 2.155,
      "step": 2958
    },
    {
      "epoch": 0.6522290185705626,
      "grad_norm": 0.7433266639709473,
      "learning_rate": 5.923756952647982e-05,
      "loss": 2.0439,
      "step": 2959
    },
    {
      "epoch": 0.6524494406789001,
      "grad_norm": 0.7173716425895691,
      "learning_rate": 5.917102016434494e-05,
      "loss": 2.0843,
      "step": 2960
    },
    {
      "epoch": 0.6526698627872376,
      "grad_norm": 0.7700660228729248,
      "learning_rate": 5.9104492494915254e-05,
      "loss": 2.0835,
      "step": 2961
    },
    {
      "epoch": 0.652890284895575,
      "grad_norm": 0.7564086318016052,
      "learning_rate": 5.903798655353738e-05,
      "loss": 2.0403,
      "step": 2962
    },
    {
      "epoch": 0.6531107070039125,
      "grad_norm": 0.6480859518051147,
      "learning_rate": 5.8971502375546336e-05,
      "loss": 2.1212,
      "step": 2963
    },
    {
      "epoch": 0.65333112911225,
      "grad_norm": 0.6917101144790649,
      "learning_rate": 5.8905039996265624e-05,
      "loss": 2.1357,
      "step": 2964
    },
    {
      "epoch": 0.6535515512205874,
      "grad_norm": 0.6772432327270508,
      "learning_rate": 5.883859945100716e-05,
      "loss": 2.1865,
      "step": 2965
    },
    {
      "epoch": 0.6537719733289249,
      "grad_norm": 0.7465474605560303,
      "learning_rate": 5.8772180775071174e-05,
      "loss": 2.0916,
      "step": 2966
    },
    {
      "epoch": 0.6539923954372624,
      "grad_norm": 0.7724354863166809,
      "learning_rate": 5.87057840037464e-05,
      "loss": 2.1359,
      "step": 2967
    },
    {
      "epoch": 0.6542128175455998,
      "grad_norm": 0.735632061958313,
      "learning_rate": 5.863940917230986e-05,
      "loss": 2.1639,
      "step": 2968
    },
    {
      "epoch": 0.6544332396539373,
      "grad_norm": 0.7477682828903198,
      "learning_rate": 5.857305631602692e-05,
      "loss": 2.0256,
      "step": 2969
    },
    {
      "epoch": 0.6546536617622748,
      "grad_norm": 0.7738300561904907,
      "learning_rate": 5.850672547015134e-05,
      "loss": 2.275,
      "step": 2970
    },
    {
      "epoch": 0.6548740838706122,
      "grad_norm": 0.7073681354522705,
      "learning_rate": 5.84404166699251e-05,
      "loss": 2.0635,
      "step": 2971
    },
    {
      "epoch": 0.6550945059789497,
      "grad_norm": 0.6796169877052307,
      "learning_rate": 5.8374129950578503e-05,
      "loss": 2.0989,
      "step": 2972
    },
    {
      "epoch": 0.6553149280872872,
      "grad_norm": 0.7016650438308716,
      "learning_rate": 5.830786534733012e-05,
      "loss": 2.1056,
      "step": 2973
    },
    {
      "epoch": 0.6555353501956246,
      "grad_norm": 0.7666334509849548,
      "learning_rate": 5.824162289538676e-05,
      "loss": 2.1243,
      "step": 2974
    },
    {
      "epoch": 0.6557557723039621,
      "grad_norm": 0.7061583995819092,
      "learning_rate": 5.8175402629943496e-05,
      "loss": 1.9987,
      "step": 2975
    },
    {
      "epoch": 0.6559761944122996,
      "grad_norm": 0.7272177338600159,
      "learning_rate": 5.810920458618356e-05,
      "loss": 1.969,
      "step": 2976
    },
    {
      "epoch": 0.656196616520637,
      "grad_norm": 0.7255079746246338,
      "learning_rate": 5.8043028799278446e-05,
      "loss": 1.9941,
      "step": 2977
    },
    {
      "epoch": 0.6564170386289745,
      "grad_norm": 0.7042426466941833,
      "learning_rate": 5.7976875304387756e-05,
      "loss": 2.0903,
      "step": 2978
    },
    {
      "epoch": 0.656637460737312,
      "grad_norm": 0.7679576277732849,
      "learning_rate": 5.791074413665932e-05,
      "loss": 2.0306,
      "step": 2979
    },
    {
      "epoch": 0.6568578828456494,
      "grad_norm": 0.748751699924469,
      "learning_rate": 5.784463533122899e-05,
      "loss": 2.0748,
      "step": 2980
    },
    {
      "epoch": 0.6570783049539869,
      "grad_norm": 0.7553817629814148,
      "learning_rate": 5.7778548923220844e-05,
      "loss": 2.1275,
      "step": 2981
    },
    {
      "epoch": 0.6572987270623244,
      "grad_norm": 0.7478180527687073,
      "learning_rate": 5.771248494774699e-05,
      "loss": 1.968,
      "step": 2982
    },
    {
      "epoch": 0.6575191491706618,
      "grad_norm": 0.7663314938545227,
      "learning_rate": 5.7646443439907715e-05,
      "loss": 2.0279,
      "step": 2983
    },
    {
      "epoch": 0.6577395712789993,
      "grad_norm": 0.7646080851554871,
      "learning_rate": 5.758042443479128e-05,
      "loss": 2.1587,
      "step": 2984
    },
    {
      "epoch": 0.6579599933873368,
      "grad_norm": 0.735347330570221,
      "learning_rate": 5.7514427967474026e-05,
      "loss": 2.1234,
      "step": 2985
    },
    {
      "epoch": 0.6581804154956742,
      "grad_norm": 0.7195048928260803,
      "learning_rate": 5.744845407302028e-05,
      "loss": 2.0767,
      "step": 2986
    },
    {
      "epoch": 0.6584008376040117,
      "grad_norm": 0.7577106952667236,
      "learning_rate": 5.738250278648245e-05,
      "loss": 2.068,
      "step": 2987
    },
    {
      "epoch": 0.6586212597123492,
      "grad_norm": 0.7048833966255188,
      "learning_rate": 5.731657414290085e-05,
      "loss": 2.0141,
      "step": 2988
    },
    {
      "epoch": 0.6588416818206866,
      "grad_norm": 0.7207455635070801,
      "learning_rate": 5.725066817730385e-05,
      "loss": 2.0242,
      "step": 2989
    },
    {
      "epoch": 0.6590621039290241,
      "grad_norm": 0.7409653663635254,
      "learning_rate": 5.718478492470765e-05,
      "loss": 2.1059,
      "step": 2990
    },
    {
      "epoch": 0.6592825260373616,
      "grad_norm": 0.7787463068962097,
      "learning_rate": 5.711892442011648e-05,
      "loss": 2.0679,
      "step": 2991
    },
    {
      "epoch": 0.659502948145699,
      "grad_norm": 0.721291184425354,
      "learning_rate": 5.705308669852247e-05,
      "loss": 2.1185,
      "step": 2992
    },
    {
      "epoch": 0.6597233702540365,
      "grad_norm": 0.7927785515785217,
      "learning_rate": 5.698727179490565e-05,
      "loss": 2.2269,
      "step": 2993
    },
    {
      "epoch": 0.659943792362374,
      "grad_norm": 0.7021698951721191,
      "learning_rate": 5.6921479744233874e-05,
      "loss": 2.2898,
      "step": 2994
    },
    {
      "epoch": 0.6601642144707114,
      "grad_norm": 0.7544586658477783,
      "learning_rate": 5.685571058146293e-05,
      "loss": 2.0909,
      "step": 2995
    },
    {
      "epoch": 0.6603846365790489,
      "grad_norm": 0.7370157241821289,
      "learning_rate": 5.6789964341536375e-05,
      "loss": 1.9811,
      "step": 2996
    },
    {
      "epoch": 0.6606050586873864,
      "grad_norm": 0.7332147359848022,
      "learning_rate": 5.672424105938563e-05,
      "loss": 2.0007,
      "step": 2997
    },
    {
      "epoch": 0.6608254807957238,
      "grad_norm": 0.6672928929328918,
      "learning_rate": 5.665854076992991e-05,
      "loss": 2.1227,
      "step": 2998
    },
    {
      "epoch": 0.6610459029040613,
      "grad_norm": 0.7413603067398071,
      "learning_rate": 5.6592863508076197e-05,
      "loss": 2.2205,
      "step": 2999
    },
    {
      "epoch": 0.6612663250123988,
      "grad_norm": 0.7109661102294922,
      "learning_rate": 5.6527209308719306e-05,
      "loss": 2.1028,
      "step": 3000
    },
    {
      "epoch": 0.6612663250123988,
      "eval_bpb": 1.082199664815407,
      "eval_loss": 2.0859575271606445,
      "eval_runtime": 184.3007,
      "eval_samples_per_second": 62.181,
      "eval_steps_per_second": 7.775,
      "step": 3000
    },
    {
      "epoch": 0.6614867471207362,
      "grad_norm": 0.7366294264793396,
      "learning_rate": 5.646157820674167e-05,
      "loss": 2.0001,
      "step": 3001
    },
    {
      "epoch": 0.6617071692290737,
      "grad_norm": 0.7309939861297607,
      "learning_rate": 5.639597023701355e-05,
      "loss": 1.9511,
      "step": 3002
    },
    {
      "epoch": 0.6619275913374112,
      "grad_norm": 0.6854066848754883,
      "learning_rate": 5.6330385434392896e-05,
      "loss": 2.0247,
      "step": 3003
    },
    {
      "epoch": 0.6621480134457486,
      "grad_norm": 0.6652411222457886,
      "learning_rate": 5.626482383372534e-05,
      "loss": 1.972,
      "step": 3004
    },
    {
      "epoch": 0.6623684355540861,
      "grad_norm": 0.7634361386299133,
      "learning_rate": 5.61992854698442e-05,
      "loss": 2.0984,
      "step": 3005
    },
    {
      "epoch": 0.6625888576624236,
      "grad_norm": 0.7446879148483276,
      "learning_rate": 5.613377037757044e-05,
      "loss": 2.1057,
      "step": 3006
    },
    {
      "epoch": 0.662809279770761,
      "grad_norm": 0.7614639401435852,
      "learning_rate": 5.606827859171264e-05,
      "loss": 2.1308,
      "step": 3007
    },
    {
      "epoch": 0.6630297018790985,
      "grad_norm": 0.7633336782455444,
      "learning_rate": 5.600281014706703e-05,
      "loss": 2.0163,
      "step": 3008
    },
    {
      "epoch": 0.663250123987436,
      "grad_norm": 0.735965371131897,
      "learning_rate": 5.5937365078417404e-05,
      "loss": 2.0814,
      "step": 3009
    },
    {
      "epoch": 0.6634705460957734,
      "grad_norm": 0.8183976411819458,
      "learning_rate": 5.587194342053524e-05,
      "loss": 2.0258,
      "step": 3010
    },
    {
      "epoch": 0.6636909682041109,
      "grad_norm": 0.7061830163002014,
      "learning_rate": 5.5806545208179406e-05,
      "loss": 2.134,
      "step": 3011
    },
    {
      "epoch": 0.6639113903124484,
      "grad_norm": 0.7769129276275635,
      "learning_rate": 5.5741170476096415e-05,
      "loss": 2.1844,
      "step": 3012
    },
    {
      "epoch": 0.6641318124207858,
      "grad_norm": 0.7672994136810303,
      "learning_rate": 5.56758192590203e-05,
      "loss": 2.0587,
      "step": 3013
    },
    {
      "epoch": 0.6643522345291233,
      "grad_norm": 0.7826948761940002,
      "learning_rate": 5.561049159167262e-05,
      "loss": 2.2472,
      "step": 3014
    },
    {
      "epoch": 0.6645726566374608,
      "grad_norm": 0.7515006065368652,
      "learning_rate": 5.554518750876234e-05,
      "loss": 2.144,
      "step": 3015
    },
    {
      "epoch": 0.6647930787457982,
      "grad_norm": 0.6935786008834839,
      "learning_rate": 5.5479907044986004e-05,
      "loss": 2.0251,
      "step": 3016
    },
    {
      "epoch": 0.6650135008541357,
      "grad_norm": 0.7365511059761047,
      "learning_rate": 5.541465023502752e-05,
      "loss": 1.978,
      "step": 3017
    },
    {
      "epoch": 0.6652339229624732,
      "grad_norm": 0.7079964280128479,
      "learning_rate": 5.5349417113558254e-05,
      "loss": 2.1767,
      "step": 3018
    },
    {
      "epoch": 0.6654543450708106,
      "grad_norm": 0.7057011723518372,
      "learning_rate": 5.5284207715237014e-05,
      "loss": 2.1184,
      "step": 3019
    },
    {
      "epoch": 0.6656747671791481,
      "grad_norm": 0.7483015060424805,
      "learning_rate": 5.521902207470997e-05,
      "loss": 2.0698,
      "step": 3020
    },
    {
      "epoch": 0.6658951892874856,
      "grad_norm": 0.7887651920318604,
      "learning_rate": 5.5153860226610734e-05,
      "loss": 2.1857,
      "step": 3021
    },
    {
      "epoch": 0.666115611395823,
      "grad_norm": 0.7393646240234375,
      "learning_rate": 5.508872220556012e-05,
      "loss": 2.1052,
      "step": 3022
    },
    {
      "epoch": 0.6663360335041605,
      "grad_norm": 0.7663282752037048,
      "learning_rate": 5.502360804616644e-05,
      "loss": 2.0795,
      "step": 3023
    },
    {
      "epoch": 0.666556455612498,
      "grad_norm": 0.7211540341377258,
      "learning_rate": 5.4958517783025275e-05,
      "loss": 2.0242,
      "step": 3024
    }
  ],
  "logging_steps": 1,
  "max_steps": 4537,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 378,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 1.185047088765862e+18,
  "train_batch_size": 6,
  "trial_name": null,
  "trial_params": null
}