{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 741,
  "global_step": 2961,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00033772374197906115,
      "grad_norm": 0.8691335320472717,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 1.4523,
      "step": 1
    },
    {
      "epoch": 0.00033772374197906115,
      "eval_loss": 1.2390516996383667,
      "eval_runtime": 348.2193,
      "eval_samples_per_second": 8.535,
      "eval_steps_per_second": 1.068,
      "step": 1
    },
    {
      "epoch": 0.0006754474839581223,
      "grad_norm": 0.7917861342430115,
      "learning_rate": 4.000000000000001e-06,
      "loss": 1.4245,
      "step": 2
    },
    {
      "epoch": 0.0010131712259371835,
      "grad_norm": 0.8042817115783691,
      "learning_rate": 6e-06,
      "loss": 1.2222,
      "step": 3
    },
    {
      "epoch": 0.0013508949679162446,
      "grad_norm": 0.7690820097923279,
      "learning_rate": 8.000000000000001e-06,
      "loss": 1.5285,
      "step": 4
    },
    {
      "epoch": 0.0016886187098953055,
      "grad_norm": 0.7847064137458801,
      "learning_rate": 1e-05,
      "loss": 1.4399,
      "step": 5
    },
    {
      "epoch": 0.002026342451874367,
      "grad_norm": 0.7379052639007568,
      "learning_rate": 1.2e-05,
      "loss": 1.4816,
      "step": 6
    },
    {
      "epoch": 0.002364066193853428,
      "grad_norm": 0.8600063920021057,
      "learning_rate": 1.4e-05,
      "loss": 1.3804,
      "step": 7
    },
    {
      "epoch": 0.002701789935832489,
      "grad_norm": 0.7446380853652954,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 1.4849,
      "step": 8
    },
    {
      "epoch": 0.00303951367781155,
      "grad_norm": 0.7647069096565247,
      "learning_rate": 1.8e-05,
      "loss": 1.4933,
      "step": 9
    },
    {
      "epoch": 0.003377237419790611,
      "grad_norm": 0.8156012892723083,
      "learning_rate": 2e-05,
      "loss": 1.3997,
      "step": 10
    },
    {
      "epoch": 0.0037149611617696724,
      "grad_norm": 0.7134099006652832,
      "learning_rate": 1.999999433328606e-05,
      "loss": 1.4615,
      "step": 11
    },
    {
      "epoch": 0.004052684903748734,
      "grad_norm": 0.6633527874946594,
      "learning_rate": 1.999997733315066e-05,
      "loss": 1.3827,
      "step": 12
    },
    {
      "epoch": 0.004390408645727794,
      "grad_norm": 0.7372402548789978,
      "learning_rate": 1.9999948999613074e-05,
      "loss": 1.4183,
      "step": 13
    },
    {
      "epoch": 0.004728132387706856,
      "grad_norm": 0.6915538907051086,
      "learning_rate": 1.9999909332705403e-05,
      "loss": 1.5101,
      "step": 14
    },
    {
      "epoch": 0.005065856129685917,
      "grad_norm": 0.5909135937690735,
      "learning_rate": 1.999985833247261e-05,
      "loss": 1.5618,
      "step": 15
    },
    {
      "epoch": 0.005403579871664978,
      "grad_norm": 0.6857839822769165,
      "learning_rate": 1.999979599897249e-05,
      "loss": 1.2789,
      "step": 16
    },
    {
      "epoch": 0.005741303613644039,
      "grad_norm": 0.6145290732383728,
      "learning_rate": 1.99997223322757e-05,
      "loss": 1.4605,
      "step": 17
    },
    {
      "epoch": 0.0060790273556231,
      "grad_norm": 0.6638191938400269,
      "learning_rate": 1.999963733246572e-05,
      "loss": 1.4164,
      "step": 18
    },
    {
      "epoch": 0.006416751097602162,
      "grad_norm": 0.7354663014411926,
      "learning_rate": 1.9999540999638882e-05,
      "loss": 1.3611,
      "step": 19
    },
    {
      "epoch": 0.006754474839581222,
      "grad_norm": 0.7099593281745911,
      "learning_rate": 1.999943333390437e-05,
      "loss": 1.4501,
      "step": 20
    },
    {
      "epoch": 0.0070921985815602835,
      "grad_norm": 0.6876184344291687,
      "learning_rate": 1.9999314335384205e-05,
      "loss": 1.3233,
      "step": 21
    },
    {
      "epoch": 0.007429922323539345,
      "grad_norm": 0.6709492802619934,
      "learning_rate": 1.9999184004213253e-05,
      "loss": 1.3322,
      "step": 22
    },
    {
      "epoch": 0.007767646065518406,
      "grad_norm": 0.7476434111595154,
      "learning_rate": 1.999904234053922e-05,
      "loss": 1.251,
      "step": 23
    },
    {
      "epoch": 0.008105369807497468,
      "grad_norm": 0.6507943272590637,
      "learning_rate": 1.9998889344522668e-05,
      "loss": 1.4471,
      "step": 24
    },
    {
      "epoch": 0.008443093549476529,
      "grad_norm": 0.6675460338592529,
      "learning_rate": 1.9998725016336982e-05,
      "loss": 1.3671,
      "step": 25
    },
    {
      "epoch": 0.008780817291455589,
      "grad_norm": 0.6601463556289673,
      "learning_rate": 1.9998549356168415e-05,
      "loss": 1.36,
      "step": 26
    },
    {
      "epoch": 0.00911854103343465,
      "grad_norm": 0.6329362988471985,
      "learning_rate": 1.999836236421604e-05,
      "loss": 1.3317,
      "step": 27
    },
    {
      "epoch": 0.009456264775413711,
      "grad_norm": 0.7787251472473145,
      "learning_rate": 1.9998164040691787e-05,
      "loss": 1.22,
      "step": 28
    },
    {
      "epoch": 0.009793988517392773,
      "grad_norm": 0.752509593963623,
      "learning_rate": 1.999795438582043e-05,
      "loss": 1.2085,
      "step": 29
    },
    {
      "epoch": 0.010131712259371834,
      "grad_norm": 0.6960368156433105,
      "learning_rate": 1.999773339983957e-05,
      "loss": 1.1876,
      "step": 30
    },
    {
      "epoch": 0.010469436001350895,
      "grad_norm": 0.7400830388069153,
      "learning_rate": 1.9997501082999663e-05,
      "loss": 1.2809,
      "step": 31
    },
    {
      "epoch": 0.010807159743329957,
      "grad_norm": 0.7028891444206238,
      "learning_rate": 1.9997257435564008e-05,
      "loss": 1.3177,
      "step": 32
    },
    {
      "epoch": 0.011144883485309016,
      "grad_norm": 0.7228677868843079,
      "learning_rate": 1.9997002457808735e-05,
      "loss": 1.4775,
      "step": 33
    },
    {
      "epoch": 0.011482607227288078,
      "grad_norm": 0.6896461248397827,
      "learning_rate": 1.9996736150022826e-05,
      "loss": 1.3044,
      "step": 34
    },
    {
      "epoch": 0.01182033096926714,
      "grad_norm": 0.8067850470542908,
      "learning_rate": 1.9996458512508095e-05,
      "loss": 1.1402,
      "step": 35
    },
    {
      "epoch": 0.0121580547112462,
      "grad_norm": 0.733795702457428,
      "learning_rate": 1.9996169545579205e-05,
      "loss": 1.228,
      "step": 36
    },
    {
      "epoch": 0.012495778453225262,
      "grad_norm": 0.7718669176101685,
      "learning_rate": 1.9995869249563653e-05,
      "loss": 1.1756,
      "step": 37
    },
    {
      "epoch": 0.012833502195204323,
      "grad_norm": 0.6810094118118286,
      "learning_rate": 1.999555762480178e-05,
      "loss": 1.3362,
      "step": 38
    },
    {
      "epoch": 0.013171225937183385,
      "grad_norm": 0.735789954662323,
      "learning_rate": 1.9995234671646753e-05,
      "loss": 1.2539,
      "step": 39
    },
    {
      "epoch": 0.013508949679162444,
      "grad_norm": 0.9424735307693481,
      "learning_rate": 1.9994900390464598e-05,
      "loss": 0.9794,
      "step": 40
    },
    {
      "epoch": 0.013846673421141506,
      "grad_norm": 0.8318026065826416,
      "learning_rate": 1.9994554781634168e-05,
      "loss": 1.1921,
      "step": 41
    },
    {
      "epoch": 0.014184397163120567,
      "grad_norm": 0.8098856806755066,
      "learning_rate": 1.9994197845547158e-05,
      "loss": 1.2707,
      "step": 42
    },
    {
      "epoch": 0.014522120905099628,
      "grad_norm": 0.7989124059677124,
      "learning_rate": 1.9993829582608093e-05,
      "loss": 1.1952,
      "step": 43
    },
    {
      "epoch": 0.01485984464707869,
      "grad_norm": 0.7633487582206726,
      "learning_rate": 1.9993449993234352e-05,
      "loss": 1.338,
      "step": 44
    },
    {
      "epoch": 0.015197568389057751,
      "grad_norm": 0.7575072050094604,
      "learning_rate": 1.999305907785613e-05,
      "loss": 1.1802,
      "step": 45
    },
    {
      "epoch": 0.015535292131036813,
      "grad_norm": 0.8352583646774292,
      "learning_rate": 1.999265683691647e-05,
      "loss": 1.153,
      "step": 46
    },
    {
      "epoch": 0.015873015873015872,
      "grad_norm": 0.8807387948036194,
      "learning_rate": 1.999224327087125e-05,
      "loss": 1.1068,
      "step": 47
    },
    {
      "epoch": 0.016210739614994935,
      "grad_norm": 0.8409246802330017,
      "learning_rate": 1.9991818380189183e-05,
      "loss": 1.1885,
      "step": 48
    },
    {
      "epoch": 0.016548463356973995,
      "grad_norm": 0.8024547696113586,
      "learning_rate": 1.9991382165351816e-05,
      "loss": 1.2001,
      "step": 49
    },
    {
      "epoch": 0.016886187098953058,
      "grad_norm": 0.8989960551261902,
      "learning_rate": 1.9990934626853527e-05,
      "loss": 1.2516,
      "step": 50
    },
    {
      "epoch": 0.017223910840932118,
      "grad_norm": 0.8436941504478455,
      "learning_rate": 1.9990475765201537e-05,
      "loss": 1.223,
      "step": 51
    },
    {
      "epoch": 0.017561634582911177,
      "grad_norm": 1.0476566553115845,
      "learning_rate": 1.9990005580915887e-05,
      "loss": 1.2454,
      "step": 52
    },
    {
      "epoch": 0.01789935832489024,
      "grad_norm": 0.87506502866745,
      "learning_rate": 1.9989524074529455e-05,
      "loss": 1.2538,
      "step": 53
    },
    {
      "epoch": 0.0182370820668693,
      "grad_norm": 0.9177961945533752,
      "learning_rate": 1.9989031246587963e-05,
      "loss": 1.2544,
      "step": 54
    },
    {
      "epoch": 0.018574805808848363,
      "grad_norm": 0.795853853225708,
      "learning_rate": 1.9988527097649943e-05,
      "loss": 1.2602,
      "step": 55
    },
    {
      "epoch": 0.018912529550827423,
      "grad_norm": 0.9396657347679138,
      "learning_rate": 1.9988011628286776e-05,
      "loss": 1.1491,
      "step": 56
    },
    {
      "epoch": 0.019250253292806486,
      "grad_norm": 1.0028746128082275,
      "learning_rate": 1.9987484839082666e-05,
      "loss": 1.1566,
      "step": 57
    },
    {
      "epoch": 0.019587977034785545,
      "grad_norm": 1.0788178443908691,
      "learning_rate": 1.9986946730634638e-05,
      "loss": 1.1646,
      "step": 58
    },
    {
      "epoch": 0.019925700776764605,
      "grad_norm": 1.1312810182571411,
      "learning_rate": 1.9986397303552557e-05,
      "loss": 1.0336,
      "step": 59
    },
    {
      "epoch": 0.020263424518743668,
      "grad_norm": 0.9251755475997925,
      "learning_rate": 1.9985836558459118e-05,
      "loss": 1.2061,
      "step": 60
    },
    {
      "epoch": 0.020601148260722728,
      "grad_norm": 0.9717414379119873,
      "learning_rate": 1.998526449598983e-05,
      "loss": 1.2481,
      "step": 61
    },
    {
      "epoch": 0.02093887200270179,
      "grad_norm": 1.037160038948059,
      "learning_rate": 1.998468111679304e-05,
      "loss": 1.1242,
      "step": 62
    },
    {
      "epoch": 0.02127659574468085,
      "grad_norm": 1.0893983840942383,
      "learning_rate": 1.9984086421529913e-05,
      "loss": 1.2112,
      "step": 63
    },
    {
      "epoch": 0.021614319486659914,
      "grad_norm": 1.0214881896972656,
      "learning_rate": 1.9983480410874444e-05,
      "loss": 1.0843,
      "step": 64
    },
    {
      "epoch": 0.021952043228638973,
      "grad_norm": 1.002325415611267,
      "learning_rate": 1.9982863085513455e-05,
      "loss": 1.1636,
      "step": 65
    },
    {
      "epoch": 0.022289766970618033,
      "grad_norm": 1.0469549894332886,
      "learning_rate": 1.998223444614658e-05,
      "loss": 1.0943,
      "step": 66
    },
    {
      "epoch": 0.022627490712597096,
      "grad_norm": 1.1922545433044434,
      "learning_rate": 1.998159449348629e-05,
      "loss": 1.1239,
      "step": 67
    },
    {
      "epoch": 0.022965214454576156,
      "grad_norm": 1.025060772895813,
      "learning_rate": 1.9980943228257867e-05,
      "loss": 1.2297,
      "step": 68
    },
    {
      "epoch": 0.02330293819655522,
      "grad_norm": 1.2323729991912842,
      "learning_rate": 1.998028065119942e-05,
      "loss": 1.0179,
      "step": 69
    },
    {
      "epoch": 0.02364066193853428,
      "grad_norm": 1.190319299697876,
      "learning_rate": 1.9979606763061866e-05,
      "loss": 1.1217,
      "step": 70
    },
    {
      "epoch": 0.02397838568051334,
      "grad_norm": 1.1796809434890747,
      "learning_rate": 1.9978921564608968e-05,
      "loss": 1.21,
      "step": 71
    },
    {
      "epoch": 0.0243161094224924,
      "grad_norm": 1.1037620306015015,
      "learning_rate": 1.9978225056617278e-05,
      "loss": 1.1952,
      "step": 72
    },
    {
      "epoch": 0.02465383316447146,
      "grad_norm": 1.0818530321121216,
      "learning_rate": 1.997751723987618e-05,
      "loss": 1.2662,
      "step": 73
    },
    {
      "epoch": 0.024991556906450524,
      "grad_norm": 1.2175343036651611,
      "learning_rate": 1.997679811518788e-05,
      "loss": 1.0332,
      "step": 74
    },
    {
      "epoch": 0.025329280648429583,
      "grad_norm": 1.0987571477890015,
      "learning_rate": 1.9976067683367388e-05,
      "loss": 1.2153,
      "step": 75
    },
    {
      "epoch": 0.025667004390408647,
      "grad_norm": 1.198886752128601,
      "learning_rate": 1.997532594524253e-05,
      "loss": 0.9888,
      "step": 76
    },
    {
      "epoch": 0.026004728132387706,
      "grad_norm": 1.145399570465088,
      "learning_rate": 1.997457290165396e-05,
      "loss": 1.1934,
      "step": 77
    },
    {
      "epoch": 0.02634245187436677,
      "grad_norm": 1.0562129020690918,
      "learning_rate": 1.997380855345512e-05,
      "loss": 1.1233,
      "step": 78
    },
    {
      "epoch": 0.02668017561634583,
      "grad_norm": 1.0568044185638428,
      "learning_rate": 1.997303290151229e-05,
      "loss": 1.0638,
      "step": 79
    },
    {
      "epoch": 0.02701789935832489,
      "grad_norm": 1.056656002998352,
      "learning_rate": 1.9972245946704545e-05,
      "loss": 1.1706,
      "step": 80
    },
    {
      "epoch": 0.02735562310030395,
      "grad_norm": 0.9692465662956238,
      "learning_rate": 1.9971447689923777e-05,
      "loss": 1.2065,
      "step": 81
    },
    {
      "epoch": 0.02769334684228301,
      "grad_norm": 1.0248713493347168,
      "learning_rate": 1.9970638132074682e-05,
      "loss": 1.1078,
      "step": 82
    },
    {
      "epoch": 0.028031070584262074,
      "grad_norm": 1.0774048566818237,
      "learning_rate": 1.996981727407477e-05,
      "loss": 1.0891,
      "step": 83
    },
    {
      "epoch": 0.028368794326241134,
      "grad_norm": 1.0754584074020386,
      "learning_rate": 1.9968985116854348e-05,
      "loss": 1.2111,
      "step": 84
    },
    {
      "epoch": 0.028706518068220197,
      "grad_norm": 1.2687808275222778,
      "learning_rate": 1.996814166135654e-05,
      "loss": 1.1219,
      "step": 85
    },
    {
      "epoch": 0.029044241810199257,
      "grad_norm": 1.130565881729126,
      "learning_rate": 1.9967286908537273e-05,
      "loss": 1.0994,
      "step": 86
    },
    {
      "epoch": 0.029381965552178316,
      "grad_norm": 1.0333300828933716,
      "learning_rate": 1.996642085936527e-05,
      "loss": 1.0979,
      "step": 87
    },
    {
      "epoch": 0.02971968929415738,
      "grad_norm": 1.180769443511963,
      "learning_rate": 1.9965543514822063e-05,
      "loss": 1.0679,
      "step": 88
    },
    {
      "epoch": 0.03005741303613644,
      "grad_norm": 1.2959948778152466,
      "learning_rate": 1.9964654875901985e-05,
      "loss": 1.0858,
      "step": 89
    },
    {
      "epoch": 0.030395136778115502,
      "grad_norm": 1.0110646486282349,
      "learning_rate": 1.9963754943612167e-05,
      "loss": 1.0614,
      "step": 90
    },
    {
      "epoch": 0.030732860520094562,
      "grad_norm": 1.069658875465393,
      "learning_rate": 1.996284371897254e-05,
      "loss": 1.1384,
      "step": 91
    },
    {
      "epoch": 0.031070584262073625,
      "grad_norm": 1.125742793083191,
      "learning_rate": 1.9961921203015837e-05,
      "loss": 1.1249,
      "step": 92
    },
    {
      "epoch": 0.031408308004052685,
      "grad_norm": 0.9892888069152832,
      "learning_rate": 1.9960987396787582e-05,
      "loss": 1.1239,
      "step": 93
    },
    {
      "epoch": 0.031746031746031744,
      "grad_norm": 1.1462578773498535,
      "learning_rate": 1.99600423013461e-05,
      "loss": 1.1451,
      "step": 94
    },
    {
      "epoch": 0.032083755488010804,
      "grad_norm": 1.2448674440383911,
      "learning_rate": 1.9959085917762508e-05,
      "loss": 1.0432,
      "step": 95
    },
    {
      "epoch": 0.03242147922998987,
      "grad_norm": 1.1088314056396484,
      "learning_rate": 1.9958118247120714e-05,
      "loss": 1.1271,
      "step": 96
    },
    {
      "epoch": 0.03275920297196893,
      "grad_norm": 1.0089964866638184,
      "learning_rate": 1.9957139290517423e-05,
      "loss": 1.1616,
      "step": 97
    },
    {
      "epoch": 0.03309692671394799,
      "grad_norm": 1.2244312763214111,
      "learning_rate": 1.9956149049062125e-05,
      "loss": 1.0231,
      "step": 98
    },
    {
      "epoch": 0.03343465045592705,
      "grad_norm": 1.097711443901062,
      "learning_rate": 1.995514752387711e-05,
      "loss": 1.0811,
      "step": 99
    },
    {
      "epoch": 0.033772374197906116,
      "grad_norm": 1.0650869607925415,
      "learning_rate": 1.995413471609744e-05,
      "loss": 1.1214,
      "step": 100
    },
    {
      "epoch": 0.034110097939885176,
      "grad_norm": 1.2128289937973022,
      "learning_rate": 1.995311062687098e-05,
      "loss": 1.1209,
      "step": 101
    },
    {
      "epoch": 0.034447821681864235,
      "grad_norm": 1.330643653869629,
      "learning_rate": 1.995207525735837e-05,
      "loss": 0.955,
      "step": 102
    },
    {
      "epoch": 0.034785545423843295,
      "grad_norm": 1.4415818452835083,
      "learning_rate": 1.9951028608733044e-05,
      "loss": 1.0421,
      "step": 103
    },
    {
      "epoch": 0.035123269165822354,
      "grad_norm": 1.206916093826294,
      "learning_rate": 1.9949970682181208e-05,
      "loss": 1.1352,
      "step": 104
    },
    {
      "epoch": 0.03546099290780142,
      "grad_norm": 1.2657822370529175,
      "learning_rate": 1.9948901478901858e-05,
      "loss": 1.0408,
      "step": 105
    },
    {
      "epoch": 0.03579871664978048,
      "grad_norm": 1.0985996723175049,
      "learning_rate": 1.994782100010677e-05,
      "loss": 1.0235,
      "step": 106
    },
    {
      "epoch": 0.03613644039175954,
      "grad_norm": 1.2382546663284302,
      "learning_rate": 1.9946729247020492e-05,
      "loss": 1.0977,
      "step": 107
    },
    {
      "epoch": 0.0364741641337386,
      "grad_norm": 1.2296465635299683,
      "learning_rate": 1.994562622088036e-05,
      "loss": 1.1056,
      "step": 108
    },
    {
      "epoch": 0.03681188787571766,
      "grad_norm": 1.1169533729553223,
      "learning_rate": 1.9944511922936476e-05,
      "loss": 0.9911,
      "step": 109
    },
    {
      "epoch": 0.037149611617696726,
      "grad_norm": 1.2000555992126465,
      "learning_rate": 1.9943386354451722e-05,
      "loss": 1.1155,
      "step": 110
    },
    {
      "epoch": 0.037487335359675786,
      "grad_norm": 1.4214962720870972,
      "learning_rate": 1.9942249516701757e-05,
      "loss": 0.9501,
      "step": 111
    },
    {
      "epoch": 0.037825059101654845,
      "grad_norm": 1.138381004333496,
      "learning_rate": 1.9941101410975e-05,
      "loss": 1.1555,
      "step": 112
    },
    {
      "epoch": 0.038162782843633905,
      "grad_norm": 1.1567991971969604,
      "learning_rate": 1.9939942038572657e-05,
      "loss": 1.0941,
      "step": 113
    },
    {
      "epoch": 0.03850050658561297,
      "grad_norm": 1.2985361814498901,
      "learning_rate": 1.9938771400808693e-05,
      "loss": 1.0178,
      "step": 114
    },
    {
      "epoch": 0.03883823032759203,
      "grad_norm": 1.1133500337600708,
      "learning_rate": 1.9937589499009834e-05,
      "loss": 1.1013,
      "step": 115
    },
    {
      "epoch": 0.03917595406957109,
      "grad_norm": 1.2184875011444092,
      "learning_rate": 1.993639633451559e-05,
      "loss": 1.116,
      "step": 116
    },
    {
      "epoch": 0.03951367781155015,
      "grad_norm": 1.1061135530471802,
      "learning_rate": 1.9935191908678224e-05,
      "loss": 1.1254,
      "step": 117
    },
    {
      "epoch": 0.03985140155352921,
      "grad_norm": 1.2577052116394043,
      "learning_rate": 1.9933976222862754e-05,
      "loss": 1.0621,
      "step": 118
    },
    {
      "epoch": 0.04018912529550828,
      "grad_norm": 1.1311990022659302,
      "learning_rate": 1.9932749278446978e-05,
      "loss": 1.0789,
      "step": 119
    },
    {
      "epoch": 0.040526849037487336,
      "grad_norm": 1.2264856100082397,
      "learning_rate": 1.993151107682144e-05,
      "loss": 1.1294,
      "step": 120
    },
    {
      "epoch": 0.040864572779466396,
      "grad_norm": 1.1693328619003296,
      "learning_rate": 1.9930261619389446e-05,
      "loss": 1.0168,
      "step": 121
    },
    {
      "epoch": 0.041202296521445456,
      "grad_norm": 1.3191126585006714,
      "learning_rate": 1.9929000907567067e-05,
      "loss": 1.1116,
      "step": 122
    },
    {
      "epoch": 0.04154002026342452,
      "grad_norm": 1.3865387439727783,
      "learning_rate": 1.9927728942783112e-05,
      "loss": 0.9587,
      "step": 123
    },
    {
      "epoch": 0.04187774400540358,
      "grad_norm": 1.2523542642593384,
      "learning_rate": 1.9926445726479162e-05,
      "loss": 1.0787,
      "step": 124
    },
    {
      "epoch": 0.04221546774738264,
      "grad_norm": 1.1161696910858154,
      "learning_rate": 1.9925151260109533e-05,
      "loss": 1.0525,
      "step": 125
    },
    {
      "epoch": 0.0425531914893617,
      "grad_norm": 1.4163349866867065,
      "learning_rate": 1.9923845545141308e-05,
      "loss": 1.0917,
      "step": 126
    },
    {
      "epoch": 0.04289091523134076,
      "grad_norm": 1.244499921798706,
      "learning_rate": 1.99225285830543e-05,
      "loss": 1.0705,
      "step": 127
    },
    {
      "epoch": 0.04322863897331983,
      "grad_norm": 1.2816716432571411,
      "learning_rate": 1.9921200375341087e-05,
      "loss": 1.1192,
      "step": 128
    },
    {
      "epoch": 0.04356636271529889,
      "grad_norm": 1.2304255962371826,
      "learning_rate": 1.9919860923506977e-05,
      "loss": 1.0569,
      "step": 129
    },
    {
      "epoch": 0.04390408645727795,
      "grad_norm": 1.1381237506866455,
      "learning_rate": 1.991851022907003e-05,
      "loss": 0.9996,
      "step": 130
    },
    {
      "epoch": 0.044241810199257006,
      "grad_norm": 1.3635597229003906,
      "learning_rate": 1.9917148293561048e-05,
      "loss": 1.0294,
      "step": 131
    },
    {
      "epoch": 0.044579533941236066,
      "grad_norm": 1.4003256559371948,
      "learning_rate": 1.9915775118523567e-05,
      "loss": 1.0703,
      "step": 132
    },
    {
      "epoch": 0.04491725768321513,
      "grad_norm": 1.2415904998779297,
      "learning_rate": 1.991439070551387e-05,
      "loss": 1.1172,
      "step": 133
    },
    {
      "epoch": 0.04525498142519419,
      "grad_norm": 1.3725273609161377,
      "learning_rate": 1.9912995056100968e-05,
      "loss": 1.0632,
      "step": 134
    },
    {
      "epoch": 0.04559270516717325,
      "grad_norm": 1.1847169399261475,
      "learning_rate": 1.9911588171866612e-05,
      "loss": 1.0401,
      "step": 135
    },
    {
      "epoch": 0.04593042890915231,
      "grad_norm": 1.3664716482162476,
      "learning_rate": 1.991017005440528e-05,
      "loss": 0.989,
      "step": 136
    },
    {
      "epoch": 0.04626815265113138,
      "grad_norm": 1.1029189825057983,
      "learning_rate": 1.990874070532419e-05,
      "loss": 1.1216,
      "step": 137
    },
    {
      "epoch": 0.04660587639311044,
      "grad_norm": 1.1539902687072754,
      "learning_rate": 1.9907300126243286e-05,
      "loss": 1.1808,
      "step": 138
    },
    {
      "epoch": 0.0469436001350895,
      "grad_norm": 1.439884901046753,
      "learning_rate": 1.990584831879523e-05,
      "loss": 1.135,
      "step": 139
    },
    {
      "epoch": 0.04728132387706856,
      "grad_norm": 1.2660239934921265,
      "learning_rate": 1.9904385284625426e-05,
      "loss": 1.1224,
      "step": 140
    },
    {
      "epoch": 0.047619047619047616,
      "grad_norm": 1.1815766096115112,
      "learning_rate": 1.9902911025391987e-05,
      "loss": 1.0502,
      "step": 141
    },
    {
      "epoch": 0.04795677136102668,
      "grad_norm": 1.2217198610305786,
      "learning_rate": 1.9901425542765757e-05,
      "loss": 1.05,
      "step": 142
    },
    {
      "epoch": 0.04829449510300574,
      "grad_norm": 1.2554545402526855,
      "learning_rate": 1.9899928838430297e-05,
      "loss": 0.9519,
      "step": 143
    },
    {
      "epoch": 0.0486322188449848,
      "grad_norm": 1.3783280849456787,
      "learning_rate": 1.989842091408188e-05,
      "loss": 1.0126,
      "step": 144
    },
    {
      "epoch": 0.04896994258696386,
      "grad_norm": 1.287166714668274,
      "learning_rate": 1.9896901771429516e-05,
      "loss": 1.0803,
      "step": 145
    },
    {
      "epoch": 0.04930766632894292,
      "grad_norm": 1.4803354740142822,
      "learning_rate": 1.98953714121949e-05,
      "loss": 0.8707,
      "step": 146
    },
    {
      "epoch": 0.04964539007092199,
      "grad_norm": 1.6217652559280396,
      "learning_rate": 1.989382983811246e-05,
      "loss": 0.9855,
      "step": 147
    },
    {
      "epoch": 0.04998311381290105,
      "grad_norm": 1.4599343538284302,
      "learning_rate": 1.9892277050929326e-05,
      "loss": 1.0917,
      "step": 148
    },
    {
      "epoch": 0.05032083755488011,
      "grad_norm": 1.2973390817642212,
      "learning_rate": 1.989071305240534e-05,
      "loss": 1.1261,
      "step": 149
    },
    {
      "epoch": 0.05065856129685917,
      "grad_norm": 1.288674235343933,
      "learning_rate": 1.9889137844313047e-05,
      "loss": 1.0314,
      "step": 150
    },
    {
      "epoch": 0.050996285038838234,
      "grad_norm": 1.186750888824463,
      "learning_rate": 1.9887551428437696e-05,
      "loss": 1.0433,
      "step": 151
    },
    {
      "epoch": 0.05133400878081729,
      "grad_norm": 1.2350986003875732,
      "learning_rate": 1.988595380657724e-05,
      "loss": 1.0406,
      "step": 152
    },
    {
      "epoch": 0.05167173252279635,
      "grad_norm": 1.2333608865737915,
      "learning_rate": 1.9884344980542337e-05,
      "loss": 1.0127,
      "step": 153
    },
    {
      "epoch": 0.05200945626477541,
      "grad_norm": 1.5522956848144531,
      "learning_rate": 1.9882724952156337e-05,
      "loss": 1.0323,
      "step": 154
    },
    {
      "epoch": 0.05234718000675447,
      "grad_norm": 1.5638422966003418,
      "learning_rate": 1.9881093723255283e-05,
      "loss": 1.0168,
      "step": 155
    },
    {
      "epoch": 0.05268490374873354,
      "grad_norm": 1.3767153024673462,
      "learning_rate": 1.9879451295687927e-05,
      "loss": 1.0402,
      "step": 156
    },
    {
      "epoch": 0.0530226274907126,
      "grad_norm": 1.442148208618164,
      "learning_rate": 1.9877797671315688e-05,
      "loss": 0.937,
      "step": 157
    },
    {
      "epoch": 0.05336035123269166,
      "grad_norm": 1.166099190711975,
      "learning_rate": 1.9876132852012697e-05,
      "loss": 1.0421,
      "step": 158
    },
    {
      "epoch": 0.05369807497467072,
      "grad_norm": 1.299013614654541,
      "learning_rate": 1.9874456839665767e-05,
      "loss": 0.9955,
      "step": 159
    },
    {
      "epoch": 0.05403579871664978,
      "grad_norm": 1.1731435060501099,
      "learning_rate": 1.9872769636174384e-05,
      "loss": 1.0713,
      "step": 160
    },
    {
      "epoch": 0.054373522458628844,
      "grad_norm": 1.2361280918121338,
      "learning_rate": 1.9871071243450745e-05,
      "loss": 0.9948,
      "step": 161
    },
    {
      "epoch": 0.0547112462006079,
      "grad_norm": 1.2518837451934814,
      "learning_rate": 1.9869361663419697e-05,
      "loss": 1.0896,
      "step": 162
    },
    {
      "epoch": 0.05504896994258696,
      "grad_norm": 1.2696311473846436,
      "learning_rate": 1.9867640898018786e-05,
      "loss": 1.0898,
      "step": 163
    },
    {
      "epoch": 0.05538669368456602,
      "grad_norm": 1.466525912284851,
      "learning_rate": 1.9865908949198223e-05,
      "loss": 1.0251,
      "step": 164
    },
    {
      "epoch": 0.05572441742654509,
      "grad_norm": 1.4841068983078003,
      "learning_rate": 1.9864165818920913e-05,
      "loss": 1.0882,
      "step": 165
    },
    {
      "epoch": 0.05606214116852415,
      "grad_norm": 1.2612347602844238,
      "learning_rate": 1.9862411509162406e-05,
      "loss": 1.0708,
      "step": 166
    },
    {
      "epoch": 0.05639986491050321,
      "grad_norm": 1.5272561311721802,
      "learning_rate": 1.9860646021910947e-05,
      "loss": 0.9337,
      "step": 167
    },
    {
      "epoch": 0.05673758865248227,
      "grad_norm": 1.3568799495697021,
      "learning_rate": 1.985886935916743e-05,
      "loss": 1.1045,
      "step": 168
    },
    {
      "epoch": 0.05707531239446133,
      "grad_norm": 1.3452101945877075,
      "learning_rate": 1.9857081522945427e-05,
      "loss": 1.0561,
      "step": 169
    },
    {
      "epoch": 0.057413036136440394,
      "grad_norm": 1.410446047782898,
      "learning_rate": 1.985528251527117e-05,
      "loss": 1.1261,
      "step": 170
    },
    {
      "epoch": 0.057750759878419454,
      "grad_norm": 1.2828803062438965,
      "learning_rate": 1.9853472338183544e-05,
      "loss": 1.1957,
      "step": 171
    },
    {
      "epoch": 0.058088483620398514,
      "grad_norm": 1.6754173040390015,
      "learning_rate": 1.9851650993734112e-05,
      "loss": 0.9598,
      "step": 172
    },
    {
      "epoch": 0.05842620736237757,
      "grad_norm": 1.5594916343688965,
      "learning_rate": 1.9849818483987074e-05,
      "loss": 1.0111,
      "step": 173
    },
    {
      "epoch": 0.05876393110435663,
      "grad_norm": 1.280019998550415,
      "learning_rate": 1.9847974811019294e-05,
      "loss": 1.0047,
      "step": 174
    },
    {
      "epoch": 0.0591016548463357,
      "grad_norm": 1.2656089067459106,
      "learning_rate": 1.9846119976920284e-05,
      "loss": 1.0408,
      "step": 175
    },
    {
      "epoch": 0.05943937858831476,
      "grad_norm": 1.293466567993164,
      "learning_rate": 1.984425398379221e-05,
      "loss": 1.0983,
      "step": 176
    },
    {
      "epoch": 0.05977710233029382,
      "grad_norm": 1.196362853050232,
      "learning_rate": 1.9842376833749878e-05,
      "loss": 0.9904,
      "step": 177
    },
    {
      "epoch": 0.06011482607227288,
      "grad_norm": 1.346326231956482,
      "learning_rate": 1.9840488528920743e-05,
      "loss": 0.9417,
      "step": 178
    },
    {
      "epoch": 0.060452549814251945,
      "grad_norm": 1.2747066020965576,
      "learning_rate": 1.9838589071444905e-05,
      "loss": 1.068,
      "step": 179
    },
    {
      "epoch": 0.060790273556231005,
      "grad_norm": 1.286851167678833,
      "learning_rate": 1.9836678463475096e-05,
      "loss": 0.9681,
      "step": 180
    },
    {
      "epoch": 0.061127997298210064,
      "grad_norm": 1.250110387802124,
      "learning_rate": 1.983475670717669e-05,
      "loss": 1.0312,
      "step": 181
    },
    {
      "epoch": 0.061465721040189124,
      "grad_norm": 1.5548391342163086,
      "learning_rate": 1.98328238047277e-05,
      "loss": 1.022,
      "step": 182
    },
    {
      "epoch": 0.06180344478216818,
      "grad_norm": 1.4442728757858276,
      "learning_rate": 1.9830879758318765e-05,
      "loss": 1.1751,
      "step": 183
    },
    {
      "epoch": 0.06214116852414725,
      "grad_norm": 1.144952654838562,
      "learning_rate": 1.9828924570153155e-05,
      "loss": 1.1576,
      "step": 184
    },
    {
      "epoch": 0.06247889226612631,
      "grad_norm": 1.2939969301223755,
      "learning_rate": 1.982695824244677e-05,
      "loss": 1.0007,
      "step": 185
    },
    {
      "epoch": 0.06281661600810537,
      "grad_norm": 1.291254997253418,
      "learning_rate": 1.9824980777428133e-05,
      "loss": 1.0811,
      "step": 186
    },
    {
      "epoch": 0.06315433975008443,
      "grad_norm": 1.5193324089050293,
      "learning_rate": 1.9822992177338383e-05,
      "loss": 0.9987,
      "step": 187
    },
    {
      "epoch": 0.06349206349206349,
      "grad_norm": 1.6055748462677002,
      "learning_rate": 1.9820992444431296e-05,
      "loss": 0.978,
      "step": 188
    },
    {
      "epoch": 0.06382978723404255,
      "grad_norm": 1.3080774545669556,
      "learning_rate": 1.9818981580973254e-05,
      "loss": 1.0504,
      "step": 189
    },
    {
      "epoch": 0.06416751097602161,
      "grad_norm": 1.6906174421310425,
      "learning_rate": 1.981695958924325e-05,
      "loss": 0.9122,
      "step": 190
    },
    {
      "epoch": 0.06450523471800068,
      "grad_norm": 1.316392421722412,
      "learning_rate": 1.9814926471532888e-05,
      "loss": 1.0985,
      "step": 191
    },
    {
      "epoch": 0.06484295845997974,
      "grad_norm": 1.4848136901855469,
      "learning_rate": 1.98128822301464e-05,
      "loss": 1.0883,
      "step": 192
    },
    {
      "epoch": 0.0651806822019588,
      "grad_norm": 1.4925001859664917,
      "learning_rate": 1.9810826867400605e-05,
      "loss": 1.0623,
      "step": 193
    },
    {
      "epoch": 0.06551840594393786,
      "grad_norm": 1.3369628190994263,
      "learning_rate": 1.9808760385624932e-05,
      "loss": 1.1153,
      "step": 194
    },
    {
      "epoch": 0.06585612968591692,
      "grad_norm": 1.4233779907226562,
      "learning_rate": 1.9806682787161418e-05,
      "loss": 1.0171,
      "step": 195
    },
    {
      "epoch": 0.06619385342789598,
      "grad_norm": 1.3952828645706177,
      "learning_rate": 1.9804594074364692e-05,
      "loss": 1.0192,
      "step": 196
    },
    {
      "epoch": 0.06653157716987504,
      "grad_norm": 1.5116938352584839,
      "learning_rate": 1.9802494249601984e-05,
      "loss": 0.9419,
      "step": 197
    },
    {
      "epoch": 0.0668693009118541,
      "grad_norm": 1.390758991241455,
      "learning_rate": 1.9800383315253115e-05,
      "loss": 1.1492,
      "step": 198
    },
    {
      "epoch": 0.06720702465383316,
      "grad_norm": 1.4588640928268433,
      "learning_rate": 1.979826127371049e-05,
      "loss": 0.9627,
      "step": 199
    },
    {
      "epoch": 0.06754474839581223,
      "grad_norm": 1.2791355848312378,
      "learning_rate": 1.9796128127379117e-05,
      "loss": 1.0945,
      "step": 200
    },
    {
      "epoch": 0.06788247213779129,
      "grad_norm": 1.2112960815429688,
      "learning_rate": 1.9793983878676578e-05,
      "loss": 1.0788,
      "step": 201
    },
    {
      "epoch": 0.06822019587977035,
      "grad_norm": 1.3849095106124878,
      "learning_rate": 1.9791828530033048e-05,
      "loss": 1.1808,
      "step": 202
    },
    {
      "epoch": 0.06855791962174941,
      "grad_norm": 1.289900302886963,
      "learning_rate": 1.978966208389127e-05,
      "loss": 1.1193,
      "step": 203
    },
    {
      "epoch": 0.06889564336372847,
      "grad_norm": 1.4258767366409302,
      "learning_rate": 1.978748454270657e-05,
      "loss": 1.007,
      "step": 204
    },
    {
      "epoch": 0.06923336710570753,
      "grad_norm": 1.3458566665649414,
      "learning_rate": 1.978529590894685e-05,
      "loss": 1.0839,
      "step": 205
    },
    {
      "epoch": 0.06957109084768659,
      "grad_norm": 1.3136255741119385,
      "learning_rate": 1.9783096185092582e-05,
      "loss": 1.054,
      "step": 206
    },
    {
      "epoch": 0.06990881458966565,
      "grad_norm": 1.3639452457427979,
      "learning_rate": 1.978088537363681e-05,
      "loss": 0.961,
      "step": 207
    },
    {
      "epoch": 0.07024653833164471,
      "grad_norm": 1.3582463264465332,
      "learning_rate": 1.9778663477085137e-05,
      "loss": 1.0477,
      "step": 208
    },
    {
      "epoch": 0.07058426207362378,
      "grad_norm": 1.46528959274292,
      "learning_rate": 1.9776430497955736e-05,
      "loss": 1.031,
      "step": 209
    },
    {
      "epoch": 0.07092198581560284,
      "grad_norm": 1.0939236879348755,
      "learning_rate": 1.9774186438779336e-05,
      "loss": 0.6518,
      "step": 210
    },
    {
      "epoch": 0.0712597095575819,
      "grad_norm": 1.621648907661438,
      "learning_rate": 1.9771931302099228e-05,
      "loss": 0.9673,
      "step": 211
    },
    {
      "epoch": 0.07159743329956096,
      "grad_norm": 1.4661349058151245,
      "learning_rate": 1.9769665090471253e-05,
      "loss": 1.034,
      "step": 212
    },
    {
      "epoch": 0.07193515704154002,
      "grad_norm": 1.2988148927688599,
      "learning_rate": 1.9767387806463807e-05,
      "loss": 0.6753,
      "step": 213
    },
    {
      "epoch": 0.07227288078351908,
      "grad_norm": 1.3928842544555664,
      "learning_rate": 1.976509945265783e-05,
      "loss": 1.1259,
      "step": 214
    },
    {
      "epoch": 0.07261060452549814,
      "grad_norm": 1.6413177251815796,
      "learning_rate": 1.9762800031646814e-05,
      "loss": 0.9387,
      "step": 215
    },
    {
      "epoch": 0.0729483282674772,
      "grad_norm": 1.6180251836776733,
      "learning_rate": 1.9760489546036793e-05,
      "loss": 1.0458,
      "step": 216
    },
    {
      "epoch": 0.07328605200945626,
      "grad_norm": 1.4046472311019897,
      "learning_rate": 1.9758167998446334e-05,
      "loss": 0.9696,
      "step": 217
    },
    {
      "epoch": 0.07362377575143532,
      "grad_norm": 1.4379267692565918,
      "learning_rate": 1.975583539150655e-05,
      "loss": 1.0832,
      "step": 218
    },
    {
      "epoch": 0.07396149949341439,
      "grad_norm": 1.3287361860275269,
      "learning_rate": 1.9753491727861085e-05,
      "loss": 1.0125,
      "step": 219
    },
    {
      "epoch": 0.07429922323539345,
      "grad_norm": 1.3055379390716553,
      "learning_rate": 1.975113701016611e-05,
      "loss": 1.1367,
      "step": 220
    },
    {
      "epoch": 0.07463694697737251,
      "grad_norm": 1.4172956943511963,
      "learning_rate": 1.974877124109033e-05,
      "loss": 1.0,
      "step": 221
    },
    {
      "epoch": 0.07497467071935157,
      "grad_norm": 1.3897404670715332,
      "learning_rate": 1.9746394423314973e-05,
      "loss": 0.9993,
      "step": 222
    },
    {
      "epoch": 0.07531239446133063,
      "grad_norm": 1.2462091445922852,
      "learning_rate": 1.9744006559533782e-05,
      "loss": 1.0552,
      "step": 223
    },
    {
      "epoch": 0.07565011820330969,
      "grad_norm": 1.4593998193740845,
      "learning_rate": 1.9741607652453033e-05,
      "loss": 0.8495,
      "step": 224
    },
    {
      "epoch": 0.07598784194528875,
      "grad_norm": 1.3893877267837524,
      "learning_rate": 1.9739197704791508e-05,
      "loss": 0.9648,
      "step": 225
    },
    {
      "epoch": 0.07632556568726781,
      "grad_norm": 1.4935283660888672,
      "learning_rate": 1.9736776719280502e-05,
      "loss": 0.9438,
      "step": 226
    },
    {
      "epoch": 0.07666328942924687,
      "grad_norm": 1.4032809734344482,
      "learning_rate": 1.9734344698663823e-05,
      "loss": 0.9616,
      "step": 227
    },
    {
      "epoch": 0.07700101317122594,
      "grad_norm": 1.3284560441970825,
      "learning_rate": 1.973190164569778e-05,
      "loss": 1.0369,
      "step": 228
    },
    {
      "epoch": 0.077338736913205,
      "grad_norm": 1.634592056274414,
      "learning_rate": 1.9729447563151193e-05,
      "loss": 0.8822,
      "step": 229
    },
    {
      "epoch": 0.07767646065518406,
      "grad_norm": 1.3971806764602661,
      "learning_rate": 1.9726982453805383e-05,
      "loss": 1.0102,
      "step": 230
    },
    {
      "epoch": 0.07801418439716312,
      "grad_norm": 1.3324154615402222,
      "learning_rate": 1.9724506320454153e-05,
      "loss": 1.0613,
      "step": 231
    },
    {
      "epoch": 0.07835190813914218,
      "grad_norm": 1.3715732097625732,
      "learning_rate": 1.972201916590382e-05,
      "loss": 0.9861,
      "step": 232
    },
    {
      "epoch": 0.07868963188112124,
      "grad_norm": 1.3813071250915527,
      "learning_rate": 1.971952099297318e-05,
      "loss": 1.0753,
      "step": 233
    },
    {
      "epoch": 0.0790273556231003,
      "grad_norm": 1.7483845949172974,
      "learning_rate": 1.971701180449352e-05,
      "loss": 0.9567,
      "step": 234
    },
    {
      "epoch": 0.07936507936507936,
      "grad_norm": 1.3579798936843872,
      "learning_rate": 1.971449160330861e-05,
      "loss": 1.1004,
      "step": 235
    },
    {
      "epoch": 0.07970280310705842,
      "grad_norm": 1.3669630289077759,
      "learning_rate": 1.9711960392274698e-05,
      "loss": 1.0097,
      "step": 236
    },
    {
      "epoch": 0.0800405268490375,
      "grad_norm": 1.3327555656433105,
      "learning_rate": 1.9709418174260523e-05,
      "loss": 1.0684,
      "step": 237
    },
    {
      "epoch": 0.08037825059101655,
      "grad_norm": 1.3267375230789185,
      "learning_rate": 1.970686495214728e-05,
      "loss": 0.9903,
      "step": 238
    },
    {
      "epoch": 0.08071597433299561,
      "grad_norm": 1.643053650856018,
      "learning_rate": 1.970430072882865e-05,
      "loss": 1.0131,
      "step": 239
    },
    {
      "epoch": 0.08105369807497467,
      "grad_norm": 1.4109166860580444,
      "learning_rate": 1.9701725507210776e-05,
      "loss": 1.0438,
      "step": 240
    },
    {
      "epoch": 0.08139142181695373,
      "grad_norm": 1.5236960649490356,
      "learning_rate": 1.9699139290212267e-05,
      "loss": 0.9117,
      "step": 241
    },
    {
      "epoch": 0.08172914555893279,
      "grad_norm": 1.5459891557693481,
      "learning_rate": 1.969654208076419e-05,
      "loss": 0.9182,
      "step": 242
    },
    {
      "epoch": 0.08206686930091185,
      "grad_norm": 1.3404607772827148,
      "learning_rate": 1.969393388181008e-05,
      "loss": 1.0593,
      "step": 243
    },
    {
      "epoch": 0.08240459304289091,
      "grad_norm": 1.4050788879394531,
      "learning_rate": 1.9691314696305915e-05,
      "loss": 1.047,
      "step": 244
    },
    {
      "epoch": 0.08274231678486997,
      "grad_norm": 1.356428623199463,
      "learning_rate": 1.9688684527220133e-05,
      "loss": 1.1039,
      "step": 245
    },
    {
      "epoch": 0.08308004052684904,
      "grad_norm": 1.4300233125686646,
      "learning_rate": 1.9686043377533615e-05,
      "loss": 1.1485,
      "step": 246
    },
    {
      "epoch": 0.0834177642688281,
      "grad_norm": 1.4281103610992432,
      "learning_rate": 1.9683391250239693e-05,
      "loss": 0.9888,
      "step": 247
    },
    {
      "epoch": 0.08375548801080716,
      "grad_norm": 1.4770859479904175,
      "learning_rate": 1.9680728148344132e-05,
      "loss": 1.0017,
      "step": 248
    },
    {
      "epoch": 0.08409321175278622,
      "grad_norm": 1.289460301399231,
      "learning_rate": 1.9678054074865137e-05,
      "loss": 1.0654,
      "step": 249
    },
    {
      "epoch": 0.08443093549476528,
      "grad_norm": 1.124488353729248,
      "learning_rate": 1.967536903283336e-05,
      "loss": 0.6338,
      "step": 250
    },
    {
      "epoch": 0.08476865923674434,
      "grad_norm": 1.22398841381073,
      "learning_rate": 1.9672673025291863e-05,
      "loss": 0.6657,
      "step": 251
    },
    {
      "epoch": 0.0851063829787234,
      "grad_norm": 1.4435992240905762,
      "learning_rate": 1.9669966055296154e-05,
      "loss": 1.0658,
      "step": 252
    },
    {
      "epoch": 0.08544410672070246,
      "grad_norm": 1.499653697013855,
      "learning_rate": 1.9667248125914153e-05,
      "loss": 1.1536,
      "step": 253
    },
    {
      "epoch": 0.08578183046268152,
      "grad_norm": 1.4434421062469482,
      "learning_rate": 1.966451924022621e-05,
      "loss": 1.0493,
      "step": 254
    },
    {
      "epoch": 0.08611955420466058,
      "grad_norm": 1.4568160772323608,
      "learning_rate": 1.9661779401325086e-05,
      "loss": 1.0196,
      "step": 255
    },
    {
      "epoch": 0.08645727794663965,
      "grad_norm": 1.4852335453033447,
      "learning_rate": 1.965902861231596e-05,
      "loss": 1.1893,
      "step": 256
    },
    {
      "epoch": 0.08679500168861871,
      "grad_norm": 1.370621681213379,
      "learning_rate": 1.965626687631641e-05,
      "loss": 0.9938,
      "step": 257
    },
    {
      "epoch": 0.08713272543059777,
      "grad_norm": 1.2765510082244873,
      "learning_rate": 1.9653494196456442e-05,
      "loss": 1.062,
      "step": 258
    },
    {
      "epoch": 0.08747044917257683,
      "grad_norm": 1.5824520587921143,
      "learning_rate": 1.9650710575878445e-05,
      "loss": 1.1016,
      "step": 259
    },
    {
      "epoch": 0.0878081729145559,
      "grad_norm": 1.4297126531600952,
      "learning_rate": 1.9647916017737218e-05,
      "loss": 1.0511,
      "step": 260
    },
    {
      "epoch": 0.08814589665653495,
      "grad_norm": 1.6792505979537964,
      "learning_rate": 1.964511052519995e-05,
      "loss": 0.9311,
      "step": 261
    },
    {
      "epoch": 0.08848362039851401,
      "grad_norm": 1.302420735359192,
      "learning_rate": 1.964229410144623e-05,
      "loss": 1.0242,
      "step": 262
    },
    {
      "epoch": 0.08882134414049307,
      "grad_norm": 1.3385374546051025,
      "learning_rate": 1.963946674966803e-05,
      "loss": 1.1204,
      "step": 263
    },
    {
      "epoch": 0.08915906788247213,
      "grad_norm": 1.4421699047088623,
      "learning_rate": 1.9636628473069704e-05,
      "loss": 0.969,
      "step": 264
    },
    {
      "epoch": 0.0894967916244512,
      "grad_norm": 1.465737223625183,
      "learning_rate": 1.9633779274868003e-05,
      "loss": 1.0395,
      "step": 265
    },
    {
      "epoch": 0.08983451536643026,
      "grad_norm": 1.0596539974212646,
      "learning_rate": 1.9630919158292037e-05,
      "loss": 0.6803,
      "step": 266
    },
    {
      "epoch": 0.09017223910840932,
      "grad_norm": 1.8407821655273438,
      "learning_rate": 1.9628048126583298e-05,
      "loss": 0.7807,
      "step": 267
    },
    {
      "epoch": 0.09050996285038838,
      "grad_norm": 1.4079521894454956,
      "learning_rate": 1.9625166182995654e-05,
      "loss": 0.9937,
      "step": 268
    },
    {
      "epoch": 0.09084768659236744,
      "grad_norm": 1.6088870763778687,
      "learning_rate": 1.962227333079533e-05,
      "loss": 1.0248,
      "step": 269
    },
    {
      "epoch": 0.0911854103343465,
      "grad_norm": 1.413586974143982,
      "learning_rate": 1.9619369573260924e-05,
      "loss": 0.9148,
      "step": 270
    },
    {
      "epoch": 0.09152313407632556,
      "grad_norm": 1.3591145277023315,
      "learning_rate": 1.9616454913683383e-05,
      "loss": 1.078,
      "step": 271
    },
    {
      "epoch": 0.09186085781830462,
      "grad_norm": 1.5194146633148193,
      "learning_rate": 1.9613529355366026e-05,
      "loss": 0.9426,
      "step": 272
    },
    {
      "epoch": 0.09219858156028368,
      "grad_norm": 1.3751038312911987,
      "learning_rate": 1.96105929016245e-05,
      "loss": 1.1102,
      "step": 273
    },
    {
      "epoch": 0.09253630530226276,
      "grad_norm": 1.4757986068725586,
      "learning_rate": 1.9607645555786823e-05,
      "loss": 0.9543,
      "step": 274
    },
    {
      "epoch": 0.09287402904424182,
      "grad_norm": 1.5707237720489502,
      "learning_rate": 1.9604687321193342e-05,
      "loss": 1.0697,
      "step": 275
    },
    {
      "epoch": 0.09321175278622088,
      "grad_norm": 1.7457220554351807,
      "learning_rate": 1.9601718201196757e-05,
      "loss": 1.0523,
      "step": 276
    },
    {
      "epoch": 0.09354947652819993,
      "grad_norm": 1.4106884002685547,
      "learning_rate": 1.9598738199162094e-05,
      "loss": 1.0469,
      "step": 277
    },
    {
      "epoch": 0.093887200270179,
      "grad_norm": 1.386497139930725,
      "learning_rate": 1.959574731846672e-05,
      "loss": 1.087,
      "step": 278
    },
    {
      "epoch": 0.09422492401215805,
      "grad_norm": 1.3651485443115234,
      "learning_rate": 1.9592745562500326e-05,
      "loss": 1.081,
      "step": 279
    },
    {
      "epoch": 0.09456264775413711,
      "grad_norm": 1.6645921468734741,
      "learning_rate": 1.958973293466493e-05,
      "loss": 1.0591,
      "step": 280
    },
    {
      "epoch": 0.09490037149611617,
      "grad_norm": 1.5544499158859253,
      "learning_rate": 1.9586709438374873e-05,
      "loss": 1.0791,
      "step": 281
    },
    {
      "epoch": 0.09523809523809523,
      "grad_norm": 1.4313130378723145,
      "learning_rate": 1.958367507705681e-05,
      "loss": 0.9816,
      "step": 282
    },
    {
      "epoch": 0.09557581898007429,
      "grad_norm": 1.490135669708252,
      "learning_rate": 1.958062985414972e-05,
      "loss": 1.028,
      "step": 283
    },
    {
      "epoch": 0.09591354272205337,
      "grad_norm": 1.6701232194900513,
      "learning_rate": 1.9577573773104877e-05,
      "loss": 0.9252,
      "step": 284
    },
    {
      "epoch": 0.09625126646403243,
      "grad_norm": 1.545169711112976,
      "learning_rate": 1.9574506837385872e-05,
      "loss": 1.0003,
      "step": 285
    },
    {
      "epoch": 0.09658899020601149,
      "grad_norm": 1.523728847503662,
      "learning_rate": 1.9571429050468594e-05,
      "loss": 1.1214,
      "step": 286
    },
    {
      "epoch": 0.09692671394799054,
      "grad_norm": 1.4326605796813965,
      "learning_rate": 1.956834041584123e-05,
      "loss": 1.0819,
      "step": 287
    },
    {
      "epoch": 0.0972644376899696,
      "grad_norm": 1.4165427684783936,
      "learning_rate": 1.956524093700426e-05,
      "loss": 1.0076,
      "step": 288
    },
    {
      "epoch": 0.09760216143194866,
      "grad_norm": 1.2945036888122559,
      "learning_rate": 1.9562130617470463e-05,
      "loss": 1.1276,
      "step": 289
    },
    {
      "epoch": 0.09793988517392772,
      "grad_norm": 1.4874722957611084,
      "learning_rate": 1.9559009460764885e-05,
      "loss": 0.8736,
      "step": 290
    },
    {
      "epoch": 0.09827760891590678,
      "grad_norm": 1.342305064201355,
      "learning_rate": 1.955587747042488e-05,
      "loss": 1.06,
      "step": 291
    },
    {
      "epoch": 0.09861533265788584,
      "grad_norm": 1.470832109451294,
      "learning_rate": 1.955273465000006e-05,
      "loss": 1.0064,
      "step": 292
    },
    {
      "epoch": 0.09895305639986492,
      "grad_norm": 1.3302005529403687,
      "learning_rate": 1.9549581003052315e-05,
      "loss": 1.0491,
      "step": 293
    },
    {
      "epoch": 0.09929078014184398,
      "grad_norm": 1.5110094547271729,
      "learning_rate": 1.954641653315581e-05,
      "loss": 0.9659,
      "step": 294
    },
    {
      "epoch": 0.09962850388382304,
      "grad_norm": 1.3879752159118652,
      "learning_rate": 1.9543241243896976e-05,
      "loss": 0.9564,
      "step": 295
    },
    {
      "epoch": 0.0999662276258021,
      "grad_norm": 1.281111717224121,
      "learning_rate": 1.9540055138874504e-05,
      "loss": 0.9773,
      "step": 296
    },
    {
      "epoch": 0.10030395136778116,
      "grad_norm": 1.5493919849395752,
      "learning_rate": 1.9536858221699342e-05,
      "loss": 0.9208,
      "step": 297
    },
    {
      "epoch": 0.10064167510976021,
      "grad_norm": 1.5539532899856567,
      "learning_rate": 1.9533650495994697e-05,
      "loss": 0.928,
      "step": 298
    },
    {
      "epoch": 0.10097939885173927,
      "grad_norm": 1.4157252311706543,
      "learning_rate": 1.9530431965396013e-05,
      "loss": 0.9867,
      "step": 299
    },
    {
      "epoch": 0.10131712259371833,
      "grad_norm": 1.5017726421356201,
      "learning_rate": 1.9527202633551e-05,
      "loss": 1.0077,
      "step": 300
    },
    {
      "epoch": 0.1016548463356974,
      "grad_norm": 1.423443078994751,
      "learning_rate": 1.9523962504119588e-05,
      "loss": 1.0641,
      "step": 301
    },
    {
      "epoch": 0.10199257007767647,
      "grad_norm": 1.534997820854187,
      "learning_rate": 1.952071158077396e-05,
      "loss": 0.9852,
      "step": 302
    },
    {
      "epoch": 0.10233029381965553,
      "grad_norm": 1.3870246410369873,
      "learning_rate": 1.9517449867198523e-05,
      "loss": 1.0142,
      "step": 303
    },
    {
      "epoch": 0.10266801756163459,
      "grad_norm": 1.1131715774536133,
      "learning_rate": 1.9514177367089918e-05,
      "loss": 0.6895,
      "step": 304
    },
    {
      "epoch": 0.10300574130361365,
      "grad_norm": 1.5246529579162598,
      "learning_rate": 1.951089408415701e-05,
      "loss": 1.0109,
      "step": 305
    },
    {
      "epoch": 0.1033434650455927,
      "grad_norm": 1.449523687362671,
      "learning_rate": 1.950760002212088e-05,
      "loss": 0.9801,
      "step": 306
    },
    {
      "epoch": 0.10368118878757177,
      "grad_norm": 1.6088491678237915,
      "learning_rate": 1.9504295184714838e-05,
      "loss": 0.9538,
      "step": 307
    },
    {
      "epoch": 0.10401891252955082,
      "grad_norm": 1.5267025232315063,
      "learning_rate": 1.9500979575684392e-05,
      "loss": 0.9342,
      "step": 308
    },
    {
      "epoch": 0.10435663627152988,
      "grad_norm": 1.8209447860717773,
      "learning_rate": 1.9497653198787265e-05,
      "loss": 1.0505,
      "step": 309
    },
    {
      "epoch": 0.10469436001350894,
      "grad_norm": 1.3485527038574219,
      "learning_rate": 1.949431605779338e-05,
      "loss": 1.0387,
      "step": 310
    },
    {
      "epoch": 0.10503208375548802,
      "grad_norm": 1.4648553133010864,
      "learning_rate": 1.9490968156484867e-05,
      "loss": 0.9623,
      "step": 311
    },
    {
      "epoch": 0.10536980749746708,
      "grad_norm": 1.5027012825012207,
      "learning_rate": 1.948760949865604e-05,
      "loss": 0.96,
      "step": 312
    },
    {
      "epoch": 0.10570753123944614,
      "grad_norm": 1.5433413982391357,
      "learning_rate": 1.9484240088113407e-05,
      "loss": 1.0235,
      "step": 313
    },
    {
      "epoch": 0.1060452549814252,
      "grad_norm": 1.3185620307922363,
      "learning_rate": 1.9480859928675674e-05,
      "loss": 1.0635,
      "step": 314
    },
    {
      "epoch": 0.10638297872340426,
      "grad_norm": 1.7005455493927002,
      "learning_rate": 1.9477469024173717e-05,
      "loss": 0.9529,
      "step": 315
    },
    {
      "epoch": 0.10672070246538332,
      "grad_norm": 1.787432312965393,
      "learning_rate": 1.947406737845059e-05,
      "loss": 0.9144,
      "step": 316
    },
    {
      "epoch": 0.10705842620736238,
      "grad_norm": 1.4893176555633545,
      "learning_rate": 1.9470654995361526e-05,
      "loss": 1.0576,
      "step": 317
    },
    {
      "epoch": 0.10739614994934144,
      "grad_norm": 1.3912529945373535,
      "learning_rate": 1.9467231878773925e-05,
      "loss": 0.9881,
      "step": 318
    },
    {
      "epoch": 0.1077338736913205,
      "grad_norm": 1.3835701942443848,
      "learning_rate": 1.9463798032567352e-05,
      "loss": 0.9456,
      "step": 319
    },
    {
      "epoch": 0.10807159743329955,
      "grad_norm": 1.8590047359466553,
      "learning_rate": 1.946035346063353e-05,
      "loss": 1.0943,
      "step": 320
    },
    {
      "epoch": 0.10840932117527863,
      "grad_norm": 1.3064184188842773,
      "learning_rate": 1.9456898166876344e-05,
      "loss": 1.0626,
      "step": 321
    },
    {
      "epoch": 0.10874704491725769,
      "grad_norm": 1.4224311113357544,
      "learning_rate": 1.945343215521182e-05,
      "loss": 0.9684,
      "step": 322
    },
    {
      "epoch": 0.10908476865923675,
      "grad_norm": 1.3517730236053467,
      "learning_rate": 1.944995542956814e-05,
      "loss": 1.0418,
      "step": 323
    },
    {
      "epoch": 0.1094224924012158,
      "grad_norm": 1.4266732931137085,
      "learning_rate": 1.944646799388563e-05,
      "loss": 1.0416,
      "step": 324
    },
    {
      "epoch": 0.10976021614319487,
      "grad_norm": 1.420160174369812,
      "learning_rate": 1.9442969852116743e-05,
      "loss": 1.0252,
      "step": 325
    },
    {
      "epoch": 0.11009793988517393,
      "grad_norm": 1.7210063934326172,
      "learning_rate": 1.9439461008226078e-05,
      "loss": 0.8756,
      "step": 326
    },
    {
      "epoch": 0.11043566362715299,
      "grad_norm": 1.7844417095184326,
      "learning_rate": 1.9435941466190353e-05,
      "loss": 1.0676,
      "step": 327
    },
    {
      "epoch": 0.11077338736913205,
      "grad_norm": 1.3618022203445435,
      "learning_rate": 1.943241122999842e-05,
      "loss": 1.0737,
      "step": 328
    },
    {
      "epoch": 0.1111111111111111,
      "grad_norm": 1.0146949291229248,
      "learning_rate": 1.9428870303651246e-05,
      "loss": 0.6509,
      "step": 329
    },
    {
      "epoch": 0.11144883485309018,
      "grad_norm": 1.4110677242279053,
      "learning_rate": 1.9425318691161915e-05,
      "loss": 1.1214,
      "step": 330
    },
    {
      "epoch": 0.11178655859506924,
      "grad_norm": 1.3130214214324951,
      "learning_rate": 1.942175639655562e-05,
      "loss": 0.9811,
      "step": 331
    },
    {
      "epoch": 0.1121242823370483,
      "grad_norm": 1.5635089874267578,
      "learning_rate": 1.941818342386966e-05,
      "loss": 1.0225,
      "step": 332
    },
    {
      "epoch": 0.11246200607902736,
      "grad_norm": 1.4778531789779663,
      "learning_rate": 1.9414599777153443e-05,
      "loss": 0.9036,
      "step": 333
    },
    {
      "epoch": 0.11279972982100642,
      "grad_norm": 1.5236148834228516,
      "learning_rate": 1.941100546046846e-05,
      "loss": 1.0043,
      "step": 334
    },
    {
      "epoch": 0.11313745356298548,
      "grad_norm": 1.349365234375,
      "learning_rate": 1.9407400477888315e-05,
      "loss": 1.0342,
      "step": 335
    },
    {
      "epoch": 0.11347517730496454,
      "grad_norm": 1.5294772386550903,
      "learning_rate": 1.9403784833498684e-05,
      "loss": 1.0328,
      "step": 336
    },
    {
      "epoch": 0.1138129010469436,
      "grad_norm": 1.4891618490219116,
      "learning_rate": 1.940015853139733e-05,
      "loss": 1.0309,
      "step": 337
    },
    {
      "epoch": 0.11415062478892266,
      "grad_norm": 1.730509638786316,
      "learning_rate": 1.93965215756941e-05,
      "loss": 0.9821,
      "step": 338
    },
    {
      "epoch": 0.11448834853090173,
      "grad_norm": 1.5002323389053345,
      "learning_rate": 1.9392873970510905e-05,
      "loss": 0.9416,
      "step": 339
    },
    {
      "epoch": 0.11482607227288079,
      "grad_norm": 1.7629708051681519,
      "learning_rate": 1.938921571998174e-05,
      "loss": 0.9382,
      "step": 340
    },
    {
      "epoch": 0.11516379601485985,
      "grad_norm": 1.6019755601882935,
      "learning_rate": 1.9385546828252653e-05,
      "loss": 0.9601,
      "step": 341
    },
    {
      "epoch": 0.11550151975683891,
      "grad_norm": 1.7641814947128296,
      "learning_rate": 1.9381867299481754e-05,
      "loss": 1.089,
      "step": 342
    },
    {
      "epoch": 0.11583924349881797,
      "grad_norm": 1.5759930610656738,
      "learning_rate": 1.9378177137839214e-05,
      "loss": 0.9215,
      "step": 343
    },
    {
      "epoch": 0.11617696724079703,
      "grad_norm": 1.630548357963562,
      "learning_rate": 1.9374476347507247e-05,
      "loss": 1.0236,
      "step": 344
    },
    {
      "epoch": 0.11651469098277609,
      "grad_norm": 1.4785456657409668,
      "learning_rate": 1.937076493268012e-05,
      "loss": 1.0571,
      "step": 345
    },
    {
      "epoch": 0.11685241472475515,
      "grad_norm": 1.3563222885131836,
      "learning_rate": 1.9367042897564142e-05,
      "loss": 0.9682,
      "step": 346
    },
    {
      "epoch": 0.1171901384667342,
      "grad_norm": 1.4236825704574585,
      "learning_rate": 1.9363310246377648e-05,
      "loss": 1.1513,
      "step": 347
    },
    {
      "epoch": 0.11752786220871327,
      "grad_norm": 1.7846503257751465,
      "learning_rate": 1.9359566983351015e-05,
      "loss": 0.8615,
      "step": 348
    },
    {
      "epoch": 0.11786558595069234,
      "grad_norm": 1.5891832113265991,
      "learning_rate": 1.935581311272664e-05,
      "loss": 0.9746,
      "step": 349
    },
    {
      "epoch": 0.1182033096926714,
      "grad_norm": 1.6135145425796509,
      "learning_rate": 1.9352048638758948e-05,
      "loss": 1.0531,
      "step": 350
    },
    {
      "epoch": 0.11854103343465046,
      "grad_norm": 1.8050966262817383,
      "learning_rate": 1.9348273565714378e-05,
      "loss": 0.9069,
      "step": 351
    },
    {
      "epoch": 0.11887875717662952,
      "grad_norm": 1.7311831712722778,
      "learning_rate": 1.934448789787138e-05,
      "loss": 0.9866,
      "step": 352
    },
    {
      "epoch": 0.11921648091860858,
      "grad_norm": 1.3858376741409302,
      "learning_rate": 1.9340691639520414e-05,
      "loss": 1.0352,
      "step": 353
    },
    {
      "epoch": 0.11955420466058764,
      "grad_norm": 1.3783371448516846,
      "learning_rate": 1.9336884794963946e-05,
      "loss": 1.0727,
      "step": 354
    },
    {
      "epoch": 0.1198919284025667,
      "grad_norm": 1.586795687675476,
      "learning_rate": 1.9333067368516428e-05,
      "loss": 1.1081,
      "step": 355
    },
    {
      "epoch": 0.12022965214454576,
      "grad_norm": 1.392831802368164,
      "learning_rate": 1.9329239364504323e-05,
      "loss": 1.0262,
      "step": 356
    },
    {
      "epoch": 0.12056737588652482,
      "grad_norm": 1.422114610671997,
      "learning_rate": 1.9325400787266065e-05,
      "loss": 1.0495,
      "step": 357
    },
    {
      "epoch": 0.12090509962850389,
      "grad_norm": 1.4294555187225342,
      "learning_rate": 1.9321551641152076e-05,
      "loss": 0.9719,
      "step": 358
    },
    {
      "epoch": 0.12124282337048295,
      "grad_norm": 1.3650896549224854,
      "learning_rate": 1.9317691930524763e-05,
      "loss": 1.0016,
      "step": 359
    },
    {
      "epoch": 0.12158054711246201,
      "grad_norm": 1.5222194194793701,
      "learning_rate": 1.9313821659758498e-05,
      "loss": 1.0295,
      "step": 360
    },
    {
      "epoch": 0.12191827085444107,
      "grad_norm": 1.5866661071777344,
      "learning_rate": 1.9309940833239628e-05,
      "loss": 0.8911,
      "step": 361
    },
    {
      "epoch": 0.12225599459642013,
      "grad_norm": 1.5298398733139038,
      "learning_rate": 1.9306049455366453e-05,
      "loss": 0.9782,
      "step": 362
    },
    {
      "epoch": 0.12259371833839919,
      "grad_norm": 1.773601770401001,
      "learning_rate": 1.9302147530549245e-05,
      "loss": 1.0153,
      "step": 363
    },
    {
      "epoch": 0.12293144208037825,
      "grad_norm": 1.406482219696045,
      "learning_rate": 1.929823506321022e-05,
      "loss": 1.0614,
      "step": 364
    },
    {
      "epoch": 0.12326916582235731,
      "grad_norm": 1.5339316129684448,
      "learning_rate": 1.9294312057783543e-05,
      "loss": 1.0036,
      "step": 365
    },
    {
      "epoch": 0.12360688956433637,
      "grad_norm": 1.456047534942627,
      "learning_rate": 1.9290378518715327e-05,
      "loss": 0.9798,
      "step": 366
    },
    {
      "epoch": 0.12394461330631544,
      "grad_norm": 1.4721204042434692,
      "learning_rate": 1.928643445046362e-05,
      "loss": 1.0214,
      "step": 367
    },
    {
      "epoch": 0.1242823370482945,
      "grad_norm": 1.1126680374145508,
      "learning_rate": 1.9282479857498395e-05,
      "loss": 0.5866,
      "step": 368
    },
    {
      "epoch": 0.12462006079027356,
      "grad_norm": 1.523270845413208,
      "learning_rate": 1.9278514744301573e-05,
      "loss": 0.9177,
      "step": 369
    },
    {
      "epoch": 0.12495778453225262,
      "grad_norm": 1.4886374473571777,
      "learning_rate": 1.927453911536698e-05,
      "loss": 1.0699,
      "step": 370
    },
    {
      "epoch": 0.12529550827423167,
      "grad_norm": 1.6067849397659302,
      "learning_rate": 1.927055297520037e-05,
      "loss": 0.9982,
      "step": 371
    },
    {
      "epoch": 0.12563323201621074,
      "grad_norm": 1.3580256700515747,
      "learning_rate": 1.9266556328319402e-05,
      "loss": 1.0022,
      "step": 372
    },
    {
      "epoch": 0.1259709557581898,
      "grad_norm": 1.614216923713684,
      "learning_rate": 1.926254917925365e-05,
      "loss": 1.0241,
      "step": 373
    },
    {
      "epoch": 0.12630867950016886,
      "grad_norm": 1.722751259803772,
      "learning_rate": 1.9258531532544586e-05,
      "loss": 0.9687,
      "step": 374
    },
    {
      "epoch": 0.12664640324214793,
      "grad_norm": 1.6124260425567627,
      "learning_rate": 1.9254503392745584e-05,
      "loss": 0.9543,
      "step": 375
    },
    {
      "epoch": 0.12698412698412698,
      "grad_norm": 1.4939568042755127,
      "learning_rate": 1.9250464764421905e-05,
      "loss": 1.0683,
      "step": 376
    },
    {
      "epoch": 0.12732185072610605,
      "grad_norm": 1.5119009017944336,
      "learning_rate": 1.9246415652150694e-05,
      "loss": 1.0656,
      "step": 377
    },
    {
      "epoch": 0.1276595744680851,
      "grad_norm": 1.3209376335144043,
      "learning_rate": 1.9242356060520992e-05,
      "loss": 0.9558,
      "step": 378
    },
    {
      "epoch": 0.12799729821006417,
      "grad_norm": 1.6999938488006592,
      "learning_rate": 1.92382859941337e-05,
      "loss": 1.0239,
      "step": 379
    },
    {
      "epoch": 0.12833502195204322,
      "grad_norm": 1.7199757099151611,
      "learning_rate": 1.9234205457601608e-05,
      "loss": 1.042,
      "step": 380
    },
    {
      "epoch": 0.1286727456940223,
      "grad_norm": 1.466601848602295,
      "learning_rate": 1.9230114455549355e-05,
      "loss": 0.9706,
      "step": 381
    },
    {
      "epoch": 0.12901046943600136,
      "grad_norm": 1.4318078756332397,
      "learning_rate": 1.922601299261345e-05,
      "loss": 1.0345,
      "step": 382
    },
    {
      "epoch": 0.1293481931779804,
      "grad_norm": 1.4411404132843018,
      "learning_rate": 1.922190107344226e-05,
      "loss": 1.018,
      "step": 383
    },
    {
      "epoch": 0.12968591691995948,
      "grad_norm": 1.8353866338729858,
      "learning_rate": 1.921777870269599e-05,
      "loss": 0.8481,
      "step": 384
    },
    {
      "epoch": 0.13002364066193853,
      "grad_norm": 1.7054131031036377,
      "learning_rate": 1.921364588504671e-05,
      "loss": 0.9717,
      "step": 385
    },
    {
      "epoch": 0.1303613644039176,
      "grad_norm": 1.662449836730957,
      "learning_rate": 1.920950262517831e-05,
      "loss": 1.0683,
      "step": 386
    },
    {
      "epoch": 0.13069908814589665,
      "grad_norm": 1.590417504310608,
      "learning_rate": 1.9205348927786533e-05,
      "loss": 0.9932,
      "step": 387
    },
    {
      "epoch": 0.13103681188787572,
      "grad_norm": 1.5271410942077637,
      "learning_rate": 1.9201184797578937e-05,
      "loss": 0.971,
      "step": 388
    },
    {
      "epoch": 0.13137453562985477,
      "grad_norm": 1.4209152460098267,
      "learning_rate": 1.919701023927491e-05,
      "loss": 0.9814,
      "step": 389
    },
    {
      "epoch": 0.13171225937183384,
      "grad_norm": 1.4118589162826538,
      "learning_rate": 1.919282525760565e-05,
      "loss": 0.9293,
      "step": 390
    },
    {
      "epoch": 0.1320499831138129,
      "grad_norm": 1.4969202280044556,
      "learning_rate": 1.9188629857314186e-05,
      "loss": 0.9934,
      "step": 391
    },
    {
      "epoch": 0.13238770685579196,
      "grad_norm": 1.4557698965072632,
      "learning_rate": 1.918442404315534e-05,
      "loss": 0.976,
      "step": 392
    },
    {
      "epoch": 0.13272543059777103,
      "grad_norm": 1.524240493774414,
      "learning_rate": 1.9180207819895745e-05,
      "loss": 0.9948,
      "step": 393
    },
    {
      "epoch": 0.13306315433975008,
      "grad_norm": 1.8682528734207153,
      "learning_rate": 1.9175981192313822e-05,
      "loss": 0.895,
      "step": 394
    },
    {
      "epoch": 0.13340087808172915,
      "grad_norm": 1.7607568502426147,
      "learning_rate": 1.9171744165199792e-05,
      "loss": 0.9007,
      "step": 395
    },
    {
      "epoch": 0.1337386018237082,
      "grad_norm": 1.7495015859603882,
      "learning_rate": 1.916749674335566e-05,
      "loss": 0.8882,
      "step": 396
    },
    {
      "epoch": 0.13407632556568727,
      "grad_norm": 1.5924961566925049,
      "learning_rate": 1.9163238931595205e-05,
      "loss": 1.0028,
      "step": 397
    },
    {
      "epoch": 0.13441404930766632,
      "grad_norm": 1.534024715423584,
      "learning_rate": 1.9158970734743994e-05,
      "loss": 1.1062,
      "step": 398
    },
    {
      "epoch": 0.1347517730496454,
      "grad_norm": 1.60129714012146,
      "learning_rate": 1.9154692157639357e-05,
      "loss": 0.9326,
      "step": 399
    },
    {
      "epoch": 0.13508949679162446,
      "grad_norm": 1.5313612222671509,
      "learning_rate": 1.9150403205130384e-05,
      "loss": 0.968,
      "step": 400
    },
    {
      "epoch": 0.1354272205336035,
      "grad_norm": 1.6020941734313965,
      "learning_rate": 1.914610388207793e-05,
      "loss": 0.9906,
      "step": 401
    },
    {
      "epoch": 0.13576494427558258,
      "grad_norm": 1.4641673564910889,
      "learning_rate": 1.91417941933546e-05,
      "loss": 1.004,
      "step": 402
    },
    {
      "epoch": 0.13610266801756163,
      "grad_norm": 1.4751704931259155,
      "learning_rate": 1.9137474143844758e-05,
      "loss": 0.9484,
      "step": 403
    },
    {
      "epoch": 0.1364403917595407,
      "grad_norm": 1.5967063903808594,
      "learning_rate": 1.9133143738444485e-05,
      "loss": 1.1317,
      "step": 404
    },
    {
      "epoch": 0.13677811550151975,
      "grad_norm": 1.8189454078674316,
      "learning_rate": 1.9128802982061628e-05,
      "loss": 0.9014,
      "step": 405
    },
    {
      "epoch": 0.13711583924349882,
      "grad_norm": 1.3561347723007202,
      "learning_rate": 1.912445187961575e-05,
      "loss": 1.021,
      "step": 406
    },
    {
      "epoch": 0.13745356298547787,
      "grad_norm": 1.3975261449813843,
      "learning_rate": 1.9120090436038137e-05,
      "loss": 1.0304,
      "step": 407
    },
    {
      "epoch": 0.13779128672745694,
      "grad_norm": 1.4179965257644653,
      "learning_rate": 1.9115718656271803e-05,
      "loss": 0.9831,
      "step": 408
    },
    {
      "epoch": 0.13812901046943601,
      "grad_norm": 1.1856952905654907,
      "learning_rate": 1.911133654527147e-05,
      "loss": 0.6458,
      "step": 409
    },
    {
      "epoch": 0.13846673421141506,
      "grad_norm": 1.1118652820587158,
      "learning_rate": 1.9106944108003574e-05,
      "loss": 0.6916,
      "step": 410
    },
    {
      "epoch": 0.13880445795339413,
      "grad_norm": 1.5765239000320435,
      "learning_rate": 1.9102541349446252e-05,
      "loss": 1.0141,
      "step": 411
    },
    {
      "epoch": 0.13914218169537318,
      "grad_norm": 1.3771705627441406,
      "learning_rate": 1.909812827458934e-05,
      "loss": 0.9753,
      "step": 412
    },
    {
      "epoch": 0.13947990543735225,
      "grad_norm": 1.5254392623901367,
      "learning_rate": 1.909370488843436e-05,
      "loss": 1.0779,
      "step": 413
    },
    {
      "epoch": 0.1398176291793313,
      "grad_norm": 1.7858482599258423,
      "learning_rate": 1.908927119599453e-05,
      "loss": 0.9657,
      "step": 414
    },
    {
      "epoch": 0.14015535292131037,
      "grad_norm": 1.3768337965011597,
      "learning_rate": 1.9084827202294742e-05,
      "loss": 0.9824,
      "step": 415
    },
    {
      "epoch": 0.14049307666328942,
      "grad_norm": 1.3407484292984009,
      "learning_rate": 1.9080372912371562e-05,
      "loss": 1.0127,
      "step": 416
    },
    {
      "epoch": 0.1408308004052685,
      "grad_norm": 1.7983311414718628,
      "learning_rate": 1.907590833127323e-05,
      "loss": 0.8017,
      "step": 417
    },
    {
      "epoch": 0.14116852414724756,
      "grad_norm": 1.4739267826080322,
      "learning_rate": 1.9071433464059647e-05,
      "loss": 1.03,
      "step": 418
    },
    {
      "epoch": 0.1415062478892266,
      "grad_norm": 1.7468547821044922,
      "learning_rate": 1.9066948315802363e-05,
      "loss": 0.8483,
      "step": 419
    },
    {
      "epoch": 0.14184397163120568,
      "grad_norm": 1.332759141921997,
      "learning_rate": 1.90624528915846e-05,
      "loss": 1.0492,
      "step": 420
    },
    {
      "epoch": 0.14218169537318473,
      "grad_norm": 1.4422720670700073,
      "learning_rate": 1.9057947196501213e-05,
      "loss": 1.0105,
      "step": 421
    },
    {
      "epoch": 0.1425194191151638,
      "grad_norm": 1.5826495885849,
      "learning_rate": 1.9053431235658695e-05,
      "loss": 0.9631,
      "step": 422
    },
    {
      "epoch": 0.14285714285714285,
      "grad_norm": 1.694396734237671,
      "learning_rate": 1.9048905014175178e-05,
      "loss": 0.8336,
      "step": 423
    },
    {
      "epoch": 0.14319486659912192,
      "grad_norm": 1.5038524866104126,
      "learning_rate": 1.9044368537180424e-05,
      "loss": 0.9966,
      "step": 424
    },
    {
      "epoch": 0.14353259034110097,
      "grad_norm": 1.5683435201644897,
      "learning_rate": 1.903982180981581e-05,
      "loss": 1.0206,
      "step": 425
    },
    {
      "epoch": 0.14387031408308004,
      "grad_norm": 1.5082347393035889,
      "learning_rate": 1.9035264837234347e-05,
      "loss": 0.9925,
      "step": 426
    },
    {
      "epoch": 0.14420803782505912,
      "grad_norm": 1.7863913774490356,
      "learning_rate": 1.9030697624600644e-05,
      "loss": 0.9431,
      "step": 427
    },
    {
      "epoch": 0.14454576156703816,
      "grad_norm": 1.570063591003418,
      "learning_rate": 1.9026120177090915e-05,
      "loss": 0.9308,
      "step": 428
    },
    {
      "epoch": 0.14488348530901723,
      "grad_norm": 1.3439537286758423,
      "learning_rate": 1.9021532499892982e-05,
      "loss": 0.9581,
      "step": 429
    },
    {
      "epoch": 0.14522120905099628,
      "grad_norm": 1.5545659065246582,
      "learning_rate": 1.901693459820625e-05,
      "loss": 0.9606,
      "step": 430
    },
    {
      "epoch": 0.14555893279297535,
      "grad_norm": 1.7860249280929565,
      "learning_rate": 1.901232647724172e-05,
      "loss": 1.0994,
      "step": 431
    },
    {
      "epoch": 0.1458966565349544,
      "grad_norm": 1.5848186016082764,
      "learning_rate": 1.9007708142221975e-05,
      "loss": 1.0094,
      "step": 432
    },
    {
      "epoch": 0.14623438027693347,
      "grad_norm": 1.560357928276062,
      "learning_rate": 1.9003079598381167e-05,
      "loss": 0.9539,
      "step": 433
    },
    {
      "epoch": 0.14657210401891252,
      "grad_norm": 1.552776575088501,
      "learning_rate": 1.8998440850965027e-05,
      "loss": 0.9927,
      "step": 434
    },
    {
      "epoch": 0.1469098277608916,
      "grad_norm": 1.481494426727295,
      "learning_rate": 1.8993791905230846e-05,
      "loss": 0.9884,
      "step": 435
    },
    {
      "epoch": 0.14724755150287064,
      "grad_norm": 1.5752434730529785,
      "learning_rate": 1.8989132766447474e-05,
      "loss": 0.9467,
      "step": 436
    },
    {
      "epoch": 0.1475852752448497,
      "grad_norm": 1.869407296180725,
      "learning_rate": 1.8984463439895305e-05,
      "loss": 0.9387,
      "step": 437
    },
    {
      "epoch": 0.14792299898682879,
      "grad_norm": 1.5623244047164917,
      "learning_rate": 1.89797839308663e-05,
      "loss": 0.9431,
      "step": 438
    },
    {
      "epoch": 0.14826072272880783,
      "grad_norm": 1.8062375783920288,
      "learning_rate": 1.897509424466393e-05,
      "loss": 0.9283,
      "step": 439
    },
    {
      "epoch": 0.1485984464707869,
      "grad_norm": 1.496391773223877,
      "learning_rate": 1.897039438660323e-05,
      "loss": 0.9474,
      "step": 440
    },
    {
      "epoch": 0.14893617021276595,
      "grad_norm": 1.8720375299453735,
      "learning_rate": 1.8965684362010745e-05,
      "loss": 0.9655,
      "step": 441
    },
    {
      "epoch": 0.14927389395474502,
      "grad_norm": 1.592070460319519,
      "learning_rate": 1.8960964176224546e-05,
      "loss": 0.9259,
      "step": 442
    },
    {
      "epoch": 0.14961161769672407,
      "grad_norm": 1.7660655975341797,
      "learning_rate": 1.8956233834594225e-05,
      "loss": 0.9902,
      "step": 443
    },
    {
      "epoch": 0.14994934143870314,
      "grad_norm": 1.5559897422790527,
      "learning_rate": 1.895149334248088e-05,
      "loss": 1.1129,
      "step": 444
    },
    {
      "epoch": 0.1502870651806822,
      "grad_norm": 1.4823886156082153,
      "learning_rate": 1.8946742705257112e-05,
      "loss": 0.9204,
      "step": 445
    },
    {
      "epoch": 0.15062478892266126,
      "grad_norm": 1.8868902921676636,
      "learning_rate": 1.8941981928307023e-05,
      "loss": 1.0001,
      "step": 446
    },
    {
      "epoch": 0.15096251266464034,
      "grad_norm": 1.7174978256225586,
      "learning_rate": 1.8937211017026205e-05,
      "loss": 0.9826,
      "step": 447
    },
    {
      "epoch": 0.15130023640661938,
      "grad_norm": 1.8747516870498657,
      "learning_rate": 1.8932429976821733e-05,
      "loss": 0.8949,
      "step": 448
    },
    {
      "epoch": 0.15163796014859846,
      "grad_norm": 1.8017258644104004,
      "learning_rate": 1.892763881311217e-05,
      "loss": 0.879,
      "step": 449
    },
    {
      "epoch": 0.1519756838905775,
      "grad_norm": 1.8290693759918213,
      "learning_rate": 1.892283753132754e-05,
      "loss": 0.929,
      "step": 450
    },
    {
      "epoch": 0.15231340763255657,
      "grad_norm": 1.475809097290039,
      "learning_rate": 1.891802613690935e-05,
      "loss": 0.9705,
      "step": 451
    },
    {
      "epoch": 0.15265113137453562,
      "grad_norm": 1.7010360956192017,
      "learning_rate": 1.8913204635310548e-05,
      "loss": 0.9999,
      "step": 452
    },
    {
      "epoch": 0.1529888551165147,
      "grad_norm": 2.004819631576538,
      "learning_rate": 1.8908373031995558e-05,
      "loss": 0.8903,
      "step": 453
    },
    {
      "epoch": 0.15332657885849374,
      "grad_norm": 1.6799027919769287,
      "learning_rate": 1.8903531332440236e-05,
      "loss": 1.0354,
      "step": 454
    },
    {
      "epoch": 0.1536643026004728,
      "grad_norm": 1.9644033908843994,
      "learning_rate": 1.8898679542131896e-05,
      "loss": 0.925,
      "step": 455
    },
    {
      "epoch": 0.1540020263424519,
      "grad_norm": 1.5758953094482422,
      "learning_rate": 1.8893817666569266e-05,
      "loss": 0.9803,
      "step": 456
    },
    {
      "epoch": 0.15433975008443093,
      "grad_norm": 1.5964207649230957,
      "learning_rate": 1.8888945711262532e-05,
      "loss": 1.0137,
      "step": 457
    },
    {
      "epoch": 0.15467747382641,
      "grad_norm": 1.7565548419952393,
      "learning_rate": 1.8884063681733278e-05,
      "loss": 0.9274,
      "step": 458
    },
    {
      "epoch": 0.15501519756838905,
      "grad_norm": 1.4115689992904663,
      "learning_rate": 1.8879171583514524e-05,
      "loss": 0.9136,
      "step": 459
    },
    {
      "epoch": 0.15535292131036813,
      "grad_norm": 1.6017147302627563,
      "learning_rate": 1.887426942215069e-05,
      "loss": 0.9333,
      "step": 460
    },
    {
      "epoch": 0.15569064505234717,
      "grad_norm": 1.5487878322601318,
      "learning_rate": 1.886935720319761e-05,
      "loss": 0.9818,
      "step": 461
    },
    {
      "epoch": 0.15602836879432624,
      "grad_norm": 2.036057233810425,
      "learning_rate": 1.88644349322225e-05,
      "loss": 0.8964,
      "step": 462
    },
    {
      "epoch": 0.1563660925363053,
      "grad_norm": 1.7509297132492065,
      "learning_rate": 1.8859502614804e-05,
      "loss": 0.9564,
      "step": 463
    },
    {
      "epoch": 0.15670381627828436,
      "grad_norm": 1.4644593000411987,
      "learning_rate": 1.8854560256532098e-05,
      "loss": 1.0363,
      "step": 464
    },
    {
      "epoch": 0.15704154002026344,
      "grad_norm": 1.4420108795166016,
      "learning_rate": 1.884960786300819e-05,
      "loss": 1.0113,
      "step": 465
    },
    {
      "epoch": 0.15737926376224248,
      "grad_norm": 1.55623197555542,
      "learning_rate": 1.884464543984504e-05,
      "loss": 0.9531,
      "step": 466
    },
    {
      "epoch": 0.15771698750422156,
      "grad_norm": 1.7645419836044312,
      "learning_rate": 1.8839672992666763e-05,
      "loss": 0.9253,
      "step": 467
    },
    {
      "epoch": 0.1580547112462006,
      "grad_norm": 1.5681535005569458,
      "learning_rate": 1.8834690527108847e-05,
      "loss": 1.0326,
      "step": 468
    },
    {
      "epoch": 0.15839243498817968,
      "grad_norm": 1.7348835468292236,
      "learning_rate": 1.8829698048818142e-05,
      "loss": 0.9444,
      "step": 469
    },
    {
      "epoch": 0.15873015873015872,
      "grad_norm": 1.5393171310424805,
      "learning_rate": 1.882469556345283e-05,
      "loss": 0.9953,
      "step": 470
    },
    {
      "epoch": 0.1590678824721378,
      "grad_norm": 2.0154402256011963,
      "learning_rate": 1.8819683076682444e-05,
      "loss": 0.8975,
      "step": 471
    },
    {
      "epoch": 0.15940560621411684,
      "grad_norm": 1.5245670080184937,
      "learning_rate": 1.881466059418785e-05,
      "loss": 0.9831,
      "step": 472
    },
    {
      "epoch": 0.15974332995609591,
      "grad_norm": 1.6321852207183838,
      "learning_rate": 1.8809628121661247e-05,
      "loss": 0.9532,
      "step": 473
    },
    {
      "epoch": 0.160081053698075,
      "grad_norm": 1.7098428010940552,
      "learning_rate": 1.8804585664806143e-05,
      "loss": 0.9892,
      "step": 474
    },
    {
      "epoch": 0.16041877744005403,
      "grad_norm": 1.5354005098342896,
      "learning_rate": 1.879953322933737e-05,
      "loss": 0.9775,
      "step": 475
    },
    {
      "epoch": 0.1607565011820331,
      "grad_norm": 1.3887295722961426,
      "learning_rate": 1.8794470820981083e-05,
      "loss": 1.0574,
      "step": 476
    },
    {
      "epoch": 0.16109422492401215,
      "grad_norm": 1.5925416946411133,
      "learning_rate": 1.878939844547471e-05,
      "loss": 0.9661,
      "step": 477
    },
    {
      "epoch": 0.16143194866599123,
      "grad_norm": 1.6338387727737427,
      "learning_rate": 1.8784316108566994e-05,
      "loss": 0.951,
      "step": 478
    },
    {
      "epoch": 0.16176967240797027,
      "grad_norm": 1.496110200881958,
      "learning_rate": 1.8779223816017973e-05,
      "loss": 0.9938,
      "step": 479
    },
    {
      "epoch": 0.16210739614994935,
      "grad_norm": 1.6870453357696533,
      "learning_rate": 1.8774121573598956e-05,
      "loss": 0.8811,
      "step": 480
    },
    {
      "epoch": 0.1624451198919284,
      "grad_norm": 1.867260456085205,
      "learning_rate": 1.8769009387092533e-05,
      "loss": 0.9585,
      "step": 481
    },
    {
      "epoch": 0.16278284363390746,
      "grad_norm": 1.6198762655258179,
      "learning_rate": 1.8763887262292557e-05,
      "loss": 1.0059,
      "step": 482
    },
    {
      "epoch": 0.16312056737588654,
      "grad_norm": 1.6184449195861816,
      "learning_rate": 1.875875520500416e-05,
      "loss": 0.9373,
      "step": 483
    },
    {
      "epoch": 0.16345829111786558,
      "grad_norm": 1.4818060398101807,
      "learning_rate": 1.8753613221043722e-05,
      "loss": 0.9916,
      "step": 484
    },
    {
      "epoch": 0.16379601485984466,
      "grad_norm": 1.6538429260253906,
      "learning_rate": 1.8748461316238865e-05,
      "loss": 1.0418,
      "step": 485
    },
    {
      "epoch": 0.1641337386018237,
      "grad_norm": 1.6286081075668335,
      "learning_rate": 1.874329949642847e-05,
      "loss": 0.9622,
      "step": 486
    },
    {
      "epoch": 0.16447146234380278,
      "grad_norm": 1.3797705173492432,
      "learning_rate": 1.873812776746265e-05,
      "loss": 1.0605,
      "step": 487
    },
    {
      "epoch": 0.16480918608578182,
      "grad_norm": 1.473914623260498,
      "learning_rate": 1.873294613520274e-05,
      "loss": 0.9471,
      "step": 488
    },
    {
      "epoch": 0.1651469098277609,
      "grad_norm": 1.8592287302017212,
      "learning_rate": 1.8727754605521307e-05,
      "loss": 0.9974,
      "step": 489
    },
    {
      "epoch": 0.16548463356973994,
      "grad_norm": 1.6448007822036743,
      "learning_rate": 1.872255318430214e-05,
      "loss": 1.0302,
      "step": 490
    },
    {
      "epoch": 0.16582235731171902,
      "grad_norm": 1.6616442203521729,
      "learning_rate": 1.8717341877440227e-05,
      "loss": 0.9735,
      "step": 491
    },
    {
      "epoch": 0.1661600810536981,
      "grad_norm": 1.7255498170852661,
      "learning_rate": 1.8712120690841767e-05,
      "loss": 1.0061,
      "step": 492
    },
    {
      "epoch": 0.16649780479567713,
      "grad_norm": 1.5620553493499756,
      "learning_rate": 1.870688963042415e-05,
      "loss": 1.0274,
      "step": 493
    },
    {
      "epoch": 0.1668355285376562,
      "grad_norm": 1.6843756437301636,
      "learning_rate": 1.8701648702115966e-05,
      "loss": 1.0585,
      "step": 494
    },
    {
      "epoch": 0.16717325227963525,
      "grad_norm": 1.6597129106521606,
      "learning_rate": 1.8696397911856983e-05,
      "loss": 0.9156,
      "step": 495
    },
    {
      "epoch": 0.16751097602161433,
      "grad_norm": 1.564494013786316,
      "learning_rate": 1.8691137265598144e-05,
      "loss": 0.9558,
      "step": 496
    },
    {
      "epoch": 0.16784869976359337,
      "grad_norm": 1.89838445186615,
      "learning_rate": 1.868586676930156e-05,
      "loss": 0.8052,
      "step": 497
    },
    {
      "epoch": 0.16818642350557245,
      "grad_norm": 1.3948090076446533,
      "learning_rate": 1.868058642894052e-05,
      "loss": 0.9775,
      "step": 498
    },
    {
      "epoch": 0.1685241472475515,
      "grad_norm": 1.884446620941162,
      "learning_rate": 1.8675296250499457e-05,
      "loss": 0.9029,
      "step": 499
    },
    {
      "epoch": 0.16886187098953057,
      "grad_norm": 1.5154876708984375,
      "learning_rate": 1.8669996239973946e-05,
      "loss": 0.9713,
      "step": 500
    },
    {
      "epoch": 0.1691995947315096,
      "grad_norm": 1.8604340553283691,
      "learning_rate": 1.866468640337073e-05,
      "loss": 0.8028,
      "step": 501
    },
    {
      "epoch": 0.16953731847348869,
      "grad_norm": 1.4779167175292969,
      "learning_rate": 1.8659366746707665e-05,
      "loss": 1.0923,
      "step": 502
    },
    {
      "epoch": 0.16987504221546776,
      "grad_norm": 1.552448034286499,
      "learning_rate": 1.865403727601375e-05,
      "loss": 0.9258,
      "step": 503
    },
    {
      "epoch": 0.1702127659574468,
      "grad_norm": 1.6407873630523682,
      "learning_rate": 1.86486979973291e-05,
      "loss": 1.0117,
      "step": 504
    },
    {
      "epoch": 0.17055048969942588,
      "grad_norm": 1.4104267358779907,
      "learning_rate": 1.8643348916704947e-05,
      "loss": 0.8797,
      "step": 505
    },
    {
      "epoch": 0.17088821344140492,
      "grad_norm": 1.5362735986709595,
      "learning_rate": 1.863799004020364e-05,
      "loss": 0.9478,
      "step": 506
    },
    {
      "epoch": 0.171225937183384,
      "grad_norm": 1.9138734340667725,
      "learning_rate": 1.863262137389861e-05,
      "loss": 0.9242,
      "step": 507
    },
    {
      "epoch": 0.17156366092536304,
      "grad_norm": 1.7430192232131958,
      "learning_rate": 1.862724292387441e-05,
      "loss": 0.9525,
      "step": 508
    },
    {
      "epoch": 0.17190138466734212,
      "grad_norm": 1.726288080215454,
      "learning_rate": 1.862185469622666e-05,
      "loss": 0.9031,
      "step": 509
    },
    {
      "epoch": 0.17223910840932116,
      "grad_norm": 1.5301762819290161,
      "learning_rate": 1.8616456697062068e-05,
      "loss": 1.0518,
      "step": 510
    },
    {
      "epoch": 0.17257683215130024,
      "grad_norm": 1.5603861808776855,
      "learning_rate": 1.861104893249842e-05,
      "loss": 0.9528,
      "step": 511
    },
    {
      "epoch": 0.1729145558932793,
      "grad_norm": 1.609186053276062,
      "learning_rate": 1.8605631408664566e-05,
      "loss": 0.9681,
      "step": 512
    },
    {
      "epoch": 0.17325227963525835,
      "grad_norm": 1.5156545639038086,
      "learning_rate": 1.860020413170042e-05,
      "loss": 1.008,
      "step": 513
    },
    {
      "epoch": 0.17359000337723743,
      "grad_norm": 1.6036454439163208,
      "learning_rate": 1.8594767107756943e-05,
      "loss": 0.9897,
      "step": 514
    },
    {
      "epoch": 0.17392772711921647,
      "grad_norm": 1.6897298097610474,
      "learning_rate": 1.8589320342996153e-05,
      "loss": 0.9567,
      "step": 515
    },
    {
      "epoch": 0.17426545086119555,
      "grad_norm": 1.6624140739440918,
      "learning_rate": 1.8583863843591095e-05,
      "loss": 1.0006,
      "step": 516
    },
    {
      "epoch": 0.1746031746031746,
      "grad_norm": 1.5293560028076172,
      "learning_rate": 1.8578397615725857e-05,
      "loss": 0.9494,
      "step": 517
    },
    {
      "epoch": 0.17494089834515367,
      "grad_norm": 1.5202938318252563,
      "learning_rate": 1.857292166559555e-05,
      "loss": 0.9813,
      "step": 518
    },
    {
      "epoch": 0.1752786220871327,
      "grad_norm": 1.7682570219039917,
      "learning_rate": 1.8567435999406297e-05,
      "loss": 1.0424,
      "step": 519
    },
    {
      "epoch": 0.1756163458291118,
      "grad_norm": 1.4950370788574219,
      "learning_rate": 1.8561940623375245e-05,
      "loss": 0.9477,
      "step": 520
    },
    {
      "epoch": 0.17595406957109086,
      "grad_norm": 2.034437656402588,
      "learning_rate": 1.8556435543730537e-05,
      "loss": 0.8438,
      "step": 521
    },
    {
      "epoch": 0.1762917933130699,
      "grad_norm": 1.329776406288147,
      "learning_rate": 1.855092076671131e-05,
      "loss": 0.9924,
      "step": 522
    },
    {
      "epoch": 0.17662951705504898,
      "grad_norm": 1.3395782709121704,
      "learning_rate": 1.85453962985677e-05,
      "loss": 1.0502,
      "step": 523
    },
    {
      "epoch": 0.17696724079702802,
      "grad_norm": 1.5812408924102783,
      "learning_rate": 1.8539862145560828e-05,
      "loss": 0.9642,
      "step": 524
    },
    {
      "epoch": 0.1773049645390071,
      "grad_norm": 1.8044716119766235,
      "learning_rate": 1.853431831396278e-05,
      "loss": 0.921,
      "step": 525
    },
    {
      "epoch": 0.17764268828098614,
      "grad_norm": 1.855987310409546,
      "learning_rate": 1.8528764810056618e-05,
      "loss": 1.046,
      "step": 526
    },
    {
      "epoch": 0.17798041202296522,
      "grad_norm": 1.6714807748794556,
      "learning_rate": 1.8523201640136366e-05,
      "loss": 1.0735,
      "step": 527
    },
    {
      "epoch": 0.17831813576494426,
      "grad_norm": 1.8209046125411987,
      "learning_rate": 1.851762881050701e-05,
      "loss": 0.9023,
      "step": 528
    },
    {
      "epoch": 0.17865585950692334,
      "grad_norm": 1.7834218740463257,
      "learning_rate": 1.8512046327484466e-05,
      "loss": 0.9413,
      "step": 529
    },
    {
      "epoch": 0.1789935832489024,
      "grad_norm": 1.464054822921753,
      "learning_rate": 1.8506454197395608e-05,
      "loss": 1.0273,
      "step": 530
    },
    {
      "epoch": 0.17933130699088146,
      "grad_norm": 1.6190582513809204,
      "learning_rate": 1.8500852426578233e-05,
      "loss": 0.8873,
      "step": 531
    },
    {
      "epoch": 0.17966903073286053,
      "grad_norm": 1.8413068056106567,
      "learning_rate": 1.8495241021381064e-05,
      "loss": 0.9872,
      "step": 532
    },
    {
      "epoch": 0.18000675447483958,
      "grad_norm": 1.578251838684082,
      "learning_rate": 1.8489619988163756e-05,
      "loss": 0.9947,
      "step": 533
    },
    {
      "epoch": 0.18034447821681865,
      "grad_norm": 1.7031548023223877,
      "learning_rate": 1.8483989333296858e-05,
      "loss": 0.9695,
      "step": 534
    },
    {
      "epoch": 0.1806822019587977,
      "grad_norm": 1.8782185316085815,
      "learning_rate": 1.8478349063161834e-05,
      "loss": 0.7757,
      "step": 535
    },
    {
      "epoch": 0.18101992570077677,
      "grad_norm": 1.8156107664108276,
      "learning_rate": 1.8472699184151045e-05,
      "loss": 1.028,
      "step": 536
    },
    {
      "epoch": 0.18135764944275581,
      "grad_norm": 1.5370242595672607,
      "learning_rate": 1.846703970266774e-05,
      "loss": 0.9745,
      "step": 537
    },
    {
      "epoch": 0.1816953731847349,
      "grad_norm": 1.6713438034057617,
      "learning_rate": 1.8461370625126053e-05,
      "loss": 0.9936,
      "step": 538
    },
    {
      "epoch": 0.18203309692671396,
      "grad_norm": 1.8412811756134033,
      "learning_rate": 1.8455691957950992e-05,
      "loss": 1.0226,
      "step": 539
    },
    {
      "epoch": 0.182370820668693,
      "grad_norm": 2.0089471340179443,
      "learning_rate": 1.845000370757843e-05,
      "loss": 1.0445,
      "step": 540
    },
    {
      "epoch": 0.18270854441067208,
      "grad_norm": 1.7078709602355957,
      "learning_rate": 1.8444305880455108e-05,
      "loss": 0.9486,
      "step": 541
    },
    {
      "epoch": 0.18304626815265113,
      "grad_norm": 1.9150617122650146,
      "learning_rate": 1.8438598483038615e-05,
      "loss": 0.8995,
      "step": 542
    },
    {
      "epoch": 0.1833839918946302,
      "grad_norm": 1.6875557899475098,
      "learning_rate": 1.843288152179739e-05,
      "loss": 0.9298,
      "step": 543
    },
    {
      "epoch": 0.18372171563660925,
      "grad_norm": 1.8333115577697754,
      "learning_rate": 1.842715500321071e-05,
      "loss": 0.9052,
      "step": 544
    },
    {
      "epoch": 0.18405943937858832,
      "grad_norm": 2.083130359649658,
      "learning_rate": 1.8421418933768683e-05,
      "loss": 1.0277,
      "step": 545
    },
    {
      "epoch": 0.18439716312056736,
      "grad_norm": 1.6788403987884521,
      "learning_rate": 1.8415673319972246e-05,
      "loss": 0.9974,
      "step": 546
    },
    {
      "epoch": 0.18473488686254644,
      "grad_norm": 1.7654380798339844,
      "learning_rate": 1.840991816833314e-05,
      "loss": 1.0563,
      "step": 547
    },
    {
      "epoch": 0.1850726106045255,
      "grad_norm": 1.597407341003418,
      "learning_rate": 1.840415348537393e-05,
      "loss": 1.0107,
      "step": 548
    },
    {
      "epoch": 0.18541033434650456,
      "grad_norm": 1.5563991069793701,
      "learning_rate": 1.839837927762798e-05,
      "loss": 0.9042,
      "step": 549
    },
    {
      "epoch": 0.18574805808848363,
      "grad_norm": 1.7205748558044434,
      "learning_rate": 1.8392595551639443e-05,
      "loss": 0.8364,
      "step": 550
    },
    {
      "epoch": 0.18608578183046268,
      "grad_norm": 1.6478700637817383,
      "learning_rate": 1.8386802313963262e-05,
      "loss": 0.8956,
      "step": 551
    },
    {
      "epoch": 0.18642350557244175,
      "grad_norm": 1.8371875286102295,
      "learning_rate": 1.838099957116517e-05,
      "loss": 0.9429,
      "step": 552
    },
    {
      "epoch": 0.1867612293144208,
      "grad_norm": 1.7303746938705444,
      "learning_rate": 1.837518732982165e-05,
      "loss": 0.9506,
      "step": 553
    },
    {
      "epoch": 0.18709895305639987,
      "grad_norm": 1.606325387954712,
      "learning_rate": 1.8369365596519973e-05,
      "loss": 0.8852,
      "step": 554
    },
    {
      "epoch": 0.18743667679837892,
      "grad_norm": 1.6578656435012817,
      "learning_rate": 1.836353437785816e-05,
      "loss": 0.9402,
      "step": 555
    },
    {
      "epoch": 0.187774400540358,
      "grad_norm": 1.713122010231018,
      "learning_rate": 1.8357693680444978e-05,
      "loss": 0.9213,
      "step": 556
    },
    {
      "epoch": 0.18811212428233706,
      "grad_norm": 1.6525804996490479,
      "learning_rate": 1.8351843510899933e-05,
      "loss": 0.9977,
      "step": 557
    },
    {
      "epoch": 0.1884498480243161,
      "grad_norm": 1.6452568769454956,
      "learning_rate": 1.834598387585328e-05,
      "loss": 0.8998,
      "step": 558
    },
    {
      "epoch": 0.18878757176629518,
      "grad_norm": 1.643561601638794,
      "learning_rate": 1.8340114781945993e-05,
      "loss": 0.995,
      "step": 559
    },
    {
      "epoch": 0.18912529550827423,
      "grad_norm": 1.4392259120941162,
      "learning_rate": 1.8334236235829764e-05,
      "loss": 0.9965,
      "step": 560
    },
    {
      "epoch": 0.1894630192502533,
      "grad_norm": 1.6428143978118896,
      "learning_rate": 1.8328348244167005e-05,
      "loss": 0.9189,
      "step": 561
    },
    {
      "epoch": 0.18980074299223235,
      "grad_norm": 1.807239294052124,
      "learning_rate": 1.8322450813630826e-05,
      "loss": 0.9478,
      "step": 562
    },
    {
      "epoch": 0.19013846673421142,
      "grad_norm": 1.8083332777023315,
      "learning_rate": 1.831654395090504e-05,
      "loss": 0.8151,
      "step": 563
    },
    {
      "epoch": 0.19047619047619047,
      "grad_norm": 1.7193403244018555,
      "learning_rate": 1.8310627662684145e-05,
      "loss": 0.9987,
      "step": 564
    },
    {
      "epoch": 0.19081391421816954,
      "grad_norm": 1.7191373109817505,
      "learning_rate": 1.8304701955673322e-05,
      "loss": 0.9495,
      "step": 565
    },
    {
      "epoch": 0.19115163796014858,
      "grad_norm": 1.5498125553131104,
      "learning_rate": 1.829876683658843e-05,
      "loss": 0.9234,
      "step": 566
    },
    {
      "epoch": 0.19148936170212766,
      "grad_norm": 1.5695441961288452,
      "learning_rate": 1.8292822312155997e-05,
      "loss": 0.9652,
      "step": 567
    },
    {
      "epoch": 0.19182708544410673,
      "grad_norm": 1.586310625076294,
      "learning_rate": 1.8286868389113202e-05,
      "loss": 0.9207,
      "step": 568
    },
    {
      "epoch": 0.19216480918608578,
      "grad_norm": 1.5127084255218506,
      "learning_rate": 1.8280905074207886e-05,
      "loss": 0.9235,
      "step": 569
    },
    {
      "epoch": 0.19250253292806485,
      "grad_norm": 1.7495170831680298,
      "learning_rate": 1.827493237419852e-05,
      "loss": 1.0483,
      "step": 570
    },
    {
      "epoch": 0.1928402566700439,
      "grad_norm": 1.518812656402588,
      "learning_rate": 1.826895029585423e-05,
      "loss": 0.9699,
      "step": 571
    },
    {
      "epoch": 0.19317798041202297,
      "grad_norm": 1.9115099906921387,
      "learning_rate": 1.8262958845954755e-05,
      "loss": 0.9492,
      "step": 572
    },
    {
      "epoch": 0.19351570415400202,
      "grad_norm": 1.5862581729888916,
      "learning_rate": 1.8256958031290464e-05,
      "loss": 0.9909,
      "step": 573
    },
    {
      "epoch": 0.1938534278959811,
      "grad_norm": 1.8429068326950073,
      "learning_rate": 1.8250947858662337e-05,
      "loss": 1.0641,
      "step": 574
    },
    {
      "epoch": 0.19419115163796014,
      "grad_norm": 1.6671022176742554,
      "learning_rate": 1.8244928334881962e-05,
      "loss": 0.8949,
      "step": 575
    },
    {
      "epoch": 0.1945288753799392,
      "grad_norm": 1.9761885404586792,
      "learning_rate": 1.823889946677152e-05,
      "loss": 0.9599,
      "step": 576
    },
    {
      "epoch": 0.19486659912191828,
      "grad_norm": 1.6153950691223145,
      "learning_rate": 1.8232861261163784e-05,
      "loss": 1.0179,
      "step": 577
    },
    {
      "epoch": 0.19520432286389733,
      "grad_norm": 2.112407922744751,
      "learning_rate": 1.822681372490211e-05,
      "loss": 0.9317,
      "step": 578
    },
    {
      "epoch": 0.1955420466058764,
      "grad_norm": 1.4841777086257935,
      "learning_rate": 1.8220756864840436e-05,
      "loss": 0.8798,
      "step": 579
    },
    {
      "epoch": 0.19587977034785545,
      "grad_norm": 1.5745129585266113,
      "learning_rate": 1.8214690687843255e-05,
      "loss": 0.9736,
      "step": 580
    },
    {
      "epoch": 0.19621749408983452,
      "grad_norm": 2.0149619579315186,
      "learning_rate": 1.8208615200785628e-05,
      "loss": 0.9209,
      "step": 581
    },
    {
      "epoch": 0.19655521783181357,
      "grad_norm": 1.5334727764129639,
      "learning_rate": 1.8202530410553162e-05,
      "loss": 0.9308,
      "step": 582
    },
    {
      "epoch": 0.19689294157379264,
      "grad_norm": 1.9031733274459839,
      "learning_rate": 1.819643632404201e-05,
      "loss": 1.0482,
      "step": 583
    },
    {
      "epoch": 0.19723066531577169,
      "grad_norm": 1.5342808961868286,
      "learning_rate": 1.8190332948158868e-05,
      "loss": 0.9666,
      "step": 584
    },
    {
      "epoch": 0.19756838905775076,
      "grad_norm": 1.537971019744873,
      "learning_rate": 1.818422028982094e-05,
      "loss": 1.0,
      "step": 585
    },
    {
      "epoch": 0.19790611279972983,
      "grad_norm": 1.3718063831329346,
      "learning_rate": 1.8178098355955976e-05,
      "loss": 0.9514,
      "step": 586
    },
    {
      "epoch": 0.19824383654170888,
      "grad_norm": 1.5259064435958862,
      "learning_rate": 1.817196715350222e-05,
      "loss": 1.0383,
      "step": 587
    },
    {
      "epoch": 0.19858156028368795,
      "grad_norm": 1.6124998331069946,
      "learning_rate": 1.8165826689408423e-05,
      "loss": 0.9582,
      "step": 588
    },
    {
      "epoch": 0.198919284025667,
      "grad_norm": 1.6474043130874634,
      "learning_rate": 1.815967697063384e-05,
      "loss": 0.978,
      "step": 589
    },
    {
      "epoch": 0.19925700776764607,
      "grad_norm": 1.494575023651123,
      "learning_rate": 1.8153518004148213e-05,
      "loss": 0.9402,
      "step": 590
    },
    {
      "epoch": 0.19959473150962512,
      "grad_norm": 1.8232345581054688,
      "learning_rate": 1.8147349796931754e-05,
      "loss": 0.9886,
      "step": 591
    },
    {
      "epoch": 0.1999324552516042,
      "grad_norm": 1.6692698001861572,
      "learning_rate": 1.8141172355975163e-05,
      "loss": 0.8999,
      "step": 592
    },
    {
      "epoch": 0.20027017899358324,
      "grad_norm": 1.5491018295288086,
      "learning_rate": 1.8134985688279597e-05,
      "loss": 1.08,
      "step": 593
    },
    {
      "epoch": 0.2006079027355623,
      "grad_norm": 1.9052106142044067,
      "learning_rate": 1.8128789800856666e-05,
      "loss": 1.0565,
      "step": 594
    },
    {
      "epoch": 0.20094562647754138,
      "grad_norm": 1.4865642786026,
      "learning_rate": 1.8122584700728444e-05,
      "loss": 0.9275,
      "step": 595
    },
    {
      "epoch": 0.20128335021952043,
      "grad_norm": 1.4526065587997437,
      "learning_rate": 1.8116370394927424e-05,
      "loss": 1.0406,
      "step": 596
    },
    {
      "epoch": 0.2016210739614995,
      "grad_norm": 1.417462944984436,
      "learning_rate": 1.8110146890496556e-05,
      "loss": 0.6454,
      "step": 597
    },
    {
      "epoch": 0.20195879770347855,
      "grad_norm": 1.6990950107574463,
      "learning_rate": 1.81039141944892e-05,
      "loss": 1.0037,
      "step": 598
    },
    {
      "epoch": 0.20229652144545762,
      "grad_norm": 1.7993725538253784,
      "learning_rate": 1.8097672313969135e-05,
      "loss": 0.8797,
      "step": 599
    },
    {
      "epoch": 0.20263424518743667,
      "grad_norm": 1.8546535968780518,
      "learning_rate": 1.8091421256010553e-05,
      "loss": 1.0385,
      "step": 600
    },
    {
      "epoch": 0.20297196892941574,
      "grad_norm": 1.5344856977462769,
      "learning_rate": 1.8085161027698046e-05,
      "loss": 0.9766,
      "step": 601
    },
    {
      "epoch": 0.2033096926713948,
      "grad_norm": 1.6322252750396729,
      "learning_rate": 1.8078891636126598e-05,
      "loss": 0.9157,
      "step": 602
    },
    {
      "epoch": 0.20364741641337386,
      "grad_norm": 2.0640347003936768,
      "learning_rate": 1.8072613088401575e-05,
      "loss": 0.981,
      "step": 603
    },
    {
      "epoch": 0.20398514015535293,
      "grad_norm": 1.928788423538208,
      "learning_rate": 1.8066325391638733e-05,
      "loss": 0.8622,
      "step": 604
    },
    {
      "epoch": 0.20432286389733198,
      "grad_norm": 1.5575127601623535,
      "learning_rate": 1.806002855296418e-05,
      "loss": 1.0438,
      "step": 605
    },
    {
      "epoch": 0.20466058763931105,
      "grad_norm": 1.6592882871627808,
      "learning_rate": 1.8053722579514396e-05,
      "loss": 0.98,
      "step": 606
    },
    {
      "epoch": 0.2049983113812901,
      "grad_norm": 1.656277060508728,
      "learning_rate": 1.8047407478436208e-05,
      "loss": 0.9753,
      "step": 607
    },
    {
      "epoch": 0.20533603512326917,
      "grad_norm": 1.5151846408843994,
      "learning_rate": 1.804108325688679e-05,
      "loss": 0.9485,
      "step": 608
    },
    {
      "epoch": 0.20567375886524822,
      "grad_norm": 1.8576091527938843,
      "learning_rate": 1.8034749922033655e-05,
      "loss": 0.9773,
      "step": 609
    },
    {
      "epoch": 0.2060114826072273,
      "grad_norm": 1.584010362625122,
      "learning_rate": 1.8028407481054643e-05,
      "loss": 0.9487,
      "step": 610
    },
    {
      "epoch": 0.20634920634920634,
      "grad_norm": 1.4198172092437744,
      "learning_rate": 1.8022055941137916e-05,
      "loss": 1.0212,
      "step": 611
    },
    {
      "epoch": 0.2066869300911854,
      "grad_norm": 1.612269401550293,
      "learning_rate": 1.8015695309481936e-05,
      "loss": 0.9287,
      "step": 612
    },
    {
      "epoch": 0.20702465383316448,
      "grad_norm": 1.3393043279647827,
      "learning_rate": 1.800932559329549e-05,
      "loss": 0.6778,
      "step": 613
    },
    {
      "epoch": 0.20736237757514353,
      "grad_norm": 1.967218041419983,
      "learning_rate": 1.8002946799797646e-05,
      "loss": 0.9549,
      "step": 614
    },
    {
      "epoch": 0.2077001013171226,
      "grad_norm": 1.711090326309204,
      "learning_rate": 1.799655893621776e-05,
      "loss": 0.8831,
      "step": 615
    },
    {
      "epoch": 0.20803782505910165,
      "grad_norm": 1.6661006212234497,
      "learning_rate": 1.7990162009795477e-05,
      "loss": 1.0317,
      "step": 616
    },
    {
      "epoch": 0.20837554880108072,
      "grad_norm": 1.69973623752594,
      "learning_rate": 1.7983756027780704e-05,
      "loss": 1.0171,
      "step": 617
    },
    {
      "epoch": 0.20871327254305977,
      "grad_norm": 1.592516303062439,
      "learning_rate": 1.7977340997433616e-05,
      "loss": 0.916,
      "step": 618
    },
    {
      "epoch": 0.20905099628503884,
      "grad_norm": 1.703519344329834,
      "learning_rate": 1.797091692602464e-05,
      "loss": 0.9714,
      "step": 619
    },
    {
      "epoch": 0.2093887200270179,
      "grad_norm": 1.7290936708450317,
      "learning_rate": 1.796448382083445e-05,
      "loss": 0.9887,
      "step": 620
    },
    {
      "epoch": 0.20972644376899696,
      "grad_norm": 1.582982063293457,
      "learning_rate": 1.7958041689153963e-05,
      "loss": 1.0013,
      "step": 621
    },
    {
      "epoch": 0.21006416751097604,
      "grad_norm": 1.749436855316162,
      "learning_rate": 1.795159053828432e-05,
      "loss": 0.9256,
      "step": 622
    },
    {
      "epoch": 0.21040189125295508,
      "grad_norm": 1.7141071557998657,
      "learning_rate": 1.7945130375536886e-05,
      "loss": 0.9881,
      "step": 623
    },
    {
      "epoch": 0.21073961499493415,
      "grad_norm": 1.5005500316619873,
      "learning_rate": 1.793866120823324e-05,
      "loss": 0.9808,
      "step": 624
    },
    {
      "epoch": 0.2110773387369132,
      "grad_norm": 1.6043682098388672,
      "learning_rate": 1.793218304370517e-05,
      "loss": 0.9551,
      "step": 625
    },
    {
      "epoch": 0.21141506247889227,
      "grad_norm": 1.4688178300857544,
      "learning_rate": 1.792569588929465e-05,
      "loss": 1.0461,
      "step": 626
    },
    {
      "epoch": 0.21175278622087132,
      "grad_norm": 1.6248085498809814,
      "learning_rate": 1.7919199752353854e-05,
      "loss": 0.9945,
      "step": 627
    },
    {
      "epoch": 0.2120905099628504,
      "grad_norm": 1.8134740591049194,
      "learning_rate": 1.7912694640245133e-05,
      "loss": 0.8224,
      "step": 628
    },
    {
      "epoch": 0.21242823370482944,
      "grad_norm": 1.827818751335144,
      "learning_rate": 1.7906180560341006e-05,
      "loss": 0.8681,
      "step": 629
    },
    {
      "epoch": 0.2127659574468085,
      "grad_norm": 1.4634346961975098,
      "learning_rate": 1.789965752002416e-05,
      "loss": 0.9307,
      "step": 630
    },
    {
      "epoch": 0.21310368118878756,
      "grad_norm": 1.7828854322433472,
      "learning_rate": 1.789312552668744e-05,
      "loss": 0.9193,
      "step": 631
    },
    {
      "epoch": 0.21344140493076663,
      "grad_norm": 1.5666884183883667,
      "learning_rate": 1.7886584587733823e-05,
      "loss": 1.0102,
      "step": 632
    },
    {
      "epoch": 0.2137791286727457,
      "grad_norm": 1.6436760425567627,
      "learning_rate": 1.7880034710576442e-05,
      "loss": 0.982,
      "step": 633
    },
    {
      "epoch": 0.21411685241472475,
      "grad_norm": 1.517155647277832,
      "learning_rate": 1.7873475902638552e-05,
      "loss": 0.9064,
      "step": 634
    },
    {
      "epoch": 0.21445457615670382,
      "grad_norm": 1.5884089469909668,
      "learning_rate": 1.7866908171353534e-05,
      "loss": 1.0428,
      "step": 635
    },
    {
      "epoch": 0.21479229989868287,
      "grad_norm": 1.7295477390289307,
      "learning_rate": 1.7860331524164873e-05,
      "loss": 0.9774,
      "step": 636
    },
    {
      "epoch": 0.21513002364066194,
      "grad_norm": 1.8955925703048706,
      "learning_rate": 1.7853745968526167e-05,
      "loss": 0.939,
      "step": 637
    },
    {
      "epoch": 0.215467747382641,
      "grad_norm": 1.7358198165893555,
      "learning_rate": 1.784715151190111e-05,
      "loss": 1.1168,
      "step": 638
    },
    {
      "epoch": 0.21580547112462006,
      "grad_norm": 1.5801705121994019,
      "learning_rate": 1.7840548161763477e-05,
      "loss": 1.0106,
      "step": 639
    },
    {
      "epoch": 0.2161431948665991,
      "grad_norm": 1.6363626718521118,
      "learning_rate": 1.783393592559713e-05,
      "loss": 1.0126,
      "step": 640
    },
    {
      "epoch": 0.21648091860857818,
      "grad_norm": 1.552738070487976,
      "learning_rate": 1.7827314810896004e-05,
      "loss": 0.9086,
      "step": 641
    },
    {
      "epoch": 0.21681864235055726,
      "grad_norm": 1.576757788658142,
      "learning_rate": 1.7820684825164084e-05,
      "loss": 0.9074,
      "step": 642
    },
    {
      "epoch": 0.2171563660925363,
      "grad_norm": 1.396670937538147,
      "learning_rate": 1.7814045975915423e-05,
      "loss": 1.0055,
      "step": 643
    },
    {
      "epoch": 0.21749408983451538,
      "grad_norm": 1.3884540796279907,
      "learning_rate": 1.7807398270674108e-05,
      "loss": 0.9937,
      "step": 644
    },
    {
      "epoch": 0.21783181357649442,
      "grad_norm": 1.5955413579940796,
      "learning_rate": 1.7800741716974266e-05,
      "loss": 1.0226,
      "step": 645
    },
    {
      "epoch": 0.2181695373184735,
      "grad_norm": 1.562610387802124,
      "learning_rate": 1.7794076322360063e-05,
      "loss": 0.9158,
      "step": 646
    },
    {
      "epoch": 0.21850726106045254,
      "grad_norm": 1.5183926820755005,
      "learning_rate": 1.7787402094385665e-05,
      "loss": 1.0495,
      "step": 647
    },
    {
      "epoch": 0.2188449848024316,
      "grad_norm": 2.2183873653411865,
      "learning_rate": 1.778071904061527e-05,
      "loss": 0.9927,
      "step": 648
    },
    {
      "epoch": 0.21918270854441066,
      "grad_norm": 1.6374105215072632,
      "learning_rate": 1.7774027168623064e-05,
      "loss": 0.8947,
      "step": 649
    },
    {
      "epoch": 0.21952043228638973,
      "grad_norm": 1.4358652830123901,
      "learning_rate": 1.7767326485993227e-05,
      "loss": 0.934,
      "step": 650
    },
    {
      "epoch": 0.2198581560283688,
      "grad_norm": 1.8155231475830078,
      "learning_rate": 1.776061700031994e-05,
      "loss": 0.9797,
      "step": 651
    },
    {
      "epoch": 0.22019587977034785,
      "grad_norm": 1.74410879611969,
      "learning_rate": 1.775389871920734e-05,
      "loss": 0.902,
      "step": 652
    },
    {
      "epoch": 0.22053360351232693,
      "grad_norm": 1.5893549919128418,
      "learning_rate": 1.774717165026955e-05,
      "loss": 0.9979,
      "step": 653
    },
    {
      "epoch": 0.22087132725430597,
      "grad_norm": 1.5434657335281372,
      "learning_rate": 1.7740435801130646e-05,
      "loss": 0.9993,
      "step": 654
    },
    {
      "epoch": 0.22120905099628504,
      "grad_norm": 2.164044141769409,
      "learning_rate": 1.7733691179424646e-05,
      "loss": 0.8445,
      "step": 655
    },
    {
      "epoch": 0.2215467747382641,
      "grad_norm": 1.9465259313583374,
      "learning_rate": 1.7726937792795523e-05,
      "loss": 0.8733,
      "step": 656
    },
    {
      "epoch": 0.22188449848024316,
      "grad_norm": 1.7270108461380005,
      "learning_rate": 1.7720175648897177e-05,
      "loss": 0.9346,
      "step": 657
    },
    {
      "epoch": 0.2222222222222222,
      "grad_norm": 1.7372462749481201,
      "learning_rate": 1.771340475539344e-05,
      "loss": 1.0387,
      "step": 658
    },
    {
      "epoch": 0.22255994596420128,
      "grad_norm": 2.0999910831451416,
      "learning_rate": 1.7706625119958052e-05,
      "loss": 0.8103,
      "step": 659
    },
    {
      "epoch": 0.22289766970618036,
      "grad_norm": 1.7823400497436523,
      "learning_rate": 1.769983675027466e-05,
      "loss": 1.0026,
      "step": 660
    },
    {
      "epoch": 0.2232353934481594,
      "grad_norm": 1.8012356758117676,
      "learning_rate": 1.7693039654036824e-05,
      "loss": 0.9478,
      "step": 661
    },
    {
      "epoch": 0.22357311719013848,
      "grad_norm": 1.9153497219085693,
      "learning_rate": 1.7686233838947974e-05,
      "loss": 0.9943,
      "step": 662
    },
    {
      "epoch": 0.22391084093211752,
      "grad_norm": 1.5731364488601685,
      "learning_rate": 1.7679419312721434e-05,
      "loss": 0.9317,
      "step": 663
    },
    {
      "epoch": 0.2242485646740966,
      "grad_norm": 1.8192622661590576,
      "learning_rate": 1.76725960830804e-05,
      "loss": 0.8187,
      "step": 664
    },
    {
      "epoch": 0.22458628841607564,
      "grad_norm": 1.6074172258377075,
      "learning_rate": 1.766576415775793e-05,
      "loss": 0.966,
      "step": 665
    },
    {
      "epoch": 0.22492401215805471,
      "grad_norm": 1.6496984958648682,
      "learning_rate": 1.7658923544496937e-05,
      "loss": 0.9587,
      "step": 666
    },
    {
      "epoch": 0.22526173590003376,
      "grad_norm": 1.642215371131897,
      "learning_rate": 1.7652074251050183e-05,
      "loss": 1.0225,
      "step": 667
    },
    {
      "epoch": 0.22559945964201283,
      "grad_norm": 1.5923503637313843,
      "learning_rate": 1.764521628518026e-05,
      "loss": 0.9702,
      "step": 668
    },
    {
      "epoch": 0.2259371833839919,
      "grad_norm": 1.6317005157470703,
      "learning_rate": 1.7638349654659596e-05,
      "loss": 0.9831,
      "step": 669
    },
    {
      "epoch": 0.22627490712597095,
      "grad_norm": 1.5705472230911255,
      "learning_rate": 1.763147436727044e-05,
      "loss": 1.0232,
      "step": 670
    },
    {
      "epoch": 0.22661263086795003,
      "grad_norm": 1.9368271827697754,
      "learning_rate": 1.7624590430804843e-05,
      "loss": 0.9847,
      "step": 671
    },
    {
      "epoch": 0.22695035460992907,
      "grad_norm": 1.7606356143951416,
      "learning_rate": 1.7617697853064677e-05,
      "loss": 0.9356,
      "step": 672
    },
    {
      "epoch": 0.22728807835190815,
      "grad_norm": 1.4509592056274414,
      "learning_rate": 1.7610796641861584e-05,
      "loss": 1.0573,
      "step": 673
    },
    {
      "epoch": 0.2276258020938872,
      "grad_norm": 1.76920485496521,
      "learning_rate": 1.7603886805017004e-05,
      "loss": 0.9323,
      "step": 674
    },
    {
      "epoch": 0.22796352583586627,
      "grad_norm": 1.7435137033462524,
      "learning_rate": 1.7596968350362155e-05,
      "loss": 0.9284,
      "step": 675
    },
    {
      "epoch": 0.2283012495778453,
      "grad_norm": 2.0984134674072266,
      "learning_rate": 1.759004128573801e-05,
      "loss": 0.8099,
      "step": 676
    },
    {
      "epoch": 0.22863897331982438,
      "grad_norm": 1.665867805480957,
      "learning_rate": 1.7583105618995318e-05,
      "loss": 0.969,
      "step": 677
    },
    {
      "epoch": 0.22897669706180346,
      "grad_norm": 1.560520887374878,
      "learning_rate": 1.7576161357994558e-05,
      "loss": 0.9738,
      "step": 678
    },
    {
      "epoch": 0.2293144208037825,
      "grad_norm": 1.7215228080749512,
      "learning_rate": 1.7569208510605965e-05,
      "loss": 1.0384,
      "step": 679
    },
    {
      "epoch": 0.22965214454576158,
      "grad_norm": 1.5817105770111084,
      "learning_rate": 1.7562247084709493e-05,
      "loss": 0.9233,
      "step": 680
    },
    {
      "epoch": 0.22998986828774062,
      "grad_norm": 2.0537335872650146,
      "learning_rate": 1.7555277088194827e-05,
      "loss": 0.9444,
      "step": 681
    },
    {
      "epoch": 0.2303275920297197,
      "grad_norm": 1.7047560214996338,
      "learning_rate": 1.754829852896136e-05,
      "loss": 0.9873,
      "step": 682
    },
    {
      "epoch": 0.23066531577169874,
      "grad_norm": 1.8259193897247314,
      "learning_rate": 1.7541311414918192e-05,
      "loss": 0.9948,
      "step": 683
    },
    {
      "epoch": 0.23100303951367782,
      "grad_norm": 1.574973464012146,
      "learning_rate": 1.7534315753984125e-05,
      "loss": 1.0116,
      "step": 684
    },
    {
      "epoch": 0.23134076325565686,
      "grad_norm": 1.7668733596801758,
      "learning_rate": 1.752731155408763e-05,
      "loss": 0.8736,
      "step": 685
    },
    {
      "epoch": 0.23167848699763594,
      "grad_norm": 1.5341562032699585,
      "learning_rate": 1.7520298823166873e-05,
      "loss": 0.9592,
      "step": 686
    },
    {
      "epoch": 0.232016210739615,
      "grad_norm": 1.5290080308914185,
      "learning_rate": 1.751327756916968e-05,
      "loss": 0.9494,
      "step": 687
    },
    {
      "epoch": 0.23235393448159405,
      "grad_norm": 1.6091424226760864,
      "learning_rate": 1.750624780005354e-05,
      "loss": 0.8683,
      "step": 688
    },
    {
      "epoch": 0.23269165822357313,
      "grad_norm": 1.5730867385864258,
      "learning_rate": 1.7499209523785593e-05,
      "loss": 0.9836,
      "step": 689
    },
    {
      "epoch": 0.23302938196555217,
      "grad_norm": 1.9672855138778687,
      "learning_rate": 1.749216274834261e-05,
      "loss": 0.9426,
      "step": 690
    },
    {
      "epoch": 0.23336710570753125,
      "grad_norm": 1.6657418012619019,
      "learning_rate": 1.7485107481711014e-05,
      "loss": 0.9568,
      "step": 691
    },
    {
      "epoch": 0.2337048294495103,
      "grad_norm": 1.8412368297576904,
      "learning_rate": 1.747804373188683e-05,
      "loss": 0.9294,
      "step": 692
    },
    {
      "epoch": 0.23404255319148937,
      "grad_norm": 1.6791536808013916,
      "learning_rate": 1.747097150687572e-05,
      "loss": 0.9927,
      "step": 693
    },
    {
      "epoch": 0.2343802769334684,
      "grad_norm": 1.6063833236694336,
      "learning_rate": 1.746389081469293e-05,
      "loss": 0.9519,
      "step": 694
    },
    {
      "epoch": 0.23471800067544749,
      "grad_norm": 1.8460699319839478,
      "learning_rate": 1.7456801663363317e-05,
      "loss": 0.9173,
      "step": 695
    },
    {
      "epoch": 0.23505572441742653,
      "grad_norm": 1.707335114479065,
      "learning_rate": 1.7449704060921314e-05,
      "loss": 0.9907,
      "step": 696
    },
    {
      "epoch": 0.2353934481594056,
      "grad_norm": 2.0749149322509766,
      "learning_rate": 1.744259801541094e-05,
      "loss": 0.9238,
      "step": 697
    },
    {
      "epoch": 0.23573117190138468,
      "grad_norm": 1.805444598197937,
      "learning_rate": 1.743548353488578e-05,
      "loss": 0.964,
      "step": 698
    },
    {
      "epoch": 0.23606889564336372,
      "grad_norm": 1.5846120119094849,
      "learning_rate": 1.742836062740898e-05,
      "loss": 0.9849,
      "step": 699
    },
    {
      "epoch": 0.2364066193853428,
      "grad_norm": 1.885827660560608,
      "learning_rate": 1.742122930105324e-05,
      "loss": 0.6704,
      "step": 700
    },
    {
      "epoch": 0.23674434312732184,
      "grad_norm": 1.6241611242294312,
      "learning_rate": 1.741408956390079e-05,
      "loss": 1.0637,
      "step": 701
    },
    {
      "epoch": 0.23708206686930092,
      "grad_norm": 1.6633431911468506,
      "learning_rate": 1.7406941424043402e-05,
      "loss": 0.9597,
      "step": 702
    },
    {
      "epoch": 0.23741979061127996,
      "grad_norm": 1.8707470893859863,
      "learning_rate": 1.739978488958237e-05,
      "loss": 0.9871,
      "step": 703
    },
    {
      "epoch": 0.23775751435325904,
      "grad_norm": 1.804038405418396,
      "learning_rate": 1.73926199686285e-05,
      "loss": 0.9628,
      "step": 704
    },
    {
      "epoch": 0.23809523809523808,
      "grad_norm": 1.8133996725082397,
      "learning_rate": 1.7385446669302105e-05,
      "loss": 0.9877,
      "step": 705
    },
    {
      "epoch": 0.23843296183721716,
      "grad_norm": 1.507765293121338,
      "learning_rate": 1.737826499973299e-05,
      "loss": 0.968,
      "step": 706
    },
    {
      "epoch": 0.23877068557919623,
      "grad_norm": 1.70377516746521,
      "learning_rate": 1.737107496806045e-05,
      "loss": 0.9479,
      "step": 707
    },
    {
      "epoch": 0.23910840932117527,
      "grad_norm": 1.6430121660232544,
      "learning_rate": 1.7363876582433254e-05,
      "loss": 0.9772,
      "step": 708
    },
    {
      "epoch": 0.23944613306315435,
      "grad_norm": 1.5794414281845093,
      "learning_rate": 1.7356669851009636e-05,
      "loss": 0.9557,
      "step": 709
    },
    {
      "epoch": 0.2397838568051334,
      "grad_norm": 1.4723002910614014,
      "learning_rate": 1.7349454781957305e-05,
      "loss": 1.0041,
      "step": 710
    },
    {
      "epoch": 0.24012158054711247,
      "grad_norm": 1.907857060432434,
      "learning_rate": 1.7342231383453397e-05,
      "loss": 0.9612,
      "step": 711
    },
    {
      "epoch": 0.2404593042890915,
      "grad_norm": 2.029594898223877,
      "learning_rate": 1.7334999663684504e-05,
      "loss": 0.7922,
      "step": 712
    },
    {
      "epoch": 0.2407970280310706,
      "grad_norm": 1.6132161617279053,
      "learning_rate": 1.7327759630846644e-05,
      "loss": 0.8914,
      "step": 713
    },
    {
      "epoch": 0.24113475177304963,
      "grad_norm": 1.784063696861267,
      "learning_rate": 1.7320511293145246e-05,
      "loss": 0.8831,
      "step": 714
    },
    {
      "epoch": 0.2414724755150287,
      "grad_norm": 1.6230236291885376,
      "learning_rate": 1.731325465879518e-05,
      "loss": 0.8684,
      "step": 715
    },
    {
      "epoch": 0.24181019925700778,
      "grad_norm": 1.8663990497589111,
      "learning_rate": 1.730598973602068e-05,
      "loss": 0.9823,
      "step": 716
    },
    {
      "epoch": 0.24214792299898683,
      "grad_norm": 1.6423972845077515,
      "learning_rate": 1.7298716533055406e-05,
      "loss": 0.9929,
      "step": 717
    },
    {
      "epoch": 0.2424856467409659,
      "grad_norm": 1.774942398071289,
      "learning_rate": 1.729143505814239e-05,
      "loss": 0.9258,
      "step": 718
    },
    {
      "epoch": 0.24282337048294494,
      "grad_norm": 1.6613595485687256,
      "learning_rate": 1.728414531953403e-05,
      "loss": 0.9222,
      "step": 719
    },
    {
      "epoch": 0.24316109422492402,
      "grad_norm": 1.6560906171798706,
      "learning_rate": 1.727684732549211e-05,
      "loss": 0.9106,
      "step": 720
    },
    {
      "epoch": 0.24349881796690306,
      "grad_norm": 1.667242169380188,
      "learning_rate": 1.726954108428775e-05,
      "loss": 1.0285,
      "step": 721
    },
    {
      "epoch": 0.24383654170888214,
      "grad_norm": 1.9235655069351196,
      "learning_rate": 1.7262226604201432e-05,
      "loss": 0.9394,
      "step": 722
    },
    {
      "epoch": 0.24417426545086118,
      "grad_norm": 1.7309292554855347,
      "learning_rate": 1.725490389352297e-05,
      "loss": 1.0066,
      "step": 723
    },
    {
      "epoch": 0.24451198919284026,
      "grad_norm": 1.6281174421310425,
      "learning_rate": 1.7247572960551497e-05,
      "loss": 1.0058,
      "step": 724
    },
    {
      "epoch": 0.24484971293481933,
      "grad_norm": 1.706660509109497,
      "learning_rate": 1.724023381359548e-05,
      "loss": 1.0691,
      "step": 725
    },
    {
      "epoch": 0.24518743667679838,
      "grad_norm": 1.788631796836853,
      "learning_rate": 1.7232886460972685e-05,
      "loss": 0.963,
      "step": 726
    },
    {
      "epoch": 0.24552516041877745,
      "grad_norm": 1.899524450302124,
      "learning_rate": 1.7225530911010184e-05,
      "loss": 0.9423,
      "step": 727
    },
    {
      "epoch": 0.2458628841607565,
      "grad_norm": 1.776076078414917,
      "learning_rate": 1.721816717204433e-05,
      "loss": 1.0046,
      "step": 728
    },
    {
      "epoch": 0.24620060790273557,
      "grad_norm": 1.9414191246032715,
      "learning_rate": 1.7210795252420776e-05,
      "loss": 0.8755,
      "step": 729
    },
    {
      "epoch": 0.24653833164471461,
      "grad_norm": 1.771082878112793,
      "learning_rate": 1.7203415160494427e-05,
      "loss": 0.9126,
      "step": 730
    },
    {
      "epoch": 0.2468760553866937,
      "grad_norm": 1.6341469287872314,
      "learning_rate": 1.7196026904629453e-05,
      "loss": 1.0091,
      "step": 731
    },
    {
      "epoch": 0.24721377912867273,
      "grad_norm": 1.9980156421661377,
      "learning_rate": 1.7188630493199284e-05,
      "loss": 0.8063,
      "step": 732
    },
    {
      "epoch": 0.2475515028706518,
      "grad_norm": 1.6002767086029053,
      "learning_rate": 1.7181225934586588e-05,
      "loss": 0.984,
      "step": 733
    },
    {
      "epoch": 0.24788922661263088,
      "grad_norm": 1.8467609882354736,
      "learning_rate": 1.717381323718327e-05,
      "loss": 0.9571,
      "step": 734
    },
    {
      "epoch": 0.24822695035460993,
      "grad_norm": 1.7542591094970703,
      "learning_rate": 1.716639240939046e-05,
      "loss": 0.8965,
      "step": 735
    },
    {
      "epoch": 0.248564674096589,
      "grad_norm": 1.509108543395996,
      "learning_rate": 1.7158963459618493e-05,
      "loss": 0.9262,
      "step": 736
    },
    {
      "epoch": 0.24890239783856805,
      "grad_norm": 1.6256673336029053,
      "learning_rate": 1.715152639628692e-05,
      "loss": 0.9345,
      "step": 737
    },
    {
      "epoch": 0.24924012158054712,
      "grad_norm": 1.6046345233917236,
      "learning_rate": 1.7144081227824482e-05,
      "loss": 0.9861,
      "step": 738
    },
    {
      "epoch": 0.24957784532252617,
      "grad_norm": 1.6616566181182861,
      "learning_rate": 1.7136627962669104e-05,
      "loss": 1.0075,
      "step": 739
    },
    {
      "epoch": 0.24991556906450524,
      "grad_norm": 1.6407862901687622,
      "learning_rate": 1.71291666092679e-05,
      "loss": 0.9475,
      "step": 740
    },
    {
      "epoch": 0.2502532928064843,
      "grad_norm": 1.9198567867279053,
      "learning_rate": 1.712169717607713e-05,
      "loss": 0.9863,
      "step": 741
    },
    {
      "epoch": 0.2502532928064843,
      "eval_loss": 0.8352074027061462,
      "eval_runtime": 349.1012,
      "eval_samples_per_second": 8.513,
      "eval_steps_per_second": 1.066,
      "step": 741
    },
    {
      "epoch": 0.25059101654846333,
      "grad_norm": 1.6872812509536743,
      "learning_rate": 1.711421967156223e-05,
      "loss": 0.9607,
      "step": 742
    },
    {
      "epoch": 0.25092874029044243,
      "grad_norm": 1.7971426248550415,
      "learning_rate": 1.7106734104197768e-05,
      "loss": 0.9321,
      "step": 743
    },
    {
      "epoch": 0.2512664640324215,
      "grad_norm": 2.0841782093048096,
      "learning_rate": 1.7099240482467463e-05,
      "loss": 0.9322,
      "step": 744
    },
    {
      "epoch": 0.2516041877744005,
      "grad_norm": 1.6395483016967773,
      "learning_rate": 1.709173881486416e-05,
      "loss": 0.9661,
      "step": 745
    },
    {
      "epoch": 0.2519419115163796,
      "grad_norm": 1.7422521114349365,
      "learning_rate": 1.7084229109889813e-05,
      "loss": 0.9374,
      "step": 746
    },
    {
      "epoch": 0.25227963525835867,
      "grad_norm": 1.689939260482788,
      "learning_rate": 1.70767113760555e-05,
      "loss": 0.9243,
      "step": 747
    },
    {
      "epoch": 0.2526173590003377,
      "grad_norm": 1.5937788486480713,
      "learning_rate": 1.706918562188138e-05,
      "loss": 0.9573,
      "step": 748
    },
    {
      "epoch": 0.25295508274231676,
      "grad_norm": 1.6884655952453613,
      "learning_rate": 1.706165185589672e-05,
      "loss": 0.9622,
      "step": 749
    },
    {
      "epoch": 0.25329280648429586,
      "grad_norm": 1.6605758666992188,
      "learning_rate": 1.705411008663986e-05,
      "loss": 0.961,
      "step": 750
    },
    {
      "epoch": 0.2536305302262749,
      "grad_norm": 1.8007186651229858,
      "learning_rate": 1.70465603226582e-05,
      "loss": 0.9445,
      "step": 751
    },
    {
      "epoch": 0.25396825396825395,
      "grad_norm": 1.8559463024139404,
      "learning_rate": 1.7039002572508227e-05,
      "loss": 1.0253,
      "step": 752
    },
    {
      "epoch": 0.25430597771023306,
      "grad_norm": 1.9192028045654297,
      "learning_rate": 1.7031436844755448e-05,
      "loss": 0.9332,
      "step": 753
    },
    {
      "epoch": 0.2546437014522121,
      "grad_norm": 1.7730717658996582,
      "learning_rate": 1.702386314797443e-05,
      "loss": 0.9052,
      "step": 754
    },
    {
      "epoch": 0.25498142519419115,
      "grad_norm": 1.8629709482192993,
      "learning_rate": 1.7016281490748768e-05,
      "loss": 0.9488,
      "step": 755
    },
    {
      "epoch": 0.2553191489361702,
      "grad_norm": 1.6746963262557983,
      "learning_rate": 1.7008691881671083e-05,
      "loss": 0.9028,
      "step": 756
    },
    {
      "epoch": 0.2556568726781493,
      "grad_norm": 1.6942139863967896,
      "learning_rate": 1.7001094329343e-05,
      "loss": 0.8795,
      "step": 757
    },
    {
      "epoch": 0.25599459642012834,
      "grad_norm": 1.6107038259506226,
      "learning_rate": 1.699348884237515e-05,
      "loss": 0.9777,
      "step": 758
    },
    {
      "epoch": 0.2563323201621074,
      "grad_norm": 1.8247896432876587,
      "learning_rate": 1.6985875429387155e-05,
      "loss": 0.9964,
      "step": 759
    },
    {
      "epoch": 0.25667004390408643,
      "grad_norm": 1.8404369354248047,
      "learning_rate": 1.6978254099007625e-05,
      "loss": 0.9397,
      "step": 760
    },
    {
      "epoch": 0.25700776764606553,
      "grad_norm": 2.00410795211792,
      "learning_rate": 1.6970624859874138e-05,
      "loss": 1.009,
      "step": 761
    },
    {
      "epoch": 0.2573454913880446,
      "grad_norm": 1.5542668104171753,
      "learning_rate": 1.696298772063324e-05,
      "loss": 0.9608,
      "step": 762
    },
    {
      "epoch": 0.2576832151300236,
      "grad_norm": 1.582123041152954,
      "learning_rate": 1.6955342689940423e-05,
      "loss": 0.9808,
      "step": 763
    },
    {
      "epoch": 0.2580209388720027,
      "grad_norm": 2.1611881256103516,
      "learning_rate": 1.694768977646013e-05,
      "loss": 0.9963,
      "step": 764
    },
    {
      "epoch": 0.25835866261398177,
      "grad_norm": 1.8269307613372803,
      "learning_rate": 1.6940028988865735e-05,
      "loss": 0.9306,
      "step": 765
    },
    {
      "epoch": 0.2586963863559608,
      "grad_norm": 1.9661474227905273,
      "learning_rate": 1.693236033583954e-05,
      "loss": 0.7756,
      "step": 766
    },
    {
      "epoch": 0.25903411009793986,
      "grad_norm": 1.7735875844955444,
      "learning_rate": 1.6924683826072752e-05,
      "loss": 0.9278,
      "step": 767
    },
    {
      "epoch": 0.25937183383991896,
      "grad_norm": 2.0303139686584473,
      "learning_rate": 1.6916999468265487e-05,
      "loss": 0.8306,
      "step": 768
    },
    {
      "epoch": 0.259709557581898,
      "grad_norm": 1.8015457391738892,
      "learning_rate": 1.6909307271126763e-05,
      "loss": 0.9343,
      "step": 769
    },
    {
      "epoch": 0.26004728132387706,
      "grad_norm": 1.573121428489685,
      "learning_rate": 1.690160724337447e-05,
      "loss": 0.9203,
      "step": 770
    },
    {
      "epoch": 0.26038500506585616,
      "grad_norm": 1.7579994201660156,
      "learning_rate": 1.6893899393735383e-05,
      "loss": 0.9518,
      "step": 771
    },
    {
      "epoch": 0.2607227288078352,
      "grad_norm": 1.6328119039535522,
      "learning_rate": 1.6886183730945138e-05,
      "loss": 0.9258,
      "step": 772
    },
    {
      "epoch": 0.26106045254981425,
      "grad_norm": 1.7829252481460571,
      "learning_rate": 1.6878460263748223e-05,
      "loss": 0.9418,
      "step": 773
    },
    {
      "epoch": 0.2613981762917933,
      "grad_norm": 1.7189514636993408,
      "learning_rate": 1.6870729000897977e-05,
      "loss": 0.9143,
      "step": 774
    },
    {
      "epoch": 0.2617359000337724,
      "grad_norm": 1.6750832796096802,
      "learning_rate": 1.6862989951156567e-05,
      "loss": 1.0029,
      "step": 775
    },
    {
      "epoch": 0.26207362377575144,
      "grad_norm": 1.4781761169433594,
      "learning_rate": 1.6855243123294992e-05,
      "loss": 1.0103,
      "step": 776
    },
    {
      "epoch": 0.2624113475177305,
      "grad_norm": 1.6477673053741455,
      "learning_rate": 1.684748852609306e-05,
      "loss": 0.9866,
      "step": 777
    },
    {
      "epoch": 0.26274907125970953,
      "grad_norm": 1.7204375267028809,
      "learning_rate": 1.6839726168339394e-05,
      "loss": 0.978,
      "step": 778
    },
    {
      "epoch": 0.26308679500168863,
      "grad_norm": 1.5131142139434814,
      "learning_rate": 1.6831956058831402e-05,
      "loss": 0.9431,
      "step": 779
    },
    {
      "epoch": 0.2634245187436677,
      "grad_norm": 1.6253634691238403,
      "learning_rate": 1.6824178206375282e-05,
      "loss": 0.9199,
      "step": 780
    },
    {
      "epoch": 0.2637622424856467,
      "grad_norm": 1.688163161277771,
      "learning_rate": 1.6816392619786007e-05,
      "loss": 0.978,
      "step": 781
    },
    {
      "epoch": 0.2640999662276258,
      "grad_norm": 1.6951016187667847,
      "learning_rate": 1.6808599307887316e-05,
      "loss": 0.9697,
      "step": 782
    },
    {
      "epoch": 0.26443768996960487,
      "grad_norm": 2.4601340293884277,
      "learning_rate": 1.6800798279511702e-05,
      "loss": 0.8792,
      "step": 783
    },
    {
      "epoch": 0.2647754137115839,
      "grad_norm": 1.6054311990737915,
      "learning_rate": 1.6792989543500405e-05,
      "loss": 0.9455,
      "step": 784
    },
    {
      "epoch": 0.26511313745356296,
      "grad_norm": 1.5519051551818848,
      "learning_rate": 1.67851731087034e-05,
      "loss": 0.9889,
      "step": 785
    },
    {
      "epoch": 0.26545086119554206,
      "grad_norm": 1.8028554916381836,
      "learning_rate": 1.6777348983979385e-05,
      "loss": 0.9082,
      "step": 786
    },
    {
      "epoch": 0.2657885849375211,
      "grad_norm": 1.8364535570144653,
      "learning_rate": 1.676951717819578e-05,
      "loss": 0.7947,
      "step": 787
    },
    {
      "epoch": 0.26612630867950016,
      "grad_norm": 1.7280329465866089,
      "learning_rate": 1.67616777002287e-05,
      "loss": 0.9632,
      "step": 788
    },
    {
      "epoch": 0.26646403242147926,
      "grad_norm": 1.709215521812439,
      "learning_rate": 1.6753830558962962e-05,
      "loss": 1.0047,
      "step": 789
    },
    {
      "epoch": 0.2668017561634583,
      "grad_norm": 1.7154841423034668,
      "learning_rate": 1.6745975763292072e-05,
      "loss": 0.9449,
      "step": 790
    },
    {
      "epoch": 0.26713947990543735,
      "grad_norm": 1.714390516281128,
      "learning_rate": 1.6738113322118197e-05,
      "loss": 1.0677,
      "step": 791
    },
    {
      "epoch": 0.2674772036474164,
      "grad_norm": 1.6912754774093628,
      "learning_rate": 1.6730243244352185e-05,
      "loss": 1.029,
      "step": 792
    },
    {
      "epoch": 0.2678149273893955,
      "grad_norm": 1.7450389862060547,
      "learning_rate": 1.672236553891353e-05,
      "loss": 0.9683,
      "step": 793
    },
    {
      "epoch": 0.26815265113137454,
      "grad_norm": 1.5921680927276611,
      "learning_rate": 1.671448021473038e-05,
      "loss": 0.9965,
      "step": 794
    },
    {
      "epoch": 0.2684903748733536,
      "grad_norm": 1.7931723594665527,
      "learning_rate": 1.6706587280739494e-05,
      "loss": 0.9823,
      "step": 795
    },
    {
      "epoch": 0.26882809861533263,
      "grad_norm": 2.034852981567383,
      "learning_rate": 1.6698686745886285e-05,
      "loss": 0.9124,
      "step": 796
    },
    {
      "epoch": 0.26916582235731173,
      "grad_norm": 1.8566904067993164,
      "learning_rate": 1.6690778619124763e-05,
      "loss": 0.8774,
      "step": 797
    },
    {
      "epoch": 0.2695035460992908,
      "grad_norm": 1.6134916543960571,
      "learning_rate": 1.6682862909417542e-05,
      "loss": 0.6526,
      "step": 798
    },
    {
      "epoch": 0.2698412698412698,
      "grad_norm": 1.5902119874954224,
      "learning_rate": 1.6674939625735844e-05,
      "loss": 0.9846,
      "step": 799
    },
    {
      "epoch": 0.2701789935832489,
      "grad_norm": 1.6434234380722046,
      "learning_rate": 1.666700877705946e-05,
      "loss": 1.0368,
      "step": 800
    },
    {
      "epoch": 0.270516717325228,
      "grad_norm": 1.5020389556884766,
      "learning_rate": 1.665907037237676e-05,
      "loss": 0.9552,
      "step": 801
    },
    {
      "epoch": 0.270854441067207,
      "grad_norm": 2.0392677783966064,
      "learning_rate": 1.6651124420684676e-05,
      "loss": 0.9449,
      "step": 802
    },
    {
      "epoch": 0.27119216480918606,
      "grad_norm": 1.9003113508224487,
      "learning_rate": 1.66431709309887e-05,
      "loss": 0.9543,
      "step": 803
    },
    {
      "epoch": 0.27152988855116517,
      "grad_norm": 1.9123886823654175,
      "learning_rate": 1.6635209912302857e-05,
      "loss": 1.0118,
      "step": 804
    },
    {
      "epoch": 0.2718676122931442,
      "grad_norm": 1.569347620010376,
      "learning_rate": 1.662724137364971e-05,
      "loss": 0.95,
      "step": 805
    },
    {
      "epoch": 0.27220533603512326,
      "grad_norm": 1.7663401365280151,
      "learning_rate": 1.661926532406035e-05,
      "loss": 0.9204,
      "step": 806
    },
    {
      "epoch": 0.2725430597771023,
      "grad_norm": 1.8114197254180908,
      "learning_rate": 1.6611281772574376e-05,
      "loss": 0.8783,
      "step": 807
    },
    {
      "epoch": 0.2728807835190814,
      "grad_norm": 1.78357994556427,
      "learning_rate": 1.6603290728239883e-05,
      "loss": 0.9397,
      "step": 808
    },
    {
      "epoch": 0.27321850726106045,
      "grad_norm": 1.9610450267791748,
      "learning_rate": 1.6595292200113467e-05,
      "loss": 0.9004,
      "step": 809
    },
    {
      "epoch": 0.2735562310030395,
      "grad_norm": 1.6399847269058228,
      "learning_rate": 1.65872861972602e-05,
      "loss": 0.9668,
      "step": 810
    },
    {
      "epoch": 0.2738939547450186,
      "grad_norm": 1.9651689529418945,
      "learning_rate": 1.6579272728753628e-05,
      "loss": 0.8243,
      "step": 811
    },
    {
      "epoch": 0.27423167848699764,
      "grad_norm": 1.8175697326660156,
      "learning_rate": 1.6571251803675758e-05,
      "loss": 0.9832,
      "step": 812
    },
    {
      "epoch": 0.2745694022289767,
      "grad_norm": 1.8699359893798828,
      "learning_rate": 1.6563223431117046e-05,
      "loss": 0.9434,
      "step": 813
    },
    {
      "epoch": 0.27490712597095573,
      "grad_norm": 1.9184753894805908,
      "learning_rate": 1.6555187620176394e-05,
      "loss": 0.9394,
      "step": 814
    },
    {
      "epoch": 0.27524484971293484,
      "grad_norm": 2.2123491764068604,
      "learning_rate": 1.6547144379961128e-05,
      "loss": 0.8595,
      "step": 815
    },
    {
      "epoch": 0.2755825734549139,
      "grad_norm": 1.6461082696914673,
      "learning_rate": 1.6539093719586998e-05,
      "loss": 1.0135,
      "step": 816
    },
    {
      "epoch": 0.2759202971968929,
      "grad_norm": 1.6762102842330933,
      "learning_rate": 1.6531035648178157e-05,
      "loss": 0.8971,
      "step": 817
    },
    {
      "epoch": 0.27625802093887203,
      "grad_norm": 1.736091136932373,
      "learning_rate": 1.6522970174867166e-05,
      "loss": 0.9163,
      "step": 818
    },
    {
      "epoch": 0.2765957446808511,
      "grad_norm": 1.8698163032531738,
      "learning_rate": 1.6514897308794973e-05,
      "loss": 0.9734,
      "step": 819
    },
    {
      "epoch": 0.2769334684228301,
      "grad_norm": 1.8918684720993042,
      "learning_rate": 1.6506817059110898e-05,
      "loss": 0.8814,
      "step": 820
    },
    {
      "epoch": 0.27727119216480917,
      "grad_norm": 1.5647212266921997,
      "learning_rate": 1.6498729434972636e-05,
      "loss": 0.9906,
      "step": 821
    },
    {
      "epoch": 0.27760891590678827,
      "grad_norm": 1.78870689868927,
      "learning_rate": 1.649063444554624e-05,
      "loss": 0.9707,
      "step": 822
    },
    {
      "epoch": 0.2779466396487673,
      "grad_norm": 1.8421725034713745,
      "learning_rate": 1.64825321000061e-05,
      "loss": 0.8972,
      "step": 823
    },
    {
      "epoch": 0.27828436339074636,
      "grad_norm": 1.6814625263214111,
      "learning_rate": 1.6474422407534964e-05,
      "loss": 0.9706,
      "step": 824
    },
    {
      "epoch": 0.2786220871327254,
      "grad_norm": 1.698258876800537,
      "learning_rate": 1.6466305377323876e-05,
      "loss": 0.988,
      "step": 825
    },
    {
      "epoch": 0.2789598108747045,
      "grad_norm": 1.971911907196045,
      "learning_rate": 1.645818101857223e-05,
      "loss": 0.9131,
      "step": 826
    },
    {
      "epoch": 0.27929753461668355,
      "grad_norm": 1.9593485593795776,
      "learning_rate": 1.64500493404877e-05,
      "loss": 1.0166,
      "step": 827
    },
    {
      "epoch": 0.2796352583586626,
      "grad_norm": 1.7925101518630981,
      "learning_rate": 1.6441910352286264e-05,
      "loss": 0.8551,
      "step": 828
    },
    {
      "epoch": 0.2799729821006417,
      "grad_norm": 1.650335431098938,
      "learning_rate": 1.6433764063192195e-05,
      "loss": 0.9382,
      "step": 829
    },
    {
      "epoch": 0.28031070584262074,
      "grad_norm": 1.6871994733810425,
      "learning_rate": 1.642561048243802e-05,
      "loss": 0.9557,
      "step": 830
    },
    {
      "epoch": 0.2806484295845998,
      "grad_norm": 1.9687548875808716,
      "learning_rate": 1.641744961926455e-05,
      "loss": 0.8738,
      "step": 831
    },
    {
      "epoch": 0.28098615332657884,
      "grad_norm": 1.5919392108917236,
      "learning_rate": 1.6409281482920833e-05,
      "loss": 0.8967,
      "step": 832
    },
    {
      "epoch": 0.28132387706855794,
      "grad_norm": 1.7013665437698364,
      "learning_rate": 1.6401106082664172e-05,
      "loss": 0.9284,
      "step": 833
    },
    {
      "epoch": 0.281661600810537,
      "grad_norm": 1.676510214805603,
      "learning_rate": 1.63929234277601e-05,
      "loss": 0.9245,
      "step": 834
    },
    {
      "epoch": 0.28199932455251603,
      "grad_norm": 1.6504560708999634,
      "learning_rate": 1.638473352748236e-05,
      "loss": 1.0036,
      "step": 835
    },
    {
      "epoch": 0.28233704829449513,
      "grad_norm": 1.688187599182129,
      "learning_rate": 1.6376536391112927e-05,
      "loss": 1.0176,
      "step": 836
    },
    {
      "epoch": 0.2826747720364742,
      "grad_norm": 1.8376169204711914,
      "learning_rate": 1.636833202794196e-05,
      "loss": 1.0009,
      "step": 837
    },
    {
      "epoch": 0.2830124957784532,
      "grad_norm": 1.8317110538482666,
      "learning_rate": 1.636012044726782e-05,
      "loss": 0.9017,
      "step": 838
    },
    {
      "epoch": 0.28335021952043227,
      "grad_norm": 1.8462111949920654,
      "learning_rate": 1.6351901658397037e-05,
      "loss": 0.9247,
      "step": 839
    },
    {
      "epoch": 0.28368794326241137,
      "grad_norm": 1.8604130744934082,
      "learning_rate": 1.6343675670644322e-05,
      "loss": 0.9548,
      "step": 840
    },
    {
      "epoch": 0.2840256670043904,
      "grad_norm": 1.6301411390304565,
      "learning_rate": 1.6335442493332535e-05,
      "loss": 0.9888,
      "step": 841
    },
    {
      "epoch": 0.28436339074636946,
      "grad_norm": 1.7351330518722534,
      "learning_rate": 1.6327202135792687e-05,
      "loss": 0.9953,
      "step": 842
    },
    {
      "epoch": 0.2847011144883485,
      "grad_norm": 1.7734582424163818,
      "learning_rate": 1.631895460736393e-05,
      "loss": 0.9825,
      "step": 843
    },
    {
      "epoch": 0.2850388382303276,
      "grad_norm": 1.6907756328582764,
      "learning_rate": 1.6310699917393538e-05,
      "loss": 0.9133,
      "step": 844
    },
    {
      "epoch": 0.28537656197230665,
      "grad_norm": 1.685553789138794,
      "learning_rate": 1.630243807523691e-05,
      "loss": 0.9258,
      "step": 845
    },
    {
      "epoch": 0.2857142857142857,
      "grad_norm": 1.7675367593765259,
      "learning_rate": 1.6294169090257543e-05,
      "loss": 0.9048,
      "step": 846
    },
    {
      "epoch": 0.2860520094562648,
      "grad_norm": 1.79044508934021,
      "learning_rate": 1.6285892971827034e-05,
      "loss": 1.0181,
      "step": 847
    },
    {
      "epoch": 0.28638973319824385,
      "grad_norm": 1.8971859216690063,
      "learning_rate": 1.6277609729325054e-05,
      "loss": 0.8825,
      "step": 848
    },
    {
      "epoch": 0.2867274569402229,
      "grad_norm": 1.6097294092178345,
      "learning_rate": 1.6269319372139357e-05,
      "loss": 0.9809,
      "step": 849
    },
    {
      "epoch": 0.28706518068220194,
      "grad_norm": 1.6697542667388916,
      "learning_rate": 1.6261021909665772e-05,
      "loss": 0.961,
      "step": 850
    },
    {
      "epoch": 0.28740290442418104,
      "grad_norm": 1.4712474346160889,
      "learning_rate": 1.6252717351308156e-05,
      "loss": 0.8794,
      "step": 851
    },
    {
      "epoch": 0.2877406281661601,
      "grad_norm": 1.6370964050292969,
      "learning_rate": 1.6244405706478425e-05,
      "loss": 0.9287,
      "step": 852
    },
    {
      "epoch": 0.28807835190813913,
      "grad_norm": 1.8078553676605225,
      "learning_rate": 1.6236086984596516e-05,
      "loss": 0.9771,
      "step": 853
    },
    {
      "epoch": 0.28841607565011823,
      "grad_norm": 1.696596384048462,
      "learning_rate": 1.6227761195090404e-05,
      "loss": 0.9091,
      "step": 854
    },
    {
      "epoch": 0.2887537993920973,
      "grad_norm": 1.6049069166183472,
      "learning_rate": 1.6219428347396055e-05,
      "loss": 1.0198,
      "step": 855
    },
    {
      "epoch": 0.2890915231340763,
      "grad_norm": 2.1233584880828857,
      "learning_rate": 1.6211088450957436e-05,
      "loss": 0.8786,
      "step": 856
    },
    {
      "epoch": 0.28942924687605537,
      "grad_norm": 1.9633384943008423,
      "learning_rate": 1.6202741515226522e-05,
      "loss": 0.9074,
      "step": 857
    },
    {
      "epoch": 0.28976697061803447,
      "grad_norm": 1.7439830303192139,
      "learning_rate": 1.619438754966324e-05,
      "loss": 0.9872,
      "step": 858
    },
    {
      "epoch": 0.2901046943600135,
      "grad_norm": 1.705034613609314,
      "learning_rate": 1.618602656373551e-05,
      "loss": 0.8997,
      "step": 859
    },
    {
      "epoch": 0.29044241810199256,
      "grad_norm": 1.8590495586395264,
      "learning_rate": 1.617765856691918e-05,
      "loss": 1.003,
      "step": 860
    },
    {
      "epoch": 0.2907801418439716,
      "grad_norm": 1.7539708614349365,
      "learning_rate": 1.6169283568698068e-05,
      "loss": 1.0508,
      "step": 861
    },
    {
      "epoch": 0.2911178655859507,
      "grad_norm": 1.7079880237579346,
      "learning_rate": 1.6160901578563916e-05,
      "loss": 0.9757,
      "step": 862
    },
    {
      "epoch": 0.29145558932792975,
      "grad_norm": 1.814475655555725,
      "learning_rate": 1.6152512606016394e-05,
      "loss": 0.9278,
      "step": 863
    },
    {
      "epoch": 0.2917933130699088,
      "grad_norm": 1.9058188199996948,
      "learning_rate": 1.6144116660563076e-05,
      "loss": 0.913,
      "step": 864
    },
    {
      "epoch": 0.2921310368118879,
      "grad_norm": 1.7010709047317505,
      "learning_rate": 1.6135713751719454e-05,
      "loss": 0.9871,
      "step": 865
    },
    {
      "epoch": 0.29246876055386695,
      "grad_norm": 1.8642061948776245,
      "learning_rate": 1.6127303889008907e-05,
      "loss": 0.9579,
      "step": 866
    },
    {
      "epoch": 0.292806484295846,
      "grad_norm": 1.854447603225708,
      "learning_rate": 1.6118887081962683e-05,
      "loss": 1.0098,
      "step": 867
    },
    {
      "epoch": 0.29314420803782504,
      "grad_norm": 2.043466567993164,
      "learning_rate": 1.6110463340119917e-05,
      "loss": 0.8941,
      "step": 868
    },
    {
      "epoch": 0.29348193177980414,
      "grad_norm": 1.94557785987854,
      "learning_rate": 1.6102032673027586e-05,
      "loss": 0.9711,
      "step": 869
    },
    {
      "epoch": 0.2938196555217832,
      "grad_norm": 1.6541953086853027,
      "learning_rate": 1.6093595090240536e-05,
      "loss": 0.9605,
      "step": 870
    },
    {
      "epoch": 0.29415737926376223,
      "grad_norm": 1.739141583442688,
      "learning_rate": 1.608515060132144e-05,
      "loss": 0.9811,
      "step": 871
    },
    {
      "epoch": 0.2944951030057413,
      "grad_norm": 1.6079906225204468,
      "learning_rate": 1.607669921584079e-05,
      "loss": 0.9147,
      "step": 872
    },
    {
      "epoch": 0.2948328267477204,
      "grad_norm": 1.8069438934326172,
      "learning_rate": 1.606824094337691e-05,
      "loss": 0.9051,
      "step": 873
    },
    {
      "epoch": 0.2951705504896994,
      "grad_norm": 1.538088321685791,
      "learning_rate": 1.6059775793515925e-05,
      "loss": 0.9096,
      "step": 874
    },
    {
      "epoch": 0.29550827423167847,
      "grad_norm": 1.688720941543579,
      "learning_rate": 1.6051303775851742e-05,
      "loss": 1.0802,
      "step": 875
    },
    {
      "epoch": 0.29584599797365757,
      "grad_norm": 1.7103081941604614,
      "learning_rate": 1.6042824899986066e-05,
      "loss": 1.0012,
      "step": 876
    },
    {
      "epoch": 0.2961837217156366,
      "grad_norm": 1.7190202474594116,
      "learning_rate": 1.6034339175528374e-05,
      "loss": 0.9019,
      "step": 877
    },
    {
      "epoch": 0.29652144545761566,
      "grad_norm": 2.028632402420044,
      "learning_rate": 1.6025846612095895e-05,
      "loss": 0.819,
      "step": 878
    },
    {
      "epoch": 0.2968591691995947,
      "grad_norm": 2.1882991790771484,
      "learning_rate": 1.6017347219313613e-05,
      "loss": 0.7723,
      "step": 879
    },
    {
      "epoch": 0.2971968929415738,
      "grad_norm": 1.793570637702942,
      "learning_rate": 1.6008841006814265e-05,
      "loss": 0.9971,
      "step": 880
    },
    {
      "epoch": 0.29753461668355285,
      "grad_norm": 1.8553962707519531,
      "learning_rate": 1.6000327984238292e-05,
      "loss": 0.9674,
      "step": 881
    },
    {
      "epoch": 0.2978723404255319,
      "grad_norm": 1.6118420362472534,
      "learning_rate": 1.5991808161233874e-05,
      "loss": 0.8691,
      "step": 882
    },
    {
      "epoch": 0.298210064167511,
      "grad_norm": 1.716882586479187,
      "learning_rate": 1.5983281547456884e-05,
      "loss": 0.9546,
      "step": 883
    },
    {
      "epoch": 0.29854778790949005,
      "grad_norm": 1.7618470191955566,
      "learning_rate": 1.5974748152570908e-05,
      "loss": 1.0572,
      "step": 884
    },
    {
      "epoch": 0.2988855116514691,
      "grad_norm": 1.984457015991211,
      "learning_rate": 1.59662079862472e-05,
      "loss": 0.9006,
      "step": 885
    },
    {
      "epoch": 0.29922323539344814,
      "grad_norm": 2.3071436882019043,
      "learning_rate": 1.5957661058164697e-05,
      "loss": 0.7871,
      "step": 886
    },
    {
      "epoch": 0.29956095913542724,
      "grad_norm": 1.6607407331466675,
      "learning_rate": 1.5949107378010003e-05,
      "loss": 1.0372,
      "step": 887
    },
    {
      "epoch": 0.2998986828774063,
      "grad_norm": 1.671355128288269,
      "learning_rate": 1.5940546955477366e-05,
      "loss": 0.9307,
      "step": 888
    },
    {
      "epoch": 0.30023640661938533,
      "grad_norm": 1.8619440793991089,
      "learning_rate": 1.5931979800268684e-05,
      "loss": 0.9783,
      "step": 889
    },
    {
      "epoch": 0.3005741303613644,
      "grad_norm": 1.8045966625213623,
      "learning_rate": 1.592340592209347e-05,
      "loss": 0.9386,
      "step": 890
    },
    {
      "epoch": 0.3009118541033435,
      "grad_norm": 1.678651213645935,
      "learning_rate": 1.5914825330668876e-05,
      "loss": 1.0097,
      "step": 891
    },
    {
      "epoch": 0.3012495778453225,
      "grad_norm": 1.8553318977355957,
      "learning_rate": 1.590623803571965e-05,
      "loss": 0.9907,
      "step": 892
    },
    {
      "epoch": 0.30158730158730157,
      "grad_norm": 1.6557639837265015,
      "learning_rate": 1.5897644046978143e-05,
      "loss": 0.9686,
      "step": 893
    },
    {
      "epoch": 0.30192502532928067,
      "grad_norm": 1.725522756576538,
      "learning_rate": 1.5889043374184286e-05,
      "loss": 1.0101,
      "step": 894
    },
    {
      "epoch": 0.3022627490712597,
      "grad_norm": 1.6362131834030151,
      "learning_rate": 1.5880436027085596e-05,
      "loss": 0.9371,
      "step": 895
    },
    {
      "epoch": 0.30260047281323876,
      "grad_norm": 1.7060288190841675,
      "learning_rate": 1.5871822015437145e-05,
      "loss": 0.9741,
      "step": 896
    },
    {
      "epoch": 0.3029381965552178,
      "grad_norm": 1.6950194835662842,
      "learning_rate": 1.5863201349001556e-05,
      "loss": 0.9721,
      "step": 897
    },
    {
      "epoch": 0.3032759202971969,
      "grad_norm": 1.879228115081787,
      "learning_rate": 1.5854574037549006e-05,
      "loss": 1.0407,
      "step": 898
    },
    {
      "epoch": 0.30361364403917596,
      "grad_norm": 1.9987359046936035,
      "learning_rate": 1.5845940090857192e-05,
      "loss": 0.847,
      "step": 899
    },
    {
      "epoch": 0.303951367781155,
      "grad_norm": 1.754077672958374,
      "learning_rate": 1.5837299518711334e-05,
      "loss": 0.9402,
      "step": 900
    },
    {
      "epoch": 0.3042890915231341,
      "grad_norm": 1.760243535041809,
      "learning_rate": 1.582865233090417e-05,
      "loss": 0.9955,
      "step": 901
    },
    {
      "epoch": 0.30462681526511315,
      "grad_norm": 1.694116473197937,
      "learning_rate": 1.5819998537235918e-05,
      "loss": 0.8992,
      "step": 902
    },
    {
      "epoch": 0.3049645390070922,
      "grad_norm": 1.999349594116211,
      "learning_rate": 1.58113381475143e-05,
      "loss": 0.8825,
      "step": 903
    },
    {
      "epoch": 0.30530226274907124,
      "grad_norm": 1.7163528203964233,
      "learning_rate": 1.58026711715545e-05,
      "loss": 0.9226,
      "step": 904
    },
    {
      "epoch": 0.30563998649105034,
      "grad_norm": 1.834763526916504,
      "learning_rate": 1.5793997619179184e-05,
      "loss": 0.9717,
      "step": 905
    },
    {
      "epoch": 0.3059777102330294,
      "grad_norm": 2.2868542671203613,
      "learning_rate": 1.5785317500218444e-05,
      "loss": 0.8923,
      "step": 906
    },
    {
      "epoch": 0.30631543397500843,
      "grad_norm": 1.9197564125061035,
      "learning_rate": 1.5776630824509843e-05,
      "loss": 0.947,
      "step": 907
    },
    {
      "epoch": 0.3066531577169875,
      "grad_norm": 1.7714134454727173,
      "learning_rate": 1.5767937601898356e-05,
      "loss": 0.8877,
      "step": 908
    },
    {
      "epoch": 0.3069908814589666,
      "grad_norm": 1.8209251165390015,
      "learning_rate": 1.575923784223638e-05,
      "loss": 0.9994,
      "step": 909
    },
    {
      "epoch": 0.3073286052009456,
      "grad_norm": 1.831709384918213,
      "learning_rate": 1.575053155538374e-05,
      "loss": 0.9991,
      "step": 910
    },
    {
      "epoch": 0.30766632894292467,
      "grad_norm": 1.8738268613815308,
      "learning_rate": 1.574181875120763e-05,
      "loss": 0.8878,
      "step": 911
    },
    {
      "epoch": 0.3080040526849038,
      "grad_norm": 1.6752723455429077,
      "learning_rate": 1.573309943958265e-05,
      "loss": 0.9342,
      "step": 912
    },
    {
      "epoch": 0.3083417764268828,
      "grad_norm": 1.6565552949905396,
      "learning_rate": 1.5724373630390767e-05,
      "loss": 0.9251,
      "step": 913
    },
    {
      "epoch": 0.30867950016886186,
      "grad_norm": 1.7624239921569824,
      "learning_rate": 1.571564133352131e-05,
      "loss": 0.9417,
      "step": 914
    },
    {
      "epoch": 0.3090172239108409,
      "grad_norm": 1.6646729707717896,
      "learning_rate": 1.5706902558870972e-05,
      "loss": 0.9735,
      "step": 915
    },
    {
      "epoch": 0.30935494765282,
      "grad_norm": 2.0742392539978027,
      "learning_rate": 1.5698157316343774e-05,
      "loss": 0.9046,
      "step": 916
    },
    {
      "epoch": 0.30969267139479906,
      "grad_norm": 2.111624002456665,
      "learning_rate": 1.568940561585108e-05,
      "loss": 0.796,
      "step": 917
    },
    {
      "epoch": 0.3100303951367781,
      "grad_norm": 1.695008397102356,
      "learning_rate": 1.568064746731156e-05,
      "loss": 1.0126,
      "step": 918
    },
    {
      "epoch": 0.3103681188787572,
      "grad_norm": 1.664841651916504,
      "learning_rate": 1.5671882880651203e-05,
      "loss": 0.9853,
      "step": 919
    },
    {
      "epoch": 0.31070584262073625,
      "grad_norm": 1.728818655014038,
      "learning_rate": 1.5663111865803285e-05,
      "loss": 0.9416,
      "step": 920
    },
    {
      "epoch": 0.3110435663627153,
      "grad_norm": 1.88104248046875,
      "learning_rate": 1.565433443270838e-05,
      "loss": 0.9991,
      "step": 921
    },
    {
      "epoch": 0.31138129010469434,
      "grad_norm": 2.0281503200531006,
      "learning_rate": 1.5645550591314322e-05,
      "loss": 0.8841,
      "step": 922
    },
    {
      "epoch": 0.31171901384667344,
      "grad_norm": 1.8744511604309082,
      "learning_rate": 1.5636760351576217e-05,
      "loss": 1.0247,
      "step": 923
    },
    {
      "epoch": 0.3120567375886525,
      "grad_norm": 1.8174734115600586,
      "learning_rate": 1.562796372345642e-05,
      "loss": 0.9657,
      "step": 924
    },
    {
      "epoch": 0.31239446133063153,
      "grad_norm": 1.9457026720046997,
      "learning_rate": 1.5619160716924528e-05,
      "loss": 0.8835,
      "step": 925
    },
    {
      "epoch": 0.3127321850726106,
      "grad_norm": 1.704619288444519,
      "learning_rate": 1.561035134195736e-05,
      "loss": 0.943,
      "step": 926
    },
    {
      "epoch": 0.3130699088145897,
      "grad_norm": 1.5926496982574463,
      "learning_rate": 1.5601535608538958e-05,
      "loss": 0.95,
      "step": 927
    },
    {
      "epoch": 0.3134076325565687,
      "grad_norm": 1.6686426401138306,
      "learning_rate": 1.5592713526660573e-05,
      "loss": 0.8993,
      "step": 928
    },
    {
      "epoch": 0.3137453562985478,
      "grad_norm": 1.9565778970718384,
      "learning_rate": 1.558388510632065e-05,
      "loss": 0.9959,
      "step": 929
    },
    {
      "epoch": 0.3140830800405269,
      "grad_norm": 1.7225892543792725,
      "learning_rate": 1.5575050357524807e-05,
      "loss": 0.8316,
      "step": 930
    },
    {
      "epoch": 0.3144208037825059,
      "grad_norm": 1.7764719724655151,
      "learning_rate": 1.5566209290285853e-05,
      "loss": 0.9269,
      "step": 931
    },
    {
      "epoch": 0.31475852752448497,
      "grad_norm": 1.6721168756484985,
      "learning_rate": 1.555736191462374e-05,
      "loss": 0.9813,
      "step": 932
    },
    {
      "epoch": 0.315096251266464,
      "grad_norm": 1.6121631860733032,
      "learning_rate": 1.5548508240565584e-05,
      "loss": 0.8988,
      "step": 933
    },
    {
      "epoch": 0.3154339750084431,
      "grad_norm": 1.9957661628723145,
      "learning_rate": 1.5539648278145624e-05,
      "loss": 0.9775,
      "step": 934
    },
    {
      "epoch": 0.31577169875042216,
      "grad_norm": 1.8715879917144775,
      "learning_rate": 1.5530782037405244e-05,
      "loss": 1.0317,
      "step": 935
    },
    {
      "epoch": 0.3161094224924012,
      "grad_norm": 1.7323411703109741,
      "learning_rate": 1.5521909528392928e-05,
      "loss": 0.9688,
      "step": 936
    },
    {
      "epoch": 0.31644714623438025,
      "grad_norm": 1.8136608600616455,
      "learning_rate": 1.551303076116427e-05,
      "loss": 0.9607,
      "step": 937
    },
    {
      "epoch": 0.31678486997635935,
      "grad_norm": 1.9968019723892212,
      "learning_rate": 1.5504145745781962e-05,
      "loss": 0.9563,
      "step": 938
    },
    {
      "epoch": 0.3171225937183384,
      "grad_norm": 1.6916370391845703,
      "learning_rate": 1.5495254492315764e-05,
      "loss": 0.9165,
      "step": 939
    },
    {
      "epoch": 0.31746031746031744,
      "grad_norm": 1.6760390996932983,
      "learning_rate": 1.548635701084252e-05,
      "loss": 0.9426,
      "step": 940
    },
    {
      "epoch": 0.31779804120229654,
      "grad_norm": 1.9274353981018066,
      "learning_rate": 1.5477453311446127e-05,
      "loss": 0.8865,
      "step": 941
    },
    {
      "epoch": 0.3181357649442756,
      "grad_norm": 1.5673449039459229,
      "learning_rate": 1.5468543404217526e-05,
      "loss": 0.9822,
      "step": 942
    },
    {
      "epoch": 0.31847348868625464,
      "grad_norm": 1.8774478435516357,
      "learning_rate": 1.5459627299254695e-05,
      "loss": 0.9847,
      "step": 943
    },
    {
      "epoch": 0.3188112124282337,
      "grad_norm": 1.3237369060516357,
      "learning_rate": 1.545070500666264e-05,
      "loss": 0.6387,
      "step": 944
    },
    {
      "epoch": 0.3191489361702128,
      "grad_norm": 1.5955337285995483,
      "learning_rate": 1.5441776536553376e-05,
      "loss": 0.9537,
      "step": 945
    },
    {
      "epoch": 0.31948665991219183,
      "grad_norm": 1.58523690700531,
      "learning_rate": 1.543284189904592e-05,
      "loss": 0.8754,
      "step": 946
    },
    {
      "epoch": 0.3198243836541709,
      "grad_norm": 1.8454502820968628,
      "learning_rate": 1.5423901104266275e-05,
      "loss": 0.894,
      "step": 947
    },
    {
      "epoch": 0.32016210739615,
      "grad_norm": 1.895353078842163,
      "learning_rate": 1.5414954162347435e-05,
      "loss": 0.9618,
      "step": 948
    },
    {
      "epoch": 0.320499831138129,
      "grad_norm": 1.7890435457229614,
      "learning_rate": 1.5406001083429348e-05,
      "loss": 0.9039,
      "step": 949
    },
    {
      "epoch": 0.32083755488010807,
      "grad_norm": 1.7549837827682495,
      "learning_rate": 1.539704187765892e-05,
      "loss": 0.895,
      "step": 950
    },
    {
      "epoch": 0.3211752786220871,
      "grad_norm": 1.6020547151565552,
      "learning_rate": 1.5388076555190002e-05,
      "loss": 0.8485,
      "step": 951
    },
    {
      "epoch": 0.3215130023640662,
      "grad_norm": 2.1602346897125244,
      "learning_rate": 1.537910512618338e-05,
      "loss": 0.9414,
      "step": 952
    },
    {
      "epoch": 0.32185072610604526,
      "grad_norm": 1.689802885055542,
      "learning_rate": 1.5370127600806752e-05,
      "loss": 0.9042,
      "step": 953
    },
    {
      "epoch": 0.3221884498480243,
      "grad_norm": 1.9341133832931519,
      "learning_rate": 1.5361143989234745e-05,
      "loss": 0.9252,
      "step": 954
    },
    {
      "epoch": 0.32252617359000335,
      "grad_norm": 1.8917310237884521,
      "learning_rate": 1.5352154301648855e-05,
      "loss": 1.0126,
      "step": 955
    },
    {
      "epoch": 0.32286389733198245,
      "grad_norm": 1.5514098405838013,
      "learning_rate": 1.534315854823749e-05,
      "loss": 1.0272,
      "step": 956
    },
    {
      "epoch": 0.3232016210739615,
      "grad_norm": 2.052022695541382,
      "learning_rate": 1.5334156739195915e-05,
      "loss": 0.8465,
      "step": 957
    },
    {
      "epoch": 0.32353934481594054,
      "grad_norm": 1.5995906591415405,
      "learning_rate": 1.5325148884726274e-05,
      "loss": 0.8842,
      "step": 958
    },
    {
      "epoch": 0.32387706855791965,
      "grad_norm": 1.6747844219207764,
      "learning_rate": 1.5316134995037545e-05,
      "loss": 0.8907,
      "step": 959
    },
    {
      "epoch": 0.3242147922998987,
      "grad_norm": 1.7473276853561401,
      "learning_rate": 1.5307115080345563e-05,
      "loss": 0.9505,
      "step": 960
    },
    {
      "epoch": 0.32455251604187774,
      "grad_norm": 2.097384452819824,
      "learning_rate": 1.529808915087298e-05,
      "loss": 0.897,
      "step": 961
    },
    {
      "epoch": 0.3248902397838568,
      "grad_norm": 1.7922123670578003,
      "learning_rate": 1.5289057216849262e-05,
      "loss": 0.9068,
      "step": 962
    },
    {
      "epoch": 0.3252279635258359,
      "grad_norm": 1.9021642208099365,
      "learning_rate": 1.5280019288510696e-05,
      "loss": 0.8906,
      "step": 963
    },
    {
      "epoch": 0.32556568726781493,
      "grad_norm": 1.9048819541931152,
      "learning_rate": 1.5270975376100343e-05,
      "loss": 1.0274,
      "step": 964
    },
    {
      "epoch": 0.325903411009794,
      "grad_norm": 1.8295190334320068,
      "learning_rate": 1.5261925489868066e-05,
      "loss": 0.9162,
      "step": 965
    },
    {
      "epoch": 0.3262411347517731,
      "grad_norm": 1.8592089414596558,
      "learning_rate": 1.5252869640070485e-05,
      "loss": 0.9516,
      "step": 966
    },
    {
      "epoch": 0.3265788584937521,
      "grad_norm": 1.9672023057937622,
      "learning_rate": 1.5243807836970979e-05,
      "loss": 1.0005,
      "step": 967
    },
    {
      "epoch": 0.32691658223573117,
      "grad_norm": 2.3448212146759033,
      "learning_rate": 1.5234740090839677e-05,
      "loss": 0.8631,
      "step": 968
    },
    {
      "epoch": 0.3272543059777102,
      "grad_norm": 1.7024081945419312,
      "learning_rate": 1.5225666411953448e-05,
      "loss": 1.0522,
      "step": 969
    },
    {
      "epoch": 0.3275920297196893,
      "grad_norm": 1.9224388599395752,
      "learning_rate": 1.521658681059588e-05,
      "loss": 0.9567,
      "step": 970
    },
    {
      "epoch": 0.32792975346166836,
      "grad_norm": 1.9978340864181519,
      "learning_rate": 1.5207501297057268e-05,
      "loss": 0.8758,
      "step": 971
    },
    {
      "epoch": 0.3282674772036474,
      "grad_norm": 1.8140151500701904,
      "learning_rate": 1.5198409881634617e-05,
      "loss": 0.8496,
      "step": 972
    },
    {
      "epoch": 0.32860520094562645,
      "grad_norm": 1.751039981842041,
      "learning_rate": 1.5189312574631619e-05,
      "loss": 0.9007,
      "step": 973
    },
    {
      "epoch": 0.32894292468760555,
      "grad_norm": 1.9297765493392944,
      "learning_rate": 1.5180209386358641e-05,
      "loss": 0.8341,
      "step": 974
    },
    {
      "epoch": 0.3292806484295846,
      "grad_norm": 1.9308090209960938,
      "learning_rate": 1.5171100327132711e-05,
      "loss": 0.9158,
      "step": 975
    },
    {
      "epoch": 0.32961837217156364,
      "grad_norm": 1.9035474061965942,
      "learning_rate": 1.5161985407277523e-05,
      "loss": 0.8861,
      "step": 976
    },
    {
      "epoch": 0.32995609591354275,
      "grad_norm": 1.8591562509536743,
      "learning_rate": 1.5152864637123397e-05,
      "loss": 0.9426,
      "step": 977
    },
    {
      "epoch": 0.3302938196555218,
      "grad_norm": 1.8116110563278198,
      "learning_rate": 1.5143738027007298e-05,
      "loss": 0.9548,
      "step": 978
    },
    {
      "epoch": 0.33063154339750084,
      "grad_norm": 1.9246737957000732,
      "learning_rate": 1.5134605587272804e-05,
      "loss": 0.8065,
      "step": 979
    },
    {
      "epoch": 0.3309692671394799,
      "grad_norm": 1.9244173765182495,
      "learning_rate": 1.5125467328270098e-05,
      "loss": 1.0167,
      "step": 980
    },
    {
      "epoch": 0.331306990881459,
      "grad_norm": 1.6728218793869019,
      "learning_rate": 1.5116323260355957e-05,
      "loss": 0.6305,
      "step": 981
    },
    {
      "epoch": 0.33164471462343803,
      "grad_norm": 1.7474414110183716,
      "learning_rate": 1.510717339389375e-05,
      "loss": 0.9841,
      "step": 982
    },
    {
      "epoch": 0.3319824383654171,
      "grad_norm": 1.666841983795166,
      "learning_rate": 1.5098017739253407e-05,
      "loss": 0.9369,
      "step": 983
    },
    {
      "epoch": 0.3323201621073962,
      "grad_norm": 1.8118096590042114,
      "learning_rate": 1.5088856306811432e-05,
      "loss": 0.9759,
      "step": 984
    },
    {
      "epoch": 0.3326578858493752,
      "grad_norm": 1.8391228914260864,
      "learning_rate": 1.5079689106950855e-05,
      "loss": 0.9599,
      "step": 985
    },
    {
      "epoch": 0.33299560959135427,
      "grad_norm": 2.210848331451416,
      "learning_rate": 1.5070516150061268e-05,
      "loss": 0.9556,
      "step": 986
    },
    {
      "epoch": 0.3333333333333333,
      "grad_norm": 2.022447347640991,
      "learning_rate": 1.5061337446538767e-05,
      "loss": 0.9913,
      "step": 987
    },
    {
      "epoch": 0.3336710570753124,
      "grad_norm": 1.7348984479904175,
      "learning_rate": 1.5052153006785977e-05,
      "loss": 0.9434,
      "step": 988
    },
    {
      "epoch": 0.33400878081729146,
      "grad_norm": 1.8453718423843384,
      "learning_rate": 1.5042962841212007e-05,
      "loss": 0.9649,
      "step": 989
    },
    {
      "epoch": 0.3343465045592705,
      "grad_norm": 1.634851336479187,
      "learning_rate": 1.5033766960232474e-05,
      "loss": 0.95,
      "step": 990
    },
    {
      "epoch": 0.33468422830124955,
      "grad_norm": 1.8091766834259033,
      "learning_rate": 1.5024565374269462e-05,
      "loss": 0.8841,
      "step": 991
    },
    {
      "epoch": 0.33502195204322865,
      "grad_norm": 1.6516820192337036,
      "learning_rate": 1.5015358093751514e-05,
      "loss": 0.9671,
      "step": 992
    },
    {
      "epoch": 0.3353596757852077,
      "grad_norm": 1.7631797790527344,
      "learning_rate": 1.5006145129113645e-05,
      "loss": 0.9945,
      "step": 993
    },
    {
      "epoch": 0.33569739952718675,
      "grad_norm": 1.8884351253509521,
      "learning_rate": 1.4996926490797297e-05,
      "loss": 0.9113,
      "step": 994
    },
    {
      "epoch": 0.33603512326916585,
      "grad_norm": 1.9042267799377441,
      "learning_rate": 1.498770218925035e-05,
      "loss": 0.8493,
      "step": 995
    },
    {
      "epoch": 0.3363728470111449,
      "grad_norm": 1.980736494064331,
      "learning_rate": 1.4978472234927094e-05,
      "loss": 0.8945,
      "step": 996
    },
    {
      "epoch": 0.33671057075312394,
      "grad_norm": 2.016295909881592,
      "learning_rate": 1.4969236638288238e-05,
      "loss": 0.9279,
      "step": 997
    },
    {
      "epoch": 0.337048294495103,
      "grad_norm": 1.8109267950057983,
      "learning_rate": 1.4959995409800874e-05,
      "loss": 0.9913,
      "step": 998
    },
    {
      "epoch": 0.3373860182370821,
      "grad_norm": 1.9335198402404785,
      "learning_rate": 1.4950748559938484e-05,
      "loss": 0.9741,
      "step": 999
    },
    {
      "epoch": 0.33772374197906113,
      "grad_norm": 2.0967869758605957,
      "learning_rate": 1.4941496099180919e-05,
      "loss": 0.8921,
      "step": 1000
    },
    {
      "epoch": 0.3380614657210402,
      "grad_norm": 1.6611685752868652,
      "learning_rate": 1.4932238038014388e-05,
      "loss": 0.9733,
      "step": 1001
    },
    {
      "epoch": 0.3383991894630192,
      "grad_norm": 1.8603452444076538,
      "learning_rate": 1.4922974386931445e-05,
      "loss": 0.9429,
      "step": 1002
    },
    {
      "epoch": 0.3387369132049983,
      "grad_norm": 1.9593448638916016,
      "learning_rate": 1.4913705156430986e-05,
      "loss": 0.954,
      "step": 1003
    },
    {
      "epoch": 0.33907463694697737,
      "grad_norm": 1.8208675384521484,
      "learning_rate": 1.4904430357018225e-05,
      "loss": 0.8961,
      "step": 1004
    },
    {
      "epoch": 0.3394123606889564,
      "grad_norm": 1.997043490409851,
      "learning_rate": 1.4895149999204693e-05,
      "loss": 0.9356,
      "step": 1005
    },
    {
      "epoch": 0.3397500844309355,
      "grad_norm": 1.8070285320281982,
      "learning_rate": 1.4885864093508208e-05,
      "loss": 0.9266,
      "step": 1006
    },
    {
      "epoch": 0.34008780817291456,
      "grad_norm": 2.0907399654388428,
      "learning_rate": 1.487657265045289e-05,
      "loss": 0.8906,
      "step": 1007
    },
    {
      "epoch": 0.3404255319148936,
      "grad_norm": 1.743486762046814,
      "learning_rate": 1.4867275680569129e-05,
      "loss": 0.9332,
      "step": 1008
    },
    {
      "epoch": 0.34076325565687265,
      "grad_norm": 1.6807969808578491,
      "learning_rate": 1.485797319439358e-05,
      "loss": 0.8382,
      "step": 1009
    },
    {
      "epoch": 0.34110097939885176,
      "grad_norm": 1.2802186012268066,
      "learning_rate": 1.4848665202469143e-05,
      "loss": 0.6016,
      "step": 1010
    },
    {
      "epoch": 0.3414387031408308,
      "grad_norm": 1.879190444946289,
      "learning_rate": 1.4839351715344967e-05,
      "loss": 0.9728,
      "step": 1011
    },
    {
      "epoch": 0.34177642688280985,
      "grad_norm": 2.0218772888183594,
      "learning_rate": 1.4830032743576427e-05,
      "loss": 0.994,
      "step": 1012
    },
    {
      "epoch": 0.34211415062478895,
      "grad_norm": 1.7348403930664062,
      "learning_rate": 1.482070829772511e-05,
      "loss": 0.8922,
      "step": 1013
    },
    {
      "epoch": 0.342451874366768,
      "grad_norm": 1.9652162790298462,
      "learning_rate": 1.4811378388358814e-05,
      "loss": 0.9668,
      "step": 1014
    },
    {
      "epoch": 0.34278959810874704,
      "grad_norm": 1.7499802112579346,
      "learning_rate": 1.4802043026051516e-05,
      "loss": 0.9372,
      "step": 1015
    },
    {
      "epoch": 0.3431273218507261,
      "grad_norm": 1.8248952627182007,
      "learning_rate": 1.4792702221383391e-05,
      "loss": 0.9717,
      "step": 1016
    },
    {
      "epoch": 0.3434650455927052,
      "grad_norm": 1.6907544136047363,
      "learning_rate": 1.4783355984940767e-05,
      "loss": 0.9207,
      "step": 1017
    },
    {
      "epoch": 0.34380276933468423,
      "grad_norm": 1.7490170001983643,
      "learning_rate": 1.4774004327316132e-05,
      "loss": 0.939,
      "step": 1018
    },
    {
      "epoch": 0.3441404930766633,
      "grad_norm": 1.6336274147033691,
      "learning_rate": 1.4764647259108122e-05,
      "loss": 0.9311,
      "step": 1019
    },
    {
      "epoch": 0.3444782168186423,
      "grad_norm": 1.9612398147583008,
      "learning_rate": 1.4755284790921504e-05,
      "loss": 0.9363,
      "step": 1020
    },
    {
      "epoch": 0.3448159405606214,
      "grad_norm": 2.318145275115967,
      "learning_rate": 1.4745916933367166e-05,
      "loss": 0.7937,
      "step": 1021
    },
    {
      "epoch": 0.34515366430260047,
      "grad_norm": 2.1588714122772217,
      "learning_rate": 1.4736543697062096e-05,
      "loss": 0.9681,
      "step": 1022
    },
    {
      "epoch": 0.3454913880445795,
      "grad_norm": 1.6315913200378418,
      "learning_rate": 1.4727165092629388e-05,
      "loss": 0.912,
      "step": 1023
    },
    {
      "epoch": 0.3458291117865586,
      "grad_norm": 1.9684486389160156,
      "learning_rate": 1.4717781130698212e-05,
      "loss": 0.9306,
      "step": 1024
    },
    {
      "epoch": 0.34616683552853766,
      "grad_norm": 1.8798085451126099,
      "learning_rate": 1.4708391821903818e-05,
      "loss": 0.9164,
      "step": 1025
    },
    {
      "epoch": 0.3465045592705167,
      "grad_norm": 1.6425312757492065,
      "learning_rate": 1.4698997176887511e-05,
      "loss": 1.0078,
      "step": 1026
    },
    {
      "epoch": 0.34684228301249576,
      "grad_norm": 1.9080438613891602,
      "learning_rate": 1.468959720629664e-05,
      "loss": 0.9325,
      "step": 1027
    },
    {
      "epoch": 0.34718000675447486,
      "grad_norm": 1.7191963195800781,
      "learning_rate": 1.4680191920784595e-05,
      "loss": 0.9049,
      "step": 1028
    },
    {
      "epoch": 0.3475177304964539,
      "grad_norm": 1.664200782775879,
      "learning_rate": 1.467078133101079e-05,
      "loss": 0.9916,
      "step": 1029
    },
    {
      "epoch": 0.34785545423843295,
      "grad_norm": 1.8735160827636719,
      "learning_rate": 1.4661365447640652e-05,
      "loss": 0.9703,
      "step": 1030
    },
    {
      "epoch": 0.34819317798041205,
      "grad_norm": 1.382426142692566,
      "learning_rate": 1.4651944281345598e-05,
      "loss": 0.558,
      "step": 1031
    },
    {
      "epoch": 0.3485309017223911,
      "grad_norm": 2.2660768032073975,
      "learning_rate": 1.4642517842803043e-05,
      "loss": 0.859,
      "step": 1032
    },
    {
      "epoch": 0.34886862546437014,
      "grad_norm": 1.781118631362915,
      "learning_rate": 1.4633086142696373e-05,
      "loss": 0.9778,
      "step": 1033
    },
    {
      "epoch": 0.3492063492063492,
      "grad_norm": 1.946829915046692,
      "learning_rate": 1.4623649191714938e-05,
      "loss": 1.0114,
      "step": 1034
    },
    {
      "epoch": 0.3495440729483283,
      "grad_norm": 2.004974842071533,
      "learning_rate": 1.4614207000554036e-05,
      "loss": 0.8925,
      "step": 1035
    },
    {
      "epoch": 0.34988179669030733,
      "grad_norm": 1.8297479152679443,
      "learning_rate": 1.4604759579914906e-05,
      "loss": 0.9563,
      "step": 1036
    },
    {
      "epoch": 0.3502195204322864,
      "grad_norm": 1.7602097988128662,
      "learning_rate": 1.4595306940504717e-05,
      "loss": 0.9809,
      "step": 1037
    },
    {
      "epoch": 0.3505572441742654,
      "grad_norm": 1.7393555641174316,
      "learning_rate": 1.4585849093036546e-05,
      "loss": 0.9434,
      "step": 1038
    },
    {
      "epoch": 0.3508949679162445,
      "grad_norm": 1.925390362739563,
      "learning_rate": 1.4576386048229383e-05,
      "loss": 0.9358,
      "step": 1039
    },
    {
      "epoch": 0.3512326916582236,
      "grad_norm": 1.7319129705429077,
      "learning_rate": 1.4566917816808092e-05,
      "loss": 0.8856,
      "step": 1040
    },
    {
      "epoch": 0.3515704154002026,
      "grad_norm": 1.8821488618850708,
      "learning_rate": 1.4557444409503427e-05,
      "loss": 0.9939,
      "step": 1041
    },
    {
      "epoch": 0.3519081391421817,
      "grad_norm": 1.963468074798584,
      "learning_rate": 1.4547965837052015e-05,
      "loss": 0.9102,
      "step": 1042
    },
    {
      "epoch": 0.35224586288416077,
      "grad_norm": 1.7378478050231934,
      "learning_rate": 1.4538482110196319e-05,
      "loss": 0.9552,
      "step": 1043
    },
    {
      "epoch": 0.3525835866261398,
      "grad_norm": 1.991645336151123,
      "learning_rate": 1.4528993239684654e-05,
      "loss": 0.8696,
      "step": 1044
    },
    {
      "epoch": 0.35292131036811886,
      "grad_norm": 1.6152653694152832,
      "learning_rate": 1.4519499236271165e-05,
      "loss": 0.9142,
      "step": 1045
    },
    {
      "epoch": 0.35325903411009796,
      "grad_norm": 1.6878371238708496,
      "learning_rate": 1.4510000110715809e-05,
      "loss": 0.9304,
      "step": 1046
    },
    {
      "epoch": 0.353596757852077,
      "grad_norm": 1.8225268125534058,
      "learning_rate": 1.4500495873784354e-05,
      "loss": 0.8681,
      "step": 1047
    },
    {
      "epoch": 0.35393448159405605,
      "grad_norm": 1.8811557292938232,
      "learning_rate": 1.4490986536248359e-05,
      "loss": 0.8276,
      "step": 1048
    },
    {
      "epoch": 0.35427220533603515,
      "grad_norm": 1.8217591047286987,
      "learning_rate": 1.4481472108885159e-05,
      "loss": 0.9987,
      "step": 1049
    },
    {
      "epoch": 0.3546099290780142,
      "grad_norm": 1.6272413730621338,
      "learning_rate": 1.4471952602477866e-05,
      "loss": 0.9684,
      "step": 1050
    },
    {
      "epoch": 0.35494765281999324,
      "grad_norm": 2.0465359687805176,
      "learning_rate": 1.4462428027815345e-05,
      "loss": 0.9998,
      "step": 1051
    },
    {
      "epoch": 0.3552853765619723,
      "grad_norm": 1.9257118701934814,
      "learning_rate": 1.4452898395692198e-05,
      "loss": 0.8704,
      "step": 1052
    },
    {
      "epoch": 0.3556231003039514,
      "grad_norm": 1.7191137075424194,
      "learning_rate": 1.444336371690877e-05,
      "loss": 1.013,
      "step": 1053
    },
    {
      "epoch": 0.35596082404593044,
      "grad_norm": 2.064124584197998,
      "learning_rate": 1.4433824002271117e-05,
      "loss": 0.9506,
      "step": 1054
    },
    {
      "epoch": 0.3562985477879095,
      "grad_norm": 1.8792414665222168,
      "learning_rate": 1.4424279262591008e-05,
      "loss": 0.9191,
      "step": 1055
    },
    {
      "epoch": 0.3566362715298885,
      "grad_norm": 1.6206231117248535,
      "learning_rate": 1.4414729508685905e-05,
      "loss": 0.9407,
      "step": 1056
    },
    {
      "epoch": 0.35697399527186763,
      "grad_norm": 1.943588376045227,
      "learning_rate": 1.4405174751378953e-05,
      "loss": 0.9652,
      "step": 1057
    },
    {
      "epoch": 0.3573117190138467,
      "grad_norm": 1.866969108581543,
      "learning_rate": 1.4395615001498963e-05,
      "loss": 0.9762,
      "step": 1058
    },
    {
      "epoch": 0.3576494427558257,
      "grad_norm": 1.7495476007461548,
      "learning_rate": 1.4386050269880414e-05,
      "loss": 0.9097,
      "step": 1059
    },
    {
      "epoch": 0.3579871664978048,
      "grad_norm": 1.955832600593567,
      "learning_rate": 1.4376480567363426e-05,
      "loss": 1.0359,
      "step": 1060
    },
    {
      "epoch": 0.35832489023978387,
      "grad_norm": 1.8877729177474976,
      "learning_rate": 1.4366905904793747e-05,
      "loss": 0.9504,
      "step": 1061
    },
    {
      "epoch": 0.3586626139817629,
      "grad_norm": 1.7843695878982544,
      "learning_rate": 1.4357326293022756e-05,
      "loss": 0.9439,
      "step": 1062
    },
    {
      "epoch": 0.35900033772374196,
      "grad_norm": 1.8945456743240356,
      "learning_rate": 1.4347741742907433e-05,
      "loss": 0.9844,
      "step": 1063
    },
    {
      "epoch": 0.35933806146572106,
      "grad_norm": 2.41679310798645,
      "learning_rate": 1.4338152265310362e-05,
      "loss": 0.8292,
      "step": 1064
    },
    {
      "epoch": 0.3596757852077001,
      "grad_norm": 1.7947648763656616,
      "learning_rate": 1.432855787109971e-05,
      "loss": 0.9354,
      "step": 1065
    },
    {
      "epoch": 0.36001350894967915,
      "grad_norm": 1.7991578578948975,
      "learning_rate": 1.4318958571149208e-05,
      "loss": 0.9491,
      "step": 1066
    },
    {
      "epoch": 0.3603512326916582,
      "grad_norm": 1.984706163406372,
      "learning_rate": 1.4309354376338163e-05,
      "loss": 0.994,
      "step": 1067
    },
    {
      "epoch": 0.3606889564336373,
      "grad_norm": 1.9382905960083008,
      "learning_rate": 1.4299745297551408e-05,
      "loss": 1.0259,
      "step": 1068
    },
    {
      "epoch": 0.36102668017561634,
      "grad_norm": 1.8896520137786865,
      "learning_rate": 1.4290131345679337e-05,
      "loss": 0.8826,
      "step": 1069
    },
    {
      "epoch": 0.3613644039175954,
      "grad_norm": 1.6711127758026123,
      "learning_rate": 1.428051253161784e-05,
      "loss": 1.0467,
      "step": 1070
    },
    {
      "epoch": 0.3617021276595745,
      "grad_norm": 1.619842290878296,
      "learning_rate": 1.4270888866268335e-05,
      "loss": 0.9971,
      "step": 1071
    },
    {
      "epoch": 0.36203985140155354,
      "grad_norm": 2.2485158443450928,
      "learning_rate": 1.4261260360537738e-05,
      "loss": 0.754,
      "step": 1072
    },
    {
      "epoch": 0.3623775751435326,
      "grad_norm": 2.2402751445770264,
      "learning_rate": 1.4251627025338443e-05,
      "loss": 0.8075,
      "step": 1073
    },
    {
      "epoch": 0.36271529888551163,
      "grad_norm": 1.987387776374817,
      "learning_rate": 1.4241988871588321e-05,
      "loss": 0.841,
      "step": 1074
    },
    {
      "epoch": 0.36305302262749073,
      "grad_norm": 2.461583137512207,
      "learning_rate": 1.4232345910210703e-05,
      "loss": 0.8265,
      "step": 1075
    },
    {
      "epoch": 0.3633907463694698,
      "grad_norm": 1.8633638620376587,
      "learning_rate": 1.4222698152134373e-05,
      "loss": 0.9232,
      "step": 1076
    },
    {
      "epoch": 0.3637284701114488,
      "grad_norm": 1.909400463104248,
      "learning_rate": 1.4213045608293546e-05,
      "loss": 0.8302,
      "step": 1077
    },
    {
      "epoch": 0.3640661938534279,
      "grad_norm": 2.5440402030944824,
      "learning_rate": 1.4203388289627862e-05,
      "loss": 0.6835,
      "step": 1078
    },
    {
      "epoch": 0.36440391759540697,
      "grad_norm": 1.7967454195022583,
      "learning_rate": 1.4193726207082375e-05,
      "loss": 0.9731,
      "step": 1079
    },
    {
      "epoch": 0.364741641337386,
      "grad_norm": 1.947107195854187,
      "learning_rate": 1.4184059371607535e-05,
      "loss": 0.9522,
      "step": 1080
    },
    {
      "epoch": 0.36507936507936506,
      "grad_norm": 1.971351146697998,
      "learning_rate": 1.4174387794159183e-05,
      "loss": 0.9348,
      "step": 1081
    },
    {
      "epoch": 0.36541708882134416,
      "grad_norm": 2.1089208126068115,
      "learning_rate": 1.4164711485698528e-05,
      "loss": 0.824,
      "step": 1082
    },
    {
      "epoch": 0.3657548125633232,
      "grad_norm": 2.046424388885498,
      "learning_rate": 1.4155030457192147e-05,
      "loss": 0.9164,
      "step": 1083
    },
    {
      "epoch": 0.36609253630530225,
      "grad_norm": 1.939206600189209,
      "learning_rate": 1.4145344719611962e-05,
      "loss": 0.9437,
      "step": 1084
    },
    {
      "epoch": 0.3664302600472813,
      "grad_norm": 1.7749134302139282,
      "learning_rate": 1.4135654283935236e-05,
      "loss": 0.941,
      "step": 1085
    },
    {
      "epoch": 0.3667679837892604,
      "grad_norm": 1.9709738492965698,
      "learning_rate": 1.4125959161144555e-05,
      "loss": 0.9481,
      "step": 1086
    },
    {
      "epoch": 0.36710570753123944,
      "grad_norm": 2.0064754486083984,
      "learning_rate": 1.4116259362227816e-05,
      "loss": 0.9896,
      "step": 1087
    },
    {
      "epoch": 0.3674434312732185,
      "grad_norm": 1.6240124702453613,
      "learning_rate": 1.410655489817821e-05,
      "loss": 0.8913,
      "step": 1088
    },
    {
      "epoch": 0.3677811550151976,
      "grad_norm": 1.9543485641479492,
      "learning_rate": 1.409684577999423e-05,
      "loss": 0.8643,
      "step": 1089
    },
    {
      "epoch": 0.36811887875717664,
      "grad_norm": 2.008417844772339,
      "learning_rate": 1.4087132018679633e-05,
      "loss": 0.9782,
      "step": 1090
    },
    {
      "epoch": 0.3684566024991557,
      "grad_norm": 1.8004097938537598,
      "learning_rate": 1.4077413625243434e-05,
      "loss": 0.9404,
      "step": 1091
    },
    {
      "epoch": 0.36879432624113473,
      "grad_norm": 2.173180341720581,
      "learning_rate": 1.406769061069991e-05,
      "loss": 0.9996,
      "step": 1092
    },
    {
      "epoch": 0.36913204998311383,
      "grad_norm": 1.8467398881912231,
      "learning_rate": 1.405796298606857e-05,
      "loss": 0.9714,
      "step": 1093
    },
    {
      "epoch": 0.3694697737250929,
      "grad_norm": 2.0710389614105225,
      "learning_rate": 1.4048230762374145e-05,
      "loss": 0.8169,
      "step": 1094
    },
    {
      "epoch": 0.3698074974670719,
      "grad_norm": 1.8315657377243042,
      "learning_rate": 1.4038493950646582e-05,
      "loss": 0.9191,
      "step": 1095
    },
    {
      "epoch": 0.370145221209051,
      "grad_norm": 1.6651959419250488,
      "learning_rate": 1.4028752561921023e-05,
      "loss": 0.9108,
      "step": 1096
    },
    {
      "epoch": 0.37048294495103007,
      "grad_norm": 1.998687744140625,
      "learning_rate": 1.4019006607237804e-05,
      "loss": 0.8898,
      "step": 1097
    },
    {
      "epoch": 0.3708206686930091,
      "grad_norm": 1.8711721897125244,
      "learning_rate": 1.4009256097642432e-05,
      "loss": 0.9353,
      "step": 1098
    },
    {
      "epoch": 0.37115839243498816,
      "grad_norm": 1.9383724927902222,
      "learning_rate": 1.3999501044185577e-05,
      "loss": 0.7289,
      "step": 1099
    },
    {
      "epoch": 0.37149611617696726,
      "grad_norm": 2.0143282413482666,
      "learning_rate": 1.3989741457923053e-05,
      "loss": 0.9654,
      "step": 1100
    },
    {
      "epoch": 0.3718338399189463,
      "grad_norm": 2.0706212520599365,
      "learning_rate": 1.3979977349915825e-05,
      "loss": 0.9364,
      "step": 1101
    },
    {
      "epoch": 0.37217156366092535,
      "grad_norm": 1.6361027956008911,
      "learning_rate": 1.3970208731229975e-05,
      "loss": 0.9608,
      "step": 1102
    },
    {
      "epoch": 0.3725092874029044,
      "grad_norm": 2.104321241378784,
      "learning_rate": 1.3960435612936687e-05,
      "loss": 0.9581,
      "step": 1103
    },
    {
      "epoch": 0.3728470111448835,
      "grad_norm": 1.8511741161346436,
      "learning_rate": 1.3950658006112261e-05,
      "loss": 0.8706,
      "step": 1104
    },
    {
      "epoch": 0.37318473488686255,
      "grad_norm": 2.1196465492248535,
      "learning_rate": 1.3940875921838073e-05,
      "loss": 0.9559,
      "step": 1105
    },
    {
      "epoch": 0.3735224586288416,
      "grad_norm": 1.7070279121398926,
      "learning_rate": 1.3931089371200586e-05,
      "loss": 0.9708,
      "step": 1106
    },
    {
      "epoch": 0.3738601823708207,
      "grad_norm": 1.8689969778060913,
      "learning_rate": 1.3921298365291306e-05,
      "loss": 1.0305,
      "step": 1107
    },
    {
      "epoch": 0.37419790611279974,
      "grad_norm": 1.8969258069992065,
      "learning_rate": 1.3911502915206805e-05,
      "loss": 0.9344,
      "step": 1108
    },
    {
      "epoch": 0.3745356298547788,
      "grad_norm": 1.9760124683380127,
      "learning_rate": 1.3901703032048683e-05,
      "loss": 0.9687,
      "step": 1109
    },
    {
      "epoch": 0.37487335359675783,
      "grad_norm": 1.8447012901306152,
      "learning_rate": 1.3891898726923571e-05,
      "loss": 0.9623,
      "step": 1110
    },
    {
      "epoch": 0.37521107733873693,
      "grad_norm": 1.7218875885009766,
      "learning_rate": 1.3882090010943108e-05,
      "loss": 1.0012,
      "step": 1111
    },
    {
      "epoch": 0.375548801080716,
      "grad_norm": 1.9626065492630005,
      "learning_rate": 1.3872276895223923e-05,
      "loss": 0.9483,
      "step": 1112
    },
    {
      "epoch": 0.375886524822695,
      "grad_norm": 1.8915529251098633,
      "learning_rate": 1.3862459390887647e-05,
      "loss": 0.9726,
      "step": 1113
    },
    {
      "epoch": 0.3762242485646741,
      "grad_norm": 2.1333682537078857,
      "learning_rate": 1.3852637509060875e-05,
      "loss": 0.8612,
      "step": 1114
    },
    {
      "epoch": 0.37656197230665317,
      "grad_norm": 1.9874072074890137,
      "learning_rate": 1.384281126087517e-05,
      "loss": 0.8696,
      "step": 1115
    },
    {
      "epoch": 0.3768996960486322,
      "grad_norm": 1.7438346147537231,
      "learning_rate": 1.3832980657467036e-05,
      "loss": 0.8983,
      "step": 1116
    },
    {
      "epoch": 0.37723741979061126,
      "grad_norm": 2.0936760902404785,
      "learning_rate": 1.3823145709977913e-05,
      "loss": 0.9154,
      "step": 1117
    },
    {
      "epoch": 0.37757514353259036,
      "grad_norm": 1.7423537969589233,
      "learning_rate": 1.3813306429554177e-05,
      "loss": 0.904,
      "step": 1118
    },
    {
      "epoch": 0.3779128672745694,
      "grad_norm": 1.9526287317276,
      "learning_rate": 1.3803462827347099e-05,
      "loss": 0.9205,
      "step": 1119
    },
    {
      "epoch": 0.37825059101654845,
      "grad_norm": 2.153003454208374,
      "learning_rate": 1.3793614914512855e-05,
      "loss": 0.9992,
      "step": 1120
    },
    {
      "epoch": 0.3785883147585275,
      "grad_norm": 1.6474826335906982,
      "learning_rate": 1.3783762702212507e-05,
      "loss": 0.9827,
      "step": 1121
    },
    {
      "epoch": 0.3789260385005066,
      "grad_norm": 1.8821412324905396,
      "learning_rate": 1.3773906201611987e-05,
      "loss": 0.8822,
      "step": 1122
    },
    {
      "epoch": 0.37926376224248565,
      "grad_norm": 1.6695109605789185,
      "learning_rate": 1.3764045423882092e-05,
      "loss": 0.9855,
      "step": 1123
    },
    {
      "epoch": 0.3796014859844647,
      "grad_norm": 1.8657668828964233,
      "learning_rate": 1.3754180380198463e-05,
      "loss": 1.0308,
      "step": 1124
    },
    {
      "epoch": 0.3799392097264438,
      "grad_norm": 2.0930817127227783,
      "learning_rate": 1.3744311081741576e-05,
      "loss": 0.7643,
      "step": 1125
    },
    {
      "epoch": 0.38027693346842284,
      "grad_norm": 2.4427947998046875,
      "learning_rate": 1.3734437539696724e-05,
      "loss": 0.9053,
      "step": 1126
    },
    {
      "epoch": 0.3806146572104019,
      "grad_norm": 1.949517846107483,
      "learning_rate": 1.3724559765254024e-05,
      "loss": 0.853,
      "step": 1127
    },
    {
      "epoch": 0.38095238095238093,
      "grad_norm": 1.8175265789031982,
      "learning_rate": 1.371467776960837e-05,
      "loss": 0.8667,
      "step": 1128
    },
    {
      "epoch": 0.38129010469436003,
      "grad_norm": 1.8508483171463013,
      "learning_rate": 1.3704791563959461e-05,
      "loss": 0.9486,
      "step": 1129
    },
    {
      "epoch": 0.3816278284363391,
      "grad_norm": 1.7791342735290527,
      "learning_rate": 1.3694901159511746e-05,
      "loss": 0.9388,
      "step": 1130
    },
    {
      "epoch": 0.3819655521783181,
      "grad_norm": 1.941375494003296,
      "learning_rate": 1.3685006567474453e-05,
      "loss": 0.9062,
      "step": 1131
    },
    {
      "epoch": 0.38230327592029717,
      "grad_norm": 2.0593764781951904,
      "learning_rate": 1.3675107799061544e-05,
      "loss": 0.9164,
      "step": 1132
    },
    {
      "epoch": 0.38264099966227627,
      "grad_norm": 1.8946325778961182,
      "learning_rate": 1.3665204865491713e-05,
      "loss": 0.8968,
      "step": 1133
    },
    {
      "epoch": 0.3829787234042553,
      "grad_norm": 1.8426826000213623,
      "learning_rate": 1.3655297777988381e-05,
      "loss": 0.8373,
      "step": 1134
    },
    {
      "epoch": 0.38331644714623436,
      "grad_norm": 1.9077823162078857,
      "learning_rate": 1.3645386547779675e-05,
      "loss": 0.9465,
      "step": 1135
    },
    {
      "epoch": 0.38365417088821346,
      "grad_norm": 1.8025373220443726,
      "learning_rate": 1.3635471186098415e-05,
      "loss": 0.8414,
      "step": 1136
    },
    {
      "epoch": 0.3839918946301925,
      "grad_norm": 2.006598711013794,
      "learning_rate": 1.3625551704182108e-05,
      "loss": 0.901,
      "step": 1137
    },
    {
      "epoch": 0.38432961837217156,
      "grad_norm": 1.6973176002502441,
      "learning_rate": 1.3615628113272919e-05,
      "loss": 0.914,
      "step": 1138
    },
    {
      "epoch": 0.3846673421141506,
      "grad_norm": 2.147218942642212,
      "learning_rate": 1.3605700424617685e-05,
      "loss": 0.9115,
      "step": 1139
    },
    {
      "epoch": 0.3850050658561297,
      "grad_norm": 1.8710441589355469,
      "learning_rate": 1.359576864946788e-05,
      "loss": 0.9808,
      "step": 1140
    },
    {
      "epoch": 0.38534278959810875,
      "grad_norm": 1.9602144956588745,
      "learning_rate": 1.358583279907961e-05,
      "loss": 0.9062,
      "step": 1141
    },
    {
      "epoch": 0.3856805133400878,
      "grad_norm": 1.952062964439392,
      "learning_rate": 1.3575892884713597e-05,
      "loss": 0.9159,
      "step": 1142
    },
    {
      "epoch": 0.3860182370820669,
      "grad_norm": 1.7300454378128052,
      "learning_rate": 1.3565948917635169e-05,
      "loss": 1.0286,
      "step": 1143
    },
    {
      "epoch": 0.38635596082404594,
      "grad_norm": 1.683313012123108,
      "learning_rate": 1.3556000909114256e-05,
      "loss": 0.9495,
      "step": 1144
    },
    {
      "epoch": 0.386693684566025,
      "grad_norm": 1.6555192470550537,
      "learning_rate": 1.3546048870425356e-05,
      "loss": 0.9396,
      "step": 1145
    },
    {
      "epoch": 0.38703140830800403,
      "grad_norm": 2.154519557952881,
      "learning_rate": 1.3536092812847548e-05,
      "loss": 0.8995,
      "step": 1146
    },
    {
      "epoch": 0.38736913204998313,
      "grad_norm": 1.9346210956573486,
      "learning_rate": 1.3526132747664446e-05,
      "loss": 0.9435,
      "step": 1147
    },
    {
      "epoch": 0.3877068557919622,
      "grad_norm": 1.8418406248092651,
      "learning_rate": 1.3516168686164228e-05,
      "loss": 0.8877,
      "step": 1148
    },
    {
      "epoch": 0.3880445795339412,
      "grad_norm": 1.6585381031036377,
      "learning_rate": 1.3506200639639586e-05,
      "loss": 0.9824,
      "step": 1149
    },
    {
      "epoch": 0.38838230327592027,
      "grad_norm": 2.033780574798584,
      "learning_rate": 1.349622861938774e-05,
      "loss": 1.0155,
      "step": 1150
    },
    {
      "epoch": 0.38872002701789937,
      "grad_norm": 2.485130786895752,
      "learning_rate": 1.3486252636710398e-05,
      "loss": 0.9225,
      "step": 1151
    },
    {
      "epoch": 0.3890577507598784,
      "grad_norm": 1.9694938659667969,
      "learning_rate": 1.3476272702913771e-05,
      "loss": 0.8916,
      "step": 1152
    },
    {
      "epoch": 0.38939547450185746,
      "grad_norm": 2.0955631732940674,
      "learning_rate": 1.3466288829308551e-05,
      "loss": 0.7706,
      "step": 1153
    },
    {
      "epoch": 0.38973319824383656,
      "grad_norm": 1.9798814058303833,
      "learning_rate": 1.3456301027209884e-05,
      "loss": 0.8921,
      "step": 1154
    },
    {
      "epoch": 0.3900709219858156,
      "grad_norm": 1.8467589616775513,
      "learning_rate": 1.3446309307937378e-05,
      "loss": 0.9097,
      "step": 1155
    },
    {
      "epoch": 0.39040864572779466,
      "grad_norm": 1.7739150524139404,
      "learning_rate": 1.3436313682815066e-05,
      "loss": 0.9751,
      "step": 1156
    },
    {
      "epoch": 0.3907463694697737,
      "grad_norm": 1.3572055101394653,
      "learning_rate": 1.3426314163171428e-05,
      "loss": 0.6169,
      "step": 1157
    },
    {
      "epoch": 0.3910840932117528,
      "grad_norm": 1.858239769935608,
      "learning_rate": 1.3416310760339342e-05,
      "loss": 0.9291,
      "step": 1158
    },
    {
      "epoch": 0.39142181695373185,
      "grad_norm": 2.3136749267578125,
      "learning_rate": 1.3406303485656094e-05,
      "loss": 0.9306,
      "step": 1159
    },
    {
      "epoch": 0.3917595406957109,
      "grad_norm": 1.8299657106399536,
      "learning_rate": 1.3396292350463355e-05,
      "loss": 0.8967,
      "step": 1160
    },
    {
      "epoch": 0.39209726443769,
      "grad_norm": 1.7923498153686523,
      "learning_rate": 1.3386277366107177e-05,
      "loss": 0.9226,
      "step": 1161
    },
    {
      "epoch": 0.39243498817966904,
      "grad_norm": 1.7742265462875366,
      "learning_rate": 1.3376258543937968e-05,
      "loss": 0.9873,
      "step": 1162
    },
    {
      "epoch": 0.3927727119216481,
      "grad_norm": 1.8665355443954468,
      "learning_rate": 1.3366235895310486e-05,
      "loss": 0.9312,
      "step": 1163
    },
    {
      "epoch": 0.39311043566362713,
      "grad_norm": 1.7860281467437744,
      "learning_rate": 1.3356209431583831e-05,
      "loss": 0.9623,
      "step": 1164
    },
    {
      "epoch": 0.39344815940560623,
      "grad_norm": 1.6853506565093994,
      "learning_rate": 1.3346179164121418e-05,
      "loss": 0.8801,
      "step": 1165
    },
    {
      "epoch": 0.3937858831475853,
      "grad_norm": 1.8677129745483398,
      "learning_rate": 1.3336145104290985e-05,
      "loss": 0.9477,
      "step": 1166
    },
    {
      "epoch": 0.3941236068895643,
      "grad_norm": 1.9344143867492676,
      "learning_rate": 1.3326107263464559e-05,
      "loss": 0.9269,
      "step": 1167
    },
    {
      "epoch": 0.39446133063154337,
      "grad_norm": 2.2069036960601807,
      "learning_rate": 1.331606565301845e-05,
      "loss": 0.8375,
      "step": 1168
    },
    {
      "epoch": 0.3947990543735225,
      "grad_norm": 2.011625289916992,
      "learning_rate": 1.330602028433325e-05,
      "loss": 0.9246,
      "step": 1169
    },
    {
      "epoch": 0.3951367781155015,
      "grad_norm": 1.895991325378418,
      "learning_rate": 1.3295971168793802e-05,
      "loss": 0.9329,
      "step": 1170
    },
    {
      "epoch": 0.39547450185748056,
      "grad_norm": 1.9360542297363281,
      "learning_rate": 1.3285918317789203e-05,
      "loss": 1.0057,
      "step": 1171
    },
    {
      "epoch": 0.39581222559945967,
      "grad_norm": 1.979845643043518,
      "learning_rate": 1.3275861742712777e-05,
      "loss": 0.9098,
      "step": 1172
    },
    {
      "epoch": 0.3961499493414387,
      "grad_norm": 2.233832836151123,
      "learning_rate": 1.3265801454962066e-05,
      "loss": 0.888,
      "step": 1173
    },
    {
      "epoch": 0.39648767308341776,
      "grad_norm": 2.030988931655884,
      "learning_rate": 1.3255737465938832e-05,
      "loss": 0.9663,
      "step": 1174
    },
    {
      "epoch": 0.3968253968253968,
      "grad_norm": 2.3149735927581787,
      "learning_rate": 1.3245669787049021e-05,
      "loss": 0.8756,
      "step": 1175
    },
    {
      "epoch": 0.3971631205673759,
      "grad_norm": 2.0783510208129883,
      "learning_rate": 1.3235598429702762e-05,
      "loss": 0.9235,
      "step": 1176
    },
    {
      "epoch": 0.39750084430935495,
      "grad_norm": 1.9827739000320435,
      "learning_rate": 1.322552340531436e-05,
      "loss": 0.853,
      "step": 1177
    },
    {
      "epoch": 0.397838568051334,
      "grad_norm": 1.8293756246566772,
      "learning_rate": 1.3215444725302269e-05,
      "loss": 0.9823,
      "step": 1178
    },
    {
      "epoch": 0.3981762917933131,
      "grad_norm": 1.7556391954421997,
      "learning_rate": 1.3205362401089086e-05,
      "loss": 0.9299,
      "step": 1179
    },
    {
      "epoch": 0.39851401553529214,
      "grad_norm": 1.8551040887832642,
      "learning_rate": 1.3195276444101546e-05,
      "loss": 0.9619,
      "step": 1180
    },
    {
      "epoch": 0.3988517392772712,
      "grad_norm": 1.9954276084899902,
      "learning_rate": 1.3185186865770489e-05,
      "loss": 0.9306,
      "step": 1181
    },
    {
      "epoch": 0.39918946301925023,
      "grad_norm": 1.956292748451233,
      "learning_rate": 1.3175093677530868e-05,
      "loss": 1.0099,
      "step": 1182
    },
    {
      "epoch": 0.39952718676122934,
      "grad_norm": 1.8072935342788696,
      "learning_rate": 1.316499689082173e-05,
      "loss": 1.0059,
      "step": 1183
    },
    {
      "epoch": 0.3998649105032084,
      "grad_norm": 1.8124750852584839,
      "learning_rate": 1.3154896517086188e-05,
      "loss": 0.9898,
      "step": 1184
    },
    {
      "epoch": 0.4002026342451874,
      "grad_norm": 2.139202833175659,
      "learning_rate": 1.3144792567771436e-05,
      "loss": 0.9057,
      "step": 1185
    },
    {
      "epoch": 0.4005403579871665,
      "grad_norm": 1.8282634019851685,
      "learning_rate": 1.3134685054328702e-05,
      "loss": 0.9188,
      "step": 1186
    },
    {
      "epoch": 0.4008780817291456,
      "grad_norm": 1.9855763912200928,
      "learning_rate": 1.3124573988213272e-05,
      "loss": 0.9345,
      "step": 1187
    },
    {
      "epoch": 0.4012158054711246,
      "grad_norm": 1.8315223455429077,
      "learning_rate": 1.3114459380884446e-05,
      "loss": 1.0299,
      "step": 1188
    },
    {
      "epoch": 0.40155352921310367,
      "grad_norm": 2.085352659225464,
      "learning_rate": 1.310434124380554e-05,
      "loss": 0.8648,
      "step": 1189
    },
    {
      "epoch": 0.40189125295508277,
      "grad_norm": 1.9488624334335327,
      "learning_rate": 1.3094219588443874e-05,
      "loss": 0.8925,
      "step": 1190
    },
    {
      "epoch": 0.4022289766970618,
      "grad_norm": 1.827561855316162,
      "learning_rate": 1.3084094426270755e-05,
      "loss": 0.9087,
      "step": 1191
    },
    {
      "epoch": 0.40256670043904086,
      "grad_norm": 1.792706847190857,
      "learning_rate": 1.3073965768761461e-05,
      "loss": 0.9773,
      "step": 1192
    },
    {
      "epoch": 0.4029044241810199,
      "grad_norm": 1.8823603391647339,
      "learning_rate": 1.3063833627395231e-05,
      "loss": 0.8209,
      "step": 1193
    },
    {
      "epoch": 0.403242147922999,
      "grad_norm": 1.9375311136245728,
      "learning_rate": 1.3053698013655253e-05,
      "loss": 0.8997,
      "step": 1194
    },
    {
      "epoch": 0.40357987166497805,
      "grad_norm": 2.2598185539245605,
      "learning_rate": 1.3043558939028657e-05,
      "loss": 0.7642,
      "step": 1195
    },
    {
      "epoch": 0.4039175954069571,
      "grad_norm": 2.2706687450408936,
      "learning_rate": 1.3033416415006486e-05,
      "loss": 0.7874,
      "step": 1196
    },
    {
      "epoch": 0.40425531914893614,
      "grad_norm": 2.1890311241149902,
      "learning_rate": 1.3023270453083701e-05,
      "loss": 0.8122,
      "step": 1197
    },
    {
      "epoch": 0.40459304289091524,
      "grad_norm": 1.8818801641464233,
      "learning_rate": 1.3013121064759148e-05,
      "loss": 1.0118,
      "step": 1198
    },
    {
      "epoch": 0.4049307666328943,
      "grad_norm": 1.7273484468460083,
      "learning_rate": 1.3002968261535566e-05,
      "loss": 0.9519,
      "step": 1199
    },
    {
      "epoch": 0.40526849037487334,
      "grad_norm": 2.0114917755126953,
      "learning_rate": 1.2992812054919566e-05,
      "loss": 0.8603,
      "step": 1200
    },
    {
      "epoch": 0.40560621411685244,
      "grad_norm": 1.8165820837020874,
      "learning_rate": 1.2982652456421607e-05,
      "loss": 0.846,
      "step": 1201
    },
    {
      "epoch": 0.4059439378588315,
      "grad_norm": 2.1232945919036865,
      "learning_rate": 1.2972489477555995e-05,
      "loss": 0.9422,
      "step": 1202
    },
    {
      "epoch": 0.40628166160081053,
      "grad_norm": 2.146646022796631,
      "learning_rate": 1.296232312984087e-05,
      "loss": 0.9057,
      "step": 1203
    },
    {
      "epoch": 0.4066193853427896,
      "grad_norm": 1.6918911933898926,
      "learning_rate": 1.2952153424798193e-05,
      "loss": 0.92,
      "step": 1204
    },
    {
      "epoch": 0.4069571090847687,
      "grad_norm": 1.8909733295440674,
      "learning_rate": 1.2941980373953719e-05,
      "loss": 0.8628,
      "step": 1205
    },
    {
      "epoch": 0.4072948328267477,
      "grad_norm": 1.810035228729248,
      "learning_rate": 1.293180398883701e-05,
      "loss": 0.9572,
      "step": 1206
    },
    {
      "epoch": 0.40763255656872677,
      "grad_norm": 1.8162001371383667,
      "learning_rate": 1.292162428098139e-05,
      "loss": 0.9787,
      "step": 1207
    },
    {
      "epoch": 0.40797028031070587,
      "grad_norm": 1.9078537225723267,
      "learning_rate": 1.2911441261923965e-05,
      "loss": 0.8925,
      "step": 1208
    },
    {
      "epoch": 0.4083080040526849,
      "grad_norm": 2.091458559036255,
      "learning_rate": 1.2901254943205582e-05,
      "loss": 0.8954,
      "step": 1209
    },
    {
      "epoch": 0.40864572779466396,
      "grad_norm": 1.9199728965759277,
      "learning_rate": 1.289106533637083e-05,
      "loss": 0.861,
      "step": 1210
    },
    {
      "epoch": 0.408983451536643,
      "grad_norm": 2.108295440673828,
      "learning_rate": 1.2880872452968034e-05,
      "loss": 0.9107,
      "step": 1211
    },
    {
      "epoch": 0.4093211752786221,
      "grad_norm": 1.86684250831604,
      "learning_rate": 1.2870676304549217e-05,
      "loss": 0.8663,
      "step": 1212
    },
    {
      "epoch": 0.40965889902060115,
      "grad_norm": 1.7666953802108765,
      "learning_rate": 1.2860476902670116e-05,
      "loss": 0.8907,
      "step": 1213
    },
    {
      "epoch": 0.4099966227625802,
      "grad_norm": 1.8922408819198608,
      "learning_rate": 1.2850274258890144e-05,
      "loss": 0.8949,
      "step": 1214
    },
    {
      "epoch": 0.41033434650455924,
      "grad_norm": 1.8432509899139404,
      "learning_rate": 1.2840068384772399e-05,
      "loss": 0.9345,
      "step": 1215
    },
    {
      "epoch": 0.41067207024653835,
      "grad_norm": 2.022132158279419,
      "learning_rate": 1.2829859291883629e-05,
      "loss": 0.8814,
      "step": 1216
    },
    {
      "epoch": 0.4110097939885174,
      "grad_norm": 1.7533247470855713,
      "learning_rate": 1.2819646991794242e-05,
      "loss": 0.9257,
      "step": 1217
    },
    {
      "epoch": 0.41134751773049644,
      "grad_norm": 1.7666136026382446,
      "learning_rate": 1.280943149607827e-05,
      "loss": 0.9574,
      "step": 1218
    },
    {
      "epoch": 0.41168524147247554,
      "grad_norm": 1.9618439674377441,
      "learning_rate": 1.2799212816313375e-05,
      "loss": 0.9889,
      "step": 1219
    },
    {
      "epoch": 0.4120229652144546,
      "grad_norm": 1.8391505479812622,
      "learning_rate": 1.2788990964080824e-05,
      "loss": 0.9626,
      "step": 1220
    },
    {
      "epoch": 0.41236068895643363,
      "grad_norm": 1.89092218875885,
      "learning_rate": 1.2778765950965473e-05,
      "loss": 0.9528,
      "step": 1221
    },
    {
      "epoch": 0.4126984126984127,
      "grad_norm": 1.8667736053466797,
      "learning_rate": 1.2768537788555777e-05,
      "loss": 0.893,
      "step": 1222
    },
    {
      "epoch": 0.4130361364403918,
      "grad_norm": 1.9640095233917236,
      "learning_rate": 1.2758306488443744e-05,
      "loss": 0.8942,
      "step": 1223
    },
    {
      "epoch": 0.4133738601823708,
      "grad_norm": 1.8662136793136597,
      "learning_rate": 1.274807206222494e-05,
      "loss": 0.9543,
      "step": 1224
    },
    {
      "epoch": 0.41371158392434987,
      "grad_norm": 1.838829755783081,
      "learning_rate": 1.2737834521498489e-05,
      "loss": 0.806,
      "step": 1225
    },
    {
      "epoch": 0.41404930766632897,
      "grad_norm": 2.156665325164795,
      "learning_rate": 1.2727593877867024e-05,
      "loss": 0.8979,
      "step": 1226
    },
    {
      "epoch": 0.414387031408308,
      "grad_norm": 2.411325216293335,
      "learning_rate": 1.2717350142936713e-05,
      "loss": 0.8855,
      "step": 1227
    },
    {
      "epoch": 0.41472475515028706,
      "grad_norm": 2.144016981124878,
      "learning_rate": 1.2707103328317211e-05,
      "loss": 0.9919,
      "step": 1228
    },
    {
      "epoch": 0.4150624788922661,
      "grad_norm": 2.289569139480591,
      "learning_rate": 1.2696853445621677e-05,
      "loss": 0.8534,
      "step": 1229
    },
    {
      "epoch": 0.4154002026342452,
      "grad_norm": 2.1525347232818604,
      "learning_rate": 1.268660050646674e-05,
      "loss": 0.8544,
      "step": 1230
    },
    {
      "epoch": 0.41573792637622425,
      "grad_norm": 1.860866904258728,
      "learning_rate": 1.2676344522472497e-05,
      "loss": 0.9853,
      "step": 1231
    },
    {
      "epoch": 0.4160756501182033,
      "grad_norm": 1.9257713556289673,
      "learning_rate": 1.2666085505262486e-05,
      "loss": 0.8685,
      "step": 1232
    },
    {
      "epoch": 0.41641337386018235,
      "grad_norm": 2.4670870304107666,
      "learning_rate": 1.2655823466463697e-05,
      "loss": 0.742,
      "step": 1233
    },
    {
      "epoch": 0.41675109760216145,
      "grad_norm": 2.6579794883728027,
      "learning_rate": 1.2645558417706539e-05,
      "loss": 0.8255,
      "step": 1234
    },
    {
      "epoch": 0.4170888213441405,
      "grad_norm": 1.7902908325195312,
      "learning_rate": 1.2635290370624823e-05,
      "loss": 0.9728,
      "step": 1235
    },
    {
      "epoch": 0.41742654508611954,
      "grad_norm": 1.801051378250122,
      "learning_rate": 1.2625019336855774e-05,
      "loss": 0.9262,
      "step": 1236
    },
    {
      "epoch": 0.41776426882809864,
      "grad_norm": 1.9690940380096436,
      "learning_rate": 1.261474532803999e-05,
      "loss": 0.9424,
      "step": 1237
    },
    {
      "epoch": 0.4181019925700777,
      "grad_norm": 2.043015718460083,
      "learning_rate": 1.2604468355821446e-05,
      "loss": 0.9335,
      "step": 1238
    },
    {
      "epoch": 0.41843971631205673,
      "grad_norm": 1.7600940465927124,
      "learning_rate": 1.2594188431847476e-05,
      "loss": 0.8925,
      "step": 1239
    },
    {
      "epoch": 0.4187774400540358,
      "grad_norm": 2.0738720893859863,
      "learning_rate": 1.2583905567768753e-05,
      "loss": 0.863,
      "step": 1240
    },
    {
      "epoch": 0.4191151637960149,
      "grad_norm": 2.285740375518799,
      "learning_rate": 1.2573619775239295e-05,
      "loss": 0.7894,
      "step": 1241
    },
    {
      "epoch": 0.4194528875379939,
      "grad_norm": 1.9206831455230713,
      "learning_rate": 1.2563331065916418e-05,
      "loss": 0.9738,
      "step": 1242
    },
    {
      "epoch": 0.41979061127997297,
      "grad_norm": 1.964751124382019,
      "learning_rate": 1.2553039451460768e-05,
      "loss": 0.9371,
      "step": 1243
    },
    {
      "epoch": 0.42012833502195207,
      "grad_norm": 2.053690195083618,
      "learning_rate": 1.2542744943536266e-05,
      "loss": 0.9425,
      "step": 1244
    },
    {
      "epoch": 0.4204660587639311,
      "grad_norm": 1.8210452795028687,
      "learning_rate": 1.2532447553810125e-05,
      "loss": 0.9167,
      "step": 1245
    },
    {
      "epoch": 0.42080378250591016,
      "grad_norm": 1.800986886024475,
      "learning_rate": 1.2522147293952806e-05,
      "loss": 0.9871,
      "step": 1246
    },
    {
      "epoch": 0.4211415062478892,
      "grad_norm": 1.8193479776382446,
      "learning_rate": 1.2511844175638046e-05,
      "loss": 0.9633,
      "step": 1247
    },
    {
      "epoch": 0.4214792299898683,
      "grad_norm": 2.0621864795684814,
      "learning_rate": 1.2501538210542805e-05,
      "loss": 0.921,
      "step": 1248
    },
    {
      "epoch": 0.42181695373184735,
      "grad_norm": 2.0097219944000244,
      "learning_rate": 1.2491229410347272e-05,
      "loss": 0.9767,
      "step": 1249
    },
    {
      "epoch": 0.4221546774738264,
      "grad_norm": 2.014012575149536,
      "learning_rate": 1.2480917786734855e-05,
      "loss": 0.9383,
      "step": 1250
    },
    {
      "epoch": 0.42249240121580545,
      "grad_norm": 2.158416986465454,
      "learning_rate": 1.2470603351392154e-05,
      "loss": 0.8202,
      "step": 1251
    },
    {
      "epoch": 0.42283012495778455,
      "grad_norm": 1.8901411294937134,
      "learning_rate": 1.2460286116008963e-05,
      "loss": 0.9365,
      "step": 1252
    },
    {
      "epoch": 0.4231678486997636,
      "grad_norm": 2.4007437229156494,
      "learning_rate": 1.2449966092278247e-05,
      "loss": 0.8298,
      "step": 1253
    },
    {
      "epoch": 0.42350557244174264,
      "grad_norm": 2.4351422786712646,
      "learning_rate": 1.243964329189613e-05,
      "loss": 1.0076,
      "step": 1254
    },
    {
      "epoch": 0.42384329618372174,
      "grad_norm": 1.853796124458313,
      "learning_rate": 1.242931772656188e-05,
      "loss": 0.8867,
      "step": 1255
    },
    {
      "epoch": 0.4241810199257008,
      "grad_norm": 2.0209298133850098,
      "learning_rate": 1.2418989407977906e-05,
      "loss": 0.8121,
      "step": 1256
    },
    {
      "epoch": 0.42451874366767983,
      "grad_norm": 1.950087308883667,
      "learning_rate": 1.2408658347849735e-05,
      "loss": 0.9011,
      "step": 1257
    },
    {
      "epoch": 0.4248564674096589,
      "grad_norm": 1.927813172340393,
      "learning_rate": 1.2398324557885994e-05,
      "loss": 0.9811,
      "step": 1258
    },
    {
      "epoch": 0.425194191151638,
      "grad_norm": 1.734915018081665,
      "learning_rate": 1.2387988049798412e-05,
      "loss": 0.905,
      "step": 1259
    },
    {
      "epoch": 0.425531914893617,
      "grad_norm": 1.9399888515472412,
      "learning_rate": 1.2377648835301795e-05,
      "loss": 1.0745,
      "step": 1260
    },
    {
      "epoch": 0.42586963863559607,
      "grad_norm": 1.9512065649032593,
      "learning_rate": 1.236730692611402e-05,
      "loss": 0.9372,
      "step": 1261
    },
    {
      "epoch": 0.4262073623775751,
      "grad_norm": 1.7759578227996826,
      "learning_rate": 1.235696233395601e-05,
      "loss": 0.9051,
      "step": 1262
    },
    {
      "epoch": 0.4265450861195542,
      "grad_norm": 2.0102717876434326,
      "learning_rate": 1.234661507055174e-05,
      "loss": 0.9119,
      "step": 1263
    },
    {
      "epoch": 0.42688280986153326,
      "grad_norm": 2.3365261554718018,
      "learning_rate": 1.2336265147628202e-05,
      "loss": 0.921,
      "step": 1264
    },
    {
      "epoch": 0.4272205336035123,
      "grad_norm": 2.1019299030303955,
      "learning_rate": 1.2325912576915407e-05,
      "loss": 1.0094,
      "step": 1265
    },
    {
      "epoch": 0.4275582573454914,
      "grad_norm": 2.3696141242980957,
      "learning_rate": 1.2315557370146371e-05,
      "loss": 0.8332,
      "step": 1266
    },
    {
      "epoch": 0.42789598108747046,
      "grad_norm": 1.7975027561187744,
      "learning_rate": 1.2305199539057086e-05,
      "loss": 0.9832,
      "step": 1267
    },
    {
      "epoch": 0.4282337048294495,
      "grad_norm": 1.7722046375274658,
      "learning_rate": 1.2294839095386528e-05,
      "loss": 0.965,
      "step": 1268
    },
    {
      "epoch": 0.42857142857142855,
      "grad_norm": 2.064667224884033,
      "learning_rate": 1.2284476050876633e-05,
      "loss": 0.9724,
      "step": 1269
    },
    {
      "epoch": 0.42890915231340765,
      "grad_norm": 1.866690993309021,
      "learning_rate": 1.227411041727228e-05,
      "loss": 0.9438,
      "step": 1270
    },
    {
      "epoch": 0.4292468760553867,
      "grad_norm": 1.839307427406311,
      "learning_rate": 1.2263742206321287e-05,
      "loss": 0.8904,
      "step": 1271
    },
    {
      "epoch": 0.42958459979736574,
      "grad_norm": 2.0098676681518555,
      "learning_rate": 1.225337142977439e-05,
      "loss": 0.8605,
      "step": 1272
    },
    {
      "epoch": 0.42992232353934484,
      "grad_norm": 2.028200387954712,
      "learning_rate": 1.2242998099385234e-05,
      "loss": 0.8602,
      "step": 1273
    },
    {
      "epoch": 0.4302600472813239,
      "grad_norm": 2.1327691078186035,
      "learning_rate": 1.2232622226910361e-05,
      "loss": 0.9003,
      "step": 1274
    },
    {
      "epoch": 0.43059777102330293,
      "grad_norm": 2.0550925731658936,
      "learning_rate": 1.2222243824109189e-05,
      "loss": 0.928,
      "step": 1275
    },
    {
      "epoch": 0.430935494765282,
      "grad_norm": 1.9582988023757935,
      "learning_rate": 1.2211862902743998e-05,
      "loss": 0.9936,
      "step": 1276
    },
    {
      "epoch": 0.4312732185072611,
      "grad_norm": 1.8253018856048584,
      "learning_rate": 1.2201479474579946e-05,
      "loss": 0.9635,
      "step": 1277
    },
    {
      "epoch": 0.4316109422492401,
      "grad_norm": 1.712349772453308,
      "learning_rate": 1.2191093551385004e-05,
      "loss": 0.925,
      "step": 1278
    },
    {
      "epoch": 0.43194866599121917,
      "grad_norm": 1.7755755186080933,
      "learning_rate": 1.2180705144929989e-05,
      "loss": 0.8731,
      "step": 1279
    },
    {
      "epoch": 0.4322863897331982,
      "grad_norm": 1.9583232402801514,
      "learning_rate": 1.2170314266988525e-05,
      "loss": 0.7861,
      "step": 1280
    },
    {
      "epoch": 0.4326241134751773,
      "grad_norm": 1.9060887098312378,
      "learning_rate": 1.215992092933704e-05,
      "loss": 0.8586,
      "step": 1281
    },
    {
      "epoch": 0.43296183721715636,
      "grad_norm": 1.937493085861206,
      "learning_rate": 1.2149525143754748e-05,
      "loss": 0.9178,
      "step": 1282
    },
    {
      "epoch": 0.4332995609591354,
      "grad_norm": 1.8489172458648682,
      "learning_rate": 1.2139126922023633e-05,
      "loss": 0.9897,
      "step": 1283
    },
    {
      "epoch": 0.4336372847011145,
      "grad_norm": 1.9744218587875366,
      "learning_rate": 1.212872627592845e-05,
      "loss": 0.9194,
      "step": 1284
    },
    {
      "epoch": 0.43397500844309356,
      "grad_norm": 2.1447272300720215,
      "learning_rate": 1.2118323217256694e-05,
      "loss": 0.888,
      "step": 1285
    },
    {
      "epoch": 0.4343127321850726,
      "grad_norm": 1.792178750038147,
      "learning_rate": 1.2107917757798595e-05,
      "loss": 0.8844,
      "step": 1286
    },
    {
      "epoch": 0.43465045592705165,
      "grad_norm": 1.8723993301391602,
      "learning_rate": 1.2097509909347112e-05,
      "loss": 0.8607,
      "step": 1287
    },
    {
      "epoch": 0.43498817966903075,
      "grad_norm": 1.9561115503311157,
      "learning_rate": 1.2087099683697899e-05,
      "loss": 0.9937,
      "step": 1288
    },
    {
      "epoch": 0.4353259034110098,
      "grad_norm": 2.203263282775879,
      "learning_rate": 1.207668709264931e-05,
      "loss": 0.8598,
      "step": 1289
    },
    {
      "epoch": 0.43566362715298884,
      "grad_norm": 1.8464620113372803,
      "learning_rate": 1.206627214800238e-05,
      "loss": 0.8224,
      "step": 1290
    },
    {
      "epoch": 0.43600135089496794,
      "grad_norm": 1.9765788316726685,
      "learning_rate": 1.2055854861560814e-05,
      "loss": 0.9204,
      "step": 1291
    },
    {
      "epoch": 0.436339074636947,
      "grad_norm": 2.3137168884277344,
      "learning_rate": 1.204543524513097e-05,
      "loss": 0.9857,
      "step": 1292
    },
    {
      "epoch": 0.43667679837892603,
      "grad_norm": 2.2410762310028076,
      "learning_rate": 1.203501331052184e-05,
      "loss": 0.9416,
      "step": 1293
    },
    {
      "epoch": 0.4370145221209051,
      "grad_norm": 1.948917269706726,
      "learning_rate": 1.2024589069545049e-05,
      "loss": 0.9387,
      "step": 1294
    },
    {
      "epoch": 0.4373522458628842,
      "grad_norm": 1.9878791570663452,
      "learning_rate": 1.2014162534014839e-05,
      "loss": 0.8924,
      "step": 1295
    },
    {
      "epoch": 0.4376899696048632,
      "grad_norm": 2.2056660652160645,
      "learning_rate": 1.2003733715748052e-05,
      "loss": 0.946,
      "step": 1296
    },
    {
      "epoch": 0.4380276933468423,
      "grad_norm": 2.1312437057495117,
      "learning_rate": 1.1993302626564103e-05,
      "loss": 0.8995,
      "step": 1297
    },
    {
      "epoch": 0.4383654170888213,
      "grad_norm": 1.8763830661773682,
      "learning_rate": 1.1982869278284996e-05,
      "loss": 0.9721,
      "step": 1298
    },
    {
      "epoch": 0.4387031408308004,
      "grad_norm": 1.8853307962417603,
      "learning_rate": 1.1972433682735299e-05,
      "loss": 0.8882,
      "step": 1299
    },
    {
      "epoch": 0.43904086457277947,
      "grad_norm": 2.0810678005218506,
      "learning_rate": 1.196199585174211e-05,
      "loss": 0.9519,
      "step": 1300
    },
    {
      "epoch": 0.4393785883147585,
      "grad_norm": 2.0313189029693604,
      "learning_rate": 1.1951555797135073e-05,
      "loss": 0.9843,
      "step": 1301
    },
    {
      "epoch": 0.4397163120567376,
      "grad_norm": 2.1687819957733154,
      "learning_rate": 1.1941113530746344e-05,
      "loss": 0.9771,
      "step": 1302
    },
    {
      "epoch": 0.44005403579871666,
      "grad_norm": 2.061176061630249,
      "learning_rate": 1.1930669064410599e-05,
      "loss": 0.9124,
      "step": 1303
    },
    {
      "epoch": 0.4403917595406957,
      "grad_norm": 1.784262776374817,
      "learning_rate": 1.1920222409964992e-05,
      "loss": 0.8809,
      "step": 1304
    },
    {
      "epoch": 0.44072948328267475,
      "grad_norm": 2.0833351612091064,
      "learning_rate": 1.1909773579249166e-05,
      "loss": 0.9105,
      "step": 1305
    },
    {
      "epoch": 0.44106720702465385,
      "grad_norm": 2.0073049068450928,
      "learning_rate": 1.1899322584105225e-05,
      "loss": 0.8086,
      "step": 1306
    },
    {
      "epoch": 0.4414049307666329,
      "grad_norm": 2.2003703117370605,
      "learning_rate": 1.188886943637773e-05,
      "loss": 0.8247,
      "step": 1307
    },
    {
      "epoch": 0.44174265450861194,
      "grad_norm": 2.0165679454803467,
      "learning_rate": 1.1878414147913685e-05,
      "loss": 1.0411,
      "step": 1308
    },
    {
      "epoch": 0.44208037825059104,
      "grad_norm": 2.1784286499023438,
      "learning_rate": 1.186795673056251e-05,
      "loss": 0.91,
      "step": 1309
    },
    {
      "epoch": 0.4424181019925701,
      "grad_norm": 1.9215080738067627,
      "learning_rate": 1.1857497196176049e-05,
      "loss": 0.8561,
      "step": 1310
    },
    {
      "epoch": 0.44275582573454914,
      "grad_norm": 2.08202862739563,
      "learning_rate": 1.1847035556608532e-05,
      "loss": 0.8452,
      "step": 1311
    },
    {
      "epoch": 0.4430935494765282,
      "grad_norm": 1.9002983570098877,
      "learning_rate": 1.1836571823716594e-05,
      "loss": 0.9132,
      "step": 1312
    },
    {
      "epoch": 0.4434312732185073,
      "grad_norm": 2.2023849487304688,
      "learning_rate": 1.1826106009359221e-05,
      "loss": 0.85,
      "step": 1313
    },
    {
      "epoch": 0.44376899696048633,
      "grad_norm": 2.1278507709503174,
      "learning_rate": 1.1815638125397773e-05,
      "loss": 0.8989,
      "step": 1314
    },
    {
      "epoch": 0.4441067207024654,
      "grad_norm": 2.1192264556884766,
      "learning_rate": 1.1805168183695948e-05,
      "loss": 0.9039,
      "step": 1315
    },
    {
      "epoch": 0.4444444444444444,
      "grad_norm": 2.150742292404175,
      "learning_rate": 1.179469619611978e-05,
      "loss": 0.966,
      "step": 1316
    },
    {
      "epoch": 0.4447821681864235,
      "grad_norm": 1.9691756963729858,
      "learning_rate": 1.1784222174537627e-05,
      "loss": 0.9672,
      "step": 1317
    },
    {
      "epoch": 0.44511989192840257,
      "grad_norm": 1.7529911994934082,
      "learning_rate": 1.1773746130820137e-05,
      "loss": 0.9259,
      "step": 1318
    },
    {
      "epoch": 0.4454576156703816,
      "grad_norm": 1.8914545774459839,
      "learning_rate": 1.176326807684026e-05,
      "loss": 0.9457,
      "step": 1319
    },
    {
      "epoch": 0.4457953394123607,
      "grad_norm": 1.976816177368164,
      "learning_rate": 1.1752788024473227e-05,
      "loss": 0.9471,
      "step": 1320
    },
    {
      "epoch": 0.44613306315433976,
      "grad_norm": 2.179889678955078,
      "learning_rate": 1.1742305985596526e-05,
      "loss": 1.0221,
      "step": 1321
    },
    {
      "epoch": 0.4464707868963188,
      "grad_norm": 1.7898480892181396,
      "learning_rate": 1.1731821972089902e-05,
      "loss": 0.9861,
      "step": 1322
    },
    {
      "epoch": 0.44680851063829785,
      "grad_norm": 1.8143142461776733,
      "learning_rate": 1.1721335995835336e-05,
      "loss": 0.931,
      "step": 1323
    },
    {
      "epoch": 0.44714623438027695,
      "grad_norm": 2.0183262825012207,
      "learning_rate": 1.1710848068717033e-05,
      "loss": 0.906,
      "step": 1324
    },
    {
      "epoch": 0.447483958122256,
      "grad_norm": 1.9314905405044556,
      "learning_rate": 1.1700358202621412e-05,
      "loss": 0.8519,
      "step": 1325
    },
    {
      "epoch": 0.44782168186423504,
      "grad_norm": 1.79714035987854,
      "learning_rate": 1.1689866409437084e-05,
      "loss": 0.8746,
      "step": 1326
    },
    {
      "epoch": 0.4481594056062141,
      "grad_norm": 1.8785738945007324,
      "learning_rate": 1.1679372701054846e-05,
      "loss": 0.891,
      "step": 1327
    },
    {
      "epoch": 0.4484971293481932,
      "grad_norm": 1.9698909521102905,
      "learning_rate": 1.1668877089367668e-05,
      "loss": 0.947,
      "step": 1328
    },
    {
      "epoch": 0.44883485309017224,
      "grad_norm": 1.9419656991958618,
      "learning_rate": 1.1658379586270679e-05,
      "loss": 0.9278,
      "step": 1329
    },
    {
      "epoch": 0.4491725768321513,
      "grad_norm": 1.8932734727859497,
      "learning_rate": 1.1647880203661147e-05,
      "loss": 0.9076,
      "step": 1330
    },
    {
      "epoch": 0.4495103005741304,
      "grad_norm": 1.9739705324172974,
      "learning_rate": 1.1637378953438469e-05,
      "loss": 0.8946,
      "step": 1331
    },
    {
      "epoch": 0.44984802431610943,
      "grad_norm": 1.9140440225601196,
      "learning_rate": 1.162687584750416e-05,
      "loss": 0.9681,
      "step": 1332
    },
    {
      "epoch": 0.4501857480580885,
      "grad_norm": 1.9482139348983765,
      "learning_rate": 1.1616370897761846e-05,
      "loss": 0.912,
      "step": 1333
    },
    {
      "epoch": 0.4505234718000675,
      "grad_norm": 1.7124406099319458,
      "learning_rate": 1.1605864116117231e-05,
      "loss": 0.9546,
      "step": 1334
    },
    {
      "epoch": 0.4508611955420466,
      "grad_norm": 1.9819090366363525,
      "learning_rate": 1.1595355514478101e-05,
      "loss": 0.8915,
      "step": 1335
    },
    {
      "epoch": 0.45119891928402567,
      "grad_norm": 1.9874001741409302,
      "learning_rate": 1.1584845104754305e-05,
      "loss": 0.9365,
      "step": 1336
    },
    {
      "epoch": 0.4515366430260047,
      "grad_norm": 2.264918565750122,
      "learning_rate": 1.1574332898857738e-05,
      "loss": 0.7504,
      "step": 1337
    },
    {
      "epoch": 0.4518743667679838,
      "grad_norm": 1.8209718465805054,
      "learning_rate": 1.1563818908702338e-05,
      "loss": 0.9023,
      "step": 1338
    },
    {
      "epoch": 0.45221209050996286,
      "grad_norm": 2.164968490600586,
      "learning_rate": 1.1553303146204051e-05,
      "loss": 0.8107,
      "step": 1339
    },
    {
      "epoch": 0.4525498142519419,
      "grad_norm": 2.274294853210449,
      "learning_rate": 1.1542785623280847e-05,
      "loss": 0.9699,
      "step": 1340
    },
    {
      "epoch": 0.45288753799392095,
      "grad_norm": 2.0331029891967773,
      "learning_rate": 1.1532266351852685e-05,
      "loss": 0.7868,
      "step": 1341
    },
    {
      "epoch": 0.45322526173590005,
      "grad_norm": 2.0142533779144287,
      "learning_rate": 1.1521745343841502e-05,
      "loss": 0.9008,
      "step": 1342
    },
    {
      "epoch": 0.4535629854778791,
      "grad_norm": 1.9881572723388672,
      "learning_rate": 1.1511222611171212e-05,
      "loss": 0.9203,
      "step": 1343
    },
    {
      "epoch": 0.45390070921985815,
      "grad_norm": 2.25355863571167,
      "learning_rate": 1.150069816576767e-05,
      "loss": 0.8339,
      "step": 1344
    },
    {
      "epoch": 0.4542384329618372,
      "grad_norm": 2.178528070449829,
      "learning_rate": 1.1490172019558688e-05,
      "loss": 0.7892,
      "step": 1345
    },
    {
      "epoch": 0.4545761567038163,
      "grad_norm": 2.0951755046844482,
      "learning_rate": 1.147964418447399e-05,
      "loss": 0.7699,
      "step": 1346
    },
    {
      "epoch": 0.45491388044579534,
      "grad_norm": 1.9430228471755981,
      "learning_rate": 1.1469114672445232e-05,
      "loss": 1.0185,
      "step": 1347
    },
    {
      "epoch": 0.4552516041877744,
      "grad_norm": 2.1091206073760986,
      "learning_rate": 1.1458583495405953e-05,
      "loss": 0.9408,
      "step": 1348
    },
    {
      "epoch": 0.4555893279297535,
      "grad_norm": 1.8600050210952759,
      "learning_rate": 1.1448050665291587e-05,
      "loss": 0.9816,
      "step": 1349
    },
    {
      "epoch": 0.45592705167173253,
      "grad_norm": 1.91958749294281,
      "learning_rate": 1.1437516194039445e-05,
      "loss": 0.9396,
      "step": 1350
    },
    {
      "epoch": 0.4562647754137116,
      "grad_norm": 1.7879693508148193,
      "learning_rate": 1.1426980093588688e-05,
      "loss": 0.9652,
      "step": 1351
    },
    {
      "epoch": 0.4566024991556906,
      "grad_norm": 2.2525665760040283,
      "learning_rate": 1.141644237588034e-05,
      "loss": 0.9218,
      "step": 1352
    },
    {
      "epoch": 0.4569402228976697,
      "grad_norm": 2.252046585083008,
      "learning_rate": 1.1405903052857233e-05,
      "loss": 0.9493,
      "step": 1353
    },
    {
      "epoch": 0.45727794663964877,
      "grad_norm": 1.9324638843536377,
      "learning_rate": 1.1395362136464045e-05,
      "loss": 0.8943,
      "step": 1354
    },
    {
      "epoch": 0.4576156703816278,
      "grad_norm": 2.0035860538482666,
      "learning_rate": 1.1384819638647242e-05,
      "loss": 0.9201,
      "step": 1355
    },
    {
      "epoch": 0.4579533941236069,
      "grad_norm": 1.9937076568603516,
      "learning_rate": 1.137427557135509e-05,
      "loss": 0.8877,
      "step": 1356
    },
    {
      "epoch": 0.45829111786558596,
      "grad_norm": 2.2411949634552,
      "learning_rate": 1.1363729946537626e-05,
      "loss": 0.5921,
      "step": 1357
    },
    {
      "epoch": 0.458628841607565,
      "grad_norm": 2.0183629989624023,
      "learning_rate": 1.1353182776146665e-05,
      "loss": 0.9564,
      "step": 1358
    },
    {
      "epoch": 0.45896656534954405,
      "grad_norm": 2.107069253921509,
      "learning_rate": 1.1342634072135767e-05,
      "loss": 0.9095,
      "step": 1359
    },
    {
      "epoch": 0.45930428909152315,
      "grad_norm": 2.0457348823547363,
      "learning_rate": 1.1332083846460222e-05,
      "loss": 0.8769,
      "step": 1360
    },
    {
      "epoch": 0.4596420128335022,
      "grad_norm": 1.8746775388717651,
      "learning_rate": 1.132153211107706e-05,
      "loss": 0.8921,
      "step": 1361
    },
    {
      "epoch": 0.45997973657548125,
      "grad_norm": 1.869728922843933,
      "learning_rate": 1.1310978877945007e-05,
      "loss": 0.9616,
      "step": 1362
    },
    {
      "epoch": 0.4603174603174603,
      "grad_norm": 1.8770689964294434,
      "learning_rate": 1.1300424159024498e-05,
      "loss": 0.9759,
      "step": 1363
    },
    {
      "epoch": 0.4606551840594394,
      "grad_norm": 1.8766754865646362,
      "learning_rate": 1.128986796627765e-05,
      "loss": 0.9154,
      "step": 1364
    },
    {
      "epoch": 0.46099290780141844,
      "grad_norm": 1.85992431640625,
      "learning_rate": 1.127931031166824e-05,
      "loss": 0.9941,
      "step": 1365
    },
    {
      "epoch": 0.4613306315433975,
      "grad_norm": 1.717789649963379,
      "learning_rate": 1.1268751207161718e-05,
      "loss": 0.9363,
      "step": 1366
    },
    {
      "epoch": 0.4616683552853766,
      "grad_norm": 1.8278369903564453,
      "learning_rate": 1.1258190664725164e-05,
      "loss": 0.8747,
      "step": 1367
    },
    {
      "epoch": 0.46200607902735563,
      "grad_norm": 1.8778209686279297,
      "learning_rate": 1.1247628696327297e-05,
      "loss": 0.939,
      "step": 1368
    },
    {
      "epoch": 0.4623438027693347,
      "grad_norm": 2.2979867458343506,
      "learning_rate": 1.1237065313938442e-05,
      "loss": 0.9284,
      "step": 1369
    },
    {
      "epoch": 0.4626815265113137,
      "grad_norm": 2.042623996734619,
      "learning_rate": 1.1226500529530535e-05,
      "loss": 0.8814,
      "step": 1370
    },
    {
      "epoch": 0.4630192502532928,
      "grad_norm": 2.0781402587890625,
      "learning_rate": 1.1215934355077098e-05,
      "loss": 0.9196,
      "step": 1371
    },
    {
      "epoch": 0.46335697399527187,
      "grad_norm": 1.936281442642212,
      "learning_rate": 1.1205366802553231e-05,
      "loss": 0.9538,
      "step": 1372
    },
    {
      "epoch": 0.4636946977372509,
      "grad_norm": 1.7294816970825195,
      "learning_rate": 1.1194797883935593e-05,
      "loss": 0.8852,
      "step": 1373
    },
    {
      "epoch": 0.46403242147923,
      "grad_norm": 1.985040307044983,
      "learning_rate": 1.1184227611202386e-05,
      "loss": 0.8456,
      "step": 1374
    },
    {
      "epoch": 0.46437014522120906,
      "grad_norm": 1.9678064584732056,
      "learning_rate": 1.1173655996333356e-05,
      "loss": 0.9065,
      "step": 1375
    },
    {
      "epoch": 0.4647078689631881,
      "grad_norm": 2.090934991836548,
      "learning_rate": 1.1163083051309768e-05,
      "loss": 0.799,
      "step": 1376
    },
    {
      "epoch": 0.46504559270516715,
      "grad_norm": 2.3231401443481445,
      "learning_rate": 1.1152508788114395e-05,
      "loss": 0.78,
      "step": 1377
    },
    {
      "epoch": 0.46538331644714626,
      "grad_norm": 1.5755906105041504,
      "learning_rate": 1.1141933218731498e-05,
      "loss": 0.5633,
      "step": 1378
    },
    {
      "epoch": 0.4657210401891253,
      "grad_norm": 2.4686338901519775,
      "learning_rate": 1.1131356355146818e-05,
      "loss": 0.8248,
      "step": 1379
    },
    {
      "epoch": 0.46605876393110435,
      "grad_norm": 1.9969300031661987,
      "learning_rate": 1.1120778209347578e-05,
      "loss": 0.9907,
      "step": 1380
    },
    {
      "epoch": 0.4663964876730834,
      "grad_norm": 1.941463828086853,
      "learning_rate": 1.1110198793322433e-05,
      "loss": 0.9667,
      "step": 1381
    },
    {
      "epoch": 0.4667342114150625,
      "grad_norm": 1.9269962310791016,
      "learning_rate": 1.1099618119061498e-05,
      "loss": 0.9318,
      "step": 1382
    },
    {
      "epoch": 0.46707193515704154,
      "grad_norm": 2.2035114765167236,
      "learning_rate": 1.1089036198556292e-05,
      "loss": 0.8578,
      "step": 1383
    },
    {
      "epoch": 0.4674096588990206,
      "grad_norm": 1.813893437385559,
      "learning_rate": 1.1078453043799766e-05,
      "loss": 0.9939,
      "step": 1384
    },
    {
      "epoch": 0.4677473826409997,
      "grad_norm": 2.3865420818328857,
      "learning_rate": 1.1067868666786263e-05,
      "loss": 0.8424,
      "step": 1385
    },
    {
      "epoch": 0.46808510638297873,
      "grad_norm": 1.6896467208862305,
      "learning_rate": 1.1057283079511506e-05,
      "loss": 0.8912,
      "step": 1386
    },
    {
      "epoch": 0.4684228301249578,
      "grad_norm": 1.8420244455337524,
      "learning_rate": 1.1046696293972594e-05,
      "loss": 0.8285,
      "step": 1387
    },
    {
      "epoch": 0.4687605538669368,
      "grad_norm": 2.124197006225586,
      "learning_rate": 1.1036108322167988e-05,
      "loss": 0.9112,
      "step": 1388
    },
    {
      "epoch": 0.4690982776089159,
      "grad_norm": 1.6444261074066162,
      "learning_rate": 1.102551917609749e-05,
      "loss": 0.9357,
      "step": 1389
    },
    {
      "epoch": 0.46943600135089497,
      "grad_norm": 1.9190261363983154,
      "learning_rate": 1.1014928867762227e-05,
      "loss": 0.8942,
      "step": 1390
    },
    {
      "epoch": 0.469773725092874,
      "grad_norm": 1.992695927619934,
      "learning_rate": 1.100433740916465e-05,
      "loss": 0.8718,
      "step": 1391
    },
    {
      "epoch": 0.47011144883485306,
      "grad_norm": 2.3133132457733154,
      "learning_rate": 1.0993744812308516e-05,
      "loss": 0.8441,
      "step": 1392
    },
    {
      "epoch": 0.47044917257683216,
      "grad_norm": 1.8236665725708008,
      "learning_rate": 1.0983151089198867e-05,
      "loss": 0.9072,
      "step": 1393
    },
    {
      "epoch": 0.4707868963188112,
      "grad_norm": 1.864425539970398,
      "learning_rate": 1.097255625184202e-05,
      "loss": 0.8753,
      "step": 1394
    },
    {
      "epoch": 0.47112462006079026,
      "grad_norm": 1.9490926265716553,
      "learning_rate": 1.0961960312245557e-05,
      "loss": 0.9223,
      "step": 1395
    },
    {
      "epoch": 0.47146234380276936,
      "grad_norm": 2.0624613761901855,
      "learning_rate": 1.0951363282418312e-05,
      "loss": 0.8373,
      "step": 1396
    },
    {
      "epoch": 0.4718000675447484,
      "grad_norm": 2.5410656929016113,
      "learning_rate": 1.0940765174370353e-05,
      "loss": 0.8191,
      "step": 1397
    },
    {
      "epoch": 0.47213779128672745,
      "grad_norm": 2.1857717037200928,
      "learning_rate": 1.093016600011297e-05,
      "loss": 1.0242,
      "step": 1398
    },
    {
      "epoch": 0.4724755150287065,
      "grad_norm": 1.8861407041549683,
      "learning_rate": 1.0919565771658657e-05,
      "loss": 0.9714,
      "step": 1399
    },
    {
      "epoch": 0.4728132387706856,
      "grad_norm": 1.9982528686523438,
      "learning_rate": 1.0908964501021108e-05,
      "loss": 0.919,
      "step": 1400
    },
    {
      "epoch": 0.47315096251266464,
      "grad_norm": 2.063215732574463,
      "learning_rate": 1.0898362200215199e-05,
      "loss": 0.9846,
      "step": 1401
    },
    {
      "epoch": 0.4734886862546437,
      "grad_norm": 1.8089967966079712,
      "learning_rate": 1.0887758881256967e-05,
      "loss": 1.0112,
      "step": 1402
    },
    {
      "epoch": 0.4738264099966228,
      "grad_norm": 2.208112955093384,
      "learning_rate": 1.0877154556163614e-05,
      "loss": 0.9537,
      "step": 1403
    },
    {
      "epoch": 0.47416413373860183,
      "grad_norm": 1.9283421039581299,
      "learning_rate": 1.0866549236953468e-05,
      "loss": 0.871,
      "step": 1404
    },
    {
      "epoch": 0.4745018574805809,
      "grad_norm": 1.8044226169586182,
      "learning_rate": 1.085594293564599e-05,
      "loss": 0.9732,
      "step": 1405
    },
    {
      "epoch": 0.4748395812225599,
      "grad_norm": 1.731278657913208,
      "learning_rate": 1.0845335664261762e-05,
      "loss": 0.9539,
      "step": 1406
    },
    {
      "epoch": 0.475177304964539,
      "grad_norm": 1.9852811098098755,
      "learning_rate": 1.0834727434822458e-05,
      "loss": 0.9117,
      "step": 1407
    },
    {
      "epoch": 0.4755150287065181,
      "grad_norm": 1.9068949222564697,
      "learning_rate": 1.0824118259350834e-05,
      "loss": 0.9227,
      "step": 1408
    },
    {
      "epoch": 0.4758527524484971,
      "grad_norm": 2.0608766078948975,
      "learning_rate": 1.0813508149870722e-05,
      "loss": 0.9375,
      "step": 1409
    },
    {
      "epoch": 0.47619047619047616,
      "grad_norm": 2.132451295852661,
      "learning_rate": 1.0802897118407018e-05,
      "loss": 0.9269,
      "step": 1410
    },
    {
      "epoch": 0.47652819993245527,
      "grad_norm": 1.8073441982269287,
      "learning_rate": 1.0792285176985653e-05,
      "loss": 0.9135,
      "step": 1411
    },
    {
      "epoch": 0.4768659236744343,
      "grad_norm": 2.008596181869507,
      "learning_rate": 1.07816723376336e-05,
      "loss": 0.8795,
      "step": 1412
    },
    {
      "epoch": 0.47720364741641336,
      "grad_norm": 2.044818878173828,
      "learning_rate": 1.0771058612378837e-05,
      "loss": 0.9416,
      "step": 1413
    },
    {
      "epoch": 0.47754137115839246,
      "grad_norm": 1.97854483127594,
      "learning_rate": 1.076044401325036e-05,
      "loss": 0.9211,
      "step": 1414
    },
    {
      "epoch": 0.4778790949003715,
      "grad_norm": 2.307147264480591,
      "learning_rate": 1.0749828552278141e-05,
      "loss": 0.8738,
      "step": 1415
    },
    {
      "epoch": 0.47821681864235055,
      "grad_norm": 1.8890177011489868,
      "learning_rate": 1.0739212241493142e-05,
      "loss": 0.8821,
      "step": 1416
    },
    {
      "epoch": 0.4785545423843296,
      "grad_norm": 1.9954496622085571,
      "learning_rate": 1.0728595092927279e-05,
      "loss": 0.8509,
      "step": 1417
    },
    {
      "epoch": 0.4788922661263087,
      "grad_norm": 1.9925594329833984,
      "learning_rate": 1.0717977118613424e-05,
      "loss": 0.8663,
      "step": 1418
    },
    {
      "epoch": 0.47922998986828774,
      "grad_norm": 2.040476083755493,
      "learning_rate": 1.070735833058538e-05,
      "loss": 0.882,
      "step": 1419
    },
    {
      "epoch": 0.4795677136102668,
      "grad_norm": 2.19811749458313,
      "learning_rate": 1.0696738740877872e-05,
      "loss": 0.9275,
      "step": 1420
    },
    {
      "epoch": 0.4799054373522459,
      "grad_norm": 2.0865771770477295,
      "learning_rate": 1.0686118361526537e-05,
      "loss": 1.0142,
      "step": 1421
    },
    {
      "epoch": 0.48024316109422494,
      "grad_norm": 2.0237812995910645,
      "learning_rate": 1.0675497204567904e-05,
      "loss": 0.9529,
      "step": 1422
    },
    {
      "epoch": 0.480580884836204,
      "grad_norm": 2.1303398609161377,
      "learning_rate": 1.0664875282039386e-05,
      "loss": 0.9528,
      "step": 1423
    },
    {
      "epoch": 0.480918608578183,
      "grad_norm": 2.4577667713165283,
      "learning_rate": 1.0654252605979266e-05,
      "loss": 0.7443,
      "step": 1424
    },
    {
      "epoch": 0.48125633232016213,
      "grad_norm": 1.9244482517242432,
      "learning_rate": 1.0643629188426669e-05,
      "loss": 0.9972,
      "step": 1425
    },
    {
      "epoch": 0.4815940560621412,
      "grad_norm": 2.123914957046509,
      "learning_rate": 1.0633005041421573e-05,
      "loss": 0.9864,
      "step": 1426
    },
    {
      "epoch": 0.4819317798041202,
      "grad_norm": 1.7781258821487427,
      "learning_rate": 1.062238017700478e-05,
      "loss": 0.9651,
      "step": 1427
    },
    {
      "epoch": 0.48226950354609927,
      "grad_norm": 2.0538721084594727,
      "learning_rate": 1.0611754607217906e-05,
      "loss": 0.8766,
      "step": 1428
    },
    {
      "epoch": 0.48260722728807837,
      "grad_norm": 2.3012802600860596,
      "learning_rate": 1.0601128344103354e-05,
      "loss": 0.9408,
      "step": 1429
    },
    {
      "epoch": 0.4829449510300574,
      "grad_norm": 2.1147348880767822,
      "learning_rate": 1.0590501399704332e-05,
      "loss": 0.8904,
      "step": 1430
    },
    {
      "epoch": 0.48328267477203646,
      "grad_norm": 2.0623388290405273,
      "learning_rate": 1.0579873786064804e-05,
      "loss": 0.8835,
      "step": 1431
    },
    {
      "epoch": 0.48362039851401556,
      "grad_norm": 2.1415388584136963,
      "learning_rate": 1.0569245515229505e-05,
      "loss": 0.8407,
      "step": 1432
    },
    {
      "epoch": 0.4839581222559946,
      "grad_norm": 2.024411201477051,
      "learning_rate": 1.0558616599243907e-05,
      "loss": 0.9264,
      "step": 1433
    },
    {
      "epoch": 0.48429584599797365,
      "grad_norm": 1.7645560503005981,
      "learning_rate": 1.0547987050154214e-05,
      "loss": 0.9687,
      "step": 1434
    },
    {
      "epoch": 0.4846335697399527,
      "grad_norm": 1.8392291069030762,
      "learning_rate": 1.053735688000735e-05,
      "loss": 0.8924,
      "step": 1435
    },
    {
      "epoch": 0.4849712934819318,
      "grad_norm": 2.565171480178833,
      "learning_rate": 1.052672610085094e-05,
      "loss": 0.8486,
      "step": 1436
    },
    {
      "epoch": 0.48530901722391084,
      "grad_norm": 2.082021474838257,
      "learning_rate": 1.0516094724733305e-05,
      "loss": 0.9053,
      "step": 1437
    },
    {
      "epoch": 0.4856467409658899,
      "grad_norm": 2.0885109901428223,
      "learning_rate": 1.0505462763703434e-05,
      "loss": 0.8817,
      "step": 1438
    },
    {
      "epoch": 0.485984464707869,
      "grad_norm": 2.1836016178131104,
      "learning_rate": 1.049483022981098e-05,
      "loss": 0.8546,
      "step": 1439
    },
    {
      "epoch": 0.48632218844984804,
      "grad_norm": 1.8616808652877808,
      "learning_rate": 1.0484197135106265e-05,
      "loss": 0.9866,
      "step": 1440
    },
    {
      "epoch": 0.4866599121918271,
      "grad_norm": 2.181602716445923,
      "learning_rate": 1.0473563491640211e-05,
      "loss": 0.8033,
      "step": 1441
    },
    {
      "epoch": 0.48699763593380613,
      "grad_norm": 2.450058698654175,
      "learning_rate": 1.0462929311464393e-05,
      "loss": 0.8076,
      "step": 1442
    },
    {
      "epoch": 0.48733535967578523,
      "grad_norm": 2.2067461013793945,
      "learning_rate": 1.0452294606630975e-05,
      "loss": 0.9615,
      "step": 1443
    },
    {
      "epoch": 0.4876730834177643,
      "grad_norm": 2.20280122756958,
      "learning_rate": 1.0441659389192728e-05,
      "loss": 0.9241,
      "step": 1444
    },
    {
      "epoch": 0.4880108071597433,
      "grad_norm": 2.240480422973633,
      "learning_rate": 1.0431023671202997e-05,
      "loss": 0.9471,
      "step": 1445
    },
    {
      "epoch": 0.48834853090172237,
      "grad_norm": 2.230299949645996,
      "learning_rate": 1.0420387464715698e-05,
      "loss": 0.915,
      "step": 1446
    },
    {
      "epoch": 0.48868625464370147,
      "grad_norm": 1.975317120552063,
      "learning_rate": 1.0409750781785298e-05,
      "loss": 0.9326,
      "step": 1447
    },
    {
      "epoch": 0.4890239783856805,
      "grad_norm": 2.232386350631714,
      "learning_rate": 1.0399113634466803e-05,
      "loss": 1.0094,
      "step": 1448
    },
    {
      "epoch": 0.48936170212765956,
      "grad_norm": 2.6009745597839355,
      "learning_rate": 1.038847603481575e-05,
      "loss": 0.8268,
      "step": 1449
    },
    {
      "epoch": 0.48969942586963866,
      "grad_norm": 2.101100444793701,
      "learning_rate": 1.0377837994888183e-05,
      "loss": 0.9102,
      "step": 1450
    },
    {
      "epoch": 0.4900371496116177,
      "grad_norm": 1.9767736196517944,
      "learning_rate": 1.036719952674065e-05,
      "loss": 0.9757,
      "step": 1451
    },
    {
      "epoch": 0.49037487335359675,
      "grad_norm": 2.022456645965576,
      "learning_rate": 1.035656064243018e-05,
      "loss": 0.9228,
      "step": 1452
    },
    {
      "epoch": 0.4907125970955758,
      "grad_norm": 2.277435541152954,
      "learning_rate": 1.0345921354014279e-05,
      "loss": 0.8845,
      "step": 1453
    },
    {
      "epoch": 0.4910503208375549,
      "grad_norm": 1.9421972036361694,
      "learning_rate": 1.033528167355091e-05,
      "loss": 0.8761,
      "step": 1454
    },
    {
      "epoch": 0.49138804457953394,
      "grad_norm": 2.136847496032715,
      "learning_rate": 1.032464161309847e-05,
      "loss": 0.9518,
      "step": 1455
    },
    {
      "epoch": 0.491725768321513,
      "grad_norm": 1.7513453960418701,
      "learning_rate": 1.0314001184715798e-05,
      "loss": 0.9703,
      "step": 1456
    },
    {
      "epoch": 0.49206349206349204,
      "grad_norm": 2.0479624271392822,
      "learning_rate": 1.030336040046215e-05,
      "loss": 0.9816,
      "step": 1457
    },
    {
      "epoch": 0.49240121580547114,
      "grad_norm": 2.0126564502716064,
      "learning_rate": 1.0292719272397182e-05,
      "loss": 0.9312,
      "step": 1458
    },
    {
      "epoch": 0.4927389395474502,
      "grad_norm": 1.8984626531600952,
      "learning_rate": 1.0282077812580936e-05,
      "loss": 0.9475,
      "step": 1459
    },
    {
      "epoch": 0.49307666328942923,
      "grad_norm": 2.190065383911133,
      "learning_rate": 1.0271436033073834e-05,
      "loss": 0.9447,
      "step": 1460
    },
    {
      "epoch": 0.49341438703140833,
      "grad_norm": 2.63460373878479,
      "learning_rate": 1.026079394593666e-05,
      "loss": 0.8908,
      "step": 1461
    },
    {
      "epoch": 0.4937521107733874,
      "grad_norm": 1.8003696203231812,
      "learning_rate": 1.025015156323055e-05,
      "loss": 0.9541,
      "step": 1462
    },
    {
      "epoch": 0.4940898345153664,
      "grad_norm": 2.0622365474700928,
      "learning_rate": 1.0239508897016974e-05,
      "loss": 0.891,
      "step": 1463
    },
    {
      "epoch": 0.49442755825734547,
      "grad_norm": 1.9204559326171875,
      "learning_rate": 1.0228865959357712e-05,
      "loss": 0.9384,
      "step": 1464
    },
    {
      "epoch": 0.49476528199932457,
      "grad_norm": 2.0527544021606445,
      "learning_rate": 1.0218222762314867e-05,
      "loss": 0.8966,
      "step": 1465
    },
    {
      "epoch": 0.4951030057413036,
      "grad_norm": 1.9934475421905518,
      "learning_rate": 1.0207579317950826e-05,
      "loss": 0.9178,
      "step": 1466
    },
    {
      "epoch": 0.49544072948328266,
      "grad_norm": 2.3906826972961426,
      "learning_rate": 1.0196935638328267e-05,
      "loss": 0.8649,
      "step": 1467
    },
    {
      "epoch": 0.49577845322526176,
      "grad_norm": 2.249781608581543,
      "learning_rate": 1.018629173551012e-05,
      "loss": 0.8615,
      "step": 1468
    },
    {
      "epoch": 0.4961161769672408,
      "grad_norm": 2.2300660610198975,
      "learning_rate": 1.0175647621559575e-05,
      "loss": 0.8643,
      "step": 1469
    },
    {
      "epoch": 0.49645390070921985,
      "grad_norm": 1.8089981079101562,
      "learning_rate": 1.0165003308540069e-05,
      "loss": 0.9733,
      "step": 1470
    },
    {
      "epoch": 0.4967916244511989,
      "grad_norm": 2.226328134536743,
      "learning_rate": 1.015435880851525e-05,
      "loss": 0.8639,
      "step": 1471
    },
    {
      "epoch": 0.497129348193178,
      "grad_norm": 2.4592671394348145,
      "learning_rate": 1.0143714133548992e-05,
      "loss": 0.8724,
      "step": 1472
    },
    {
      "epoch": 0.49746707193515705,
      "grad_norm": 2.418508768081665,
      "learning_rate": 1.0133069295705353e-05,
      "loss": 0.9318,
      "step": 1473
    },
    {
      "epoch": 0.4978047956771361,
      "grad_norm": 1.9920225143432617,
      "learning_rate": 1.0122424307048593e-05,
      "loss": 0.8945,
      "step": 1474
    },
    {
      "epoch": 0.49814251941911514,
      "grad_norm": 1.9483217000961304,
      "learning_rate": 1.0111779179643124e-05,
      "loss": 0.9134,
      "step": 1475
    },
    {
      "epoch": 0.49848024316109424,
      "grad_norm": 2.5371334552764893,
      "learning_rate": 1.0101133925553527e-05,
      "loss": 0.7877,
      "step": 1476
    },
    {
      "epoch": 0.4988179669030733,
      "grad_norm": 1.8763883113861084,
      "learning_rate": 1.0090488556844526e-05,
      "loss": 0.9187,
      "step": 1477
    },
    {
      "epoch": 0.49915569064505233,
      "grad_norm": 2.427953004837036,
      "learning_rate": 1.0079843085580972e-05,
      "loss": 0.8706,
      "step": 1478
    },
    {
      "epoch": 0.49949341438703143,
      "grad_norm": 1.7668119668960571,
      "learning_rate": 1.0069197523827835e-05,
      "loss": 0.9936,
      "step": 1479
    },
    {
      "epoch": 0.4998311381290105,
      "grad_norm": 2.088651180267334,
      "learning_rate": 1.005855188365018e-05,
      "loss": 0.9423,
      "step": 1480
    },
    {
      "epoch": 0.5001688618709895,
      "grad_norm": 1.9717845916748047,
      "learning_rate": 1.0047906177113172e-05,
      "loss": 1.0039,
      "step": 1481
    },
    {
      "epoch": 0.5005065856129686,
      "grad_norm": 2.5525548458099365,
      "learning_rate": 1.0037260416282042e-05,
      "loss": 0.9459,
      "step": 1482
    },
    {
      "epoch": 0.5005065856129686,
      "eval_loss": 0.8175700902938843,
      "eval_runtime": 348.1718,
      "eval_samples_per_second": 8.536,
      "eval_steps_per_second": 1.068,
      "step": 1482
    },
    {
      "epoch": 0.5008443093549476,
      "grad_norm": 2.0213067531585693,
      "learning_rate": 1.0026614613222087e-05,
      "loss": 0.8827,
      "step": 1483
    },
    {
      "epoch": 0.5011820330969267,
      "grad_norm": 1.9109587669372559,
      "learning_rate": 1.0015968779998654e-05,
      "loss": 0.9446,
      "step": 1484
    },
    {
      "epoch": 0.5015197568389058,
      "grad_norm": 1.8046602010726929,
      "learning_rate": 1.000532292867712e-05,
      "loss": 0.9352,
      "step": 1485
    },
    {
      "epoch": 0.5018574805808849,
      "grad_norm": 2.33417010307312,
      "learning_rate": 9.994677071322883e-06,
      "loss": 0.9345,
      "step": 1486
    },
    {
      "epoch": 0.5021952043228639,
      "grad_norm": 2.217038631439209,
      "learning_rate": 9.984031220001349e-06,
      "loss": 0.9533,
      "step": 1487
    },
    {
      "epoch": 0.502532928064843,
      "grad_norm": 2.0383055210113525,
      "learning_rate": 9.973385386777913e-06,
      "loss": 1.0034,
      "step": 1488
    },
    {
      "epoch": 0.502870651806822,
      "grad_norm": 1.8766435384750366,
      "learning_rate": 9.962739583717962e-06,
      "loss": 0.9843,
      "step": 1489
    },
    {
      "epoch": 0.503208375548801,
      "grad_norm": 2.241734027862549,
      "learning_rate": 9.952093822886833e-06,
      "loss": 0.9067,
      "step": 1490
    },
    {
      "epoch": 0.5035460992907801,
      "grad_norm": 2.3242456912994385,
      "learning_rate": 9.94144811634982e-06,
      "loss": 0.937,
      "step": 1491
    },
    {
      "epoch": 0.5038838230327592,
      "grad_norm": 2.147385835647583,
      "learning_rate": 9.930802476172169e-06,
      "loss": 0.9984,
      "step": 1492
    },
    {
      "epoch": 0.5042215467747383,
      "grad_norm": 1.8983402252197266,
      "learning_rate": 9.920156914419031e-06,
      "loss": 0.9046,
      "step": 1493
    },
    {
      "epoch": 0.5045592705167173,
      "grad_norm": 2.199707508087158,
      "learning_rate": 9.909511443155477e-06,
      "loss": 0.9172,
      "step": 1494
    },
    {
      "epoch": 0.5048969942586964,
      "grad_norm": 2.165229082107544,
      "learning_rate": 9.898866074446473e-06,
      "loss": 0.9347,
      "step": 1495
    },
    {
      "epoch": 0.5052347180006754,
      "grad_norm": 2.0696499347686768,
      "learning_rate": 9.888220820356879e-06,
      "loss": 0.9741,
      "step": 1496
    },
    {
      "epoch": 0.5055724417426545,
      "grad_norm": 1.7935636043548584,
      "learning_rate": 9.877575692951412e-06,
      "loss": 0.8422,
      "step": 1497
    },
    {
      "epoch": 0.5059101654846335,
      "grad_norm": 2.173367977142334,
      "learning_rate": 9.866930704294647e-06,
      "loss": 0.8958,
      "step": 1498
    },
    {
      "epoch": 0.5062478892266127,
      "grad_norm": 2.18229603767395,
      "learning_rate": 9.85628586645101e-06,
      "loss": 0.8764,
      "step": 1499
    },
    {
      "epoch": 0.5065856129685917,
      "grad_norm": 1.9013248682022095,
      "learning_rate": 9.845641191484752e-06,
      "loss": 0.8896,
      "step": 1500
    },
    {
      "epoch": 0.5069233367105708,
      "grad_norm": 1.730664610862732,
      "learning_rate": 9.834996691459938e-06,
      "loss": 0.9446,
      "step": 1501
    },
    {
      "epoch": 0.5072610604525498,
      "grad_norm": 1.957341194152832,
      "learning_rate": 9.824352378440427e-06,
      "loss": 0.8892,
      "step": 1502
    },
    {
      "epoch": 0.5075987841945289,
      "grad_norm": 2.1916897296905518,
      "learning_rate": 9.813708264489884e-06,
      "loss": 0.9016,
      "step": 1503
    },
    {
      "epoch": 0.5079365079365079,
      "grad_norm": 2.1283984184265137,
      "learning_rate": 9.803064361671738e-06,
      "loss": 0.9363,
      "step": 1504
    },
    {
      "epoch": 0.508274231678487,
      "grad_norm": 1.9124133586883545,
      "learning_rate": 9.792420682049174e-06,
      "loss": 0.9665,
      "step": 1505
    },
    {
      "epoch": 0.5086119554204661,
      "grad_norm": 1.9941717386245728,
      "learning_rate": 9.781777237685137e-06,
      "loss": 0.9343,
      "step": 1506
    },
    {
      "epoch": 0.5089496791624452,
      "grad_norm": 1.8717682361602783,
      "learning_rate": 9.771134040642291e-06,
      "loss": 0.9696,
      "step": 1507
    },
    {
      "epoch": 0.5092874029044242,
      "grad_norm": 2.200096368789673,
      "learning_rate": 9.760491102983033e-06,
      "loss": 0.8594,
      "step": 1508
    },
    {
      "epoch": 0.5096251266464032,
      "grad_norm": 2.4973738193511963,
      "learning_rate": 9.74984843676945e-06,
      "loss": 0.9566,
      "step": 1509
    },
    {
      "epoch": 0.5099628503883823,
      "grad_norm": 2.1460494995117188,
      "learning_rate": 9.739206054063341e-06,
      "loss": 0.8733,
      "step": 1510
    },
    {
      "epoch": 0.5103005741303613,
      "grad_norm": 2.0771334171295166,
      "learning_rate": 9.72856396692617e-06,
      "loss": 0.8856,
      "step": 1511
    },
    {
      "epoch": 0.5106382978723404,
      "grad_norm": 2.209477663040161,
      "learning_rate": 9.717922187419068e-06,
      "loss": 0.8868,
      "step": 1512
    },
    {
      "epoch": 0.5109760216143195,
      "grad_norm": 2.293940544128418,
      "learning_rate": 9.707280727602821e-06,
      "loss": 0.9764,
      "step": 1513
    },
    {
      "epoch": 0.5113137453562986,
      "grad_norm": 2.3166861534118652,
      "learning_rate": 9.696639599537853e-06,
      "loss": 0.8414,
      "step": 1514
    },
    {
      "epoch": 0.5116514690982776,
      "grad_norm": 2.0471506118774414,
      "learning_rate": 9.685998815284204e-06,
      "loss": 0.9605,
      "step": 1515
    },
    {
      "epoch": 0.5119891928402567,
      "grad_norm": 2.108478546142578,
      "learning_rate": 9.675358386901534e-06,
      "loss": 0.8987,
      "step": 1516
    },
    {
      "epoch": 0.5123269165822357,
      "grad_norm": 2.0791547298431396,
      "learning_rate": 9.664718326449093e-06,
      "loss": 0.9491,
      "step": 1517
    },
    {
      "epoch": 0.5126646403242148,
      "grad_norm": 1.9483215808868408,
      "learning_rate": 9.654078645985723e-06,
      "loss": 0.9877,
      "step": 1518
    },
    {
      "epoch": 0.5130023640661938,
      "grad_norm": 2.6539015769958496,
      "learning_rate": 9.643439357569819e-06,
      "loss": 0.8742,
      "step": 1519
    },
    {
      "epoch": 0.5133400878081729,
      "grad_norm": 2.0702359676361084,
      "learning_rate": 9.632800473259355e-06,
      "loss": 0.8976,
      "step": 1520
    },
    {
      "epoch": 0.513677811550152,
      "grad_norm": 2.258368968963623,
      "learning_rate": 9.62216200511182e-06,
      "loss": 0.9146,
      "step": 1521
    },
    {
      "epoch": 0.5140155352921311,
      "grad_norm": 1.8676854372024536,
      "learning_rate": 9.611523965184256e-06,
      "loss": 0.9513,
      "step": 1522
    },
    {
      "epoch": 0.5143532590341101,
      "grad_norm": 2.0028960704803467,
      "learning_rate": 9.600886365533199e-06,
      "loss": 0.8337,
      "step": 1523
    },
    {
      "epoch": 0.5146909827760892,
      "grad_norm": 2.3330938816070557,
      "learning_rate": 9.590249218214707e-06,
      "loss": 0.9294,
      "step": 1524
    },
    {
      "epoch": 0.5150287065180682,
      "grad_norm": 2.347144842147827,
      "learning_rate": 9.579612535284306e-06,
      "loss": 0.9342,
      "step": 1525
    },
    {
      "epoch": 0.5153664302600472,
      "grad_norm": 2.160086154937744,
      "learning_rate": 9.568976328797003e-06,
      "loss": 0.8036,
      "step": 1526
    },
    {
      "epoch": 0.5157041540020263,
      "grad_norm": 1.9515527486801147,
      "learning_rate": 9.558340610807273e-06,
      "loss": 1.0133,
      "step": 1527
    },
    {
      "epoch": 0.5160418777440055,
      "grad_norm": 2.0629584789276123,
      "learning_rate": 9.547705393369028e-06,
      "loss": 0.9113,
      "step": 1528
    },
    {
      "epoch": 0.5163796014859845,
      "grad_norm": 1.9612088203430176,
      "learning_rate": 9.537070688535612e-06,
      "loss": 0.9572,
      "step": 1529
    },
    {
      "epoch": 0.5167173252279635,
      "grad_norm": 2.366767644882202,
      "learning_rate": 9.52643650835979e-06,
      "loss": 0.8173,
      "step": 1530
    },
    {
      "epoch": 0.5170550489699426,
      "grad_norm": 2.1534788608551025,
      "learning_rate": 9.51580286489374e-06,
      "loss": 0.9327,
      "step": 1531
    },
    {
      "epoch": 0.5173927727119216,
      "grad_norm": 1.8814557790756226,
      "learning_rate": 9.50516977018902e-06,
      "loss": 0.9724,
      "step": 1532
    },
    {
      "epoch": 0.5177304964539007,
      "grad_norm": 1.5354766845703125,
      "learning_rate": 9.494537236296571e-06,
      "loss": 0.6706,
      "step": 1533
    },
    {
      "epoch": 0.5180682201958797,
      "grad_norm": 1.8877885341644287,
      "learning_rate": 9.483905275266698e-06,
      "loss": 0.8544,
      "step": 1534
    },
    {
      "epoch": 0.5184059439378589,
      "grad_norm": 1.7796337604522705,
      "learning_rate": 9.473273899149064e-06,
      "loss": 0.9313,
      "step": 1535
    },
    {
      "epoch": 0.5187436676798379,
      "grad_norm": 1.827683687210083,
      "learning_rate": 9.462643119992657e-06,
      "loss": 0.9464,
      "step": 1536
    },
    {
      "epoch": 0.519081391421817,
      "grad_norm": 2.18498158454895,
      "learning_rate": 9.45201294984579e-06,
      "loss": 0.9218,
      "step": 1537
    },
    {
      "epoch": 0.519419115163796,
      "grad_norm": 2.1132256984710693,
      "learning_rate": 9.441383400756096e-06,
      "loss": 0.9311,
      "step": 1538
    },
    {
      "epoch": 0.5197568389057751,
      "grad_norm": 2.269808292388916,
      "learning_rate": 9.430754484770498e-06,
      "loss": 0.9054,
      "step": 1539
    },
    {
      "epoch": 0.5200945626477541,
      "grad_norm": 1.7456409931182861,
      "learning_rate": 9.4201262139352e-06,
      "loss": 0.9813,
      "step": 1540
    },
    {
      "epoch": 0.5204322863897332,
      "grad_norm": 2.3699538707733154,
      "learning_rate": 9.409498600295672e-06,
      "loss": 0.8969,
      "step": 1541
    },
    {
      "epoch": 0.5207700101317123,
      "grad_norm": 2.5312106609344482,
      "learning_rate": 9.398871655896647e-06,
      "loss": 0.8545,
      "step": 1542
    },
    {
      "epoch": 0.5211077338736914,
      "grad_norm": 1.940110206604004,
      "learning_rate": 9.388245392782099e-06,
      "loss": 0.8301,
      "step": 1543
    },
    {
      "epoch": 0.5214454576156704,
      "grad_norm": 1.9892146587371826,
      "learning_rate": 9.37761982299522e-06,
      "loss": 0.9477,
      "step": 1544
    },
    {
      "epoch": 0.5217831813576495,
      "grad_norm": 2.048942804336548,
      "learning_rate": 9.366994958578429e-06,
      "loss": 0.8706,
      "step": 1545
    },
    {
      "epoch": 0.5221209050996285,
      "grad_norm": 2.390434503555298,
      "learning_rate": 9.356370811573333e-06,
      "loss": 0.9602,
      "step": 1546
    },
    {
      "epoch": 0.5224586288416075,
      "grad_norm": 2.873443126678467,
      "learning_rate": 9.34574739402074e-06,
      "loss": 0.9026,
      "step": 1547
    },
    {
      "epoch": 0.5227963525835866,
      "grad_norm": 2.1880173683166504,
      "learning_rate": 9.335124717960614e-06,
      "loss": 0.8287,
      "step": 1548
    },
    {
      "epoch": 0.5231340763255656,
      "grad_norm": 1.9307271242141724,
      "learning_rate": 9.324502795432099e-06,
      "loss": 0.8967,
      "step": 1549
    },
    {
      "epoch": 0.5234718000675448,
      "grad_norm": 2.307063341140747,
      "learning_rate": 9.313881638473468e-06,
      "loss": 0.871,
      "step": 1550
    },
    {
      "epoch": 0.5238095238095238,
      "grad_norm": 2.550196886062622,
      "learning_rate": 9.30326125912213e-06,
      "loss": 0.9261,
      "step": 1551
    },
    {
      "epoch": 0.5241472475515029,
      "grad_norm": 2.270210027694702,
      "learning_rate": 9.292641669414624e-06,
      "loss": 0.9259,
      "step": 1552
    },
    {
      "epoch": 0.5244849712934819,
      "grad_norm": 2.077427864074707,
      "learning_rate": 9.28202288138658e-06,
      "loss": 0.9417,
      "step": 1553
    },
    {
      "epoch": 0.524822695035461,
      "grad_norm": 2.234992504119873,
      "learning_rate": 9.271404907072723e-06,
      "loss": 0.9432,
      "step": 1554
    },
    {
      "epoch": 0.52516041877744,
      "grad_norm": 2.2260167598724365,
      "learning_rate": 9.260787758506858e-06,
      "loss": 0.8012,
      "step": 1555
    },
    {
      "epoch": 0.5254981425194191,
      "grad_norm": 1.967956781387329,
      "learning_rate": 9.25017144772186e-06,
      "loss": 0.9797,
      "step": 1556
    },
    {
      "epoch": 0.5258358662613982,
      "grad_norm": 2.0938894748687744,
      "learning_rate": 9.239555986749645e-06,
      "loss": 0.8434,
      "step": 1557
    },
    {
      "epoch": 0.5261735900033773,
      "grad_norm": 2.1352832317352295,
      "learning_rate": 9.228941387621165e-06,
      "loss": 0.9059,
      "step": 1558
    },
    {
      "epoch": 0.5265113137453563,
      "grad_norm": 1.8967257738113403,
      "learning_rate": 9.218327662366402e-06,
      "loss": 0.9464,
      "step": 1559
    },
    {
      "epoch": 0.5268490374873354,
      "grad_norm": 2.008976459503174,
      "learning_rate": 9.207714823014349e-06,
      "loss": 0.88,
      "step": 1560
    },
    {
      "epoch": 0.5271867612293144,
      "grad_norm": 2.136796712875366,
      "learning_rate": 9.197102881592987e-06,
      "loss": 0.9452,
      "step": 1561
    },
    {
      "epoch": 0.5275244849712935,
      "grad_norm": 1.8848307132720947,
      "learning_rate": 9.18649185012928e-06,
      "loss": 0.9781,
      "step": 1562
    },
    {
      "epoch": 0.5278622087132725,
      "grad_norm": 2.117826223373413,
      "learning_rate": 9.175881740649168e-06,
      "loss": 0.8181,
      "step": 1563
    },
    {
      "epoch": 0.5281999324552517,
      "grad_norm": 2.369727849960327,
      "learning_rate": 9.165272565177547e-06,
      "loss": 0.909,
      "step": 1564
    },
    {
      "epoch": 0.5285376561972307,
      "grad_norm": 1.7786369323730469,
      "learning_rate": 9.154664335738236e-06,
      "loss": 0.9509,
      "step": 1565
    },
    {
      "epoch": 0.5288753799392097,
      "grad_norm": 2.1855576038360596,
      "learning_rate": 9.144057064354011e-06,
      "loss": 0.921,
      "step": 1566
    },
    {
      "epoch": 0.5292131036811888,
      "grad_norm": 2.042402982711792,
      "learning_rate": 9.133450763046537e-06,
      "loss": 0.9028,
      "step": 1567
    },
    {
      "epoch": 0.5295508274231678,
      "grad_norm": 2.0760905742645264,
      "learning_rate": 9.122845443836392e-06,
      "loss": 0.9545,
      "step": 1568
    },
    {
      "epoch": 0.5298885511651469,
      "grad_norm": 2.046173334121704,
      "learning_rate": 9.112241118743034e-06,
      "loss": 0.8862,
      "step": 1569
    },
    {
      "epoch": 0.5302262749071259,
      "grad_norm": 1.9728723764419556,
      "learning_rate": 9.101637799784805e-06,
      "loss": 0.9076,
      "step": 1570
    },
    {
      "epoch": 0.5305639986491051,
      "grad_norm": 2.5120463371276855,
      "learning_rate": 9.091035498978894e-06,
      "loss": 0.7446,
      "step": 1571
    },
    {
      "epoch": 0.5309017223910841,
      "grad_norm": 1.9758186340332031,
      "learning_rate": 9.080434228341345e-06,
      "loss": 0.9405,
      "step": 1572
    },
    {
      "epoch": 0.5312394461330632,
      "grad_norm": 2.1183483600616455,
      "learning_rate": 9.069833999887033e-06,
      "loss": 0.869,
      "step": 1573
    },
    {
      "epoch": 0.5315771698750422,
      "grad_norm": 2.072654962539673,
      "learning_rate": 9.05923482562965e-06,
      "loss": 0.9616,
      "step": 1574
    },
    {
      "epoch": 0.5319148936170213,
      "grad_norm": 2.318828582763672,
      "learning_rate": 9.048636717581692e-06,
      "loss": 0.8465,
      "step": 1575
    },
    {
      "epoch": 0.5322526173590003,
      "grad_norm": 1.991982340812683,
      "learning_rate": 9.038039687754445e-06,
      "loss": 0.9056,
      "step": 1576
    },
    {
      "epoch": 0.5325903411009794,
      "grad_norm": 2.0290725231170654,
      "learning_rate": 9.027443748157985e-06,
      "loss": 0.9292,
      "step": 1577
    },
    {
      "epoch": 0.5329280648429585,
      "grad_norm": 2.187969923019409,
      "learning_rate": 9.01684891080114e-06,
      "loss": 0.9371,
      "step": 1578
    },
    {
      "epoch": 0.5332657885849376,
      "grad_norm": 2.387437343597412,
      "learning_rate": 9.006255187691485e-06,
      "loss": 0.6733,
      "step": 1579
    },
    {
      "epoch": 0.5336035123269166,
      "grad_norm": 2.3176658153533936,
      "learning_rate": 8.995662590835353e-06,
      "loss": 0.7633,
      "step": 1580
    },
    {
      "epoch": 0.5339412360688957,
      "grad_norm": 2.064039707183838,
      "learning_rate": 8.985071132237776e-06,
      "loss": 0.7966,
      "step": 1581
    },
    {
      "epoch": 0.5342789598108747,
      "grad_norm": 1.9774638414382935,
      "learning_rate": 8.974480823902514e-06,
      "loss": 0.9097,
      "step": 1582
    },
    {
      "epoch": 0.5346166835528537,
      "grad_norm": 1.9442496299743652,
      "learning_rate": 8.963891677832012e-06,
      "loss": 0.9923,
      "step": 1583
    },
    {
      "epoch": 0.5349544072948328,
      "grad_norm": 2.2192673683166504,
      "learning_rate": 8.953303706027407e-06,
      "loss": 0.938,
      "step": 1584
    },
    {
      "epoch": 0.5352921310368118,
      "grad_norm": 2.0519487857818604,
      "learning_rate": 8.942716920488499e-06,
      "loss": 0.8868,
      "step": 1585
    },
    {
      "epoch": 0.535629854778791,
      "grad_norm": 2.2097697257995605,
      "learning_rate": 8.932131333213739e-06,
      "loss": 0.9899,
      "step": 1586
    },
    {
      "epoch": 0.53596757852077,
      "grad_norm": 2.6935625076293945,
      "learning_rate": 8.921546956200236e-06,
      "loss": 0.732,
      "step": 1587
    },
    {
      "epoch": 0.5363053022627491,
      "grad_norm": 2.0861599445343018,
      "learning_rate": 8.910963801443713e-06,
      "loss": 0.8695,
      "step": 1588
    },
    {
      "epoch": 0.5366430260047281,
      "grad_norm": 1.9669183492660522,
      "learning_rate": 8.900381880938507e-06,
      "loss": 0.9088,
      "step": 1589
    },
    {
      "epoch": 0.5369807497467072,
      "grad_norm": 2.4032747745513916,
      "learning_rate": 8.889801206677566e-06,
      "loss": 0.9123,
      "step": 1590
    },
    {
      "epoch": 0.5373184734886862,
      "grad_norm": 2.0466840267181396,
      "learning_rate": 8.879221790652426e-06,
      "loss": 1.0115,
      "step": 1591
    },
    {
      "epoch": 0.5376561972306653,
      "grad_norm": 2.250793695449829,
      "learning_rate": 8.868643644853185e-06,
      "loss": 0.9538,
      "step": 1592
    },
    {
      "epoch": 0.5379939209726444,
      "grad_norm": 2.3968069553375244,
      "learning_rate": 8.858066781268506e-06,
      "loss": 0.7895,
      "step": 1593
    },
    {
      "epoch": 0.5383316447146235,
      "grad_norm": 2.0200259685516357,
      "learning_rate": 8.847491211885607e-06,
      "loss": 0.889,
      "step": 1594
    },
    {
      "epoch": 0.5386693684566025,
      "grad_norm": 1.8661391735076904,
      "learning_rate": 8.836916948690234e-06,
      "loss": 0.8862,
      "step": 1595
    },
    {
      "epoch": 0.5390070921985816,
      "grad_norm": 1.995679497718811,
      "learning_rate": 8.826344003666647e-06,
      "loss": 1.0001,
      "step": 1596
    },
    {
      "epoch": 0.5393448159405606,
      "grad_norm": 2.0383059978485107,
      "learning_rate": 8.815772388797616e-06,
      "loss": 0.928,
      "step": 1597
    },
    {
      "epoch": 0.5396825396825397,
      "grad_norm": 2.0352134704589844,
      "learning_rate": 8.80520211606441e-06,
      "loss": 0.9031,
      "step": 1598
    },
    {
      "epoch": 0.5400202634245187,
      "grad_norm": 1.9612479209899902,
      "learning_rate": 8.79463319744677e-06,
      "loss": 0.8621,
      "step": 1599
    },
    {
      "epoch": 0.5403579871664979,
      "grad_norm": 1.9854050874710083,
      "learning_rate": 8.7840656449229e-06,
      "loss": 0.9291,
      "step": 1600
    },
    {
      "epoch": 0.5406957109084769,
      "grad_norm": 1.8677308559417725,
      "learning_rate": 8.773499470469467e-06,
      "loss": 0.9154,
      "step": 1601
    },
    {
      "epoch": 0.541033434650456,
      "grad_norm": 2.255582809448242,
      "learning_rate": 8.762934686061561e-06,
      "loss": 0.8541,
      "step": 1602
    },
    {
      "epoch": 0.541371158392435,
      "grad_norm": 2.1378121376037598,
      "learning_rate": 8.752371303672708e-06,
      "loss": 0.9834,
      "step": 1603
    },
    {
      "epoch": 0.541708882134414,
      "grad_norm": 2.056579351425171,
      "learning_rate": 8.741809335274835e-06,
      "loss": 0.9479,
      "step": 1604
    },
    {
      "epoch": 0.5420466058763931,
      "grad_norm": 2.073805570602417,
      "learning_rate": 8.731248792838285e-06,
      "loss": 0.9537,
      "step": 1605
    },
    {
      "epoch": 0.5423843296183721,
      "grad_norm": 2.1310765743255615,
      "learning_rate": 8.720689688331761e-06,
      "loss": 0.8547,
      "step": 1606
    },
    {
      "epoch": 0.5427220533603513,
      "grad_norm": 2.2561984062194824,
      "learning_rate": 8.710132033722351e-06,
      "loss": 0.9075,
      "step": 1607
    },
    {
      "epoch": 0.5430597771023303,
      "grad_norm": 2.318345546722412,
      "learning_rate": 8.699575840975503e-06,
      "loss": 1.0249,
      "step": 1608
    },
    {
      "epoch": 0.5433975008443094,
      "grad_norm": 2.307399034500122,
      "learning_rate": 8.689021122054996e-06,
      "loss": 0.8123,
      "step": 1609
    },
    {
      "epoch": 0.5437352245862884,
      "grad_norm": 2.401289939880371,
      "learning_rate": 8.678467888922944e-06,
      "loss": 0.9329,
      "step": 1610
    },
    {
      "epoch": 0.5440729483282675,
      "grad_norm": 2.175964832305908,
      "learning_rate": 8.667916153539778e-06,
      "loss": 0.8803,
      "step": 1611
    },
    {
      "epoch": 0.5444106720702465,
      "grad_norm": 1.868950605392456,
      "learning_rate": 8.657365927864236e-06,
      "loss": 1.0252,
      "step": 1612
    },
    {
      "epoch": 0.5447483958122256,
      "grad_norm": 2.2036266326904297,
      "learning_rate": 8.646817223853337e-06,
      "loss": 0.991,
      "step": 1613
    },
    {
      "epoch": 0.5450861195542046,
      "grad_norm": 1.9032148122787476,
      "learning_rate": 8.636270053462375e-06,
      "loss": 0.9037,
      "step": 1614
    },
    {
      "epoch": 0.5454238432961838,
      "grad_norm": 1.9057902097702026,
      "learning_rate": 8.625724428644913e-06,
      "loss": 0.9596,
      "step": 1615
    },
    {
      "epoch": 0.5457615670381628,
      "grad_norm": 2.4022998809814453,
      "learning_rate": 8.61518036135276e-06,
      "loss": 0.829,
      "step": 1616
    },
    {
      "epoch": 0.5460992907801419,
      "grad_norm": 1.943615436553955,
      "learning_rate": 8.604637863535958e-06,
      "loss": 0.8647,
      "step": 1617
    },
    {
      "epoch": 0.5464370145221209,
      "grad_norm": 2.3932230472564697,
      "learning_rate": 8.594096947142767e-06,
      "loss": 0.8181,
      "step": 1618
    },
    {
      "epoch": 0.5467747382641,
      "grad_norm": 1.9482349157333374,
      "learning_rate": 8.583557624119664e-06,
      "loss": 0.943,
      "step": 1619
    },
    {
      "epoch": 0.547112462006079,
      "grad_norm": 2.1437089443206787,
      "learning_rate": 8.573019906411313e-06,
      "loss": 0.7898,
      "step": 1620
    },
    {
      "epoch": 0.547450185748058,
      "grad_norm": 2.0396831035614014,
      "learning_rate": 8.56248380596056e-06,
      "loss": 0.88,
      "step": 1621
    },
    {
      "epoch": 0.5477879094900372,
      "grad_norm": 2.2517213821411133,
      "learning_rate": 8.551949334708416e-06,
      "loss": 0.92,
      "step": 1622
    },
    {
      "epoch": 0.5481256332320162,
      "grad_norm": 1.7327356338500977,
      "learning_rate": 8.541416504594052e-06,
      "loss": 0.9288,
      "step": 1623
    },
    {
      "epoch": 0.5484633569739953,
      "grad_norm": 2.0105037689208984,
      "learning_rate": 8.530885327554773e-06,
      "loss": 0.8625,
      "step": 1624
    },
    {
      "epoch": 0.5488010807159743,
      "grad_norm": 1.9824519157409668,
      "learning_rate": 8.520355815526011e-06,
      "loss": 0.832,
      "step": 1625
    },
    {
      "epoch": 0.5491388044579534,
      "grad_norm": 1.9794176816940308,
      "learning_rate": 8.509827980441316e-06,
      "loss": 0.861,
      "step": 1626
    },
    {
      "epoch": 0.5494765281999324,
      "grad_norm": 2.1880228519439697,
      "learning_rate": 8.499301834232332e-06,
      "loss": 0.907,
      "step": 1627
    },
    {
      "epoch": 0.5498142519419115,
      "grad_norm": 2.343493938446045,
      "learning_rate": 8.488777388828793e-06,
      "loss": 0.8633,
      "step": 1628
    },
    {
      "epoch": 0.5501519756838906,
      "grad_norm": 1.960121512413025,
      "learning_rate": 8.478254656158498e-06,
      "loss": 0.9385,
      "step": 1629
    },
    {
      "epoch": 0.5504896994258697,
      "grad_norm": 2.1544158458709717,
      "learning_rate": 8.467733648147318e-06,
      "loss": 0.9277,
      "step": 1630
    },
    {
      "epoch": 0.5508274231678487,
      "grad_norm": 2.313514232635498,
      "learning_rate": 8.457214376719154e-06,
      "loss": 0.8739,
      "step": 1631
    },
    {
      "epoch": 0.5511651469098278,
      "grad_norm": 2.1777889728546143,
      "learning_rate": 8.446696853795949e-06,
      "loss": 0.8889,
      "step": 1632
    },
    {
      "epoch": 0.5515028706518068,
      "grad_norm": 2.0003085136413574,
      "learning_rate": 8.436181091297665e-06,
      "loss": 0.8681,
      "step": 1633
    },
    {
      "epoch": 0.5518405943937859,
      "grad_norm": 2.312922239303589,
      "learning_rate": 8.425667101142264e-06,
      "loss": 0.8833,
      "step": 1634
    },
    {
      "epoch": 0.5521783181357649,
      "grad_norm": 2.1937663555145264,
      "learning_rate": 8.415154895245698e-06,
      "loss": 0.9127,
      "step": 1635
    },
    {
      "epoch": 0.5525160418777441,
      "grad_norm": 2.202589273452759,
      "learning_rate": 8.404644485521899e-06,
      "loss": 0.8447,
      "step": 1636
    },
    {
      "epoch": 0.5528537656197231,
      "grad_norm": 2.3558027744293213,
      "learning_rate": 8.39413588388277e-06,
      "loss": 0.9622,
      "step": 1637
    },
    {
      "epoch": 0.5531914893617021,
      "grad_norm": 2.294466495513916,
      "learning_rate": 8.383629102238157e-06,
      "loss": 0.8214,
      "step": 1638
    },
    {
      "epoch": 0.5535292131036812,
      "grad_norm": 2.2326254844665527,
      "learning_rate": 8.373124152495841e-06,
      "loss": 0.9415,
      "step": 1639
    },
    {
      "epoch": 0.5538669368456602,
      "grad_norm": 2.200256109237671,
      "learning_rate": 8.362621046561534e-06,
      "loss": 0.8697,
      "step": 1640
    },
    {
      "epoch": 0.5542046605876393,
      "grad_norm": 2.218677520751953,
      "learning_rate": 8.352119796338858e-06,
      "loss": 0.9324,
      "step": 1641
    },
    {
      "epoch": 0.5545423843296183,
      "grad_norm": 2.152667999267578,
      "learning_rate": 8.341620413729326e-06,
      "loss": 0.9291,
      "step": 1642
    },
    {
      "epoch": 0.5548801080715975,
      "grad_norm": 2.320580244064331,
      "learning_rate": 8.331122910632334e-06,
      "loss": 0.8655,
      "step": 1643
    },
    {
      "epoch": 0.5552178318135765,
      "grad_norm": 2.049086332321167,
      "learning_rate": 8.320627298945156e-06,
      "loss": 0.9566,
      "step": 1644
    },
    {
      "epoch": 0.5555555555555556,
      "grad_norm": 1.898208737373352,
      "learning_rate": 8.310133590562922e-06,
      "loss": 0.9466,
      "step": 1645
    },
    {
      "epoch": 0.5558932792975346,
      "grad_norm": 1.9477075338363647,
      "learning_rate": 8.29964179737859e-06,
      "loss": 0.9176,
      "step": 1646
    },
    {
      "epoch": 0.5562310030395137,
      "grad_norm": 2.3717169761657715,
      "learning_rate": 8.289151931282969e-06,
      "loss": 0.8439,
      "step": 1647
    },
    {
      "epoch": 0.5565687267814927,
      "grad_norm": 2.031773090362549,
      "learning_rate": 8.278664004164665e-06,
      "loss": 0.9002,
      "step": 1648
    },
    {
      "epoch": 0.5569064505234718,
      "grad_norm": 1.9806246757507324,
      "learning_rate": 8.268178027910101e-06,
      "loss": 0.8697,
      "step": 1649
    },
    {
      "epoch": 0.5572441742654508,
      "grad_norm": 2.036160707473755,
      "learning_rate": 8.257694014403474e-06,
      "loss": 0.9453,
      "step": 1650
    },
    {
      "epoch": 0.55758189800743,
      "grad_norm": 2.2009780406951904,
      "learning_rate": 8.247211975526776e-06,
      "loss": 0.8797,
      "step": 1651
    },
    {
      "epoch": 0.557919621749409,
      "grad_norm": 2.8980820178985596,
      "learning_rate": 8.236731923159744e-06,
      "loss": 0.8837,
      "step": 1652
    },
    {
      "epoch": 0.5582573454913881,
      "grad_norm": 2.106335401535034,
      "learning_rate": 8.226253869179865e-06,
      "loss": 0.9713,
      "step": 1653
    },
    {
      "epoch": 0.5585950692333671,
      "grad_norm": 2.0530128479003906,
      "learning_rate": 8.215777825462375e-06,
      "loss": 0.8925,
      "step": 1654
    },
    {
      "epoch": 0.5589327929753461,
      "grad_norm": 2.0262086391448975,
      "learning_rate": 8.205303803880221e-06,
      "loss": 0.8723,
      "step": 1655
    },
    {
      "epoch": 0.5592705167173252,
      "grad_norm": 1.9922345876693726,
      "learning_rate": 8.194831816304057e-06,
      "loss": 0.9299,
      "step": 1656
    },
    {
      "epoch": 0.5596082404593042,
      "grad_norm": 1.9655886888504028,
      "learning_rate": 8.18436187460223e-06,
      "loss": 0.8191,
      "step": 1657
    },
    {
      "epoch": 0.5599459642012834,
      "grad_norm": 2.097193717956543,
      "learning_rate": 8.173893990640782e-06,
      "loss": 0.9287,
      "step": 1658
    },
    {
      "epoch": 0.5602836879432624,
      "grad_norm": 2.0568060874938965,
      "learning_rate": 8.163428176283411e-06,
      "loss": 0.9968,
      "step": 1659
    },
    {
      "epoch": 0.5606214116852415,
      "grad_norm": 2.266770601272583,
      "learning_rate": 8.152964443391466e-06,
      "loss": 0.9401,
      "step": 1660
    },
    {
      "epoch": 0.5609591354272205,
      "grad_norm": 2.385958194732666,
      "learning_rate": 8.142502803823954e-06,
      "loss": 0.7995,
      "step": 1661
    },
    {
      "epoch": 0.5612968591691996,
      "grad_norm": 2.346012830734253,
      "learning_rate": 8.132043269437491e-06,
      "loss": 0.924,
      "step": 1662
    },
    {
      "epoch": 0.5616345829111786,
      "grad_norm": 2.438565254211426,
      "learning_rate": 8.12158585208632e-06,
      "loss": 0.9517,
      "step": 1663
    },
    {
      "epoch": 0.5619723066531577,
      "grad_norm": 2.374269962310791,
      "learning_rate": 8.11113056362227e-06,
      "loss": 0.9694,
      "step": 1664
    },
    {
      "epoch": 0.5623100303951368,
      "grad_norm": 2.4003591537475586,
      "learning_rate": 8.10067741589478e-06,
      "loss": 0.9133,
      "step": 1665
    },
    {
      "epoch": 0.5626477541371159,
      "grad_norm": 1.986120343208313,
      "learning_rate": 8.09022642075084e-06,
      "loss": 0.9036,
      "step": 1666
    },
    {
      "epoch": 0.5629854778790949,
      "grad_norm": 1.9208234548568726,
      "learning_rate": 8.07977759003501e-06,
      "loss": 0.9339,
      "step": 1667
    },
    {
      "epoch": 0.563323201621074,
      "grad_norm": 2.115947723388672,
      "learning_rate": 8.069330935589403e-06,
      "loss": 0.9524,
      "step": 1668
    },
    {
      "epoch": 0.563660925363053,
      "grad_norm": 2.4844136238098145,
      "learning_rate": 8.058886469253658e-06,
      "loss": 0.8905,
      "step": 1669
    },
    {
      "epoch": 0.5639986491050321,
      "grad_norm": 2.253383159637451,
      "learning_rate": 8.048444202864932e-06,
      "loss": 0.8262,
      "step": 1670
    },
    {
      "epoch": 0.5643363728470111,
      "grad_norm": 2.7413434982299805,
      "learning_rate": 8.038004148257892e-06,
      "loss": 0.7139,
      "step": 1671
    },
    {
      "epoch": 0.5646740965889903,
      "grad_norm": 2.153628349304199,
      "learning_rate": 8.027566317264704e-06,
      "loss": 0.8672,
      "step": 1672
    },
    {
      "epoch": 0.5650118203309693,
      "grad_norm": 2.2476258277893066,
      "learning_rate": 8.017130721715006e-06,
      "loss": 0.9549,
      "step": 1673
    },
    {
      "epoch": 0.5653495440729484,
      "grad_norm": 1.9218089580535889,
      "learning_rate": 8.0066973734359e-06,
      "loss": 0.926,
      "step": 1674
    },
    {
      "epoch": 0.5656872678149274,
      "grad_norm": 2.3433077335357666,
      "learning_rate": 7.996266284251952e-06,
      "loss": 0.9498,
      "step": 1675
    },
    {
      "epoch": 0.5660249915569064,
      "grad_norm": 2.7445151805877686,
      "learning_rate": 7.985837465985163e-06,
      "loss": 0.9342,
      "step": 1676
    },
    {
      "epoch": 0.5663627152988855,
      "grad_norm": 2.0426998138427734,
      "learning_rate": 7.975410930454954e-06,
      "loss": 0.869,
      "step": 1677
    },
    {
      "epoch": 0.5667004390408645,
      "grad_norm": 2.010939598083496,
      "learning_rate": 7.964986689478164e-06,
      "loss": 0.894,
      "step": 1678
    },
    {
      "epoch": 0.5670381627828436,
      "grad_norm": 2.144061326980591,
      "learning_rate": 7.954564754869034e-06,
      "loss": 0.9449,
      "step": 1679
    },
    {
      "epoch": 0.5673758865248227,
      "grad_norm": 2.0930023193359375,
      "learning_rate": 7.944145138439189e-06,
      "loss": 0.958,
      "step": 1680
    },
    {
      "epoch": 0.5677136102668018,
      "grad_norm": 2.1209235191345215,
      "learning_rate": 7.933727851997621e-06,
      "loss": 0.9168,
      "step": 1681
    },
    {
      "epoch": 0.5680513340087808,
      "grad_norm": 2.08486270904541,
      "learning_rate": 7.923312907350694e-06,
      "loss": 0.8859,
      "step": 1682
    },
    {
      "epoch": 0.5683890577507599,
      "grad_norm": 1.9459245204925537,
      "learning_rate": 7.912900316302106e-06,
      "loss": 0.9433,
      "step": 1683
    },
    {
      "epoch": 0.5687267814927389,
      "grad_norm": 1.8624752759933472,
      "learning_rate": 7.902490090652892e-06,
      "loss": 0.9592,
      "step": 1684
    },
    {
      "epoch": 0.569064505234718,
      "grad_norm": 2.1397950649261475,
      "learning_rate": 7.892082242201405e-06,
      "loss": 0.8845,
      "step": 1685
    },
    {
      "epoch": 0.569402228976697,
      "grad_norm": 2.2607498168945312,
      "learning_rate": 7.88167678274331e-06,
      "loss": 0.8036,
      "step": 1686
    },
    {
      "epoch": 0.5697399527186762,
      "grad_norm": 1.9214880466461182,
      "learning_rate": 7.871273724071553e-06,
      "loss": 0.9067,
      "step": 1687
    },
    {
      "epoch": 0.5700776764606552,
      "grad_norm": 2.1255006790161133,
      "learning_rate": 7.860873077976366e-06,
      "loss": 0.8889,
      "step": 1688
    },
    {
      "epoch": 0.5704154002026343,
      "grad_norm": 2.189648389816284,
      "learning_rate": 7.850474856245255e-06,
      "loss": 0.9139,
      "step": 1689
    },
    {
      "epoch": 0.5707531239446133,
      "grad_norm": 2.2092716693878174,
      "learning_rate": 7.840079070662962e-06,
      "loss": 0.9272,
      "step": 1690
    },
    {
      "epoch": 0.5710908476865924,
      "grad_norm": 2.1536898612976074,
      "learning_rate": 7.829685733011477e-06,
      "loss": 0.8146,
      "step": 1691
    },
    {
      "epoch": 0.5714285714285714,
      "grad_norm": 2.291381597518921,
      "learning_rate": 7.81929485507001e-06,
      "loss": 0.8777,
      "step": 1692
    },
    {
      "epoch": 0.5717662951705504,
      "grad_norm": 2.3900275230407715,
      "learning_rate": 7.808906448614997e-06,
      "loss": 0.8488,
      "step": 1693
    },
    {
      "epoch": 0.5721040189125296,
      "grad_norm": 2.0692286491394043,
      "learning_rate": 7.79852052542006e-06,
      "loss": 0.9248,
      "step": 1694
    },
    {
      "epoch": 0.5724417426545086,
      "grad_norm": 2.281797409057617,
      "learning_rate": 7.788137097256003e-06,
      "loss": 0.7534,
      "step": 1695
    },
    {
      "epoch": 0.5727794663964877,
      "grad_norm": 2.1658692359924316,
      "learning_rate": 7.777756175890815e-06,
      "loss": 0.9687,
      "step": 1696
    },
    {
      "epoch": 0.5731171901384667,
      "grad_norm": 2.2884953022003174,
      "learning_rate": 7.767377773089642e-06,
      "loss": 0.8972,
      "step": 1697
    },
    {
      "epoch": 0.5734549138804458,
      "grad_norm": 2.310586452484131,
      "learning_rate": 7.757001900614769e-06,
      "loss": 0.8869,
      "step": 1698
    },
    {
      "epoch": 0.5737926376224248,
      "grad_norm": 1.9316874742507935,
      "learning_rate": 7.746628570225612e-06,
      "loss": 0.8866,
      "step": 1699
    },
    {
      "epoch": 0.5741303613644039,
      "grad_norm": 2.201829671859741,
      "learning_rate": 7.736257793678714e-06,
      "loss": 0.8467,
      "step": 1700
    },
    {
      "epoch": 0.574468085106383,
      "grad_norm": 2.088134527206421,
      "learning_rate": 7.725889582727724e-06,
      "loss": 0.9034,
      "step": 1701
    },
    {
      "epoch": 0.5748058088483621,
      "grad_norm": 1.8187178373336792,
      "learning_rate": 7.715523949123372e-06,
      "loss": 0.9437,
      "step": 1702
    },
    {
      "epoch": 0.5751435325903411,
      "grad_norm": 1.9805182218551636,
      "learning_rate": 7.705160904613474e-06,
      "loss": 0.8606,
      "step": 1703
    },
    {
      "epoch": 0.5754812563323202,
      "grad_norm": 2.049921751022339,
      "learning_rate": 7.694800460942917e-06,
      "loss": 0.9477,
      "step": 1704
    },
    {
      "epoch": 0.5758189800742992,
      "grad_norm": 2.239813804626465,
      "learning_rate": 7.684442629853634e-06,
      "loss": 0.8809,
      "step": 1705
    },
    {
      "epoch": 0.5761567038162783,
      "grad_norm": 2.3173892498016357,
      "learning_rate": 7.674087423084593e-06,
      "loss": 0.8176,
      "step": 1706
    },
    {
      "epoch": 0.5764944275582573,
      "grad_norm": 2.383143424987793,
      "learning_rate": 7.663734852371802e-06,
      "loss": 0.8691,
      "step": 1707
    },
    {
      "epoch": 0.5768321513002365,
      "grad_norm": 1.9560092687606812,
      "learning_rate": 7.653384929448264e-06,
      "loss": 0.8804,
      "step": 1708
    },
    {
      "epoch": 0.5771698750422155,
      "grad_norm": 2.8453991413116455,
      "learning_rate": 7.643037666043993e-06,
      "loss": 0.9003,
      "step": 1709
    },
    {
      "epoch": 0.5775075987841946,
      "grad_norm": 2.0371031761169434,
      "learning_rate": 7.632693073885982e-06,
      "loss": 0.9062,
      "step": 1710
    },
    {
      "epoch": 0.5778453225261736,
      "grad_norm": 1.9739131927490234,
      "learning_rate": 7.622351164698209e-06,
      "loss": 0.9874,
      "step": 1711
    },
    {
      "epoch": 0.5781830462681526,
      "grad_norm": 2.4006271362304688,
      "learning_rate": 7.612011950201592e-06,
      "loss": 0.9968,
      "step": 1712
    },
    {
      "epoch": 0.5785207700101317,
      "grad_norm": 2.3302085399627686,
      "learning_rate": 7.601675442114009e-06,
      "loss": 0.8823,
      "step": 1713
    },
    {
      "epoch": 0.5788584937521107,
      "grad_norm": 2.272998094558716,
      "learning_rate": 7.591341652150268e-06,
      "loss": 0.9322,
      "step": 1714
    },
    {
      "epoch": 0.5791962174940898,
      "grad_norm": 2.137287139892578,
      "learning_rate": 7.581010592022096e-06,
      "loss": 0.9733,
      "step": 1715
    },
    {
      "epoch": 0.5795339412360689,
      "grad_norm": 2.3357043266296387,
      "learning_rate": 7.570682273438122e-06,
      "loss": 0.9498,
      "step": 1716
    },
    {
      "epoch": 0.579871664978048,
      "grad_norm": 2.6811578273773193,
      "learning_rate": 7.560356708103873e-06,
      "loss": 0.9001,
      "step": 1717
    },
    {
      "epoch": 0.580209388720027,
      "grad_norm": 2.273045063018799,
      "learning_rate": 7.550033907721754e-06,
      "loss": 0.8973,
      "step": 1718
    },
    {
      "epoch": 0.5805471124620061,
      "grad_norm": 2.472449541091919,
      "learning_rate": 7.53971388399104e-06,
      "loss": 0.8634,
      "step": 1719
    },
    {
      "epoch": 0.5808848362039851,
      "grad_norm": 2.100856065750122,
      "learning_rate": 7.529396648607848e-06,
      "loss": 0.8988,
      "step": 1720
    },
    {
      "epoch": 0.5812225599459642,
      "grad_norm": 2.185356378555298,
      "learning_rate": 7.51908221326515e-06,
      "loss": 0.9632,
      "step": 1721
    },
    {
      "epoch": 0.5815602836879432,
      "grad_norm": 2.2246310710906982,
      "learning_rate": 7.508770589652731e-06,
      "loss": 0.9229,
      "step": 1722
    },
    {
      "epoch": 0.5818980074299224,
      "grad_norm": 1.9937264919281006,
      "learning_rate": 7.498461789457201e-06,
      "loss": 0.9053,
      "step": 1723
    },
    {
      "epoch": 0.5822357311719014,
      "grad_norm": 2.263745069503784,
      "learning_rate": 7.488155824361955e-06,
      "loss": 0.8488,
      "step": 1724
    },
    {
      "epoch": 0.5825734549138805,
      "grad_norm": 1.8541616201400757,
      "learning_rate": 7.477852706047195e-06,
      "loss": 0.945,
      "step": 1725
    },
    {
      "epoch": 0.5829111786558595,
      "grad_norm": 2.170732259750366,
      "learning_rate": 7.467552446189879e-06,
      "loss": 0.871,
      "step": 1726
    },
    {
      "epoch": 0.5832489023978386,
      "grad_norm": 2.1467292308807373,
      "learning_rate": 7.457255056463732e-06,
      "loss": 0.9162,
      "step": 1727
    },
    {
      "epoch": 0.5835866261398176,
      "grad_norm": 2.2860095500946045,
      "learning_rate": 7.446960548539235e-06,
      "loss": 0.8995,
      "step": 1728
    },
    {
      "epoch": 0.5839243498817966,
      "grad_norm": 2.016458511352539,
      "learning_rate": 7.4366689340835865e-06,
      "loss": 0.9526,
      "step": 1729
    },
    {
      "epoch": 0.5842620736237758,
      "grad_norm": 1.9850744009017944,
      "learning_rate": 7.426380224760712e-06,
      "loss": 0.8824,
      "step": 1730
    },
    {
      "epoch": 0.5845997973657548,
      "grad_norm": 1.9019895792007446,
      "learning_rate": 7.416094432231247e-06,
      "loss": 0.9042,
      "step": 1731
    },
    {
      "epoch": 0.5849375211077339,
      "grad_norm": 2.049781084060669,
      "learning_rate": 7.405811568152528e-06,
      "loss": 0.8822,
      "step": 1732
    },
    {
      "epoch": 0.5852752448497129,
      "grad_norm": 2.1685612201690674,
      "learning_rate": 7.395531644178558e-06,
      "loss": 0.8826,
      "step": 1733
    },
    {
      "epoch": 0.585612968591692,
      "grad_norm": 2.1419284343719482,
      "learning_rate": 7.385254671960012e-06,
      "loss": 0.984,
      "step": 1734
    },
    {
      "epoch": 0.585950692333671,
      "grad_norm": 2.1284854412078857,
      "learning_rate": 7.374980663144226e-06,
      "loss": 0.8489,
      "step": 1735
    },
    {
      "epoch": 0.5862884160756501,
      "grad_norm": 2.247516632080078,
      "learning_rate": 7.36470962937518e-06,
      "loss": 0.9894,
      "step": 1736
    },
    {
      "epoch": 0.5866261398176292,
      "grad_norm": 2.281723976135254,
      "learning_rate": 7.354441582293468e-06,
      "loss": 0.9193,
      "step": 1737
    },
    {
      "epoch": 0.5869638635596083,
      "grad_norm": 2.308698892593384,
      "learning_rate": 7.344176533536305e-06,
      "loss": 0.9273,
      "step": 1738
    },
    {
      "epoch": 0.5873015873015873,
      "grad_norm": 2.126769781112671,
      "learning_rate": 7.3339144947375155e-06,
      "loss": 0.8878,
      "step": 1739
    },
    {
      "epoch": 0.5876393110435664,
      "grad_norm": 2.231560707092285,
      "learning_rate": 7.323655477527509e-06,
      "loss": 0.9695,
      "step": 1740
    },
    {
      "epoch": 0.5879770347855454,
      "grad_norm": 2.1032183170318604,
      "learning_rate": 7.3133994935332605e-06,
      "loss": 0.8796,
      "step": 1741
    },
    {
      "epoch": 0.5883147585275245,
      "grad_norm": 2.1094555854797363,
      "learning_rate": 7.303146554378325e-06,
      "loss": 0.963,
      "step": 1742
    },
    {
      "epoch": 0.5886524822695035,
      "grad_norm": 2.0848007202148438,
      "learning_rate": 7.292896671682791e-06,
      "loss": 0.8955,
      "step": 1743
    },
    {
      "epoch": 0.5889902060114826,
      "grad_norm": 1.9999600648880005,
      "learning_rate": 7.282649857063292e-06,
      "loss": 0.9531,
      "step": 1744
    },
    {
      "epoch": 0.5893279297534617,
      "grad_norm": 1.9510157108306885,
      "learning_rate": 7.272406122132976e-06,
      "loss": 0.9035,
      "step": 1745
    },
    {
      "epoch": 0.5896656534954408,
      "grad_norm": 2.4807143211364746,
      "learning_rate": 7.262165478501515e-06,
      "loss": 0.7668,
      "step": 1746
    },
    {
      "epoch": 0.5900033772374198,
      "grad_norm": 2.124891519546509,
      "learning_rate": 7.2519279377750605e-06,
      "loss": 0.9721,
      "step": 1747
    },
    {
      "epoch": 0.5903411009793988,
      "grad_norm": 2.4939804077148438,
      "learning_rate": 7.241693511556258e-06,
      "loss": 0.8076,
      "step": 1748
    },
    {
      "epoch": 0.5906788247213779,
      "grad_norm": 2.1625943183898926,
      "learning_rate": 7.231462211444226e-06,
      "loss": 0.8134,
      "step": 1749
    },
    {
      "epoch": 0.5910165484633569,
      "grad_norm": 2.1695523262023926,
      "learning_rate": 7.221234049034529e-06,
      "loss": 0.8787,
      "step": 1750
    },
    {
      "epoch": 0.591354272205336,
      "grad_norm": 2.3628430366516113,
      "learning_rate": 7.211009035919181e-06,
      "loss": 0.8555,
      "step": 1751
    },
    {
      "epoch": 0.5916919959473151,
      "grad_norm": 1.8637280464172363,
      "learning_rate": 7.200787183686625e-06,
      "loss": 0.9656,
      "step": 1752
    },
    {
      "epoch": 0.5920297196892942,
      "grad_norm": 2.0604541301727295,
      "learning_rate": 7.190568503921731e-06,
      "loss": 0.9758,
      "step": 1753
    },
    {
      "epoch": 0.5923674434312732,
      "grad_norm": 2.2500455379486084,
      "learning_rate": 7.180353008205763e-06,
      "loss": 0.9587,
      "step": 1754
    },
    {
      "epoch": 0.5927051671732523,
      "grad_norm": 2.195838689804077,
      "learning_rate": 7.170140708116373e-06,
      "loss": 0.9829,
      "step": 1755
    },
    {
      "epoch": 0.5930428909152313,
      "grad_norm": 2.237177610397339,
      "learning_rate": 7.159931615227604e-06,
      "loss": 0.8671,
      "step": 1756
    },
    {
      "epoch": 0.5933806146572104,
      "grad_norm": 2.3936445713043213,
      "learning_rate": 7.14972574110986e-06,
      "loss": 0.8303,
      "step": 1757
    },
    {
      "epoch": 0.5937183383991894,
      "grad_norm": 1.8947376012802124,
      "learning_rate": 7.13952309732989e-06,
      "loss": 0.9206,
      "step": 1758
    },
    {
      "epoch": 0.5940560621411686,
      "grad_norm": 2.177638530731201,
      "learning_rate": 7.129323695450785e-06,
      "loss": 0.9165,
      "step": 1759
    },
    {
      "epoch": 0.5943937858831476,
      "grad_norm": 2.0789384841918945,
      "learning_rate": 7.119127547031967e-06,
      "loss": 0.8987,
      "step": 1760
    },
    {
      "epoch": 0.5947315096251267,
      "grad_norm": 1.8517448902130127,
      "learning_rate": 7.108934663629171e-06,
      "loss": 0.9031,
      "step": 1761
    },
    {
      "epoch": 0.5950692333671057,
      "grad_norm": 2.200853109359741,
      "learning_rate": 7.09874505679442e-06,
      "loss": 0.9863,
      "step": 1762
    },
    {
      "epoch": 0.5954069571090848,
      "grad_norm": 2.3959531784057617,
      "learning_rate": 7.088558738076037e-06,
      "loss": 0.9081,
      "step": 1763
    },
    {
      "epoch": 0.5957446808510638,
      "grad_norm": 2.084338426589966,
      "learning_rate": 7.0783757190186106e-06,
      "loss": 0.9564,
      "step": 1764
    },
    {
      "epoch": 0.5960824045930428,
      "grad_norm": 1.9701474905014038,
      "learning_rate": 7.068196011162994e-06,
      "loss": 0.9431,
      "step": 1765
    },
    {
      "epoch": 0.596420128335022,
      "grad_norm": 1.9587982892990112,
      "learning_rate": 7.058019626046281e-06,
      "loss": 0.854,
      "step": 1766
    },
    {
      "epoch": 0.596757852077001,
      "grad_norm": 1.8701039552688599,
      "learning_rate": 7.047846575201811e-06,
      "loss": 0.8648,
      "step": 1767
    },
    {
      "epoch": 0.5970955758189801,
      "grad_norm": 2.1876587867736816,
      "learning_rate": 7.037676870159131e-06,
      "loss": 0.9565,
      "step": 1768
    },
    {
      "epoch": 0.5974332995609591,
      "grad_norm": 2.2911555767059326,
      "learning_rate": 7.027510522444008e-06,
      "loss": 0.8579,
      "step": 1769
    },
    {
      "epoch": 0.5977710233029382,
      "grad_norm": 1.9534614086151123,
      "learning_rate": 7.017347543578396e-06,
      "loss": 0.9064,
      "step": 1770
    },
    {
      "epoch": 0.5981087470449172,
      "grad_norm": 2.0604896545410156,
      "learning_rate": 7.007187945080438e-06,
      "loss": 0.9507,
      "step": 1771
    },
    {
      "epoch": 0.5984464707868963,
      "grad_norm": 2.0710809230804443,
      "learning_rate": 6.997031738464435e-06,
      "loss": 0.8951,
      "step": 1772
    },
    {
      "epoch": 0.5987841945288754,
      "grad_norm": 1.9995189905166626,
      "learning_rate": 6.986878935240855e-06,
      "loss": 0.869,
      "step": 1773
    },
    {
      "epoch": 0.5991219182708545,
      "grad_norm": 2.1634485721588135,
      "learning_rate": 6.976729546916302e-06,
      "loss": 0.8228,
      "step": 1774
    },
    {
      "epoch": 0.5994596420128335,
      "grad_norm": 2.1357860565185547,
      "learning_rate": 6.9665835849935155e-06,
      "loss": 0.9244,
      "step": 1775
    },
    {
      "epoch": 0.5997973657548126,
      "grad_norm": 2.621122121810913,
      "learning_rate": 6.956441060971346e-06,
      "loss": 0.7681,
      "step": 1776
    },
    {
      "epoch": 0.6001350894967916,
      "grad_norm": 2.1879477500915527,
      "learning_rate": 6.946301986344749e-06,
      "loss": 0.9339,
      "step": 1777
    },
    {
      "epoch": 0.6004728132387707,
      "grad_norm": 2.2887489795684814,
      "learning_rate": 6.936166372604773e-06,
      "loss": 0.8763,
      "step": 1778
    },
    {
      "epoch": 0.6008105369807497,
      "grad_norm": 2.167616128921509,
      "learning_rate": 6.926034231238544e-06,
      "loss": 0.9342,
      "step": 1779
    },
    {
      "epoch": 0.6011482607227288,
      "grad_norm": 2.1249804496765137,
      "learning_rate": 6.915905573729246e-06,
      "loss": 0.9217,
      "step": 1780
    },
    {
      "epoch": 0.6014859844647079,
      "grad_norm": 2.1896371841430664,
      "learning_rate": 6.905780411556128e-06,
      "loss": 0.8861,
      "step": 1781
    },
    {
      "epoch": 0.601823708206687,
      "grad_norm": 2.069077253341675,
      "learning_rate": 6.895658756194463e-06,
      "loss": 0.9096,
      "step": 1782
    },
    {
      "epoch": 0.602161431948666,
      "grad_norm": 2.4963550567626953,
      "learning_rate": 6.88554061911556e-06,
      "loss": 0.8855,
      "step": 1783
    },
    {
      "epoch": 0.602499155690645,
      "grad_norm": 2.309236764907837,
      "learning_rate": 6.875426011786731e-06,
      "loss": 0.9292,
      "step": 1784
    },
    {
      "epoch": 0.6028368794326241,
      "grad_norm": 1.8941360712051392,
      "learning_rate": 6.865314945671302e-06,
      "loss": 0.8666,
      "step": 1785
    },
    {
      "epoch": 0.6031746031746031,
      "grad_norm": 2.0691680908203125,
      "learning_rate": 6.85520743222857e-06,
      "loss": 0.8494,
      "step": 1786
    },
    {
      "epoch": 0.6035123269165822,
      "grad_norm": 2.071002244949341,
      "learning_rate": 6.845103482913813e-06,
      "loss": 0.9134,
      "step": 1787
    },
    {
      "epoch": 0.6038500506585613,
      "grad_norm": 2.2587225437164307,
      "learning_rate": 6.835003109178273e-06,
      "loss": 0.981,
      "step": 1788
    },
    {
      "epoch": 0.6041877744005404,
      "grad_norm": 2.0450143814086914,
      "learning_rate": 6.8249063224691355e-06,
      "loss": 0.928,
      "step": 1789
    },
    {
      "epoch": 0.6045254981425194,
      "grad_norm": 2.0472524166107178,
      "learning_rate": 6.814813134229516e-06,
      "loss": 0.8549,
      "step": 1790
    },
    {
      "epoch": 0.6048632218844985,
      "grad_norm": 2.046198844909668,
      "learning_rate": 6.804723555898458e-06,
      "loss": 0.9974,
      "step": 1791
    },
    {
      "epoch": 0.6052009456264775,
      "grad_norm": 1.916806936264038,
      "learning_rate": 6.794637598910917e-06,
      "loss": 0.9846,
      "step": 1792
    },
    {
      "epoch": 0.6055386693684566,
      "grad_norm": 2.353759527206421,
      "learning_rate": 6.784555274697737e-06,
      "loss": 0.8766,
      "step": 1793
    },
    {
      "epoch": 0.6058763931104356,
      "grad_norm": 2.3988232612609863,
      "learning_rate": 6.7744765946856415e-06,
      "loss": 0.8575,
      "step": 1794
    },
    {
      "epoch": 0.6062141168524148,
      "grad_norm": 2.3079159259796143,
      "learning_rate": 6.764401570297239e-06,
      "loss": 0.8794,
      "step": 1795
    },
    {
      "epoch": 0.6065518405943938,
      "grad_norm": 2.070265769958496,
      "learning_rate": 6.754330212950984e-06,
      "loss": 0.9861,
      "step": 1796
    },
    {
      "epoch": 0.6068895643363729,
      "grad_norm": 2.287381410598755,
      "learning_rate": 6.744262534061173e-06,
      "loss": 0.9272,
      "step": 1797
    },
    {
      "epoch": 0.6072272880783519,
      "grad_norm": 2.00732684135437,
      "learning_rate": 6.734198545037935e-06,
      "loss": 0.9224,
      "step": 1798
    },
    {
      "epoch": 0.607565011820331,
      "grad_norm": 1.9060947895050049,
      "learning_rate": 6.724138257287227e-06,
      "loss": 0.9239,
      "step": 1799
    },
    {
      "epoch": 0.60790273556231,
      "grad_norm": 1.8695586919784546,
      "learning_rate": 6.7140816822108e-06,
      "loss": 0.927,
      "step": 1800
    },
    {
      "epoch": 0.608240459304289,
      "grad_norm": 2.3061318397521973,
      "learning_rate": 6.704028831206197e-06,
      "loss": 0.9765,
      "step": 1801
    },
    {
      "epoch": 0.6085781830462682,
      "grad_norm": 2.1096982955932617,
      "learning_rate": 6.693979715666753e-06,
      "loss": 0.8672,
      "step": 1802
    },
    {
      "epoch": 0.6089159067882473,
      "grad_norm": 2.3348734378814697,
      "learning_rate": 6.683934346981553e-06,
      "loss": 0.8313,
      "step": 1803
    },
    {
      "epoch": 0.6092536305302263,
      "grad_norm": 2.040879249572754,
      "learning_rate": 6.673892736535448e-06,
      "loss": 0.9163,
      "step": 1804
    },
    {
      "epoch": 0.6095913542722053,
      "grad_norm": 2.2922797203063965,
      "learning_rate": 6.663854895709017e-06,
      "loss": 0.9154,
      "step": 1805
    },
    {
      "epoch": 0.6099290780141844,
      "grad_norm": 2.617279291152954,
      "learning_rate": 6.653820835878584e-06,
      "loss": 0.9148,
      "step": 1806
    },
    {
      "epoch": 0.6102668017561634,
      "grad_norm": 2.1217141151428223,
      "learning_rate": 6.643790568416173e-06,
      "loss": 0.8771,
      "step": 1807
    },
    {
      "epoch": 0.6106045254981425,
      "grad_norm": 2.266434907913208,
      "learning_rate": 6.6337641046895155e-06,
      "loss": 0.9117,
      "step": 1808
    },
    {
      "epoch": 0.6109422492401215,
      "grad_norm": 1.849165439605713,
      "learning_rate": 6.6237414560620335e-06,
      "loss": 0.9752,
      "step": 1809
    },
    {
      "epoch": 0.6112799729821007,
      "grad_norm": 2.086050510406494,
      "learning_rate": 6.613722633892826e-06,
      "loss": 0.972,
      "step": 1810
    },
    {
      "epoch": 0.6116176967240797,
      "grad_norm": 2.7654125690460205,
      "learning_rate": 6.603707649536647e-06,
      "loss": 0.7942,
      "step": 1811
    },
    {
      "epoch": 0.6119554204660588,
      "grad_norm": 2.1671764850616455,
      "learning_rate": 6.593696514343908e-06,
      "loss": 0.8873,
      "step": 1812
    },
    {
      "epoch": 0.6122931442080378,
      "grad_norm": 2.041618585586548,
      "learning_rate": 6.583689239660663e-06,
      "loss": 0.9598,
      "step": 1813
    },
    {
      "epoch": 0.6126308679500169,
      "grad_norm": 1.920276165008545,
      "learning_rate": 6.573685836828578e-06,
      "loss": 0.9528,
      "step": 1814
    },
    {
      "epoch": 0.6129685916919959,
      "grad_norm": 1.9856504201889038,
      "learning_rate": 6.563686317184937e-06,
      "loss": 0.9097,
      "step": 1815
    },
    {
      "epoch": 0.613306315433975,
      "grad_norm": 2.2021074295043945,
      "learning_rate": 6.553690692062626e-06,
      "loss": 0.9335,
      "step": 1816
    },
    {
      "epoch": 0.6136440391759541,
      "grad_norm": 2.13700795173645,
      "learning_rate": 6.543698972790118e-06,
      "loss": 0.8085,
      "step": 1817
    },
    {
      "epoch": 0.6139817629179332,
      "grad_norm": 1.9793208837509155,
      "learning_rate": 6.533711170691453e-06,
      "loss": 0.8681,
      "step": 1818
    },
    {
      "epoch": 0.6143194866599122,
      "grad_norm": 1.9172335863113403,
      "learning_rate": 6.52372729708623e-06,
      "loss": 0.9335,
      "step": 1819
    },
    {
      "epoch": 0.6146572104018913,
      "grad_norm": 2.2456419467926025,
      "learning_rate": 6.513747363289606e-06,
      "loss": 0.8959,
      "step": 1820
    },
    {
      "epoch": 0.6149949341438703,
      "grad_norm": 2.379347085952759,
      "learning_rate": 6.503771380612267e-06,
      "loss": 0.9085,
      "step": 1821
    },
    {
      "epoch": 0.6153326578858493,
      "grad_norm": 1.8554575443267822,
      "learning_rate": 6.493799360360414e-06,
      "loss": 0.8501,
      "step": 1822
    },
    {
      "epoch": 0.6156703816278284,
      "grad_norm": 2.8844170570373535,
      "learning_rate": 6.483831313835776e-06,
      "loss": 0.9154,
      "step": 1823
    },
    {
      "epoch": 0.6160081053698075,
      "grad_norm": 2.3491733074188232,
      "learning_rate": 6.4738672523355565e-06,
      "loss": 0.863,
      "step": 1824
    },
    {
      "epoch": 0.6163458291117866,
      "grad_norm": 2.301781415939331,
      "learning_rate": 6.463907187152458e-06,
      "loss": 0.911,
      "step": 1825
    },
    {
      "epoch": 0.6166835528537656,
      "grad_norm": 2.172818422317505,
      "learning_rate": 6.453951129574644e-06,
      "loss": 0.8763,
      "step": 1826
    },
    {
      "epoch": 0.6170212765957447,
      "grad_norm": 2.067887544631958,
      "learning_rate": 6.443999090885746e-06,
      "loss": 0.8922,
      "step": 1827
    },
    {
      "epoch": 0.6173590003377237,
      "grad_norm": 2.1434357166290283,
      "learning_rate": 6.434051082364832e-06,
      "loss": 0.8937,
      "step": 1828
    },
    {
      "epoch": 0.6176967240797028,
      "grad_norm": 1.948878526687622,
      "learning_rate": 6.424107115286404e-06,
      "loss": 0.8752,
      "step": 1829
    },
    {
      "epoch": 0.6180344478216818,
      "grad_norm": 2.281795024871826,
      "learning_rate": 6.414167200920392e-06,
      "loss": 0.9853,
      "step": 1830
    },
    {
      "epoch": 0.618372171563661,
      "grad_norm": 2.1348700523376465,
      "learning_rate": 6.404231350532122e-06,
      "loss": 0.9082,
      "step": 1831
    },
    {
      "epoch": 0.61870989530564,
      "grad_norm": 2.401895761489868,
      "learning_rate": 6.3942995753823165e-06,
      "loss": 0.9092,
      "step": 1832
    },
    {
      "epoch": 0.6190476190476191,
      "grad_norm": 2.215165376663208,
      "learning_rate": 6.384371886727083e-06,
      "loss": 0.9963,
      "step": 1833
    },
    {
      "epoch": 0.6193853427895981,
      "grad_norm": 2.100831985473633,
      "learning_rate": 6.3744482958178966e-06,
      "loss": 0.9438,
      "step": 1834
    },
    {
      "epoch": 0.6197230665315772,
      "grad_norm": 2.176938533782959,
      "learning_rate": 6.364528813901587e-06,
      "loss": 0.9119,
      "step": 1835
    },
    {
      "epoch": 0.6200607902735562,
      "grad_norm": 2.083984851837158,
      "learning_rate": 6.354613452220326e-06,
      "loss": 0.9821,
      "step": 1836
    },
    {
      "epoch": 0.6203985140155353,
      "grad_norm": 2.266484498977661,
      "learning_rate": 6.344702222011621e-06,
      "loss": 0.9371,
      "step": 1837
    },
    {
      "epoch": 0.6207362377575144,
      "grad_norm": 1.9595890045166016,
      "learning_rate": 6.3347951345082895e-06,
      "loss": 0.8819,
      "step": 1838
    },
    {
      "epoch": 0.6210739614994935,
      "grad_norm": 2.30241060256958,
      "learning_rate": 6.324892200938461e-06,
      "loss": 0.7973,
      "step": 1839
    },
    {
      "epoch": 0.6214116852414725,
      "grad_norm": 2.224616765975952,
      "learning_rate": 6.314993432525547e-06,
      "loss": 0.9598,
      "step": 1840
    },
    {
      "epoch": 0.6217494089834515,
      "grad_norm": 2.03629469871521,
      "learning_rate": 6.305098840488255e-06,
      "loss": 0.8905,
      "step": 1841
    },
    {
      "epoch": 0.6220871327254306,
      "grad_norm": 2.4267795085906982,
      "learning_rate": 6.295208436040543e-06,
      "loss": 0.9393,
      "step": 1842
    },
    {
      "epoch": 0.6224248564674096,
      "grad_norm": 2.1655142307281494,
      "learning_rate": 6.285322230391629e-06,
      "loss": 0.9899,
      "step": 1843
    },
    {
      "epoch": 0.6227625802093887,
      "grad_norm": 1.91446053981781,
      "learning_rate": 6.2754402347459795e-06,
      "loss": 0.9699,
      "step": 1844
    },
    {
      "epoch": 0.6231003039513677,
      "grad_norm": 1.9776164293289185,
      "learning_rate": 6.265562460303278e-06,
      "loss": 0.8972,
      "step": 1845
    },
    {
      "epoch": 0.6234380276933469,
      "grad_norm": 2.0903191566467285,
      "learning_rate": 6.255688918258429e-06,
      "loss": 0.8948,
      "step": 1846
    },
    {
      "epoch": 0.6237757514353259,
      "grad_norm": 2.19762921333313,
      "learning_rate": 6.245819619801538e-06,
      "loss": 0.9365,
      "step": 1847
    },
    {
      "epoch": 0.624113475177305,
      "grad_norm": 2.1051478385925293,
      "learning_rate": 6.23595457611791e-06,
      "loss": 0.9818,
      "step": 1848
    },
    {
      "epoch": 0.624451198919284,
      "grad_norm": 2.45239520072937,
      "learning_rate": 6.226093798388017e-06,
      "loss": 0.8955,
      "step": 1849
    },
    {
      "epoch": 0.6247889226612631,
      "grad_norm": 2.1102776527404785,
      "learning_rate": 6.216237297787495e-06,
      "loss": 0.8882,
      "step": 1850
    },
    {
      "epoch": 0.6251266464032421,
      "grad_norm": 2.2518310546875,
      "learning_rate": 6.206385085487148e-06,
      "loss": 0.9108,
      "step": 1851
    },
    {
      "epoch": 0.6254643701452212,
      "grad_norm": 2.483504056930542,
      "learning_rate": 6.196537172652905e-06,
      "loss": 0.9215,
      "step": 1852
    },
    {
      "epoch": 0.6258020938872003,
      "grad_norm": 2.486008882522583,
      "learning_rate": 6.186693570445829e-06,
      "loss": 0.8816,
      "step": 1853
    },
    {
      "epoch": 0.6261398176291794,
      "grad_norm": 2.163830518722534,
      "learning_rate": 6.176854290022087e-06,
      "loss": 0.9876,
      "step": 1854
    },
    {
      "epoch": 0.6264775413711584,
      "grad_norm": 2.341820478439331,
      "learning_rate": 6.167019342532967e-06,
      "loss": 0.9677,
      "step": 1855
    },
    {
      "epoch": 0.6268152651131375,
      "grad_norm": 1.951225757598877,
      "learning_rate": 6.157188739124834e-06,
      "loss": 0.8962,
      "step": 1856
    },
    {
      "epoch": 0.6271529888551165,
      "grad_norm": 1.9839675426483154,
      "learning_rate": 6.147362490939128e-06,
      "loss": 0.9557,
      "step": 1857
    },
    {
      "epoch": 0.6274907125970955,
      "grad_norm": 2.268717050552368,
      "learning_rate": 6.137540609112356e-06,
      "loss": 0.8853,
      "step": 1858
    },
    {
      "epoch": 0.6278284363390746,
      "grad_norm": 2.0686283111572266,
      "learning_rate": 6.12772310477608e-06,
      "loss": 0.8903,
      "step": 1859
    },
    {
      "epoch": 0.6281661600810537,
      "grad_norm": 2.118307113647461,
      "learning_rate": 6.117909989056899e-06,
      "loss": 1.017,
      "step": 1860
    },
    {
      "epoch": 0.6285038838230328,
      "grad_norm": 2.570741653442383,
      "learning_rate": 6.1081012730764285e-06,
      "loss": 0.7749,
      "step": 1861
    },
    {
      "epoch": 0.6288416075650118,
      "grad_norm": 2.6835007667541504,
      "learning_rate": 6.098296967951318e-06,
      "loss": 0.7137,
      "step": 1862
    },
    {
      "epoch": 0.6291793313069909,
      "grad_norm": 2.122966766357422,
      "learning_rate": 6.088497084793197e-06,
      "loss": 0.8632,
      "step": 1863
    },
    {
      "epoch": 0.6295170550489699,
      "grad_norm": 2.1530489921569824,
      "learning_rate": 6.0787016347086995e-06,
      "loss": 0.9405,
      "step": 1864
    },
    {
      "epoch": 0.629854778790949,
      "grad_norm": 1.999572515487671,
      "learning_rate": 6.068910628799418e-06,
      "loss": 0.8329,
      "step": 1865
    },
    {
      "epoch": 0.630192502532928,
      "grad_norm": 2.0898401737213135,
      "learning_rate": 6.059124078161929e-06,
      "loss": 1.0016,
      "step": 1866
    },
    {
      "epoch": 0.6305302262749072,
      "grad_norm": 2.052664041519165,
      "learning_rate": 6.049341993887743e-06,
      "loss": 0.9163,
      "step": 1867
    },
    {
      "epoch": 0.6308679500168862,
      "grad_norm": 2.04396915435791,
      "learning_rate": 6.039564387063315e-06,
      "loss": 0.9511,
      "step": 1868
    },
    {
      "epoch": 0.6312056737588653,
      "grad_norm": 2.4519169330596924,
      "learning_rate": 6.029791268770029e-06,
      "loss": 0.906,
      "step": 1869
    },
    {
      "epoch": 0.6315433975008443,
      "grad_norm": 2.272944450378418,
      "learning_rate": 6.020022650084176e-06,
      "loss": 0.9736,
      "step": 1870
    },
    {
      "epoch": 0.6318811212428234,
      "grad_norm": 2.854095458984375,
      "learning_rate": 6.0102585420769475e-06,
      "loss": 0.7372,
      "step": 1871
    },
    {
      "epoch": 0.6322188449848024,
      "grad_norm": 2.1190836429595947,
      "learning_rate": 6.000498955814426e-06,
      "loss": 1.0144,
      "step": 1872
    },
    {
      "epoch": 0.6325565687267815,
      "grad_norm": 2.312664747238159,
      "learning_rate": 5.990743902357571e-06,
      "loss": 0.8996,
      "step": 1873
    },
    {
      "epoch": 0.6328942924687605,
      "grad_norm": 2.1206462383270264,
      "learning_rate": 5.9809933927622e-06,
      "loss": 0.8708,
      "step": 1874
    },
    {
      "epoch": 0.6332320162107397,
      "grad_norm": 2.228893518447876,
      "learning_rate": 5.97124743807898e-06,
      "loss": 0.9679,
      "step": 1875
    },
    {
      "epoch": 0.6335697399527187,
      "grad_norm": 2.611494541168213,
      "learning_rate": 5.96150604935342e-06,
      "loss": 0.7808,
      "step": 1876
    },
    {
      "epoch": 0.6339074636946977,
      "grad_norm": 2.1471853256225586,
      "learning_rate": 5.951769237625859e-06,
      "loss": 0.8681,
      "step": 1877
    },
    {
      "epoch": 0.6342451874366768,
      "grad_norm": 1.8136999607086182,
      "learning_rate": 5.942037013931434e-06,
      "loss": 0.9307,
      "step": 1878
    },
    {
      "epoch": 0.6345829111786558,
      "grad_norm": 2.2189857959747314,
      "learning_rate": 5.932309389300092e-06,
      "loss": 0.9257,
      "step": 1879
    },
    {
      "epoch": 0.6349206349206349,
      "grad_norm": 2.330842971801758,
      "learning_rate": 5.922586374756568e-06,
      "loss": 0.9483,
      "step": 1880
    },
    {
      "epoch": 0.6352583586626139,
      "grad_norm": 2.1410586833953857,
      "learning_rate": 5.912867981320374e-06,
      "loss": 0.9295,
      "step": 1881
    },
    {
      "epoch": 0.6355960824045931,
      "grad_norm": 2.1761696338653564,
      "learning_rate": 5.903154220005771e-06,
      "loss": 0.8567,
      "step": 1882
    },
    {
      "epoch": 0.6359338061465721,
      "grad_norm": 1.9287163019180298,
      "learning_rate": 5.8934451018217916e-06,
      "loss": 0.9342,
      "step": 1883
    },
    {
      "epoch": 0.6362715298885512,
      "grad_norm": 2.1592161655426025,
      "learning_rate": 5.883740637772188e-06,
      "loss": 0.932,
      "step": 1884
    },
    {
      "epoch": 0.6366092536305302,
      "grad_norm": 2.3960421085357666,
      "learning_rate": 5.874040838855448e-06,
      "loss": 0.8524,
      "step": 1885
    },
    {
      "epoch": 0.6369469773725093,
      "grad_norm": 2.14560604095459,
      "learning_rate": 5.8643457160647634e-06,
      "loss": 0.943,
      "step": 1886
    },
    {
      "epoch": 0.6372847011144883,
      "grad_norm": 2.3031256198883057,
      "learning_rate": 5.85465528038804e-06,
      "loss": 0.8669,
      "step": 1887
    },
    {
      "epoch": 0.6376224248564674,
      "grad_norm": 2.2758572101593018,
      "learning_rate": 5.844969542807856e-06,
      "loss": 0.814,
      "step": 1888
    },
    {
      "epoch": 0.6379601485984465,
      "grad_norm": 2.2265443801879883,
      "learning_rate": 5.835288514301473e-06,
      "loss": 0.8524,
      "step": 1889
    },
    {
      "epoch": 0.6382978723404256,
      "grad_norm": 2.1327388286590576,
      "learning_rate": 5.825612205840822e-06,
      "loss": 0.8575,
      "step": 1890
    },
    {
      "epoch": 0.6386355960824046,
      "grad_norm": 2.1798312664031982,
      "learning_rate": 5.815940628392469e-06,
      "loss": 0.9767,
      "step": 1891
    },
    {
      "epoch": 0.6389733198243837,
      "grad_norm": 2.0838444232940674,
      "learning_rate": 5.806273792917629e-06,
      "loss": 0.9325,
      "step": 1892
    },
    {
      "epoch": 0.6393110435663627,
      "grad_norm": 2.278316020965576,
      "learning_rate": 5.796611710372142e-06,
      "loss": 0.835,
      "step": 1893
    },
    {
      "epoch": 0.6396487673083417,
      "grad_norm": 2.139503002166748,
      "learning_rate": 5.786954391706458e-06,
      "loss": 0.8929,
      "step": 1894
    },
    {
      "epoch": 0.6399864910503208,
      "grad_norm": 2.010955572128296,
      "learning_rate": 5.77730184786563e-06,
      "loss": 0.9782,
      "step": 1895
    },
    {
      "epoch": 0.6403242147923,
      "grad_norm": 2.068819999694824,
      "learning_rate": 5.767654089789299e-06,
      "loss": 0.872,
      "step": 1896
    },
    {
      "epoch": 0.640661938534279,
      "grad_norm": 2.307697296142578,
      "learning_rate": 5.75801112841168e-06,
      "loss": 0.946,
      "step": 1897
    },
    {
      "epoch": 0.640999662276258,
      "grad_norm": 2.10174822807312,
      "learning_rate": 5.748372974661563e-06,
      "loss": 0.9227,
      "step": 1898
    },
    {
      "epoch": 0.6413373860182371,
      "grad_norm": 2.420623779296875,
      "learning_rate": 5.738739639462267e-06,
      "loss": 0.8546,
      "step": 1899
    },
    {
      "epoch": 0.6416751097602161,
      "grad_norm": 2.4562671184539795,
      "learning_rate": 5.729111133731664e-06,
      "loss": 0.8894,
      "step": 1900
    },
    {
      "epoch": 0.6420128335021952,
      "grad_norm": 2.2119877338409424,
      "learning_rate": 5.719487468382166e-06,
      "loss": 0.8733,
      "step": 1901
    },
    {
      "epoch": 0.6423505572441742,
      "grad_norm": 2.3638525009155273,
      "learning_rate": 5.709868654320669e-06,
      "loss": 0.9449,
      "step": 1902
    },
    {
      "epoch": 0.6426882809861534,
      "grad_norm": 2.623934507369995,
      "learning_rate": 5.700254702448593e-06,
      "loss": 0.8424,
      "step": 1903
    },
    {
      "epoch": 0.6430260047281324,
      "grad_norm": 2.4989233016967773,
      "learning_rate": 5.690645623661841e-06,
      "loss": 0.8407,
      "step": 1904
    },
    {
      "epoch": 0.6433637284701115,
      "grad_norm": 2.449920892715454,
      "learning_rate": 5.681041428850793e-06,
      "loss": 0.9145,
      "step": 1905
    },
    {
      "epoch": 0.6437014522120905,
      "grad_norm": 2.0799996852874756,
      "learning_rate": 5.671442128900298e-06,
      "loss": 1.0124,
      "step": 1906
    },
    {
      "epoch": 0.6440391759540696,
      "grad_norm": 2.1359193325042725,
      "learning_rate": 5.6618477346896384e-06,
      "loss": 0.9139,
      "step": 1907
    },
    {
      "epoch": 0.6443768996960486,
      "grad_norm": 2.4779677391052246,
      "learning_rate": 5.652258257092569e-06,
      "loss": 0.8878,
      "step": 1908
    },
    {
      "epoch": 0.6447146234380277,
      "grad_norm": 1.8984113931655884,
      "learning_rate": 5.642673706977251e-06,
      "loss": 0.9169,
      "step": 1909
    },
    {
      "epoch": 0.6450523471800067,
      "grad_norm": 2.1943469047546387,
      "learning_rate": 5.633094095206254e-06,
      "loss": 0.9461,
      "step": 1910
    },
    {
      "epoch": 0.6453900709219859,
      "grad_norm": 2.405614137649536,
      "learning_rate": 5.62351943263658e-06,
      "loss": 0.8651,
      "step": 1911
    },
    {
      "epoch": 0.6457277946639649,
      "grad_norm": 2.1085307598114014,
      "learning_rate": 5.613949730119588e-06,
      "loss": 0.8848,
      "step": 1912
    },
    {
      "epoch": 0.646065518405944,
      "grad_norm": 2.6126227378845215,
      "learning_rate": 5.604384998501039e-06,
      "loss": 0.9289,
      "step": 1913
    },
    {
      "epoch": 0.646403242147923,
      "grad_norm": 2.096825122833252,
      "learning_rate": 5.594825248621051e-06,
      "loss": 0.8772,
      "step": 1914
    },
    {
      "epoch": 0.646740965889902,
      "grad_norm": 2.344665288925171,
      "learning_rate": 5.585270491314096e-06,
      "loss": 0.7947,
      "step": 1915
    },
    {
      "epoch": 0.6470786896318811,
      "grad_norm": 2.2012763023376465,
      "learning_rate": 5.575720737408994e-06,
      "loss": 0.9246,
      "step": 1916
    },
    {
      "epoch": 0.6474164133738601,
      "grad_norm": 2.4091103076934814,
      "learning_rate": 5.566175997728884e-06,
      "loss": 0.8348,
      "step": 1917
    },
    {
      "epoch": 0.6477541371158393,
      "grad_norm": 2.3580985069274902,
      "learning_rate": 5.556636283091232e-06,
      "loss": 0.9016,
      "step": 1918
    },
    {
      "epoch": 0.6480918608578183,
      "grad_norm": 2.1496996879577637,
      "learning_rate": 5.547101604307808e-06,
      "loss": 0.8842,
      "step": 1919
    },
    {
      "epoch": 0.6484295845997974,
      "grad_norm": 2.0653743743896484,
      "learning_rate": 5.537571972184661e-06,
      "loss": 0.9205,
      "step": 1920
    },
    {
      "epoch": 0.6487673083417764,
      "grad_norm": 2.009680986404419,
      "learning_rate": 5.5280473975221324e-06,
      "loss": 0.8586,
      "step": 1921
    },
    {
      "epoch": 0.6491050320837555,
      "grad_norm": 1.9056396484375,
      "learning_rate": 5.5185278911148435e-06,
      "loss": 0.9522,
      "step": 1922
    },
    {
      "epoch": 0.6494427558257345,
      "grad_norm": 2.039597511291504,
      "learning_rate": 5.509013463751645e-06,
      "loss": 0.8952,
      "step": 1923
    },
    {
      "epoch": 0.6497804795677136,
      "grad_norm": 1.9870561361312866,
      "learning_rate": 5.499504126215648e-06,
      "loss": 0.9239,
      "step": 1924
    },
    {
      "epoch": 0.6501182033096927,
      "grad_norm": 1.9598612785339355,
      "learning_rate": 5.489999889284192e-06,
      "loss": 0.8642,
      "step": 1925
    },
    {
      "epoch": 0.6504559270516718,
      "grad_norm": 1.9224967956542969,
      "learning_rate": 5.480500763728838e-06,
      "loss": 0.9597,
      "step": 1926
    },
    {
      "epoch": 0.6507936507936508,
      "grad_norm": 1.9950709342956543,
      "learning_rate": 5.471006760315352e-06,
      "loss": 0.9806,
      "step": 1927
    },
    {
      "epoch": 0.6511313745356299,
      "grad_norm": 2.5497794151306152,
      "learning_rate": 5.461517889803683e-06,
      "loss": 0.8521,
      "step": 1928
    },
    {
      "epoch": 0.6514690982776089,
      "grad_norm": 2.521477460861206,
      "learning_rate": 5.452034162947985e-06,
      "loss": 0.8866,
      "step": 1929
    },
    {
      "epoch": 0.651806822019588,
      "grad_norm": 2.50911283493042,
      "learning_rate": 5.442555590496574e-06,
      "loss": 0.8056,
      "step": 1930
    },
    {
      "epoch": 0.652144545761567,
      "grad_norm": 2.297373056411743,
      "learning_rate": 5.433082183191909e-06,
      "loss": 0.8173,
      "step": 1931
    },
    {
      "epoch": 0.6524822695035462,
      "grad_norm": 2.2929182052612305,
      "learning_rate": 5.423613951770623e-06,
      "loss": 0.9346,
      "step": 1932
    },
    {
      "epoch": 0.6528199932455252,
      "grad_norm": 2.4846928119659424,
      "learning_rate": 5.414150906963455e-06,
      "loss": 0.9262,
      "step": 1933
    },
    {
      "epoch": 0.6531577169875042,
      "grad_norm": 1.5869812965393066,
      "learning_rate": 5.404693059495285e-06,
      "loss": 0.5801,
      "step": 1934
    },
    {
      "epoch": 0.6534954407294833,
      "grad_norm": 2.7539358139038086,
      "learning_rate": 5.3952404200850964e-06,
      "loss": 0.8683,
      "step": 1935
    },
    {
      "epoch": 0.6538331644714623,
      "grad_norm": 2.150520086288452,
      "learning_rate": 5.385792999445967e-06,
      "loss": 0.9526,
      "step": 1936
    },
    {
      "epoch": 0.6541708882134414,
      "grad_norm": 2.2529349327087402,
      "learning_rate": 5.376350808285065e-06,
      "loss": 0.9936,
      "step": 1937
    },
    {
      "epoch": 0.6545086119554204,
      "grad_norm": 2.061922073364258,
      "learning_rate": 5.366913857303631e-06,
      "loss": 0.9653,
      "step": 1938
    },
    {
      "epoch": 0.6548463356973995,
      "grad_norm": 2.624249219894409,
      "learning_rate": 5.357482157196956e-06,
      "loss": 0.808,
      "step": 1939
    },
    {
      "epoch": 0.6551840594393786,
      "grad_norm": 2.4176673889160156,
      "learning_rate": 5.348055718654406e-06,
      "loss": 0.8513,
      "step": 1940
    },
    {
      "epoch": 0.6555217831813577,
      "grad_norm": 2.1617352962493896,
      "learning_rate": 5.338634552359353e-06,
      "loss": 0.9699,
      "step": 1941
    },
    {
      "epoch": 0.6558595069233367,
      "grad_norm": 2.1925055980682373,
      "learning_rate": 5.3292186689892135e-06,
      "loss": 0.8819,
      "step": 1942
    },
    {
      "epoch": 0.6561972306653158,
      "grad_norm": 2.205291986465454,
      "learning_rate": 5.319808079215409e-06,
      "loss": 0.88,
      "step": 1943
    },
    {
      "epoch": 0.6565349544072948,
      "grad_norm": 2.2272748947143555,
      "learning_rate": 5.310402793703365e-06,
      "loss": 0.9325,
      "step": 1944
    },
    {
      "epoch": 0.6568726781492739,
      "grad_norm": 2.0562326908111572,
      "learning_rate": 5.301002823112497e-06,
      "loss": 0.9045,
      "step": 1945
    },
    {
      "epoch": 0.6572104018912529,
      "grad_norm": 2.0758743286132812,
      "learning_rate": 5.291608178096183e-06,
      "loss": 0.8873,
      "step": 1946
    },
    {
      "epoch": 0.6575481256332321,
      "grad_norm": 2.0963387489318848,
      "learning_rate": 5.282218869301788e-06,
      "loss": 0.9491,
      "step": 1947
    },
    {
      "epoch": 0.6578858493752111,
      "grad_norm": 1.869484782218933,
      "learning_rate": 5.272834907370618e-06,
      "loss": 0.9077,
      "step": 1948
    },
    {
      "epoch": 0.6582235731171902,
      "grad_norm": 2.1737680435180664,
      "learning_rate": 5.263456302937904e-06,
      "loss": 0.894,
      "step": 1949
    },
    {
      "epoch": 0.6585612968591692,
      "grad_norm": 2.090510606765747,
      "learning_rate": 5.254083066632838e-06,
      "loss": 0.8421,
      "step": 1950
    },
    {
      "epoch": 0.6588990206011482,
      "grad_norm": 1.9546269178390503,
      "learning_rate": 5.244715209078498e-06,
      "loss": 0.8708,
      "step": 1951
    },
    {
      "epoch": 0.6592367443431273,
      "grad_norm": 2.332157611846924,
      "learning_rate": 5.235352740891882e-06,
      "loss": 0.8355,
      "step": 1952
    },
    {
      "epoch": 0.6595744680851063,
      "grad_norm": 2.038799285888672,
      "learning_rate": 5.225995672683872e-06,
      "loss": 0.8495,
      "step": 1953
    },
    {
      "epoch": 0.6599121918270855,
      "grad_norm": 2.2928173542022705,
      "learning_rate": 5.2166440150592375e-06,
      "loss": 0.8711,
      "step": 1954
    },
    {
      "epoch": 0.6602499155690645,
      "grad_norm": 2.093309164047241,
      "learning_rate": 5.207297778616612e-06,
      "loss": 0.886,
      "step": 1955
    },
    {
      "epoch": 0.6605876393110436,
      "grad_norm": 2.240039110183716,
      "learning_rate": 5.1979569739484835e-06,
      "loss": 0.9417,
      "step": 1956
    },
    {
      "epoch": 0.6609253630530226,
      "grad_norm": 2.0255680084228516,
      "learning_rate": 5.188621611641188e-06,
      "loss": 0.9462,
      "step": 1957
    },
    {
      "epoch": 0.6612630867950017,
      "grad_norm": 2.316746234893799,
      "learning_rate": 5.179291702274894e-06,
      "loss": 0.8784,
      "step": 1958
    },
    {
      "epoch": 0.6616008105369807,
      "grad_norm": 2.371103048324585,
      "learning_rate": 5.169967256423576e-06,
      "loss": 0.8154,
      "step": 1959
    },
    {
      "epoch": 0.6619385342789598,
      "grad_norm": 2.3669769763946533,
      "learning_rate": 5.160648284655032e-06,
      "loss": 0.9756,
      "step": 1960
    },
    {
      "epoch": 0.6622762580209389,
      "grad_norm": 2.140087842941284,
      "learning_rate": 5.15133479753086e-06,
      "loss": 0.8717,
      "step": 1961
    },
    {
      "epoch": 0.662613981762918,
      "grad_norm": 2.565234422683716,
      "learning_rate": 5.1420268056064246e-06,
      "loss": 0.9612,
      "step": 1962
    },
    {
      "epoch": 0.662951705504897,
      "grad_norm": 2.2368409633636475,
      "learning_rate": 5.132724319430873e-06,
      "loss": 0.8437,
      "step": 1963
    },
    {
      "epoch": 0.6632894292468761,
      "grad_norm": 2.1018495559692383,
      "learning_rate": 5.123427349547112e-06,
      "loss": 0.8392,
      "step": 1964
    },
    {
      "epoch": 0.6636271529888551,
      "grad_norm": 2.220388174057007,
      "learning_rate": 5.114135906491795e-06,
      "loss": 0.9297,
      "step": 1965
    },
    {
      "epoch": 0.6639648767308342,
      "grad_norm": 2.325301170349121,
      "learning_rate": 5.104850000795314e-06,
      "loss": 0.8377,
      "step": 1966
    },
    {
      "epoch": 0.6643026004728132,
      "grad_norm": 2.1345479488372803,
      "learning_rate": 5.0955696429817745e-06,
      "loss": 0.9449,
      "step": 1967
    },
    {
      "epoch": 0.6646403242147924,
      "grad_norm": 2.860790729522705,
      "learning_rate": 5.0862948435690155e-06,
      "loss": 0.88,
      "step": 1968
    },
    {
      "epoch": 0.6649780479567714,
      "grad_norm": 2.103815793991089,
      "learning_rate": 5.077025613068559e-06,
      "loss": 0.9449,
      "step": 1969
    },
    {
      "epoch": 0.6653157716987504,
      "grad_norm": 2.687592029571533,
      "learning_rate": 5.067761961985613e-06,
      "loss": 0.8061,
      "step": 1970
    },
    {
      "epoch": 0.6656534954407295,
      "grad_norm": 2.2916858196258545,
      "learning_rate": 5.058503900819085e-06,
      "loss": 0.8364,
      "step": 1971
    },
    {
      "epoch": 0.6659912191827085,
      "grad_norm": 2.176361560821533,
      "learning_rate": 5.049251440061519e-06,
      "loss": 0.9706,
      "step": 1972
    },
    {
      "epoch": 0.6663289429246876,
      "grad_norm": 2.380934953689575,
      "learning_rate": 5.040004590199128e-06,
      "loss": 0.86,
      "step": 1973
    },
    {
      "epoch": 0.6666666666666666,
      "grad_norm": 2.7465901374816895,
      "learning_rate": 5.030763361711766e-06,
      "loss": 0.792,
      "step": 1974
    },
    {
      "epoch": 0.6670043904086457,
      "grad_norm": 2.2876007556915283,
      "learning_rate": 5.0215277650729074e-06,
      "loss": 0.9254,
      "step": 1975
    },
    {
      "epoch": 0.6673421141506248,
      "grad_norm": 2.096707344055176,
      "learning_rate": 5.0122978107496535e-06,
      "loss": 0.9621,
      "step": 1976
    },
    {
      "epoch": 0.6676798378926039,
      "grad_norm": 2.0514087677001953,
      "learning_rate": 5.003073509202704e-06,
      "loss": 0.8807,
      "step": 1977
    },
    {
      "epoch": 0.6680175616345829,
      "grad_norm": 2.1223366260528564,
      "learning_rate": 4.993854870886355e-06,
      "loss": 0.9959,
      "step": 1978
    },
    {
      "epoch": 0.668355285376562,
      "grad_norm": 2.675635814666748,
      "learning_rate": 4.98464190624849e-06,
      "loss": 0.808,
      "step": 1979
    },
    {
      "epoch": 0.668693009118541,
      "grad_norm": 2.353671073913574,
      "learning_rate": 4.975434625730544e-06,
      "loss": 0.8944,
      "step": 1980
    },
    {
      "epoch": 0.6690307328605201,
      "grad_norm": 2.232428550720215,
      "learning_rate": 4.9662330397675254e-06,
      "loss": 0.8574,
      "step": 1981
    },
    {
      "epoch": 0.6693684566024991,
      "grad_norm": 2.210773468017578,
      "learning_rate": 4.957037158787995e-06,
      "loss": 0.8067,
      "step": 1982
    },
    {
      "epoch": 0.6697061803444783,
      "grad_norm": 2.585771083831787,
      "learning_rate": 4.947846993214028e-06,
      "loss": 0.7682,
      "step": 1983
    },
    {
      "epoch": 0.6700439040864573,
      "grad_norm": 2.355372667312622,
      "learning_rate": 4.9386625534612355e-06,
      "loss": 0.7989,
      "step": 1984
    },
    {
      "epoch": 0.6703816278284364,
      "grad_norm": 2.4133353233337402,
      "learning_rate": 4.9294838499387354e-06,
      "loss": 0.9204,
      "step": 1985
    },
    {
      "epoch": 0.6707193515704154,
      "grad_norm": 2.690869092941284,
      "learning_rate": 4.920310893049146e-06,
      "loss": 0.7141,
      "step": 1986
    },
    {
      "epoch": 0.6710570753123944,
      "grad_norm": 2.162081480026245,
      "learning_rate": 4.911143693188576e-06,
      "loss": 0.9089,
      "step": 1987
    },
    {
      "epoch": 0.6713947990543735,
      "grad_norm": 2.0533034801483154,
      "learning_rate": 4.901982260746593e-06,
      "loss": 0.8047,
      "step": 1988
    },
    {
      "epoch": 0.6717325227963525,
      "grad_norm": 2.6076178550720215,
      "learning_rate": 4.892826606106251e-06,
      "loss": 0.8378,
      "step": 1989
    },
    {
      "epoch": 0.6720702465383317,
      "grad_norm": 2.3445820808410645,
      "learning_rate": 4.883676739644047e-06,
      "loss": 0.8989,
      "step": 1990
    },
    {
      "epoch": 0.6724079702803107,
      "grad_norm": 2.4133026599884033,
      "learning_rate": 4.874532671729904e-06,
      "loss": 0.97,
      "step": 1991
    },
    {
      "epoch": 0.6727456940222898,
      "grad_norm": 2.3604164123535156,
      "learning_rate": 4.8653944127272e-06,
      "loss": 0.9019,
      "step": 1992
    },
    {
      "epoch": 0.6730834177642688,
      "grad_norm": 2.3890974521636963,
      "learning_rate": 4.856261972992705e-06,
      "loss": 0.9437,
      "step": 1993
    },
    {
      "epoch": 0.6734211415062479,
      "grad_norm": 2.5097169876098633,
      "learning_rate": 4.847135362876606e-06,
      "loss": 0.7619,
      "step": 1994
    },
    {
      "epoch": 0.6737588652482269,
      "grad_norm": 2.2338852882385254,
      "learning_rate": 4.838014592722482e-06,
      "loss": 0.907,
      "step": 1995
    },
    {
      "epoch": 0.674096588990206,
      "grad_norm": 2.476188898086548,
      "learning_rate": 4.828899672867291e-06,
      "loss": 0.9119,
      "step": 1996
    },
    {
      "epoch": 0.6744343127321851,
      "grad_norm": 2.193742036819458,
      "learning_rate": 4.819790613641362e-06,
      "loss": 0.9252,
      "step": 1997
    },
    {
      "epoch": 0.6747720364741642,
      "grad_norm": 2.1354148387908936,
      "learning_rate": 4.810687425368382e-06,
      "loss": 0.8706,
      "step": 1998
    },
    {
      "epoch": 0.6751097602161432,
      "grad_norm": 2.230577230453491,
      "learning_rate": 4.801590118365384e-06,
      "loss": 0.8606,
      "step": 1999
    },
    {
      "epoch": 0.6754474839581223,
      "grad_norm": 2.118468999862671,
      "learning_rate": 4.7924987029427375e-06,
      "loss": 0.9167,
      "step": 2000
    },
    {
      "epoch": 0.6757852077001013,
      "grad_norm": 2.478400945663452,
      "learning_rate": 4.7834131894041255e-06,
      "loss": 0.8832,
      "step": 2001
    },
    {
      "epoch": 0.6761229314420804,
      "grad_norm": 2.1356163024902344,
      "learning_rate": 4.774333588046554e-06,
      "loss": 0.9041,
      "step": 2002
    },
    {
      "epoch": 0.6764606551840594,
      "grad_norm": 2.0840580463409424,
      "learning_rate": 4.765259909160325e-06,
      "loss": 0.9193,
      "step": 2003
    },
    {
      "epoch": 0.6767983789260384,
      "grad_norm": 2.1799819469451904,
      "learning_rate": 4.756192163029024e-06,
      "loss": 0.8874,
      "step": 2004
    },
    {
      "epoch": 0.6771361026680176,
      "grad_norm": 2.249936819076538,
      "learning_rate": 4.747130359929517e-06,
      "loss": 0.9209,
      "step": 2005
    },
    {
      "epoch": 0.6774738264099966,
      "grad_norm": 2.4650964736938477,
      "learning_rate": 4.7380745101319335e-06,
      "loss": 0.8639,
      "step": 2006
    },
    {
      "epoch": 0.6778115501519757,
      "grad_norm": 2.2307426929473877,
      "learning_rate": 4.729024623899657e-06,
      "loss": 0.9025,
      "step": 2007
    },
    {
      "epoch": 0.6781492738939547,
      "grad_norm": 2.021679162979126,
      "learning_rate": 4.71998071148931e-06,
      "loss": 0.8778,
      "step": 2008
    },
    {
      "epoch": 0.6784869976359338,
      "grad_norm": 2.2024154663085938,
      "learning_rate": 4.710942783150738e-06,
      "loss": 0.8617,
      "step": 2009
    },
    {
      "epoch": 0.6788247213779128,
      "grad_norm": 1.9139302968978882,
      "learning_rate": 4.701910849127026e-06,
      "loss": 0.9437,
      "step": 2010
    },
    {
      "epoch": 0.6791624451198919,
      "grad_norm": 2.542754888534546,
      "learning_rate": 4.69288491965444e-06,
      "loss": 0.9987,
      "step": 2011
    },
    {
      "epoch": 0.679500168861871,
      "grad_norm": 2.19413685798645,
      "learning_rate": 4.683865004962452e-06,
      "loss": 0.8842,
      "step": 2012
    },
    {
      "epoch": 0.6798378926038501,
      "grad_norm": 2.5622613430023193,
      "learning_rate": 4.674851115273729e-06,
      "loss": 0.8796,
      "step": 2013
    },
    {
      "epoch": 0.6801756163458291,
      "grad_norm": 2.1969311237335205,
      "learning_rate": 4.665843260804086e-06,
      "loss": 0.931,
      "step": 2014
    },
    {
      "epoch": 0.6805133400878082,
      "grad_norm": 2.3127245903015137,
      "learning_rate": 4.656841451762514e-06,
      "loss": 0.8276,
      "step": 2015
    },
    {
      "epoch": 0.6808510638297872,
      "grad_norm": 2.048332929611206,
      "learning_rate": 4.647845698351147e-06,
      "loss": 0.9337,
      "step": 2016
    },
    {
      "epoch": 0.6811887875717663,
      "grad_norm": 2.172200918197632,
      "learning_rate": 4.638856010765259e-06,
      "loss": 0.8244,
      "step": 2017
    },
    {
      "epoch": 0.6815265113137453,
      "grad_norm": 1.9820847511291504,
      "learning_rate": 4.62987239919325e-06,
      "loss": 0.8645,
      "step": 2018
    },
    {
      "epoch": 0.6818642350557245,
      "grad_norm": 2.090216636657715,
      "learning_rate": 4.620894873816626e-06,
      "loss": 0.8759,
      "step": 2019
    },
    {
      "epoch": 0.6822019587977035,
      "grad_norm": 2.495551586151123,
      "learning_rate": 4.6119234448099996e-06,
      "loss": 0.809,
      "step": 2020
    },
    {
      "epoch": 0.6825396825396826,
      "grad_norm": 2.276416063308716,
      "learning_rate": 4.602958122341084e-06,
      "loss": 0.9242,
      "step": 2021
    },
    {
      "epoch": 0.6828774062816616,
      "grad_norm": 2.243605852127075,
      "learning_rate": 4.593998916570656e-06,
      "loss": 0.9143,
      "step": 2022
    },
    {
      "epoch": 0.6832151300236406,
      "grad_norm": 2.319047212600708,
      "learning_rate": 4.585045837652566e-06,
      "loss": 0.8912,
      "step": 2023
    },
    {
      "epoch": 0.6835528537656197,
      "grad_norm": 2.3731746673583984,
      "learning_rate": 4.5760988957337245e-06,
      "loss": 0.8074,
      "step": 2024
    },
    {
      "epoch": 0.6838905775075987,
      "grad_norm": 2.1798858642578125,
      "learning_rate": 4.567158100954084e-06,
      "loss": 0.8782,
      "step": 2025
    },
    {
      "epoch": 0.6842283012495779,
      "grad_norm": 2.209730625152588,
      "learning_rate": 4.558223463446629e-06,
      "loss": 0.9051,
      "step": 2026
    },
    {
      "epoch": 0.6845660249915569,
      "grad_norm": 2.3937106132507324,
      "learning_rate": 4.549294993337361e-06,
      "loss": 0.9114,
      "step": 2027
    },
    {
      "epoch": 0.684903748733536,
      "grad_norm": 2.17600679397583,
      "learning_rate": 4.540372700745305e-06,
      "loss": 0.9306,
      "step": 2028
    },
    {
      "epoch": 0.685241472475515,
      "grad_norm": 2.387052059173584,
      "learning_rate": 4.531456595782479e-06,
      "loss": 0.923,
      "step": 2029
    },
    {
      "epoch": 0.6855791962174941,
      "grad_norm": 2.229837417602539,
      "learning_rate": 4.5225466885538734e-06,
      "loss": 0.8996,
      "step": 2030
    },
    {
      "epoch": 0.6859169199594731,
      "grad_norm": 1.9835578203201294,
      "learning_rate": 4.513642989157483e-06,
      "loss": 0.8958,
      "step": 2031
    },
    {
      "epoch": 0.6862546437014522,
      "grad_norm": 2.4297168254852295,
      "learning_rate": 4.504745507684239e-06,
      "loss": 0.9263,
      "step": 2032
    },
    {
      "epoch": 0.6865923674434313,
      "grad_norm": 2.437748908996582,
      "learning_rate": 4.495854254218043e-06,
      "loss": 0.8023,
      "step": 2033
    },
    {
      "epoch": 0.6869300911854104,
      "grad_norm": 2.310101270675659,
      "learning_rate": 4.486969238835732e-06,
      "loss": 0.8752,
      "step": 2034
    },
    {
      "epoch": 0.6872678149273894,
      "grad_norm": 2.5739047527313232,
      "learning_rate": 4.478090471607075e-06,
      "loss": 0.8536,
      "step": 2035
    },
    {
      "epoch": 0.6876055386693685,
      "grad_norm": 2.4084553718566895,
      "learning_rate": 4.469217962594759e-06,
      "loss": 0.8544,
      "step": 2036
    },
    {
      "epoch": 0.6879432624113475,
      "grad_norm": 2.1092429161071777,
      "learning_rate": 4.460351721854377e-06,
      "loss": 0.8905,
      "step": 2037
    },
    {
      "epoch": 0.6882809861533266,
      "grad_norm": 2.0682880878448486,
      "learning_rate": 4.4514917594344186e-06,
      "loss": 0.862,
      "step": 2038
    },
    {
      "epoch": 0.6886187098953056,
      "grad_norm": 2.272036552429199,
      "learning_rate": 4.442638085376264e-06,
      "loss": 0.9461,
      "step": 2039
    },
    {
      "epoch": 0.6889564336372846,
      "grad_norm": 2.1004793643951416,
      "learning_rate": 4.433790709714152e-06,
      "loss": 0.8774,
      "step": 2040
    },
    {
      "epoch": 0.6892941573792638,
      "grad_norm": 2.205986499786377,
      "learning_rate": 4.4249496424751915e-06,
      "loss": 0.9224,
      "step": 2041
    },
    {
      "epoch": 0.6896318811212429,
      "grad_norm": 1.9250695705413818,
      "learning_rate": 4.416114893679355e-06,
      "loss": 0.9417,
      "step": 2042
    },
    {
      "epoch": 0.6899696048632219,
      "grad_norm": 2.1743650436401367,
      "learning_rate": 4.4072864733394295e-06,
      "loss": 0.8808,
      "step": 2043
    },
    {
      "epoch": 0.6903073286052009,
      "grad_norm": 2.3830106258392334,
      "learning_rate": 4.398464391461045e-06,
      "loss": 0.91,
      "step": 2044
    },
    {
      "epoch": 0.69064505234718,
      "grad_norm": 2.2202558517456055,
      "learning_rate": 4.389648658042644e-06,
      "loss": 1.0179,
      "step": 2045
    },
    {
      "epoch": 0.690982776089159,
      "grad_norm": 2.1960337162017822,
      "learning_rate": 4.3808392830754765e-06,
      "loss": 0.9118,
      "step": 2046
    },
    {
      "epoch": 0.6913204998311381,
      "grad_norm": 2.2039921283721924,
      "learning_rate": 4.372036276543586e-06,
      "loss": 0.939,
      "step": 2047
    },
    {
      "epoch": 0.6916582235731172,
      "grad_norm": 2.4202194213867188,
      "learning_rate": 4.363239648423784e-06,
      "loss": 0.8723,
      "step": 2048
    },
    {
      "epoch": 0.6919959473150963,
      "grad_norm": 2.357401132583618,
      "learning_rate": 4.354449408685679e-06,
      "loss": 0.8502,
      "step": 2049
    },
    {
      "epoch": 0.6923336710570753,
      "grad_norm": 2.4816739559173584,
      "learning_rate": 4.345665567291625e-06,
      "loss": 0.8424,
      "step": 2050
    },
    {
      "epoch": 0.6926713947990544,
      "grad_norm": 2.231994390487671,
      "learning_rate": 4.3368881341967135e-06,
      "loss": 0.9155,
      "step": 2051
    },
    {
      "epoch": 0.6930091185410334,
      "grad_norm": 2.37893009185791,
      "learning_rate": 4.3281171193488015e-06,
      "loss": 0.9651,
      "step": 2052
    },
    {
      "epoch": 0.6933468422830125,
      "grad_norm": 2.3825724124908447,
      "learning_rate": 4.319352532688444e-06,
      "loss": 0.8329,
      "step": 2053
    },
    {
      "epoch": 0.6936845660249915,
      "grad_norm": 2.0765316486358643,
      "learning_rate": 4.310594384148924e-06,
      "loss": 0.9349,
      "step": 2054
    },
    {
      "epoch": 0.6940222897669707,
      "grad_norm": 2.1223695278167725,
      "learning_rate": 4.301842683656228e-06,
      "loss": 0.8875,
      "step": 2055
    },
    {
      "epoch": 0.6943600135089497,
      "grad_norm": 2.5488202571868896,
      "learning_rate": 4.29309744112903e-06,
      "loss": 0.8402,
      "step": 2056
    },
    {
      "epoch": 0.6946977372509288,
      "grad_norm": 2.3655552864074707,
      "learning_rate": 4.28435866647869e-06,
      "loss": 0.8582,
      "step": 2057
    },
    {
      "epoch": 0.6950354609929078,
      "grad_norm": 2.3298375606536865,
      "learning_rate": 4.275626369609236e-06,
      "loss": 0.9426,
      "step": 2058
    },
    {
      "epoch": 0.6953731847348869,
      "grad_norm": 2.0038905143737793,
      "learning_rate": 4.266900560417351e-06,
      "loss": 0.8374,
      "step": 2059
    },
    {
      "epoch": 0.6957109084768659,
      "grad_norm": 2.532590627670288,
      "learning_rate": 4.258181248792374e-06,
      "loss": 0.8985,
      "step": 2060
    },
    {
      "epoch": 0.6960486322188449,
      "grad_norm": 2.024226665496826,
      "learning_rate": 4.249468444616264e-06,
      "loss": 0.9628,
      "step": 2061
    },
    {
      "epoch": 0.6963863559608241,
      "grad_norm": 2.2409470081329346,
      "learning_rate": 4.2407621577636205e-06,
      "loss": 0.8115,
      "step": 2062
    },
    {
      "epoch": 0.6967240797028031,
      "grad_norm": 1.981762170791626,
      "learning_rate": 4.23206239810165e-06,
      "loss": 0.9469,
      "step": 2063
    },
    {
      "epoch": 0.6970618034447822,
      "grad_norm": 2.087700843811035,
      "learning_rate": 4.223369175490162e-06,
      "loss": 0.8501,
      "step": 2064
    },
    {
      "epoch": 0.6973995271867612,
      "grad_norm": 2.3302414417266846,
      "learning_rate": 4.21468249978156e-06,
      "loss": 0.8852,
      "step": 2065
    },
    {
      "epoch": 0.6977372509287403,
      "grad_norm": 2.5850419998168945,
      "learning_rate": 4.206002380820821e-06,
      "loss": 0.8991,
      "step": 2066
    },
    {
      "epoch": 0.6980749746707193,
      "grad_norm": 2.2061843872070312,
      "learning_rate": 4.197328828445501e-06,
      "loss": 0.9875,
      "step": 2067
    },
    {
      "epoch": 0.6984126984126984,
      "grad_norm": 2.07354474067688,
      "learning_rate": 4.188661852485706e-06,
      "loss": 0.8496,
      "step": 2068
    },
    {
      "epoch": 0.6987504221546774,
      "grad_norm": 2.35353422164917,
      "learning_rate": 4.180001462764082e-06,
      "loss": 0.869,
      "step": 2069
    },
    {
      "epoch": 0.6990881458966566,
      "grad_norm": 2.2031638622283936,
      "learning_rate": 4.171347669095835e-06,
      "loss": 0.9205,
      "step": 2070
    },
    {
      "epoch": 0.6994258696386356,
      "grad_norm": 2.430985927581787,
      "learning_rate": 4.162700481288668e-06,
      "loss": 0.9132,
      "step": 2071
    },
    {
      "epoch": 0.6997635933806147,
      "grad_norm": 2.4217703342437744,
      "learning_rate": 4.154059909142809e-06,
      "loss": 0.8169,
      "step": 2072
    },
    {
      "epoch": 0.7001013171225937,
      "grad_norm": 2.5649189949035645,
      "learning_rate": 4.1454259624509986e-06,
      "loss": 0.8556,
      "step": 2073
    },
    {
      "epoch": 0.7004390408645728,
      "grad_norm": 1.941725492477417,
      "learning_rate": 4.136798650998446e-06,
      "loss": 0.9726,
      "step": 2074
    },
    {
      "epoch": 0.7007767646065518,
      "grad_norm": 2.7518253326416016,
      "learning_rate": 4.128177984562859e-06,
      "loss": 0.9036,
      "step": 2075
    },
    {
      "epoch": 0.7011144883485309,
      "grad_norm": 2.128253936767578,
      "learning_rate": 4.119563972914405e-06,
      "loss": 0.9237,
      "step": 2076
    },
    {
      "epoch": 0.70145221209051,
      "grad_norm": 2.499741792678833,
      "learning_rate": 4.110956625815713e-06,
      "loss": 0.9372,
      "step": 2077
    },
    {
      "epoch": 0.701789935832489,
      "grad_norm": 2.266214609146118,
      "learning_rate": 4.102355953021862e-06,
      "loss": 0.7857,
      "step": 2078
    },
    {
      "epoch": 0.7021276595744681,
      "grad_norm": 2.182671070098877,
      "learning_rate": 4.0937619642803495e-06,
      "loss": 0.9183,
      "step": 2079
    },
    {
      "epoch": 0.7024653833164471,
      "grad_norm": 2.6268868446350098,
      "learning_rate": 4.0851746693311244e-06,
      "loss": 0.9031,
      "step": 2080
    },
    {
      "epoch": 0.7028031070584262,
      "grad_norm": 2.3605175018310547,
      "learning_rate": 4.076594077906534e-06,
      "loss": 0.8408,
      "step": 2081
    },
    {
      "epoch": 0.7031408308004052,
      "grad_norm": 2.287648916244507,
      "learning_rate": 4.0680201997313214e-06,
      "loss": 0.8284,
      "step": 2082
    },
    {
      "epoch": 0.7034785545423843,
      "grad_norm": 2.836381435394287,
      "learning_rate": 4.059453044522636e-06,
      "loss": 0.8116,
      "step": 2083
    },
    {
      "epoch": 0.7038162782843634,
      "grad_norm": 2.3921515941619873,
      "learning_rate": 4.050892621989998e-06,
      "loss": 0.863,
      "step": 2084
    },
    {
      "epoch": 0.7041540020263425,
      "grad_norm": 2.374224901199341,
      "learning_rate": 4.042338941835304e-06,
      "loss": 0.8251,
      "step": 2085
    },
    {
      "epoch": 0.7044917257683215,
      "grad_norm": 2.2647366523742676,
      "learning_rate": 4.033792013752803e-06,
      "loss": 0.9187,
      "step": 2086
    },
    {
      "epoch": 0.7048294495103006,
      "grad_norm": 2.364478588104248,
      "learning_rate": 4.025251847429096e-06,
      "loss": 0.9564,
      "step": 2087
    },
    {
      "epoch": 0.7051671732522796,
      "grad_norm": 2.62589430809021,
      "learning_rate": 4.016718452543118e-06,
      "loss": 0.935,
      "step": 2088
    },
    {
      "epoch": 0.7055048969942587,
      "grad_norm": 1.9200074672698975,
      "learning_rate": 4.008191838766135e-06,
      "loss": 0.8858,
      "step": 2089
    },
    {
      "epoch": 0.7058426207362377,
      "grad_norm": 2.3263633251190186,
      "learning_rate": 3.9996720157617094e-06,
      "loss": 0.9793,
      "step": 2090
    },
    {
      "epoch": 0.7061803444782169,
      "grad_norm": 2.679243326187134,
      "learning_rate": 3.9911589931857406e-06,
      "loss": 0.8671,
      "step": 2091
    },
    {
      "epoch": 0.7065180682201959,
      "grad_norm": 2.1167654991149902,
      "learning_rate": 3.982652780686387e-06,
      "loss": 0.8777,
      "step": 2092
    },
    {
      "epoch": 0.706855791962175,
      "grad_norm": 2.248894214630127,
      "learning_rate": 3.974153387904104e-06,
      "loss": 0.9298,
      "step": 2093
    },
    {
      "epoch": 0.707193515704154,
      "grad_norm": 2.530035972595215,
      "learning_rate": 3.965660824471627e-06,
      "loss": 0.8946,
      "step": 2094
    },
    {
      "epoch": 0.707531239446133,
      "grad_norm": 2.2246837615966797,
      "learning_rate": 3.9571751000139336e-06,
      "loss": 0.8702,
      "step": 2095
    },
    {
      "epoch": 0.7078689631881121,
      "grad_norm": 2.0223522186279297,
      "learning_rate": 3.948696224148259e-06,
      "loss": 0.8849,
      "step": 2096
    },
    {
      "epoch": 0.7082066869300911,
      "grad_norm": 2.67899489402771,
      "learning_rate": 3.940224206484079e-06,
      "loss": 0.9487,
      "step": 2097
    },
    {
      "epoch": 0.7085444106720703,
      "grad_norm": 2.632047653198242,
      "learning_rate": 3.93175905662309e-06,
      "loss": 0.8383,
      "step": 2098
    },
    {
      "epoch": 0.7088821344140493,
      "grad_norm": 2.1614739894866943,
      "learning_rate": 3.923300784159214e-06,
      "loss": 0.9236,
      "step": 2099
    },
    {
      "epoch": 0.7092198581560284,
      "grad_norm": 2.0535237789154053,
      "learning_rate": 3.914849398678563e-06,
      "loss": 0.9999,
      "step": 2100
    },
    {
      "epoch": 0.7095575818980074,
      "grad_norm": 2.2405240535736084,
      "learning_rate": 3.9064049097594645e-06,
      "loss": 0.9119,
      "step": 2101
    },
    {
      "epoch": 0.7098953056399865,
      "grad_norm": 1.9875352382659912,
      "learning_rate": 3.897967326972418e-06,
      "loss": 0.8647,
      "step": 2102
    },
    {
      "epoch": 0.7102330293819655,
      "grad_norm": 2.188563585281372,
      "learning_rate": 3.88953665988009e-06,
      "loss": 1.0046,
      "step": 2103
    },
    {
      "epoch": 0.7105707531239446,
      "grad_norm": 2.4211080074310303,
      "learning_rate": 3.881112918037321e-06,
      "loss": 0.8546,
      "step": 2104
    },
    {
      "epoch": 0.7109084768659236,
      "grad_norm": 2.180091142654419,
      "learning_rate": 3.872696110991096e-06,
      "loss": 0.8612,
      "step": 2105
    },
    {
      "epoch": 0.7112462006079028,
      "grad_norm": 2.6614630222320557,
      "learning_rate": 3.8642862482805454e-06,
      "loss": 0.9373,
      "step": 2106
    },
    {
      "epoch": 0.7115839243498818,
      "grad_norm": 2.633596420288086,
      "learning_rate": 3.855883339436928e-06,
      "loss": 0.7795,
      "step": 2107
    },
    {
      "epoch": 0.7119216480918609,
      "grad_norm": 2.2152504920959473,
      "learning_rate": 3.847487393983609e-06,
      "loss": 0.9077,
      "step": 2108
    },
    {
      "epoch": 0.7122593718338399,
      "grad_norm": 2.2634880542755127,
      "learning_rate": 3.839098421436085e-06,
      "loss": 0.9596,
      "step": 2109
    },
    {
      "epoch": 0.712597095575819,
      "grad_norm": 2.2546422481536865,
      "learning_rate": 3.830716431301936e-06,
      "loss": 0.9469,
      "step": 2110
    },
    {
      "epoch": 0.712934819317798,
      "grad_norm": 2.2387077808380127,
      "learning_rate": 3.822341433080821e-06,
      "loss": 0.8937,
      "step": 2111
    },
    {
      "epoch": 0.713272543059777,
      "grad_norm": 1.99384343624115,
      "learning_rate": 3.8139734362644966e-06,
      "loss": 0.9242,
      "step": 2112
    },
    {
      "epoch": 0.7136102668017562,
      "grad_norm": 2.308274030685425,
      "learning_rate": 3.805612450336761e-06,
      "loss": 0.8536,
      "step": 2113
    },
    {
      "epoch": 0.7139479905437353,
      "grad_norm": 2.3821144104003906,
      "learning_rate": 3.7972584847734815e-06,
      "loss": 0.9018,
      "step": 2114
    },
    {
      "epoch": 0.7142857142857143,
      "grad_norm": 2.1594913005828857,
      "learning_rate": 3.7889115490425644e-06,
      "loss": 0.8656,
      "step": 2115
    },
    {
      "epoch": 0.7146234380276933,
      "grad_norm": 2.1017870903015137,
      "learning_rate": 3.7805716526039492e-06,
      "loss": 0.9263,
      "step": 2116
    },
    {
      "epoch": 0.7149611617696724,
      "grad_norm": 2.167569637298584,
      "learning_rate": 3.772238804909598e-06,
      "loss": 0.9262,
      "step": 2117
    },
    {
      "epoch": 0.7152988855116514,
      "grad_norm": 2.223496675491333,
      "learning_rate": 3.7639130154034833e-06,
      "loss": 0.8439,
      "step": 2118
    },
    {
      "epoch": 0.7156366092536305,
      "grad_norm": 2.4693150520324707,
      "learning_rate": 3.755594293521577e-06,
      "loss": 0.8324,
      "step": 2119
    },
    {
      "epoch": 0.7159743329956096,
      "grad_norm": 2.2335903644561768,
      "learning_rate": 3.747282648691849e-06,
      "loss": 0.9096,
      "step": 2120
    },
    {
      "epoch": 0.7163120567375887,
      "grad_norm": 2.7649967670440674,
      "learning_rate": 3.738978090334232e-06,
      "loss": 0.7821,
      "step": 2121
    },
    {
      "epoch": 0.7166497804795677,
      "grad_norm": 2.454958915710449,
      "learning_rate": 3.73068062786064e-06,
      "loss": 0.9326,
      "step": 2122
    },
    {
      "epoch": 0.7169875042215468,
      "grad_norm": 2.0256831645965576,
      "learning_rate": 3.7223902706749504e-06,
      "loss": 0.8049,
      "step": 2123
    },
    {
      "epoch": 0.7173252279635258,
      "grad_norm": 2.0120720863342285,
      "learning_rate": 3.714107028172972e-06,
      "loss": 0.8697,
      "step": 2124
    },
    {
      "epoch": 0.7176629517055049,
      "grad_norm": 2.306400775909424,
      "learning_rate": 3.705830909742458e-06,
      "loss": 0.8765,
      "step": 2125
    },
    {
      "epoch": 0.7180006754474839,
      "grad_norm": 2.711920976638794,
      "learning_rate": 3.69756192476309e-06,
      "loss": 0.8209,
      "step": 2126
    },
    {
      "epoch": 0.7183383991894631,
      "grad_norm": 2.2785186767578125,
      "learning_rate": 3.6893000826064617e-06,
      "loss": 0.9146,
      "step": 2127
    },
    {
      "epoch": 0.7186761229314421,
      "grad_norm": 2.038022756576538,
      "learning_rate": 3.6810453926360768e-06,
      "loss": 0.9396,
      "step": 2128
    },
    {
      "epoch": 0.7190138466734212,
      "grad_norm": 1.9913356304168701,
      "learning_rate": 3.672797864207316e-06,
      "loss": 0.5593,
      "step": 2129
    },
    {
      "epoch": 0.7193515704154002,
      "grad_norm": 2.3595211505889893,
      "learning_rate": 3.664557506667472e-06,
      "loss": 0.9507,
      "step": 2130
    },
    {
      "epoch": 0.7196892941573793,
      "grad_norm": 2.1933913230895996,
      "learning_rate": 3.6563243293556836e-06,
      "loss": 0.9183,
      "step": 2131
    },
    {
      "epoch": 0.7200270178993583,
      "grad_norm": 2.3389265537261963,
      "learning_rate": 3.6480983416029626e-06,
      "loss": 0.8785,
      "step": 2132
    },
    {
      "epoch": 0.7203647416413373,
      "grad_norm": 2.4853029251098633,
      "learning_rate": 3.6398795527321838e-06,
      "loss": 1.0074,
      "step": 2133
    },
    {
      "epoch": 0.7207024653833164,
      "grad_norm": 2.2510271072387695,
      "learning_rate": 3.631667972058042e-06,
      "loss": 0.8821,
      "step": 2134
    },
    {
      "epoch": 0.7210401891252955,
      "grad_norm": 2.782541275024414,
      "learning_rate": 3.6234636088870754e-06,
      "loss": 0.8118,
      "step": 2135
    },
    {
      "epoch": 0.7213779128672746,
      "grad_norm": 2.1631734371185303,
      "learning_rate": 3.615266472517641e-06,
      "loss": 0.9016,
      "step": 2136
    },
    {
      "epoch": 0.7217156366092536,
      "grad_norm": 2.5555121898651123,
      "learning_rate": 3.6070765722399037e-06,
      "loss": 0.9051,
      "step": 2137
    },
    {
      "epoch": 0.7220533603512327,
      "grad_norm": 2.265713691711426,
      "learning_rate": 3.598893917335832e-06,
      "loss": 0.8126,
      "step": 2138
    },
    {
      "epoch": 0.7223910840932117,
      "grad_norm": 2.6724672317504883,
      "learning_rate": 3.5907185170791657e-06,
      "loss": 0.8618,
      "step": 2139
    },
    {
      "epoch": 0.7227288078351908,
      "grad_norm": 2.738072156906128,
      "learning_rate": 3.58255038073545e-06,
      "loss": 0.7868,
      "step": 2140
    },
    {
      "epoch": 0.7230665315771698,
      "grad_norm": 2.135471820831299,
      "learning_rate": 3.5743895175619814e-06,
      "loss": 0.8648,
      "step": 2141
    },
    {
      "epoch": 0.723404255319149,
      "grad_norm": 2.2589986324310303,
      "learning_rate": 3.5662359368078083e-06,
      "loss": 0.9367,
      "step": 2142
    },
    {
      "epoch": 0.723741979061128,
      "grad_norm": 2.008265733718872,
      "learning_rate": 3.558089647713736e-06,
      "loss": 0.8566,
      "step": 2143
    },
    {
      "epoch": 0.7240797028031071,
      "grad_norm": 2.057110548019409,
      "learning_rate": 3.5499506595123035e-06,
      "loss": 0.9439,
      "step": 2144
    },
    {
      "epoch": 0.7244174265450861,
      "grad_norm": 2.6089847087860107,
      "learning_rate": 3.5418189814277735e-06,
      "loss": 0.8605,
      "step": 2145
    },
    {
      "epoch": 0.7247551502870652,
      "grad_norm": 2.1783204078674316,
      "learning_rate": 3.533694622676125e-06,
      "loss": 0.9008,
      "step": 2146
    },
    {
      "epoch": 0.7250928740290442,
      "grad_norm": 2.445638418197632,
      "learning_rate": 3.5255775924650416e-06,
      "loss": 0.8672,
      "step": 2147
    },
    {
      "epoch": 0.7254305977710233,
      "grad_norm": 1.9164636135101318,
      "learning_rate": 3.5174678999939003e-06,
      "loss": 0.8812,
      "step": 2148
    },
    {
      "epoch": 0.7257683215130024,
      "grad_norm": 2.0231008529663086,
      "learning_rate": 3.5093655544537666e-06,
      "loss": 0.8948,
      "step": 2149
    },
    {
      "epoch": 0.7261060452549815,
      "grad_norm": 2.4377524852752686,
      "learning_rate": 3.501270565027364e-06,
      "loss": 0.8588,
      "step": 2150
    },
    {
      "epoch": 0.7264437689969605,
      "grad_norm": 2.324773073196411,
      "learning_rate": 3.493182940889106e-06,
      "loss": 0.8366,
      "step": 2151
    },
    {
      "epoch": 0.7267814927389395,
      "grad_norm": 2.1994378566741943,
      "learning_rate": 3.485102691205031e-06,
      "loss": 0.8644,
      "step": 2152
    },
    {
      "epoch": 0.7271192164809186,
      "grad_norm": 2.114354372024536,
      "learning_rate": 3.4770298251328326e-06,
      "loss": 0.9357,
      "step": 2153
    },
    {
      "epoch": 0.7274569402228976,
      "grad_norm": 2.4573519229888916,
      "learning_rate": 3.468964351821846e-06,
      "loss": 0.892,
      "step": 2154
    },
    {
      "epoch": 0.7277946639648767,
      "grad_norm": 2.344160556793213,
      "learning_rate": 3.4609062804130066e-06,
      "loss": 0.929,
      "step": 2155
    },
    {
      "epoch": 0.7281323877068558,
      "grad_norm": 2.2475080490112305,
      "learning_rate": 3.4528556200388754e-06,
      "loss": 0.7921,
      "step": 2156
    },
    {
      "epoch": 0.7284701114488349,
      "grad_norm": 2.0246737003326416,
      "learning_rate": 3.4448123798236076e-06,
      "loss": 1.0341,
      "step": 2157
    },
    {
      "epoch": 0.7288078351908139,
      "grad_norm": 2.7074389457702637,
      "learning_rate": 3.4367765688829556e-06,
      "loss": 0.819,
      "step": 2158
    },
    {
      "epoch": 0.729145558932793,
      "grad_norm": 1.9794524908065796,
      "learning_rate": 3.4287481963242474e-06,
      "loss": 0.892,
      "step": 2159
    },
    {
      "epoch": 0.729483282674772,
      "grad_norm": 2.743108034133911,
      "learning_rate": 3.4207272712463744e-06,
      "loss": 0.8341,
      "step": 2160
    },
    {
      "epoch": 0.7298210064167511,
      "grad_norm": 2.1096954345703125,
      "learning_rate": 3.412713802739801e-06,
      "loss": 0.8924,
      "step": 2161
    },
    {
      "epoch": 0.7301587301587301,
      "grad_norm": 2.255678176879883,
      "learning_rate": 3.4047077998865362e-06,
      "loss": 0.9031,
      "step": 2162
    },
    {
      "epoch": 0.7304964539007093,
      "grad_norm": 2.161868095397949,
      "learning_rate": 3.396709271760119e-06,
      "loss": 0.9202,
      "step": 2163
    },
    {
      "epoch": 0.7308341776426883,
      "grad_norm": 2.4716079235076904,
      "learning_rate": 3.388718227425626e-06,
      "loss": 0.7917,
      "step": 2164
    },
    {
      "epoch": 0.7311719013846674,
      "grad_norm": 2.205186128616333,
      "learning_rate": 3.3807346759396486e-06,
      "loss": 0.9246,
      "step": 2165
    },
    {
      "epoch": 0.7315096251266464,
      "grad_norm": 2.145702600479126,
      "learning_rate": 3.37275862635029e-06,
      "loss": 0.9223,
      "step": 2166
    },
    {
      "epoch": 0.7318473488686255,
      "grad_norm": 2.035405397415161,
      "learning_rate": 3.3647900876971463e-06,
      "loss": 0.903,
      "step": 2167
    },
    {
      "epoch": 0.7321850726106045,
      "grad_norm": 2.6831305027008057,
      "learning_rate": 3.3568290690113037e-06,
      "loss": 0.8774,
      "step": 2168
    },
    {
      "epoch": 0.7325227963525835,
      "grad_norm": 2.5106282234191895,
      "learning_rate": 3.348875579315326e-06,
      "loss": 0.8223,
      "step": 2169
    },
    {
      "epoch": 0.7328605200945626,
      "grad_norm": 2.112215995788574,
      "learning_rate": 3.340929627623246e-06,
      "loss": 0.8575,
      "step": 2170
    },
    {
      "epoch": 0.7331982438365418,
      "grad_norm": 2.075366973876953,
      "learning_rate": 3.3329912229405427e-06,
      "loss": 0.9132,
      "step": 2171
    },
    {
      "epoch": 0.7335359675785208,
      "grad_norm": 2.060065746307373,
      "learning_rate": 3.3250603742641597e-06,
      "loss": 0.9012,
      "step": 2172
    },
    {
      "epoch": 0.7338736913204998,
      "grad_norm": 2.305737257003784,
      "learning_rate": 3.3171370905824597e-06,
      "loss": 0.9544,
      "step": 2173
    },
    {
      "epoch": 0.7342114150624789,
      "grad_norm": 2.243670701980591,
      "learning_rate": 3.3092213808752384e-06,
      "loss": 0.8651,
      "step": 2174
    },
    {
      "epoch": 0.7345491388044579,
      "grad_norm": 2.135791778564453,
      "learning_rate": 3.3013132541137185e-06,
      "loss": 0.9172,
      "step": 2175
    },
    {
      "epoch": 0.734886862546437,
      "grad_norm": 2.787170171737671,
      "learning_rate": 3.293412719260507e-06,
      "loss": 0.8792,
      "step": 2176
    },
    {
      "epoch": 0.735224586288416,
      "grad_norm": 2.2687695026397705,
      "learning_rate": 3.285519785269624e-06,
      "loss": 0.8844,
      "step": 2177
    },
    {
      "epoch": 0.7355623100303952,
      "grad_norm": 2.6627016067504883,
      "learning_rate": 3.2776344610864674e-06,
      "loss": 0.8123,
      "step": 2178
    },
    {
      "epoch": 0.7359000337723742,
      "grad_norm": 2.3692190647125244,
      "learning_rate": 3.2697567556478137e-06,
      "loss": 0.8013,
      "step": 2179
    },
    {
      "epoch": 0.7362377575143533,
      "grad_norm": 2.441559076309204,
      "learning_rate": 3.261886677881806e-06,
      "loss": 0.9736,
      "step": 2180
    },
    {
      "epoch": 0.7365754812563323,
      "grad_norm": 2.381810188293457,
      "learning_rate": 3.25402423670793e-06,
      "loss": 0.9432,
      "step": 2181
    },
    {
      "epoch": 0.7369132049983114,
      "grad_norm": 2.2515718936920166,
      "learning_rate": 3.246169441037037e-06,
      "loss": 0.7725,
      "step": 2182
    },
    {
      "epoch": 0.7372509287402904,
      "grad_norm": 2.0359139442443848,
      "learning_rate": 3.2383222997713036e-06,
      "loss": 0.9055,
      "step": 2183
    },
    {
      "epoch": 0.7375886524822695,
      "grad_norm": 2.6650893688201904,
      "learning_rate": 3.230482821804224e-06,
      "loss": 0.8519,
      "step": 2184
    },
    {
      "epoch": 0.7379263762242486,
      "grad_norm": 2.175403594970703,
      "learning_rate": 3.222651016020617e-06,
      "loss": 0.9422,
      "step": 2185
    },
    {
      "epoch": 0.7382640999662277,
      "grad_norm": 2.1851320266723633,
      "learning_rate": 3.214826891296603e-06,
      "loss": 0.902,
      "step": 2186
    },
    {
      "epoch": 0.7386018237082067,
      "grad_norm": 2.29379940032959,
      "learning_rate": 3.2070104564995976e-06,
      "loss": 0.9208,
      "step": 2187
    },
    {
      "epoch": 0.7389395474501858,
      "grad_norm": 2.5971128940582275,
      "learning_rate": 3.199201720488303e-06,
      "loss": 0.9964,
      "step": 2188
    },
    {
      "epoch": 0.7392772711921648,
      "grad_norm": 2.0851998329162598,
      "learning_rate": 3.191400692112686e-06,
      "loss": 0.9302,
      "step": 2189
    },
    {
      "epoch": 0.7396149949341438,
      "grad_norm": 2.1983766555786133,
      "learning_rate": 3.1836073802139944e-06,
      "loss": 0.8825,
      "step": 2190
    },
    {
      "epoch": 0.7399527186761229,
      "grad_norm": 2.304858922958374,
      "learning_rate": 3.1758217936247227e-06,
      "loss": 0.8255,
      "step": 2191
    },
    {
      "epoch": 0.740290442418102,
      "grad_norm": 2.241874933242798,
      "learning_rate": 3.1680439411685994e-06,
      "loss": 0.84,
      "step": 2192
    },
    {
      "epoch": 0.7406281661600811,
      "grad_norm": 2.1936841011047363,
      "learning_rate": 3.1602738316606096e-06,
      "loss": 0.911,
      "step": 2193
    },
    {
      "epoch": 0.7409658899020601,
      "grad_norm": 2.094891309738159,
      "learning_rate": 3.1525114739069418e-06,
      "loss": 0.8588,
      "step": 2194
    },
    {
      "epoch": 0.7413036136440392,
      "grad_norm": 2.4792709350585938,
      "learning_rate": 3.144756876705012e-06,
      "loss": 0.9249,
      "step": 2195
    },
    {
      "epoch": 0.7416413373860182,
      "grad_norm": 2.0860207080841064,
      "learning_rate": 3.1370100488434363e-06,
      "loss": 0.8588,
      "step": 2196
    },
    {
      "epoch": 0.7419790611279973,
      "grad_norm": 2.0951738357543945,
      "learning_rate": 3.1292709991020266e-06,
      "loss": 0.919,
      "step": 2197
    },
    {
      "epoch": 0.7423167848699763,
      "grad_norm": 2.1614620685577393,
      "learning_rate": 3.121539736251781e-06,
      "loss": 0.994,
      "step": 2198
    },
    {
      "epoch": 0.7426545086119554,
      "grad_norm": 2.389741897583008,
      "learning_rate": 3.1138162690548625e-06,
      "loss": 0.8624,
      "step": 2199
    },
    {
      "epoch": 0.7429922323539345,
      "grad_norm": 2.365753412246704,
      "learning_rate": 3.106100606264616e-06,
      "loss": 0.8495,
      "step": 2200
    },
    {
      "epoch": 0.7433299560959136,
      "grad_norm": 2.780808687210083,
      "learning_rate": 3.0983927566255324e-06,
      "loss": 0.8159,
      "step": 2201
    },
    {
      "epoch": 0.7436676798378926,
      "grad_norm": 2.5312860012054443,
      "learning_rate": 3.090692728873241e-06,
      "loss": 0.8177,
      "step": 2202
    },
    {
      "epoch": 0.7440054035798717,
      "grad_norm": 1.9862521886825562,
      "learning_rate": 3.083000531734516e-06,
      "loss": 0.9504,
      "step": 2203
    },
    {
      "epoch": 0.7443431273218507,
      "grad_norm": 2.398495674133301,
      "learning_rate": 3.075316173927253e-06,
      "loss": 0.8261,
      "step": 2204
    },
    {
      "epoch": 0.7446808510638298,
      "grad_norm": 2.250046491622925,
      "learning_rate": 3.067639664160463e-06,
      "loss": 0.8464,
      "step": 2205
    },
    {
      "epoch": 0.7450185748058088,
      "grad_norm": 2.3744239807128906,
      "learning_rate": 3.059971011134266e-06,
      "loss": 0.9099,
      "step": 2206
    },
    {
      "epoch": 0.745356298547788,
      "grad_norm": 2.732987880706787,
      "learning_rate": 3.0523102235398716e-06,
      "loss": 0.7424,
      "step": 2207
    },
    {
      "epoch": 0.745694022289767,
      "grad_norm": 2.098093032836914,
      "learning_rate": 3.044657310059579e-06,
      "loss": 0.9096,
      "step": 2208
    },
    {
      "epoch": 0.746031746031746,
      "grad_norm": 2.0368969440460205,
      "learning_rate": 3.0370122793667658e-06,
      "loss": 0.8876,
      "step": 2209
    },
    {
      "epoch": 0.7463694697737251,
      "grad_norm": 2.1883294582366943,
      "learning_rate": 3.029375140125862e-06,
      "loss": 0.909,
      "step": 2210
    },
    {
      "epoch": 0.7467071935157041,
      "grad_norm": 2.4476590156555176,
      "learning_rate": 3.021745900992379e-06,
      "loss": 0.9218,
      "step": 2211
    },
    {
      "epoch": 0.7470449172576832,
      "grad_norm": 2.689948558807373,
      "learning_rate": 3.0141245706128487e-06,
      "loss": 0.8694,
      "step": 2212
    },
    {
      "epoch": 0.7473826409996622,
      "grad_norm": 2.4732143878936768,
      "learning_rate": 3.0065111576248516e-06,
      "loss": 0.8621,
      "step": 2213
    },
    {
      "epoch": 0.7477203647416414,
      "grad_norm": 2.2958459854125977,
      "learning_rate": 2.9989056706570043e-06,
      "loss": 0.8703,
      "step": 2214
    },
    {
      "epoch": 0.7480580884836204,
      "grad_norm": 2.2433371543884277,
      "learning_rate": 2.9913081183289195e-06,
      "loss": 0.9097,
      "step": 2215
    },
    {
      "epoch": 0.7483958122255995,
      "grad_norm": 2.265828847885132,
      "learning_rate": 2.9837185092512334e-06,
      "loss": 0.8227,
      "step": 2216
    },
    {
      "epoch": 0.7487335359675785,
      "grad_norm": 2.5929372310638428,
      "learning_rate": 2.976136852025573e-06,
      "loss": 0.8768,
      "step": 2217
    },
    {
      "epoch": 0.7490712597095576,
      "grad_norm": 2.430474042892456,
      "learning_rate": 2.968563155244556e-06,
      "loss": 0.8729,
      "step": 2218
    },
    {
      "epoch": 0.7494089834515366,
      "grad_norm": 1.9676469564437866,
      "learning_rate": 2.9609974274917808e-06,
      "loss": 0.8332,
      "step": 2219
    },
    {
      "epoch": 0.7497467071935157,
      "grad_norm": 2.6544787883758545,
      "learning_rate": 2.9534396773417996e-06,
      "loss": 0.8579,
      "step": 2220
    },
    {
      "epoch": 0.7500844309354948,
      "grad_norm": 2.018902540206909,
      "learning_rate": 2.9458899133601438e-06,
      "loss": 0.9269,
      "step": 2221
    },
    {
      "epoch": 0.7504221546774739,
      "grad_norm": 2.028341054916382,
      "learning_rate": 2.9383481441032834e-06,
      "loss": 0.9088,
      "step": 2222
    },
    {
      "epoch": 0.7507598784194529,
      "grad_norm": 2.530329704284668,
      "learning_rate": 2.930814378118624e-06,
      "loss": 0.908,
      "step": 2223
    },
    {
      "epoch": 0.7507598784194529,
      "eval_loss": 0.8084899187088013,
      "eval_runtime": 347.301,
      "eval_samples_per_second": 8.557,
      "eval_steps_per_second": 1.071,
      "step": 2223
    },
    {
      "epoch": 0.751097602161432,
      "grad_norm": 2.0819270610809326,
      "learning_rate": 2.9232886239445056e-06,
      "loss": 0.9082,
      "step": 2224
    },
    {
      "epoch": 0.751435325903411,
      "grad_norm": 2.0199506282806396,
      "learning_rate": 2.915770890110189e-06,
      "loss": 0.9029,
      "step": 2225
    },
    {
      "epoch": 0.75177304964539,
      "grad_norm": 2.442213773727417,
      "learning_rate": 2.908261185135843e-06,
      "loss": 0.9158,
      "step": 2226
    },
    {
      "epoch": 0.7521107733873691,
      "grad_norm": 2.1880266666412354,
      "learning_rate": 2.9007595175325375e-06,
      "loss": 0.9039,
      "step": 2227
    },
    {
      "epoch": 0.7524484971293482,
      "grad_norm": 2.2033627033233643,
      "learning_rate": 2.893265895802233e-06,
      "loss": 0.9245,
      "step": 2228
    },
    {
      "epoch": 0.7527862208713273,
      "grad_norm": 2.458392858505249,
      "learning_rate": 2.885780328437774e-06,
      "loss": 0.8789,
      "step": 2229
    },
    {
      "epoch": 0.7531239446133063,
      "grad_norm": 2.542020559310913,
      "learning_rate": 2.878302823922874e-06,
      "loss": 0.7915,
      "step": 2230
    },
    {
      "epoch": 0.7534616683552854,
      "grad_norm": 2.2780895233154297,
      "learning_rate": 2.870833390732102e-06,
      "loss": 0.8587,
      "step": 2231
    },
    {
      "epoch": 0.7537993920972644,
      "grad_norm": 2.043471336364746,
      "learning_rate": 2.863372037330897e-06,
      "loss": 0.8505,
      "step": 2232
    },
    {
      "epoch": 0.7541371158392435,
      "grad_norm": 2.2601757049560547,
      "learning_rate": 2.855918772175522e-06,
      "loss": 0.9394,
      "step": 2233
    },
    {
      "epoch": 0.7544748395812225,
      "grad_norm": 2.154200553894043,
      "learning_rate": 2.84847360371308e-06,
      "loss": 0.8864,
      "step": 2234
    },
    {
      "epoch": 0.7548125633232016,
      "grad_norm": 2.2340898513793945,
      "learning_rate": 2.84103654038151e-06,
      "loss": 0.9307,
      "step": 2235
    },
    {
      "epoch": 0.7551502870651807,
      "grad_norm": 2.136181592941284,
      "learning_rate": 2.833607590609544e-06,
      "loss": 0.9136,
      "step": 2236
    },
    {
      "epoch": 0.7554880108071598,
      "grad_norm": 2.1249923706054688,
      "learning_rate": 2.826186762816732e-06,
      "loss": 0.9192,
      "step": 2237
    },
    {
      "epoch": 0.7558257345491388,
      "grad_norm": 2.1073174476623535,
      "learning_rate": 2.818774065413414e-06,
      "loss": 0.9103,
      "step": 2238
    },
    {
      "epoch": 0.7561634582911179,
      "grad_norm": 2.1226649284362793,
      "learning_rate": 2.8113695068007185e-06,
      "loss": 0.9623,
      "step": 2239
    },
    {
      "epoch": 0.7565011820330969,
      "grad_norm": 2.13651967048645,
      "learning_rate": 2.8039730953705524e-06,
      "loss": 0.8626,
      "step": 2240
    },
    {
      "epoch": 0.756838905775076,
      "grad_norm": 2.537858486175537,
      "learning_rate": 2.796584839505575e-06,
      "loss": 0.9036,
      "step": 2241
    },
    {
      "epoch": 0.757176629517055,
      "grad_norm": 2.3505749702453613,
      "learning_rate": 2.7892047475792226e-06,
      "loss": 0.8791,
      "step": 2242
    },
    {
      "epoch": 0.7575143532590342,
      "grad_norm": 2.3673033714294434,
      "learning_rate": 2.7818328279556683e-06,
      "loss": 0.9391,
      "step": 2243
    },
    {
      "epoch": 0.7578520770010132,
      "grad_norm": 2.9157936573028564,
      "learning_rate": 2.77446908898982e-06,
      "loss": 0.8776,
      "step": 2244
    },
    {
      "epoch": 0.7581898007429922,
      "grad_norm": 2.264425277709961,
      "learning_rate": 2.7671135390273175e-06,
      "loss": 0.8896,
      "step": 2245
    },
    {
      "epoch": 0.7585275244849713,
      "grad_norm": 2.3055782318115234,
      "learning_rate": 2.7597661864045232e-06,
      "loss": 0.9739,
      "step": 2246
    },
    {
      "epoch": 0.7588652482269503,
      "grad_norm": 2.185673236846924,
      "learning_rate": 2.7524270394485065e-06,
      "loss": 0.9218,
      "step": 2247
    },
    {
      "epoch": 0.7592029719689294,
      "grad_norm": 2.1334211826324463,
      "learning_rate": 2.7450961064770344e-06,
      "loss": 0.9062,
      "step": 2248
    },
    {
      "epoch": 0.7595406957109084,
      "grad_norm": 2.2204785346984863,
      "learning_rate": 2.7377733957985687e-06,
      "loss": 0.9231,
      "step": 2249
    },
    {
      "epoch": 0.7598784194528876,
      "grad_norm": 2.5631942749023438,
      "learning_rate": 2.7304589157122498e-06,
      "loss": 0.7652,
      "step": 2250
    },
    {
      "epoch": 0.7602161431948666,
      "grad_norm": 2.298811435699463,
      "learning_rate": 2.723152674507894e-06,
      "loss": 0.8821,
      "step": 2251
    },
    {
      "epoch": 0.7605538669368457,
      "grad_norm": 2.5264346599578857,
      "learning_rate": 2.715854680465969e-06,
      "loss": 0.9261,
      "step": 2252
    },
    {
      "epoch": 0.7608915906788247,
      "grad_norm": 2.265582323074341,
      "learning_rate": 2.7085649418576153e-06,
      "loss": 0.8529,
      "step": 2253
    },
    {
      "epoch": 0.7612293144208038,
      "grad_norm": 2.2410635948181152,
      "learning_rate": 2.7012834669445966e-06,
      "loss": 0.9216,
      "step": 2254
    },
    {
      "epoch": 0.7615670381627828,
      "grad_norm": 2.330439329147339,
      "learning_rate": 2.694010263979322e-06,
      "loss": 0.8952,
      "step": 2255
    },
    {
      "epoch": 0.7619047619047619,
      "grad_norm": 2.5634233951568604,
      "learning_rate": 2.686745341204825e-06,
      "loss": 0.9509,
      "step": 2256
    },
    {
      "epoch": 0.762242485646741,
      "grad_norm": 2.477675437927246,
      "learning_rate": 2.6794887068547526e-06,
      "loss": 0.7934,
      "step": 2257
    },
    {
      "epoch": 0.7625802093887201,
      "grad_norm": 2.3950467109680176,
      "learning_rate": 2.6722403691533594e-06,
      "loss": 0.8488,
      "step": 2258
    },
    {
      "epoch": 0.7629179331306991,
      "grad_norm": 2.26098370552063,
      "learning_rate": 2.6650003363154963e-06,
      "loss": 0.9082,
      "step": 2259
    },
    {
      "epoch": 0.7632556568726782,
      "grad_norm": 2.72948956489563,
      "learning_rate": 2.657768616546602e-06,
      "loss": 0.8348,
      "step": 2260
    },
    {
      "epoch": 0.7635933806146572,
      "grad_norm": 2.796614646911621,
      "learning_rate": 2.6505452180426983e-06,
      "loss": 0.8739,
      "step": 2261
    },
    {
      "epoch": 0.7639311043566362,
      "grad_norm": 2.1549580097198486,
      "learning_rate": 2.6433301489903618e-06,
      "loss": 0.9404,
      "step": 2262
    },
    {
      "epoch": 0.7642688280986153,
      "grad_norm": 2.3220598697662354,
      "learning_rate": 2.6361234175667516e-06,
      "loss": 0.8153,
      "step": 2263
    },
    {
      "epoch": 0.7646065518405943,
      "grad_norm": 2.355344533920288,
      "learning_rate": 2.628925031939554e-06,
      "loss": 0.9094,
      "step": 2264
    },
    {
      "epoch": 0.7649442755825735,
      "grad_norm": 2.218343496322632,
      "learning_rate": 2.621735000267013e-06,
      "loss": 0.9496,
      "step": 2265
    },
    {
      "epoch": 0.7652819993245525,
      "grad_norm": 2.1747515201568604,
      "learning_rate": 2.614553330697898e-06,
      "loss": 0.9086,
      "step": 2266
    },
    {
      "epoch": 0.7656197230665316,
      "grad_norm": 2.0163192749023438,
      "learning_rate": 2.6073800313715013e-06,
      "loss": 0.8517,
      "step": 2267
    },
    {
      "epoch": 0.7659574468085106,
      "grad_norm": 2.0961902141571045,
      "learning_rate": 2.600215110417631e-06,
      "loss": 0.9049,
      "step": 2268
    },
    {
      "epoch": 0.7662951705504897,
      "grad_norm": 2.2469472885131836,
      "learning_rate": 2.5930585759566017e-06,
      "loss": 0.9049,
      "step": 2269
    },
    {
      "epoch": 0.7666328942924687,
      "grad_norm": 2.2688517570495605,
      "learning_rate": 2.5859104360992115e-06,
      "loss": 0.8799,
      "step": 2270
    },
    {
      "epoch": 0.7669706180344478,
      "grad_norm": 2.1799774169921875,
      "learning_rate": 2.5787706989467643e-06,
      "loss": 0.9662,
      "step": 2271
    },
    {
      "epoch": 0.7673083417764269,
      "grad_norm": 2.3821704387664795,
      "learning_rate": 2.5716393725910216e-06,
      "loss": 0.8225,
      "step": 2272
    },
    {
      "epoch": 0.767646065518406,
      "grad_norm": 2.0905206203460693,
      "learning_rate": 2.5645164651142195e-06,
      "loss": 0.9419,
      "step": 2273
    },
    {
      "epoch": 0.767983789260385,
      "grad_norm": 2.471987724304199,
      "learning_rate": 2.5574019845890643e-06,
      "loss": 0.7992,
      "step": 2274
    },
    {
      "epoch": 0.7683215130023641,
      "grad_norm": 2.5249152183532715,
      "learning_rate": 2.550295939078691e-06,
      "loss": 0.9545,
      "step": 2275
    },
    {
      "epoch": 0.7686592367443431,
      "grad_norm": 2.4439244270324707,
      "learning_rate": 2.5431983366366875e-06,
      "loss": 0.856,
      "step": 2276
    },
    {
      "epoch": 0.7689969604863222,
      "grad_norm": 1.7017780542373657,
      "learning_rate": 2.536109185307072e-06,
      "loss": 0.5575,
      "step": 2277
    },
    {
      "epoch": 0.7693346842283012,
      "grad_norm": 2.241992712020874,
      "learning_rate": 2.5290284931242815e-06,
      "loss": 0.8601,
      "step": 2278
    },
    {
      "epoch": 0.7696724079702804,
      "grad_norm": 2.849855661392212,
      "learning_rate": 2.5219562681131716e-06,
      "loss": 0.9231,
      "step": 2279
    },
    {
      "epoch": 0.7700101317122594,
      "grad_norm": 2.353912353515625,
      "learning_rate": 2.514892518288988e-06,
      "loss": 0.8596,
      "step": 2280
    },
    {
      "epoch": 0.7703478554542385,
      "grad_norm": 2.1409833431243896,
      "learning_rate": 2.5078372516573912e-06,
      "loss": 0.9516,
      "step": 2281
    },
    {
      "epoch": 0.7706855791962175,
      "grad_norm": 2.651157855987549,
      "learning_rate": 2.500790476214413e-06,
      "loss": 0.9298,
      "step": 2282
    },
    {
      "epoch": 0.7710233029381965,
      "grad_norm": 2.455305576324463,
      "learning_rate": 2.4937521999464633e-06,
      "loss": 0.8779,
      "step": 2283
    },
    {
      "epoch": 0.7713610266801756,
      "grad_norm": 2.0984818935394287,
      "learning_rate": 2.4867224308303228e-06,
      "loss": 0.9686,
      "step": 2284
    },
    {
      "epoch": 0.7716987504221546,
      "grad_norm": 2.233726739883423,
      "learning_rate": 2.4797011768331304e-06,
      "loss": 0.8437,
      "step": 2285
    },
    {
      "epoch": 0.7720364741641338,
      "grad_norm": 2.251816749572754,
      "learning_rate": 2.472688445912372e-06,
      "loss": 0.8753,
      "step": 2286
    },
    {
      "epoch": 0.7723741979061128,
      "grad_norm": 2.34570574760437,
      "learning_rate": 2.4656842460158783e-06,
      "loss": 0.9181,
      "step": 2287
    },
    {
      "epoch": 0.7727119216480919,
      "grad_norm": 2.6001083850860596,
      "learning_rate": 2.4586885850818076e-06,
      "loss": 0.8928,
      "step": 2288
    },
    {
      "epoch": 0.7730496453900709,
      "grad_norm": 2.0827982425689697,
      "learning_rate": 2.451701471038641e-06,
      "loss": 0.8945,
      "step": 2289
    },
    {
      "epoch": 0.77338736913205,
      "grad_norm": 2.920703411102295,
      "learning_rate": 2.4447229118051775e-06,
      "loss": 0.8503,
      "step": 2290
    },
    {
      "epoch": 0.773725092874029,
      "grad_norm": 2.3528969287872314,
      "learning_rate": 2.437752915290508e-06,
      "loss": 0.8625,
      "step": 2291
    },
    {
      "epoch": 0.7740628166160081,
      "grad_norm": 2.1248507499694824,
      "learning_rate": 2.430791489394039e-06,
      "loss": 0.9085,
      "step": 2292
    },
    {
      "epoch": 0.7744005403579872,
      "grad_norm": 2.5249507427215576,
      "learning_rate": 2.4238386420054436e-06,
      "loss": 0.9335,
      "step": 2293
    },
    {
      "epoch": 0.7747382640999663,
      "grad_norm": 2.7721898555755615,
      "learning_rate": 2.416894381004682e-06,
      "loss": 1.0284,
      "step": 2294
    },
    {
      "epoch": 0.7750759878419453,
      "grad_norm": 2.7028748989105225,
      "learning_rate": 2.409958714261991e-06,
      "loss": 0.9506,
      "step": 2295
    },
    {
      "epoch": 0.7754137115839244,
      "grad_norm": 2.7603092193603516,
      "learning_rate": 2.4030316496378493e-06,
      "loss": 0.8881,
      "step": 2296
    },
    {
      "epoch": 0.7757514353259034,
      "grad_norm": 2.421834945678711,
      "learning_rate": 2.3961131949829987e-06,
      "loss": 0.8443,
      "step": 2297
    },
    {
      "epoch": 0.7760891590678825,
      "grad_norm": 2.224776268005371,
      "learning_rate": 2.3892033581384188e-06,
      "loss": 0.9565,
      "step": 2298
    },
    {
      "epoch": 0.7764268828098615,
      "grad_norm": 2.5617058277130127,
      "learning_rate": 2.382302146935326e-06,
      "loss": 0.8867,
      "step": 2299
    },
    {
      "epoch": 0.7767646065518405,
      "grad_norm": 2.356306552886963,
      "learning_rate": 2.3754095691951585e-06,
      "loss": 0.9338,
      "step": 2300
    },
    {
      "epoch": 0.7771023302938197,
      "grad_norm": 2.0664615631103516,
      "learning_rate": 2.3685256327295627e-06,
      "loss": 0.9874,
      "step": 2301
    },
    {
      "epoch": 0.7774400540357987,
      "grad_norm": 2.184074878692627,
      "learning_rate": 2.361650345340405e-06,
      "loss": 0.9105,
      "step": 2302
    },
    {
      "epoch": 0.7777777777777778,
      "grad_norm": 2.521228551864624,
      "learning_rate": 2.354783714819745e-06,
      "loss": 0.927,
      "step": 2303
    },
    {
      "epoch": 0.7781155015197568,
      "grad_norm": 2.7180845737457275,
      "learning_rate": 2.347925748949822e-06,
      "loss": 0.8175,
      "step": 2304
    },
    {
      "epoch": 0.7784532252617359,
      "grad_norm": 2.189541816711426,
      "learning_rate": 2.3410764555030653e-06,
      "loss": 0.9504,
      "step": 2305
    },
    {
      "epoch": 0.7787909490037149,
      "grad_norm": 2.357917070388794,
      "learning_rate": 2.334235842242072e-06,
      "loss": 0.9528,
      "step": 2306
    },
    {
      "epoch": 0.779128672745694,
      "grad_norm": 2.6936287879943848,
      "learning_rate": 2.327403916919602e-06,
      "loss": 0.791,
      "step": 2307
    },
    {
      "epoch": 0.7794663964876731,
      "grad_norm": 2.6440634727478027,
      "learning_rate": 2.3205806872785686e-06,
      "loss": 0.885,
      "step": 2308
    },
    {
      "epoch": 0.7798041202296522,
      "grad_norm": 2.1481082439422607,
      "learning_rate": 2.31376616105203e-06,
      "loss": 0.9087,
      "step": 2309
    },
    {
      "epoch": 0.7801418439716312,
      "grad_norm": 2.3218395709991455,
      "learning_rate": 2.3069603459631795e-06,
      "loss": 0.8482,
      "step": 2310
    },
    {
      "epoch": 0.7804795677136103,
      "grad_norm": 2.4146180152893066,
      "learning_rate": 2.3001632497253423e-06,
      "loss": 0.9644,
      "step": 2311
    },
    {
      "epoch": 0.7808172914555893,
      "grad_norm": 2.7607650756835938,
      "learning_rate": 2.2933748800419496e-06,
      "loss": 0.8743,
      "step": 2312
    },
    {
      "epoch": 0.7811550151975684,
      "grad_norm": 2.5582823753356934,
      "learning_rate": 2.2865952446065632e-06,
      "loss": 0.9346,
      "step": 2313
    },
    {
      "epoch": 0.7814927389395474,
      "grad_norm": 2.210233211517334,
      "learning_rate": 2.279824351102824e-06,
      "loss": 0.9302,
      "step": 2314
    },
    {
      "epoch": 0.7818304626815266,
      "grad_norm": 2.378706216812134,
      "learning_rate": 2.2730622072044806e-06,
      "loss": 0.859,
      "step": 2315
    },
    {
      "epoch": 0.7821681864235056,
      "grad_norm": 2.0426790714263916,
      "learning_rate": 2.266308820575357e-06,
      "loss": 0.9062,
      "step": 2316
    },
    {
      "epoch": 0.7825059101654847,
      "grad_norm": 2.071845293045044,
      "learning_rate": 2.2595641988693572e-06,
      "loss": 0.9286,
      "step": 2317
    },
    {
      "epoch": 0.7828436339074637,
      "grad_norm": 2.1487317085266113,
      "learning_rate": 2.252828349730449e-06,
      "loss": 0.8547,
      "step": 2318
    },
    {
      "epoch": 0.7831813576494427,
      "grad_norm": 2.406669855117798,
      "learning_rate": 2.2461012807926585e-06,
      "loss": 0.9013,
      "step": 2319
    },
    {
      "epoch": 0.7835190813914218,
      "grad_norm": 2.385948657989502,
      "learning_rate": 2.239382999680062e-06,
      "loss": 1.0018,
      "step": 2320
    },
    {
      "epoch": 0.7838568051334008,
      "grad_norm": 2.4606282711029053,
      "learning_rate": 2.2326735140067755e-06,
      "loss": 0.8834,
      "step": 2321
    },
    {
      "epoch": 0.78419452887538,
      "grad_norm": 2.2426421642303467,
      "learning_rate": 2.225972831376939e-06,
      "loss": 0.8516,
      "step": 2322
    },
    {
      "epoch": 0.784532252617359,
      "grad_norm": 2.271660566329956,
      "learning_rate": 2.219280959384733e-06,
      "loss": 0.898,
      "step": 2323
    },
    {
      "epoch": 0.7848699763593381,
      "grad_norm": 2.3826022148132324,
      "learning_rate": 2.2125979056143366e-06,
      "loss": 0.9576,
      "step": 2324
    },
    {
      "epoch": 0.7852077001013171,
      "grad_norm": 2.3449206352233887,
      "learning_rate": 2.2059236776399407e-06,
      "loss": 0.8788,
      "step": 2325
    },
    {
      "epoch": 0.7855454238432962,
      "grad_norm": 2.1580145359039307,
      "learning_rate": 2.199258283025735e-06,
      "loss": 0.9125,
      "step": 2326
    },
    {
      "epoch": 0.7858831475852752,
      "grad_norm": 2.4399185180664062,
      "learning_rate": 2.192601729325895e-06,
      "loss": 0.944,
      "step": 2327
    },
    {
      "epoch": 0.7862208713272543,
      "grad_norm": 2.04860520362854,
      "learning_rate": 2.1859540240845788e-06,
      "loss": 0.9048,
      "step": 2328
    },
    {
      "epoch": 0.7865585950692333,
      "grad_norm": 2.0556511878967285,
      "learning_rate": 2.1793151748359155e-06,
      "loss": 0.8972,
      "step": 2329
    },
    {
      "epoch": 0.7868963188112125,
      "grad_norm": 2.6024322509765625,
      "learning_rate": 2.1726851891039957e-06,
      "loss": 0.8871,
      "step": 2330
    },
    {
      "epoch": 0.7872340425531915,
      "grad_norm": 2.294511318206787,
      "learning_rate": 2.1660640744028703e-06,
      "loss": 0.8711,
      "step": 2331
    },
    {
      "epoch": 0.7875717662951706,
      "grad_norm": 2.5296030044555664,
      "learning_rate": 2.159451838236526e-06,
      "loss": 0.8534,
      "step": 2332
    },
    {
      "epoch": 0.7879094900371496,
      "grad_norm": 2.8209927082061768,
      "learning_rate": 2.1528484880988908e-06,
      "loss": 0.7231,
      "step": 2333
    },
    {
      "epoch": 0.7882472137791287,
      "grad_norm": 2.232222080230713,
      "learning_rate": 2.1462540314738356e-06,
      "loss": 0.8654,
      "step": 2334
    },
    {
      "epoch": 0.7885849375211077,
      "grad_norm": 2.458707094192505,
      "learning_rate": 2.1396684758351306e-06,
      "loss": 0.9201,
      "step": 2335
    },
    {
      "epoch": 0.7889226612630867,
      "grad_norm": 2.1869614124298096,
      "learning_rate": 2.1330918286464695e-06,
      "loss": 0.8883,
      "step": 2336
    },
    {
      "epoch": 0.7892603850050659,
      "grad_norm": 2.187095880508423,
      "learning_rate": 2.126524097361449e-06,
      "loss": 0.8763,
      "step": 2337
    },
    {
      "epoch": 0.789598108747045,
      "grad_norm": 2.345210313796997,
      "learning_rate": 2.11996528942356e-06,
      "loss": 0.8281,
      "step": 2338
    },
    {
      "epoch": 0.789935832489024,
      "grad_norm": 2.2287001609802246,
      "learning_rate": 2.113415412266182e-06,
      "loss": 0.9352,
      "step": 2339
    },
    {
      "epoch": 0.790273556231003,
      "grad_norm": 2.0529768466949463,
      "learning_rate": 2.1068744733125634e-06,
      "loss": 0.9582,
      "step": 2340
    },
    {
      "epoch": 0.7906112799729821,
      "grad_norm": 2.4243195056915283,
      "learning_rate": 2.10034247997584e-06,
      "loss": 0.8981,
      "step": 2341
    },
    {
      "epoch": 0.7909490037149611,
      "grad_norm": 2.1847126483917236,
      "learning_rate": 2.093819439658996e-06,
      "loss": 0.8509,
      "step": 2342
    },
    {
      "epoch": 0.7912867274569402,
      "grad_norm": 2.314656972885132,
      "learning_rate": 2.0873053597548677e-06,
      "loss": 0.8222,
      "step": 2343
    },
    {
      "epoch": 0.7916244511989193,
      "grad_norm": 2.4287848472595215,
      "learning_rate": 2.0808002476461487e-06,
      "loss": 0.9462,
      "step": 2344
    },
    {
      "epoch": 0.7919621749408984,
      "grad_norm": 2.1777443885803223,
      "learning_rate": 2.0743041107053528e-06,
      "loss": 0.8755,
      "step": 2345
    },
    {
      "epoch": 0.7922998986828774,
      "grad_norm": 2.2278995513916016,
      "learning_rate": 2.067816956294835e-06,
      "loss": 0.8403,
      "step": 2346
    },
    {
      "epoch": 0.7926376224248565,
      "grad_norm": 2.06711483001709,
      "learning_rate": 2.0613387917667616e-06,
      "loss": 0.9188,
      "step": 2347
    },
    {
      "epoch": 0.7929753461668355,
      "grad_norm": 2.9795339107513428,
      "learning_rate": 2.054869624463116e-06,
      "loss": 0.8035,
      "step": 2348
    },
    {
      "epoch": 0.7933130699088146,
      "grad_norm": 2.4865341186523438,
      "learning_rate": 2.0484094617156823e-06,
      "loss": 0.7274,
      "step": 2349
    },
    {
      "epoch": 0.7936507936507936,
      "grad_norm": 2.485691547393799,
      "learning_rate": 2.0419583108460418e-06,
      "loss": 0.8891,
      "step": 2350
    },
    {
      "epoch": 0.7939885173927728,
      "grad_norm": 2.425727605819702,
      "learning_rate": 2.035516179165551e-06,
      "loss": 0.8844,
      "step": 2351
    },
    {
      "epoch": 0.7943262411347518,
      "grad_norm": 2.276596784591675,
      "learning_rate": 2.0290830739753653e-06,
      "loss": 0.9507,
      "step": 2352
    },
    {
      "epoch": 0.7946639648767309,
      "grad_norm": 2.2929317951202393,
      "learning_rate": 2.0226590025663883e-06,
      "loss": 0.9384,
      "step": 2353
    },
    {
      "epoch": 0.7950016886187099,
      "grad_norm": 2.529444694519043,
      "learning_rate": 2.016243972219296e-06,
      "loss": 0.8768,
      "step": 2354
    },
    {
      "epoch": 0.795339412360689,
      "grad_norm": 2.427338123321533,
      "learning_rate": 2.009837990204525e-06,
      "loss": 0.9172,
      "step": 2355
    },
    {
      "epoch": 0.795677136102668,
      "grad_norm": 2.970611095428467,
      "learning_rate": 2.003441063782241e-06,
      "loss": 1.0487,
      "step": 2356
    },
    {
      "epoch": 0.796014859844647,
      "grad_norm": 2.1466615200042725,
      "learning_rate": 1.997053200202357e-06,
      "loss": 0.9169,
      "step": 2357
    },
    {
      "epoch": 0.7963525835866262,
      "grad_norm": 2.2688703536987305,
      "learning_rate": 1.9906744067045116e-06,
      "loss": 0.9428,
      "step": 2358
    },
    {
      "epoch": 0.7966903073286052,
      "grad_norm": 2.1906068325042725,
      "learning_rate": 1.984304690518064e-06,
      "loss": 0.8859,
      "step": 2359
    },
    {
      "epoch": 0.7970280310705843,
      "grad_norm": 2.535109519958496,
      "learning_rate": 1.9779440588620903e-06,
      "loss": 0.9055,
      "step": 2360
    },
    {
      "epoch": 0.7973657548125633,
      "grad_norm": 2.494708776473999,
      "learning_rate": 1.9715925189453566e-06,
      "loss": 0.8353,
      "step": 2361
    },
    {
      "epoch": 0.7977034785545424,
      "grad_norm": 2.427567481994629,
      "learning_rate": 1.9652500779663453e-06,
      "loss": 0.8527,
      "step": 2362
    },
    {
      "epoch": 0.7980412022965214,
      "grad_norm": 2.3818278312683105,
      "learning_rate": 1.958916743113214e-06,
      "loss": 0.9156,
      "step": 2363
    },
    {
      "epoch": 0.7983789260385005,
      "grad_norm": 2.3568496704101562,
      "learning_rate": 1.9525925215637966e-06,
      "loss": 0.7816,
      "step": 2364
    },
    {
      "epoch": 0.7987166497804795,
      "grad_norm": 2.3129029273986816,
      "learning_rate": 1.9462774204856084e-06,
      "loss": 0.8496,
      "step": 2365
    },
    {
      "epoch": 0.7990543735224587,
      "grad_norm": 2.145062208175659,
      "learning_rate": 1.9399714470358235e-06,
      "loss": 0.8721,
      "step": 2366
    },
    {
      "epoch": 0.7993920972644377,
      "grad_norm": 2.277348041534424,
      "learning_rate": 1.933674608361269e-06,
      "loss": 0.8997,
      "step": 2367
    },
    {
      "epoch": 0.7997298210064168,
      "grad_norm": 2.361419916152954,
      "learning_rate": 1.9273869115984246e-06,
      "loss": 0.9623,
      "step": 2368
    },
    {
      "epoch": 0.8000675447483958,
      "grad_norm": 2.3715293407440186,
      "learning_rate": 1.921108363873404e-06,
      "loss": 0.9715,
      "step": 2369
    },
    {
      "epoch": 0.8004052684903749,
      "grad_norm": 2.580935478210449,
      "learning_rate": 1.9148389723019556e-06,
      "loss": 0.8074,
      "step": 2370
    },
    {
      "epoch": 0.8007429922323539,
      "grad_norm": 2.1488654613494873,
      "learning_rate": 1.908578743989451e-06,
      "loss": 0.9272,
      "step": 2371
    },
    {
      "epoch": 0.801080715974333,
      "grad_norm": 2.2449309825897217,
      "learning_rate": 1.902327686030866e-06,
      "loss": 0.9552,
      "step": 2372
    },
    {
      "epoch": 0.8014184397163121,
      "grad_norm": 2.4418044090270996,
      "learning_rate": 1.8960858055108045e-06,
      "loss": 0.8906,
      "step": 2373
    },
    {
      "epoch": 0.8017561634582911,
      "grad_norm": 2.520758628845215,
      "learning_rate": 1.8898531095034467e-06,
      "loss": 0.8687,
      "step": 2374
    },
    {
      "epoch": 0.8020938872002702,
      "grad_norm": 2.250593900680542,
      "learning_rate": 1.8836296050725755e-06,
      "loss": 0.9346,
      "step": 2375
    },
    {
      "epoch": 0.8024316109422492,
      "grad_norm": 2.0078232288360596,
      "learning_rate": 1.877415299271561e-06,
      "loss": 0.9266,
      "step": 2376
    },
    {
      "epoch": 0.8027693346842283,
      "grad_norm": 2.275620937347412,
      "learning_rate": 1.8712101991433362e-06,
      "loss": 1.0152,
      "step": 2377
    },
    {
      "epoch": 0.8031070584262073,
      "grad_norm": 2.348424196243286,
      "learning_rate": 1.8650143117204077e-06,
      "loss": 0.9246,
      "step": 2378
    },
    {
      "epoch": 0.8034447821681864,
      "grad_norm": 2.767559766769409,
      "learning_rate": 1.8588276440248399e-06,
      "loss": 0.6669,
      "step": 2379
    },
    {
      "epoch": 0.8037825059101655,
      "grad_norm": 2.3368520736694336,
      "learning_rate": 1.8526502030682469e-06,
      "loss": 0.8842,
      "step": 2380
    },
    {
      "epoch": 0.8041202296521446,
      "grad_norm": 2.099885940551758,
      "learning_rate": 1.846481995851792e-06,
      "loss": 0.983,
      "step": 2381
    },
    {
      "epoch": 0.8044579533941236,
      "grad_norm": 2.749765396118164,
      "learning_rate": 1.840323029366159e-06,
      "loss": 0.8095,
      "step": 2382
    },
    {
      "epoch": 0.8047956771361027,
      "grad_norm": 2.6756253242492676,
      "learning_rate": 1.8341733105915794e-06,
      "loss": 0.8334,
      "step": 2383
    },
    {
      "epoch": 0.8051334008780817,
      "grad_norm": 2.6348021030426025,
      "learning_rate": 1.8280328464977837e-06,
      "loss": 0.9492,
      "step": 2384
    },
    {
      "epoch": 0.8054711246200608,
      "grad_norm": 2.240184783935547,
      "learning_rate": 1.821901644044025e-06,
      "loss": 0.8722,
      "step": 2385
    },
    {
      "epoch": 0.8058088483620398,
      "grad_norm": 2.419062376022339,
      "learning_rate": 1.8157797101790597e-06,
      "loss": 0.9664,
      "step": 2386
    },
    {
      "epoch": 0.806146572104019,
      "grad_norm": 2.3228180408477783,
      "learning_rate": 1.8096670518411353e-06,
      "loss": 0.9519,
      "step": 2387
    },
    {
      "epoch": 0.806484295845998,
      "grad_norm": 2.3404786586761475,
      "learning_rate": 1.8035636759579888e-06,
      "loss": 0.9337,
      "step": 2388
    },
    {
      "epoch": 0.8068220195879771,
      "grad_norm": 2.5321860313415527,
      "learning_rate": 1.7974695894468385e-06,
      "loss": 0.8539,
      "step": 2389
    },
    {
      "epoch": 0.8071597433299561,
      "grad_norm": 2.269758939743042,
      "learning_rate": 1.791384799214373e-06,
      "loss": 0.904,
      "step": 2390
    },
    {
      "epoch": 0.8074974670719351,
      "grad_norm": 3.053079843521118,
      "learning_rate": 1.7853093121567477e-06,
      "loss": 0.9078,
      "step": 2391
    },
    {
      "epoch": 0.8078351908139142,
      "grad_norm": 2.0459346771240234,
      "learning_rate": 1.7792431351595674e-06,
      "loss": 0.9067,
      "step": 2392
    },
    {
      "epoch": 0.8081729145558932,
      "grad_norm": 2.205310583114624,
      "learning_rate": 1.77318627509789e-06,
      "loss": 0.9573,
      "step": 2393
    },
    {
      "epoch": 0.8085106382978723,
      "grad_norm": 2.375980854034424,
      "learning_rate": 1.7671387388362205e-06,
      "loss": 0.9011,
      "step": 2394
    },
    {
      "epoch": 0.8088483620398514,
      "grad_norm": 2.4712748527526855,
      "learning_rate": 1.7611005332284847e-06,
      "loss": 0.8507,
      "step": 2395
    },
    {
      "epoch": 0.8091860857818305,
      "grad_norm": 2.4097931385040283,
      "learning_rate": 1.755071665118041e-06,
      "loss": 0.9152,
      "step": 2396
    },
    {
      "epoch": 0.8095238095238095,
      "grad_norm": 2.158761978149414,
      "learning_rate": 1.7490521413376649e-06,
      "loss": 0.8828,
      "step": 2397
    },
    {
      "epoch": 0.8098615332657886,
      "grad_norm": 2.3166160583496094,
      "learning_rate": 1.743041968709538e-06,
      "loss": 0.8578,
      "step": 2398
    },
    {
      "epoch": 0.8101992570077676,
      "grad_norm": 2.2314670085906982,
      "learning_rate": 1.7370411540452492e-06,
      "loss": 0.8658,
      "step": 2399
    },
    {
      "epoch": 0.8105369807497467,
      "grad_norm": 2.299769401550293,
      "learning_rate": 1.731049704145773e-06,
      "loss": 0.8374,
      "step": 2400
    },
    {
      "epoch": 0.8108747044917257,
      "grad_norm": 1.9546984434127808,
      "learning_rate": 1.725067625801481e-06,
      "loss": 0.9504,
      "step": 2401
    },
    {
      "epoch": 0.8112124282337049,
      "grad_norm": 2.376016616821289,
      "learning_rate": 1.7190949257921197e-06,
      "loss": 0.8695,
      "step": 2402
    },
    {
      "epoch": 0.8115501519756839,
      "grad_norm": 2.361262798309326,
      "learning_rate": 1.7131316108867978e-06,
      "loss": 0.8438,
      "step": 2403
    },
    {
      "epoch": 0.811887875717663,
      "grad_norm": 2.4931447505950928,
      "learning_rate": 1.707177687844006e-06,
      "loss": 0.8851,
      "step": 2404
    },
    {
      "epoch": 0.812225599459642,
      "grad_norm": 2.327471971511841,
      "learning_rate": 1.7012331634115719e-06,
      "loss": 0.8654,
      "step": 2405
    },
    {
      "epoch": 0.8125633232016211,
      "grad_norm": 2.2243874073028564,
      "learning_rate": 1.695298044326681e-06,
      "loss": 0.9868,
      "step": 2406
    },
    {
      "epoch": 0.8129010469436001,
      "grad_norm": 2.7563979625701904,
      "learning_rate": 1.6893723373158588e-06,
      "loss": 0.9022,
      "step": 2407
    },
    {
      "epoch": 0.8132387706855791,
      "grad_norm": 2.69340443611145,
      "learning_rate": 1.6834560490949636e-06,
      "loss": 0.8963,
      "step": 2408
    },
    {
      "epoch": 0.8135764944275583,
      "grad_norm": 2.4481773376464844,
      "learning_rate": 1.6775491863691763e-06,
      "loss": 0.8108,
      "step": 2409
    },
    {
      "epoch": 0.8139142181695374,
      "grad_norm": 2.1881377696990967,
      "learning_rate": 1.671651755832997e-06,
      "loss": 0.9135,
      "step": 2410
    },
    {
      "epoch": 0.8142519419115164,
      "grad_norm": 3.0163321495056152,
      "learning_rate": 1.6657637641702374e-06,
      "loss": 0.8486,
      "step": 2411
    },
    {
      "epoch": 0.8145896656534954,
      "grad_norm": 1.9681172370910645,
      "learning_rate": 1.6598852180540114e-06,
      "loss": 0.9059,
      "step": 2412
    },
    {
      "epoch": 0.8149273893954745,
      "grad_norm": 2.4752144813537598,
      "learning_rate": 1.6540161241467235e-06,
      "loss": 0.9118,
      "step": 2413
    },
    {
      "epoch": 0.8152651131374535,
      "grad_norm": 2.502593994140625,
      "learning_rate": 1.6481564891000678e-06,
      "loss": 0.8499,
      "step": 2414
    },
    {
      "epoch": 0.8156028368794326,
      "grad_norm": 2.5271475315093994,
      "learning_rate": 1.6423063195550271e-06,
      "loss": 0.833,
      "step": 2415
    },
    {
      "epoch": 0.8159405606214117,
      "grad_norm": 1.7622939348220825,
      "learning_rate": 1.6364656221418417e-06,
      "loss": 0.5683,
      "step": 2416
    },
    {
      "epoch": 0.8162782843633908,
      "grad_norm": 2.1854426860809326,
      "learning_rate": 1.630634403480026e-06,
      "loss": 0.9323,
      "step": 2417
    },
    {
      "epoch": 0.8166160081053698,
      "grad_norm": 2.417870283126831,
      "learning_rate": 1.6248126701783495e-06,
      "loss": 0.966,
      "step": 2418
    },
    {
      "epoch": 0.8169537318473489,
      "grad_norm": 2.4956154823303223,
      "learning_rate": 1.6190004288348327e-06,
      "loss": 0.9651,
      "step": 2419
    },
    {
      "epoch": 0.8172914555893279,
      "grad_norm": 2.3127987384796143,
      "learning_rate": 1.6131976860367394e-06,
      "loss": 0.8383,
      "step": 2420
    },
    {
      "epoch": 0.817629179331307,
      "grad_norm": 2.3027374744415283,
      "learning_rate": 1.6074044483605577e-06,
      "loss": 0.9864,
      "step": 2421
    },
    {
      "epoch": 0.817966903073286,
      "grad_norm": 2.31717586517334,
      "learning_rate": 1.6016207223720203e-06,
      "loss": 0.9161,
      "step": 2422
    },
    {
      "epoch": 0.8183046268152652,
      "grad_norm": 2.1508214473724365,
      "learning_rate": 1.5958465146260715e-06,
      "loss": 0.9231,
      "step": 2423
    },
    {
      "epoch": 0.8186423505572442,
      "grad_norm": 2.396125078201294,
      "learning_rate": 1.5900818316668609e-06,
      "loss": 0.9109,
      "step": 2424
    },
    {
      "epoch": 0.8189800742992233,
      "grad_norm": 2.1820759773254395,
      "learning_rate": 1.584326680027758e-06,
      "loss": 0.9337,
      "step": 2425
    },
    {
      "epoch": 0.8193177980412023,
      "grad_norm": 2.2879111766815186,
      "learning_rate": 1.5785810662313172e-06,
      "loss": 0.8436,
      "step": 2426
    },
    {
      "epoch": 0.8196555217831814,
      "grad_norm": 2.146381378173828,
      "learning_rate": 1.572844996789291e-06,
      "loss": 0.8499,
      "step": 2427
    },
    {
      "epoch": 0.8199932455251604,
      "grad_norm": 2.6388208866119385,
      "learning_rate": 1.5671184782026106e-06,
      "loss": 0.8418,
      "step": 2428
    },
    {
      "epoch": 0.8203309692671394,
      "grad_norm": 3.0260109901428223,
      "learning_rate": 1.5614015169613862e-06,
      "loss": 0.8119,
      "step": 2429
    },
    {
      "epoch": 0.8206686930091185,
      "grad_norm": 2.3653459548950195,
      "learning_rate": 1.555694119544894e-06,
      "loss": 0.8997,
      "step": 2430
    },
    {
      "epoch": 0.8210064167510976,
      "grad_norm": 2.4612114429473877,
      "learning_rate": 1.5499962924215739e-06,
      "loss": 0.9219,
      "step": 2431
    },
    {
      "epoch": 0.8213441404930767,
      "grad_norm": 2.407473564147949,
      "learning_rate": 1.5443080420490097e-06,
      "loss": 0.9305,
      "step": 2432
    },
    {
      "epoch": 0.8216818642350557,
      "grad_norm": 2.6794142723083496,
      "learning_rate": 1.538629374873949e-06,
      "loss": 0.8405,
      "step": 2433
    },
    {
      "epoch": 0.8220195879770348,
      "grad_norm": 2.585465669631958,
      "learning_rate": 1.532960297332262e-06,
      "loss": 0.8445,
      "step": 2434
    },
    {
      "epoch": 0.8223573117190138,
      "grad_norm": 2.6902544498443604,
      "learning_rate": 1.5273008158489556e-06,
      "loss": 0.7897,
      "step": 2435
    },
    {
      "epoch": 0.8226950354609929,
      "grad_norm": 2.799748420715332,
      "learning_rate": 1.5216509368381693e-06,
      "loss": 0.9265,
      "step": 2436
    },
    {
      "epoch": 0.8230327592029719,
      "grad_norm": 2.0505125522613525,
      "learning_rate": 1.516010666703146e-06,
      "loss": 0.8519,
      "step": 2437
    },
    {
      "epoch": 0.8233704829449511,
      "grad_norm": 2.479980230331421,
      "learning_rate": 1.5103800118362477e-06,
      "loss": 0.8691,
      "step": 2438
    },
    {
      "epoch": 0.8237082066869301,
      "grad_norm": 2.519620656967163,
      "learning_rate": 1.5047589786189365e-06,
      "loss": 0.777,
      "step": 2439
    },
    {
      "epoch": 0.8240459304289092,
      "grad_norm": 2.1691229343414307,
      "learning_rate": 1.4991475734217697e-06,
      "loss": 0.8845,
      "step": 2440
    },
    {
      "epoch": 0.8243836541708882,
      "grad_norm": 2.382126808166504,
      "learning_rate": 1.4935458026043958e-06,
      "loss": 0.8966,
      "step": 2441
    },
    {
      "epoch": 0.8247213779128673,
      "grad_norm": 2.186721086502075,
      "learning_rate": 1.4879536725155342e-06,
      "loss": 0.9435,
      "step": 2442
    },
    {
      "epoch": 0.8250591016548463,
      "grad_norm": 2.6181178092956543,
      "learning_rate": 1.4823711894929905e-06,
      "loss": 0.8538,
      "step": 2443
    },
    {
      "epoch": 0.8253968253968254,
      "grad_norm": 2.587207794189453,
      "learning_rate": 1.4767983598636338e-06,
      "loss": 0.8919,
      "step": 2444
    },
    {
      "epoch": 0.8257345491388045,
      "grad_norm": 2.5727341175079346,
      "learning_rate": 1.4712351899433851e-06,
      "loss": 0.7593,
      "step": 2445
    },
    {
      "epoch": 0.8260722728807836,
      "grad_norm": 2.961782932281494,
      "learning_rate": 1.4656816860372246e-06,
      "loss": 0.838,
      "step": 2446
    },
    {
      "epoch": 0.8264099966227626,
      "grad_norm": 2.147956609725952,
      "learning_rate": 1.4601378544391765e-06,
      "loss": 0.897,
      "step": 2447
    },
    {
      "epoch": 0.8267477203647416,
      "grad_norm": 2.233283281326294,
      "learning_rate": 1.4546037014323023e-06,
      "loss": 0.8252,
      "step": 2448
    },
    {
      "epoch": 0.8270854441067207,
      "grad_norm": 2.657028913497925,
      "learning_rate": 1.4490792332886938e-06,
      "loss": 0.9044,
      "step": 2449
    },
    {
      "epoch": 0.8274231678486997,
      "grad_norm": 2.138300895690918,
      "learning_rate": 1.4435644562694673e-06,
      "loss": 0.8979,
      "step": 2450
    },
    {
      "epoch": 0.8277608915906788,
      "grad_norm": 2.007845401763916,
      "learning_rate": 1.4380593766247563e-06,
      "loss": 0.8682,
      "step": 2451
    },
    {
      "epoch": 0.8280986153326579,
      "grad_norm": 2.259765625,
      "learning_rate": 1.4325640005937058e-06,
      "loss": 0.936,
      "step": 2452
    },
    {
      "epoch": 0.828436339074637,
      "grad_norm": 2.2791409492492676,
      "learning_rate": 1.4270783344044515e-06,
      "loss": 0.8541,
      "step": 2453
    },
    {
      "epoch": 0.828774062816616,
      "grad_norm": 2.4329257011413574,
      "learning_rate": 1.4216023842741456e-06,
      "loss": 0.964,
      "step": 2454
    },
    {
      "epoch": 0.8291117865585951,
      "grad_norm": 2.2679944038391113,
      "learning_rate": 1.416136156408907e-06,
      "loss": 0.922,
      "step": 2455
    },
    {
      "epoch": 0.8294495103005741,
      "grad_norm": 2.4042017459869385,
      "learning_rate": 1.4106796570038495e-06,
      "loss": 0.9327,
      "step": 2456
    },
    {
      "epoch": 0.8297872340425532,
      "grad_norm": 3.0364768505096436,
      "learning_rate": 1.405232892243058e-06,
      "loss": 0.8633,
      "step": 2457
    },
    {
      "epoch": 0.8301249577845322,
      "grad_norm": 2.1611204147338867,
      "learning_rate": 1.3997958682995826e-06,
      "loss": 0.8751,
      "step": 2458
    },
    {
      "epoch": 0.8304626815265113,
      "grad_norm": 2.5465054512023926,
      "learning_rate": 1.3943685913354376e-06,
      "loss": 0.9091,
      "step": 2459
    },
    {
      "epoch": 0.8308004052684904,
      "grad_norm": 2.234417200088501,
      "learning_rate": 1.3889510675015826e-06,
      "loss": 0.8438,
      "step": 2460
    },
    {
      "epoch": 0.8311381290104695,
      "grad_norm": 2.406367778778076,
      "learning_rate": 1.3835433029379352e-06,
      "loss": 0.9774,
      "step": 2461
    },
    {
      "epoch": 0.8314758527524485,
      "grad_norm": 2.2194106578826904,
      "learning_rate": 1.378145303773346e-06,
      "loss": 0.8723,
      "step": 2462
    },
    {
      "epoch": 0.8318135764944276,
      "grad_norm": 2.698496103286743,
      "learning_rate": 1.3727570761255927e-06,
      "loss": 0.8845,
      "step": 2463
    },
    {
      "epoch": 0.8321513002364066,
      "grad_norm": 2.5093882083892822,
      "learning_rate": 1.3673786261013911e-06,
      "loss": 0.8192,
      "step": 2464
    },
    {
      "epoch": 0.8324890239783856,
      "grad_norm": 2.564704418182373,
      "learning_rate": 1.3620099597963644e-06,
      "loss": 0.8816,
      "step": 2465
    },
    {
      "epoch": 0.8328267477203647,
      "grad_norm": 2.2826569080352783,
      "learning_rate": 1.3566510832950542e-06,
      "loss": 0.8276,
      "step": 2466
    },
    {
      "epoch": 0.8331644714623438,
      "grad_norm": 2.3982207775115967,
      "learning_rate": 1.3513020026709023e-06,
      "loss": 0.9758,
      "step": 2467
    },
    {
      "epoch": 0.8335021952043229,
      "grad_norm": 2.3847405910491943,
      "learning_rate": 1.3459627239862516e-06,
      "loss": 0.8284,
      "step": 2468
    },
    {
      "epoch": 0.8338399189463019,
      "grad_norm": 3.217291831970215,
      "learning_rate": 1.340633253292336e-06,
      "loss": 0.9406,
      "step": 2469
    },
    {
      "epoch": 0.834177642688281,
      "grad_norm": 2.183720588684082,
      "learning_rate": 1.335313596629272e-06,
      "loss": 0.9025,
      "step": 2470
    },
    {
      "epoch": 0.83451536643026,
      "grad_norm": 2.346276044845581,
      "learning_rate": 1.330003760026054e-06,
      "loss": 0.9316,
      "step": 2471
    },
    {
      "epoch": 0.8348530901722391,
      "grad_norm": 2.589066743850708,
      "learning_rate": 1.3247037495005488e-06,
      "loss": 0.8969,
      "step": 2472
    },
    {
      "epoch": 0.8351908139142181,
      "grad_norm": 1.8356684446334839,
      "learning_rate": 1.3194135710594812e-06,
      "loss": 0.5605,
      "step": 2473
    },
    {
      "epoch": 0.8355285376561973,
      "grad_norm": 2.189584970474243,
      "learning_rate": 1.3141332306984378e-06,
      "loss": 0.9328,
      "step": 2474
    },
    {
      "epoch": 0.8358662613981763,
      "grad_norm": 2.120513916015625,
      "learning_rate": 1.3088627344018602e-06,
      "loss": 0.9108,
      "step": 2475
    },
    {
      "epoch": 0.8362039851401554,
      "grad_norm": 2.3951048851013184,
      "learning_rate": 1.3036020881430188e-06,
      "loss": 0.8896,
      "step": 2476
    },
    {
      "epoch": 0.8365417088821344,
      "grad_norm": 2.003607749938965,
      "learning_rate": 1.298351297884034e-06,
      "loss": 0.8927,
      "step": 2477
    },
    {
      "epoch": 0.8368794326241135,
      "grad_norm": 2.0530917644500732,
      "learning_rate": 1.2931103695758507e-06,
      "loss": 0.7597,
      "step": 2478
    },
    {
      "epoch": 0.8372171563660925,
      "grad_norm": 2.5960099697113037,
      "learning_rate": 1.2878793091582353e-06,
      "loss": 0.9177,
      "step": 2479
    },
    {
      "epoch": 0.8375548801080716,
      "grad_norm": 2.157999038696289,
      "learning_rate": 1.2826581225597767e-06,
      "loss": 0.8809,
      "step": 2480
    },
    {
      "epoch": 0.8378926038500507,
      "grad_norm": 2.703725814819336,
      "learning_rate": 1.2774468156978615e-06,
      "loss": 0.8108,
      "step": 2481
    },
    {
      "epoch": 0.8382303275920298,
      "grad_norm": 1.7827895879745483,
      "learning_rate": 1.2722453944786928e-06,
      "loss": 0.5379,
      "step": 2482
    },
    {
      "epoch": 0.8385680513340088,
      "grad_norm": 2.1449601650238037,
      "learning_rate": 1.2670538647972652e-06,
      "loss": 0.9624,
      "step": 2483
    },
    {
      "epoch": 0.8389057750759878,
      "grad_norm": 2.363133430480957,
      "learning_rate": 1.2618722325373533e-06,
      "loss": 0.9431,
      "step": 2484
    },
    {
      "epoch": 0.8392434988179669,
      "grad_norm": 2.1707346439361572,
      "learning_rate": 1.256700503571533e-06,
      "loss": 0.8937,
      "step": 2485
    },
    {
      "epoch": 0.8395812225599459,
      "grad_norm": 2.5085253715515137,
      "learning_rate": 1.2515386837611388e-06,
      "loss": 0.9574,
      "step": 2486
    },
    {
      "epoch": 0.839918946301925,
      "grad_norm": 2.956296920776367,
      "learning_rate": 1.2463867789562834e-06,
      "loss": 0.642,
      "step": 2487
    },
    {
      "epoch": 0.8402566700439041,
      "grad_norm": 2.127531051635742,
      "learning_rate": 1.2412447949958417e-06,
      "loss": 0.9121,
      "step": 2488
    },
    {
      "epoch": 0.8405943937858832,
      "grad_norm": 2.448301076889038,
      "learning_rate": 1.2361127377074445e-06,
      "loss": 0.8682,
      "step": 2489
    },
    {
      "epoch": 0.8409321175278622,
      "grad_norm": 2.169731855392456,
      "learning_rate": 1.230990612907471e-06,
      "loss": 0.8523,
      "step": 2490
    },
    {
      "epoch": 0.8412698412698413,
      "grad_norm": 2.217228889465332,
      "learning_rate": 1.2258784264010458e-06,
      "loss": 0.8909,
      "step": 2491
    },
    {
      "epoch": 0.8416075650118203,
      "grad_norm": 2.3277409076690674,
      "learning_rate": 1.2207761839820264e-06,
      "loss": 0.886,
      "step": 2492
    },
    {
      "epoch": 0.8419452887537994,
      "grad_norm": 2.239809274673462,
      "learning_rate": 1.2156838914330072e-06,
      "loss": 0.893,
      "step": 2493
    },
    {
      "epoch": 0.8422830124957784,
      "grad_norm": 2.2831108570098877,
      "learning_rate": 1.2106015545252947e-06,
      "loss": 0.9588,
      "step": 2494
    },
    {
      "epoch": 0.8426207362377575,
      "grad_norm": 2.5422134399414062,
      "learning_rate": 1.2055291790189193e-06,
      "loss": 0.8327,
      "step": 2495
    },
    {
      "epoch": 0.8429584599797366,
      "grad_norm": 2.0560832023620605,
      "learning_rate": 1.2004667706626294e-06,
      "loss": 0.8631,
      "step": 2496
    },
    {
      "epoch": 0.8432961837217157,
      "grad_norm": 2.524000406265259,
      "learning_rate": 1.1954143351938596e-06,
      "loss": 0.9456,
      "step": 2497
    },
    {
      "epoch": 0.8436339074636947,
      "grad_norm": 2.194988250732422,
      "learning_rate": 1.1903718783387563e-06,
      "loss": 0.9135,
      "step": 2498
    },
    {
      "epoch": 0.8439716312056738,
      "grad_norm": 2.210972547531128,
      "learning_rate": 1.1853394058121492e-06,
      "loss": 0.9481,
      "step": 2499
    },
    {
      "epoch": 0.8443093549476528,
      "grad_norm": 2.2208800315856934,
      "learning_rate": 1.180316923317556e-06,
      "loss": 0.8568,
      "step": 2500
    },
    {
      "epoch": 0.8446470786896318,
      "grad_norm": 2.5883102416992188,
      "learning_rate": 1.1753044365471733e-06,
      "loss": 0.9158,
      "step": 2501
    },
    {
      "epoch": 0.8449848024316109,
      "grad_norm": 2.5092554092407227,
      "learning_rate": 1.1703019511818603e-06,
      "loss": 0.9339,
      "step": 2502
    },
    {
      "epoch": 0.84532252617359,
      "grad_norm": 2.2143142223358154,
      "learning_rate": 1.165309472891153e-06,
      "loss": 0.8544,
      "step": 2503
    },
    {
      "epoch": 0.8456602499155691,
      "grad_norm": 2.0532405376434326,
      "learning_rate": 1.1603270073332428e-06,
      "loss": 0.8866,
      "step": 2504
    },
    {
      "epoch": 0.8459979736575481,
      "grad_norm": 2.3955533504486084,
      "learning_rate": 1.155354560154962e-06,
      "loss": 0.9574,
      "step": 2505
    },
    {
      "epoch": 0.8463356973995272,
      "grad_norm": 2.14290714263916,
      "learning_rate": 1.1503921369918092e-06,
      "loss": 0.8828,
      "step": 2506
    },
    {
      "epoch": 0.8466734211415062,
      "grad_norm": 2.3420231342315674,
      "learning_rate": 1.1454397434679022e-06,
      "loss": 0.8766,
      "step": 2507
    },
    {
      "epoch": 0.8470111448834853,
      "grad_norm": 2.136030435562134,
      "learning_rate": 1.1404973851960033e-06,
      "loss": 0.8651,
      "step": 2508
    },
    {
      "epoch": 0.8473488686254643,
      "grad_norm": 2.369544506072998,
      "learning_rate": 1.1355650677774987e-06,
      "loss": 0.9323,
      "step": 2509
    },
    {
      "epoch": 0.8476865923674435,
      "grad_norm": 2.2353506088256836,
      "learning_rate": 1.1306427968023948e-06,
      "loss": 0.8853,
      "step": 2510
    },
    {
      "epoch": 0.8480243161094225,
      "grad_norm": 2.4282150268554688,
      "learning_rate": 1.1257305778493122e-06,
      "loss": 0.9104,
      "step": 2511
    },
    {
      "epoch": 0.8483620398514016,
      "grad_norm": 2.2982757091522217,
      "learning_rate": 1.120828416485481e-06,
      "loss": 0.9052,
      "step": 2512
    },
    {
      "epoch": 0.8486997635933806,
      "grad_norm": 2.2985551357269287,
      "learning_rate": 1.1159363182667237e-06,
      "loss": 0.8691,
      "step": 2513
    },
    {
      "epoch": 0.8490374873353597,
      "grad_norm": 2.1996283531188965,
      "learning_rate": 1.111054288737473e-06,
      "loss": 0.9141,
      "step": 2514
    },
    {
      "epoch": 0.8493752110773387,
      "grad_norm": 2.3880622386932373,
      "learning_rate": 1.1061823334307354e-06,
      "loss": 0.8648,
      "step": 2515
    },
    {
      "epoch": 0.8497129348193178,
      "grad_norm": 1.9563425779342651,
      "learning_rate": 1.1013204578681081e-06,
      "loss": 0.5251,
      "step": 2516
    },
    {
      "epoch": 0.8500506585612969,
      "grad_norm": 2.512305498123169,
      "learning_rate": 1.0964686675597636e-06,
      "loss": 0.8832,
      "step": 2517
    },
    {
      "epoch": 0.850388382303276,
      "grad_norm": 2.2796037197113037,
      "learning_rate": 1.0916269680044434e-06,
      "loss": 0.9315,
      "step": 2518
    },
    {
      "epoch": 0.850726106045255,
      "grad_norm": 2.558786153793335,
      "learning_rate": 1.0867953646894525e-06,
      "loss": 0.7958,
      "step": 2519
    },
    {
      "epoch": 0.851063829787234,
      "grad_norm": 2.5943210124969482,
      "learning_rate": 1.081973863090653e-06,
      "loss": 0.8103,
      "step": 2520
    },
    {
      "epoch": 0.8514015535292131,
      "grad_norm": 2.052233934402466,
      "learning_rate": 1.0771624686724601e-06,
      "loss": 0.9378,
      "step": 2521
    },
    {
      "epoch": 0.8517392772711921,
      "grad_norm": 2.2113242149353027,
      "learning_rate": 1.0723611868878346e-06,
      "loss": 0.8909,
      "step": 2522
    },
    {
      "epoch": 0.8520770010131712,
      "grad_norm": 2.433570384979248,
      "learning_rate": 1.0675700231782682e-06,
      "loss": 0.891,
      "step": 2523
    },
    {
      "epoch": 0.8524147247551502,
      "grad_norm": 2.726158618927002,
      "learning_rate": 1.0627889829737991e-06,
      "loss": 0.8023,
      "step": 2524
    },
    {
      "epoch": 0.8527524484971294,
      "grad_norm": 2.033104181289673,
      "learning_rate": 1.05801807169298e-06,
      "loss": 0.9129,
      "step": 2525
    },
    {
      "epoch": 0.8530901722391084,
      "grad_norm": 2.6169629096984863,
      "learning_rate": 1.053257294742891e-06,
      "loss": 0.834,
      "step": 2526
    },
    {
      "epoch": 0.8534278959810875,
      "grad_norm": 2.578960657119751,
      "learning_rate": 1.0485066575191237e-06,
      "loss": 0.8315,
      "step": 2527
    },
    {
      "epoch": 0.8537656197230665,
      "grad_norm": 2.1021556854248047,
      "learning_rate": 1.0437661654057773e-06,
      "loss": 0.9232,
      "step": 2528
    },
    {
      "epoch": 0.8541033434650456,
      "grad_norm": 2.3449721336364746,
      "learning_rate": 1.0390358237754571e-06,
      "loss": 0.9659,
      "step": 2529
    },
    {
      "epoch": 0.8544410672070246,
      "grad_norm": 2.206014394760132,
      "learning_rate": 1.0343156379892594e-06,
      "loss": 0.8881,
      "step": 2530
    },
    {
      "epoch": 0.8547787909490037,
      "grad_norm": 2.152017593383789,
      "learning_rate": 1.0296056133967735e-06,
      "loss": 0.9775,
      "step": 2531
    },
    {
      "epoch": 0.8551165146909828,
      "grad_norm": 2.1389689445495605,
      "learning_rate": 1.0249057553360742e-06,
      "loss": 0.9373,
      "step": 2532
    },
    {
      "epoch": 0.8554542384329619,
      "grad_norm": 2.6036739349365234,
      "learning_rate": 1.020216069133707e-06,
      "loss": 0.8932,
      "step": 2533
    },
    {
      "epoch": 0.8557919621749409,
      "grad_norm": 2.368703603744507,
      "learning_rate": 1.015536560104694e-06,
      "loss": 0.8812,
      "step": 2534
    },
    {
      "epoch": 0.85612968591692,
      "grad_norm": 2.0887515544891357,
      "learning_rate": 1.0108672335525293e-06,
      "loss": 0.8866,
      "step": 2535
    },
    {
      "epoch": 0.856467409658899,
      "grad_norm": 2.6979596614837646,
      "learning_rate": 1.0062080947691543e-06,
      "loss": 0.83,
      "step": 2536
    },
    {
      "epoch": 0.856805133400878,
      "grad_norm": 2.5387330055236816,
      "learning_rate": 1.001559149034973e-06,
      "loss": 0.9568,
      "step": 2537
    },
    {
      "epoch": 0.8571428571428571,
      "grad_norm": 2.274622917175293,
      "learning_rate": 9.96920401618834e-07,
      "loss": 0.975,
      "step": 2538
    },
    {
      "epoch": 0.8574805808848363,
      "grad_norm": 2.5581774711608887,
      "learning_rate": 9.922918577780284e-07,
      "loss": 0.8401,
      "step": 2539
    },
    {
      "epoch": 0.8578183046268153,
      "grad_norm": 2.596660852432251,
      "learning_rate": 9.876735227582846e-07,
      "loss": 0.9013,
      "step": 2540
    },
    {
      "epoch": 0.8581560283687943,
      "grad_norm": 2.355773448944092,
      "learning_rate": 9.830654017937536e-07,
      "loss": 0.8557,
      "step": 2541
    },
    {
      "epoch": 0.8584937521107734,
      "grad_norm": 2.435783624649048,
      "learning_rate": 9.78467500107021e-07,
      "loss": 0.9362,
      "step": 2542
    },
    {
      "epoch": 0.8588314758527524,
      "grad_norm": 2.3510801792144775,
      "learning_rate": 9.738798229090874e-07,
      "loss": 0.8437,
      "step": 2543
    },
    {
      "epoch": 0.8591691995947315,
      "grad_norm": 2.2962543964385986,
      "learning_rate": 9.69302375399357e-07,
      "loss": 0.7991,
      "step": 2544
    },
    {
      "epoch": 0.8595069233367105,
      "grad_norm": 2.5320885181427,
      "learning_rate": 9.647351627656542e-07,
      "loss": 0.8529,
      "step": 2545
    },
    {
      "epoch": 0.8598446470786897,
      "grad_norm": 2.4155659675598145,
      "learning_rate": 9.6017819018419e-07,
      "loss": 0.9271,
      "step": 2546
    },
    {
      "epoch": 0.8601823708206687,
      "grad_norm": 3.0298893451690674,
      "learning_rate": 9.556314628195806e-07,
      "loss": 0.7908,
      "step": 2547
    },
    {
      "epoch": 0.8605200945626478,
      "grad_norm": 2.169752836227417,
      "learning_rate": 9.510949858248253e-07,
      "loss": 0.9681,
      "step": 2548
    },
    {
      "epoch": 0.8608578183046268,
      "grad_norm": 2.4905197620391846,
      "learning_rate": 9.46568764341308e-07,
      "loss": 0.8781,
      "step": 2549
    },
    {
      "epoch": 0.8611955420466059,
      "grad_norm": 2.14904522895813,
      "learning_rate": 9.420528034987885e-07,
      "loss": 0.8446,
      "step": 2550
    },
    {
      "epoch": 0.8615332657885849,
      "grad_norm": 2.471031665802002,
      "learning_rate": 9.375471084153986e-07,
      "loss": 0.9458,
      "step": 2551
    },
    {
      "epoch": 0.861870989530564,
      "grad_norm": 2.5052096843719482,
      "learning_rate": 9.330516841976356e-07,
      "loss": 0.9519,
      "step": 2552
    },
    {
      "epoch": 0.8622087132725431,
      "grad_norm": 2.4355945587158203,
      "learning_rate": 9.285665359403584e-07,
      "loss": 0.8301,
      "step": 2553
    },
    {
      "epoch": 0.8625464370145222,
      "grad_norm": 2.4321563243865967,
      "learning_rate": 9.240916687267731e-07,
      "loss": 0.8952,
      "step": 2554
    },
    {
      "epoch": 0.8628841607565012,
      "grad_norm": 2.3525125980377197,
      "learning_rate": 9.196270876284386e-07,
      "loss": 0.9013,
      "step": 2555
    },
    {
      "epoch": 0.8632218844984803,
      "grad_norm": 2.119307518005371,
      "learning_rate": 9.151727977052604e-07,
      "loss": 0.9042,
      "step": 2556
    },
    {
      "epoch": 0.8635596082404593,
      "grad_norm": 2.572552442550659,
      "learning_rate": 9.107288040054707e-07,
      "loss": 0.9243,
      "step": 2557
    },
    {
      "epoch": 0.8638973319824383,
      "grad_norm": 2.044543981552124,
      "learning_rate": 9.062951115656404e-07,
      "loss": 0.8726,
      "step": 2558
    },
    {
      "epoch": 0.8642350557244174,
      "grad_norm": 2.244361400604248,
      "learning_rate": 9.018717254106624e-07,
      "loss": 0.8782,
      "step": 2559
    },
    {
      "epoch": 0.8645727794663964,
      "grad_norm": 2.415295362472534,
      "learning_rate": 8.974586505537497e-07,
      "loss": 0.864,
      "step": 2560
    },
    {
      "epoch": 0.8649105032083756,
      "grad_norm": 2.4802701473236084,
      "learning_rate": 8.930558919964294e-07,
      "loss": 0.781,
      "step": 2561
    },
    {
      "epoch": 0.8652482269503546,
      "grad_norm": 2.405062437057495,
      "learning_rate": 8.886634547285311e-07,
      "loss": 0.8641,
      "step": 2562
    },
    {
      "epoch": 0.8655859506923337,
      "grad_norm": 2.4501090049743652,
      "learning_rate": 8.842813437281994e-07,
      "loss": 0.9348,
      "step": 2563
    },
    {
      "epoch": 0.8659236744343127,
      "grad_norm": 2.3122994899749756,
      "learning_rate": 8.79909563961866e-07,
      "loss": 0.8814,
      "step": 2564
    },
    {
      "epoch": 0.8662613981762918,
      "grad_norm": 2.6196277141571045,
      "learning_rate": 8.755481203842509e-07,
      "loss": 0.8182,
      "step": 2565
    },
    {
      "epoch": 0.8665991219182708,
      "grad_norm": 2.387298583984375,
      "learning_rate": 8.711970179383722e-07,
      "loss": 0.9678,
      "step": 2566
    },
    {
      "epoch": 0.8669368456602499,
      "grad_norm": 2.6273133754730225,
      "learning_rate": 8.668562615555154e-07,
      "loss": 0.9181,
      "step": 2567
    },
    {
      "epoch": 0.867274569402229,
      "grad_norm": 2.221956729888916,
      "learning_rate": 8.625258561552463e-07,
      "loss": 0.916,
      "step": 2568
    },
    {
      "epoch": 0.8676122931442081,
      "grad_norm": 2.6760571002960205,
      "learning_rate": 8.582058066454002e-07,
      "loss": 0.8675,
      "step": 2569
    },
    {
      "epoch": 0.8679500168861871,
      "grad_norm": 2.399322271347046,
      "learning_rate": 8.538961179220729e-07,
      "loss": 0.8513,
      "step": 2570
    },
    {
      "epoch": 0.8682877406281662,
      "grad_norm": 2.478769540786743,
      "learning_rate": 8.495967948696193e-07,
      "loss": 0.798,
      "step": 2571
    },
    {
      "epoch": 0.8686254643701452,
      "grad_norm": 2.6005032062530518,
      "learning_rate": 8.453078423606465e-07,
      "loss": 0.9057,
      "step": 2572
    },
    {
      "epoch": 0.8689631881121243,
      "grad_norm": 2.907315731048584,
      "learning_rate": 8.410292652560071e-07,
      "loss": 0.9097,
      "step": 2573
    },
    {
      "epoch": 0.8693009118541033,
      "grad_norm": 2.321065664291382,
      "learning_rate": 8.367610684047966e-07,
      "loss": 0.9839,
      "step": 2574
    },
    {
      "epoch": 0.8696386355960825,
      "grad_norm": 2.344665050506592,
      "learning_rate": 8.325032566443448e-07,
      "loss": 0.8476,
      "step": 2575
    },
    {
      "epoch": 0.8699763593380615,
      "grad_norm": 2.195733070373535,
      "learning_rate": 8.282558348002101e-07,
      "loss": 0.8038,
      "step": 2576
    },
    {
      "epoch": 0.8703140830800405,
      "grad_norm": 2.7513418197631836,
      "learning_rate": 8.240188076861799e-07,
      "loss": 0.9765,
      "step": 2577
    },
    {
      "epoch": 0.8706518068220196,
      "grad_norm": 2.938497304916382,
      "learning_rate": 8.197921801042575e-07,
      "loss": 0.7402,
      "step": 2578
    },
    {
      "epoch": 0.8709895305639986,
      "grad_norm": 2.199756383895874,
      "learning_rate": 8.155759568446608e-07,
      "loss": 0.9606,
      "step": 2579
    },
    {
      "epoch": 0.8713272543059777,
      "grad_norm": 2.6543023586273193,
      "learning_rate": 8.113701426858156e-07,
      "loss": 0.8594,
      "step": 2580
    },
    {
      "epoch": 0.8716649780479567,
      "grad_norm": 2.444899559020996,
      "learning_rate": 8.071747423943521e-07,
      "loss": 0.9063,
      "step": 2581
    },
    {
      "epoch": 0.8720027017899359,
      "grad_norm": 2.447617769241333,
      "learning_rate": 8.029897607250969e-07,
      "loss": 0.8533,
      "step": 2582
    },
    {
      "epoch": 0.8723404255319149,
      "grad_norm": 2.174772024154663,
      "learning_rate": 7.988152024210661e-07,
      "loss": 0.8341,
      "step": 2583
    },
    {
      "epoch": 0.872678149273894,
      "grad_norm": 2.2116329669952393,
      "learning_rate": 7.946510722134693e-07,
      "loss": 0.8733,
      "step": 2584
    },
    {
      "epoch": 0.873015873015873,
      "grad_norm": 2.3341622352600098,
      "learning_rate": 7.904973748216905e-07,
      "loss": 0.9076,
      "step": 2585
    },
    {
      "epoch": 0.8733535967578521,
      "grad_norm": 2.531532049179077,
      "learning_rate": 7.863541149532916e-07,
      "loss": 0.8503,
      "step": 2586
    },
    {
      "epoch": 0.8736913204998311,
      "grad_norm": 2.438175916671753,
      "learning_rate": 7.822212973040122e-07,
      "loss": 0.8548,
      "step": 2587
    },
    {
      "epoch": 0.8740290442418102,
      "grad_norm": 2.5346250534057617,
      "learning_rate": 7.780989265577454e-07,
      "loss": 0.9191,
      "step": 2588
    },
    {
      "epoch": 0.8743667679837892,
      "grad_norm": 2.448096990585327,
      "learning_rate": 7.73987007386553e-07,
      "loss": 0.9048,
      "step": 2589
    },
    {
      "epoch": 0.8747044917257684,
      "grad_norm": 2.3592946529388428,
      "learning_rate": 7.698855444506481e-07,
      "loss": 0.805,
      "step": 2590
    },
    {
      "epoch": 0.8750422154677474,
      "grad_norm": 2.4568774700164795,
      "learning_rate": 7.657945423983948e-07,
      "loss": 0.7931,
      "step": 2591
    },
    {
      "epoch": 0.8753799392097265,
      "grad_norm": 2.777737617492676,
      "learning_rate": 7.617140058663009e-07,
      "loss": 0.8409,
      "step": 2592
    },
    {
      "epoch": 0.8757176629517055,
      "grad_norm": 2.185781478881836,
      "learning_rate": 7.576439394790125e-07,
      "loss": 0.7958,
      "step": 2593
    },
    {
      "epoch": 0.8760553866936845,
      "grad_norm": 2.2117321491241455,
      "learning_rate": 7.535843478493065e-07,
      "loss": 0.9148,
      "step": 2594
    },
    {
      "epoch": 0.8763931104356636,
      "grad_norm": 2.1668050289154053,
      "learning_rate": 7.495352355781005e-07,
      "loss": 0.8822,
      "step": 2595
    },
    {
      "epoch": 0.8767308341776426,
      "grad_norm": 2.5279533863067627,
      "learning_rate": 7.454966072544189e-07,
      "loss": 0.8955,
      "step": 2596
    },
    {
      "epoch": 0.8770685579196218,
      "grad_norm": 2.1592040061950684,
      "learning_rate": 7.414684674554151e-07,
      "loss": 0.9182,
      "step": 2597
    },
    {
      "epoch": 0.8774062816616008,
      "grad_norm": 2.2824387550354004,
      "learning_rate": 7.374508207463516e-07,
      "loss": 0.9177,
      "step": 2598
    },
    {
      "epoch": 0.8777440054035799,
      "grad_norm": 2.4402899742126465,
      "learning_rate": 7.334436716805993e-07,
      "loss": 0.8998,
      "step": 2599
    },
    {
      "epoch": 0.8780817291455589,
      "grad_norm": 2.4641664028167725,
      "learning_rate": 7.294470247996333e-07,
      "loss": 0.8479,
      "step": 2600
    },
    {
      "epoch": 0.878419452887538,
      "grad_norm": 2.2728517055511475,
      "learning_rate": 7.254608846330202e-07,
      "loss": 0.8499,
      "step": 2601
    },
    {
      "epoch": 0.878757176629517,
      "grad_norm": 2.403979778289795,
      "learning_rate": 7.214852556984276e-07,
      "loss": 0.88,
      "step": 2602
    },
    {
      "epoch": 0.8790949003714961,
      "grad_norm": 2.2959272861480713,
      "learning_rate": 7.175201425016065e-07,
      "loss": 0.928,
      "step": 2603
    },
    {
      "epoch": 0.8794326241134752,
      "grad_norm": 2.6694326400756836,
      "learning_rate": 7.135655495363836e-07,
      "loss": 0.8636,
      "step": 2604
    },
    {
      "epoch": 0.8797703478554543,
      "grad_norm": 2.133862018585205,
      "learning_rate": 7.096214812846758e-07,
      "loss": 0.9607,
      "step": 2605
    },
    {
      "epoch": 0.8801080715974333,
      "grad_norm": 2.2767741680145264,
      "learning_rate": 7.056879422164586e-07,
      "loss": 0.8907,
      "step": 2606
    },
    {
      "epoch": 0.8804457953394124,
      "grad_norm": 2.073643445968628,
      "learning_rate": 7.017649367897827e-07,
      "loss": 0.8934,
      "step": 2607
    },
    {
      "epoch": 0.8807835190813914,
      "grad_norm": 3.130528450012207,
      "learning_rate": 6.978524694507571e-07,
      "loss": 0.8052,
      "step": 2608
    },
    {
      "epoch": 0.8811212428233705,
      "grad_norm": 2.2912373542785645,
      "learning_rate": 6.939505446335481e-07,
      "loss": 0.8965,
      "step": 2609
    },
    {
      "epoch": 0.8814589665653495,
      "grad_norm": 2.6678929328918457,
      "learning_rate": 6.900591667603751e-07,
      "loss": 0.804,
      "step": 2610
    },
    {
      "epoch": 0.8817966903073287,
      "grad_norm": 2.6280879974365234,
      "learning_rate": 6.861783402415035e-07,
      "loss": 0.9294,
      "step": 2611
    },
    {
      "epoch": 0.8821344140493077,
      "grad_norm": 2.893681049346924,
      "learning_rate": 6.823080694752382e-07,
      "loss": 0.8513,
      "step": 2612
    },
    {
      "epoch": 0.8824721377912867,
      "grad_norm": 2.59500789642334,
      "learning_rate": 6.784483588479263e-07,
      "loss": 0.9177,
      "step": 2613
    },
    {
      "epoch": 0.8828098615332658,
      "grad_norm": 2.1812243461608887,
      "learning_rate": 6.745992127339385e-07,
      "loss": 0.9042,
      "step": 2614
    },
    {
      "epoch": 0.8831475852752448,
      "grad_norm": 2.168172836303711,
      "learning_rate": 6.707606354956775e-07,
      "loss": 0.8653,
      "step": 2615
    },
    {
      "epoch": 0.8834853090172239,
      "grad_norm": 2.3642845153808594,
      "learning_rate": 6.669326314835711e-07,
      "loss": 0.8809,
      "step": 2616
    },
    {
      "epoch": 0.8838230327592029,
      "grad_norm": 2.504467248916626,
      "learning_rate": 6.631152050360568e-07,
      "loss": 0.8691,
      "step": 2617
    },
    {
      "epoch": 0.8841607565011821,
      "grad_norm": 2.5824992656707764,
      "learning_rate": 6.593083604795869e-07,
      "loss": 0.8651,
      "step": 2618
    },
    {
      "epoch": 0.8844984802431611,
      "grad_norm": 2.267242670059204,
      "learning_rate": 6.555121021286226e-07,
      "loss": 0.863,
      "step": 2619
    },
    {
      "epoch": 0.8848362039851402,
      "grad_norm": 2.7092318534851074,
      "learning_rate": 6.517264342856256e-07,
      "loss": 0.8019,
      "step": 2620
    },
    {
      "epoch": 0.8851739277271192,
      "grad_norm": 2.241924285888672,
      "learning_rate": 6.47951361241056e-07,
      "loss": 0.8817,
      "step": 2621
    },
    {
      "epoch": 0.8855116514690983,
      "grad_norm": 2.449155330657959,
      "learning_rate": 6.441868872733625e-07,
      "loss": 0.8907,
      "step": 2622
    },
    {
      "epoch": 0.8858493752110773,
      "grad_norm": 2.7074837684631348,
      "learning_rate": 6.40433016648988e-07,
      "loss": 0.8699,
      "step": 2623
    },
    {
      "epoch": 0.8861870989530564,
      "grad_norm": 2.287071943283081,
      "learning_rate": 6.366897536223549e-07,
      "loss": 0.9699,
      "step": 2624
    },
    {
      "epoch": 0.8865248226950354,
      "grad_norm": 2.413039445877075,
      "learning_rate": 6.329571024358594e-07,
      "loss": 0.8681,
      "step": 2625
    },
    {
      "epoch": 0.8868625464370146,
      "grad_norm": 2.3009145259857178,
      "learning_rate": 6.2923506731988e-07,
      "loss": 0.7739,
      "step": 2626
    },
    {
      "epoch": 0.8872002701789936,
      "grad_norm": 2.4422223567962646,
      "learning_rate": 6.255236524927543e-07,
      "loss": 0.9378,
      "step": 2627
    },
    {
      "epoch": 0.8875379939209727,
      "grad_norm": 2.1520895957946777,
      "learning_rate": 6.218228621607902e-07,
      "loss": 0.931,
      "step": 2628
    },
    {
      "epoch": 0.8878757176629517,
      "grad_norm": 2.1370110511779785,
      "learning_rate": 6.181327005182492e-07,
      "loss": 0.9039,
      "step": 2629
    },
    {
      "epoch": 0.8882134414049307,
      "grad_norm": 2.2669124603271484,
      "learning_rate": 6.1445317174735e-07,
      "loss": 0.965,
      "step": 2630
    },
    {
      "epoch": 0.8885511651469098,
      "grad_norm": 2.5111711025238037,
      "learning_rate": 6.107842800182617e-07,
      "loss": 0.8735,
      "step": 2631
    },
    {
      "epoch": 0.8888888888888888,
      "grad_norm": 2.0509142875671387,
      "learning_rate": 6.071260294890957e-07,
      "loss": 0.8886,
      "step": 2632
    },
    {
      "epoch": 0.889226612630868,
      "grad_norm": 2.4578421115875244,
      "learning_rate": 6.034784243059022e-07,
      "loss": 0.9,
      "step": 2633
    },
    {
      "epoch": 0.889564336372847,
      "grad_norm": 2.4420857429504395,
      "learning_rate": 5.998414686026721e-07,
      "loss": 0.9284,
      "step": 2634
    },
    {
      "epoch": 0.8899020601148261,
      "grad_norm": 2.326286792755127,
      "learning_rate": 5.962151665013183e-07,
      "loss": 0.8509,
      "step": 2635
    },
    {
      "epoch": 0.8902397838568051,
      "grad_norm": 2.345550775527954,
      "learning_rate": 5.925995221116853e-07,
      "loss": 0.8649,
      "step": 2636
    },
    {
      "epoch": 0.8905775075987842,
      "grad_norm": 2.471463918685913,
      "learning_rate": 5.889945395315399e-07,
      "loss": 0.9332,
      "step": 2637
    },
    {
      "epoch": 0.8909152313407632,
      "grad_norm": 2.0651121139526367,
      "learning_rate": 5.854002228465605e-07,
      "loss": 0.5412,
      "step": 2638
    },
    {
      "epoch": 0.8912529550827423,
      "grad_norm": 2.1916112899780273,
      "learning_rate": 5.818165761303407e-07,
      "loss": 0.9778,
      "step": 2639
    },
    {
      "epoch": 0.8915906788247214,
      "grad_norm": 2.3212058544158936,
      "learning_rate": 5.782436034443817e-07,
      "loss": 0.8473,
      "step": 2640
    },
    {
      "epoch": 0.8919284025667005,
      "grad_norm": 2.2572948932647705,
      "learning_rate": 5.746813088380865e-07,
      "loss": 0.9341,
      "step": 2641
    },
    {
      "epoch": 0.8922661263086795,
      "grad_norm": 3.0329558849334717,
      "learning_rate": 5.711296963487556e-07,
      "loss": 0.8934,
      "step": 2642
    },
    {
      "epoch": 0.8926038500506586,
      "grad_norm": 2.0531928539276123,
      "learning_rate": 5.675887700015804e-07,
      "loss": 0.9208,
      "step": 2643
    },
    {
      "epoch": 0.8929415737926376,
      "grad_norm": 2.6316936016082764,
      "learning_rate": 5.640585338096493e-07,
      "loss": 0.8233,
      "step": 2644
    },
    {
      "epoch": 0.8932792975346167,
      "grad_norm": 2.3084864616394043,
      "learning_rate": 5.605389917739257e-07,
      "loss": 0.9124,
      "step": 2645
    },
    {
      "epoch": 0.8936170212765957,
      "grad_norm": 2.565397262573242,
      "learning_rate": 5.570301478832584e-07,
      "loss": 0.8541,
      "step": 2646
    },
    {
      "epoch": 0.8939547450185749,
      "grad_norm": 2.124798059463501,
      "learning_rate": 5.535320061143734e-07,
      "loss": 0.9493,
      "step": 2647
    },
    {
      "epoch": 0.8942924687605539,
      "grad_norm": 1.9945629835128784,
      "learning_rate": 5.500445704318614e-07,
      "loss": 0.8293,
      "step": 2648
    },
    {
      "epoch": 0.894630192502533,
      "grad_norm": 2.869319200515747,
      "learning_rate": 5.465678447881828e-07,
      "loss": 0.9345,
      "step": 2649
    },
    {
      "epoch": 0.894967916244512,
      "grad_norm": 2.074357271194458,
      "learning_rate": 5.431018331236593e-07,
      "loss": 0.9343,
      "step": 2650
    },
    {
      "epoch": 0.895305639986491,
      "grad_norm": 2.4249701499938965,
      "learning_rate": 5.396465393664707e-07,
      "loss": 0.7885,
      "step": 2651
    },
    {
      "epoch": 0.8956433637284701,
      "grad_norm": 2.3490028381347656,
      "learning_rate": 5.362019674326502e-07,
      "loss": 0.8071,
      "step": 2652
    },
    {
      "epoch": 0.8959810874704491,
      "grad_norm": 2.38478422164917,
      "learning_rate": 5.32768121226076e-07,
      "loss": 0.8356,
      "step": 2653
    },
    {
      "epoch": 0.8963188112124282,
      "grad_norm": 2.425182819366455,
      "learning_rate": 5.293450046384752e-07,
      "loss": 0.9064,
      "step": 2654
    },
    {
      "epoch": 0.8966565349544073,
      "grad_norm": 2.3163719177246094,
      "learning_rate": 5.259326215494132e-07,
      "loss": 0.9035,
      "step": 2655
    },
    {
      "epoch": 0.8969942586963864,
      "grad_norm": 2.353285312652588,
      "learning_rate": 5.225309758262875e-07,
      "loss": 0.8479,
      "step": 2656
    },
    {
      "epoch": 0.8973319824383654,
      "grad_norm": 2.5590999126434326,
      "learning_rate": 5.191400713243288e-07,
      "loss": 0.9535,
      "step": 2657
    },
    {
      "epoch": 0.8976697061803445,
      "grad_norm": 2.2102458477020264,
      "learning_rate": 5.157599118865953e-07,
      "loss": 0.8579,
      "step": 2658
    },
    {
      "epoch": 0.8980074299223235,
      "grad_norm": 2.3009586334228516,
      "learning_rate": 5.123905013439656e-07,
      "loss": 0.9577,
      "step": 2659
    },
    {
      "epoch": 0.8983451536643026,
      "grad_norm": 2.426560640335083,
      "learning_rate": 5.090318435151376e-07,
      "loss": 0.9134,
      "step": 2660
    },
    {
      "epoch": 0.8986828774062816,
      "grad_norm": 3.1444251537323,
      "learning_rate": 5.056839422066206e-07,
      "loss": 0.8274,
      "step": 2661
    },
    {
      "epoch": 0.8990206011482608,
      "grad_norm": 2.0234322547912598,
      "learning_rate": 5.023468012127363e-07,
      "loss": 0.5555,
      "step": 2662
    },
    {
      "epoch": 0.8993583248902398,
      "grad_norm": 2.3326833248138428,
      "learning_rate": 4.990204243156093e-07,
      "loss": 0.8281,
      "step": 2663
    },
    {
      "epoch": 0.8996960486322189,
      "grad_norm": 2.515031099319458,
      "learning_rate": 4.957048152851618e-07,
      "loss": 0.8972,
      "step": 2664
    },
    {
      "epoch": 0.9000337723741979,
      "grad_norm": 2.923725128173828,
      "learning_rate": 4.9239997787912e-07,
      "loss": 0.7711,
      "step": 2665
    },
    {
      "epoch": 0.900371496116177,
      "grad_norm": 2.1049299240112305,
      "learning_rate": 4.891059158429934e-07,
      "loss": 0.9239,
      "step": 2666
    },
    {
      "epoch": 0.900709219858156,
      "grad_norm": 2.28342604637146,
      "learning_rate": 4.858226329100846e-07,
      "loss": 0.9636,
      "step": 2667
    },
    {
      "epoch": 0.901046943600135,
      "grad_norm": 2.5690648555755615,
      "learning_rate": 4.825501328014814e-07,
      "loss": 0.9272,
      "step": 2668
    },
    {
      "epoch": 0.9013846673421142,
      "grad_norm": 2.728187322616577,
      "learning_rate": 4.792884192260439e-07,
      "loss": 0.8205,
      "step": 2669
    },
    {
      "epoch": 0.9017223910840932,
      "grad_norm": 2.4591076374053955,
      "learning_rate": 4.760374958804148e-07,
      "loss": 0.9236,
      "step": 2670
    },
    {
      "epoch": 0.9020601148260723,
      "grad_norm": 2.684945583343506,
      "learning_rate": 4.727973664490038e-07,
      "loss": 0.7685,
      "step": 2671
    },
    {
      "epoch": 0.9023978385680513,
      "grad_norm": 2.172844886779785,
      "learning_rate": 4.695680346039866e-07,
      "loss": 0.8507,
      "step": 2672
    },
    {
      "epoch": 0.9027355623100304,
      "grad_norm": 2.3776602745056152,
      "learning_rate": 4.6634950400530676e-07,
      "loss": 0.8756,
      "step": 2673
    },
    {
      "epoch": 0.9030732860520094,
      "grad_norm": 2.192049503326416,
      "learning_rate": 4.631417783006575e-07,
      "loss": 0.9072,
      "step": 2674
    },
    {
      "epoch": 0.9034110097939885,
      "grad_norm": 2.2176408767700195,
      "learning_rate": 4.5994486112549643e-07,
      "loss": 0.9354,
      "step": 2675
    },
    {
      "epoch": 0.9037487335359676,
      "grad_norm": 2.4669077396392822,
      "learning_rate": 4.567587561030262e-07,
      "loss": 0.9097,
      "step": 2676
    },
    {
      "epoch": 0.9040864572779467,
      "grad_norm": 2.0413362979888916,
      "learning_rate": 4.5358346684419294e-07,
      "loss": 0.9916,
      "step": 2677
    },
    {
      "epoch": 0.9044241810199257,
      "grad_norm": 2.436234474182129,
      "learning_rate": 4.504189969476891e-07,
      "loss": 0.8228,
      "step": 2678
    },
    {
      "epoch": 0.9047619047619048,
      "grad_norm": 2.988387107849121,
      "learning_rate": 4.4726534999994465e-07,
      "loss": 0.839,
      "step": 2679
    },
    {
      "epoch": 0.9050996285038838,
      "grad_norm": 2.1898293495178223,
      "learning_rate": 4.441225295751217e-07,
      "loss": 0.8346,
      "step": 2680
    },
    {
      "epoch": 0.9054373522458629,
      "grad_norm": 2.242518901824951,
      "learning_rate": 4.4099053923511546e-07,
      "loss": 0.8802,
      "step": 2681
    },
    {
      "epoch": 0.9057750759878419,
      "grad_norm": 2.8509042263031006,
      "learning_rate": 4.3786938252953993e-07,
      "loss": 0.7257,
      "step": 2682
    },
    {
      "epoch": 0.9061127997298211,
      "grad_norm": 2.163294792175293,
      "learning_rate": 4.347590629957388e-07,
      "loss": 0.9273,
      "step": 2683
    },
    {
      "epoch": 0.9064505234718001,
      "grad_norm": 2.5191547870635986,
      "learning_rate": 4.316595841587723e-07,
      "loss": 0.9052,
      "step": 2684
    },
    {
      "epoch": 0.9067882472137792,
      "grad_norm": 2.310823440551758,
      "learning_rate": 4.285709495314061e-07,
      "loss": 0.8911,
      "step": 2685
    },
    {
      "epoch": 0.9071259709557582,
      "grad_norm": 2.4038872718811035,
      "learning_rate": 4.2549316261412897e-07,
      "loss": 0.9064,
      "step": 2686
    },
    {
      "epoch": 0.9074636946977372,
      "grad_norm": 2.568570852279663,
      "learning_rate": 4.224262268951229e-07,
      "loss": 0.8775,
      "step": 2687
    },
    {
      "epoch": 0.9078014184397163,
      "grad_norm": 2.599276542663574,
      "learning_rate": 4.193701458502808e-07,
      "loss": 0.8848,
      "step": 2688
    },
    {
      "epoch": 0.9081391421816953,
      "grad_norm": 2.3898417949676514,
      "learning_rate": 4.1632492294318985e-07,
      "loss": 0.8666,
      "step": 2689
    },
    {
      "epoch": 0.9084768659236744,
      "grad_norm": 2.2964627742767334,
      "learning_rate": 4.1329056162512925e-07,
      "loss": 0.8718,
      "step": 2690
    },
    {
      "epoch": 0.9088145896656535,
      "grad_norm": 2.2831995487213135,
      "learning_rate": 4.1026706533507266e-07,
      "loss": 0.9691,
      "step": 2691
    },
    {
      "epoch": 0.9091523134076326,
      "grad_norm": 2.9007649421691895,
      "learning_rate": 4.0725443749967675e-07,
      "loss": 0.8148,
      "step": 2692
    },
    {
      "epoch": 0.9094900371496116,
      "grad_norm": 2.5191545486450195,
      "learning_rate": 4.0425268153328143e-07,
      "loss": 0.9315,
      "step": 2693
    },
    {
      "epoch": 0.9098277608915907,
      "grad_norm": 2.394815683364868,
      "learning_rate": 4.0126180083790765e-07,
      "loss": 0.8635,
      "step": 2694
    },
    {
      "epoch": 0.9101654846335697,
      "grad_norm": 3.0380797386169434,
      "learning_rate": 3.98281798803245e-07,
      "loss": 0.9155,
      "step": 2695
    },
    {
      "epoch": 0.9105032083755488,
      "grad_norm": 2.2617690563201904,
      "learning_rate": 3.9531267880665746e-07,
      "loss": 0.9493,
      "step": 2696
    },
    {
      "epoch": 0.9108409321175278,
      "grad_norm": 2.5579302310943604,
      "learning_rate": 3.9235444421317883e-07,
      "loss": 0.9494,
      "step": 2697
    },
    {
      "epoch": 0.911178655859507,
      "grad_norm": 2.3290627002716064,
      "learning_rate": 3.894070983755005e-07,
      "loss": 0.8232,
      "step": 2698
    },
    {
      "epoch": 0.911516379601486,
      "grad_norm": 1.9899611473083496,
      "learning_rate": 3.864706446339772e-07,
      "loss": 0.8518,
      "step": 2699
    },
    {
      "epoch": 0.9118541033434651,
      "grad_norm": 2.6049277782440186,
      "learning_rate": 3.835450863166157e-07,
      "loss": 0.8775,
      "step": 2700
    },
    {
      "epoch": 0.9121918270854441,
      "grad_norm": 2.2708237171173096,
      "learning_rate": 3.80630426739077e-07,
      "loss": 0.9576,
      "step": 2701
    },
    {
      "epoch": 0.9125295508274232,
      "grad_norm": 2.101696729660034,
      "learning_rate": 3.7772666920467214e-07,
      "loss": 0.9076,
      "step": 2702
    },
    {
      "epoch": 0.9128672745694022,
      "grad_norm": 2.295213222503662,
      "learning_rate": 3.748338170043486e-07,
      "loss": 0.9552,
      "step": 2703
    },
    {
      "epoch": 0.9132049983113812,
      "grad_norm": 2.128922939300537,
      "learning_rate": 3.719518734167038e-07,
      "loss": 0.9046,
      "step": 2704
    },
    {
      "epoch": 0.9135427220533604,
      "grad_norm": 2.427821159362793,
      "learning_rate": 3.690808417079672e-07,
      "loss": 0.7929,
      "step": 2705
    },
    {
      "epoch": 0.9138804457953394,
      "grad_norm": 2.223255157470703,
      "learning_rate": 3.662207251319994e-07,
      "loss": 0.9006,
      "step": 2706
    },
    {
      "epoch": 0.9142181695373185,
      "grad_norm": 2.119215726852417,
      "learning_rate": 3.633715269302962e-07,
      "loss": 0.9664,
      "step": 2707
    },
    {
      "epoch": 0.9145558932792975,
      "grad_norm": 2.3042349815368652,
      "learning_rate": 3.605332503319736e-07,
      "loss": 0.9216,
      "step": 2708
    },
    {
      "epoch": 0.9148936170212766,
      "grad_norm": 2.1253037452697754,
      "learning_rate": 3.577058985537729e-07,
      "loss": 0.8872,
      "step": 2709
    },
    {
      "epoch": 0.9152313407632556,
      "grad_norm": 2.370483636856079,
      "learning_rate": 3.5488947480005197e-07,
      "loss": 0.8801,
      "step": 2710
    },
    {
      "epoch": 0.9155690645052347,
      "grad_norm": 2.2568812370300293,
      "learning_rate": 3.520839822627853e-07,
      "loss": 0.9157,
      "step": 2711
    },
    {
      "epoch": 0.9159067882472138,
      "grad_norm": 2.173921823501587,
      "learning_rate": 3.492894241215583e-07,
      "loss": 0.8276,
      "step": 2712
    },
    {
      "epoch": 0.9162445119891929,
      "grad_norm": 2.3935601711273193,
      "learning_rate": 3.465058035435587e-07,
      "loss": 0.9672,
      "step": 2713
    },
    {
      "epoch": 0.9165822357311719,
      "grad_norm": 2.4702303409576416,
      "learning_rate": 3.437331236835895e-07,
      "loss": 0.9266,
      "step": 2714
    },
    {
      "epoch": 0.916919959473151,
      "grad_norm": 2.396472454071045,
      "learning_rate": 3.409713876840437e-07,
      "loss": 0.8641,
      "step": 2715
    },
    {
      "epoch": 0.91725768321513,
      "grad_norm": 1.8143154382705688,
      "learning_rate": 3.382205986749154e-07,
      "loss": 0.5756,
      "step": 2716
    },
    {
      "epoch": 0.9175954069571091,
      "grad_norm": 2.2617056369781494,
      "learning_rate": 3.354807597737908e-07,
      "loss": 0.8585,
      "step": 2717
    },
    {
      "epoch": 0.9179331306990881,
      "grad_norm": 2.2662978172302246,
      "learning_rate": 3.327518740858471e-07,
      "loss": 0.8604,
      "step": 2718
    },
    {
      "epoch": 0.9182708544410672,
      "grad_norm": 2.339306354522705,
      "learning_rate": 3.300339447038481e-07,
      "loss": 0.8825,
      "step": 2719
    },
    {
      "epoch": 0.9186085781830463,
      "grad_norm": 2.3818092346191406,
      "learning_rate": 3.273269747081376e-07,
      "loss": 0.8572,
      "step": 2720
    },
    {
      "epoch": 0.9189463019250254,
      "grad_norm": 2.588552951812744,
      "learning_rate": 3.2463096716664254e-07,
      "loss": 0.9498,
      "step": 2721
    },
    {
      "epoch": 0.9192840256670044,
      "grad_norm": 2.124140501022339,
      "learning_rate": 3.2194592513486335e-07,
      "loss": 0.9775,
      "step": 2722
    },
    {
      "epoch": 0.9196217494089834,
      "grad_norm": 2.430352210998535,
      "learning_rate": 3.192718516558724e-07,
      "loss": 0.8295,
      "step": 2723
    },
    {
      "epoch": 0.9199594731509625,
      "grad_norm": 2.426637887954712,
      "learning_rate": 3.1660874976030984e-07,
      "loss": 0.8765,
      "step": 2724
    },
    {
      "epoch": 0.9202971968929415,
      "grad_norm": 2.655494213104248,
      "learning_rate": 3.1395662246638705e-07,
      "loss": 0.8907,
      "step": 2725
    },
    {
      "epoch": 0.9206349206349206,
      "grad_norm": 2.3422131538391113,
      "learning_rate": 3.1131547277986953e-07,
      "loss": 0.8802,
      "step": 2726
    },
    {
      "epoch": 0.9209726443768997,
      "grad_norm": 2.8375446796417236,
      "learning_rate": 3.086853036940862e-07,
      "loss": 0.768,
      "step": 2727
    },
    {
      "epoch": 0.9213103681188788,
      "grad_norm": 2.4171507358551025,
      "learning_rate": 3.060661181899227e-07,
      "loss": 0.8645,
      "step": 2728
    },
    {
      "epoch": 0.9216480918608578,
      "grad_norm": 2.6565449237823486,
      "learning_rate": 3.0345791923581113e-07,
      "loss": 0.7762,
      "step": 2729
    },
    {
      "epoch": 0.9219858156028369,
      "grad_norm": 2.686286687850952,
      "learning_rate": 3.00860709787737e-07,
      "loss": 0.9112,
      "step": 2730
    },
    {
      "epoch": 0.9223235393448159,
      "grad_norm": 2.2659363746643066,
      "learning_rate": 2.9827449278922673e-07,
      "loss": 0.8508,
      "step": 2731
    },
    {
      "epoch": 0.922661263086795,
      "grad_norm": 2.1669507026672363,
      "learning_rate": 2.956992711713524e-07,
      "loss": 0.9115,
      "step": 2732
    },
    {
      "epoch": 0.922998986828774,
      "grad_norm": 2.2087814807891846,
      "learning_rate": 2.9313504785272286e-07,
      "loss": 0.8537,
      "step": 2733
    },
    {
      "epoch": 0.9233367105707532,
      "grad_norm": 2.473419189453125,
      "learning_rate": 2.905818257394799e-07,
      "loss": 0.8683,
      "step": 2734
    },
    {
      "epoch": 0.9236744343127322,
      "grad_norm": 2.5123672485351562,
      "learning_rate": 2.8803960772530116e-07,
      "loss": 0.9132,
      "step": 2735
    },
    {
      "epoch": 0.9240121580547113,
      "grad_norm": 2.3323261737823486,
      "learning_rate": 2.8550839669139294e-07,
      "loss": 0.8742,
      "step": 2736
    },
    {
      "epoch": 0.9243498817966903,
      "grad_norm": 2.6761701107025146,
      "learning_rate": 2.829881955064828e-07,
      "loss": 0.9422,
      "step": 2737
    },
    {
      "epoch": 0.9246876055386694,
      "grad_norm": 2.4500083923339844,
      "learning_rate": 2.804790070268215e-07,
      "loss": 0.9004,
      "step": 2738
    },
    {
      "epoch": 0.9250253292806484,
      "grad_norm": 2.3617799282073975,
      "learning_rate": 2.7798083409618203e-07,
      "loss": 0.8912,
      "step": 2739
    },
    {
      "epoch": 0.9253630530226274,
      "grad_norm": 2.769685745239258,
      "learning_rate": 2.754936795458485e-07,
      "loss": 0.9392,
      "step": 2740
    },
    {
      "epoch": 0.9257007767646066,
      "grad_norm": 3.2125637531280518,
      "learning_rate": 2.7301754619462163e-07,
      "loss": 0.8892,
      "step": 2741
    },
    {
      "epoch": 0.9260385005065856,
      "grad_norm": 2.852872133255005,
      "learning_rate": 2.7055243684880885e-07,
      "loss": 0.8133,
      "step": 2742
    },
    {
      "epoch": 0.9263762242485647,
      "grad_norm": 2.432758331298828,
      "learning_rate": 2.680983543022231e-07,
      "loss": 0.8494,
      "step": 2743
    },
    {
      "epoch": 0.9267139479905437,
      "grad_norm": 2.4813942909240723,
      "learning_rate": 2.656553013361829e-07,
      "loss": 0.9477,
      "step": 2744
    },
    {
      "epoch": 0.9270516717325228,
      "grad_norm": 2.626777172088623,
      "learning_rate": 2.632232807195012e-07,
      "loss": 0.8469,
      "step": 2745
    },
    {
      "epoch": 0.9273893954745018,
      "grad_norm": 2.3429129123687744,
      "learning_rate": 2.6080229520849543e-07,
      "loss": 0.937,
      "step": 2746
    },
    {
      "epoch": 0.9277271192164809,
      "grad_norm": 2.1822566986083984,
      "learning_rate": 2.583923475469685e-07,
      "loss": 0.8405,
      "step": 2747
    },
    {
      "epoch": 0.92806484295846,
      "grad_norm": 2.324636936187744,
      "learning_rate": 2.559934404662179e-07,
      "loss": 0.8944,
      "step": 2748
    },
    {
      "epoch": 0.9284025667004391,
      "grad_norm": 2.251412868499756,
      "learning_rate": 2.5360557668503096e-07,
      "loss": 0.8781,
      "step": 2749
    },
    {
      "epoch": 0.9287402904424181,
      "grad_norm": 2.515371084213257,
      "learning_rate": 2.512287589096718e-07,
      "loss": 0.9161,
      "step": 2750
    },
    {
      "epoch": 0.9290780141843972,
      "grad_norm": 2.3633031845092773,
      "learning_rate": 2.4886298983389124e-07,
      "loss": 0.8696,
      "step": 2751
    },
    {
      "epoch": 0.9294157379263762,
      "grad_norm": 2.5399417877197266,
      "learning_rate": 2.465082721389167e-07,
      "loss": 0.8486,
      "step": 2752
    },
    {
      "epoch": 0.9297534616683553,
      "grad_norm": 2.135014057159424,
      "learning_rate": 2.4416460849345124e-07,
      "loss": 0.882,
      "step": 2753
    },
    {
      "epoch": 0.9300911854103343,
      "grad_norm": 2.269313335418701,
      "learning_rate": 2.4183200155366795e-07,
      "loss": 0.9352,
      "step": 2754
    },
    {
      "epoch": 0.9304289091523134,
      "grad_norm": 2.64288330078125,
      "learning_rate": 2.395104539632098e-07,
      "loss": 0.837,
      "step": 2755
    },
    {
      "epoch": 0.9307666328942925,
      "grad_norm": 2.938790798187256,
      "learning_rate": 2.3719996835318672e-07,
      "loss": 0.8818,
      "step": 2756
    },
    {
      "epoch": 0.9311043566362716,
      "grad_norm": 2.121652364730835,
      "learning_rate": 2.3490054734217282e-07,
      "loss": 0.8903,
      "step": 2757
    },
    {
      "epoch": 0.9314420803782506,
      "grad_norm": 2.6196908950805664,
      "learning_rate": 2.326121935361958e-07,
      "loss": 0.8491,
      "step": 2758
    },
    {
      "epoch": 0.9317798041202296,
      "grad_norm": 2.3162145614624023,
      "learning_rate": 2.3033490952874882e-07,
      "loss": 0.8904,
      "step": 2759
    },
    {
      "epoch": 0.9321175278622087,
      "grad_norm": 2.4087607860565186,
      "learning_rate": 2.2806869790077403e-07,
      "loss": 0.9675,
      "step": 2760
    },
    {
      "epoch": 0.9324552516041877,
      "grad_norm": 2.4764130115509033,
      "learning_rate": 2.2581356122066466e-07,
      "loss": 0.8893,
      "step": 2761
    },
    {
      "epoch": 0.9327929753461668,
      "grad_norm": 2.6377196311950684,
      "learning_rate": 2.2356950204426743e-07,
      "loss": 0.966,
      "step": 2762
    },
    {
      "epoch": 0.9331306990881459,
      "grad_norm": 2.46336030960083,
      "learning_rate": 2.2133652291486452e-07,
      "loss": 0.925,
      "step": 2763
    },
    {
      "epoch": 0.933468422830125,
      "grad_norm": 2.194342851638794,
      "learning_rate": 2.191146263631927e-07,
      "loss": 0.9056,
      "step": 2764
    },
    {
      "epoch": 0.933806146572104,
      "grad_norm": 2.401374340057373,
      "learning_rate": 2.1690381490741984e-07,
      "loss": 0.9392,
      "step": 2765
    },
    {
      "epoch": 0.9341438703140831,
      "grad_norm": 2.216639757156372,
      "learning_rate": 2.1470409105315283e-07,
      "loss": 0.8524,
      "step": 2766
    },
    {
      "epoch": 0.9344815940560621,
      "grad_norm": 2.7402119636535645,
      "learning_rate": 2.1251545729343403e-07,
      "loss": 0.7569,
      "step": 2767
    },
    {
      "epoch": 0.9348193177980412,
      "grad_norm": 2.6275646686553955,
      "learning_rate": 2.103379161087349e-07,
      "loss": 0.8357,
      "step": 2768
    },
    {
      "epoch": 0.9351570415400202,
      "grad_norm": 2.2012383937835693,
      "learning_rate": 2.0817146996695569e-07,
      "loss": 0.9203,
      "step": 2769
    },
    {
      "epoch": 0.9354947652819994,
      "grad_norm": 2.5973784923553467,
      "learning_rate": 2.060161213234224e-07,
      "loss": 0.8323,
      "step": 2770
    },
    {
      "epoch": 0.9358324890239784,
      "grad_norm": 2.240838050842285,
      "learning_rate": 2.0387187262088547e-07,
      "loss": 0.9496,
      "step": 2771
    },
    {
      "epoch": 0.9361702127659575,
      "grad_norm": 2.3886146545410156,
      "learning_rate": 2.0173872628951207e-07,
      "loss": 0.8232,
      "step": 2772
    },
    {
      "epoch": 0.9365079365079365,
      "grad_norm": 2.462351083755493,
      "learning_rate": 1.9961668474688943e-07,
      "loss": 0.9271,
      "step": 2773
    },
    {
      "epoch": 0.9368456602499156,
      "grad_norm": 2.2900800704956055,
      "learning_rate": 1.9750575039801712e-07,
      "loss": 0.9551,
      "step": 2774
    },
    {
      "epoch": 0.9371833839918946,
      "grad_norm": 2.489917516708374,
      "learning_rate": 1.9540592563530802e-07,
      "loss": 0.8912,
      "step": 2775
    },
    {
      "epoch": 0.9375211077338736,
      "grad_norm": 2.157254695892334,
      "learning_rate": 1.9331721283858296e-07,
      "loss": 0.9546,
      "step": 2776
    },
    {
      "epoch": 0.9378588314758528,
      "grad_norm": 2.453890800476074,
      "learning_rate": 1.9123961437506943e-07,
      "loss": 0.9173,
      "step": 2777
    },
    {
      "epoch": 0.9381965552178319,
      "grad_norm": 2.3909475803375244,
      "learning_rate": 1.891731325993984e-07,
      "loss": 0.9572,
      "step": 2778
    },
    {
      "epoch": 0.9385342789598109,
      "grad_norm": 2.3252291679382324,
      "learning_rate": 1.871177698536031e-07,
      "loss": 0.8557,
      "step": 2779
    },
    {
      "epoch": 0.9388720027017899,
      "grad_norm": 2.3540406227111816,
      "learning_rate": 1.8507352846711346e-07,
      "loss": 0.9543,
      "step": 2780
    },
    {
      "epoch": 0.939209726443769,
      "grad_norm": 2.3367300033569336,
      "learning_rate": 1.8304041075675515e-07,
      "loss": 0.9038,
      "step": 2781
    },
    {
      "epoch": 0.939547450185748,
      "grad_norm": 2.2760088443756104,
      "learning_rate": 1.8101841902674832e-07,
      "loss": 0.9241,
      "step": 2782
    },
    {
      "epoch": 0.9398851739277271,
      "grad_norm": 2.219815492630005,
      "learning_rate": 1.790075555687032e-07,
      "loss": 0.8386,
      "step": 2783
    },
    {
      "epoch": 0.9402228976697061,
      "grad_norm": 2.3013269901275635,
      "learning_rate": 1.7700782266161564e-07,
      "loss": 0.855,
      "step": 2784
    },
    {
      "epoch": 0.9405606214116853,
      "grad_norm": 2.270069122314453,
      "learning_rate": 1.7501922257187053e-07,
      "loss": 0.8609,
      "step": 2785
    },
    {
      "epoch": 0.9408983451536643,
      "grad_norm": 2.4949958324432373,
      "learning_rate": 1.7304175755323172e-07,
      "loss": 0.8809,
      "step": 2786
    },
    {
      "epoch": 0.9412360688956434,
      "grad_norm": 2.4086110591888428,
      "learning_rate": 1.7107542984684534e-07,
      "loss": 0.8464,
      "step": 2787
    },
    {
      "epoch": 0.9415737926376224,
      "grad_norm": 2.2768008708953857,
      "learning_rate": 1.6912024168123543e-07,
      "loss": 0.93,
      "step": 2788
    },
    {
      "epoch": 0.9419115163796015,
      "grad_norm": 2.608144760131836,
      "learning_rate": 1.6717619527230056e-07,
      "loss": 0.8908,
      "step": 2789
    },
    {
      "epoch": 0.9422492401215805,
      "grad_norm": 2.4605042934417725,
      "learning_rate": 1.6524329282331054e-07,
      "loss": 0.8807,
      "step": 2790
    },
    {
      "epoch": 0.9425869638635596,
      "grad_norm": 2.204819679260254,
      "learning_rate": 1.6332153652490634e-07,
      "loss": 0.9121,
      "step": 2791
    },
    {
      "epoch": 0.9429246876055387,
      "grad_norm": 2.092426061630249,
      "learning_rate": 1.6141092855509798e-07,
      "loss": 0.8989,
      "step": 2792
    },
    {
      "epoch": 0.9432624113475178,
      "grad_norm": 3.043452501296997,
      "learning_rate": 1.595114710792589e-07,
      "loss": 0.8835,
      "step": 2793
    },
    {
      "epoch": 0.9436001350894968,
      "grad_norm": 2.3303987979888916,
      "learning_rate": 1.5762316625012375e-07,
      "loss": 0.8628,
      "step": 2794
    },
    {
      "epoch": 0.9439378588314759,
      "grad_norm": 2.3846569061279297,
      "learning_rate": 1.5574601620779172e-07,
      "loss": 0.7596,
      "step": 2795
    },
    {
      "epoch": 0.9442755825734549,
      "grad_norm": 1.672577142715454,
      "learning_rate": 1.538800230797166e-07,
      "loss": 0.5331,
      "step": 2796
    },
    {
      "epoch": 0.9446133063154339,
      "grad_norm": 2.5617785453796387,
      "learning_rate": 1.5202518898070672e-07,
      "loss": 0.8664,
      "step": 2797
    },
    {
      "epoch": 0.944951030057413,
      "grad_norm": 2.3889284133911133,
      "learning_rate": 1.5018151601292718e-07,
      "loss": 1.0234,
      "step": 2798
    },
    {
      "epoch": 0.9452887537993921,
      "grad_norm": 2.604764223098755,
      "learning_rate": 1.4834900626588877e-07,
      "loss": 0.889,
      "step": 2799
    },
    {
      "epoch": 0.9456264775413712,
      "grad_norm": 2.1398873329162598,
      "learning_rate": 1.4652766181645573e-07,
      "loss": 0.9129,
      "step": 2800
    },
    {
      "epoch": 0.9459642012833502,
      "grad_norm": 2.423227071762085,
      "learning_rate": 1.4471748472883352e-07,
      "loss": 0.8368,
      "step": 2801
    },
    {
      "epoch": 0.9463019250253293,
      "grad_norm": 2.5064265727996826,
      "learning_rate": 1.4291847705457552e-07,
      "loss": 0.8806,
      "step": 2802
    },
    {
      "epoch": 0.9466396487673083,
      "grad_norm": 2.121506690979004,
      "learning_rate": 1.4113064083257188e-07,
      "loss": 0.8194,
      "step": 2803
    },
    {
      "epoch": 0.9469773725092874,
      "grad_norm": 2.3742144107818604,
      "learning_rate": 1.3935397808905627e-07,
      "loss": 0.8496,
      "step": 2804
    },
    {
      "epoch": 0.9473150962512664,
      "grad_norm": 2.6145994663238525,
      "learning_rate": 1.375884908375935e-07,
      "loss": 0.8221,
      "step": 2805
    },
    {
      "epoch": 0.9476528199932456,
      "grad_norm": 2.190964698791504,
      "learning_rate": 1.3583418107908973e-07,
      "loss": 0.947,
      "step": 2806
    },
    {
      "epoch": 0.9479905437352246,
      "grad_norm": 2.8478968143463135,
      "learning_rate": 1.3409105080177563e-07,
      "loss": 0.7279,
      "step": 2807
    },
    {
      "epoch": 0.9483282674772037,
      "grad_norm": 2.49129319190979,
      "learning_rate": 1.323591019812165e-07,
      "loss": 0.8883,
      "step": 2808
    },
    {
      "epoch": 0.9486659912191827,
      "grad_norm": 2.4546456336975098,
      "learning_rate": 1.3063833658030546e-07,
      "loss": 0.8613,
      "step": 2809
    },
    {
      "epoch": 0.9490037149611618,
      "grad_norm": 2.205420732498169,
      "learning_rate": 1.2892875654925808e-07,
      "loss": 0.8293,
      "step": 2810
    },
    {
      "epoch": 0.9493414387031408,
      "grad_norm": 2.6199357509613037,
      "learning_rate": 1.272303638256156e-07,
      "loss": 0.8948,
      "step": 2811
    },
    {
      "epoch": 0.9496791624451199,
      "grad_norm": 2.2952768802642822,
      "learning_rate": 1.255431603342372e-07,
      "loss": 0.9036,
      "step": 2812
    },
    {
      "epoch": 0.950016886187099,
      "grad_norm": 2.4168577194213867,
      "learning_rate": 1.2386714798730548e-07,
      "loss": 0.9524,
      "step": 2813
    },
    {
      "epoch": 0.950354609929078,
      "grad_norm": 2.4381957054138184,
      "learning_rate": 1.2220232868431548e-07,
      "loss": 0.9531,
      "step": 2814
    },
    {
      "epoch": 0.9506923336710571,
      "grad_norm": 2.411559581756592,
      "learning_rate": 1.2054870431207676e-07,
      "loss": 0.835,
      "step": 2815
    },
    {
      "epoch": 0.9510300574130361,
      "grad_norm": 2.3768248558044434,
      "learning_rate": 1.1890627674471466e-07,
      "loss": 0.9059,
      "step": 2816
    },
    {
      "epoch": 0.9513677811550152,
      "grad_norm": 2.2264740467071533,
      "learning_rate": 1.172750478436635e-07,
      "loss": 0.8861,
      "step": 2817
    },
    {
      "epoch": 0.9517055048969942,
      "grad_norm": 2.2351887226104736,
      "learning_rate": 1.1565501945766223e-07,
      "loss": 0.9043,
      "step": 2818
    },
    {
      "epoch": 0.9520432286389733,
      "grad_norm": 2.3357994556427,
      "learning_rate": 1.1404619342275991e-07,
      "loss": 0.9148,
      "step": 2819
    },
    {
      "epoch": 0.9523809523809523,
      "grad_norm": 2.593235492706299,
      "learning_rate": 1.1244857156230693e-07,
      "loss": 0.8345,
      "step": 2820
    },
    {
      "epoch": 0.9527186761229315,
      "grad_norm": 2.35027813911438,
      "learning_rate": 1.1086215568695712e-07,
      "loss": 0.8553,
      "step": 2821
    },
    {
      "epoch": 0.9530563998649105,
      "grad_norm": 2.3915114402770996,
      "learning_rate": 1.092869475946623e-07,
      "loss": 0.8779,
      "step": 2822
    },
    {
      "epoch": 0.9533941236068896,
      "grad_norm": 2.1561496257781982,
      "learning_rate": 1.0772294907067548e-07,
      "loss": 0.8992,
      "step": 2823
    },
    {
      "epoch": 0.9537318473488686,
      "grad_norm": 2.135462999343872,
      "learning_rate": 1.0617016188754214e-07,
      "loss": 0.8513,
      "step": 2824
    },
    {
      "epoch": 0.9540695710908477,
      "grad_norm": 2.834643840789795,
      "learning_rate": 1.0462858780510232e-07,
      "loss": 0.7957,
      "step": 2825
    },
    {
      "epoch": 0.9544072948328267,
      "grad_norm": 2.3009538650512695,
      "learning_rate": 1.0309822857048624e-07,
      "loss": 0.845,
      "step": 2826
    },
    {
      "epoch": 0.9547450185748058,
      "grad_norm": 2.454481363296509,
      "learning_rate": 1.0157908591811871e-07,
      "loss": 0.9182,
      "step": 2827
    },
    {
      "epoch": 0.9550827423167849,
      "grad_norm": 2.5917739868164062,
      "learning_rate": 1.0007116156970697e-07,
      "loss": 0.8143,
      "step": 2828
    },
    {
      "epoch": 0.955420466058764,
      "grad_norm": 2.346564292907715,
      "learning_rate": 9.857445723424619e-08,
      "loss": 0.901,
      "step": 2829
    },
    {
      "epoch": 0.955758189800743,
      "grad_norm": 2.3578438758850098,
      "learning_rate": 9.708897460801615e-08,
      "loss": 0.9301,
      "step": 2830
    },
    {
      "epoch": 0.956095913542722,
      "grad_norm": 2.477080821990967,
      "learning_rate": 9.56147153745779e-08,
      "loss": 0.8713,
      "step": 2831
    },
    {
      "epoch": 0.9564336372847011,
      "grad_norm": 2.0826821327209473,
      "learning_rate": 9.415168120477158e-08,
      "loss": 0.9775,
      "step": 2832
    },
    {
      "epoch": 0.9567713610266801,
      "grad_norm": 2.4126625061035156,
      "learning_rate": 9.269987375671752e-08,
      "loss": 0.8563,
      "step": 2833
    },
    {
      "epoch": 0.9571090847686592,
      "grad_norm": 2.2559139728546143,
      "learning_rate": 9.125929467581063e-08,
      "loss": 0.8786,
      "step": 2834
    },
    {
      "epoch": 0.9574468085106383,
      "grad_norm": 2.458094596862793,
      "learning_rate": 8.982994559472158e-08,
      "loss": 0.8314,
      "step": 2835
    },
    {
      "epoch": 0.9577845322526174,
      "grad_norm": 2.414198398590088,
      "learning_rate": 8.841182813339011e-08,
      "loss": 0.9306,
      "step": 2836
    },
    {
      "epoch": 0.9581222559945964,
      "grad_norm": 2.866331100463867,
      "learning_rate": 8.700494389903391e-08,
      "loss": 0.7814,
      "step": 2837
    },
    {
      "epoch": 0.9584599797365755,
      "grad_norm": 2.2450053691864014,
      "learning_rate": 8.560929448613197e-08,
      "loss": 0.8308,
      "step": 2838
    },
    {
      "epoch": 0.9587977034785545,
      "grad_norm": 2.1105523109436035,
      "learning_rate": 8.422488147643349e-08,
      "loss": 0.9105,
      "step": 2839
    },
    {
      "epoch": 0.9591354272205336,
      "grad_norm": 2.455264091491699,
      "learning_rate": 8.285170643895446e-08,
      "loss": 0.8467,
      "step": 2840
    },
    {
      "epoch": 0.9594731509625126,
      "grad_norm": 2.3489246368408203,
      "learning_rate": 8.148977092997224e-08,
      "loss": 0.8777,
      "step": 2841
    },
    {
      "epoch": 0.9598108747044918,
      "grad_norm": 2.4936633110046387,
      "learning_rate": 8.013907649302544e-08,
      "loss": 0.8538,
      "step": 2842
    },
    {
      "epoch": 0.9601485984464708,
      "grad_norm": 2.1687724590301514,
      "learning_rate": 7.879962465891622e-08,
      "loss": 0.9408,
      "step": 2843
    },
    {
      "epoch": 0.9604863221884499,
      "grad_norm": 2.1074111461639404,
      "learning_rate": 7.747141694570026e-08,
      "loss": 0.8891,
      "step": 2844
    },
    {
      "epoch": 0.9608240459304289,
      "grad_norm": 2.1181044578552246,
      "learning_rate": 7.615445485869455e-08,
      "loss": 0.992,
      "step": 2845
    },
    {
      "epoch": 0.961161769672408,
      "grad_norm": 2.7221548557281494,
      "learning_rate": 7.484873989046626e-08,
      "loss": 0.8446,
      "step": 2846
    },
    {
      "epoch": 0.961499493414387,
      "grad_norm": 2.3074567317962646,
      "learning_rate": 7.355427352083943e-08,
      "loss": 0.8385,
      "step": 2847
    },
    {
      "epoch": 0.961837217156366,
      "grad_norm": 2.458864450454712,
      "learning_rate": 7.227105721688831e-08,
      "loss": 0.875,
      "step": 2848
    },
    {
      "epoch": 0.9621749408983451,
      "grad_norm": 2.4778177738189697,
      "learning_rate": 7.099909243293624e-08,
      "loss": 0.8531,
      "step": 2849
    },
    {
      "epoch": 0.9625126646403243,
      "grad_norm": 2.1561803817749023,
      "learning_rate": 6.973838061055449e-08,
      "loss": 0.8894,
      "step": 2850
    },
    {
      "epoch": 0.9628503883823033,
      "grad_norm": 2.4095396995544434,
      "learning_rate": 6.848892317856348e-08,
      "loss": 0.8908,
      "step": 2851
    },
    {
      "epoch": 0.9631881121242823,
      "grad_norm": 2.389167547225952,
      "learning_rate": 6.725072155302603e-08,
      "loss": 0.8571,
      "step": 2852
    },
    {
      "epoch": 0.9635258358662614,
      "grad_norm": 2.2854974269866943,
      "learning_rate": 6.602377713724961e-08,
      "loss": 0.9792,
      "step": 2853
    },
    {
      "epoch": 0.9638635596082404,
      "grad_norm": 2.2990057468414307,
      "learning_rate": 6.480809132178078e-08,
      "loss": 0.895,
      "step": 2854
    },
    {
      "epoch": 0.9642012833502195,
      "grad_norm": 2.4433751106262207,
      "learning_rate": 6.360366548440967e-08,
      "loss": 0.9217,
      "step": 2855
    },
    {
      "epoch": 0.9645390070921985,
      "grad_norm": 2.2661097049713135,
      "learning_rate": 6.241050099016544e-08,
      "loss": 0.9877,
      "step": 2856
    },
    {
      "epoch": 0.9648767308341777,
      "grad_norm": 2.360543727874756,
      "learning_rate": 6.122859919130975e-08,
      "loss": 0.8653,
      "step": 2857
    },
    {
      "epoch": 0.9652144545761567,
      "grad_norm": 2.5477449893951416,
      "learning_rate": 6.005796142734333e-08,
      "loss": 0.8547,
      "step": 2858
    },
    {
      "epoch": 0.9655521783181358,
      "grad_norm": 2.448909044265747,
      "learning_rate": 5.889858902500045e-08,
      "loss": 0.83,
      "step": 2859
    },
    {
      "epoch": 0.9658899020601148,
      "grad_norm": 2.627228260040283,
      "learning_rate": 5.775048329824673e-08,
      "loss": 0.9055,
      "step": 2860
    },
    {
      "epoch": 0.9662276258020939,
      "grad_norm": 2.8799641132354736,
      "learning_rate": 5.661364554827908e-08,
      "loss": 0.7907,
      "step": 2861
    },
    {
      "epoch": 0.9665653495440729,
      "grad_norm": 2.3545796871185303,
      "learning_rate": 5.548807706352688e-08,
      "loss": 0.942,
      "step": 2862
    },
    {
      "epoch": 0.966903073286052,
      "grad_norm": 2.0710642337799072,
      "learning_rate": 5.437377911964303e-08,
      "loss": 0.9522,
      "step": 2863
    },
    {
      "epoch": 0.9672407970280311,
      "grad_norm": 2.9301693439483643,
      "learning_rate": 5.327075297950846e-08,
      "loss": 0.815,
      "step": 2864
    },
    {
      "epoch": 0.9675785207700102,
      "grad_norm": 2.3302087783813477,
      "learning_rate": 5.217899989323205e-08,
      "loss": 0.8117,
      "step": 2865
    },
    {
      "epoch": 0.9679162445119892,
      "grad_norm": 2.4523234367370605,
      "learning_rate": 5.109852109814295e-08,
      "loss": 0.9071,
      "step": 2866
    },
    {
      "epoch": 0.9682539682539683,
      "grad_norm": 2.400637149810791,
      "learning_rate": 5.002931781879383e-08,
      "loss": 0.8702,
      "step": 2867
    },
    {
      "epoch": 0.9685916919959473,
      "grad_norm": 2.426236152648926,
      "learning_rate": 4.8971391266957605e-08,
      "loss": 0.8504,
      "step": 2868
    },
    {
      "epoch": 0.9689294157379263,
      "grad_norm": 2.428018808364868,
      "learning_rate": 4.792474264163072e-08,
      "loss": 0.8569,
      "step": 2869
    },
    {
      "epoch": 0.9692671394799054,
      "grad_norm": 2.4327168464660645,
      "learning_rate": 4.6889373129022084e-08,
      "loss": 0.8876,
      "step": 2870
    },
    {
      "epoch": 0.9696048632218845,
      "grad_norm": 2.8769681453704834,
      "learning_rate": 4.5865283902561954e-08,
      "loss": 0.9662,
      "step": 2871
    },
    {
      "epoch": 0.9699425869638636,
      "grad_norm": 2.2810587882995605,
      "learning_rate": 4.485247612289301e-08,
      "loss": 0.9102,
      "step": 2872
    },
    {
      "epoch": 0.9702803107058426,
      "grad_norm": 2.157395362854004,
      "learning_rate": 4.385095093787484e-08,
      "loss": 0.877,
      "step": 2873
    },
    {
      "epoch": 0.9706180344478217,
      "grad_norm": 2.4449779987335205,
      "learning_rate": 4.286070948257948e-08,
      "loss": 0.7638,
      "step": 2874
    },
    {
      "epoch": 0.9709557581898007,
      "grad_norm": 2.231334686279297,
      "learning_rate": 4.1881752879286974e-08,
      "loss": 0.8926,
      "step": 2875
    },
    {
      "epoch": 0.9712934819317798,
      "grad_norm": 2.291013717651367,
      "learning_rate": 4.091408223749316e-08,
      "loss": 0.8788,
      "step": 2876
    },
    {
      "epoch": 0.9716312056737588,
      "grad_norm": 2.1834704875946045,
      "learning_rate": 3.9957698653900756e-08,
      "loss": 0.8799,
      "step": 2877
    },
    {
      "epoch": 0.971968929415738,
      "grad_norm": 2.499150037765503,
      "learning_rate": 3.901260321241829e-08,
      "loss": 0.8883,
      "step": 2878
    },
    {
      "epoch": 0.972306653157717,
      "grad_norm": 2.54516863822937,
      "learning_rate": 3.80787969841645e-08,
      "loss": 0.8746,
      "step": 2879
    },
    {
      "epoch": 0.9726443768996961,
      "grad_norm": 2.310316324234009,
      "learning_rate": 3.7156281027461714e-08,
      "loss": 0.8921,
      "step": 2880
    },
    {
      "epoch": 0.9729821006416751,
      "grad_norm": 2.2050652503967285,
      "learning_rate": 3.624505638783582e-08,
      "loss": 0.9146,
      "step": 2881
    },
    {
      "epoch": 0.9733198243836542,
      "grad_norm": 2.3736412525177,
      "learning_rate": 3.534512409801849e-08,
      "loss": 0.9587,
      "step": 2882
    },
    {
      "epoch": 0.9736575481256332,
      "grad_norm": 2.0755271911621094,
      "learning_rate": 3.445648517793943e-08,
      "loss": 0.8299,
      "step": 2883
    },
    {
      "epoch": 0.9739952718676123,
      "grad_norm": 2.1954877376556396,
      "learning_rate": 3.3579140634731885e-08,
      "loss": 0.8575,
      "step": 2884
    },
    {
      "epoch": 0.9743329956095913,
      "grad_norm": 2.4114885330200195,
      "learning_rate": 3.271309146272938e-08,
      "loss": 0.9159,
      "step": 2885
    },
    {
      "epoch": 0.9746707193515705,
      "grad_norm": 2.4825356006622314,
      "learning_rate": 3.185833864346011e-08,
      "loss": 0.8595,
      "step": 2886
    },
    {
      "epoch": 0.9750084430935495,
      "grad_norm": 2.239011287689209,
      "learning_rate": 3.10148831456536e-08,
      "loss": 0.9477,
      "step": 2887
    },
    {
      "epoch": 0.9753461668355285,
      "grad_norm": 2.3798632621765137,
      "learning_rate": 3.018272592523297e-08,
      "loss": 0.8641,
      "step": 2888
    },
    {
      "epoch": 0.9756838905775076,
      "grad_norm": 2.314568281173706,
      "learning_rate": 2.9361867925319364e-08,
      "loss": 0.921,
      "step": 2889
    },
    {
      "epoch": 0.9760216143194866,
      "grad_norm": 2.6202585697174072,
      "learning_rate": 2.8552310076224167e-08,
      "loss": 0.8307,
      "step": 2890
    },
    {
      "epoch": 0.9763593380614657,
      "grad_norm": 2.231313705444336,
      "learning_rate": 2.7754053295455662e-08,
      "loss": 0.9091,
      "step": 2891
    },
    {
      "epoch": 0.9766970618034447,
      "grad_norm": 1.9781757593154907,
      "learning_rate": 2.6967098487711285e-08,
      "loss": 0.9725,
      "step": 2892
    },
    {
      "epoch": 0.9770347855454239,
      "grad_norm": 2.547541856765747,
      "learning_rate": 2.619144654488093e-08,
      "loss": 0.8865,
      "step": 2893
    },
    {
      "epoch": 0.9773725092874029,
      "grad_norm": 2.1577930450439453,
      "learning_rate": 2.542709834604473e-08,
      "loss": 0.8966,
      "step": 2894
    },
    {
      "epoch": 0.977710233029382,
      "grad_norm": 2.137227773666382,
      "learning_rate": 2.4674054757470867e-08,
      "loss": 0.9045,
      "step": 2895
    },
    {
      "epoch": 0.978047956771361,
      "grad_norm": 2.1224982738494873,
      "learning_rate": 2.3932316632614415e-08,
      "loss": 0.8261,
      "step": 2896
    },
    {
      "epoch": 0.9783856805133401,
      "grad_norm": 2.4405243396759033,
      "learning_rate": 2.3201884812120712e-08,
      "loss": 0.9669,
      "step": 2897
    },
    {
      "epoch": 0.9787234042553191,
      "grad_norm": 2.363354206085205,
      "learning_rate": 2.248276012381978e-08,
      "loss": 0.8744,
      "step": 2898
    },
    {
      "epoch": 0.9790611279972982,
      "grad_norm": 2.2978427410125732,
      "learning_rate": 2.1774943382724124e-08,
      "loss": 0.8961,
      "step": 2899
    },
    {
      "epoch": 0.9793988517392773,
      "grad_norm": 2.239476203918457,
      "learning_rate": 2.107843539103538e-08,
      "loss": 0.8817,
      "step": 2900
    },
    {
      "epoch": 0.9797365754812564,
      "grad_norm": 2.5443384647369385,
      "learning_rate": 2.0393236938133222e-08,
      "loss": 0.9068,
      "step": 2901
    },
    {
      "epoch": 0.9800742992232354,
      "grad_norm": 2.0194156169891357,
      "learning_rate": 1.9719348800585348e-08,
      "loss": 0.9421,
      "step": 2902
    },
    {
      "epoch": 0.9804120229652145,
      "grad_norm": 2.123076915740967,
      "learning_rate": 1.9056771742135272e-08,
      "loss": 0.9092,
      "step": 2903
    },
    {
      "epoch": 0.9807497467071935,
      "grad_norm": 2.693715810775757,
      "learning_rate": 1.8405506513711203e-08,
      "loss": 0.8385,
      "step": 2904
    },
    {
      "epoch": 0.9810874704491725,
      "grad_norm": 2.781344175338745,
      "learning_rate": 1.776555385342049e-08,
      "loss": 0.8984,
      "step": 2905
    },
    {
      "epoch": 0.9814251941911516,
      "grad_norm": 2.723668098449707,
      "learning_rate": 1.7136914486547417e-08,
      "loss": 0.8578,
      "step": 2906
    },
    {
      "epoch": 0.9817629179331308,
      "grad_norm": 3.002384901046753,
      "learning_rate": 1.6519589125555402e-08,
      "loss": 0.7507,
      "step": 2907
    },
    {
      "epoch": 0.9821006416751098,
      "grad_norm": 2.252967357635498,
      "learning_rate": 1.591357847008923e-08,
      "loss": 0.8059,
      "step": 2908
    },
    {
      "epoch": 0.9824383654170888,
      "grad_norm": 2.361051082611084,
      "learning_rate": 1.531888320696284e-08,
      "loss": 0.8796,
      "step": 2909
    },
    {
      "epoch": 0.9827760891590679,
      "grad_norm": 2.7459847927093506,
      "learning_rate": 1.4735504010171542e-08,
      "loss": 0.8572,
      "step": 2910
    },
    {
      "epoch": 0.9831138129010469,
      "grad_norm": 2.520587921142578,
      "learning_rate": 1.4163441540884227e-08,
      "loss": 0.7698,
      "step": 2911
    },
    {
      "epoch": 0.983451536643026,
      "grad_norm": 2.3132646083831787,
      "learning_rate": 1.3602696447442276e-08,
      "loss": 0.8591,
      "step": 2912
    },
    {
      "epoch": 0.983789260385005,
      "grad_norm": 3.138343572616577,
      "learning_rate": 1.3053269365363997e-08,
      "loss": 0.8752,
      "step": 2913
    },
    {
      "epoch": 0.9841269841269841,
      "grad_norm": 2.3329052925109863,
      "learning_rate": 1.2515160917337954e-08,
      "loss": 0.8433,
      "step": 2914
    },
    {
      "epoch": 0.9844647078689632,
      "grad_norm": 2.2660744190216064,
      "learning_rate": 1.1988371713224089e-08,
      "loss": 0.9215,
      "step": 2915
    },
    {
      "epoch": 0.9848024316109423,
      "grad_norm": 2.734771251678467,
      "learning_rate": 1.1472902350057046e-08,
      "loss": 0.7998,
      "step": 2916
    },
    {
      "epoch": 0.9851401553529213,
      "grad_norm": 2.424440383911133,
      "learning_rate": 1.0968753412039513e-08,
      "loss": 0.8313,
      "step": 2917
    },
    {
      "epoch": 0.9854778790949004,
      "grad_norm": 2.437412977218628,
      "learning_rate": 1.0475925470545544e-08,
      "loss": 0.8149,
      "step": 2918
    },
    {
      "epoch": 0.9858156028368794,
      "grad_norm": 2.433354139328003,
      "learning_rate": 9.994419084116137e-09,
      "loss": 0.9204,
      "step": 2919
    },
    {
      "epoch": 0.9861533265788585,
      "grad_norm": 2.1572396755218506,
      "learning_rate": 9.52423479846587e-09,
      "loss": 0.8981,
      "step": 2920
    },
    {
      "epoch": 0.9864910503208375,
      "grad_norm": 2.5323219299316406,
      "learning_rate": 9.065373146472933e-09,
      "loss": 0.9406,
      "step": 2921
    },
    {
      "epoch": 0.9868287740628167,
      "grad_norm": 2.318537473678589,
      "learning_rate": 8.617834648185774e-09,
      "loss": 0.8722,
      "step": 2922
    },
    {
      "epoch": 0.9871664978047957,
      "grad_norm": 2.3500096797943115,
      "learning_rate": 8.181619810818663e-09,
      "loss": 0.8631,
      "step": 2923
    },
    {
      "epoch": 0.9875042215467748,
      "grad_norm": 2.2677595615386963,
      "learning_rate": 7.756729128752804e-09,
      "loss": 0.9085,
      "step": 2924
    },
    {
      "epoch": 0.9878419452887538,
      "grad_norm": 2.627009391784668,
      "learning_rate": 7.343163083533e-09,
      "loss": 0.9328,
      "step": 2925
    },
    {
      "epoch": 0.9881796690307328,
      "grad_norm": 2.273522138595581,
      "learning_rate": 6.940922143874318e-09,
      "loss": 0.8729,
      "step": 2926
    },
    {
      "epoch": 0.9885173927727119,
      "grad_norm": 2.3030664920806885,
      "learning_rate": 6.550006765652095e-09,
      "loss": 0.7919,
      "step": 2927
    },
    {
      "epoch": 0.9888551165146909,
      "grad_norm": 2.312722682952881,
      "learning_rate": 6.170417391907491e-09,
      "loss": 0.934,
      "step": 2928
    },
    {
      "epoch": 0.9891928402566701,
      "grad_norm": 2.7633538246154785,
      "learning_rate": 5.802154452845266e-09,
      "loss": 0.9039,
      "step": 2929
    },
    {
      "epoch": 0.9895305639986491,
      "grad_norm": 2.639570474624634,
      "learning_rate": 5.445218365833782e-09,
      "loss": 0.927,
      "step": 2930
    },
    {
      "epoch": 0.9898682877406282,
      "grad_norm": 2.3472354412078857,
      "learning_rate": 5.099609535403893e-09,
      "loss": 0.9292,
      "step": 2931
    },
    {
      "epoch": 0.9902060114826072,
      "grad_norm": 2.1085309982299805,
      "learning_rate": 4.7653283532500534e-09,
      "loss": 0.93,
      "step": 2932
    },
    {
      "epoch": 0.9905437352245863,
      "grad_norm": 2.254676342010498,
      "learning_rate": 4.4423751982247685e-09,
      "loss": 0.8822,
      "step": 2933
    },
    {
      "epoch": 0.9908814589665653,
      "grad_norm": 2.397109031677246,
      "learning_rate": 4.130750436347475e-09,
      "loss": 0.8562,
      "step": 2934
    },
    {
      "epoch": 0.9912191827085444,
      "grad_norm": 2.297647714614868,
      "learning_rate": 3.830454420794549e-09,
      "loss": 0.9064,
      "step": 2935
    },
    {
      "epoch": 0.9915569064505235,
      "grad_norm": 2.4098708629608154,
      "learning_rate": 3.54148749190375e-09,
      "loss": 0.8918,
      "step": 2936
    },
    {
      "epoch": 0.9918946301925026,
      "grad_norm": 2.8959648609161377,
      "learning_rate": 3.2638499771753262e-09,
      "loss": 0.8694,
      "step": 2937
    },
    {
      "epoch": 0.9922323539344816,
      "grad_norm": 2.420807123184204,
      "learning_rate": 2.997542191265357e-09,
      "loss": 0.9162,
      "step": 2938
    },
    {
      "epoch": 0.9925700776764607,
      "grad_norm": 2.2540388107299805,
      "learning_rate": 2.7425644359946323e-09,
      "loss": 0.9408,
      "step": 2939
    },
    {
      "epoch": 0.9929078014184397,
      "grad_norm": 2.003499746322632,
      "learning_rate": 2.4989170003375527e-09,
      "loss": 0.9048,
      "step": 2940
    },
    {
      "epoch": 0.9932455251604188,
      "grad_norm": 2.4180562496185303,
      "learning_rate": 2.26660016043323e-09,
      "loss": 0.7695,
      "step": 2941
    },
    {
      "epoch": 0.9935832489023978,
      "grad_norm": 2.0967814922332764,
      "learning_rate": 2.0456141795732744e-09,
      "loss": 0.885,
      "step": 2942
    },
    {
      "epoch": 0.993920972644377,
      "grad_norm": 2.693650007247925,
      "learning_rate": 1.835959308212898e-09,
      "loss": 0.8422,
      "step": 2943
    },
    {
      "epoch": 0.994258696386356,
      "grad_norm": 2.0724146366119385,
      "learning_rate": 1.6376357839609225e-09,
      "loss": 0.9101,
      "step": 2944
    },
    {
      "epoch": 0.994596420128335,
      "grad_norm": 2.7921407222747803,
      "learning_rate": 1.4506438315875504e-09,
      "loss": 0.947,
      "step": 2945
    },
    {
      "epoch": 0.9949341438703141,
      "grad_norm": 2.334181785583496,
      "learning_rate": 1.2749836630177038e-09,
      "loss": 0.9292,
      "step": 2946
    },
    {
      "epoch": 0.9952718676122931,
      "grad_norm": 2.4760899543762207,
      "learning_rate": 1.110655477335465e-09,
      "loss": 0.8287,
      "step": 2947
    },
    {
      "epoch": 0.9956095913542722,
      "grad_norm": 2.388657569885254,
      "learning_rate": 9.576594607807465e-10,
      "loss": 0.8482,
      "step": 2948
    },
    {
      "epoch": 0.9959473150962512,
      "grad_norm": 2.380058765411377,
      "learning_rate": 8.159957867504009e-10,
      "loss": 0.9584,
      "step": 2949
    },
    {
      "epoch": 0.9962850388382303,
      "grad_norm": 2.401569128036499,
      "learning_rate": 6.856646157971103e-10,
      "loss": 1.0034,
      "step": 2950
    },
    {
      "epoch": 0.9966227625802094,
      "grad_norm": 2.501185655593872,
      "learning_rate": 5.666660956316072e-10,
      "loss": 0.9499,
      "step": 2951
    },
    {
      "epoch": 0.9969604863221885,
      "grad_norm": 2.622105360031128,
      "learning_rate": 4.5900036111934387e-10,
      "loss": 0.8677,
      "step": 2952
    },
    {
      "epoch": 0.9972982100641675,
      "grad_norm": 2.4056553840637207,
      "learning_rate": 3.626675342838226e-10,
      "loss": 0.8842,
      "step": 2953
    },
    {
      "epoch": 0.9976359338061466,
      "grad_norm": 2.3371286392211914,
      "learning_rate": 2.7766772430215526e-10,
      "loss": 0.8573,
      "step": 2954
    },
    {
      "epoch": 0.9979736575481256,
      "grad_norm": 2.451071262359619,
      "learning_rate": 2.0400102750839368e-10,
      "loss": 0.7888,
      "step": 2955
    },
    {
      "epoch": 0.9983113812901047,
      "grad_norm": 2.5965516567230225,
      "learning_rate": 1.416675273924195e-10,
      "loss": 0.8961,
      "step": 2956
    },
    {
      "epoch": 0.9986491050320837,
      "grad_norm": 2.4250102043151855,
      "learning_rate": 9.066729459883405e-11,
      "loss": 0.8344,
      "step": 2957
    },
    {
      "epoch": 0.9989868287740629,
      "grad_norm": 2.167609930038452,
      "learning_rate": 5.1000386929178633e-11,
      "loss": 0.9751,
      "step": 2958
    },
    {
      "epoch": 0.9993245525160419,
      "grad_norm": 2.205791473388672,
      "learning_rate": 2.2666849338603968e-11,
      "loss": 0.9198,
      "step": 2959
    },
    {
      "epoch": 0.999662276258021,
      "grad_norm": 2.4993886947631836,
      "learning_rate": 5.666713940311042e-12,
      "loss": 0.869,
      "step": 2960
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.651550769805908,
      "learning_rate": 0.0,
      "loss": 0.8757,
      "step": 2961
    }
  ],
  "logging_steps": 1,
  "max_steps": 2961,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 2.259080854980526e+18,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}