{
  "best_metric": 1.8160488605499268,
  "best_model_checkpoint": "pneumonia3/checkpoint-2608",
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 2608,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0003834355828220859,
      "grad_norm": 0.8956754803657532,
      "learning_rate": 1.9157088122605365e-07,
      "loss": 0.0063,
      "step": 1
    },
    {
      "epoch": 0.0007668711656441718,
      "grad_norm": NaN,
      "learning_rate": 1.9157088122605365e-07,
      "loss": 5.584,
      "step": 2
    },
    {
      "epoch": 0.0011503067484662577,
      "grad_norm": NaN,
      "learning_rate": 1.9157088122605365e-07,
      "loss": 2.2579,
      "step": 3
    },
    {
      "epoch": 0.0015337423312883436,
      "grad_norm": NaN,
      "learning_rate": 1.9157088122605365e-07,
      "loss": 1.0205,
      "step": 4
    },
    {
      "epoch": 0.0019171779141104294,
      "grad_norm": 8.993351936340332,
      "learning_rate": 3.831417624521073e-07,
      "loss": 0.0311,
      "step": 5
    },
    {
      "epoch": 0.0023006134969325155,
      "grad_norm": NaN,
      "learning_rate": 3.831417624521073e-07,
      "loss": 1.7237,
      "step": 6
    },
    {
      "epoch": 0.0026840490797546013,
      "grad_norm": 14.72826862335205,
      "learning_rate": 5.747126436781609e-07,
      "loss": 0.1122,
      "step": 7
    },
    {
      "epoch": 0.003067484662576687,
      "grad_norm": 3.6637356281280518,
      "learning_rate": 7.662835249042146e-07,
      "loss": 0.0223,
      "step": 8
    },
    {
      "epoch": 0.003450920245398773,
      "grad_norm": Infinity,
      "learning_rate": 7.662835249042146e-07,
      "loss": 0.7598,
      "step": 9
    },
    {
      "epoch": 0.003834355828220859,
      "grad_norm": 24.12232780456543,
      "learning_rate": 9.578544061302681e-07,
      "loss": 0.6069,
      "step": 10
    },
    {
      "epoch": 0.004217791411042945,
      "grad_norm": 7.987460613250732,
      "learning_rate": 1.1494252873563219e-06,
      "loss": 0.0296,
      "step": 11
    },
    {
      "epoch": 0.004601226993865031,
      "grad_norm": 1.6977758407592773,
      "learning_rate": 1.3409961685823756e-06,
      "loss": 0.008,
      "step": 12
    },
    {
      "epoch": 0.004984662576687116,
      "grad_norm": 0.19353151321411133,
      "learning_rate": 1.5325670498084292e-06,
      "loss": 0.0017,
      "step": 13
    },
    {
      "epoch": 0.005368098159509203,
      "grad_norm": 3.7171666622161865,
      "learning_rate": 1.724137931034483e-06,
      "loss": 0.0126,
      "step": 14
    },
    {
      "epoch": 0.005751533742331288,
      "grad_norm": Infinity,
      "learning_rate": 1.724137931034483e-06,
      "loss": 1.3653,
      "step": 15
    },
    {
      "epoch": 0.006134969325153374,
      "grad_norm": 6.29285192489624,
      "learning_rate": 1.9157088122605362e-06,
      "loss": 0.0114,
      "step": 16
    },
    {
      "epoch": 0.00651840490797546,
      "grad_norm": 213.4350128173828,
      "learning_rate": 2.1072796934865904e-06,
      "loss": 5.9951,
      "step": 17
    },
    {
      "epoch": 0.006901840490797546,
      "grad_norm": 0.3442775011062622,
      "learning_rate": 2.2988505747126437e-06,
      "loss": 0.0021,
      "step": 18
    },
    {
      "epoch": 0.0072852760736196315,
      "grad_norm": 2.453491449356079,
      "learning_rate": 2.4904214559386975e-06,
      "loss": 0.0215,
      "step": 19
    },
    {
      "epoch": 0.007668711656441718,
      "grad_norm": 212.2965850830078,
      "learning_rate": 2.6819923371647512e-06,
      "loss": 5.3477,
      "step": 20
    },
    {
      "epoch": 0.008052147239263804,
      "grad_norm": 2.1834559440612793,
      "learning_rate": 2.8735632183908046e-06,
      "loss": 0.0067,
      "step": 21
    },
    {
      "epoch": 0.00843558282208589,
      "grad_norm": 0.46670106053352356,
      "learning_rate": 3.0651340996168583e-06,
      "loss": 0.0031,
      "step": 22
    },
    {
      "epoch": 0.008819018404907975,
      "grad_norm": NaN,
      "learning_rate": 3.0651340996168583e-06,
      "loss": 4.2522,
      "step": 23
    },
    {
      "epoch": 0.009202453987730062,
      "grad_norm": 4.1455769538879395,
      "learning_rate": 3.2567049808429117e-06,
      "loss": 0.0382,
      "step": 24
    },
    {
      "epoch": 0.009585889570552147,
      "grad_norm": 4.376644134521484,
      "learning_rate": 3.448275862068966e-06,
      "loss": 0.0156,
      "step": 25
    },
    {
      "epoch": 0.009969325153374233,
      "grad_norm": 6.021722793579102,
      "learning_rate": 3.6398467432950196e-06,
      "loss": 0.0658,
      "step": 26
    },
    {
      "epoch": 0.010352760736196318,
      "grad_norm": 2.8319826126098633,
      "learning_rate": 3.8314176245210725e-06,
      "loss": 0.0228,
      "step": 27
    },
    {
      "epoch": 0.010736196319018405,
      "grad_norm": 7.251446723937988,
      "learning_rate": 4.022988505747127e-06,
      "loss": 0.0726,
      "step": 28
    },
    {
      "epoch": 0.01111963190184049,
      "grad_norm": 96.88240051269531,
      "learning_rate": 4.214559386973181e-06,
      "loss": 6.416,
      "step": 29
    },
    {
      "epoch": 0.011503067484662576,
      "grad_norm": 1.113573431968689,
      "learning_rate": 4.406130268199233e-06,
      "loss": 0.0034,
      "step": 30
    },
    {
      "epoch": 0.011886503067484663,
      "grad_norm": 76.65680694580078,
      "learning_rate": 4.5977011494252875e-06,
      "loss": 0.9038,
      "step": 31
    },
    {
      "epoch": 0.012269938650306749,
      "grad_norm": 86.12126159667969,
      "learning_rate": 4.789272030651342e-06,
      "loss": 1.2334,
      "step": 32
    },
    {
      "epoch": 0.012653374233128834,
      "grad_norm": 5.0126142501831055,
      "learning_rate": 4.980842911877395e-06,
      "loss": 0.025,
      "step": 33
    },
    {
      "epoch": 0.01303680981595092,
      "grad_norm": 74.3577651977539,
      "learning_rate": 5.172413793103448e-06,
      "loss": 1.2891,
      "step": 34
    },
    {
      "epoch": 0.013420245398773007,
      "grad_norm": 40.25752258300781,
      "learning_rate": 5.3639846743295025e-06,
      "loss": 0.9468,
      "step": 35
    },
    {
      "epoch": 0.013803680981595092,
      "grad_norm": 0.10465328395366669,
      "learning_rate": 5.555555555555556e-06,
      "loss": 0.0015,
      "step": 36
    },
    {
      "epoch": 0.014187116564417178,
      "grad_norm": 2.5959291458129883,
      "learning_rate": 5.747126436781609e-06,
      "loss": 0.0092,
      "step": 37
    },
    {
      "epoch": 0.014570552147239263,
      "grad_norm": 0.26874759793281555,
      "learning_rate": 5.938697318007663e-06,
      "loss": 0.0018,
      "step": 38
    },
    {
      "epoch": 0.01495398773006135,
      "grad_norm": 0.9604064226150513,
      "learning_rate": 6.130268199233717e-06,
      "loss": 0.0031,
      "step": 39
    },
    {
      "epoch": 0.015337423312883436,
      "grad_norm": 60.562522888183594,
      "learning_rate": 6.321839080459771e-06,
      "loss": 6.1016,
      "step": 40
    },
    {
      "epoch": 0.01572085889570552,
      "grad_norm": 54.024776458740234,
      "learning_rate": 6.513409961685823e-06,
      "loss": 1.0664,
      "step": 41
    },
    {
      "epoch": 0.016104294478527608,
      "grad_norm": 5.381654262542725,
      "learning_rate": 6.7049808429118775e-06,
      "loss": 0.0345,
      "step": 42
    },
    {
      "epoch": 0.016487730061349692,
      "grad_norm": 123.08094787597656,
      "learning_rate": 6.896551724137932e-06,
      "loss": 1.4581,
      "step": 43
    },
    {
      "epoch": 0.01687116564417178,
      "grad_norm": 8.839736938476562,
      "learning_rate": 7.088122605363985e-06,
      "loss": 0.0295,
      "step": 44
    },
    {
      "epoch": 0.017254601226993866,
      "grad_norm": 2.2964253425598145,
      "learning_rate": 7.279693486590039e-06,
      "loss": 0.0139,
      "step": 45
    },
    {
      "epoch": 0.01763803680981595,
      "grad_norm": 1.5887353420257568,
      "learning_rate": 7.4712643678160925e-06,
      "loss": 0.0085,
      "step": 46
    },
    {
      "epoch": 0.018021472392638037,
      "grad_norm": 65.72738647460938,
      "learning_rate": 7.662835249042145e-06,
      "loss": 1.3535,
      "step": 47
    },
    {
      "epoch": 0.018404907975460124,
      "grad_norm": 266.64569091796875,
      "learning_rate": 7.854406130268199e-06,
      "loss": 1.6163,
      "step": 48
    },
    {
      "epoch": 0.018788343558282208,
      "grad_norm": 10.696005821228027,
      "learning_rate": 8.045977011494253e-06,
      "loss": 0.1241,
      "step": 49
    },
    {
      "epoch": 0.019171779141104295,
      "grad_norm": 3.5583226680755615,
      "learning_rate": 8.237547892720307e-06,
      "loss": 0.0094,
      "step": 50
    },
    {
      "epoch": 0.019555214723926382,
      "grad_norm": 2.1616084575653076,
      "learning_rate": 8.429118773946362e-06,
      "loss": 0.0086,
      "step": 51
    },
    {
      "epoch": 0.019938650306748466,
      "grad_norm": 0.8845393061637878,
      "learning_rate": 8.620689655172414e-06,
      "loss": 0.0052,
      "step": 52
    },
    {
      "epoch": 0.020322085889570553,
      "grad_norm": 62.269100189208984,
      "learning_rate": 8.812260536398467e-06,
      "loss": 1.0137,
      "step": 53
    },
    {
      "epoch": 0.020705521472392636,
      "grad_norm": 1.511942744255066,
      "learning_rate": 9.00383141762452e-06,
      "loss": 0.0042,
      "step": 54
    },
    {
      "epoch": 0.021088957055214724,
      "grad_norm": 35.93381881713867,
      "learning_rate": 9.195402298850575e-06,
      "loss": 0.7896,
      "step": 55
    },
    {
      "epoch": 0.02147239263803681,
      "grad_norm": 61.91488265991211,
      "learning_rate": 9.386973180076629e-06,
      "loss": 0.8931,
      "step": 56
    },
    {
      "epoch": 0.021855828220858894,
      "grad_norm": 182.94107055664062,
      "learning_rate": 9.578544061302683e-06,
      "loss": 2.389,
      "step": 57
    },
    {
      "epoch": 0.02223926380368098,
      "grad_norm": 0.3494167923927307,
      "learning_rate": 9.770114942528738e-06,
      "loss": 0.0024,
      "step": 58
    },
    {
      "epoch": 0.02262269938650307,
      "grad_norm": 114.19107818603516,
      "learning_rate": 9.96168582375479e-06,
      "loss": 5.874,
      "step": 59
    },
    {
      "epoch": 0.023006134969325152,
      "grad_norm": 106.06359100341797,
      "learning_rate": 1.0153256704980842e-05,
      "loss": 1.3262,
      "step": 60
    },
    {
      "epoch": 0.02338957055214724,
      "grad_norm": 2.7763619422912598,
      "learning_rate": 1.0344827586206897e-05,
      "loss": 0.0113,
      "step": 61
    },
    {
      "epoch": 0.023773006134969327,
      "grad_norm": 64.10795593261719,
      "learning_rate": 1.053639846743295e-05,
      "loss": 0.7485,
      "step": 62
    },
    {
      "epoch": 0.02415644171779141,
      "grad_norm": 89.75699615478516,
      "learning_rate": 1.0727969348659005e-05,
      "loss": 5.8164,
      "step": 63
    },
    {
      "epoch": 0.024539877300613498,
      "grad_norm": 173.2172393798828,
      "learning_rate": 1.091954022988506e-05,
      "loss": 6.1543,
      "step": 64
    },
    {
      "epoch": 0.02492331288343558,
      "grad_norm": 55.35068893432617,
      "learning_rate": 1.1111111111111112e-05,
      "loss": 0.9883,
      "step": 65
    },
    {
      "epoch": 0.02530674846625767,
      "grad_norm": 6.057254791259766,
      "learning_rate": 1.1302681992337164e-05,
      "loss": 0.0216,
      "step": 66
    },
    {
      "epoch": 0.025690184049079755,
      "grad_norm": 2.4507741928100586,
      "learning_rate": 1.1494252873563218e-05,
      "loss": 0.0169,
      "step": 67
    },
    {
      "epoch": 0.02607361963190184,
      "grad_norm": 80.08985137939453,
      "learning_rate": 1.1685823754789272e-05,
      "loss": 1.002,
      "step": 68
    },
    {
      "epoch": 0.026457055214723926,
      "grad_norm": 188.5212860107422,
      "learning_rate": 1.1877394636015327e-05,
      "loss": 1.4981,
      "step": 69
    },
    {
      "epoch": 0.026840490797546013,
      "grad_norm": 61.822757720947266,
      "learning_rate": 1.206896551724138e-05,
      "loss": 1.2676,
      "step": 70
    },
    {
      "epoch": 0.027223926380368097,
      "grad_norm": 64.24207305908203,
      "learning_rate": 1.2260536398467433e-05,
      "loss": 0.8086,
      "step": 71
    },
    {
      "epoch": 0.027607361963190184,
      "grad_norm": 323.5947265625,
      "learning_rate": 1.2452107279693487e-05,
      "loss": 2.217,
      "step": 72
    },
    {
      "epoch": 0.02799079754601227,
      "grad_norm": 129.00328063964844,
      "learning_rate": 1.2643678160919542e-05,
      "loss": 5.5635,
      "step": 73
    },
    {
      "epoch": 0.028374233128834355,
      "grad_norm": 11.802986145019531,
      "learning_rate": 1.2835249042145594e-05,
      "loss": 0.0487,
      "step": 74
    },
    {
      "epoch": 0.028757668711656442,
      "grad_norm": 0.5383877754211426,
      "learning_rate": 1.3026819923371647e-05,
      "loss": 0.0032,
      "step": 75
    },
    {
      "epoch": 0.029141104294478526,
      "grad_norm": 159.20738220214844,
      "learning_rate": 1.3218390804597702e-05,
      "loss": 5.7246,
      "step": 76
    },
    {
      "epoch": 0.029524539877300613,
      "grad_norm": 60.51280212402344,
      "learning_rate": 1.3409961685823755e-05,
      "loss": 1.0391,
      "step": 77
    },
    {
      "epoch": 0.0299079754601227,
      "grad_norm": 73.97840118408203,
      "learning_rate": 1.360153256704981e-05,
      "loss": 0.9321,
      "step": 78
    },
    {
      "epoch": 0.030291411042944784,
      "grad_norm": 1.2738165855407715,
      "learning_rate": 1.3793103448275863e-05,
      "loss": 0.0052,
      "step": 79
    },
    {
      "epoch": 0.03067484662576687,
      "grad_norm": 0.34937584400177,
      "learning_rate": 1.3984674329501916e-05,
      "loss": 0.0024,
      "step": 80
    },
    {
      "epoch": 0.031058282208588958,
      "grad_norm": 67.0838851928711,
      "learning_rate": 1.417624521072797e-05,
      "loss": 1.3536,
      "step": 81
    },
    {
      "epoch": 0.03144171779141104,
      "grad_norm": 0.8529216051101685,
      "learning_rate": 1.4367816091954022e-05,
      "loss": 0.0033,
      "step": 82
    },
    {
      "epoch": 0.031825153374233126,
      "grad_norm": 37.8706169128418,
      "learning_rate": 1.4559386973180078e-05,
      "loss": 1.1436,
      "step": 83
    },
    {
      "epoch": 0.032208588957055216,
      "grad_norm": 76.21135711669922,
      "learning_rate": 1.475095785440613e-05,
      "loss": 1.5489,
      "step": 84
    },
    {
      "epoch": 0.0325920245398773,
      "grad_norm": 1.0968785285949707,
      "learning_rate": 1.4942528735632185e-05,
      "loss": 0.003,
      "step": 85
    },
    {
      "epoch": 0.032975460122699383,
      "grad_norm": 0.46162062883377075,
      "learning_rate": 1.5134099616858237e-05,
      "loss": 0.0024,
      "step": 86
    },
    {
      "epoch": 0.033358895705521474,
      "grad_norm": 90.52841186523438,
      "learning_rate": 1.532567049808429e-05,
      "loss": 1.2198,
      "step": 87
    },
    {
      "epoch": 0.03374233128834356,
      "grad_norm": 0.3833475410938263,
      "learning_rate": 1.5517241379310346e-05,
      "loss": 0.0024,
      "step": 88
    },
    {
      "epoch": 0.03412576687116564,
      "grad_norm": 314.6944580078125,
      "learning_rate": 1.5708812260536398e-05,
      "loss": 5.8516,
      "step": 89
    },
    {
      "epoch": 0.03450920245398773,
      "grad_norm": 2.8447794914245605,
      "learning_rate": 1.5900383141762454e-05,
      "loss": 0.0128,
      "step": 90
    },
    {
      "epoch": 0.034892638036809816,
      "grad_norm": 149.35581970214844,
      "learning_rate": 1.6091954022988507e-05,
      "loss": 2.0236,
      "step": 91
    },
    {
      "epoch": 0.0352760736196319,
      "grad_norm": 55.89840316772461,
      "learning_rate": 1.628352490421456e-05,
      "loss": 1.1123,
      "step": 92
    },
    {
      "epoch": 0.03565950920245399,
      "grad_norm": 20.16428565979004,
      "learning_rate": 1.6475095785440615e-05,
      "loss": 0.0356,
      "step": 93
    },
    {
      "epoch": 0.036042944785276074,
      "grad_norm": 0.32057636976242065,
      "learning_rate": 1.6666666666666667e-05,
      "loss": 0.0018,
      "step": 94
    },
    {
      "epoch": 0.03642638036809816,
      "grad_norm": 12.989239692687988,
      "learning_rate": 1.6858237547892723e-05,
      "loss": 0.0375,
      "step": 95
    },
    {
      "epoch": 0.03680981595092025,
      "grad_norm": 3.8411009311676025,
      "learning_rate": 1.7049808429118776e-05,
      "loss": 0.0294,
      "step": 96
    },
    {
      "epoch": 0.03719325153374233,
      "grad_norm": 76.6435775756836,
      "learning_rate": 1.7241379310344828e-05,
      "loss": 1.3907,
      "step": 97
    },
    {
      "epoch": 0.037576687116564415,
      "grad_norm": 111.41454315185547,
      "learning_rate": 1.743295019157088e-05,
      "loss": 5.5,
      "step": 98
    },
    {
      "epoch": 0.037960122699386506,
      "grad_norm": 2.332939624786377,
      "learning_rate": 1.7624521072796933e-05,
      "loss": 0.008,
      "step": 99
    },
    {
      "epoch": 0.03834355828220859,
      "grad_norm": 0.13917133212089539,
      "learning_rate": 1.781609195402299e-05,
      "loss": 0.0011,
      "step": 100
    },
    {
      "epoch": 0.03872699386503067,
      "grad_norm": 1.1241354942321777,
      "learning_rate": 1.800766283524904e-05,
      "loss": 0.0037,
      "step": 101
    },
    {
      "epoch": 0.039110429447852764,
      "grad_norm": 0.4089910686016083,
      "learning_rate": 1.8199233716475097e-05,
      "loss": 0.0024,
      "step": 102
    },
    {
      "epoch": 0.03949386503067485,
      "grad_norm": 147.89845275878906,
      "learning_rate": 1.839080459770115e-05,
      "loss": 1.4602,
      "step": 103
    },
    {
      "epoch": 0.03987730061349693,
      "grad_norm": 2.5901684761047363,
      "learning_rate": 1.8582375478927206e-05,
      "loss": 0.0127,
      "step": 104
    },
    {
      "epoch": 0.040260736196319015,
      "grad_norm": 2.627976655960083,
      "learning_rate": 1.8773946360153258e-05,
      "loss": 0.0043,
      "step": 105
    },
    {
      "epoch": 0.040644171779141106,
      "grad_norm": 0.39192333817481995,
      "learning_rate": 1.896551724137931e-05,
      "loss": 0.0052,
      "step": 106
    },
    {
      "epoch": 0.04102760736196319,
      "grad_norm": 12.433937072753906,
      "learning_rate": 1.9157088122605367e-05,
      "loss": 0.0861,
      "step": 107
    },
    {
      "epoch": 0.04141104294478527,
      "grad_norm": 3.929945230484009,
      "learning_rate": 1.934865900383142e-05,
      "loss": 0.0098,
      "step": 108
    },
    {
      "epoch": 0.041794478527607364,
      "grad_norm": 100.13874816894531,
      "learning_rate": 1.9540229885057475e-05,
      "loss": 5.9512,
      "step": 109
    },
    {
      "epoch": 0.04217791411042945,
      "grad_norm": 0.7809451222419739,
      "learning_rate": 1.9731800766283527e-05,
      "loss": 0.0057,
      "step": 110
    },
    {
      "epoch": 0.04256134969325153,
      "grad_norm": 38.45713424682617,
      "learning_rate": 1.992337164750958e-05,
      "loss": 0.7549,
      "step": 111
    },
    {
      "epoch": 0.04294478527607362,
      "grad_norm": 12.913437843322754,
      "learning_rate": 2.0114942528735632e-05,
      "loss": 0.0696,
      "step": 112
    },
    {
      "epoch": 0.043328220858895705,
      "grad_norm": 161.37461853027344,
      "learning_rate": 2.0306513409961685e-05,
      "loss": 5.5381,
      "step": 113
    },
    {
      "epoch": 0.04371165644171779,
      "grad_norm": 13.583711624145508,
      "learning_rate": 2.049808429118774e-05,
      "loss": 0.0394,
      "step": 114
    },
    {
      "epoch": 0.04409509202453988,
      "grad_norm": 109.6414566040039,
      "learning_rate": 2.0689655172413793e-05,
      "loss": 1.2266,
      "step": 115
    },
    {
      "epoch": 0.04447852760736196,
      "grad_norm": 18.10565948486328,
      "learning_rate": 2.088122605363985e-05,
      "loss": 0.0208,
      "step": 116
    },
    {
      "epoch": 0.04486196319018405,
      "grad_norm": 81.25139617919922,
      "learning_rate": 2.10727969348659e-05,
      "loss": 1.2227,
      "step": 117
    },
    {
      "epoch": 0.04524539877300614,
      "grad_norm": 30.773693084716797,
      "learning_rate": 2.1264367816091954e-05,
      "loss": 0.8042,
      "step": 118
    },
    {
      "epoch": 0.04562883435582822,
      "grad_norm": 62.7272834777832,
      "learning_rate": 2.145593869731801e-05,
      "loss": 0.9893,
      "step": 119
    },
    {
      "epoch": 0.046012269938650305,
      "grad_norm": 1.4851429462432861,
      "learning_rate": 2.1647509578544062e-05,
      "loss": 0.0034,
      "step": 120
    },
    {
      "epoch": 0.046395705521472395,
      "grad_norm": 120.005126953125,
      "learning_rate": 2.183908045977012e-05,
      "loss": 1.3565,
      "step": 121
    },
    {
      "epoch": 0.04677914110429448,
      "grad_norm": 296.0460510253906,
      "learning_rate": 2.203065134099617e-05,
      "loss": 7.1914,
      "step": 122
    },
    {
      "epoch": 0.04716257668711656,
      "grad_norm": 128.2917022705078,
      "learning_rate": 2.2222222222222223e-05,
      "loss": 2.8809,
      "step": 123
    },
    {
      "epoch": 0.04754601226993865,
      "grad_norm": 86.54306030273438,
      "learning_rate": 2.2413793103448276e-05,
      "loss": 1.6328,
      "step": 124
    },
    {
      "epoch": 0.04792944785276074,
      "grad_norm": 0.41690361499786377,
      "learning_rate": 2.2605363984674328e-05,
      "loss": 0.0036,
      "step": 125
    },
    {
      "epoch": 0.04831288343558282,
      "grad_norm": 113.13068389892578,
      "learning_rate": 2.2796934865900384e-05,
      "loss": 0.8526,
      "step": 126
    },
    {
      "epoch": 0.04869631901840491,
      "grad_norm": 157.20811462402344,
      "learning_rate": 2.2988505747126437e-05,
      "loss": 4.1294,
      "step": 127
    },
    {
      "epoch": 0.049079754601226995,
      "grad_norm": 76.66633605957031,
      "learning_rate": 2.3180076628352492e-05,
      "loss": 0.6802,
      "step": 128
    },
    {
      "epoch": 0.04946319018404908,
      "grad_norm": 1.2853686809539795,
      "learning_rate": 2.3371647509578545e-05,
      "loss": 0.0071,
      "step": 129
    },
    {
      "epoch": 0.04984662576687116,
      "grad_norm": 110.15524291992188,
      "learning_rate": 2.3563218390804597e-05,
      "loss": 5.2832,
      "step": 130
    },
    {
      "epoch": 0.05023006134969325,
      "grad_norm": 89.71929931640625,
      "learning_rate": 2.3754789272030653e-05,
      "loss": 1.0645,
      "step": 131
    },
    {
      "epoch": 0.05061349693251534,
      "grad_norm": 126.18505096435547,
      "learning_rate": 2.3946360153256706e-05,
      "loss": 1.3311,
      "step": 132
    },
    {
      "epoch": 0.05099693251533742,
      "grad_norm": 6.359533309936523,
      "learning_rate": 2.413793103448276e-05,
      "loss": 0.0912,
      "step": 133
    },
    {
      "epoch": 0.05138036809815951,
      "grad_norm": 2.9896469116210938,
      "learning_rate": 2.4329501915708814e-05,
      "loss": 0.012,
      "step": 134
    },
    {
      "epoch": 0.051763803680981595,
      "grad_norm": 6.923796653747559,
      "learning_rate": 2.4521072796934867e-05,
      "loss": 0.008,
      "step": 135
    },
    {
      "epoch": 0.05214723926380368,
      "grad_norm": 167.52798461914062,
      "learning_rate": 2.4712643678160922e-05,
      "loss": 3.4893,
      "step": 136
    },
    {
      "epoch": 0.05253067484662577,
      "grad_norm": 7.542374610900879,
      "learning_rate": 2.4904214559386975e-05,
      "loss": 0.1313,
      "step": 137
    },
    {
      "epoch": 0.05291411042944785,
      "grad_norm": 100.85726928710938,
      "learning_rate": 2.5095785440613027e-05,
      "loss": 1.0762,
      "step": 138
    },
    {
      "epoch": 0.053297546012269936,
      "grad_norm": 1.1239936351776123,
      "learning_rate": 2.5287356321839083e-05,
      "loss": 0.0035,
      "step": 139
    },
    {
      "epoch": 0.05368098159509203,
      "grad_norm": 0.7821186780929565,
      "learning_rate": 2.5478927203065132e-05,
      "loss": 0.0046,
      "step": 140
    },
    {
      "epoch": 0.05406441717791411,
      "grad_norm": 70.37541961669922,
      "learning_rate": 2.5670498084291188e-05,
      "loss": 1.0537,
      "step": 141
    },
    {
      "epoch": 0.054447852760736194,
      "grad_norm": 0.31846246123313904,
      "learning_rate": 2.5862068965517244e-05,
      "loss": 0.0029,
      "step": 142
    },
    {
      "epoch": 0.054831288343558285,
      "grad_norm": 0.34014156460762024,
      "learning_rate": 2.6053639846743293e-05,
      "loss": 0.0019,
      "step": 143
    },
    {
      "epoch": 0.05521472392638037,
      "grad_norm": 104.0475082397461,
      "learning_rate": 2.624521072796935e-05,
      "loss": 6.5469,
      "step": 144
    },
    {
      "epoch": 0.05559815950920245,
      "grad_norm": 9.99521541595459,
      "learning_rate": 2.6436781609195405e-05,
      "loss": 0.0142,
      "step": 145
    },
    {
      "epoch": 0.05598159509202454,
      "grad_norm": 82.37095642089844,
      "learning_rate": 2.662835249042146e-05,
      "loss": 1.2188,
      "step": 146
    },
    {
      "epoch": 0.056365030674846626,
      "grad_norm": 99.57766723632812,
      "learning_rate": 2.681992337164751e-05,
      "loss": 1.4502,
      "step": 147
    },
    {
      "epoch": 0.05674846625766871,
      "grad_norm": 0.3052326440811157,
      "learning_rate": 2.7011494252873566e-05,
      "loss": 0.0032,
      "step": 148
    },
    {
      "epoch": 0.0571319018404908,
      "grad_norm": 1.5312012434005737,
      "learning_rate": 2.720306513409962e-05,
      "loss": 0.0033,
      "step": 149
    },
    {
      "epoch": 0.057515337423312884,
      "grad_norm": 4.41005802154541,
      "learning_rate": 2.739463601532567e-05,
      "loss": 0.0119,
      "step": 150
    },
    {
      "epoch": 0.05789877300613497,
      "grad_norm": 7.4667277336120605,
      "learning_rate": 2.7586206896551727e-05,
      "loss": 0.0504,
      "step": 151
    },
    {
      "epoch": 0.05828220858895705,
      "grad_norm": 1.4329824447631836,
      "learning_rate": 2.777777777777778e-05,
      "loss": 0.0035,
      "step": 152
    },
    {
      "epoch": 0.05866564417177914,
      "grad_norm": 2.555189609527588,
      "learning_rate": 2.796934865900383e-05,
      "loss": 0.0063,
      "step": 153
    },
    {
      "epoch": 0.059049079754601226,
      "grad_norm": 83.7321548461914,
      "learning_rate": 2.8160919540229884e-05,
      "loss": 1.5372,
      "step": 154
    },
    {
      "epoch": 0.05943251533742331,
      "grad_norm": 0.33591243624687195,
      "learning_rate": 2.835249042145594e-05,
      "loss": 0.0023,
      "step": 155
    },
    {
      "epoch": 0.0598159509202454,
      "grad_norm": 2.3094089031219482,
      "learning_rate": 2.8544061302681996e-05,
      "loss": 0.0108,
      "step": 156
    },
    {
      "epoch": 0.060199386503067484,
      "grad_norm": 1.295427918434143,
      "learning_rate": 2.8735632183908045e-05,
      "loss": 0.0058,
      "step": 157
    },
    {
      "epoch": 0.06058282208588957,
      "grad_norm": 2.300220251083374,
      "learning_rate": 2.89272030651341e-05,
      "loss": 0.0097,
      "step": 158
    },
    {
      "epoch": 0.06096625766871166,
      "grad_norm": 0.4389096200466156,
      "learning_rate": 2.9118773946360157e-05,
      "loss": 0.0021,
      "step": 159
    },
    {
      "epoch": 0.06134969325153374,
      "grad_norm": 1.887438416481018,
      "learning_rate": 2.9310344827586206e-05,
      "loss": 0.0112,
      "step": 160
    },
    {
      "epoch": 0.061733128834355826,
      "grad_norm": 103.8377456665039,
      "learning_rate": 2.950191570881226e-05,
      "loss": 1.5381,
      "step": 161
    },
    {
      "epoch": 0.062116564417177916,
      "grad_norm": 68.52552795410156,
      "learning_rate": 2.9693486590038317e-05,
      "loss": 1.0391,
      "step": 162
    },
    {
      "epoch": 0.0625,
      "grad_norm": 84.98541259765625,
      "learning_rate": 2.988505747126437e-05,
      "loss": 1.629,
      "step": 163
    },
    {
      "epoch": 0.06288343558282208,
      "grad_norm": 120.75799560546875,
      "learning_rate": 3.0076628352490422e-05,
      "loss": 1.2385,
      "step": 164
    },
    {
      "epoch": 0.06326687116564417,
      "grad_norm": 50.883365631103516,
      "learning_rate": 3.0268199233716475e-05,
      "loss": 1.0186,
      "step": 165
    },
    {
      "epoch": 0.06365030674846625,
      "grad_norm": 40.614540100097656,
      "learning_rate": 3.045977011494253e-05,
      "loss": 0.9351,
      "step": 166
    },
    {
      "epoch": 0.06403374233128835,
      "grad_norm": 1.2160786390304565,
      "learning_rate": 3.065134099616858e-05,
      "loss": 0.0038,
      "step": 167
    },
    {
      "epoch": 0.06441717791411043,
      "grad_norm": 1.191338062286377,
      "learning_rate": 3.084291187739464e-05,
      "loss": 0.0026,
      "step": 168
    },
    {
      "epoch": 0.06480061349693252,
      "grad_norm": 40.91733932495117,
      "learning_rate": 3.103448275862069e-05,
      "loss": 0.1254,
      "step": 169
    },
    {
      "epoch": 0.0651840490797546,
      "grad_norm": 20.858539581298828,
      "learning_rate": 3.1226053639846744e-05,
      "loss": 0.6699,
      "step": 170
    },
    {
      "epoch": 0.06556748466257668,
      "grad_norm": 4.083669185638428,
      "learning_rate": 3.1417624521072797e-05,
      "loss": 0.0045,
      "step": 171
    },
    {
      "epoch": 0.06595092024539877,
      "grad_norm": 25.001136779785156,
      "learning_rate": 3.160919540229885e-05,
      "loss": 0.9185,
      "step": 172
    },
    {
      "epoch": 0.06633435582822086,
      "grad_norm": 68.41410827636719,
      "learning_rate": 3.180076628352491e-05,
      "loss": 0.8237,
      "step": 173
    },
    {
      "epoch": 0.06671779141104295,
      "grad_norm": 6.041451454162598,
      "learning_rate": 3.1992337164750954e-05,
      "loss": 0.0305,
      "step": 174
    },
    {
      "epoch": 0.06710122699386503,
      "grad_norm": 0.3134997487068176,
      "learning_rate": 3.218390804597701e-05,
      "loss": 0.0014,
      "step": 175
    },
    {
      "epoch": 0.06748466257668712,
      "grad_norm": 74.52098083496094,
      "learning_rate": 3.2375478927203066e-05,
      "loss": 1.5216,
      "step": 176
    },
    {
      "epoch": 0.0678680981595092,
      "grad_norm": 7.908259868621826,
      "learning_rate": 3.256704980842912e-05,
      "loss": 0.0355,
      "step": 177
    },
    {
      "epoch": 0.06825153374233128,
      "grad_norm": 69.83658599853516,
      "learning_rate": 3.275862068965517e-05,
      "loss": 1.6026,
      "step": 178
    },
    {
      "epoch": 0.06863496932515338,
      "grad_norm": 20.584131240844727,
      "learning_rate": 3.295019157088123e-05,
      "loss": 0.8193,
      "step": 179
    },
    {
      "epoch": 0.06901840490797546,
      "grad_norm": 0.837039589881897,
      "learning_rate": 3.314176245210728e-05,
      "loss": 0.0022,
      "step": 180
    },
    {
      "epoch": 0.06940184049079755,
      "grad_norm": 138.28211975097656,
      "learning_rate": 3.3333333333333335e-05,
      "loss": 3.8125,
      "step": 181
    },
    {
      "epoch": 0.06978527607361963,
      "grad_norm": 121.50843048095703,
      "learning_rate": 3.352490421455939e-05,
      "loss": 0.9912,
      "step": 182
    },
    {
      "epoch": 0.07016871165644172,
      "grad_norm": 1.1271746158599854,
      "learning_rate": 3.371647509578545e-05,
      "loss": 0.0033,
      "step": 183
    },
    {
      "epoch": 0.0705521472392638,
      "grad_norm": 36.171268463134766,
      "learning_rate": 3.390804597701149e-05,
      "loss": 0.6187,
      "step": 184
    },
    {
      "epoch": 0.0709355828220859,
      "grad_norm": 177.51988220214844,
      "learning_rate": 3.409961685823755e-05,
      "loss": 6.2168,
      "step": 185
    },
    {
      "epoch": 0.07131901840490798,
      "grad_norm": 186.8992462158203,
      "learning_rate": 3.4291187739463604e-05,
      "loss": 5.8516,
      "step": 186
    },
    {
      "epoch": 0.07170245398773006,
      "grad_norm": 12.703483581542969,
      "learning_rate": 3.4482758620689657e-05,
      "loss": 0.079,
      "step": 187
    },
    {
      "epoch": 0.07208588957055215,
      "grad_norm": 93.80378723144531,
      "learning_rate": 3.467432950191571e-05,
      "loss": 2.0041,
      "step": 188
    },
    {
      "epoch": 0.07246932515337423,
      "grad_norm": 95.0303955078125,
      "learning_rate": 3.486590038314176e-05,
      "loss": 0.9229,
      "step": 189
    },
    {
      "epoch": 0.07285276073619631,
      "grad_norm": 122.20203399658203,
      "learning_rate": 3.505747126436782e-05,
      "loss": 2.1648,
      "step": 190
    },
    {
      "epoch": 0.0732361963190184,
      "grad_norm": 0.7854564189910889,
      "learning_rate": 3.5249042145593867e-05,
      "loss": 0.0028,
      "step": 191
    },
    {
      "epoch": 0.0736196319018405,
      "grad_norm": 1.2210944890975952,
      "learning_rate": 3.5440613026819926e-05,
      "loss": 0.0054,
      "step": 192
    },
    {
      "epoch": 0.07400306748466258,
      "grad_norm": 20.203170776367188,
      "learning_rate": 3.563218390804598e-05,
      "loss": 0.6558,
      "step": 193
    },
    {
      "epoch": 0.07438650306748466,
      "grad_norm": 149.05807495117188,
      "learning_rate": 3.582375478927204e-05,
      "loss": 2.5322,
      "step": 194
    },
    {
      "epoch": 0.07476993865030675,
      "grad_norm": 186.68226623535156,
      "learning_rate": 3.601532567049808e-05,
      "loss": 2.225,
      "step": 195
    },
    {
      "epoch": 0.07515337423312883,
      "grad_norm": 155.98397827148438,
      "learning_rate": 3.620689655172414e-05,
      "loss": 1.7022,
      "step": 196
    },
    {
      "epoch": 0.07553680981595091,
      "grad_norm": 1.276442050933838,
      "learning_rate": 3.6398467432950195e-05,
      "loss": 0.0058,
      "step": 197
    },
    {
      "epoch": 0.07592024539877301,
      "grad_norm": 10.517167091369629,
      "learning_rate": 3.659003831417625e-05,
      "loss": 0.0524,
      "step": 198
    },
    {
      "epoch": 0.0763036809815951,
      "grad_norm": 9.035552978515625,
      "learning_rate": 3.67816091954023e-05,
      "loss": 0.0112,
      "step": 199
    },
    {
      "epoch": 0.07668711656441718,
      "grad_norm": 2.937300205230713,
      "learning_rate": 3.697318007662835e-05,
      "loss": 0.0202,
      "step": 200
    },
    {
      "epoch": 0.07707055214723926,
      "grad_norm": 0.48456454277038574,
      "learning_rate": 3.716475095785441e-05,
      "loss": 0.0031,
      "step": 201
    },
    {
      "epoch": 0.07745398773006135,
      "grad_norm": 1.5013498067855835,
      "learning_rate": 3.735632183908046e-05,
      "loss": 0.0046,
      "step": 202
    },
    {
      "epoch": 0.07783742331288343,
      "grad_norm": 5.882223606109619,
      "learning_rate": 3.7547892720306517e-05,
      "loss": 0.0604,
      "step": 203
    },
    {
      "epoch": 0.07822085889570553,
      "grad_norm": 35.46553421020508,
      "learning_rate": 3.773946360153257e-05,
      "loss": 0.9717,
      "step": 204
    },
    {
      "epoch": 0.07860429447852761,
      "grad_norm": 5.151394844055176,
      "learning_rate": 3.793103448275862e-05,
      "loss": 0.0493,
      "step": 205
    },
    {
      "epoch": 0.0789877300613497,
      "grad_norm": 154.71730041503906,
      "learning_rate": 3.8122605363984674e-05,
      "loss": 1.7569,
      "step": 206
    },
    {
      "epoch": 0.07937116564417178,
      "grad_norm": 31.566425323486328,
      "learning_rate": 3.831417624521073e-05,
      "loss": 0.1257,
      "step": 207
    },
    {
      "epoch": 0.07975460122699386,
      "grad_norm": 542.98974609375,
      "learning_rate": 3.850574712643678e-05,
      "loss": 5.7324,
      "step": 208
    },
    {
      "epoch": 0.08013803680981595,
      "grad_norm": 150.66156005859375,
      "learning_rate": 3.869731800766284e-05,
      "loss": 5.9746,
      "step": 209
    },
    {
      "epoch": 0.08052147239263803,
      "grad_norm": 0.5510711669921875,
      "learning_rate": 3.888888888888889e-05,
      "loss": 0.0027,
      "step": 210
    },
    {
      "epoch": 0.08090490797546013,
      "grad_norm": 138.30918884277344,
      "learning_rate": 3.908045977011495e-05,
      "loss": 5.8896,
      "step": 211
    },
    {
      "epoch": 0.08128834355828221,
      "grad_norm": 114.1432876586914,
      "learning_rate": 3.9272030651340996e-05,
      "loss": 5.4746,
      "step": 212
    },
    {
      "epoch": 0.0816717791411043,
      "grad_norm": 92.57093048095703,
      "learning_rate": 3.9463601532567055e-05,
      "loss": 1.5499,
      "step": 213
    },
    {
      "epoch": 0.08205521472392638,
      "grad_norm": 1.4484443664550781,
      "learning_rate": 3.965517241379311e-05,
      "loss": 0.0134,
      "step": 214
    },
    {
      "epoch": 0.08243865030674846,
      "grad_norm": 123.93447875976562,
      "learning_rate": 3.984674329501916e-05,
      "loss": 2.5042,
      "step": 215
    },
    {
      "epoch": 0.08282208588957055,
      "grad_norm": 1.8793611526489258,
      "learning_rate": 4.003831417624521e-05,
      "loss": 0.0092,
      "step": 216
    },
    {
      "epoch": 0.08320552147239264,
      "grad_norm": 0.6625899076461792,
      "learning_rate": 4.0229885057471265e-05,
      "loss": 0.0024,
      "step": 217
    },
    {
      "epoch": 0.08358895705521473,
      "grad_norm": 12.705831527709961,
      "learning_rate": 4.0421455938697324e-05,
      "loss": 0.0755,
      "step": 218
    },
    {
      "epoch": 0.08397239263803681,
      "grad_norm": 13.084242820739746,
      "learning_rate": 4.061302681992337e-05,
      "loss": 0.0665,
      "step": 219
    },
    {
      "epoch": 0.0843558282208589,
      "grad_norm": 103.4574203491211,
      "learning_rate": 4.080459770114943e-05,
      "loss": 5.9307,
      "step": 220
    },
    {
      "epoch": 0.08473926380368098,
      "grad_norm": 1.5646629333496094,
      "learning_rate": 4.099616858237548e-05,
      "loss": 0.0032,
      "step": 221
    },
    {
      "epoch": 0.08512269938650306,
      "grad_norm": 13.707032203674316,
      "learning_rate": 4.1187739463601534e-05,
      "loss": 0.1379,
      "step": 222
    },
    {
      "epoch": 0.08550613496932516,
      "grad_norm": 4.0889668464660645,
      "learning_rate": 4.1379310344827587e-05,
      "loss": 0.0058,
      "step": 223
    },
    {
      "epoch": 0.08588957055214724,
      "grad_norm": 1.3660260438919067,
      "learning_rate": 4.1570881226053646e-05,
      "loss": 0.0049,
      "step": 224
    },
    {
      "epoch": 0.08627300613496933,
      "grad_norm": 9.488546371459961,
      "learning_rate": 4.17624521072797e-05,
      "loss": 0.0082,
      "step": 225
    },
    {
      "epoch": 0.08665644171779141,
      "grad_norm": 0.5113505125045776,
      "learning_rate": 4.195402298850575e-05,
      "loss": 0.0033,
      "step": 226
    },
    {
      "epoch": 0.0870398773006135,
      "grad_norm": 1.202599048614502,
      "learning_rate": 4.21455938697318e-05,
      "loss": 0.0024,
      "step": 227
    },
    {
      "epoch": 0.08742331288343558,
      "grad_norm": 191.9002685546875,
      "learning_rate": 4.2337164750957856e-05,
      "loss": 1.5363,
      "step": 228
    },
    {
      "epoch": 0.08780674846625768,
      "grad_norm": 2.334892511367798,
      "learning_rate": 4.252873563218391e-05,
      "loss": 0.0096,
      "step": 229
    },
    {
      "epoch": 0.08819018404907976,
      "grad_norm": 1.950028896331787,
      "learning_rate": 4.272030651340996e-05,
      "loss": 0.0141,
      "step": 230
    },
    {
      "epoch": 0.08857361963190184,
      "grad_norm": 7.90887451171875,
      "learning_rate": 4.291187739463602e-05,
      "loss": 0.0993,
      "step": 231
    },
    {
      "epoch": 0.08895705521472393,
      "grad_norm": 5.312363147735596,
      "learning_rate": 4.3103448275862066e-05,
      "loss": 0.0222,
      "step": 232
    },
    {
      "epoch": 0.08934049079754601,
      "grad_norm": 26.44396209716797,
      "learning_rate": 4.3295019157088125e-05,
      "loss": 0.9038,
      "step": 233
    },
    {
      "epoch": 0.0897239263803681,
      "grad_norm": 1.0253314971923828,
      "learning_rate": 4.348659003831418e-05,
      "loss": 0.0079,
      "step": 234
    },
    {
      "epoch": 0.09010736196319018,
      "grad_norm": 1.3432118892669678,
      "learning_rate": 4.367816091954024e-05,
      "loss": 0.0079,
      "step": 235
    },
    {
      "epoch": 0.09049079754601227,
      "grad_norm": 223.07232666015625,
      "learning_rate": 4.386973180076628e-05,
      "loss": 6.1641,
      "step": 236
    },
    {
      "epoch": 0.09087423312883436,
      "grad_norm": 5.819482326507568,
      "learning_rate": 4.406130268199234e-05,
      "loss": 0.0063,
      "step": 237
    },
    {
      "epoch": 0.09125766871165644,
      "grad_norm": 63.8910026550293,
      "learning_rate": 4.4252873563218394e-05,
      "loss": 1.0508,
      "step": 238
    },
    {
      "epoch": 0.09164110429447853,
      "grad_norm": 23.658857345581055,
      "learning_rate": 4.4444444444444447e-05,
      "loss": 0.1659,
      "step": 239
    },
    {
      "epoch": 0.09202453987730061,
      "grad_norm": 37.76922607421875,
      "learning_rate": 4.46360153256705e-05,
      "loss": 0.6724,
      "step": 240
    },
    {
      "epoch": 0.0924079754601227,
      "grad_norm": 4.410397529602051,
      "learning_rate": 4.482758620689655e-05,
      "loss": 0.0246,
      "step": 241
    },
    {
      "epoch": 0.09279141104294479,
      "grad_norm": 0.8323369026184082,
      "learning_rate": 4.501915708812261e-05,
      "loss": 0.0048,
      "step": 242
    },
    {
      "epoch": 0.09317484662576687,
      "grad_norm": 10.32751178741455,
      "learning_rate": 4.5210727969348656e-05,
      "loss": 0.0696,
      "step": 243
    },
    {
      "epoch": 0.09355828220858896,
      "grad_norm": 128.6432647705078,
      "learning_rate": 4.5402298850574716e-05,
      "loss": 1.2657,
      "step": 244
    },
    {
      "epoch": 0.09394171779141104,
      "grad_norm": 35.12845993041992,
      "learning_rate": 4.559386973180077e-05,
      "loss": 0.6641,
      "step": 245
    },
    {
      "epoch": 0.09432515337423313,
      "grad_norm": 5.09098482131958,
      "learning_rate": 4.578544061302682e-05,
      "loss": 0.0231,
      "step": 246
    },
    {
      "epoch": 0.09470858895705521,
      "grad_norm": 10.12911605834961,
      "learning_rate": 4.597701149425287e-05,
      "loss": 0.1411,
      "step": 247
    },
    {
      "epoch": 0.0950920245398773,
      "grad_norm": 0.48184308409690857,
      "learning_rate": 4.616858237547893e-05,
      "loss": 0.0031,
      "step": 248
    },
    {
      "epoch": 0.09547546012269939,
      "grad_norm": 1.3390072584152222,
      "learning_rate": 4.6360153256704985e-05,
      "loss": 0.0055,
      "step": 249
    },
    {
      "epoch": 0.09585889570552147,
      "grad_norm": 1.018207311630249,
      "learning_rate": 4.655172413793104e-05,
      "loss": 0.0025,
      "step": 250
    },
    {
      "epoch": 0.09624233128834356,
      "grad_norm": 0.7538869380950928,
      "learning_rate": 4.674329501915709e-05,
      "loss": 0.0035,
      "step": 251
    },
    {
      "epoch": 0.09662576687116564,
      "grad_norm": 2.483527421951294,
      "learning_rate": 4.693486590038315e-05,
      "loss": 0.0044,
      "step": 252
    },
    {
      "epoch": 0.09700920245398773,
      "grad_norm": 138.2169952392578,
      "learning_rate": 4.7126436781609195e-05,
      "loss": 2.0765,
      "step": 253
    },
    {
      "epoch": 0.09739263803680982,
      "grad_norm": 20.763120651245117,
      "learning_rate": 4.7318007662835254e-05,
      "loss": 0.142,
      "step": 254
    },
    {
      "epoch": 0.0977760736196319,
      "grad_norm": 190.92677307128906,
      "learning_rate": 4.7509578544061307e-05,
      "loss": 2.5516,
      "step": 255
    },
    {
      "epoch": 0.09815950920245399,
      "grad_norm": 3.3017444610595703,
      "learning_rate": 4.770114942528736e-05,
      "loss": 0.0114,
      "step": 256
    },
    {
      "epoch": 0.09854294478527607,
      "grad_norm": 122.03047180175781,
      "learning_rate": 4.789272030651341e-05,
      "loss": 5.3203,
      "step": 257
    },
    {
      "epoch": 0.09892638036809816,
      "grad_norm": 0.15875087678432465,
      "learning_rate": 4.8084291187739464e-05,
      "loss": 0.0017,
      "step": 258
    },
    {
      "epoch": 0.09930981595092024,
      "grad_norm": 1.3615827560424805,
      "learning_rate": 4.827586206896552e-05,
      "loss": 0.004,
      "step": 259
    },
    {
      "epoch": 0.09969325153374232,
      "grad_norm": 1.167940378189087,
      "learning_rate": 4.846743295019157e-05,
      "loss": 0.0071,
      "step": 260
    },
    {
      "epoch": 0.10007668711656442,
      "grad_norm": 0.7109537124633789,
      "learning_rate": 4.865900383141763e-05,
      "loss": 0.0034,
      "step": 261
    },
    {
      "epoch": 0.1004601226993865,
      "grad_norm": 0.3329644203186035,
      "learning_rate": 4.885057471264368e-05,
      "loss": 0.0038,
      "step": 262
    },
    {
      "epoch": 0.10084355828220859,
      "grad_norm": 102.13310241699219,
      "learning_rate": 4.904214559386973e-05,
      "loss": 2.0372,
      "step": 263
    },
    {
      "epoch": 0.10122699386503067,
      "grad_norm": 103.58604431152344,
      "learning_rate": 4.9233716475095786e-05,
      "loss": 1.6924,
      "step": 264
    },
    {
      "epoch": 0.10161042944785276,
      "grad_norm": 3.0650875568389893,
      "learning_rate": 4.9425287356321845e-05,
      "loss": 0.0152,
      "step": 265
    },
    {
      "epoch": 0.10199386503067484,
      "grad_norm": 0.4443052113056183,
      "learning_rate": 4.96168582375479e-05,
      "loss": 0.0024,
      "step": 266
    },
    {
      "epoch": 0.10237730061349694,
      "grad_norm": 0.8741576075553894,
      "learning_rate": 4.980842911877395e-05,
      "loss": 0.0033,
      "step": 267
    },
    {
      "epoch": 0.10276073619631902,
      "grad_norm": 0.7627093195915222,
      "learning_rate": 5e-05,
      "loss": 0.0032,
      "step": 268
    },
    {
      "epoch": 0.1031441717791411,
      "grad_norm": 57.72730255126953,
      "learning_rate": 4.997869620792501e-05,
      "loss": 1.0674,
      "step": 269
    },
    {
      "epoch": 0.10352760736196319,
      "grad_norm": 76.46302795410156,
      "learning_rate": 4.995739241585002e-05,
      "loss": 1.4453,
      "step": 270
    },
    {
      "epoch": 0.10391104294478527,
      "grad_norm": 274.11651611328125,
      "learning_rate": 4.993608862377504e-05,
      "loss": 2.428,
      "step": 271
    },
    {
      "epoch": 0.10429447852760736,
      "grad_norm": 162.5637664794922,
      "learning_rate": 4.9914784831700046e-05,
      "loss": 1.8976,
      "step": 272
    },
    {
      "epoch": 0.10467791411042945,
      "grad_norm": 60.692230224609375,
      "learning_rate": 4.9893481039625056e-05,
      "loss": 1.1085,
      "step": 273
    },
    {
      "epoch": 0.10506134969325154,
      "grad_norm": 130.2882537841797,
      "learning_rate": 4.9872177247550065e-05,
      "loss": 1.9282,
      "step": 274
    },
    {
      "epoch": 0.10544478527607362,
      "grad_norm": 24.1922664642334,
      "learning_rate": 4.985087345547508e-05,
      "loss": 0.0541,
      "step": 275
    },
    {
      "epoch": 0.1058282208588957,
      "grad_norm": 105.14942169189453,
      "learning_rate": 4.982956966340009e-05,
      "loss": 1.2168,
      "step": 276
    },
    {
      "epoch": 0.10621165644171779,
      "grad_norm": 182.097412109375,
      "learning_rate": 4.98082658713251e-05,
      "loss": 1.5978,
      "step": 277
    },
    {
      "epoch": 0.10659509202453987,
      "grad_norm": 22.384475708007812,
      "learning_rate": 4.978696207925011e-05,
      "loss": 0.647,
      "step": 278
    },
    {
      "epoch": 0.10697852760736197,
      "grad_norm": 15.088644981384277,
      "learning_rate": 4.976565828717512e-05,
      "loss": 0.1837,
      "step": 279
    },
    {
      "epoch": 0.10736196319018405,
      "grad_norm": 0.3369879424571991,
      "learning_rate": 4.9744354495100134e-05,
      "loss": 0.0031,
      "step": 280
    },
    {
      "epoch": 0.10774539877300614,
      "grad_norm": 199.0261688232422,
      "learning_rate": 4.9723050703025144e-05,
      "loss": 0.8648,
      "step": 281
    },
    {
      "epoch": 0.10812883435582822,
      "grad_norm": 3.4141273498535156,
      "learning_rate": 4.970174691095015e-05,
      "loss": 0.0096,
      "step": 282
    },
    {
      "epoch": 0.1085122699386503,
      "grad_norm": 1.4798744916915894,
      "learning_rate": 4.968044311887516e-05,
      "loss": 0.0094,
      "step": 283
    },
    {
      "epoch": 0.10889570552147239,
      "grad_norm": 0.30312761664390564,
      "learning_rate": 4.965913932680017e-05,
      "loss": 0.0032,
      "step": 284
    },
    {
      "epoch": 0.10927914110429447,
      "grad_norm": 0.24992112815380096,
      "learning_rate": 4.963783553472519e-05,
      "loss": 0.0017,
      "step": 285
    },
    {
      "epoch": 0.10966257668711657,
      "grad_norm": 248.25474548339844,
      "learning_rate": 4.96165317426502e-05,
      "loss": 1.9727,
      "step": 286
    },
    {
      "epoch": 0.11004601226993865,
      "grad_norm": 67.61776733398438,
      "learning_rate": 4.9595227950575206e-05,
      "loss": 0.8365,
      "step": 287
    },
    {
      "epoch": 0.11042944785276074,
      "grad_norm": 75.35614013671875,
      "learning_rate": 4.9573924158500216e-05,
      "loss": 0.0323,
      "step": 288
    },
    {
      "epoch": 0.11081288343558282,
      "grad_norm": 0.5646623373031616,
      "learning_rate": 4.955262036642523e-05,
      "loss": 0.0036,
      "step": 289
    },
    {
      "epoch": 0.1111963190184049,
      "grad_norm": 1.4315290451049805,
      "learning_rate": 4.953131657435024e-05,
      "loss": 0.0041,
      "step": 290
    },
    {
      "epoch": 0.11157975460122699,
      "grad_norm": 4.68117618560791,
      "learning_rate": 4.9510012782275244e-05,
      "loss": 0.0249,
      "step": 291
    },
    {
      "epoch": 0.11196319018404909,
      "grad_norm": 0.2679153084754944,
      "learning_rate": 4.948870899020025e-05,
      "loss": 0.0015,
      "step": 292
    },
    {
      "epoch": 0.11234662576687117,
      "grad_norm": 141.1145782470703,
      "learning_rate": 4.946740519812527e-05,
      "loss": 5.8057,
      "step": 293
    },
    {
      "epoch": 0.11273006134969325,
      "grad_norm": 54.23237228393555,
      "learning_rate": 4.944610140605028e-05,
      "loss": 1.1983,
      "step": 294
    },
    {
      "epoch": 0.11311349693251534,
      "grad_norm": 0.5340811610221863,
      "learning_rate": 4.942479761397529e-05,
      "loss": 0.0022,
      "step": 295
    },
    {
      "epoch": 0.11349693251533742,
      "grad_norm": 2.3286099433898926,
      "learning_rate": 4.94034938219003e-05,
      "loss": 0.0026,
      "step": 296
    },
    {
      "epoch": 0.1138803680981595,
      "grad_norm": 68.02838897705078,
      "learning_rate": 4.9382190029825306e-05,
      "loss": 0.7764,
      "step": 297
    },
    {
      "epoch": 0.1142638036809816,
      "grad_norm": 1.8361034393310547,
      "learning_rate": 4.936088623775032e-05,
      "loss": 0.0073,
      "step": 298
    },
    {
      "epoch": 0.11464723926380369,
      "grad_norm": 101.44828796386719,
      "learning_rate": 4.933958244567533e-05,
      "loss": 6.2578,
      "step": 299
    },
    {
      "epoch": 0.11503067484662577,
      "grad_norm": 1.1895840167999268,
      "learning_rate": 4.931827865360034e-05,
      "loss": 0.0022,
      "step": 300
    },
    {
      "epoch": 0.11541411042944785,
      "grad_norm": 157.90872192382812,
      "learning_rate": 4.929697486152535e-05,
      "loss": 1.6612,
      "step": 301
    },
    {
      "epoch": 0.11579754601226994,
      "grad_norm": 51.457881927490234,
      "learning_rate": 4.9275671069450366e-05,
      "loss": 1.0479,
      "step": 302
    },
    {
      "epoch": 0.11618098159509202,
      "grad_norm": 8.253484725952148,
      "learning_rate": 4.9254367277375376e-05,
      "loss": 0.0332,
      "step": 303
    },
    {
      "epoch": 0.1165644171779141,
      "grad_norm": 2.1906232833862305,
      "learning_rate": 4.9233063485300385e-05,
      "loss": 0.0039,
      "step": 304
    },
    {
      "epoch": 0.1169478527607362,
      "grad_norm": 5.369943141937256,
      "learning_rate": 4.9211759693225394e-05,
      "loss": 0.034,
      "step": 305
    },
    {
      "epoch": 0.11733128834355828,
      "grad_norm": 90.70282745361328,
      "learning_rate": 4.9190455901150404e-05,
      "loss": 2.1387,
      "step": 306
    },
    {
      "epoch": 0.11771472392638037,
      "grad_norm": 2.9550065994262695,
      "learning_rate": 4.916915210907542e-05,
      "loss": 0.0074,
      "step": 307
    },
    {
      "epoch": 0.11809815950920245,
      "grad_norm": 2.424849033355713,
      "learning_rate": 4.914784831700043e-05,
      "loss": 0.0033,
      "step": 308
    },
    {
      "epoch": 0.11848159509202454,
      "grad_norm": 2.2970473766326904,
      "learning_rate": 4.912654452492544e-05,
      "loss": 0.0086,
      "step": 309
    },
    {
      "epoch": 0.11886503067484662,
      "grad_norm": 3.276686429977417,
      "learning_rate": 4.910524073285045e-05,
      "loss": 0.0124,
      "step": 310
    },
    {
      "epoch": 0.11924846625766872,
      "grad_norm": 42.38562774658203,
      "learning_rate": 4.908393694077546e-05,
      "loss": 0.6455,
      "step": 311
    },
    {
      "epoch": 0.1196319018404908,
      "grad_norm": 5.350261211395264,
      "learning_rate": 4.906263314870047e-05,
      "loss": 0.0195,
      "step": 312
    },
    {
      "epoch": 0.12001533742331288,
      "grad_norm": 0.3664326071739197,
      "learning_rate": 4.904132935662548e-05,
      "loss": 0.0021,
      "step": 313
    },
    {
      "epoch": 0.12039877300613497,
      "grad_norm": 0.39626434445381165,
      "learning_rate": 4.902002556455049e-05,
      "loss": 0.0037,
      "step": 314
    },
    {
      "epoch": 0.12078220858895705,
      "grad_norm": 2.8121376037597656,
      "learning_rate": 4.89987217724755e-05,
      "loss": 0.0184,
      "step": 315
    },
    {
      "epoch": 0.12116564417177914,
      "grad_norm": 3.150080919265747,
      "learning_rate": 4.897741798040052e-05,
      "loss": 0.006,
      "step": 316
    },
    {
      "epoch": 0.12154907975460123,
      "grad_norm": 0.7951256632804871,
      "learning_rate": 4.8956114188325526e-05,
      "loss": 0.0028,
      "step": 317
    },
    {
      "epoch": 0.12193251533742332,
      "grad_norm": 3.72554349899292,
      "learning_rate": 4.8934810396250536e-05,
      "loss": 0.0047,
      "step": 318
    },
    {
      "epoch": 0.1223159509202454,
      "grad_norm": 1.347676157951355,
      "learning_rate": 4.8913506604175545e-05,
      "loss": 0.0042,
      "step": 319
    },
    {
      "epoch": 0.12269938650306748,
      "grad_norm": 16.904159545898438,
      "learning_rate": 4.8892202812100554e-05,
      "loss": 0.6064,
      "step": 320
    },
    {
      "epoch": 0.12308282208588957,
      "grad_norm": 7.208027362823486,
      "learning_rate": 4.887089902002557e-05,
      "loss": 0.0188,
      "step": 321
    },
    {
      "epoch": 0.12346625766871165,
      "grad_norm": 1.5120185613632202,
      "learning_rate": 4.884959522795058e-05,
      "loss": 0.0061,
      "step": 322
    },
    {
      "epoch": 0.12384969325153375,
      "grad_norm": 1.062395453453064,
      "learning_rate": 4.882829143587559e-05,
      "loss": 0.004,
      "step": 323
    },
    {
      "epoch": 0.12423312883435583,
      "grad_norm": 1.7264481782913208,
      "learning_rate": 4.88069876438006e-05,
      "loss": 0.0023,
      "step": 324
    },
    {
      "epoch": 0.12461656441717792,
      "grad_norm": 1.1984316110610962,
      "learning_rate": 4.878568385172561e-05,
      "loss": 0.0044,
      "step": 325
    },
    {
      "epoch": 0.125,
      "grad_norm": 0.9304490089416504,
      "learning_rate": 4.8764380059650624e-05,
      "loss": 0.0041,
      "step": 326
    },
    {
      "epoch": 0.12538343558282208,
      "grad_norm": 0.3947988748550415,
      "learning_rate": 4.874307626757563e-05,
      "loss": 0.0022,
      "step": 327
    },
    {
      "epoch": 0.12576687116564417,
      "grad_norm": 1.1371535062789917,
      "learning_rate": 4.872177247550064e-05,
      "loss": 0.0065,
      "step": 328
    },
    {
      "epoch": 0.12615030674846625,
      "grad_norm": 0.2662462592124939,
      "learning_rate": 4.870046868342565e-05,
      "loss": 0.0013,
      "step": 329
    },
    {
      "epoch": 0.12653374233128833,
      "grad_norm": 82.56263732910156,
      "learning_rate": 4.867916489135067e-05,
      "loss": 1.4531,
      "step": 330
    },
    {
      "epoch": 0.12691717791411042,
      "grad_norm": 269.9754638671875,
      "learning_rate": 4.865786109927568e-05,
      "loss": 2.0977,
      "step": 331
    },
    {
      "epoch": 0.1273006134969325,
      "grad_norm": 2.6148176193237305,
      "learning_rate": 4.8636557307200686e-05,
      "loss": 0.0096,
      "step": 332
    },
    {
      "epoch": 0.1276840490797546,
      "grad_norm": 220.37306213378906,
      "learning_rate": 4.8615253515125696e-05,
      "loss": 2.211,
      "step": 333
    },
    {
      "epoch": 0.1280674846625767,
      "grad_norm": 0.2989043891429901,
      "learning_rate": 4.8593949723050705e-05,
      "loss": 0.0025,
      "step": 334
    },
    {
      "epoch": 0.12845092024539878,
      "grad_norm": 113.23271942138672,
      "learning_rate": 4.857264593097572e-05,
      "loss": 1.9356,
      "step": 335
    },
    {
      "epoch": 0.12883435582822086,
      "grad_norm": 69.79551696777344,
      "learning_rate": 4.8551342138900724e-05,
      "loss": 1.0248,
      "step": 336
    },
    {
      "epoch": 0.12921779141104295,
      "grad_norm": 5.828656196594238,
      "learning_rate": 4.853003834682573e-05,
      "loss": 0.0179,
      "step": 337
    },
    {
      "epoch": 0.12960122699386503,
      "grad_norm": 8.144012451171875,
      "learning_rate": 4.850873455475074e-05,
      "loss": 0.0617,
      "step": 338
    },
    {
      "epoch": 0.12998466257668712,
      "grad_norm": 22.994976043701172,
      "learning_rate": 4.848743076267576e-05,
      "loss": 0.0673,
      "step": 339
    },
    {
      "epoch": 0.1303680981595092,
      "grad_norm": 40.77123260498047,
      "learning_rate": 4.846612697060077e-05,
      "loss": 0.8135,
      "step": 340
    },
    {
      "epoch": 0.13075153374233128,
      "grad_norm": 0.17868733406066895,
      "learning_rate": 4.844482317852578e-05,
      "loss": 0.0012,
      "step": 341
    },
    {
      "epoch": 0.13113496932515337,
      "grad_norm": 89.55314636230469,
      "learning_rate": 4.8423519386450786e-05,
      "loss": 1.2852,
      "step": 342
    },
    {
      "epoch": 0.13151840490797545,
      "grad_norm": 85.80487823486328,
      "learning_rate": 4.84022155943758e-05,
      "loss": 1.4805,
      "step": 343
    },
    {
      "epoch": 0.13190184049079753,
      "grad_norm": 39.61482238769531,
      "learning_rate": 4.838091180230081e-05,
      "loss": 1.0889,
      "step": 344
    },
    {
      "epoch": 0.13228527607361965,
      "grad_norm": 22.226238250732422,
      "learning_rate": 4.835960801022582e-05,
      "loss": 0.6861,
      "step": 345
    },
    {
      "epoch": 0.13266871165644173,
      "grad_norm": 100.61962127685547,
      "learning_rate": 4.833830421815083e-05,
      "loss": 1.5889,
      "step": 346
    },
    {
      "epoch": 0.1330521472392638,
      "grad_norm": 0.11054955422878265,
      "learning_rate": 4.831700042607584e-05,
      "loss": 0.0011,
      "step": 347
    },
    {
      "epoch": 0.1334355828220859,
      "grad_norm": 10.697381019592285,
      "learning_rate": 4.8295696634000856e-05,
      "loss": 0.0584,
      "step": 348
    },
    {
      "epoch": 0.13381901840490798,
      "grad_norm": 3.288479804992676,
      "learning_rate": 4.8274392841925865e-05,
      "loss": 0.0072,
      "step": 349
    },
    {
      "epoch": 0.13420245398773006,
      "grad_norm": 130.9109649658203,
      "learning_rate": 4.8253089049850874e-05,
      "loss": 1.1807,
      "step": 350
    },
    {
      "epoch": 0.13458588957055215,
      "grad_norm": 0.38042566180229187,
      "learning_rate": 4.8231785257775884e-05,
      "loss": 0.0021,
      "step": 351
    },
    {
      "epoch": 0.13496932515337423,
      "grad_norm": 128.2454376220703,
      "learning_rate": 4.821048146570089e-05,
      "loss": 1.8008,
      "step": 352
    },
    {
      "epoch": 0.13535276073619631,
      "grad_norm": 24.663909912109375,
      "learning_rate": 4.818917767362591e-05,
      "loss": 0.8667,
      "step": 353
    },
    {
      "epoch": 0.1357361963190184,
      "grad_norm": 80.28264617919922,
      "learning_rate": 4.816787388155092e-05,
      "loss": 0.9551,
      "step": 354
    },
    {
      "epoch": 0.13611963190184048,
      "grad_norm": Infinity,
      "learning_rate": 4.816787388155092e-05,
      "loss": 2.5081,
      "step": 355
    },
    {
      "epoch": 0.13650306748466257,
      "grad_norm": 0.5318364500999451,
      "learning_rate": 4.814657008947593e-05,
      "loss": 0.0032,
      "step": 356
    },
    {
      "epoch": 0.13688650306748465,
      "grad_norm": 0.21274921298027039,
      "learning_rate": 4.812526629740094e-05,
      "loss": 0.0022,
      "step": 357
    },
    {
      "epoch": 0.13726993865030676,
      "grad_norm": 2.563629627227783,
      "learning_rate": 4.810396250532595e-05,
      "loss": 0.0094,
      "step": 358
    },
    {
      "epoch": 0.13765337423312884,
      "grad_norm": 0.35965976119041443,
      "learning_rate": 4.808265871325096e-05,
      "loss": 0.0016,
      "step": 359
    },
    {
      "epoch": 0.13803680981595093,
      "grad_norm": 4.237149238586426,
      "learning_rate": 4.806135492117597e-05,
      "loss": 0.017,
      "step": 360
    },
    {
      "epoch": 0.138420245398773,
      "grad_norm": 3.40621280670166,
      "learning_rate": 4.804005112910098e-05,
      "loss": 0.0052,
      "step": 361
    },
    {
      "epoch": 0.1388036809815951,
      "grad_norm": 38.05250930786133,
      "learning_rate": 4.801874733702599e-05,
      "loss": 0.6519,
      "step": 362
    },
    {
      "epoch": 0.13918711656441718,
      "grad_norm": 35.53778076171875,
      "learning_rate": 4.7997443544951006e-05,
      "loss": 0.8091,
      "step": 363
    },
    {
      "epoch": 0.13957055214723926,
      "grad_norm": 93.23106384277344,
      "learning_rate": 4.7976139752876016e-05,
      "loss": 1.4649,
      "step": 364
    },
    {
      "epoch": 0.13995398773006135,
      "grad_norm": 10.391888618469238,
      "learning_rate": 4.7954835960801025e-05,
      "loss": 0.0904,
      "step": 365
    },
    {
      "epoch": 0.14033742331288343,
      "grad_norm": 87.87434387207031,
      "learning_rate": 4.7933532168726034e-05,
      "loss": 0.8394,
      "step": 366
    },
    {
      "epoch": 0.14072085889570551,
      "grad_norm": 186.20010375976562,
      "learning_rate": 4.791222837665105e-05,
      "loss": 3.8704,
      "step": 367
    },
    {
      "epoch": 0.1411042944785276,
      "grad_norm": 9.14842700958252,
      "learning_rate": 4.789092458457606e-05,
      "loss": 0.0408,
      "step": 368
    },
    {
      "epoch": 0.14148773006134968,
      "grad_norm": 17.004417419433594,
      "learning_rate": 4.786962079250107e-05,
      "loss": 0.1443,
      "step": 369
    },
    {
      "epoch": 0.1418711656441718,
      "grad_norm": 65.06951141357422,
      "learning_rate": 4.784831700042608e-05,
      "loss": 1.1856,
      "step": 370
    },
    {
      "epoch": 0.14225460122699388,
      "grad_norm": 139.8549346923828,
      "learning_rate": 4.782701320835109e-05,
      "loss": 6.0264,
      "step": 371
    },
    {
      "epoch": 0.14263803680981596,
      "grad_norm": 0.6574199795722961,
      "learning_rate": 4.7805709416276104e-05,
      "loss": 0.0021,
      "step": 372
    },
    {
      "epoch": 0.14302147239263804,
      "grad_norm": 3.178603410720825,
      "learning_rate": 4.778440562420111e-05,
      "loss": 0.011,
      "step": 373
    },
    {
      "epoch": 0.14340490797546013,
      "grad_norm": 47.10575866699219,
      "learning_rate": 4.776310183212612e-05,
      "loss": 0.6182,
      "step": 374
    },
    {
      "epoch": 0.1437883435582822,
      "grad_norm": 0.21282723546028137,
      "learning_rate": 4.774179804005113e-05,
      "loss": 0.0016,
      "step": 375
    },
    {
      "epoch": 0.1441717791411043,
      "grad_norm": 1.779011845588684,
      "learning_rate": 4.772049424797614e-05,
      "loss": 0.0061,
      "step": 376
    },
    {
      "epoch": 0.14455521472392638,
      "grad_norm": 0.5587969422340393,
      "learning_rate": 4.769919045590116e-05,
      "loss": 0.0016,
      "step": 377
    },
    {
      "epoch": 0.14493865030674846,
      "grad_norm": 171.7791290283203,
      "learning_rate": 4.7677886663826166e-05,
      "loss": 1.4456,
      "step": 378
    },
    {
      "epoch": 0.14532208588957055,
      "grad_norm": 20.514801025390625,
      "learning_rate": 4.7656582871751176e-05,
      "loss": 0.0603,
      "step": 379
    },
    {
      "epoch": 0.14570552147239263,
      "grad_norm": 23.678197860717773,
      "learning_rate": 4.7635279079676185e-05,
      "loss": 0.5738,
      "step": 380
    },
    {
      "epoch": 0.1460889570552147,
      "grad_norm": 4.139899253845215,
      "learning_rate": 4.76139752876012e-05,
      "loss": 0.0305,
      "step": 381
    },
    {
      "epoch": 0.1464723926380368,
      "grad_norm": 1.6574698686599731,
      "learning_rate": 4.7592671495526204e-05,
      "loss": 0.0023,
      "step": 382
    },
    {
      "epoch": 0.1468558282208589,
      "grad_norm": 0.47582539916038513,
      "learning_rate": 4.757136770345121e-05,
      "loss": 0.0028,
      "step": 383
    },
    {
      "epoch": 0.147239263803681,
      "grad_norm": 8.048569679260254,
      "learning_rate": 4.755006391137622e-05,
      "loss": 0.1513,
      "step": 384
    },
    {
      "epoch": 0.14762269938650308,
      "grad_norm": 262.814208984375,
      "learning_rate": 4.752876011930124e-05,
      "loss": 5.9023,
      "step": 385
    },
    {
      "epoch": 0.14800613496932516,
      "grad_norm": 0.6565395593643188,
      "learning_rate": 4.750745632722625e-05,
      "loss": 0.0025,
      "step": 386
    },
    {
      "epoch": 0.14838957055214724,
      "grad_norm": 6.76060152053833,
      "learning_rate": 4.748615253515126e-05,
      "loss": 0.2013,
      "step": 387
    },
    {
      "epoch": 0.14877300613496933,
      "grad_norm": 254.4228515625,
      "learning_rate": 4.7464848743076266e-05,
      "loss": 1.6953,
      "step": 388
    },
    {
      "epoch": 0.1491564417177914,
      "grad_norm": 36.017982482910156,
      "learning_rate": 4.7443544951001276e-05,
      "loss": 0.6831,
      "step": 389
    },
    {
      "epoch": 0.1495398773006135,
      "grad_norm": 1.1480233669281006,
      "learning_rate": 4.742224115892629e-05,
      "loss": 0.0034,
      "step": 390
    },
    {
      "epoch": 0.14992331288343558,
      "grad_norm": 0.12915252149105072,
      "learning_rate": 4.74009373668513e-05,
      "loss": 0.0012,
      "step": 391
    },
    {
      "epoch": 0.15030674846625766,
      "grad_norm": 0.6576260924339294,
      "learning_rate": 4.737963357477631e-05,
      "loss": 0.0039,
      "step": 392
    },
    {
      "epoch": 0.15069018404907975,
      "grad_norm": 0.5759190917015076,
      "learning_rate": 4.735832978270132e-05,
      "loss": 0.0029,
      "step": 393
    },
    {
      "epoch": 0.15107361963190183,
      "grad_norm": 9.682991981506348,
      "learning_rate": 4.7337025990626336e-05,
      "loss": 0.0185,
      "step": 394
    },
    {
      "epoch": 0.15145705521472394,
      "grad_norm": 146.7510986328125,
      "learning_rate": 4.7315722198551345e-05,
      "loss": 6.0566,
      "step": 395
    },
    {
      "epoch": 0.15184049079754602,
      "grad_norm": 105.06553649902344,
      "learning_rate": 4.7294418406476354e-05,
      "loss": 1.7637,
      "step": 396
    },
    {
      "epoch": 0.1522239263803681,
      "grad_norm": 0.605169415473938,
      "learning_rate": 4.7273114614401364e-05,
      "loss": 0.0023,
      "step": 397
    },
    {
      "epoch": 0.1526073619631902,
      "grad_norm": 7.71505880355835,
      "learning_rate": 4.725181082232637e-05,
      "loss": 0.0183,
      "step": 398
    },
    {
      "epoch": 0.15299079754601227,
      "grad_norm": 9.905267715454102,
      "learning_rate": 4.723050703025139e-05,
      "loss": 0.015,
      "step": 399
    },
    {
      "epoch": 0.15337423312883436,
      "grad_norm": 142.18589782714844,
      "learning_rate": 4.72092032381764e-05,
      "loss": 1.415,
      "step": 400
    },
    {
      "epoch": 0.15375766871165644,
      "grad_norm": 1.3648536205291748,
      "learning_rate": 4.718789944610141e-05,
      "loss": 0.0038,
      "step": 401
    },
    {
      "epoch": 0.15414110429447853,
      "grad_norm": 126.9765396118164,
      "learning_rate": 4.716659565402642e-05,
      "loss": 1.7979,
      "step": 402
    },
    {
      "epoch": 0.1545245398773006,
      "grad_norm": 45.10152053833008,
      "learning_rate": 4.7145291861951426e-05,
      "loss": 0.7524,
      "step": 403
    },
    {
      "epoch": 0.1549079754601227,
      "grad_norm": 0.40764355659484863,
      "learning_rate": 4.712398806987644e-05,
      "loss": 0.0022,
      "step": 404
    },
    {
      "epoch": 0.15529141104294478,
      "grad_norm": 2.1609463691711426,
      "learning_rate": 4.710268427780145e-05,
      "loss": 0.0041,
      "step": 405
    },
    {
      "epoch": 0.15567484662576686,
      "grad_norm": 4.3262434005737305,
      "learning_rate": 4.708138048572646e-05,
      "loss": 0.0314,
      "step": 406
    },
    {
      "epoch": 0.15605828220858894,
      "grad_norm": 0.5598259568214417,
      "learning_rate": 4.706007669365147e-05,
      "loss": 0.0028,
      "step": 407
    },
    {
      "epoch": 0.15644171779141106,
      "grad_norm": 1.6508883237838745,
      "learning_rate": 4.7038772901576486e-05,
      "loss": 0.0066,
      "step": 408
    },
    {
      "epoch": 0.15682515337423314,
      "grad_norm": 11.46448040008545,
      "learning_rate": 4.7017469109501496e-05,
      "loss": 0.0401,
      "step": 409
    },
    {
      "epoch": 0.15720858895705522,
      "grad_norm": 165.40843200683594,
      "learning_rate": 4.6996165317426505e-05,
      "loss": 3.3642,
      "step": 410
    },
    {
      "epoch": 0.1575920245398773,
      "grad_norm": 78.23701477050781,
      "learning_rate": 4.6974861525351514e-05,
      "loss": 1.3516,
      "step": 411
    },
    {
      "epoch": 0.1579754601226994,
      "grad_norm": 0.5470488667488098,
      "learning_rate": 4.6953557733276524e-05,
      "loss": 0.0019,
      "step": 412
    },
    {
      "epoch": 0.15835889570552147,
      "grad_norm": 298.1001281738281,
      "learning_rate": 4.693225394120154e-05,
      "loss": 1.5518,
      "step": 413
    },
    {
      "epoch": 0.15874233128834356,
      "grad_norm": 0.11850805580615997,
      "learning_rate": 4.691095014912655e-05,
      "loss": 0.0018,
      "step": 414
    },
    {
      "epoch": 0.15912576687116564,
      "grad_norm": 133.16050720214844,
      "learning_rate": 4.688964635705156e-05,
      "loss": 1.5381,
      "step": 415
    },
    {
      "epoch": 0.15950920245398773,
      "grad_norm": 0.5266661643981934,
      "learning_rate": 4.686834256497657e-05,
      "loss": 0.0033,
      "step": 416
    },
    {
      "epoch": 0.1598926380368098,
      "grad_norm": 28.62251091003418,
      "learning_rate": 4.684703877290158e-05,
      "loss": 0.583,
      "step": 417
    },
    {
      "epoch": 0.1602760736196319,
      "grad_norm": 3.70107364654541,
      "learning_rate": 4.682573498082659e-05,
      "loss": 0.0058,
      "step": 418
    },
    {
      "epoch": 0.16065950920245398,
      "grad_norm": 2.993849754333496,
      "learning_rate": 4.68044311887516e-05,
      "loss": 0.017,
      "step": 419
    },
    {
      "epoch": 0.16104294478527606,
      "grad_norm": 95.91368103027344,
      "learning_rate": 4.678312739667661e-05,
      "loss": 1.3858,
      "step": 420
    },
    {
      "epoch": 0.16142638036809817,
      "grad_norm": 379.5052185058594,
      "learning_rate": 4.676182360460162e-05,
      "loss": 6.4629,
      "step": 421
    },
    {
      "epoch": 0.16180981595092025,
      "grad_norm": 12.950861930847168,
      "learning_rate": 4.674051981252664e-05,
      "loss": 0.0474,
      "step": 422
    },
    {
      "epoch": 0.16219325153374234,
      "grad_norm": 161.7283935546875,
      "learning_rate": 4.6719216020451646e-05,
      "loss": 2.4474,
      "step": 423
    },
    {
      "epoch": 0.16257668711656442,
      "grad_norm": 0.9702209830284119,
      "learning_rate": 4.6697912228376656e-05,
      "loss": 0.0026,
      "step": 424
    },
    {
      "epoch": 0.1629601226993865,
      "grad_norm": 0.5118975043296814,
      "learning_rate": 4.6676608436301665e-05,
      "loss": 0.002,
      "step": 425
    },
    {
      "epoch": 0.1633435582822086,
      "grad_norm": 24.19846534729004,
      "learning_rate": 4.6655304644226674e-05,
      "loss": 0.6104,
      "step": 426
    },
    {
      "epoch": 0.16372699386503067,
      "grad_norm": 74.22824096679688,
      "learning_rate": 4.6634000852151684e-05,
      "loss": 0.7046,
      "step": 427
    },
    {
      "epoch": 0.16411042944785276,
      "grad_norm": 0.7992326617240906,
      "learning_rate": 4.661269706007669e-05,
      "loss": 0.0033,
      "step": 428
    },
    {
      "epoch": 0.16449386503067484,
      "grad_norm": 0.5454573631286621,
      "learning_rate": 4.65913932680017e-05,
      "loss": 0.0028,
      "step": 429
    },
    {
      "epoch": 0.16487730061349692,
      "grad_norm": 0.4567127525806427,
      "learning_rate": 4.657008947592671e-05,
      "loss": 0.0021,
      "step": 430
    },
    {
      "epoch": 0.165260736196319,
      "grad_norm": 149.2803955078125,
      "learning_rate": 4.654878568385173e-05,
      "loss": 1.0586,
      "step": 431
    },
    {
      "epoch": 0.1656441717791411,
      "grad_norm": 134.65061950683594,
      "learning_rate": 4.652748189177674e-05,
      "loss": 1.1211,
      "step": 432
    },
    {
      "epoch": 0.1660276073619632,
      "grad_norm": 138.16680908203125,
      "learning_rate": 4.6506178099701746e-05,
      "loss": 5.3633,
      "step": 433
    },
    {
      "epoch": 0.1664110429447853,
      "grad_norm": 1.2007067203521729,
      "learning_rate": 4.6484874307626756e-05,
      "loss": 0.0057,
      "step": 434
    },
    {
      "epoch": 0.16679447852760737,
      "grad_norm": 0.8257370591163635,
      "learning_rate": 4.646357051555177e-05,
      "loss": 0.0028,
      "step": 435
    },
    {
      "epoch": 0.16717791411042945,
      "grad_norm": 375.18988037109375,
      "learning_rate": 4.644226672347678e-05,
      "loss": 5.9131,
      "step": 436
    },
    {
      "epoch": 0.16756134969325154,
      "grad_norm": 17.690185546875,
      "learning_rate": 4.642096293140179e-05,
      "loss": 0.0275,
      "step": 437
    },
    {
      "epoch": 0.16794478527607362,
      "grad_norm": 134.84230041503906,
      "learning_rate": 4.63996591393268e-05,
      "loss": 1.8117,
      "step": 438
    },
    {
      "epoch": 0.1683282208588957,
      "grad_norm": 178.44369506835938,
      "learning_rate": 4.637835534725181e-05,
      "loss": 5.9438,
      "step": 439
    },
    {
      "epoch": 0.1687116564417178,
      "grad_norm": 9.95478630065918,
      "learning_rate": 4.6357051555176825e-05,
      "loss": 0.0455,
      "step": 440
    },
    {
      "epoch": 0.16909509202453987,
      "grad_norm": 65.96708679199219,
      "learning_rate": 4.6335747763101834e-05,
      "loss": 1.1426,
      "step": 441
    },
    {
      "epoch": 0.16947852760736196,
      "grad_norm": 36.904945373535156,
      "learning_rate": 4.6314443971026844e-05,
      "loss": 0.6465,
      "step": 442
    },
    {
      "epoch": 0.16986196319018404,
      "grad_norm": 3.5175623893737793,
      "learning_rate": 4.629314017895185e-05,
      "loss": 0.0048,
      "step": 443
    },
    {
      "epoch": 0.17024539877300612,
      "grad_norm": 1.2675668001174927,
      "learning_rate": 4.627183638687686e-05,
      "loss": 0.0065,
      "step": 444
    },
    {
      "epoch": 0.1706288343558282,
      "grad_norm": 95.76133728027344,
      "learning_rate": 4.625053259480188e-05,
      "loss": 5.6777,
      "step": 445
    },
    {
      "epoch": 0.17101226993865032,
      "grad_norm": 1.1783291101455688,
      "learning_rate": 4.622922880272689e-05,
      "loss": 0.0027,
      "step": 446
    },
    {
      "epoch": 0.1713957055214724,
      "grad_norm": 47.477622985839844,
      "learning_rate": 4.62079250106519e-05,
      "loss": 0.9526,
      "step": 447
    },
    {
      "epoch": 0.17177914110429449,
      "grad_norm": 20.091808319091797,
      "learning_rate": 4.6186621218576906e-05,
      "loss": 0.5361,
      "step": 448
    },
    {
      "epoch": 0.17216257668711657,
      "grad_norm": 0.798278272151947,
      "learning_rate": 4.616531742650192e-05,
      "loss": 0.0032,
      "step": 449
    },
    {
      "epoch": 0.17254601226993865,
      "grad_norm": 7.0091776847839355,
      "learning_rate": 4.614401363442693e-05,
      "loss": 0.0322,
      "step": 450
    },
    {
      "epoch": 0.17292944785276074,
      "grad_norm": 11.215709686279297,
      "learning_rate": 4.612270984235194e-05,
      "loss": 0.5532,
      "step": 451
    },
    {
      "epoch": 0.17331288343558282,
      "grad_norm": 11.881379127502441,
      "learning_rate": 4.610140605027695e-05,
      "loss": 0.1141,
      "step": 452
    },
    {
      "epoch": 0.1736963190184049,
      "grad_norm": 3.7086267471313477,
      "learning_rate": 4.608010225820196e-05,
      "loss": 0.0232,
      "step": 453
    },
    {
      "epoch": 0.174079754601227,
      "grad_norm": 116.72730255126953,
      "learning_rate": 4.6058798466126976e-05,
      "loss": 5.5176,
      "step": 454
    },
    {
      "epoch": 0.17446319018404907,
      "grad_norm": 150.83160400390625,
      "learning_rate": 4.6037494674051985e-05,
      "loss": 6.5215,
      "step": 455
    },
    {
      "epoch": 0.17484662576687116,
      "grad_norm": 0.8595237135887146,
      "learning_rate": 4.6016190881976994e-05,
      "loss": 0.0028,
      "step": 456
    },
    {
      "epoch": 0.17523006134969324,
      "grad_norm": 0.31871017813682556,
      "learning_rate": 4.5994887089902004e-05,
      "loss": 0.0015,
      "step": 457
    },
    {
      "epoch": 0.17561349693251535,
      "grad_norm": 0.3064028322696686,
      "learning_rate": 4.597358329782702e-05,
      "loss": 0.0022,
      "step": 458
    },
    {
      "epoch": 0.17599693251533743,
      "grad_norm": 7.268013954162598,
      "learning_rate": 4.595227950575203e-05,
      "loss": 0.0172,
      "step": 459
    },
    {
      "epoch": 0.17638036809815952,
      "grad_norm": 141.68385314941406,
      "learning_rate": 4.593097571367704e-05,
      "loss": 1.1113,
      "step": 460
    },
    {
      "epoch": 0.1767638036809816,
      "grad_norm": 179.11373901367188,
      "learning_rate": 4.590967192160205e-05,
      "loss": 1.4815,
      "step": 461
    },
    {
      "epoch": 0.17714723926380369,
      "grad_norm": 88.66524505615234,
      "learning_rate": 4.588836812952706e-05,
      "loss": 0.9683,
      "step": 462
    },
    {
      "epoch": 0.17753067484662577,
      "grad_norm": 61.23609161376953,
      "learning_rate": 4.586706433745207e-05,
      "loss": 0.6357,
      "step": 463
    },
    {
      "epoch": 0.17791411042944785,
      "grad_norm": 4.37770414352417,
      "learning_rate": 4.584576054537708e-05,
      "loss": 0.031,
      "step": 464
    },
    {
      "epoch": 0.17829754601226994,
      "grad_norm": 3.8708748817443848,
      "learning_rate": 4.582445675330209e-05,
      "loss": 0.0144,
      "step": 465
    },
    {
      "epoch": 0.17868098159509202,
      "grad_norm": 0.21280257403850555,
      "learning_rate": 4.58031529612271e-05,
      "loss": 0.0017,
      "step": 466
    },
    {
      "epoch": 0.1790644171779141,
      "grad_norm": 35.33317947387695,
      "learning_rate": 4.578184916915211e-05,
      "loss": 0.6538,
      "step": 467
    },
    {
      "epoch": 0.1794478527607362,
      "grad_norm": 47.658287048339844,
      "learning_rate": 4.5760545377077126e-05,
      "loss": 0.9697,
      "step": 468
    },
    {
      "epoch": 0.17983128834355827,
      "grad_norm": 24.110111236572266,
      "learning_rate": 4.5739241585002136e-05,
      "loss": 0.6646,
      "step": 469
    },
    {
      "epoch": 0.18021472392638035,
      "grad_norm": 0.593888521194458,
      "learning_rate": 4.5717937792927145e-05,
      "loss": 0.002,
      "step": 470
    },
    {
      "epoch": 0.18059815950920247,
      "grad_norm": 1.9042633771896362,
      "learning_rate": 4.5696634000852154e-05,
      "loss": 0.0037,
      "step": 471
    },
    {
      "epoch": 0.18098159509202455,
      "grad_norm": 3.080111026763916,
      "learning_rate": 4.567533020877717e-05,
      "loss": 0.0084,
      "step": 472
    },
    {
      "epoch": 0.18136503067484663,
      "grad_norm": 1.5383565425872803,
      "learning_rate": 4.565402641670217e-05,
      "loss": 0.0064,
      "step": 473
    },
    {
      "epoch": 0.18174846625766872,
      "grad_norm": 0.2783668637275696,
      "learning_rate": 4.563272262462718e-05,
      "loss": 0.0025,
      "step": 474
    },
    {
      "epoch": 0.1821319018404908,
      "grad_norm": 97.4244613647461,
      "learning_rate": 4.561141883255219e-05,
      "loss": 6.6191,
      "step": 475
    },
    {
      "epoch": 0.18251533742331288,
      "grad_norm": 55.277164459228516,
      "learning_rate": 4.559011504047721e-05,
      "loss": 1.1631,
      "step": 476
    },
    {
      "epoch": 0.18289877300613497,
      "grad_norm": 6.295434951782227,
      "learning_rate": 4.556881124840222e-05,
      "loss": 0.0353,
      "step": 477
    },
    {
      "epoch": 0.18328220858895705,
      "grad_norm": 19.35044288635254,
      "learning_rate": 4.5547507456327226e-05,
      "loss": 0.7227,
      "step": 478
    },
    {
      "epoch": 0.18366564417177914,
      "grad_norm": 1.969385027885437,
      "learning_rate": 4.5526203664252236e-05,
      "loss": 0.0086,
      "step": 479
    },
    {
      "epoch": 0.18404907975460122,
      "grad_norm": 6.183859348297119,
      "learning_rate": 4.5504899872177245e-05,
      "loss": 0.0301,
      "step": 480
    },
    {
      "epoch": 0.1844325153374233,
      "grad_norm": 49.332706451416016,
      "learning_rate": 4.548359608010226e-05,
      "loss": 1.1514,
      "step": 481
    },
    {
      "epoch": 0.1848159509202454,
      "grad_norm": 0.09144336730241776,
      "learning_rate": 4.546229228802727e-05,
      "loss": 0.001,
      "step": 482
    },
    {
      "epoch": 0.1851993865030675,
      "grad_norm": 227.00379943847656,
      "learning_rate": 4.544098849595228e-05,
      "loss": 6.0859,
      "step": 483
    },
    {
      "epoch": 0.18558282208588958,
      "grad_norm": 13.271467208862305,
      "learning_rate": 4.541968470387729e-05,
      "loss": 0.0679,
      "step": 484
    },
    {
      "epoch": 0.18596625766871167,
      "grad_norm": 6.430564880371094,
      "learning_rate": 4.5398380911802305e-05,
      "loss": 0.0404,
      "step": 485
    },
    {
      "epoch": 0.18634969325153375,
      "grad_norm": 97.63346862792969,
      "learning_rate": 4.5377077119727314e-05,
      "loss": 0.9849,
      "step": 486
    },
    {
      "epoch": 0.18673312883435583,
      "grad_norm": 12.30589485168457,
      "learning_rate": 4.5355773327652324e-05,
      "loss": 0.0462,
      "step": 487
    },
    {
      "epoch": 0.18711656441717792,
      "grad_norm": 44.606597900390625,
      "learning_rate": 4.533446953557733e-05,
      "loss": 0.6133,
      "step": 488
    },
    {
      "epoch": 0.1875,
      "grad_norm": 0.7023065686225891,
      "learning_rate": 4.531316574350234e-05,
      "loss": 0.003,
      "step": 489
    },
    {
      "epoch": 0.18788343558282208,
      "grad_norm": 407.8355407714844,
      "learning_rate": 4.529186195142736e-05,
      "loss": 2.3166,
      "step": 490
    },
    {
      "epoch": 0.18826687116564417,
      "grad_norm": 100.24420166015625,
      "learning_rate": 4.527055815935237e-05,
      "loss": 1.46,
      "step": 491
    },
    {
      "epoch": 0.18865030674846625,
      "grad_norm": 1.02791166305542,
      "learning_rate": 4.524925436727738e-05,
      "loss": 0.0033,
      "step": 492
    },
    {
      "epoch": 0.18903374233128833,
      "grad_norm": 151.98976135253906,
      "learning_rate": 4.5227950575202386e-05,
      "loss": 2.4806,
      "step": 493
    },
    {
      "epoch": 0.18941717791411042,
      "grad_norm": 0.4681469202041626,
      "learning_rate": 4.5206646783127396e-05,
      "loss": 0.0029,
      "step": 494
    },
    {
      "epoch": 0.1898006134969325,
      "grad_norm": 92.45951080322266,
      "learning_rate": 4.518534299105241e-05,
      "loss": 1.5801,
      "step": 495
    },
    {
      "epoch": 0.1901840490797546,
      "grad_norm": 44.523433685302734,
      "learning_rate": 4.516403919897742e-05,
      "loss": 1.1572,
      "step": 496
    },
    {
      "epoch": 0.1905674846625767,
      "grad_norm": 11.493820190429688,
      "learning_rate": 4.514273540690243e-05,
      "loss": 0.0422,
      "step": 497
    },
    {
      "epoch": 0.19095092024539878,
      "grad_norm": 202.0272674560547,
      "learning_rate": 4.512143161482744e-05,
      "loss": 2.0394,
      "step": 498
    },
    {
      "epoch": 0.19133435582822086,
      "grad_norm": 13.105687141418457,
      "learning_rate": 4.5100127822752456e-05,
      "loss": 0.0464,
      "step": 499
    },
    {
      "epoch": 0.19171779141104295,
      "grad_norm": 4.247876167297363,
      "learning_rate": 4.5078824030677465e-05,
      "loss": 0.0123,
      "step": 500
    },
    {
      "epoch": 0.19210122699386503,
      "grad_norm": 228.3590850830078,
      "learning_rate": 4.5057520238602474e-05,
      "loss": 1.4082,
      "step": 501
    },
    {
      "epoch": 0.19248466257668712,
      "grad_norm": 3.032365322113037,
      "learning_rate": 4.5036216446527484e-05,
      "loss": 0.0085,
      "step": 502
    },
    {
      "epoch": 0.1928680981595092,
      "grad_norm": 0.6577479243278503,
      "learning_rate": 4.501491265445249e-05,
      "loss": 0.0024,
      "step": 503
    },
    {
      "epoch": 0.19325153374233128,
      "grad_norm": 89.23050689697266,
      "learning_rate": 4.499360886237751e-05,
      "loss": 5.9355,
      "step": 504
    },
    {
      "epoch": 0.19363496932515337,
      "grad_norm": 274.9822692871094,
      "learning_rate": 4.497230507030252e-05,
      "loss": 2.0099,
      "step": 505
    },
    {
      "epoch": 0.19401840490797545,
      "grad_norm": 2.5567502975463867,
      "learning_rate": 4.495100127822753e-05,
      "loss": 0.0053,
      "step": 506
    },
    {
      "epoch": 0.19440184049079753,
      "grad_norm": 45.34825134277344,
      "learning_rate": 4.492969748615254e-05,
      "loss": 0.9722,
      "step": 507
    },
    {
      "epoch": 0.19478527607361965,
      "grad_norm": 43.673038482666016,
      "learning_rate": 4.4908393694077546e-05,
      "loss": 0.6895,
      "step": 508
    },
    {
      "epoch": 0.19516871165644173,
      "grad_norm": 179.43748474121094,
      "learning_rate": 4.488708990200256e-05,
      "loss": 0.6533,
      "step": 509
    },
    {
      "epoch": 0.1955521472392638,
      "grad_norm": 2.264289140701294,
      "learning_rate": 4.486578610992757e-05,
      "loss": 0.0089,
      "step": 510
    },
    {
      "epoch": 0.1959355828220859,
      "grad_norm": 2.1404731273651123,
      "learning_rate": 4.484448231785258e-05,
      "loss": 0.0134,
      "step": 511
    },
    {
      "epoch": 0.19631901840490798,
      "grad_norm": 217.19174194335938,
      "learning_rate": 4.482317852577759e-05,
      "loss": 3.2687,
      "step": 512
    },
    {
      "epoch": 0.19670245398773006,
      "grad_norm": 63.18962097167969,
      "learning_rate": 4.4801874733702606e-05,
      "loss": 1.0537,
      "step": 513
    },
    {
      "epoch": 0.19708588957055215,
      "grad_norm": 53.48243713378906,
      "learning_rate": 4.4780570941627616e-05,
      "loss": 0.7662,
      "step": 514
    },
    {
      "epoch": 0.19746932515337423,
      "grad_norm": 3.4618992805480957,
      "learning_rate": 4.4759267149552625e-05,
      "loss": 0.0125,
      "step": 515
    },
    {
      "epoch": 0.19785276073619631,
      "grad_norm": 4.8398756980896,
      "learning_rate": 4.4737963357477634e-05,
      "loss": 0.0358,
      "step": 516
    },
    {
      "epoch": 0.1982361963190184,
      "grad_norm": 107.578857421875,
      "learning_rate": 4.4716659565402644e-05,
      "loss": 0.959,
      "step": 517
    },
    {
      "epoch": 0.19861963190184048,
      "grad_norm": 1.9601085186004639,
      "learning_rate": 4.469535577332765e-05,
      "loss": 0.0099,
      "step": 518
    },
    {
      "epoch": 0.19900306748466257,
      "grad_norm": 17.420923233032227,
      "learning_rate": 4.467405198125266e-05,
      "loss": 0.1004,
      "step": 519
    },
    {
      "epoch": 0.19938650306748465,
      "grad_norm": 42.05288314819336,
      "learning_rate": 4.465274818917767e-05,
      "loss": 0.7842,
      "step": 520
    },
    {
      "epoch": 0.19976993865030676,
      "grad_norm": 4.815923690795898,
      "learning_rate": 4.463144439710268e-05,
      "loss": 0.0127,
      "step": 521
    },
    {
      "epoch": 0.20015337423312884,
      "grad_norm": 130.26187133789062,
      "learning_rate": 4.46101406050277e-05,
      "loss": 2.5907,
      "step": 522
    },
    {
      "epoch": 0.20053680981595093,
      "grad_norm": 133.83578491210938,
      "learning_rate": 4.4588836812952706e-05,
      "loss": 6.6577,
      "step": 523
    },
    {
      "epoch": 0.200920245398773,
      "grad_norm": 1.783935308456421,
      "learning_rate": 4.4567533020877716e-05,
      "loss": 0.0084,
      "step": 524
    },
    {
      "epoch": 0.2013036809815951,
      "grad_norm": 150.36233520507812,
      "learning_rate": 4.4546229228802725e-05,
      "loss": 1.8362,
      "step": 525
    },
    {
      "epoch": 0.20168711656441718,
      "grad_norm": 6.6197075843811035,
      "learning_rate": 4.452492543672774e-05,
      "loss": 0.0135,
      "step": 526
    },
    {
      "epoch": 0.20207055214723926,
      "grad_norm": 184.958740234375,
      "learning_rate": 4.450362164465275e-05,
      "loss": 1.9093,
      "step": 527
    },
    {
      "epoch": 0.20245398773006135,
      "grad_norm": 76.64862060546875,
      "learning_rate": 4.448231785257776e-05,
      "loss": 1.2227,
      "step": 528
    },
    {
      "epoch": 0.20283742331288343,
      "grad_norm": 0.4043022394180298,
      "learning_rate": 4.446101406050277e-05,
      "loss": 0.0019,
      "step": 529
    },
    {
      "epoch": 0.20322085889570551,
      "grad_norm": 26.354055404663086,
      "learning_rate": 4.443971026842778e-05,
      "loss": 1.2315,
      "step": 530
    },
    {
      "epoch": 0.2036042944785276,
      "grad_norm": 85.78749084472656,
      "learning_rate": 4.4418406476352794e-05,
      "loss": 0.9126,
      "step": 531
    },
    {
      "epoch": 0.20398773006134968,
      "grad_norm": 0.17065221071243286,
      "learning_rate": 4.4397102684277804e-05,
      "loss": 0.0014,
      "step": 532
    },
    {
      "epoch": 0.2043711656441718,
      "grad_norm": 59.73079299926758,
      "learning_rate": 4.437579889220281e-05,
      "loss": 1.2969,
      "step": 533
    },
    {
      "epoch": 0.20475460122699388,
      "grad_norm": 3.527113437652588,
      "learning_rate": 4.435449510012782e-05,
      "loss": 0.0127,
      "step": 534
    },
    {
      "epoch": 0.20513803680981596,
      "grad_norm": 2.873439073562622,
      "learning_rate": 4.433319130805284e-05,
      "loss": 0.012,
      "step": 535
    },
    {
      "epoch": 0.20552147239263804,
      "grad_norm": 11.59239673614502,
      "learning_rate": 4.431188751597785e-05,
      "loss": 0.1285,
      "step": 536
    },
    {
      "epoch": 0.20590490797546013,
      "grad_norm": 0.4274378716945648,
      "learning_rate": 4.429058372390286e-05,
      "loss": 0.0013,
      "step": 537
    },
    {
      "epoch": 0.2062883435582822,
      "grad_norm": 19.355199813842773,
      "learning_rate": 4.4269279931827866e-05,
      "loss": 0.0459,
      "step": 538
    },
    {
      "epoch": 0.2066717791411043,
      "grad_norm": 2.7791860103607178,
      "learning_rate": 4.4247976139752876e-05,
      "loss": 0.0166,
      "step": 539
    },
    {
      "epoch": 0.20705521472392638,
      "grad_norm": 11.108470916748047,
      "learning_rate": 4.422667234767789e-05,
      "loss": 0.1113,
      "step": 540
    },
    {
      "epoch": 0.20743865030674846,
      "grad_norm": 29.45231819152832,
      "learning_rate": 4.42053685556029e-05,
      "loss": 0.7363,
      "step": 541
    },
    {
      "epoch": 0.20782208588957055,
      "grad_norm": 136.5975799560547,
      "learning_rate": 4.418406476352791e-05,
      "loss": 6.1973,
      "step": 542
    },
    {
      "epoch": 0.20820552147239263,
      "grad_norm": 3.0369179248809814,
      "learning_rate": 4.416276097145292e-05,
      "loss": 0.0062,
      "step": 543
    },
    {
      "epoch": 0.2085889570552147,
      "grad_norm": 29.516555786132812,
      "learning_rate": 4.414145717937793e-05,
      "loss": 0.8848,
      "step": 544
    },
    {
      "epoch": 0.2089723926380368,
      "grad_norm": 5.653842926025391,
      "learning_rate": 4.4120153387302945e-05,
      "loss": 0.0231,
      "step": 545
    },
    {
      "epoch": 0.2093558282208589,
      "grad_norm": 111.30487823486328,
      "learning_rate": 4.4098849595227954e-05,
      "loss": 2.0919,
      "step": 546
    },
    {
      "epoch": 0.209739263803681,
      "grad_norm": 137.13394165039062,
      "learning_rate": 4.4077545803152964e-05,
      "loss": 1.6612,
      "step": 547
    },
    {
      "epoch": 0.21012269938650308,
      "grad_norm": 0.26130568981170654,
      "learning_rate": 4.405624201107797e-05,
      "loss": 0.0022,
      "step": 548
    },
    {
      "epoch": 0.21050613496932516,
      "grad_norm": 117.121826171875,
      "learning_rate": 4.403493821900299e-05,
      "loss": 1.127,
      "step": 549
    },
    {
      "epoch": 0.21088957055214724,
      "grad_norm": 0.17888779938220978,
      "learning_rate": 4.4013634426928e-05,
      "loss": 0.0019,
      "step": 550
    },
    {
      "epoch": 0.21127300613496933,
      "grad_norm": 2.819636821746826,
      "learning_rate": 4.399233063485301e-05,
      "loss": 0.0068,
      "step": 551
    },
    {
      "epoch": 0.2116564417177914,
      "grad_norm": 11.347158432006836,
      "learning_rate": 4.397102684277802e-05,
      "loss": 0.0568,
      "step": 552
    },
    {
      "epoch": 0.2120398773006135,
      "grad_norm": 1.9594674110412598,
      "learning_rate": 4.3949723050703026e-05,
      "loss": 0.0094,
      "step": 553
    },
    {
      "epoch": 0.21242331288343558,
      "grad_norm": 0.2811185419559479,
      "learning_rate": 4.392841925862804e-05,
      "loss": 0.0017,
      "step": 554
    },
    {
      "epoch": 0.21280674846625766,
      "grad_norm": 1.277053713798523,
      "learning_rate": 4.390711546655305e-05,
      "loss": 0.0055,
      "step": 555
    },
    {
      "epoch": 0.21319018404907975,
      "grad_norm": 197.99415588378906,
      "learning_rate": 4.388581167447806e-05,
      "loss": 1.8321,
      "step": 556
    },
    {
      "epoch": 0.21357361963190183,
      "grad_norm": 128.40423583984375,
      "learning_rate": 4.386450788240307e-05,
      "loss": 1.7901,
      "step": 557
    },
    {
      "epoch": 0.21395705521472394,
      "grad_norm": 1.1741713285446167,
      "learning_rate": 4.384320409032808e-05,
      "loss": 0.0036,
      "step": 558
    },
    {
      "epoch": 0.21434049079754602,
      "grad_norm": 0.503452718257904,
      "learning_rate": 4.3821900298253096e-05,
      "loss": 0.0032,
      "step": 559
    },
    {
      "epoch": 0.2147239263803681,
      "grad_norm": 48.687339782714844,
      "learning_rate": 4.3800596506178105e-05,
      "loss": 0.6504,
      "step": 560
    },
    {
      "epoch": 0.2151073619631902,
      "grad_norm": 9.530661582946777,
      "learning_rate": 4.3779292714103114e-05,
      "loss": 0.082,
      "step": 561
    },
    {
      "epoch": 0.21549079754601227,
      "grad_norm": 0.6268191933631897,
      "learning_rate": 4.3757988922028124e-05,
      "loss": 0.0038,
      "step": 562
    },
    {
      "epoch": 0.21587423312883436,
      "grad_norm": 11.719134330749512,
      "learning_rate": 4.373668512995313e-05,
      "loss": 0.1386,
      "step": 563
    },
    {
      "epoch": 0.21625766871165644,
      "grad_norm": 1.8017444610595703,
      "learning_rate": 4.371538133787814e-05,
      "loss": 0.0058,
      "step": 564
    },
    {
      "epoch": 0.21664110429447853,
      "grad_norm": 59.35349655151367,
      "learning_rate": 4.369407754580315e-05,
      "loss": 1.1719,
      "step": 565
    },
    {
      "epoch": 0.2170245398773006,
      "grad_norm": 13.164896965026855,
      "learning_rate": 4.367277375372816e-05,
      "loss": 0.5293,
      "step": 566
    },
    {
      "epoch": 0.2174079754601227,
      "grad_norm": 97.70745086669922,
      "learning_rate": 4.365146996165318e-05,
      "loss": 2.3205,
      "step": 567
    },
    {
      "epoch": 0.21779141104294478,
      "grad_norm": 47.05236053466797,
      "learning_rate": 4.3630166169578186e-05,
      "loss": 0.6914,
      "step": 568
    },
    {
      "epoch": 0.21817484662576686,
      "grad_norm": 69.89529418945312,
      "learning_rate": 4.3608862377503196e-05,
      "loss": 1.9659,
      "step": 569
    },
    {
      "epoch": 0.21855828220858894,
      "grad_norm": 139.23280334472656,
      "learning_rate": 4.3587558585428205e-05,
      "loss": 1.8878,
      "step": 570
    },
    {
      "epoch": 0.21894171779141106,
      "grad_norm": 1.1655089855194092,
      "learning_rate": 4.3566254793353214e-05,
      "loss": 0.0067,
      "step": 571
    },
    {
      "epoch": 0.21932515337423314,
      "grad_norm": 100.34245300292969,
      "learning_rate": 4.354495100127823e-05,
      "loss": 0.8511,
      "step": 572
    },
    {
      "epoch": 0.21970858895705522,
      "grad_norm": 87.71229553222656,
      "learning_rate": 4.352364720920324e-05,
      "loss": 1.1719,
      "step": 573
    },
    {
      "epoch": 0.2200920245398773,
      "grad_norm": 54.652565002441406,
      "learning_rate": 4.350234341712825e-05,
      "loss": 0.8374,
      "step": 574
    },
    {
      "epoch": 0.2204754601226994,
      "grad_norm": 0.6231522560119629,
      "learning_rate": 4.348103962505326e-05,
      "loss": 0.0037,
      "step": 575
    },
    {
      "epoch": 0.22085889570552147,
      "grad_norm": 0.3720647692680359,
      "learning_rate": 4.3459735832978274e-05,
      "loss": 0.0018,
      "step": 576
    },
    {
      "epoch": 0.22124233128834356,
      "grad_norm": 142.35353088378906,
      "learning_rate": 4.3438432040903284e-05,
      "loss": 1.7725,
      "step": 577
    },
    {
      "epoch": 0.22162576687116564,
      "grad_norm": 55.13176727294922,
      "learning_rate": 4.341712824882829e-05,
      "loss": 0.7578,
      "step": 578
    },
    {
      "epoch": 0.22200920245398773,
      "grad_norm": 1.6380528211593628,
      "learning_rate": 4.33958244567533e-05,
      "loss": 0.0051,
      "step": 579
    },
    {
      "epoch": 0.2223926380368098,
      "grad_norm": 4.648425102233887,
      "learning_rate": 4.337452066467831e-05,
      "loss": 0.0107,
      "step": 580
    },
    {
      "epoch": 0.2227760736196319,
      "grad_norm": 153.6146697998047,
      "learning_rate": 4.335321687260333e-05,
      "loss": 2.2267,
      "step": 581
    },
    {
      "epoch": 0.22315950920245398,
      "grad_norm": 132.43736267089844,
      "learning_rate": 4.333191308052834e-05,
      "loss": 1.9737,
      "step": 582
    },
    {
      "epoch": 0.22354294478527606,
      "grad_norm": 138.14703369140625,
      "learning_rate": 4.3310609288453346e-05,
      "loss": 1.2608,
      "step": 583
    },
    {
      "epoch": 0.22392638036809817,
      "grad_norm": 108.3702392578125,
      "learning_rate": 4.3289305496378356e-05,
      "loss": 5.8828,
      "step": 584
    },
    {
      "epoch": 0.22430981595092025,
      "grad_norm": 0.3305304944515228,
      "learning_rate": 4.3268001704303365e-05,
      "loss": 0.0018,
      "step": 585
    },
    {
      "epoch": 0.22469325153374234,
      "grad_norm": 26.56437110900879,
      "learning_rate": 4.324669791222838e-05,
      "loss": 0.6597,
      "step": 586
    },
    {
      "epoch": 0.22507668711656442,
      "grad_norm": 15.87643814086914,
      "learning_rate": 4.322539412015339e-05,
      "loss": 0.0642,
      "step": 587
    },
    {
      "epoch": 0.2254601226993865,
      "grad_norm": 66.76805114746094,
      "learning_rate": 4.32040903280784e-05,
      "loss": 1.0527,
      "step": 588
    },
    {
      "epoch": 0.2258435582822086,
      "grad_norm": 57.99732208251953,
      "learning_rate": 4.318278653600341e-05,
      "loss": 0.7607,
      "step": 589
    },
    {
      "epoch": 0.22622699386503067,
      "grad_norm": 3.2562761306762695,
      "learning_rate": 4.3161482743928425e-05,
      "loss": 0.0178,
      "step": 590
    },
    {
      "epoch": 0.22661042944785276,
      "grad_norm": 226.23036193847656,
      "learning_rate": 4.3140178951853434e-05,
      "loss": 1.9583,
      "step": 591
    },
    {
      "epoch": 0.22699386503067484,
      "grad_norm": 45.63861846923828,
      "learning_rate": 4.3118875159778444e-05,
      "loss": 0.6699,
      "step": 592
    },
    {
      "epoch": 0.22737730061349692,
      "grad_norm": 1.0839177370071411,
      "learning_rate": 4.309757136770345e-05,
      "loss": 0.0045,
      "step": 593
    },
    {
      "epoch": 0.227760736196319,
      "grad_norm": 10.099230766296387,
      "learning_rate": 4.307626757562846e-05,
      "loss": 0.0467,
      "step": 594
    },
    {
      "epoch": 0.2281441717791411,
      "grad_norm": 66.15339660644531,
      "learning_rate": 4.305496378355348e-05,
      "loss": 0.7017,
      "step": 595
    },
    {
      "epoch": 0.2285276073619632,
      "grad_norm": 0.2999923527240753,
      "learning_rate": 4.303365999147849e-05,
      "loss": 0.0016,
      "step": 596
    },
    {
      "epoch": 0.2289110429447853,
      "grad_norm": 13.346380233764648,
      "learning_rate": 4.30123561994035e-05,
      "loss": 0.0342,
      "step": 597
    },
    {
      "epoch": 0.22929447852760737,
      "grad_norm": 2.257432699203491,
      "learning_rate": 4.2991052407328506e-05,
      "loss": 0.0107,
      "step": 598
    },
    {
      "epoch": 0.22967791411042945,
      "grad_norm": 1.0266526937484741,
      "learning_rate": 4.2969748615253516e-05,
      "loss": 0.0033,
      "step": 599
    },
    {
      "epoch": 0.23006134969325154,
      "grad_norm": 0.8544795513153076,
      "learning_rate": 4.294844482317853e-05,
      "loss": 0.0029,
      "step": 600
    },
    {
      "epoch": 0.23044478527607362,
      "grad_norm": 23.01372718811035,
      "learning_rate": 4.292714103110354e-05,
      "loss": 0.7617,
      "step": 601
    },
    {
      "epoch": 0.2308282208588957,
      "grad_norm": 1.7772164344787598,
      "learning_rate": 4.290583723902855e-05,
      "loss": 0.0027,
      "step": 602
    },
    {
      "epoch": 0.2312116564417178,
      "grad_norm": 72.03024291992188,
      "learning_rate": 4.288453344695356e-05,
      "loss": 0.8277,
      "step": 603
    },
    {
      "epoch": 0.23159509202453987,
      "grad_norm": 16.46124839782715,
      "learning_rate": 4.2863229654878576e-05,
      "loss": 0.5488,
      "step": 604
    },
    {
      "epoch": 0.23197852760736196,
      "grad_norm": 0.7167817950248718,
      "learning_rate": 4.2841925862803585e-05,
      "loss": 0.0034,
      "step": 605
    },
    {
      "epoch": 0.23236196319018404,
      "grad_norm": 4.347182750701904,
      "learning_rate": 4.2820622070728594e-05,
      "loss": 0.0047,
      "step": 606
    },
    {
      "epoch": 0.23274539877300612,
      "grad_norm": 2.2599356174468994,
      "learning_rate": 4.2799318278653604e-05,
      "loss": 0.005,
      "step": 607
    },
    {
      "epoch": 0.2331288343558282,
      "grad_norm": 7.927770614624023,
      "learning_rate": 4.277801448657861e-05,
      "loss": 0.0365,
      "step": 608
    },
    {
      "epoch": 0.23351226993865032,
      "grad_norm": 111.94647216796875,
      "learning_rate": 4.275671069450362e-05,
      "loss": 5.792,
      "step": 609
    },
    {
      "epoch": 0.2338957055214724,
      "grad_norm": 0.46775519847869873,
      "learning_rate": 4.273540690242863e-05,
      "loss": 0.0026,
      "step": 610
    },
    {
      "epoch": 0.23427914110429449,
      "grad_norm": 72.51065063476562,
      "learning_rate": 4.271410311035364e-05,
      "loss": 0.5313,
      "step": 611
    },
    {
      "epoch": 0.23466257668711657,
      "grad_norm": 127.6706771850586,
      "learning_rate": 4.269279931827865e-05,
      "loss": 5.7559,
      "step": 612
    },
    {
      "epoch": 0.23504601226993865,
      "grad_norm": 117.30060577392578,
      "learning_rate": 4.2671495526203666e-05,
      "loss": 1.3448,
      "step": 613
    },
    {
      "epoch": 0.23542944785276074,
      "grad_norm": 51.63465881347656,
      "learning_rate": 4.2650191734128676e-05,
      "loss": 0.6577,
      "step": 614
    },
    {
      "epoch": 0.23581288343558282,
      "grad_norm": 628.3287353515625,
      "learning_rate": 4.2628887942053685e-05,
      "loss": 2.7054,
      "step": 615
    },
    {
      "epoch": 0.2361963190184049,
      "grad_norm": 8.268697738647461,
      "learning_rate": 4.2607584149978694e-05,
      "loss": 0.0254,
      "step": 616
    },
    {
      "epoch": 0.236579754601227,
      "grad_norm": 36.803314208984375,
      "learning_rate": 4.258628035790371e-05,
      "loss": 0.6128,
      "step": 617
    },
    {
      "epoch": 0.23696319018404907,
      "grad_norm": 0.6197580099105835,
      "learning_rate": 4.256497656582872e-05,
      "loss": 0.0018,
      "step": 618
    },
    {
      "epoch": 0.23734662576687116,
      "grad_norm": 0.23906821012496948,
      "learning_rate": 4.254367277375373e-05,
      "loss": 0.0013,
      "step": 619
    },
    {
      "epoch": 0.23773006134969324,
      "grad_norm": 136.54547119140625,
      "learning_rate": 4.252236898167874e-05,
      "loss": 3.1651,
      "step": 620
    },
    {
      "epoch": 0.23811349693251535,
      "grad_norm": 81.10930633544922,
      "learning_rate": 4.250106518960375e-05,
      "loss": 1.3369,
      "step": 621
    },
    {
      "epoch": 0.23849693251533743,
      "grad_norm": 0.25049564242362976,
      "learning_rate": 4.2479761397528764e-05,
      "loss": 0.0022,
      "step": 622
    },
    {
      "epoch": 0.23888036809815952,
      "grad_norm": 22.74750328063965,
      "learning_rate": 4.245845760545377e-05,
      "loss": 0.0658,
      "step": 623
    },
    {
      "epoch": 0.2392638036809816,
      "grad_norm": 2.002502918243408,
      "learning_rate": 4.243715381337878e-05,
      "loss": 0.0106,
      "step": 624
    },
    {
      "epoch": 0.23964723926380369,
      "grad_norm": 2.6754136085510254,
      "learning_rate": 4.241585002130379e-05,
      "loss": 0.0089,
      "step": 625
    },
    {
      "epoch": 0.24003067484662577,
      "grad_norm": 0.7828622460365295,
      "learning_rate": 4.239454622922881e-05,
      "loss": 0.0035,
      "step": 626
    },
    {
      "epoch": 0.24041411042944785,
      "grad_norm": 94.25086212158203,
      "learning_rate": 4.237324243715382e-05,
      "loss": 5.7812,
      "step": 627
    },
    {
      "epoch": 0.24079754601226994,
      "grad_norm": 7.338162899017334,
      "learning_rate": 4.2351938645078826e-05,
      "loss": 0.1679,
      "step": 628
    },
    {
      "epoch": 0.24118098159509202,
      "grad_norm": 5.649900913238525,
      "learning_rate": 4.2330634853003836e-05,
      "loss": 0.0223,
      "step": 629
    },
    {
      "epoch": 0.2415644171779141,
      "grad_norm": 128.02952575683594,
      "learning_rate": 4.2309331060928845e-05,
      "loss": 0.9888,
      "step": 630
    },
    {
      "epoch": 0.2419478527607362,
      "grad_norm": 181.65542602539062,
      "learning_rate": 4.228802726885386e-05,
      "loss": 5.6309,
      "step": 631
    },
    {
      "epoch": 0.24233128834355827,
      "grad_norm": 45.65901565551758,
      "learning_rate": 4.226672347677887e-05,
      "loss": 0.77,
      "step": 632
    },
    {
      "epoch": 0.24271472392638035,
      "grad_norm": 193.27027893066406,
      "learning_rate": 4.224541968470388e-05,
      "loss": 4.4053,
      "step": 633
    },
    {
      "epoch": 0.24309815950920247,
      "grad_norm": 11.706387519836426,
      "learning_rate": 4.222411589262889e-05,
      "loss": 0.07,
      "step": 634
    },
    {
      "epoch": 0.24348159509202455,
      "grad_norm": 121.58812713623047,
      "learning_rate": 4.22028121005539e-05,
      "loss": 1.4532,
      "step": 635
    },
    {
      "epoch": 0.24386503067484663,
      "grad_norm": 159.98052978515625,
      "learning_rate": 4.2181508308478914e-05,
      "loss": 6.2891,
      "step": 636
    },
    {
      "epoch": 0.24424846625766872,
      "grad_norm": 122.95700073242188,
      "learning_rate": 4.2160204516403924e-05,
      "loss": 6.1455,
      "step": 637
    },
    {
      "epoch": 0.2446319018404908,
      "grad_norm": 3.863123655319214,
      "learning_rate": 4.213890072432893e-05,
      "loss": 0.0217,
      "step": 638
    },
    {
      "epoch": 0.24501533742331288,
      "grad_norm": 143.37567138671875,
      "learning_rate": 4.211759693225394e-05,
      "loss": 5.5938,
      "step": 639
    },
    {
      "epoch": 0.24539877300613497,
      "grad_norm": 4.452521800994873,
      "learning_rate": 4.209629314017896e-05,
      "loss": 0.02,
      "step": 640
    },
    {
      "epoch": 0.24578220858895705,
      "grad_norm": 110.67408752441406,
      "learning_rate": 4.207498934810397e-05,
      "loss": 1.2315,
      "step": 641
    },
    {
      "epoch": 0.24616564417177914,
      "grad_norm": 35.09891128540039,
      "learning_rate": 4.205368555602898e-05,
      "loss": 0.6338,
      "step": 642
    },
    {
      "epoch": 0.24654907975460122,
      "grad_norm": 68.7496109008789,
      "learning_rate": 4.2032381763953986e-05,
      "loss": 5.666,
      "step": 643
    },
    {
      "epoch": 0.2469325153374233,
      "grad_norm": 0.4419560730457306,
      "learning_rate": 4.2011077971878996e-05,
      "loss": 0.0021,
      "step": 644
    },
    {
      "epoch": 0.2473159509202454,
      "grad_norm": 1.9711453914642334,
      "learning_rate": 4.198977417980401e-05,
      "loss": 0.007,
      "step": 645
    },
    {
      "epoch": 0.2476993865030675,
      "grad_norm": 170.4474639892578,
      "learning_rate": 4.196847038772902e-05,
      "loss": 4.8525,
      "step": 646
    },
    {
      "epoch": 0.24808282208588958,
      "grad_norm": 10.959206581115723,
      "learning_rate": 4.194716659565403e-05,
      "loss": 0.0409,
      "step": 647
    },
    {
      "epoch": 0.24846625766871167,
      "grad_norm": 164.01382446289062,
      "learning_rate": 4.192586280357904e-05,
      "loss": 5.5068,
      "step": 648
    },
    {
      "epoch": 0.24884969325153375,
      "grad_norm": 2.1526594161987305,
      "learning_rate": 4.190455901150405e-05,
      "loss": 0.0089,
      "step": 649
    },
    {
      "epoch": 0.24923312883435583,
      "grad_norm": 1.0481888055801392,
      "learning_rate": 4.1883255219429065e-05,
      "loss": 0.005,
      "step": 650
    },
    {
      "epoch": 0.24961656441717792,
      "grad_norm": 1.0030834674835205,
      "learning_rate": 4.1861951427354074e-05,
      "loss": 0.0046,
      "step": 651
    },
    {
      "epoch": 0.25,
      "grad_norm": 23.366331100463867,
      "learning_rate": 4.1840647635279084e-05,
      "loss": 0.5884,
      "step": 652
    },
    {
      "epoch": 0.2503834355828221,
      "grad_norm": 1.5120415687561035,
      "learning_rate": 4.181934384320409e-05,
      "loss": 0.0074,
      "step": 653
    },
    {
      "epoch": 0.25076687116564417,
      "grad_norm": 0.35710829496383667,
      "learning_rate": 4.17980400511291e-05,
      "loss": 0.0028,
      "step": 654
    },
    {
      "epoch": 0.2511503067484663,
      "grad_norm": 33.088714599609375,
      "learning_rate": 4.177673625905411e-05,
      "loss": 0.5952,
      "step": 655
    },
    {
      "epoch": 0.25153374233128833,
      "grad_norm": 144.0160675048828,
      "learning_rate": 4.175543246697912e-05,
      "loss": 5.8027,
      "step": 656
    },
    {
      "epoch": 0.25191717791411045,
      "grad_norm": 112.53301239013672,
      "learning_rate": 4.173412867490413e-05,
      "loss": 1.4445,
      "step": 657
    },
    {
      "epoch": 0.2523006134969325,
      "grad_norm": 180.619873046875,
      "learning_rate": 4.1712824882829146e-05,
      "loss": 5.5791,
      "step": 658
    },
    {
      "epoch": 0.2526840490797546,
      "grad_norm": 4.506056308746338,
      "learning_rate": 4.1691521090754156e-05,
      "loss": 0.0169,
      "step": 659
    },
    {
      "epoch": 0.25306748466257667,
      "grad_norm": 4.511041641235352,
      "learning_rate": 4.1670217298679165e-05,
      "loss": 0.0065,
      "step": 660
    },
    {
      "epoch": 0.2534509202453988,
      "grad_norm": 20.754192352294922,
      "learning_rate": 4.1648913506604174e-05,
      "loss": 0.6382,
      "step": 661
    },
    {
      "epoch": 0.25383435582822084,
      "grad_norm": 2.1220593452453613,
      "learning_rate": 4.1627609714529183e-05,
      "loss": 0.007,
      "step": 662
    },
    {
      "epoch": 0.25421779141104295,
      "grad_norm": 120.6075210571289,
      "learning_rate": 4.16063059224542e-05,
      "loss": 2.1088,
      "step": 663
    },
    {
      "epoch": 0.254601226993865,
      "grad_norm": 1.466044306755066,
      "learning_rate": 4.158500213037921e-05,
      "loss": 0.0044,
      "step": 664
    },
    {
      "epoch": 0.2549846625766871,
      "grad_norm": 41.46731185913086,
      "learning_rate": 4.156369833830422e-05,
      "loss": 0.5107,
      "step": 665
    },
    {
      "epoch": 0.2553680981595092,
      "grad_norm": 1.1529388427734375,
      "learning_rate": 4.154239454622923e-05,
      "loss": 0.002,
      "step": 666
    },
    {
      "epoch": 0.2557515337423313,
      "grad_norm": 8.65416431427002,
      "learning_rate": 4.1521090754154244e-05,
      "loss": 0.0345,
      "step": 667
    },
    {
      "epoch": 0.2561349693251534,
      "grad_norm": 22.93845558166504,
      "learning_rate": 4.149978696207925e-05,
      "loss": 0.5298,
      "step": 668
    },
    {
      "epoch": 0.25651840490797545,
      "grad_norm": 3.1691551208496094,
      "learning_rate": 4.147848317000426e-05,
      "loss": 0.0062,
      "step": 669
    },
    {
      "epoch": 0.25690184049079756,
      "grad_norm": 15.257349967956543,
      "learning_rate": 4.145717937792927e-05,
      "loss": 0.4937,
      "step": 670
    },
    {
      "epoch": 0.2572852760736196,
      "grad_norm": 3.3930716514587402,
      "learning_rate": 4.143587558585428e-05,
      "loss": 0.0194,
      "step": 671
    },
    {
      "epoch": 0.25766871165644173,
      "grad_norm": 1.1906558275222778,
      "learning_rate": 4.14145717937793e-05,
      "loss": 0.0059,
      "step": 672
    },
    {
      "epoch": 0.2580521472392638,
      "grad_norm": 14.620223045349121,
      "learning_rate": 4.1393268001704306e-05,
      "loss": 0.5498,
      "step": 673
    },
    {
      "epoch": 0.2584355828220859,
      "grad_norm": 0.23422184586524963,
      "learning_rate": 4.1371964209629316e-05,
      "loss": 0.0018,
      "step": 674
    },
    {
      "epoch": 0.25881901840490795,
      "grad_norm": 120.83528900146484,
      "learning_rate": 4.1350660417554325e-05,
      "loss": 1.7823,
      "step": 675
    },
    {
      "epoch": 0.25920245398773006,
      "grad_norm": 1.0207319259643555,
      "learning_rate": 4.1329356625479334e-05,
      "loss": 0.0049,
      "step": 676
    },
    {
      "epoch": 0.2595858895705521,
      "grad_norm": 1.2075201272964478,
      "learning_rate": 4.130805283340435e-05,
      "loss": 0.0067,
      "step": 677
    },
    {
      "epoch": 0.25996932515337423,
      "grad_norm": 0.39283424615859985,
      "learning_rate": 4.128674904132936e-05,
      "loss": 0.0022,
      "step": 678
    },
    {
      "epoch": 0.26035276073619634,
      "grad_norm": 0.2834872603416443,
      "learning_rate": 4.126544524925437e-05,
      "loss": 0.0021,
      "step": 679
    },
    {
      "epoch": 0.2607361963190184,
      "grad_norm": 202.35397338867188,
      "learning_rate": 4.124414145717938e-05,
      "loss": 1.4132,
      "step": 680
    },
    {
      "epoch": 0.2611196319018405,
      "grad_norm": 0.7038347721099854,
      "learning_rate": 4.1222837665104394e-05,
      "loss": 0.0037,
      "step": 681
    },
    {
      "epoch": 0.26150306748466257,
      "grad_norm": 189.6617889404297,
      "learning_rate": 4.1201533873029404e-05,
      "loss": 1.214,
      "step": 682
    },
    {
      "epoch": 0.2618865030674847,
      "grad_norm": 31.273454666137695,
      "learning_rate": 4.118023008095441e-05,
      "loss": 0.5957,
      "step": 683
    },
    {
      "epoch": 0.26226993865030673,
      "grad_norm": 0.646104097366333,
      "learning_rate": 4.115892628887942e-05,
      "loss": 0.0044,
      "step": 684
    },
    {
      "epoch": 0.26265337423312884,
      "grad_norm": 157.92442321777344,
      "learning_rate": 4.113762249680443e-05,
      "loss": 2.3674,
      "step": 685
    },
    {
      "epoch": 0.2630368098159509,
      "grad_norm": 38.39439392089844,
      "learning_rate": 4.111631870472945e-05,
      "loss": 0.6558,
      "step": 686
    },
    {
      "epoch": 0.263420245398773,
      "grad_norm": 2.025766611099243,
      "learning_rate": 4.109501491265446e-05,
      "loss": 0.0108,
      "step": 687
    },
    {
      "epoch": 0.26380368098159507,
      "grad_norm": 239.11248779296875,
      "learning_rate": 4.1073711120579466e-05,
      "loss": 1.8917,
      "step": 688
    },
    {
      "epoch": 0.2641871165644172,
      "grad_norm": 1.424906611442566,
      "learning_rate": 4.1052407328504476e-05,
      "loss": 0.0087,
      "step": 689
    },
    {
      "epoch": 0.2645705521472393,
      "grad_norm": 215.03074645996094,
      "learning_rate": 4.1031103536429485e-05,
      "loss": 5.6675,
      "step": 690
    },
    {
      "epoch": 0.26495398773006135,
      "grad_norm": 11.536243438720703,
      "learning_rate": 4.10097997443545e-05,
      "loss": 0.1388,
      "step": 691
    },
    {
      "epoch": 0.26533742331288346,
      "grad_norm": 102.15013122558594,
      "learning_rate": 4.098849595227951e-05,
      "loss": 5.1914,
      "step": 692
    },
    {
      "epoch": 0.2657208588957055,
      "grad_norm": 0.3701198101043701,
      "learning_rate": 4.096719216020452e-05,
      "loss": 0.0029,
      "step": 693
    },
    {
      "epoch": 0.2661042944785276,
      "grad_norm": 65.08767700195312,
      "learning_rate": 4.094588836812953e-05,
      "loss": 0.7599,
      "step": 694
    },
    {
      "epoch": 0.2664877300613497,
      "grad_norm": 9.87397289276123,
      "learning_rate": 4.0924584576054545e-05,
      "loss": 0.0347,
      "step": 695
    },
    {
      "epoch": 0.2668711656441718,
      "grad_norm": 8.291452407836914,
      "learning_rate": 4.0903280783979554e-05,
      "loss": 0.0771,
      "step": 696
    },
    {
      "epoch": 0.26725460122699385,
      "grad_norm": 209.46115112304688,
      "learning_rate": 4.0881976991904564e-05,
      "loss": 3.0864,
      "step": 697
    },
    {
      "epoch": 0.26763803680981596,
      "grad_norm": 216.21163940429688,
      "learning_rate": 4.0860673199829566e-05,
      "loss": 2.5413,
      "step": 698
    },
    {
      "epoch": 0.268021472392638,
      "grad_norm": 2.646641969680786,
      "learning_rate": 4.083936940775458e-05,
      "loss": 0.0041,
      "step": 699
    },
    {
      "epoch": 0.2684049079754601,
      "grad_norm": 156.0490264892578,
      "learning_rate": 4.081806561567959e-05,
      "loss": 1.5695,
      "step": 700
    },
    {
      "epoch": 0.2687883435582822,
      "grad_norm": 0.74263596534729,
      "learning_rate": 4.07967618236046e-05,
      "loss": 0.0024,
      "step": 701
    },
    {
      "epoch": 0.2691717791411043,
      "grad_norm": 120.81138610839844,
      "learning_rate": 4.077545803152961e-05,
      "loss": 1.086,
      "step": 702
    },
    {
      "epoch": 0.2695552147239264,
      "grad_norm": 14.250238418579102,
      "learning_rate": 4.075415423945462e-05,
      "loss": 0.055,
      "step": 703
    },
    {
      "epoch": 0.26993865030674846,
      "grad_norm": 146.56906127929688,
      "learning_rate": 4.0732850447379636e-05,
      "loss": 1.5176,
      "step": 704
    },
    {
      "epoch": 0.2703220858895706,
      "grad_norm": 3.4120378494262695,
      "learning_rate": 4.0711546655304645e-05,
      "loss": 0.0138,
      "step": 705
    },
    {
      "epoch": 0.27070552147239263,
      "grad_norm": 1.6495180130004883,
      "learning_rate": 4.0690242863229654e-05,
      "loss": 0.0087,
      "step": 706
    },
    {
      "epoch": 0.27108895705521474,
      "grad_norm": 1.0825520753860474,
      "learning_rate": 4.0668939071154663e-05,
      "loss": 0.0056,
      "step": 707
    },
    {
      "epoch": 0.2714723926380368,
      "grad_norm": 129.17660522460938,
      "learning_rate": 4.064763527907968e-05,
      "loss": 1.7218,
      "step": 708
    },
    {
      "epoch": 0.2718558282208589,
      "grad_norm": 1.3704122304916382,
      "learning_rate": 4.062633148700469e-05,
      "loss": 0.0055,
      "step": 709
    },
    {
      "epoch": 0.27223926380368096,
      "grad_norm": 19.120973587036133,
      "learning_rate": 4.06050276949297e-05,
      "loss": 0.0598,
      "step": 710
    },
    {
      "epoch": 0.2726226993865031,
      "grad_norm": 9.105579376220703,
      "learning_rate": 4.058372390285471e-05,
      "loss": 0.0208,
      "step": 711
    },
    {
      "epoch": 0.27300613496932513,
      "grad_norm": 9.893427848815918,
      "learning_rate": 4.056242011077972e-05,
      "loss": 0.034,
      "step": 712
    },
    {
      "epoch": 0.27338957055214724,
      "grad_norm": 10.373291969299316,
      "learning_rate": 4.054111631870473e-05,
      "loss": 0.0451,
      "step": 713
    },
    {
      "epoch": 0.2737730061349693,
      "grad_norm": 48.78724670410156,
      "learning_rate": 4.051981252662974e-05,
      "loss": 0.6045,
      "step": 714
    },
    {
      "epoch": 0.2741564417177914,
      "grad_norm": 45.57914733886719,
      "learning_rate": 4.049850873455475e-05,
      "loss": 0.7661,
      "step": 715
    },
    {
      "epoch": 0.2745398773006135,
      "grad_norm": 71.10076904296875,
      "learning_rate": 4.047720494247976e-05,
      "loss": 1.6184,
      "step": 716
    },
    {
      "epoch": 0.2749233128834356,
      "grad_norm": 8.53931999206543,
      "learning_rate": 4.045590115040478e-05,
      "loss": 0.0327,
      "step": 717
    },
    {
      "epoch": 0.2753067484662577,
      "grad_norm": 57.249568939208984,
      "learning_rate": 4.0434597358329786e-05,
      "loss": 0.8604,
      "step": 718
    },
    {
      "epoch": 0.27569018404907975,
      "grad_norm": 108.18406677246094,
      "learning_rate": 4.0413293566254796e-05,
      "loss": 1.3262,
      "step": 719
    },
    {
      "epoch": 0.27607361963190186,
      "grad_norm": 1.0649335384368896,
      "learning_rate": 4.0391989774179805e-05,
      "loss": 0.0041,
      "step": 720
    },
    {
      "epoch": 0.2764570552147239,
      "grad_norm": 17.896425247192383,
      "learning_rate": 4.0370685982104814e-05,
      "loss": 0.1149,
      "step": 721
    },
    {
      "epoch": 0.276840490797546,
      "grad_norm": 75.33570098876953,
      "learning_rate": 4.034938219002983e-05,
      "loss": 1.0372,
      "step": 722
    },
    {
      "epoch": 0.2772239263803681,
      "grad_norm": 293.58526611328125,
      "learning_rate": 4.032807839795484e-05,
      "loss": 5.5889,
      "step": 723
    },
    {
      "epoch": 0.2776073619631902,
      "grad_norm": 45.82387924194336,
      "learning_rate": 4.030677460587985e-05,
      "loss": 0.751,
      "step": 724
    },
    {
      "epoch": 0.27799079754601225,
      "grad_norm": 4.106308937072754,
      "learning_rate": 4.028547081380486e-05,
      "loss": 0.0154,
      "step": 725
    },
    {
      "epoch": 0.27837423312883436,
      "grad_norm": 27.708314895629883,
      "learning_rate": 4.026416702172987e-05,
      "loss": 0.6475,
      "step": 726
    },
    {
      "epoch": 0.2787576687116564,
      "grad_norm": 2.0057215690612793,
      "learning_rate": 4.0242863229654884e-05,
      "loss": 0.005,
      "step": 727
    },
    {
      "epoch": 0.2791411042944785,
      "grad_norm": 12.491474151611328,
      "learning_rate": 4.022155943757989e-05,
      "loss": 0.0745,
      "step": 728
    },
    {
      "epoch": 0.27952453987730064,
      "grad_norm": 0.5827816128730774,
      "learning_rate": 4.02002556455049e-05,
      "loss": 0.0029,
      "step": 729
    },
    {
      "epoch": 0.2799079754601227,
      "grad_norm": 123.37541198730469,
      "learning_rate": 4.017895185342991e-05,
      "loss": 1.7745,
      "step": 730
    },
    {
      "epoch": 0.2802914110429448,
      "grad_norm": 44.76090621948242,
      "learning_rate": 4.015764806135493e-05,
      "loss": 0.6943,
      "step": 731
    },
    {
      "epoch": 0.28067484662576686,
      "grad_norm": 4.828588008880615,
      "learning_rate": 4.013634426927994e-05,
      "loss": 0.0133,
      "step": 732
    },
    {
      "epoch": 0.28105828220858897,
      "grad_norm": 5.215601921081543,
      "learning_rate": 4.0115040477204946e-05,
      "loss": 0.0117,
      "step": 733
    },
    {
      "epoch": 0.28144171779141103,
      "grad_norm": 0.5461913347244263,
      "learning_rate": 4.0093736685129956e-05,
      "loss": 0.0029,
      "step": 734
    },
    {
      "epoch": 0.28182515337423314,
      "grad_norm": 125.95172882080078,
      "learning_rate": 4.0072432893054965e-05,
      "loss": 1.6723,
      "step": 735
    },
    {
      "epoch": 0.2822085889570552,
      "grad_norm": 2.944730043411255,
      "learning_rate": 4.005112910097998e-05,
      "loss": 0.0209,
      "step": 736
    },
    {
      "epoch": 0.2825920245398773,
      "grad_norm": 17.602924346923828,
      "learning_rate": 4.002982530890499e-05,
      "loss": 0.1221,
      "step": 737
    },
    {
      "epoch": 0.28297546012269936,
      "grad_norm": 3.474398612976074,
      "learning_rate": 4.000852151683e-05,
      "loss": 0.0083,
      "step": 738
    },
    {
      "epoch": 0.2833588957055215,
      "grad_norm": 191.66481018066406,
      "learning_rate": 3.998721772475501e-05,
      "loss": 1.3117,
      "step": 739
    },
    {
      "epoch": 0.2837423312883436,
      "grad_norm": 5.7077412605285645,
      "learning_rate": 3.996591393268002e-05,
      "loss": 0.0304,
      "step": 740
    },
    {
      "epoch": 0.28412576687116564,
      "grad_norm": 1.7796605825424194,
      "learning_rate": 3.9944610140605034e-05,
      "loss": 0.0048,
      "step": 741
    },
    {
      "epoch": 0.28450920245398775,
      "grad_norm": 20.046506881713867,
      "learning_rate": 3.9923306348530044e-05,
      "loss": 0.6294,
      "step": 742
    },
    {
      "epoch": 0.2848926380368098,
      "grad_norm": 8.310009956359863,
      "learning_rate": 3.9902002556455046e-05,
      "loss": 0.4993,
      "step": 743
    },
    {
      "epoch": 0.2852760736196319,
      "grad_norm": 115.0446548461914,
      "learning_rate": 3.988069876438006e-05,
      "loss": 5.3862,
      "step": 744
    },
    {
      "epoch": 0.285659509202454,
      "grad_norm": 20.69170570373535,
      "learning_rate": 3.985939497230507e-05,
      "loss": 0.5698,
      "step": 745
    },
    {
      "epoch": 0.2860429447852761,
      "grad_norm": 1.4113951921463013,
      "learning_rate": 3.983809118023008e-05,
      "loss": 0.0058,
      "step": 746
    },
    {
      "epoch": 0.28642638036809814,
      "grad_norm": 2.7105507850646973,
      "learning_rate": 3.981678738815509e-05,
      "loss": 0.0033,
      "step": 747
    },
    {
      "epoch": 0.28680981595092025,
      "grad_norm": 21.30638885498047,
      "learning_rate": 3.97954835960801e-05,
      "loss": 0.5322,
      "step": 748
    },
    {
      "epoch": 0.2871932515337423,
      "grad_norm": 16.875904083251953,
      "learning_rate": 3.9774179804005116e-05,
      "loss": 0.0605,
      "step": 749
    },
    {
      "epoch": 0.2875766871165644,
      "grad_norm": 32.32379913330078,
      "learning_rate": 3.9752876011930125e-05,
      "loss": 0.7578,
      "step": 750
    },
    {
      "epoch": 0.2879601226993865,
      "grad_norm": 124.02713012695312,
      "learning_rate": 3.9731572219855134e-05,
      "loss": 0.981,
      "step": 751
    },
    {
      "epoch": 0.2883435582822086,
      "grad_norm": 100.00495910644531,
      "learning_rate": 3.9710268427780143e-05,
      "loss": 1.5207,
      "step": 752
    },
    {
      "epoch": 0.2887269938650307,
      "grad_norm": 142.09523010253906,
      "learning_rate": 3.968896463570515e-05,
      "loss": 6.4824,
      "step": 753
    },
    {
      "epoch": 0.28911042944785276,
      "grad_norm": 0.56594318151474,
      "learning_rate": 3.966766084363017e-05,
      "loss": 0.0031,
      "step": 754
    },
    {
      "epoch": 0.28949386503067487,
      "grad_norm": 7.328930854797363,
      "learning_rate": 3.964635705155518e-05,
      "loss": 0.0452,
      "step": 755
    },
    {
      "epoch": 0.2898773006134969,
      "grad_norm": 0.901790201663971,
      "learning_rate": 3.962505325948019e-05,
      "loss": 0.0028,
      "step": 756
    },
    {
      "epoch": 0.29026073619631904,
      "grad_norm": 3.4390993118286133,
      "learning_rate": 3.96037494674052e-05,
      "loss": 0.0098,
      "step": 757
    },
    {
      "epoch": 0.2906441717791411,
      "grad_norm": 22.084123611450195,
      "learning_rate": 3.958244567533021e-05,
      "loss": 0.5352,
      "step": 758
    },
    {
      "epoch": 0.2910276073619632,
      "grad_norm": 1.4480448961257935,
      "learning_rate": 3.956114188325522e-05,
      "loss": 0.0038,
      "step": 759
    },
    {
      "epoch": 0.29141104294478526,
      "grad_norm": 0.6343742609024048,
      "learning_rate": 3.953983809118023e-05,
      "loss": 0.0027,
      "step": 760
    },
    {
      "epoch": 0.29179447852760737,
      "grad_norm": 6.620326995849609,
      "learning_rate": 3.951853429910524e-05,
      "loss": 0.2262,
      "step": 761
    },
    {
      "epoch": 0.2921779141104294,
      "grad_norm": 0.6557702422142029,
      "learning_rate": 3.949723050703025e-05,
      "loss": 0.003,
      "step": 762
    },
    {
      "epoch": 0.29256134969325154,
      "grad_norm": 87.16983795166016,
      "learning_rate": 3.9475926714955266e-05,
      "loss": 1.0498,
      "step": 763
    },
    {
      "epoch": 0.2929447852760736,
      "grad_norm": 496.9752197265625,
      "learning_rate": 3.9454622922880276e-05,
      "loss": 3.306,
      "step": 764
    },
    {
      "epoch": 0.2933282208588957,
      "grad_norm": 1.9638266563415527,
      "learning_rate": 3.9433319130805285e-05,
      "loss": 0.0062,
      "step": 765
    },
    {
      "epoch": 0.2937116564417178,
      "grad_norm": 5.150081157684326,
      "learning_rate": 3.9412015338730294e-05,
      "loss": 0.0124,
      "step": 766
    },
    {
      "epoch": 0.2940950920245399,
      "grad_norm": 89.9288558959961,
      "learning_rate": 3.9390711546655303e-05,
      "loss": 0.8052,
      "step": 767
    },
    {
      "epoch": 0.294478527607362,
      "grad_norm": 0.3339363634586334,
      "learning_rate": 3.936940775458032e-05,
      "loss": 0.0019,
      "step": 768
    },
    {
      "epoch": 0.29486196319018404,
      "grad_norm": 11.336874008178711,
      "learning_rate": 3.934810396250533e-05,
      "loss": 0.0814,
      "step": 769
    },
    {
      "epoch": 0.29524539877300615,
      "grad_norm": 7.84946870803833,
      "learning_rate": 3.932680017043034e-05,
      "loss": 0.0116,
      "step": 770
    },
    {
      "epoch": 0.2956288343558282,
      "grad_norm": 0.8124586343765259,
      "learning_rate": 3.930549637835535e-05,
      "loss": 0.0035,
      "step": 771
    },
    {
      "epoch": 0.2960122699386503,
      "grad_norm": 1.8907150030136108,
      "learning_rate": 3.9284192586280364e-05,
      "loss": 0.0031,
      "step": 772
    },
    {
      "epoch": 0.2963957055214724,
      "grad_norm": 11.590265274047852,
      "learning_rate": 3.926288879420537e-05,
      "loss": 0.0675,
      "step": 773
    },
    {
      "epoch": 0.2967791411042945,
      "grad_norm": 182.07699584960938,
      "learning_rate": 3.924158500213038e-05,
      "loss": 5.2842,
      "step": 774
    },
    {
      "epoch": 0.29716257668711654,
      "grad_norm": 367.9045104980469,
      "learning_rate": 3.922028121005539e-05,
      "loss": 5.6216,
      "step": 775
    },
    {
      "epoch": 0.29754601226993865,
      "grad_norm": 1.2980585098266602,
      "learning_rate": 3.91989774179804e-05,
      "loss": 0.0065,
      "step": 776
    },
    {
      "epoch": 0.2979294478527607,
      "grad_norm": 1.5405141115188599,
      "learning_rate": 3.917767362590542e-05,
      "loss": 0.0059,
      "step": 777
    },
    {
      "epoch": 0.2983128834355828,
      "grad_norm": 0.18221314251422882,
      "learning_rate": 3.9156369833830426e-05,
      "loss": 0.0029,
      "step": 778
    },
    {
      "epoch": 0.29869631901840493,
      "grad_norm": 42.80677032470703,
      "learning_rate": 3.9135066041755436e-05,
      "loss": 0.9043,
      "step": 779
    },
    {
      "epoch": 0.299079754601227,
      "grad_norm": 9.322417259216309,
      "learning_rate": 3.9113762249680445e-05,
      "loss": 0.0489,
      "step": 780
    },
    {
      "epoch": 0.2994631901840491,
      "grad_norm": 172.9141387939453,
      "learning_rate": 3.909245845760546e-05,
      "loss": 1.3604,
      "step": 781
    },
    {
      "epoch": 0.29984662576687116,
      "grad_norm": 2.292292833328247,
      "learning_rate": 3.907115466553047e-05,
      "loss": 0.0119,
      "step": 782
    },
    {
      "epoch": 0.30023006134969327,
      "grad_norm": 0.20979280769824982,
      "learning_rate": 3.904985087345548e-05,
      "loss": 0.0016,
      "step": 783
    },
    {
      "epoch": 0.3006134969325153,
      "grad_norm": 81.50609588623047,
      "learning_rate": 3.902854708138049e-05,
      "loss": 2.1564,
      "step": 784
    },
    {
      "epoch": 0.30099693251533743,
      "grad_norm": 0.5129420757293701,
      "learning_rate": 3.90072432893055e-05,
      "loss": 0.0024,
      "step": 785
    },
    {
      "epoch": 0.3013803680981595,
      "grad_norm": 2.1608102321624756,
      "learning_rate": 3.8985939497230514e-05,
      "loss": 0.0096,
      "step": 786
    },
    {
      "epoch": 0.3017638036809816,
      "grad_norm": 104.12256622314453,
      "learning_rate": 3.8964635705155524e-05,
      "loss": 5.7793,
      "step": 787
    },
    {
      "epoch": 0.30214723926380366,
      "grad_norm": 5.102387428283691,
      "learning_rate": 3.894333191308053e-05,
      "loss": 0.0276,
      "step": 788
    },
    {
      "epoch": 0.30253067484662577,
      "grad_norm": 143.1125030517578,
      "learning_rate": 3.8922028121005535e-05,
      "loss": 1.8458,
      "step": 789
    },
    {
      "epoch": 0.3029141104294479,
      "grad_norm": 5.7853684425354,
      "learning_rate": 3.890072432893055e-05,
      "loss": 0.0213,
      "step": 790
    },
    {
      "epoch": 0.30329754601226994,
      "grad_norm": 7.543581962585449,
      "learning_rate": 3.887942053685556e-05,
      "loss": 0.1757,
      "step": 791
    },
    {
      "epoch": 0.30368098159509205,
      "grad_norm": 2.6905198097229004,
      "learning_rate": 3.885811674478057e-05,
      "loss": 0.0089,
      "step": 792
    },
    {
      "epoch": 0.3040644171779141,
      "grad_norm": 69.44195556640625,
      "learning_rate": 3.883681295270558e-05,
      "loss": 0.7481,
      "step": 793
    },
    {
      "epoch": 0.3044478527607362,
      "grad_norm": 11.465123176574707,
      "learning_rate": 3.881550916063059e-05,
      "loss": 0.0668,
      "step": 794
    },
    {
      "epoch": 0.30483128834355827,
      "grad_norm": 210.43499755859375,
      "learning_rate": 3.8794205368555605e-05,
      "loss": 1.9943,
      "step": 795
    },
    {
      "epoch": 0.3052147239263804,
      "grad_norm": 4.897526741027832,
      "learning_rate": 3.8772901576480614e-05,
      "loss": 0.0399,
      "step": 796
    },
    {
      "epoch": 0.30559815950920244,
      "grad_norm": 0.5718501210212708,
      "learning_rate": 3.8751597784405623e-05,
      "loss": 0.0039,
      "step": 797
    },
    {
      "epoch": 0.30598159509202455,
      "grad_norm": 2.810070037841797,
      "learning_rate": 3.873029399233063e-05,
      "loss": 0.0138,
      "step": 798
    },
    {
      "epoch": 0.3063650306748466,
      "grad_norm": 142.47463989257812,
      "learning_rate": 3.870899020025565e-05,
      "loss": 2.2659,
      "step": 799
    },
    {
      "epoch": 0.3067484662576687,
      "grad_norm": 6.814948558807373,
      "learning_rate": 3.868768640818066e-05,
      "loss": 0.0326,
      "step": 800
    },
    {
      "epoch": 0.3071319018404908,
      "grad_norm": 2.7013139724731445,
      "learning_rate": 3.866638261610567e-05,
      "loss": 0.0122,
      "step": 801
    },
    {
      "epoch": 0.3075153374233129,
      "grad_norm": 0.8929374814033508,
      "learning_rate": 3.864507882403068e-05,
      "loss": 0.0043,
      "step": 802
    },
    {
      "epoch": 0.307898773006135,
      "grad_norm": 6.759843349456787,
      "learning_rate": 3.8623775031955686e-05,
      "loss": 0.0239,
      "step": 803
    },
    {
      "epoch": 0.30828220858895705,
      "grad_norm": 142.0617218017578,
      "learning_rate": 3.86024712398807e-05,
      "loss": 2.0098,
      "step": 804
    },
    {
      "epoch": 0.30866564417177916,
      "grad_norm": 99.49002838134766,
      "learning_rate": 3.858116744780571e-05,
      "loss": 0.8111,
      "step": 805
    },
    {
      "epoch": 0.3090490797546012,
      "grad_norm": 213.89450073242188,
      "learning_rate": 3.855986365573072e-05,
      "loss": 6.645,
      "step": 806
    },
    {
      "epoch": 0.30943251533742333,
      "grad_norm": 0.9271179437637329,
      "learning_rate": 3.853855986365573e-05,
      "loss": 0.0058,
      "step": 807
    },
    {
      "epoch": 0.3098159509202454,
      "grad_norm": 0.571173906326294,
      "learning_rate": 3.8517256071580746e-05,
      "loss": 0.0037,
      "step": 808
    },
    {
      "epoch": 0.3101993865030675,
      "grad_norm": 86.75624084472656,
      "learning_rate": 3.8495952279505756e-05,
      "loss": 1.6153,
      "step": 809
    },
    {
      "epoch": 0.31058282208588955,
      "grad_norm": 17.55738067626953,
      "learning_rate": 3.8474648487430765e-05,
      "loss": 0.5586,
      "step": 810
    },
    {
      "epoch": 0.31096625766871167,
      "grad_norm": 0.6597051620483398,
      "learning_rate": 3.8453344695355774e-05,
      "loss": 0.0042,
      "step": 811
    },
    {
      "epoch": 0.3113496932515337,
      "grad_norm": 112.5511703491211,
      "learning_rate": 3.8432040903280783e-05,
      "loss": 1.919,
      "step": 812
    },
    {
      "epoch": 0.31173312883435583,
      "grad_norm": 1.3902746438980103,
      "learning_rate": 3.84107371112058e-05,
      "loss": 0.0039,
      "step": 813
    },
    {
      "epoch": 0.3121165644171779,
      "grad_norm": 0.32743898034095764,
      "learning_rate": 3.838943331913081e-05,
      "loss": 0.0024,
      "step": 814
    },
    {
      "epoch": 0.3125,
      "grad_norm": 26.69757652282715,
      "learning_rate": 3.836812952705582e-05,
      "loss": 0.7124,
      "step": 815
    },
    {
      "epoch": 0.3128834355828221,
      "grad_norm": 147.48939514160156,
      "learning_rate": 3.834682573498083e-05,
      "loss": 3.6739,
      "step": 816
    },
    {
      "epoch": 0.31326687116564417,
      "grad_norm": 0.9667326807975769,
      "learning_rate": 3.832552194290584e-05,
      "loss": 0.0038,
      "step": 817
    },
    {
      "epoch": 0.3136503067484663,
      "grad_norm": 3.3136909008026123,
      "learning_rate": 3.830421815083085e-05,
      "loss": 0.0168,
      "step": 818
    },
    {
      "epoch": 0.31403374233128833,
      "grad_norm": 3.586291790008545,
      "learning_rate": 3.828291435875586e-05,
      "loss": 0.0177,
      "step": 819
    },
    {
      "epoch": 0.31441717791411045,
      "grad_norm": 0.5330338478088379,
      "learning_rate": 3.826161056668087e-05,
      "loss": 0.0022,
      "step": 820
    },
    {
      "epoch": 0.3148006134969325,
      "grad_norm": 3.030703544616699,
      "learning_rate": 3.824030677460588e-05,
      "loss": 0.0088,
      "step": 821
    },
    {
      "epoch": 0.3151840490797546,
      "grad_norm": 137.04339599609375,
      "learning_rate": 3.82190029825309e-05,
      "loss": 2.5419,
      "step": 822
    },
    {
      "epoch": 0.31556748466257667,
      "grad_norm": 6.643217086791992,
      "learning_rate": 3.8197699190455906e-05,
      "loss": 0.0334,
      "step": 823
    },
    {
      "epoch": 0.3159509202453988,
      "grad_norm": 75.90634155273438,
      "learning_rate": 3.8176395398380916e-05,
      "loss": 1.7666,
      "step": 824
    },
    {
      "epoch": 0.31633435582822084,
      "grad_norm": 14.753342628479004,
      "learning_rate": 3.8155091606305925e-05,
      "loss": 0.4866,
      "step": 825
    },
    {
      "epoch": 0.31671779141104295,
      "grad_norm": 218.08372497558594,
      "learning_rate": 3.8133787814230934e-05,
      "loss": 3.4464,
      "step": 826
    },
    {
      "epoch": 0.317101226993865,
      "grad_norm": 1.1979540586471558,
      "learning_rate": 3.811248402215595e-05,
      "loss": 0.0052,
      "step": 827
    },
    {
      "epoch": 0.3174846625766871,
      "grad_norm": 8.071738243103027,
      "learning_rate": 3.809118023008096e-05,
      "loss": 0.0327,
      "step": 828
    },
    {
      "epoch": 0.3178680981595092,
      "grad_norm": 0.4262201189994812,
      "learning_rate": 3.806987643800597e-05,
      "loss": 0.0028,
      "step": 829
    },
    {
      "epoch": 0.3182515337423313,
      "grad_norm": 239.7389678955078,
      "learning_rate": 3.804857264593098e-05,
      "loss": 2.2638,
      "step": 830
    },
    {
      "epoch": 0.3186349693251534,
      "grad_norm": 54.11473846435547,
      "learning_rate": 3.802726885385599e-05,
      "loss": 0.8097,
      "step": 831
    },
    {
      "epoch": 0.31901840490797545,
      "grad_norm": 2.1526424884796143,
      "learning_rate": 3.8005965061781004e-05,
      "loss": 0.005,
      "step": 832
    },
    {
      "epoch": 0.31940184049079756,
      "grad_norm": 0.307246595621109,
      "learning_rate": 3.798466126970601e-05,
      "loss": 0.0019,
      "step": 833
    },
    {
      "epoch": 0.3197852760736196,
      "grad_norm": 3.812610387802124,
      "learning_rate": 3.7963357477631015e-05,
      "loss": 0.0167,
      "step": 834
    },
    {
      "epoch": 0.32016871165644173,
      "grad_norm": 7.786213397979736,
      "learning_rate": 3.794205368555603e-05,
      "loss": 0.0101,
      "step": 835
    },
    {
      "epoch": 0.3205521472392638,
      "grad_norm": 1.0512534379959106,
      "learning_rate": 3.792074989348104e-05,
      "loss": 0.0072,
      "step": 836
    },
    {
      "epoch": 0.3209355828220859,
      "grad_norm": 190.12026977539062,
      "learning_rate": 3.789944610140605e-05,
      "loss": 1.3262,
      "step": 837
    },
    {
      "epoch": 0.32131901840490795,
      "grad_norm": 0.8700166344642639,
      "learning_rate": 3.787814230933106e-05,
      "loss": 0.0046,
      "step": 838
    },
    {
      "epoch": 0.32170245398773006,
      "grad_norm": 3.2446205615997314,
      "learning_rate": 3.785683851725607e-05,
      "loss": 0.0075,
      "step": 839
    },
    {
      "epoch": 0.3220858895705521,
      "grad_norm": 130.6759033203125,
      "learning_rate": 3.7835534725181085e-05,
      "loss": 1.763,
      "step": 840
    },
    {
      "epoch": 0.32246932515337423,
      "grad_norm": 212.87217712402344,
      "learning_rate": 3.7814230933106094e-05,
      "loss": 2.9789,
      "step": 841
    },
    {
      "epoch": 0.32285276073619634,
      "grad_norm": 37.23453140258789,
      "learning_rate": 3.7792927141031103e-05,
      "loss": 0.5869,
      "step": 842
    },
    {
      "epoch": 0.3232361963190184,
      "grad_norm": 5.587158203125,
      "learning_rate": 3.777162334895611e-05,
      "loss": 0.4417,
      "step": 843
    },
    {
      "epoch": 0.3236196319018405,
      "grad_norm": 115.92218017578125,
      "learning_rate": 3.775031955688112e-05,
      "loss": 6.5654,
      "step": 844
    },
    {
      "epoch": 0.32400306748466257,
      "grad_norm": 41.286556243896484,
      "learning_rate": 3.772901576480614e-05,
      "loss": 0.7354,
      "step": 845
    },
    {
      "epoch": 0.3243865030674847,
      "grad_norm": 0.4213690459728241,
      "learning_rate": 3.770771197273115e-05,
      "loss": 0.0031,
      "step": 846
    },
    {
      "epoch": 0.32476993865030673,
      "grad_norm": 6.38857364654541,
      "learning_rate": 3.768640818065616e-05,
      "loss": 0.0108,
      "step": 847
    },
    {
      "epoch": 0.32515337423312884,
      "grad_norm": 3.0541932582855225,
      "learning_rate": 3.7665104388581166e-05,
      "loss": 0.009,
      "step": 848
    },
    {
      "epoch": 0.3255368098159509,
      "grad_norm": 111.27130889892578,
      "learning_rate": 3.764380059650618e-05,
      "loss": 0.629,
      "step": 849
    },
    {
      "epoch": 0.325920245398773,
      "grad_norm": 11.071561813354492,
      "learning_rate": 3.762249680443119e-05,
      "loss": 0.157,
      "step": 850
    },
    {
      "epoch": 0.32630368098159507,
      "grad_norm": 0.9017576575279236,
      "learning_rate": 3.76011930123562e-05,
      "loss": 0.0061,
      "step": 851
    },
    {
      "epoch": 0.3266871165644172,
      "grad_norm": 0.2364967167377472,
      "learning_rate": 3.757988922028121e-05,
      "loss": 0.0018,
      "step": 852
    },
    {
      "epoch": 0.3270705521472393,
      "grad_norm": 29.81633758544922,
      "learning_rate": 3.755858542820622e-05,
      "loss": 0.8071,
      "step": 853
    },
    {
      "epoch": 0.32745398773006135,
      "grad_norm": 99.31094360351562,
      "learning_rate": 3.7537281636131236e-05,
      "loss": 1.1748,
      "step": 854
    },
    {
      "epoch": 0.32783742331288346,
      "grad_norm": 117.63552856445312,
      "learning_rate": 3.7515977844056245e-05,
      "loss": 2.9048,
      "step": 855
    },
    {
      "epoch": 0.3282208588957055,
      "grad_norm": 162.0067138671875,
      "learning_rate": 3.7494674051981254e-05,
      "loss": 1.9249,
      "step": 856
    },
    {
      "epoch": 0.3286042944785276,
      "grad_norm": 0.38351160287857056,
      "learning_rate": 3.7473370259906263e-05,
      "loss": 0.0024,
      "step": 857
    },
    {
      "epoch": 0.3289877300613497,
      "grad_norm": 2.895625114440918,
      "learning_rate": 3.745206646783127e-05,
      "loss": 0.012,
      "step": 858
    },
    {
      "epoch": 0.3293711656441718,
      "grad_norm": 1.1712878942489624,
      "learning_rate": 3.743076267575629e-05,
      "loss": 0.0037,
      "step": 859
    },
    {
      "epoch": 0.32975460122699385,
      "grad_norm": 7.501552581787109,
      "learning_rate": 3.74094588836813e-05,
      "loss": 0.046,
      "step": 860
    },
    {
      "epoch": 0.33013803680981596,
      "grad_norm": 0.43802642822265625,
      "learning_rate": 3.738815509160631e-05,
      "loss": 0.0023,
      "step": 861
    },
    {
      "epoch": 0.330521472392638,
      "grad_norm": 127.40283966064453,
      "learning_rate": 3.736685129953132e-05,
      "loss": 1.3009,
      "step": 862
    },
    {
      "epoch": 0.3309049079754601,
      "grad_norm": 17.892202377319336,
      "learning_rate": 3.734554750745633e-05,
      "loss": 0.4453,
      "step": 863
    },
    {
      "epoch": 0.3312883435582822,
      "grad_norm": 0.5644845962524414,
      "learning_rate": 3.732424371538134e-05,
      "loss": 0.0034,
      "step": 864
    },
    {
      "epoch": 0.3316717791411043,
      "grad_norm": 2.1184301376342773,
      "learning_rate": 3.730293992330635e-05,
      "loss": 0.0061,
      "step": 865
    },
    {
      "epoch": 0.3320552147239264,
      "grad_norm": 2.9578826427459717,
      "learning_rate": 3.728163613123136e-05,
      "loss": 0.0029,
      "step": 866
    },
    {
      "epoch": 0.33243865030674846,
      "grad_norm": 15.780271530151367,
      "learning_rate": 3.726033233915637e-05,
      "loss": 0.6094,
      "step": 867
    },
    {
      "epoch": 0.3328220858895706,
      "grad_norm": 108.54817962646484,
      "learning_rate": 3.7239028547081386e-05,
      "loss": 1.042,
      "step": 868
    },
    {
      "epoch": 0.33320552147239263,
      "grad_norm": 4.40326452255249,
      "learning_rate": 3.7217724755006395e-05,
      "loss": 0.0163,
      "step": 869
    },
    {
      "epoch": 0.33358895705521474,
      "grad_norm": 12.562338829040527,
      "learning_rate": 3.7196420962931405e-05,
      "loss": 0.1777,
      "step": 870
    },
    {
      "epoch": 0.3339723926380368,
      "grad_norm": 30.592666625976562,
      "learning_rate": 3.7175117170856414e-05,
      "loss": 0.5625,
      "step": 871
    },
    {
      "epoch": 0.3343558282208589,
      "grad_norm": 141.96115112304688,
      "learning_rate": 3.715381337878143e-05,
      "loss": 2.6057,
      "step": 872
    },
    {
      "epoch": 0.33473926380368096,
      "grad_norm": 95.01751708984375,
      "learning_rate": 3.713250958670644e-05,
      "loss": 1.1113,
      "step": 873
    },
    {
      "epoch": 0.3351226993865031,
      "grad_norm": 66.68183135986328,
      "learning_rate": 3.711120579463145e-05,
      "loss": 1.7306,
      "step": 874
    },
    {
      "epoch": 0.33550613496932513,
      "grad_norm": 248.12063598632812,
      "learning_rate": 3.708990200255646e-05,
      "loss": 1.7404,
      "step": 875
    },
    {
      "epoch": 0.33588957055214724,
      "grad_norm": 5.3473029136657715,
      "learning_rate": 3.706859821048147e-05,
      "loss": 0.0198,
      "step": 876
    },
    {
      "epoch": 0.3362730061349693,
      "grad_norm": 2.2328739166259766,
      "learning_rate": 3.7047294418406484e-05,
      "loss": 0.0116,
      "step": 877
    },
    {
      "epoch": 0.3366564417177914,
      "grad_norm": 22.21512222290039,
      "learning_rate": 3.702599062633149e-05,
      "loss": 0.4836,
      "step": 878
    },
    {
      "epoch": 0.3370398773006135,
      "grad_norm": 8.086519241333008,
      "learning_rate": 3.7004686834256495e-05,
      "loss": 0.0579,
      "step": 879
    },
    {
      "epoch": 0.3374233128834356,
      "grad_norm": 97.34036254882812,
      "learning_rate": 3.6983383042181505e-05,
      "loss": 1.1231,
      "step": 880
    },
    {
      "epoch": 0.3378067484662577,
      "grad_norm": 13.689335823059082,
      "learning_rate": 3.696207925010652e-05,
      "loss": 0.1815,
      "step": 881
    },
    {
      "epoch": 0.33819018404907975,
      "grad_norm": 0.08570973575115204,
      "learning_rate": 3.694077545803153e-05,
      "loss": 0.0014,
      "step": 882
    },
    {
      "epoch": 0.33857361963190186,
      "grad_norm": 14.040021896362305,
      "learning_rate": 3.691947166595654e-05,
      "loss": 0.0878,
      "step": 883
    },
    {
      "epoch": 0.3389570552147239,
      "grad_norm": 146.0187530517578,
      "learning_rate": 3.689816787388155e-05,
      "loss": 1.1436,
      "step": 884
    },
    {
      "epoch": 0.339340490797546,
      "grad_norm": 102.55371856689453,
      "learning_rate": 3.687686408180656e-05,
      "loss": 5.8774,
      "step": 885
    },
    {
      "epoch": 0.3397239263803681,
      "grad_norm": 8.883628845214844,
      "learning_rate": 3.6855560289731574e-05,
      "loss": 0.0355,
      "step": 886
    },
    {
      "epoch": 0.3401073619631902,
      "grad_norm": 199.91314697265625,
      "learning_rate": 3.6834256497656583e-05,
      "loss": 2.3519,
      "step": 887
    },
    {
      "epoch": 0.34049079754601225,
      "grad_norm": 74.7225112915039,
      "learning_rate": 3.681295270558159e-05,
      "loss": 0.9121,
      "step": 888
    },
    {
      "epoch": 0.34087423312883436,
      "grad_norm": 8.965007781982422,
      "learning_rate": 3.67916489135066e-05,
      "loss": 0.0455,
      "step": 889
    },
    {
      "epoch": 0.3412576687116564,
      "grad_norm": 21.721363067626953,
      "learning_rate": 3.677034512143162e-05,
      "loss": 0.6646,
      "step": 890
    },
    {
      "epoch": 0.3416411042944785,
      "grad_norm": 2.8286852836608887,
      "learning_rate": 3.674904132935663e-05,
      "loss": 0.0111,
      "step": 891
    },
    {
      "epoch": 0.34202453987730064,
      "grad_norm": 1.648725152015686,
      "learning_rate": 3.672773753728164e-05,
      "loss": 0.006,
      "step": 892
    },
    {
      "epoch": 0.3424079754601227,
      "grad_norm": 53.5578498840332,
      "learning_rate": 3.6706433745206646e-05,
      "loss": 0.8672,
      "step": 893
    },
    {
      "epoch": 0.3427914110429448,
      "grad_norm": 10.51111888885498,
      "learning_rate": 3.6685129953131655e-05,
      "loss": 0.0892,
      "step": 894
    },
    {
      "epoch": 0.34317484662576686,
      "grad_norm": 56.53624725341797,
      "learning_rate": 3.666382616105667e-05,
      "loss": 1.2852,
      "step": 895
    },
    {
      "epoch": 0.34355828220858897,
      "grad_norm": 20.492630004882812,
      "learning_rate": 3.664252236898168e-05,
      "loss": 0.5601,
      "step": 896
    },
    {
      "epoch": 0.34394171779141103,
      "grad_norm": 209.2409210205078,
      "learning_rate": 3.662121857690669e-05,
      "loss": 4.2934,
      "step": 897
    },
    {
      "epoch": 0.34432515337423314,
      "grad_norm": 48.036991119384766,
      "learning_rate": 3.65999147848317e-05,
      "loss": 1.4971,
      "step": 898
    },
    {
      "epoch": 0.3447085889570552,
      "grad_norm": 28.335689544677734,
      "learning_rate": 3.6578610992756715e-05,
      "loss": 0.54,
      "step": 899
    },
    {
      "epoch": 0.3450920245398773,
      "grad_norm": 16.36398696899414,
      "learning_rate": 3.6557307200681725e-05,
      "loss": 0.5205,
      "step": 900
    },
    {
      "epoch": 0.34547546012269936,
      "grad_norm": 13.94140625,
      "learning_rate": 3.6536003408606734e-05,
      "loss": 0.1626,
      "step": 901
    },
    {
      "epoch": 0.3458588957055215,
      "grad_norm": 5.451650142669678,
      "learning_rate": 3.6514699616531743e-05,
      "loss": 0.02,
      "step": 902
    },
    {
      "epoch": 0.3462423312883436,
      "grad_norm": 1.251380205154419,
      "learning_rate": 3.649339582445675e-05,
      "loss": 0.004,
      "step": 903
    },
    {
      "epoch": 0.34662576687116564,
      "grad_norm": 145.5535888671875,
      "learning_rate": 3.647209203238177e-05,
      "loss": 1.4522,
      "step": 904
    },
    {
      "epoch": 0.34700920245398775,
      "grad_norm": 59.523311614990234,
      "learning_rate": 3.645078824030678e-05,
      "loss": 1.4131,
      "step": 905
    },
    {
      "epoch": 0.3473926380368098,
      "grad_norm": 132.08468627929688,
      "learning_rate": 3.642948444823179e-05,
      "loss": 3.3542,
      "step": 906
    },
    {
      "epoch": 0.3477760736196319,
      "grad_norm": 165.15797424316406,
      "learning_rate": 3.64081806561568e-05,
      "loss": 2.2759,
      "step": 907
    },
    {
      "epoch": 0.348159509202454,
      "grad_norm": 1.4160902500152588,
      "learning_rate": 3.6386876864081806e-05,
      "loss": 0.0076,
      "step": 908
    },
    {
      "epoch": 0.3485429447852761,
      "grad_norm": 0.31354132294654846,
      "learning_rate": 3.636557307200682e-05,
      "loss": 0.0015,
      "step": 909
    },
    {
      "epoch": 0.34892638036809814,
      "grad_norm": 13.58237075805664,
      "learning_rate": 3.634426927993183e-05,
      "loss": 0.2168,
      "step": 910
    },
    {
      "epoch": 0.34930981595092025,
      "grad_norm": 45.58867263793945,
      "learning_rate": 3.632296548785684e-05,
      "loss": 0.5972,
      "step": 911
    },
    {
      "epoch": 0.3496932515337423,
      "grad_norm": 13.427237510681152,
      "learning_rate": 3.630166169578185e-05,
      "loss": 0.52,
      "step": 912
    },
    {
      "epoch": 0.3500766871165644,
      "grad_norm": 17.51856803894043,
      "learning_rate": 3.6280357903706866e-05,
      "loss": 0.5234,
      "step": 913
    },
    {
      "epoch": 0.3504601226993865,
      "grad_norm": 4.446659088134766,
      "learning_rate": 3.6259054111631875e-05,
      "loss": 0.0372,
      "step": 914
    },
    {
      "epoch": 0.3508435582822086,
      "grad_norm": 1.4216930866241455,
      "learning_rate": 3.6237750319556885e-05,
      "loss": 0.0073,
      "step": 915
    },
    {
      "epoch": 0.3512269938650307,
      "grad_norm": 77.42320251464844,
      "learning_rate": 3.6216446527481894e-05,
      "loss": 1.1709,
      "step": 916
    },
    {
      "epoch": 0.35161042944785276,
      "grad_norm": 97.61112213134766,
      "learning_rate": 3.6195142735406903e-05,
      "loss": 0.8571,
      "step": 917
    },
    {
      "epoch": 0.35199386503067487,
      "grad_norm": 135.04241943359375,
      "learning_rate": 3.617383894333192e-05,
      "loss": 2.0392,
      "step": 918
    },
    {
      "epoch": 0.3523773006134969,
      "grad_norm": 239.87155151367188,
      "learning_rate": 3.615253515125693e-05,
      "loss": 6.3501,
      "step": 919
    },
    {
      "epoch": 0.35276073619631904,
      "grad_norm": 4.7225165367126465,
      "learning_rate": 3.613123135918194e-05,
      "loss": 0.0113,
      "step": 920
    },
    {
      "epoch": 0.3531441717791411,
      "grad_norm": 288.6584167480469,
      "learning_rate": 3.610992756710695e-05,
      "loss": 1.7044,
      "step": 921
    },
    {
      "epoch": 0.3535276073619632,
      "grad_norm": 16.934629440307617,
      "learning_rate": 3.608862377503196e-05,
      "loss": 0.0699,
      "step": 922
    },
    {
      "epoch": 0.35391104294478526,
      "grad_norm": 19.55699920654297,
      "learning_rate": 3.606731998295697e-05,
      "loss": 0.5528,
      "step": 923
    },
    {
      "epoch": 0.35429447852760737,
      "grad_norm": 170.64454650878906,
      "learning_rate": 3.6046016190881975e-05,
      "loss": 5.6133,
      "step": 924
    },
    {
      "epoch": 0.3546779141104294,
      "grad_norm": 65.25788879394531,
      "learning_rate": 3.6024712398806985e-05,
      "loss": 1.1045,
      "step": 925
    },
    {
      "epoch": 0.35506134969325154,
      "grad_norm": 56.15605545043945,
      "learning_rate": 3.6003408606732e-05,
      "loss": 1.2637,
      "step": 926
    },
    {
      "epoch": 0.3554447852760736,
      "grad_norm": 18.860065460205078,
      "learning_rate": 3.598210481465701e-05,
      "loss": 0.0143,
      "step": 927
    },
    {
      "epoch": 0.3558282208588957,
      "grad_norm": 7.427710056304932,
      "learning_rate": 3.596080102258202e-05,
      "loss": 0.0345,
      "step": 928
    },
    {
      "epoch": 0.3562116564417178,
      "grad_norm": 34.25039291381836,
      "learning_rate": 3.593949723050703e-05,
      "loss": 0.8027,
      "step": 929
    },
    {
      "epoch": 0.3565950920245399,
      "grad_norm": 6.015954971313477,
      "learning_rate": 3.591819343843204e-05,
      "loss": 0.4458,
      "step": 930
    },
    {
      "epoch": 0.356978527607362,
      "grad_norm": 1.2891515493392944,
      "learning_rate": 3.5896889646357054e-05,
      "loss": 0.0023,
      "step": 931
    },
    {
      "epoch": 0.35736196319018404,
      "grad_norm": 253.78134155273438,
      "learning_rate": 3.5875585854282063e-05,
      "loss": 6.0605,
      "step": 932
    },
    {
      "epoch": 0.35774539877300615,
      "grad_norm": 2.394024610519409,
      "learning_rate": 3.585428206220707e-05,
      "loss": 0.0065,
      "step": 933
    },
    {
      "epoch": 0.3581288343558282,
      "grad_norm": 105.65200805664062,
      "learning_rate": 3.583297827013208e-05,
      "loss": 1.461,
      "step": 934
    },
    {
      "epoch": 0.3585122699386503,
      "grad_norm": 3.791978597640991,
      "learning_rate": 3.581167447805709e-05,
      "loss": 0.0257,
      "step": 935
    },
    {
      "epoch": 0.3588957055214724,
      "grad_norm": 175.56011962890625,
      "learning_rate": 3.579037068598211e-05,
      "loss": 2.4556,
      "step": 936
    },
    {
      "epoch": 0.3592791411042945,
      "grad_norm": 2.273514747619629,
      "learning_rate": 3.576906689390712e-05,
      "loss": 0.0097,
      "step": 937
    },
    {
      "epoch": 0.35966257668711654,
      "grad_norm": 30.347814559936523,
      "learning_rate": 3.5747763101832126e-05,
      "loss": 0.5523,
      "step": 938
    },
    {
      "epoch": 0.36004601226993865,
      "grad_norm": 48.91766357421875,
      "learning_rate": 3.5726459309757135e-05,
      "loss": 0.6304,
      "step": 939
    },
    {
      "epoch": 0.3604294478527607,
      "grad_norm": 12.900555610656738,
      "learning_rate": 3.570515551768215e-05,
      "loss": 0.5147,
      "step": 940
    },
    {
      "epoch": 0.3608128834355828,
      "grad_norm": 109.76214599609375,
      "learning_rate": 3.568385172560716e-05,
      "loss": 1.792,
      "step": 941
    },
    {
      "epoch": 0.36119631901840493,
      "grad_norm": 10.001334190368652,
      "learning_rate": 3.566254793353217e-05,
      "loss": 0.2402,
      "step": 942
    },
    {
      "epoch": 0.361579754601227,
      "grad_norm": 15.179523468017578,
      "learning_rate": 3.564124414145718e-05,
      "loss": 0.0724,
      "step": 943
    },
    {
      "epoch": 0.3619631901840491,
      "grad_norm": 3.511728048324585,
      "learning_rate": 3.561994034938219e-05,
      "loss": 0.0056,
      "step": 944
    },
    {
      "epoch": 0.36234662576687116,
      "grad_norm": 96.32710266113281,
      "learning_rate": 3.5598636557307205e-05,
      "loss": 6.5293,
      "step": 945
    },
    {
      "epoch": 0.36273006134969327,
      "grad_norm": 6.759649753570557,
      "learning_rate": 3.5577332765232214e-05,
      "loss": 0.0846,
      "step": 946
    },
    {
      "epoch": 0.3631134969325153,
      "grad_norm": 14.899292945861816,
      "learning_rate": 3.5556028973157223e-05,
      "loss": 0.104,
      "step": 947
    },
    {
      "epoch": 0.36349693251533743,
      "grad_norm": 8.704339027404785,
      "learning_rate": 3.553472518108223e-05,
      "loss": 0.0911,
      "step": 948
    },
    {
      "epoch": 0.3638803680981595,
      "grad_norm": 5.973624229431152,
      "learning_rate": 3.551342138900724e-05,
      "loss": 0.0193,
      "step": 949
    },
    {
      "epoch": 0.3642638036809816,
      "grad_norm": 178.04185485839844,
      "learning_rate": 3.549211759693226e-05,
      "loss": 1.0596,
      "step": 950
    },
    {
      "epoch": 0.36464723926380366,
      "grad_norm": 126.13075256347656,
      "learning_rate": 3.547081380485727e-05,
      "loss": 1.6575,
      "step": 951
    },
    {
      "epoch": 0.36503067484662577,
      "grad_norm": 101.19743347167969,
      "learning_rate": 3.544951001278228e-05,
      "loss": 5.752,
      "step": 952
    },
    {
      "epoch": 0.3654141104294479,
      "grad_norm": 11.74567985534668,
      "learning_rate": 3.5428206220707286e-05,
      "loss": 0.2307,
      "step": 953
    },
    {
      "epoch": 0.36579754601226994,
      "grad_norm": 24.42420196533203,
      "learning_rate": 3.54069024286323e-05,
      "loss": 0.1229,
      "step": 954
    },
    {
      "epoch": 0.36618098159509205,
      "grad_norm": 113.34022521972656,
      "learning_rate": 3.538559863655731e-05,
      "loss": 1.3563,
      "step": 955
    },
    {
      "epoch": 0.3665644171779141,
      "grad_norm": 16.56443214416504,
      "learning_rate": 3.536429484448232e-05,
      "loss": 0.4824,
      "step": 956
    },
    {
      "epoch": 0.3669478527607362,
      "grad_norm": 4.898117542266846,
      "learning_rate": 3.534299105240733e-05,
      "loss": 0.0257,
      "step": 957
    },
    {
      "epoch": 0.36733128834355827,
      "grad_norm": 11.129241943359375,
      "learning_rate": 3.532168726033234e-05,
      "loss": 0.0542,
      "step": 958
    },
    {
      "epoch": 0.3677147239263804,
      "grad_norm": 31.687559127807617,
      "learning_rate": 3.5300383468257355e-05,
      "loss": 0.6754,
      "step": 959
    },
    {
      "epoch": 0.36809815950920244,
      "grad_norm": 1224.97607421875,
      "learning_rate": 3.5279079676182365e-05,
      "loss": 5.4961,
      "step": 960
    },
    {
      "epoch": 0.36848159509202455,
      "grad_norm": 19.937158584594727,
      "learning_rate": 3.5257775884107374e-05,
      "loss": 0.5596,
      "step": 961
    },
    {
      "epoch": 0.3688650306748466,
      "grad_norm": 15.832524299621582,
      "learning_rate": 3.5236472092032383e-05,
      "loss": 0.09,
      "step": 962
    },
    {
      "epoch": 0.3692484662576687,
      "grad_norm": 139.12794494628906,
      "learning_rate": 3.52151682999574e-05,
      "loss": 0.9937,
      "step": 963
    },
    {
      "epoch": 0.3696319018404908,
      "grad_norm": 218.58534240722656,
      "learning_rate": 3.519386450788241e-05,
      "loss": 2.086,
      "step": 964
    },
    {
      "epoch": 0.3700153374233129,
      "grad_norm": 8.590350151062012,
      "learning_rate": 3.517256071580742e-05,
      "loss": 0.1904,
      "step": 965
    },
    {
      "epoch": 0.370398773006135,
      "grad_norm": 6.688065052032471,
      "learning_rate": 3.515125692373243e-05,
      "loss": 0.0427,
      "step": 966
    },
    {
      "epoch": 0.37078220858895705,
      "grad_norm": 24.327075958251953,
      "learning_rate": 3.512995313165744e-05,
      "loss": 0.645,
      "step": 967
    },
    {
      "epoch": 0.37116564417177916,
      "grad_norm": 2.1375648975372314,
      "learning_rate": 3.510864933958245e-05,
      "loss": 0.0069,
      "step": 968
    },
    {
      "epoch": 0.3715490797546012,
      "grad_norm": 1.738763689994812,
      "learning_rate": 3.5087345547507455e-05,
      "loss": 0.007,
      "step": 969
    },
    {
      "epoch": 0.37193251533742333,
      "grad_norm": 4.914362907409668,
      "learning_rate": 3.5066041755432465e-05,
      "loss": 0.4139,
      "step": 970
    },
    {
      "epoch": 0.3723159509202454,
      "grad_norm": 13.827409744262695,
      "learning_rate": 3.5044737963357474e-05,
      "loss": 0.5098,
      "step": 971
    },
    {
      "epoch": 0.3726993865030675,
      "grad_norm": 1.2973281145095825,
      "learning_rate": 3.502343417128249e-05,
      "loss": 0.0035,
      "step": 972
    },
    {
      "epoch": 0.37308282208588955,
      "grad_norm": 138.20770263671875,
      "learning_rate": 3.50021303792075e-05,
      "loss": 0.9175,
      "step": 973
    },
    {
      "epoch": 0.37346625766871167,
      "grad_norm": 0.4478858709335327,
      "learning_rate": 3.498082658713251e-05,
      "loss": 0.0031,
      "step": 974
    },
    {
      "epoch": 0.3738496932515337,
      "grad_norm": 0.9873825311660767,
      "learning_rate": 3.495952279505752e-05,
      "loss": 0.0056,
      "step": 975
    },
    {
      "epoch": 0.37423312883435583,
      "grad_norm": 43.3076171875,
      "learning_rate": 3.493821900298253e-05,
      "loss": 0.7267,
      "step": 976
    },
    {
      "epoch": 0.3746165644171779,
      "grad_norm": 1.7374697923660278,
      "learning_rate": 3.4916915210907543e-05,
      "loss": 0.0115,
      "step": 977
    },
    {
      "epoch": 0.375,
      "grad_norm": 28.69477081298828,
      "learning_rate": 3.489561141883255e-05,
      "loss": 0.5532,
      "step": 978
    },
    {
      "epoch": 0.3753834355828221,
      "grad_norm": 10.242043495178223,
      "learning_rate": 3.487430762675756e-05,
      "loss": 0.4641,
      "step": 979
    },
    {
      "epoch": 0.37576687116564417,
      "grad_norm": 15.66881275177002,
      "learning_rate": 3.485300383468257e-05,
      "loss": 0.1858,
      "step": 980
    },
    {
      "epoch": 0.3761503067484663,
      "grad_norm": 9.314672470092773,
      "learning_rate": 3.483170004260759e-05,
      "loss": 0.051,
      "step": 981
    },
    {
      "epoch": 0.37653374233128833,
      "grad_norm": 9.923254013061523,
      "learning_rate": 3.48103962505326e-05,
      "loss": 0.0828,
      "step": 982
    },
    {
      "epoch": 0.37691717791411045,
      "grad_norm": 1.2475645542144775,
      "learning_rate": 3.4789092458457606e-05,
      "loss": 0.0059,
      "step": 983
    },
    {
      "epoch": 0.3773006134969325,
      "grad_norm": 60.9388313293457,
      "learning_rate": 3.4767788666382615e-05,
      "loss": 1.0782,
      "step": 984
    },
    {
      "epoch": 0.3776840490797546,
      "grad_norm": 43.85203552246094,
      "learning_rate": 3.4746484874307625e-05,
      "loss": 0.9326,
      "step": 985
    },
    {
      "epoch": 0.37806748466257667,
      "grad_norm": 7.825187683105469,
      "learning_rate": 3.472518108223264e-05,
      "loss": 0.4473,
      "step": 986
    },
    {
      "epoch": 0.3784509202453988,
      "grad_norm": 3.5926787853240967,
      "learning_rate": 3.470387729015765e-05,
      "loss": 0.0061,
      "step": 987
    },
    {
      "epoch": 0.37883435582822084,
      "grad_norm": 2.565843105316162,
      "learning_rate": 3.468257349808266e-05,
      "loss": 0.016,
      "step": 988
    },
    {
      "epoch": 0.37921779141104295,
      "grad_norm": 11.560431480407715,
      "learning_rate": 3.466126970600767e-05,
      "loss": 0.0822,
      "step": 989
    },
    {
      "epoch": 0.379601226993865,
      "grad_norm": 12.063653945922852,
      "learning_rate": 3.4639965913932685e-05,
      "loss": 0.0968,
      "step": 990
    },
    {
      "epoch": 0.3799846625766871,
      "grad_norm": 5.371000289916992,
      "learning_rate": 3.4618662121857694e-05,
      "loss": 0.4231,
      "step": 991
    },
    {
      "epoch": 0.3803680981595092,
      "grad_norm": 0.4216524660587311,
      "learning_rate": 3.4597358329782703e-05,
      "loss": 0.0027,
      "step": 992
    },
    {
      "epoch": 0.3807515337423313,
      "grad_norm": 0.5369511246681213,
      "learning_rate": 3.457605453770771e-05,
      "loss": 0.0026,
      "step": 993
    },
    {
      "epoch": 0.3811349693251534,
      "grad_norm": 66.50501251220703,
      "learning_rate": 3.455475074563272e-05,
      "loss": 0.9512,
      "step": 994
    },
    {
      "epoch": 0.38151840490797545,
      "grad_norm": 31.421464920043945,
      "learning_rate": 3.453344695355774e-05,
      "loss": 0.645,
      "step": 995
    },
    {
      "epoch": 0.38190184049079756,
      "grad_norm": 57.370025634765625,
      "learning_rate": 3.451214316148275e-05,
      "loss": 0.6948,
      "step": 996
    },
    {
      "epoch": 0.3822852760736196,
      "grad_norm": 116.8099594116211,
      "learning_rate": 3.449083936940776e-05,
      "loss": 5.9358,
      "step": 997
    },
    {
      "epoch": 0.38266871165644173,
      "grad_norm": 58.960086822509766,
      "learning_rate": 3.4469535577332766e-05,
      "loss": 0.7402,
      "step": 998
    },
    {
      "epoch": 0.3830521472392638,
      "grad_norm": 2.194159984588623,
      "learning_rate": 3.4448231785257775e-05,
      "loss": 0.0069,
      "step": 999
    },
    {
      "epoch": 0.3834355828220859,
      "grad_norm": 1.4175691604614258,
      "learning_rate": 3.442692799318279e-05,
      "loss": 0.005,
      "step": 1000
    },
    {
      "epoch": 0.38381901840490795,
      "grad_norm": 1.2209025621414185,
      "learning_rate": 3.44056242011078e-05,
      "loss": 0.0048,
      "step": 1001
    },
    {
      "epoch": 0.38420245398773006,
      "grad_norm": 8.037437438964844,
      "learning_rate": 3.438432040903281e-05,
      "loss": 0.0942,
      "step": 1002
    },
    {
      "epoch": 0.3845858895705521,
      "grad_norm": 0.16226428747177124,
      "learning_rate": 3.436301661695782e-05,
      "loss": 0.0023,
      "step": 1003
    },
    {
      "epoch": 0.38496932515337423,
      "grad_norm": 10.410154342651367,
      "learning_rate": 3.4341712824882835e-05,
      "loss": 0.037,
      "step": 1004
    },
    {
      "epoch": 0.38535276073619634,
      "grad_norm": 9.644920349121094,
      "learning_rate": 3.4320409032807845e-05,
      "loss": 0.0425,
      "step": 1005
    },
    {
      "epoch": 0.3857361963190184,
      "grad_norm": 0.09576055407524109,
      "learning_rate": 3.4299105240732854e-05,
      "loss": 0.0012,
      "step": 1006
    },
    {
      "epoch": 0.3861196319018405,
      "grad_norm": 1.5106265544891357,
      "learning_rate": 3.4277801448657863e-05,
      "loss": 0.0071,
      "step": 1007
    },
    {
      "epoch": 0.38650306748466257,
      "grad_norm": 5.6399126052856445,
      "learning_rate": 3.425649765658287e-05,
      "loss": 0.0083,
      "step": 1008
    },
    {
      "epoch": 0.3868865030674847,
      "grad_norm": 0.7495969533920288,
      "learning_rate": 3.423519386450789e-05,
      "loss": 0.0038,
      "step": 1009
    },
    {
      "epoch": 0.38726993865030673,
      "grad_norm": 107.00861358642578,
      "learning_rate": 3.42138900724329e-05,
      "loss": 2.4652,
      "step": 1010
    },
    {
      "epoch": 0.38765337423312884,
      "grad_norm": 4.924862861633301,
      "learning_rate": 3.419258628035791e-05,
      "loss": 0.0218,
      "step": 1011
    },
    {
      "epoch": 0.3880368098159509,
      "grad_norm": 10.599689483642578,
      "learning_rate": 3.417128248828292e-05,
      "loss": 0.4526,
      "step": 1012
    },
    {
      "epoch": 0.388420245398773,
      "grad_norm": 54.0268669128418,
      "learning_rate": 3.4149978696207926e-05,
      "loss": 0.8702,
      "step": 1013
    },
    {
      "epoch": 0.38880368098159507,
      "grad_norm": 78.01480865478516,
      "learning_rate": 3.4128674904132935e-05,
      "loss": 1.5245,
      "step": 1014
    },
    {
      "epoch": 0.3891871165644172,
      "grad_norm": 10.43925952911377,
      "learning_rate": 3.4107371112057945e-05,
      "loss": 0.0676,
      "step": 1015
    },
    {
      "epoch": 0.3895705521472393,
      "grad_norm": 2.993187427520752,
      "learning_rate": 3.4086067319982954e-05,
      "loss": 0.0068,
      "step": 1016
    },
    {
      "epoch": 0.38995398773006135,
      "grad_norm": 0.821253776550293,
      "learning_rate": 3.406476352790797e-05,
      "loss": 0.0037,
      "step": 1017
    },
    {
      "epoch": 0.39033742331288346,
      "grad_norm": 87.67337799072266,
      "learning_rate": 3.404345973583298e-05,
      "loss": 6.043,
      "step": 1018
    },
    {
      "epoch": 0.3907208588957055,
      "grad_norm": 142.62059020996094,
      "learning_rate": 3.402215594375799e-05,
      "loss": 1.3897,
      "step": 1019
    },
    {
      "epoch": 0.3911042944785276,
      "grad_norm": 209.81443786621094,
      "learning_rate": 3.4000852151683e-05,
      "loss": 3.2858,
      "step": 1020
    },
    {
      "epoch": 0.3914877300613497,
      "grad_norm": 30.382448196411133,
      "learning_rate": 3.397954835960801e-05,
      "loss": 0.7295,
      "step": 1021
    },
    {
      "epoch": 0.3918711656441718,
      "grad_norm": 2.5746898651123047,
      "learning_rate": 3.3958244567533023e-05,
      "loss": 0.0051,
      "step": 1022
    },
    {
      "epoch": 0.39225460122699385,
      "grad_norm": 28.87703514099121,
      "learning_rate": 3.393694077545803e-05,
      "loss": 0.731,
      "step": 1023
    },
    {
      "epoch": 0.39263803680981596,
      "grad_norm": 165.62081909179688,
      "learning_rate": 3.391563698338304e-05,
      "loss": 2.7097,
      "step": 1024
    },
    {
      "epoch": 0.393021472392638,
      "grad_norm": 0.3038354814052582,
      "learning_rate": 3.389433319130805e-05,
      "loss": 0.0019,
      "step": 1025
    },
    {
      "epoch": 0.3934049079754601,
      "grad_norm": 1.9357569217681885,
      "learning_rate": 3.387302939923306e-05,
      "loss": 0.0062,
      "step": 1026
    },
    {
      "epoch": 0.3937883435582822,
      "grad_norm": 5.303816795349121,
      "learning_rate": 3.385172560715808e-05,
      "loss": 0.0374,
      "step": 1027
    },
    {
      "epoch": 0.3941717791411043,
      "grad_norm": 34.76011276245117,
      "learning_rate": 3.3830421815083086e-05,
      "loss": 0.6807,
      "step": 1028
    },
    {
      "epoch": 0.3945552147239264,
      "grad_norm": 7.539120197296143,
      "learning_rate": 3.3809118023008095e-05,
      "loss": 0.0786,
      "step": 1029
    },
    {
      "epoch": 0.39493865030674846,
      "grad_norm": 6.309747219085693,
      "learning_rate": 3.3787814230933105e-05,
      "loss": 0.0261,
      "step": 1030
    },
    {
      "epoch": 0.3953220858895706,
      "grad_norm": 106.8431396484375,
      "learning_rate": 3.376651043885812e-05,
      "loss": 1.046,
      "step": 1031
    },
    {
      "epoch": 0.39570552147239263,
      "grad_norm": 8.58929443359375,
      "learning_rate": 3.374520664678313e-05,
      "loss": 0.467,
      "step": 1032
    },
    {
      "epoch": 0.39608895705521474,
      "grad_norm": 46.64529800415039,
      "learning_rate": 3.372390285470814e-05,
      "loss": 0.645,
      "step": 1033
    },
    {
      "epoch": 0.3964723926380368,
      "grad_norm": 138.46295166015625,
      "learning_rate": 3.370259906263315e-05,
      "loss": 5.7383,
      "step": 1034
    },
    {
      "epoch": 0.3968558282208589,
      "grad_norm": 1.559178113937378,
      "learning_rate": 3.368129527055816e-05,
      "loss": 0.0107,
      "step": 1035
    },
    {
      "epoch": 0.39723926380368096,
      "grad_norm": 8.139238357543945,
      "learning_rate": 3.3659991478483174e-05,
      "loss": 0.0212,
      "step": 1036
    },
    {
      "epoch": 0.3976226993865031,
      "grad_norm": 69.52721405029297,
      "learning_rate": 3.3638687686408183e-05,
      "loss": 0.6763,
      "step": 1037
    },
    {
      "epoch": 0.39800613496932513,
      "grad_norm": 80.78510284423828,
      "learning_rate": 3.361738389433319e-05,
      "loss": 0.6592,
      "step": 1038
    },
    {
      "epoch": 0.39838957055214724,
      "grad_norm": 1.6643552780151367,
      "learning_rate": 3.35960801022582e-05,
      "loss": 0.0042,
      "step": 1039
    },
    {
      "epoch": 0.3987730061349693,
      "grad_norm": 35.950103759765625,
      "learning_rate": 3.357477631018321e-05,
      "loss": 0.6318,
      "step": 1040
    },
    {
      "epoch": 0.3991564417177914,
      "grad_norm": 5.243088722229004,
      "learning_rate": 3.355347251810823e-05,
      "loss": 0.0222,
      "step": 1041
    },
    {
      "epoch": 0.3995398773006135,
      "grad_norm": 1.2611973285675049,
      "learning_rate": 3.353216872603324e-05,
      "loss": 0.004,
      "step": 1042
    },
    {
      "epoch": 0.3999233128834356,
      "grad_norm": 7.315226078033447,
      "learning_rate": 3.3510864933958246e-05,
      "loss": 0.173,
      "step": 1043
    },
    {
      "epoch": 0.4003067484662577,
      "grad_norm": 0.6392849683761597,
      "learning_rate": 3.3489561141883255e-05,
      "loss": 0.0028,
      "step": 1044
    },
    {
      "epoch": 0.40069018404907975,
      "grad_norm": 142.23976135253906,
      "learning_rate": 3.346825734980827e-05,
      "loss": 5.9639,
      "step": 1045
    },
    {
      "epoch": 0.40107361963190186,
      "grad_norm": 27.186695098876953,
      "learning_rate": 3.344695355773328e-05,
      "loss": 0.2051,
      "step": 1046
    },
    {
      "epoch": 0.4014570552147239,
      "grad_norm": 108.6761474609375,
      "learning_rate": 3.342564976565829e-05,
      "loss": 2.6102,
      "step": 1047
    },
    {
      "epoch": 0.401840490797546,
      "grad_norm": 40.983646392822266,
      "learning_rate": 3.34043459735833e-05,
      "loss": 0.6411,
      "step": 1048
    },
    {
      "epoch": 0.4022239263803681,
      "grad_norm": 0.22776922583580017,
      "learning_rate": 3.338304218150831e-05,
      "loss": 0.0019,
      "step": 1049
    },
    {
      "epoch": 0.4026073619631902,
      "grad_norm": 2.4507944583892822,
      "learning_rate": 3.3361738389433325e-05,
      "loss": 0.0061,
      "step": 1050
    },
    {
      "epoch": 0.40299079754601225,
      "grad_norm": 13.551019668579102,
      "learning_rate": 3.3340434597358334e-05,
      "loss": 0.0541,
      "step": 1051
    },
    {
      "epoch": 0.40337423312883436,
      "grad_norm": 1.0843762159347534,
      "learning_rate": 3.3319130805283343e-05,
      "loss": 0.0046,
      "step": 1052
    },
    {
      "epoch": 0.4037576687116564,
      "grad_norm": 79.6090316772461,
      "learning_rate": 3.329782701320835e-05,
      "loss": 1.67,
      "step": 1053
    },
    {
      "epoch": 0.4041411042944785,
      "grad_norm": 221.82077026367188,
      "learning_rate": 3.327652322113337e-05,
      "loss": 2.742,
      "step": 1054
    },
    {
      "epoch": 0.40452453987730064,
      "grad_norm": 15.52837085723877,
      "learning_rate": 3.325521942905838e-05,
      "loss": 0.2072,
      "step": 1055
    },
    {
      "epoch": 0.4049079754601227,
      "grad_norm": 21.680545806884766,
      "learning_rate": 3.323391563698339e-05,
      "loss": 0.5474,
      "step": 1056
    },
    {
      "epoch": 0.4052914110429448,
      "grad_norm": 91.9194107055664,
      "learning_rate": 3.32126118449084e-05,
      "loss": 2.147,
      "step": 1057
    },
    {
      "epoch": 0.40567484662576686,
      "grad_norm": 6.127416610717773,
      "learning_rate": 3.3191308052833406e-05,
      "loss": 0.0274,
      "step": 1058
    },
    {
      "epoch": 0.40605828220858897,
      "grad_norm": 96.55035400390625,
      "learning_rate": 3.3170004260758415e-05,
      "loss": 5.8691,
      "step": 1059
    },
    {
      "epoch": 0.40644171779141103,
      "grad_norm": 129.17575073242188,
      "learning_rate": 3.3148700468683425e-05,
      "loss": 1.4494,
      "step": 1060
    },
    {
      "epoch": 0.40682515337423314,
      "grad_norm": 14.602612495422363,
      "learning_rate": 3.3127396676608434e-05,
      "loss": 0.1394,
      "step": 1061
    },
    {
      "epoch": 0.4072085889570552,
      "grad_norm": 165.3533172607422,
      "learning_rate": 3.310609288453344e-05,
      "loss": 5.4185,
      "step": 1062
    },
    {
      "epoch": 0.4075920245398773,
      "grad_norm": 114.32776641845703,
      "learning_rate": 3.308478909245846e-05,
      "loss": 0.9317,
      "step": 1063
    },
    {
      "epoch": 0.40797546012269936,
      "grad_norm": 10.328788757324219,
      "learning_rate": 3.306348530038347e-05,
      "loss": 0.0949,
      "step": 1064
    },
    {
      "epoch": 0.4083588957055215,
      "grad_norm": 73.43679809570312,
      "learning_rate": 3.304218150830848e-05,
      "loss": 1.0195,
      "step": 1065
    },
    {
      "epoch": 0.4087423312883436,
      "grad_norm": 94.94047546386719,
      "learning_rate": 3.302087771623349e-05,
      "loss": 5.0684,
      "step": 1066
    },
    {
      "epoch": 0.40912576687116564,
      "grad_norm": 34.3326301574707,
      "learning_rate": 3.2999573924158503e-05,
      "loss": 0.0419,
      "step": 1067
    },
    {
      "epoch": 0.40950920245398775,
      "grad_norm": 2.6018779277801514,
      "learning_rate": 3.297827013208351e-05,
      "loss": 0.0124,
      "step": 1068
    },
    {
      "epoch": 0.4098926380368098,
      "grad_norm": 1.7786577939987183,
      "learning_rate": 3.295696634000852e-05,
      "loss": 0.0086,
      "step": 1069
    },
    {
      "epoch": 0.4102760736196319,
      "grad_norm": 5.96002197265625,
      "learning_rate": 3.293566254793353e-05,
      "loss": 0.0331,
      "step": 1070
    },
    {
      "epoch": 0.410659509202454,
      "grad_norm": 9.824308395385742,
      "learning_rate": 3.291435875585854e-05,
      "loss": 0.4092,
      "step": 1071
    },
    {
      "epoch": 0.4110429447852761,
      "grad_norm": 1.0867952108383179,
      "learning_rate": 3.289305496378356e-05,
      "loss": 0.0044,
      "step": 1072
    },
    {
      "epoch": 0.41142638036809814,
      "grad_norm": 141.3063507080078,
      "learning_rate": 3.2871751171708566e-05,
      "loss": 2.0315,
      "step": 1073
    },
    {
      "epoch": 0.41180981595092025,
      "grad_norm": 59.421531677246094,
      "learning_rate": 3.2850447379633575e-05,
      "loss": 1.2481,
      "step": 1074
    },
    {
      "epoch": 0.4121932515337423,
      "grad_norm": 414.6046142578125,
      "learning_rate": 3.2829143587558585e-05,
      "loss": 3.2283,
      "step": 1075
    },
    {
      "epoch": 0.4125766871165644,
      "grad_norm": 8.04227352142334,
      "learning_rate": 3.2807839795483594e-05,
      "loss": 0.0501,
      "step": 1076
    },
    {
      "epoch": 0.4129601226993865,
      "grad_norm": 2.519871711730957,
      "learning_rate": 3.278653600340861e-05,
      "loss": 0.0129,
      "step": 1077
    },
    {
      "epoch": 0.4133435582822086,
      "grad_norm": 9.231657981872559,
      "learning_rate": 3.276523221133362e-05,
      "loss": 0.1472,
      "step": 1078
    },
    {
      "epoch": 0.4137269938650307,
      "grad_norm": 0.6013676524162292,
      "learning_rate": 3.274392841925863e-05,
      "loss": 0.0028,
      "step": 1079
    },
    {
      "epoch": 0.41411042944785276,
      "grad_norm": 9.617133140563965,
      "learning_rate": 3.272262462718364e-05,
      "loss": 0.033,
      "step": 1080
    },
    {
      "epoch": 0.41449386503067487,
      "grad_norm": 0.8977956771850586,
      "learning_rate": 3.2701320835108654e-05,
      "loss": 0.0046,
      "step": 1081
    },
    {
      "epoch": 0.4148773006134969,
      "grad_norm": 90.50689697265625,
      "learning_rate": 3.2680017043033663e-05,
      "loss": 5.4648,
      "step": 1082
    },
    {
      "epoch": 0.41526073619631904,
      "grad_norm": 0.3941224217414856,
      "learning_rate": 3.265871325095867e-05,
      "loss": 0.0039,
      "step": 1083
    },
    {
      "epoch": 0.4156441717791411,
      "grad_norm": 73.57256317138672,
      "learning_rate": 3.263740945888368e-05,
      "loss": 6.416,
      "step": 1084
    },
    {
      "epoch": 0.4160276073619632,
      "grad_norm": 59.00543975830078,
      "learning_rate": 3.261610566680869e-05,
      "loss": 0.7217,
      "step": 1085
    },
    {
      "epoch": 0.41641104294478526,
      "grad_norm": 3.38399600982666,
      "learning_rate": 3.259480187473371e-05,
      "loss": 0.0067,
      "step": 1086
    },
    {
      "epoch": 0.41679447852760737,
      "grad_norm": 46.32831573486328,
      "learning_rate": 3.257349808265872e-05,
      "loss": 0.6934,
      "step": 1087
    },
    {
      "epoch": 0.4171779141104294,
      "grad_norm": 146.98757934570312,
      "learning_rate": 3.2552194290583726e-05,
      "loss": 5.5771,
      "step": 1088
    },
    {
      "epoch": 0.41756134969325154,
      "grad_norm": 0.4296836256980896,
      "learning_rate": 3.2530890498508735e-05,
      "loss": 0.0025,
      "step": 1089
    },
    {
      "epoch": 0.4179447852760736,
      "grad_norm": 111.8563003540039,
      "learning_rate": 3.2509586706433745e-05,
      "loss": 5.4316,
      "step": 1090
    },
    {
      "epoch": 0.4183282208588957,
      "grad_norm": 6.3267083168029785,
      "learning_rate": 3.248828291435876e-05,
      "loss": 0.4231,
      "step": 1091
    },
    {
      "epoch": 0.4187116564417178,
      "grad_norm": 7.0891194343566895,
      "learning_rate": 3.246697912228377e-05,
      "loss": 0.0222,
      "step": 1092
    },
    {
      "epoch": 0.4190950920245399,
      "grad_norm": 1.2289345264434814,
      "learning_rate": 3.244567533020878e-05,
      "loss": 0.0039,
      "step": 1093
    },
    {
      "epoch": 0.419478527607362,
      "grad_norm": 7.595825672149658,
      "learning_rate": 3.242437153813379e-05,
      "loss": 0.0247,
      "step": 1094
    },
    {
      "epoch": 0.41986196319018404,
      "grad_norm": 3.3048288822174072,
      "learning_rate": 3.2403067746058805e-05,
      "loss": 0.0148,
      "step": 1095
    },
    {
      "epoch": 0.42024539877300615,
      "grad_norm": 79.68567657470703,
      "learning_rate": 3.2381763953983814e-05,
      "loss": 6.0049,
      "step": 1096
    },
    {
      "epoch": 0.4206288343558282,
      "grad_norm": 15.561921119689941,
      "learning_rate": 3.236046016190882e-05,
      "loss": 0.1219,
      "step": 1097
    },
    {
      "epoch": 0.4210122699386503,
      "grad_norm": 76.29059600830078,
      "learning_rate": 3.233915636983383e-05,
      "loss": 1.2828,
      "step": 1098
    },
    {
      "epoch": 0.4213957055214724,
      "grad_norm": 48.52256774902344,
      "learning_rate": 3.231785257775884e-05,
      "loss": 0.5698,
      "step": 1099
    },
    {
      "epoch": 0.4217791411042945,
      "grad_norm": 1.1661450862884521,
      "learning_rate": 3.229654878568386e-05,
      "loss": 0.0048,
      "step": 1100
    },
    {
      "epoch": 0.42216257668711654,
      "grad_norm": 0.9635507464408875,
      "learning_rate": 3.227524499360887e-05,
      "loss": 0.0042,
      "step": 1101
    },
    {
      "epoch": 0.42254601226993865,
      "grad_norm": 6.463715553283691,
      "learning_rate": 3.225394120153388e-05,
      "loss": 0.0227,
      "step": 1102
    },
    {
      "epoch": 0.4229294478527607,
      "grad_norm": 60.45621871948242,
      "learning_rate": 3.2232637409458886e-05,
      "loss": 0.8892,
      "step": 1103
    },
    {
      "epoch": 0.4233128834355828,
      "grad_norm": 2.652829647064209,
      "learning_rate": 3.2211333617383895e-05,
      "loss": 0.0126,
      "step": 1104
    },
    {
      "epoch": 0.42369631901840493,
      "grad_norm": 16.4697322845459,
      "learning_rate": 3.2190029825308905e-05,
      "loss": 0.1293,
      "step": 1105
    },
    {
      "epoch": 0.424079754601227,
      "grad_norm": 0.6932334303855896,
      "learning_rate": 3.2168726033233914e-05,
      "loss": 0.0022,
      "step": 1106
    },
    {
      "epoch": 0.4244631901840491,
      "grad_norm": 3.634219169616699,
      "learning_rate": 3.214742224115892e-05,
      "loss": 0.0057,
      "step": 1107
    },
    {
      "epoch": 0.42484662576687116,
      "grad_norm": 6.787144184112549,
      "learning_rate": 3.212611844908394e-05,
      "loss": 0.4788,
      "step": 1108
    },
    {
      "epoch": 0.42523006134969327,
      "grad_norm": 2.133728504180908,
      "learning_rate": 3.210481465700895e-05,
      "loss": 0.0099,
      "step": 1109
    },
    {
      "epoch": 0.4256134969325153,
      "grad_norm": 0.46014270186424255,
      "learning_rate": 3.208351086493396e-05,
      "loss": 0.0027,
      "step": 1110
    },
    {
      "epoch": 0.42599693251533743,
      "grad_norm": 7.713406085968018,
      "learning_rate": 3.206220707285897e-05,
      "loss": 0.032,
      "step": 1111
    },
    {
      "epoch": 0.4263803680981595,
      "grad_norm": 123.03546142578125,
      "learning_rate": 3.2040903280783977e-05,
      "loss": 0.9968,
      "step": 1112
    },
    {
      "epoch": 0.4267638036809816,
      "grad_norm": 64.26725006103516,
      "learning_rate": 3.201959948870899e-05,
      "loss": 1.3106,
      "step": 1113
    },
    {
      "epoch": 0.42714723926380366,
      "grad_norm": 0.5695100426673889,
      "learning_rate": 3.1998295696634e-05,
      "loss": 0.0039,
      "step": 1114
    },
    {
      "epoch": 0.42753067484662577,
      "grad_norm": 12.705592155456543,
      "learning_rate": 3.197699190455901e-05,
      "loss": 0.5059,
      "step": 1115
    },
    {
      "epoch": 0.4279141104294479,
      "grad_norm": 154.61781311035156,
      "learning_rate": 3.195568811248402e-05,
      "loss": 5.0117,
      "step": 1116
    },
    {
      "epoch": 0.42829754601226994,
      "grad_norm": 0.260300874710083,
      "learning_rate": 3.193438432040903e-05,
      "loss": 0.0019,
      "step": 1117
    },
    {
      "epoch": 0.42868098159509205,
      "grad_norm": 1.714156985282898,
      "learning_rate": 3.1913080528334046e-05,
      "loss": 0.0072,
      "step": 1118
    },
    {
      "epoch": 0.4290644171779141,
      "grad_norm": 7.342320442199707,
      "learning_rate": 3.1891776736259055e-05,
      "loss": 0.0997,
      "step": 1119
    },
    {
      "epoch": 0.4294478527607362,
      "grad_norm": 31.19200897216797,
      "learning_rate": 3.1870472944184065e-05,
      "loss": 0.6641,
      "step": 1120
    },
    {
      "epoch": 0.42983128834355827,
      "grad_norm": 2.5018272399902344,
      "learning_rate": 3.1849169152109074e-05,
      "loss": 0.0097,
      "step": 1121
    },
    {
      "epoch": 0.4302147239263804,
      "grad_norm": 79.78598022460938,
      "learning_rate": 3.182786536003409e-05,
      "loss": 1.3058,
      "step": 1122
    },
    {
      "epoch": 0.43059815950920244,
      "grad_norm": 0.393129825592041,
      "learning_rate": 3.18065615679591e-05,
      "loss": 0.0028,
      "step": 1123
    },
    {
      "epoch": 0.43098159509202455,
      "grad_norm": 4.799597263336182,
      "learning_rate": 3.178525777588411e-05,
      "loss": 0.0454,
      "step": 1124
    },
    {
      "epoch": 0.4313650306748466,
      "grad_norm": 3.526318311691284,
      "learning_rate": 3.176395398380912e-05,
      "loss": 0.0144,
      "step": 1125
    },
    {
      "epoch": 0.4317484662576687,
      "grad_norm": 0.4182857573032379,
      "learning_rate": 3.174265019173413e-05,
      "loss": 0.0027,
      "step": 1126
    },
    {
      "epoch": 0.4321319018404908,
      "grad_norm": 3.599029779434204,
      "learning_rate": 3.172134639965914e-05,
      "loss": 0.0185,
      "step": 1127
    },
    {
      "epoch": 0.4325153374233129,
      "grad_norm": 13.80219841003418,
      "learning_rate": 3.170004260758415e-05,
      "loss": 0.0846,
      "step": 1128
    },
    {
      "epoch": 0.432898773006135,
      "grad_norm": 30.802892684936523,
      "learning_rate": 3.167873881550916e-05,
      "loss": 0.6416,
      "step": 1129
    },
    {
      "epoch": 0.43328220858895705,
      "grad_norm": 1.0535043478012085,
      "learning_rate": 3.165743502343417e-05,
      "loss": 0.0039,
      "step": 1130
    },
    {
      "epoch": 0.43366564417177916,
      "grad_norm": 25.42630958557129,
      "learning_rate": 3.163613123135918e-05,
      "loss": 0.2013,
      "step": 1131
    },
    {
      "epoch": 0.4340490797546012,
      "grad_norm": 8.263312339782715,
      "learning_rate": 3.16148274392842e-05,
      "loss": 0.0891,
      "step": 1132
    },
    {
      "epoch": 0.43443251533742333,
      "grad_norm": 0.37161657214164734,
      "learning_rate": 3.1593523647209206e-05,
      "loss": 0.0031,
      "step": 1133
    },
    {
      "epoch": 0.4348159509202454,
      "grad_norm": 10.676396369934082,
      "learning_rate": 3.1572219855134215e-05,
      "loss": 0.0371,
      "step": 1134
    },
    {
      "epoch": 0.4351993865030675,
      "grad_norm": 7.621251583099365,
      "learning_rate": 3.1550916063059225e-05,
      "loss": 0.4434,
      "step": 1135
    },
    {
      "epoch": 0.43558282208588955,
      "grad_norm": 0.4153030514717102,
      "learning_rate": 3.152961227098424e-05,
      "loss": 0.0028,
      "step": 1136
    },
    {
      "epoch": 0.43596625766871167,
      "grad_norm": 7.068398952484131,
      "learning_rate": 3.150830847890925e-05,
      "loss": 0.0553,
      "step": 1137
    },
    {
      "epoch": 0.4363496932515337,
      "grad_norm": 78.07585144042969,
      "learning_rate": 3.148700468683426e-05,
      "loss": 1.5176,
      "step": 1138
    },
    {
      "epoch": 0.43673312883435583,
      "grad_norm": 55.385684967041016,
      "learning_rate": 3.146570089475927e-05,
      "loss": 0.9117,
      "step": 1139
    },
    {
      "epoch": 0.4371165644171779,
      "grad_norm": 5.380411624908447,
      "learning_rate": 3.144439710268428e-05,
      "loss": 0.2649,
      "step": 1140
    },
    {
      "epoch": 0.4375,
      "grad_norm": 1.1014796495437622,
      "learning_rate": 3.1423093310609294e-05,
      "loss": 0.0047,
      "step": 1141
    },
    {
      "epoch": 0.4378834355828221,
      "grad_norm": 60.870994567871094,
      "learning_rate": 3.14017895185343e-05,
      "loss": 0.8858,
      "step": 1142
    },
    {
      "epoch": 0.43826687116564417,
      "grad_norm": 7.248571395874023,
      "learning_rate": 3.138048572645931e-05,
      "loss": 0.0342,
      "step": 1143
    },
    {
      "epoch": 0.4386503067484663,
      "grad_norm": 5.0063371658325195,
      "learning_rate": 3.135918193438432e-05,
      "loss": 0.0108,
      "step": 1144
    },
    {
      "epoch": 0.43903374233128833,
      "grad_norm": 193.22222900390625,
      "learning_rate": 3.133787814230934e-05,
      "loss": 3.3618,
      "step": 1145
    },
    {
      "epoch": 0.43941717791411045,
      "grad_norm": 128.14036560058594,
      "learning_rate": 3.131657435023435e-05,
      "loss": 1.5245,
      "step": 1146
    },
    {
      "epoch": 0.4398006134969325,
      "grad_norm": 12.670878410339355,
      "learning_rate": 3.129527055815936e-05,
      "loss": 0.0353,
      "step": 1147
    },
    {
      "epoch": 0.4401840490797546,
      "grad_norm": 0.30025821924209595,
      "learning_rate": 3.1273966766084366e-05,
      "loss": 0.0028,
      "step": 1148
    },
    {
      "epoch": 0.44056748466257667,
      "grad_norm": 25.819347381591797,
      "learning_rate": 3.1252662974009375e-05,
      "loss": 0.0876,
      "step": 1149
    },
    {
      "epoch": 0.4409509202453988,
      "grad_norm": 15.05252456665039,
      "learning_rate": 3.1231359181934385e-05,
      "loss": 0.1499,
      "step": 1150
    },
    {
      "epoch": 0.44133435582822084,
      "grad_norm": 12.940930366516113,
      "learning_rate": 3.1210055389859394e-05,
      "loss": 0.0503,
      "step": 1151
    },
    {
      "epoch": 0.44171779141104295,
      "grad_norm": 27.6295166015625,
      "learning_rate": 3.11887515977844e-05,
      "loss": 0.5821,
      "step": 1152
    },
    {
      "epoch": 0.442101226993865,
      "grad_norm": 0.9625844955444336,
      "learning_rate": 3.116744780570941e-05,
      "loss": 0.0054,
      "step": 1153
    },
    {
      "epoch": 0.4424846625766871,
      "grad_norm": 19.52550506591797,
      "learning_rate": 3.114614401363443e-05,
      "loss": 0.034,
      "step": 1154
    },
    {
      "epoch": 0.4428680981595092,
      "grad_norm": 2.0176961421966553,
      "learning_rate": 3.112484022155944e-05,
      "loss": 0.0054,
      "step": 1155
    },
    {
      "epoch": 0.4432515337423313,
      "grad_norm": 209.79808044433594,
      "learning_rate": 3.110353642948445e-05,
      "loss": 2.6355,
      "step": 1156
    },
    {
      "epoch": 0.4436349693251534,
      "grad_norm": 14.19172191619873,
      "learning_rate": 3.1082232637409457e-05,
      "loss": 0.5855,
      "step": 1157
    },
    {
      "epoch": 0.44401840490797545,
      "grad_norm": 62.55099105834961,
      "learning_rate": 3.106092884533447e-05,
      "loss": 0.8145,
      "step": 1158
    },
    {
      "epoch": 0.44440184049079756,
      "grad_norm": 9.882044792175293,
      "learning_rate": 3.103962505325948e-05,
      "loss": 0.0322,
      "step": 1159
    },
    {
      "epoch": 0.4447852760736196,
      "grad_norm": 2.612950325012207,
      "learning_rate": 3.101832126118449e-05,
      "loss": 0.0114,
      "step": 1160
    },
    {
      "epoch": 0.44516871165644173,
      "grad_norm": 19.931814193725586,
      "learning_rate": 3.09970174691095e-05,
      "loss": 0.4781,
      "step": 1161
    },
    {
      "epoch": 0.4455521472392638,
      "grad_norm": 151.32928466796875,
      "learning_rate": 3.097571367703451e-05,
      "loss": 4.6748,
      "step": 1162
    },
    {
      "epoch": 0.4459355828220859,
      "grad_norm": 17.60948944091797,
      "learning_rate": 3.0954409884959526e-05,
      "loss": 0.1786,
      "step": 1163
    },
    {
      "epoch": 0.44631901840490795,
      "grad_norm": 8.400739669799805,
      "learning_rate": 3.0933106092884535e-05,
      "loss": 0.02,
      "step": 1164
    },
    {
      "epoch": 0.44670245398773006,
      "grad_norm": 7.452956199645996,
      "learning_rate": 3.0911802300809545e-05,
      "loss": 0.0238,
      "step": 1165
    },
    {
      "epoch": 0.4470858895705521,
      "grad_norm": 174.81600952148438,
      "learning_rate": 3.0890498508734554e-05,
      "loss": 1.3851,
      "step": 1166
    },
    {
      "epoch": 0.44746932515337423,
      "grad_norm": 2.653099536895752,
      "learning_rate": 3.086919471665956e-05,
      "loss": 0.0158,
      "step": 1167
    },
    {
      "epoch": 0.44785276073619634,
      "grad_norm": 0.17586538195610046,
      "learning_rate": 3.084789092458458e-05,
      "loss": 0.0015,
      "step": 1168
    },
    {
      "epoch": 0.4482361963190184,
      "grad_norm": 17.50071144104004,
      "learning_rate": 3.082658713250959e-05,
      "loss": 0.0608,
      "step": 1169
    },
    {
      "epoch": 0.4486196319018405,
      "grad_norm": 10.259678840637207,
      "learning_rate": 3.08052833404346e-05,
      "loss": 0.0739,
      "step": 1170
    },
    {
      "epoch": 0.44900306748466257,
      "grad_norm": 0.9370091557502747,
      "learning_rate": 3.078397954835961e-05,
      "loss": 0.0054,
      "step": 1171
    },
    {
      "epoch": 0.4493865030674847,
      "grad_norm": 1.6083060503005981,
      "learning_rate": 3.076267575628462e-05,
      "loss": 0.0076,
      "step": 1172
    },
    {
      "epoch": 0.44976993865030673,
      "grad_norm": 50.8786506652832,
      "learning_rate": 3.074137196420963e-05,
      "loss": 1.0371,
      "step": 1173
    },
    {
      "epoch": 0.45015337423312884,
      "grad_norm": 210.81854248046875,
      "learning_rate": 3.072006817213464e-05,
      "loss": 3.9482,
      "step": 1174
    },
    {
      "epoch": 0.4505368098159509,
      "grad_norm": 1.0925509929656982,
      "learning_rate": 3.069876438005965e-05,
      "loss": 0.0053,
      "step": 1175
    },
    {
      "epoch": 0.450920245398773,
      "grad_norm": 0.6283935308456421,
      "learning_rate": 3.067746058798466e-05,
      "loss": 0.0036,
      "step": 1176
    },
    {
      "epoch": 0.45130368098159507,
      "grad_norm": 155.3126678466797,
      "learning_rate": 3.065615679590968e-05,
      "loss": 5.6431,
      "step": 1177
    },
    {
      "epoch": 0.4516871165644172,
      "grad_norm": 4.078763484954834,
      "learning_rate": 3.0634853003834686e-05,
      "loss": 0.0166,
      "step": 1178
    },
    {
      "epoch": 0.4520705521472393,
      "grad_norm": 332.8621826171875,
      "learning_rate": 3.0613549211759695e-05,
      "loss": 3.4974,
      "step": 1179
    },
    {
      "epoch": 0.45245398773006135,
      "grad_norm": 103.15957641601562,
      "learning_rate": 3.0592245419684705e-05,
      "loss": 2.0648,
      "step": 1180
    },
    {
      "epoch": 0.45283742331288346,
      "grad_norm": 0.7433050274848938,
      "learning_rate": 3.0570941627609714e-05,
      "loss": 0.0039,
      "step": 1181
    },
    {
      "epoch": 0.4532208588957055,
      "grad_norm": 1.3463419675827026,
      "learning_rate": 3.054963783553473e-05,
      "loss": 0.0045,
      "step": 1182
    },
    {
      "epoch": 0.4536042944785276,
      "grad_norm": 62.824432373046875,
      "learning_rate": 3.052833404345974e-05,
      "loss": 1.5899,
      "step": 1183
    },
    {
      "epoch": 0.4539877300613497,
      "grad_norm": 186.881591796875,
      "learning_rate": 3.050703025138475e-05,
      "loss": 3.0674,
      "step": 1184
    },
    {
      "epoch": 0.4543711656441718,
      "grad_norm": 139.8800811767578,
      "learning_rate": 3.048572645930976e-05,
      "loss": 2.6504,
      "step": 1185
    },
    {
      "epoch": 0.45475460122699385,
      "grad_norm": 0.8735361695289612,
      "learning_rate": 3.046442266723477e-05,
      "loss": 0.0045,
      "step": 1186
    },
    {
      "epoch": 0.45513803680981596,
      "grad_norm": 4.634177207946777,
      "learning_rate": 3.0443118875159783e-05,
      "loss": 0.0315,
      "step": 1187
    },
    {
      "epoch": 0.455521472392638,
      "grad_norm": 3.494042158126831,
      "learning_rate": 3.0421815083084793e-05,
      "loss": 0.0229,
      "step": 1188
    },
    {
      "epoch": 0.4559049079754601,
      "grad_norm": 100.4810562133789,
      "learning_rate": 3.0400511291009802e-05,
      "loss": 6.0488,
      "step": 1189
    },
    {
      "epoch": 0.4562883435582822,
      "grad_norm": 0.2752269208431244,
      "learning_rate": 3.0379207498934815e-05,
      "loss": 0.0022,
      "step": 1190
    },
    {
      "epoch": 0.4566717791411043,
      "grad_norm": 8.22811508178711,
      "learning_rate": 3.0357903706859824e-05,
      "loss": 0.223,
      "step": 1191
    },
    {
      "epoch": 0.4570552147239264,
      "grad_norm": 6.318234920501709,
      "learning_rate": 3.0336599914784837e-05,
      "loss": 0.4302,
      "step": 1192
    },
    {
      "epoch": 0.45743865030674846,
      "grad_norm": 13.462141036987305,
      "learning_rate": 3.0315296122709846e-05,
      "loss": 0.0474,
      "step": 1193
    },
    {
      "epoch": 0.4578220858895706,
      "grad_norm": 23.509485244750977,
      "learning_rate": 3.029399233063486e-05,
      "loss": 0.5611,
      "step": 1194
    },
    {
      "epoch": 0.45820552147239263,
      "grad_norm": 4.083889007568359,
      "learning_rate": 3.027268853855986e-05,
      "loss": 0.0215,
      "step": 1195
    },
    {
      "epoch": 0.45858895705521474,
      "grad_norm": 232.07672119140625,
      "learning_rate": 3.0251384746484874e-05,
      "loss": 6.0332,
      "step": 1196
    },
    {
      "epoch": 0.4589723926380368,
      "grad_norm": 0.418995201587677,
      "learning_rate": 3.0230080954409883e-05,
      "loss": 0.0025,
      "step": 1197
    },
    {
      "epoch": 0.4593558282208589,
      "grad_norm": 0.27796462178230286,
      "learning_rate": 3.0208777162334896e-05,
      "loss": 0.0016,
      "step": 1198
    },
    {
      "epoch": 0.45973926380368096,
      "grad_norm": 2.5159637928009033,
      "learning_rate": 3.0187473370259905e-05,
      "loss": 0.0093,
      "step": 1199
    },
    {
      "epoch": 0.4601226993865031,
      "grad_norm": 0.2359163910150528,
      "learning_rate": 3.0166169578184918e-05,
      "loss": 0.0017,
      "step": 1200
    },
    {
      "epoch": 0.46050613496932513,
      "grad_norm": 1.0194178819656372,
      "learning_rate": 3.0144865786109927e-05,
      "loss": 0.0034,
      "step": 1201
    },
    {
      "epoch": 0.46088957055214724,
      "grad_norm": 4.287911415100098,
      "learning_rate": 3.0123561994034937e-05,
      "loss": 0.0126,
      "step": 1202
    },
    {
      "epoch": 0.4612730061349693,
      "grad_norm": 0.42968660593032837,
      "learning_rate": 3.010225820195995e-05,
      "loss": 0.0031,
      "step": 1203
    },
    {
      "epoch": 0.4616564417177914,
      "grad_norm": 22.549182891845703,
      "learning_rate": 3.008095440988496e-05,
      "loss": 0.5249,
      "step": 1204
    },
    {
      "epoch": 0.4620398773006135,
      "grad_norm": 1.3578665256500244,
      "learning_rate": 3.005965061780997e-05,
      "loss": 0.005,
      "step": 1205
    },
    {
      "epoch": 0.4624233128834356,
      "grad_norm": 0.51641446352005,
      "learning_rate": 3.003834682573498e-05,
      "loss": 0.0029,
      "step": 1206
    },
    {
      "epoch": 0.4628067484662577,
      "grad_norm": 0.573008120059967,
      "learning_rate": 3.0017043033659993e-05,
      "loss": 0.0036,
      "step": 1207
    },
    {
      "epoch": 0.46319018404907975,
      "grad_norm": 129.40185546875,
      "learning_rate": 2.9995739241585003e-05,
      "loss": 2.0801,
      "step": 1208
    },
    {
      "epoch": 0.46357361963190186,
      "grad_norm": 31.359634399414062,
      "learning_rate": 2.9974435449510012e-05,
      "loss": 0.5342,
      "step": 1209
    },
    {
      "epoch": 0.4639570552147239,
      "grad_norm": 1.4557340145111084,
      "learning_rate": 2.9953131657435025e-05,
      "loss": 0.0035,
      "step": 1210
    },
    {
      "epoch": 0.464340490797546,
      "grad_norm": 7.835709571838379,
      "learning_rate": 2.9931827865360034e-05,
      "loss": 0.1309,
      "step": 1211
    },
    {
      "epoch": 0.4647239263803681,
      "grad_norm": 1.2238229513168335,
      "learning_rate": 2.9910524073285047e-05,
      "loss": 0.0049,
      "step": 1212
    },
    {
      "epoch": 0.4651073619631902,
      "grad_norm": 2.1265645027160645,
      "learning_rate": 2.9889220281210056e-05,
      "loss": 0.0101,
      "step": 1213
    },
    {
      "epoch": 0.46549079754601225,
      "grad_norm": 2.475086212158203,
      "learning_rate": 2.986791648913507e-05,
      "loss": 0.0075,
      "step": 1214
    },
    {
      "epoch": 0.46587423312883436,
      "grad_norm": 1.0121549367904663,
      "learning_rate": 2.9846612697060078e-05,
      "loss": 0.0053,
      "step": 1215
    },
    {
      "epoch": 0.4662576687116564,
      "grad_norm": 51.59149169921875,
      "learning_rate": 2.9825308904985087e-05,
      "loss": 1.2022,
      "step": 1216
    },
    {
      "epoch": 0.4666411042944785,
      "grad_norm": 50.36254119873047,
      "learning_rate": 2.98040051129101e-05,
      "loss": 0.7691,
      "step": 1217
    },
    {
      "epoch": 0.46702453987730064,
      "grad_norm": 1.9499720335006714,
      "learning_rate": 2.978270132083511e-05,
      "loss": 0.0092,
      "step": 1218
    },
    {
      "epoch": 0.4674079754601227,
      "grad_norm": 105.05870056152344,
      "learning_rate": 2.9761397528760122e-05,
      "loss": 1.1055,
      "step": 1219
    },
    {
      "epoch": 0.4677914110429448,
      "grad_norm": 0.6851003766059875,
      "learning_rate": 2.974009373668513e-05,
      "loss": 0.0033,
      "step": 1220
    },
    {
      "epoch": 0.46817484662576686,
      "grad_norm": 9.881448745727539,
      "learning_rate": 2.9718789944610144e-05,
      "loss": 0.0343,
      "step": 1221
    },
    {
      "epoch": 0.46855828220858897,
      "grad_norm": 0.9985806941986084,
      "learning_rate": 2.9697486152535153e-05,
      "loss": 0.006,
      "step": 1222
    },
    {
      "epoch": 0.46894171779141103,
      "grad_norm": 10.051969528198242,
      "learning_rate": 2.9676182360460163e-05,
      "loss": 0.0237,
      "step": 1223
    },
    {
      "epoch": 0.46932515337423314,
      "grad_norm": 1.2723650932312012,
      "learning_rate": 2.9654878568385175e-05,
      "loss": 0.0057,
      "step": 1224
    },
    {
      "epoch": 0.4697085889570552,
      "grad_norm": 1.2021504640579224,
      "learning_rate": 2.9633574776310185e-05,
      "loss": 0.0051,
      "step": 1225
    },
    {
      "epoch": 0.4700920245398773,
      "grad_norm": 6.088982105255127,
      "learning_rate": 2.9612270984235197e-05,
      "loss": 0.0288,
      "step": 1226
    },
    {
      "epoch": 0.47047546012269936,
      "grad_norm": 1.1847741603851318,
      "learning_rate": 2.9590967192160207e-05,
      "loss": 0.0046,
      "step": 1227
    },
    {
      "epoch": 0.4708588957055215,
      "grad_norm": 1.1589510440826416,
      "learning_rate": 2.956966340008522e-05,
      "loss": 0.0042,
      "step": 1228
    },
    {
      "epoch": 0.4712423312883436,
      "grad_norm": 4.910191059112549,
      "learning_rate": 2.954835960801023e-05,
      "loss": 0.0179,
      "step": 1229
    },
    {
      "epoch": 0.47162576687116564,
      "grad_norm": 131.7100372314453,
      "learning_rate": 2.9527055815935238e-05,
      "loss": 5.6689,
      "step": 1230
    },
    {
      "epoch": 0.47200920245398775,
      "grad_norm": 48.22613525390625,
      "learning_rate": 2.950575202386025e-05,
      "loss": 0.585,
      "step": 1231
    },
    {
      "epoch": 0.4723926380368098,
      "grad_norm": 3.0125133991241455,
      "learning_rate": 2.948444823178526e-05,
      "loss": 0.0114,
      "step": 1232
    },
    {
      "epoch": 0.4727760736196319,
      "grad_norm": 109.76148986816406,
      "learning_rate": 2.9463144439710273e-05,
      "loss": 1.4414,
      "step": 1233
    },
    {
      "epoch": 0.473159509202454,
      "grad_norm": 4.895443439483643,
      "learning_rate": 2.9441840647635282e-05,
      "loss": 0.0181,
      "step": 1234
    },
    {
      "epoch": 0.4735429447852761,
      "grad_norm": 3.6181979179382324,
      "learning_rate": 2.9420536855560295e-05,
      "loss": 0.0222,
      "step": 1235
    },
    {
      "epoch": 0.47392638036809814,
      "grad_norm": 55.43559646606445,
      "learning_rate": 2.9399233063485304e-05,
      "loss": 1.0049,
      "step": 1236
    },
    {
      "epoch": 0.47430981595092025,
      "grad_norm": 12.764467239379883,
      "learning_rate": 2.9377929271410313e-05,
      "loss": 0.069,
      "step": 1237
    },
    {
      "epoch": 0.4746932515337423,
      "grad_norm": 5.7123942375183105,
      "learning_rate": 2.9356625479335326e-05,
      "loss": 0.0184,
      "step": 1238
    },
    {
      "epoch": 0.4750766871165644,
      "grad_norm": 11.40140438079834,
      "learning_rate": 2.9335321687260335e-05,
      "loss": 0.0416,
      "step": 1239
    },
    {
      "epoch": 0.4754601226993865,
      "grad_norm": 0.8403089642524719,
      "learning_rate": 2.931401789518534e-05,
      "loss": 0.0034,
      "step": 1240
    },
    {
      "epoch": 0.4758435582822086,
      "grad_norm": 3.561800479888916,
      "learning_rate": 2.9292714103110354e-05,
      "loss": 0.0037,
      "step": 1241
    },
    {
      "epoch": 0.4762269938650307,
      "grad_norm": 136.95355224609375,
      "learning_rate": 2.9271410311035363e-05,
      "loss": 2.2112,
      "step": 1242
    },
    {
      "epoch": 0.47661042944785276,
      "grad_norm": 0.960547685623169,
      "learning_rate": 2.9250106518960373e-05,
      "loss": 0.0028,
      "step": 1243
    },
    {
      "epoch": 0.47699386503067487,
      "grad_norm": 6.117381572723389,
      "learning_rate": 2.9228802726885385e-05,
      "loss": 0.0469,
      "step": 1244
    },
    {
      "epoch": 0.4773773006134969,
      "grad_norm": 19.996009826660156,
      "learning_rate": 2.9207498934810395e-05,
      "loss": 0.6602,
      "step": 1245
    },
    {
      "epoch": 0.47776073619631904,
      "grad_norm": 9.0525541305542,
      "learning_rate": 2.9186195142735407e-05,
      "loss": 0.0484,
      "step": 1246
    },
    {
      "epoch": 0.4781441717791411,
      "grad_norm": 9.098682403564453,
      "learning_rate": 2.9164891350660417e-05,
      "loss": 0.0613,
      "step": 1247
    },
    {
      "epoch": 0.4785276073619632,
      "grad_norm": 192.00765991210938,
      "learning_rate": 2.914358755858543e-05,
      "loss": 1.1309,
      "step": 1248
    },
    {
      "epoch": 0.47891104294478526,
      "grad_norm": 1.7598403692245483,
      "learning_rate": 2.912228376651044e-05,
      "loss": 0.0053,
      "step": 1249
    },
    {
      "epoch": 0.47929447852760737,
      "grad_norm": 84.8008041381836,
      "learning_rate": 2.9100979974435448e-05,
      "loss": 1.0332,
      "step": 1250
    },
    {
      "epoch": 0.4796779141104294,
      "grad_norm": 123.36064910888672,
      "learning_rate": 2.907967618236046e-05,
      "loss": 2.0742,
      "step": 1251
    },
    {
      "epoch": 0.48006134969325154,
      "grad_norm": 0.9011976718902588,
      "learning_rate": 2.905837239028547e-05,
      "loss": 0.0026,
      "step": 1252
    },
    {
      "epoch": 0.4804447852760736,
      "grad_norm": 95.64689636230469,
      "learning_rate": 2.9037068598210483e-05,
      "loss": 1.8077,
      "step": 1253
    },
    {
      "epoch": 0.4808282208588957,
      "grad_norm": 2.4204602241516113,
      "learning_rate": 2.9015764806135492e-05,
      "loss": 0.0072,
      "step": 1254
    },
    {
      "epoch": 0.4812116564417178,
      "grad_norm": 4.134298324584961,
      "learning_rate": 2.8994461014060505e-05,
      "loss": 0.022,
      "step": 1255
    },
    {
      "epoch": 0.4815950920245399,
      "grad_norm": 139.32273864746094,
      "learning_rate": 2.8973157221985514e-05,
      "loss": 6.0684,
      "step": 1256
    },
    {
      "epoch": 0.481978527607362,
      "grad_norm": 597.2622680664062,
      "learning_rate": 2.8951853429910523e-05,
      "loss": 2.1761,
      "step": 1257
    },
    {
      "epoch": 0.48236196319018404,
      "grad_norm": 5.3948163986206055,
      "learning_rate": 2.8930549637835536e-05,
      "loss": 0.022,
      "step": 1258
    },
    {
      "epoch": 0.48274539877300615,
      "grad_norm": 0.4378185272216797,
      "learning_rate": 2.8909245845760545e-05,
      "loss": 0.0024,
      "step": 1259
    },
    {
      "epoch": 0.4831288343558282,
      "grad_norm": 2.58599591255188,
      "learning_rate": 2.8887942053685558e-05,
      "loss": 0.0067,
      "step": 1260
    },
    {
      "epoch": 0.4835122699386503,
      "grad_norm": 7.638411998748779,
      "learning_rate": 2.8866638261610567e-05,
      "loss": 0.0158,
      "step": 1261
    },
    {
      "epoch": 0.4838957055214724,
      "grad_norm": 52.83462142944336,
      "learning_rate": 2.884533446953558e-05,
      "loss": 0.7105,
      "step": 1262
    },
    {
      "epoch": 0.4842791411042945,
      "grad_norm": 68.8130874633789,
      "learning_rate": 2.882403067746059e-05,
      "loss": 1.1182,
      "step": 1263
    },
    {
      "epoch": 0.48466257668711654,
      "grad_norm": 1.7231501340866089,
      "learning_rate": 2.88027268853856e-05,
      "loss": 0.0049,
      "step": 1264
    },
    {
      "epoch": 0.48504601226993865,
      "grad_norm": 1.7606278657913208,
      "learning_rate": 2.878142309331061e-05,
      "loss": 0.0068,
      "step": 1265
    },
    {
      "epoch": 0.4854294478527607,
      "grad_norm": 0.5795959830284119,
      "learning_rate": 2.876011930123562e-05,
      "loss": 0.0031,
      "step": 1266
    },
    {
      "epoch": 0.4858128834355828,
      "grad_norm": 0.928696870803833,
      "learning_rate": 2.8738815509160633e-05,
      "loss": 0.0055,
      "step": 1267
    },
    {
      "epoch": 0.48619631901840493,
      "grad_norm": 2.9062998294830322,
      "learning_rate": 2.8717511717085643e-05,
      "loss": 0.0081,
      "step": 1268
    },
    {
      "epoch": 0.486579754601227,
      "grad_norm": 0.6792395114898682,
      "learning_rate": 2.8696207925010655e-05,
      "loss": 0.0034,
      "step": 1269
    },
    {
      "epoch": 0.4869631901840491,
      "grad_norm": 89.9220962524414,
      "learning_rate": 2.8674904132935665e-05,
      "loss": 1.0928,
      "step": 1270
    },
    {
      "epoch": 0.48734662576687116,
      "grad_norm": 47.2714958190918,
      "learning_rate": 2.8653600340860677e-05,
      "loss": 0.939,
      "step": 1271
    },
    {
      "epoch": 0.48773006134969327,
      "grad_norm": 175.17723083496094,
      "learning_rate": 2.8632296548785687e-05,
      "loss": 5.8057,
      "step": 1272
    },
    {
      "epoch": 0.4881134969325153,
      "grad_norm": 47.968807220458984,
      "learning_rate": 2.8610992756710696e-05,
      "loss": 0.7549,
      "step": 1273
    },
    {
      "epoch": 0.48849693251533743,
      "grad_norm": 0.32888033986091614,
      "learning_rate": 2.858968896463571e-05,
      "loss": 0.0023,
      "step": 1274
    },
    {
      "epoch": 0.4888803680981595,
      "grad_norm": 3.0864601135253906,
      "learning_rate": 2.8568385172560718e-05,
      "loss": 0.0145,
      "step": 1275
    },
    {
      "epoch": 0.4892638036809816,
      "grad_norm": 1.845633864402771,
      "learning_rate": 2.854708138048573e-05,
      "loss": 0.005,
      "step": 1276
    },
    {
      "epoch": 0.48964723926380366,
      "grad_norm": 2.848792552947998,
      "learning_rate": 2.852577758841074e-05,
      "loss": 0.0059,
      "step": 1277
    },
    {
      "epoch": 0.49003067484662577,
      "grad_norm": 9.916733741760254,
      "learning_rate": 2.8504473796335753e-05,
      "loss": 0.0439,
      "step": 1278
    },
    {
      "epoch": 0.4904141104294479,
      "grad_norm": 22.210588455200195,
      "learning_rate": 2.8483170004260762e-05,
      "loss": 0.5889,
      "step": 1279
    },
    {
      "epoch": 0.49079754601226994,
      "grad_norm": 21.33092498779297,
      "learning_rate": 2.846186621218577e-05,
      "loss": 0.5547,
      "step": 1280
    },
    {
      "epoch": 0.49118098159509205,
      "grad_norm": 0.24328042566776276,
      "learning_rate": 2.8440562420110784e-05,
      "loss": 0.0028,
      "step": 1281
    },
    {
      "epoch": 0.4915644171779141,
      "grad_norm": 11.243773460388184,
      "learning_rate": 2.8419258628035793e-05,
      "loss": 0.1562,
      "step": 1282
    },
    {
      "epoch": 0.4919478527607362,
      "grad_norm": 8.533361434936523,
      "learning_rate": 2.8397954835960806e-05,
      "loss": 0.4775,
      "step": 1283
    },
    {
      "epoch": 0.49233128834355827,
      "grad_norm": 41.438018798828125,
      "learning_rate": 2.8376651043885815e-05,
      "loss": 0.8608,
      "step": 1284
    },
    {
      "epoch": 0.4927147239263804,
      "grad_norm": 100.1050796508789,
      "learning_rate": 2.835534725181082e-05,
      "loss": 1.001,
      "step": 1285
    },
    {
      "epoch": 0.49309815950920244,
      "grad_norm": 46.85498046875,
      "learning_rate": 2.833404345973583e-05,
      "loss": 0.5293,
      "step": 1286
    },
    {
      "epoch": 0.49348159509202455,
      "grad_norm": 1.5562477111816406,
      "learning_rate": 2.8312739667660843e-05,
      "loss": 0.006,
      "step": 1287
    },
    {
      "epoch": 0.4938650306748466,
      "grad_norm": 215.5096893310547,
      "learning_rate": 2.8291435875585853e-05,
      "loss": 4.0902,
      "step": 1288
    },
    {
      "epoch": 0.4942484662576687,
      "grad_norm": 2.6404812335968018,
      "learning_rate": 2.8270132083510865e-05,
      "loss": 0.0129,
      "step": 1289
    },
    {
      "epoch": 0.4946319018404908,
      "grad_norm": 86.96484375,
      "learning_rate": 2.8248828291435875e-05,
      "loss": 1.3555,
      "step": 1290
    },
    {
      "epoch": 0.4950153374233129,
      "grad_norm": 10.088623046875,
      "learning_rate": 2.8227524499360887e-05,
      "loss": 0.2534,
      "step": 1291
    },
    {
      "epoch": 0.495398773006135,
      "grad_norm": 87.9999008178711,
      "learning_rate": 2.8206220707285897e-05,
      "loss": 1.25,
      "step": 1292
    },
    {
      "epoch": 0.49578220858895705,
      "grad_norm": 0.6408212184906006,
      "learning_rate": 2.8184916915210906e-05,
      "loss": 0.0028,
      "step": 1293
    },
    {
      "epoch": 0.49616564417177916,
      "grad_norm": 0.568196177482605,
      "learning_rate": 2.816361312313592e-05,
      "loss": 0.0036,
      "step": 1294
    },
    {
      "epoch": 0.4965490797546012,
      "grad_norm": 21.775447845458984,
      "learning_rate": 2.8142309331060928e-05,
      "loss": 0.6846,
      "step": 1295
    },
    {
      "epoch": 0.49693251533742333,
      "grad_norm": 2.9825246334075928,
      "learning_rate": 2.812100553898594e-05,
      "loss": 0.0079,
      "step": 1296
    },
    {
      "epoch": 0.4973159509202454,
      "grad_norm": 1.3889049291610718,
      "learning_rate": 2.809970174691095e-05,
      "loss": 0.0087,
      "step": 1297
    },
    {
      "epoch": 0.4976993865030675,
      "grad_norm": 0.6602429151535034,
      "learning_rate": 2.8078397954835963e-05,
      "loss": 0.0024,
      "step": 1298
    },
    {
      "epoch": 0.49808282208588955,
      "grad_norm": 84.68866729736328,
      "learning_rate": 2.8057094162760972e-05,
      "loss": 0.8604,
      "step": 1299
    },
    {
      "epoch": 0.49846625766871167,
      "grad_norm": 1.651374340057373,
      "learning_rate": 2.803579037068598e-05,
      "loss": 0.0067,
      "step": 1300
    },
    {
      "epoch": 0.4988496932515337,
      "grad_norm": 15.142035484313965,
      "learning_rate": 2.8014486578610994e-05,
      "loss": 0.5489,
      "step": 1301
    },
    {
      "epoch": 0.49923312883435583,
      "grad_norm": 31.190401077270508,
      "learning_rate": 2.7993182786536003e-05,
      "loss": 0.1139,
      "step": 1302
    },
    {
      "epoch": 0.4996165644171779,
      "grad_norm": 4.73783540725708,
      "learning_rate": 2.7971878994461016e-05,
      "loss": 0.02,
      "step": 1303
    },
    {
      "epoch": 0.5,
      "grad_norm": 100.45794677734375,
      "learning_rate": 2.7950575202386025e-05,
      "loss": 1.3066,
      "step": 1304
    },
    {
      "epoch": 0.5003834355828221,
      "grad_norm": 6.176492214202881,
      "learning_rate": 2.7929271410311038e-05,
      "loss": 0.4216,
      "step": 1305
    },
    {
      "epoch": 0.5007668711656442,
      "grad_norm": 242.9879608154297,
      "learning_rate": 2.7907967618236047e-05,
      "loss": 2.3107,
      "step": 1306
    },
    {
      "epoch": 0.5011503067484663,
      "grad_norm": 105.03369903564453,
      "learning_rate": 2.7886663826161057e-05,
      "loss": 1.8819,
      "step": 1307
    },
    {
      "epoch": 0.5015337423312883,
      "grad_norm": 19.288230895996094,
      "learning_rate": 2.786536003408607e-05,
      "loss": 0.0878,
      "step": 1308
    },
    {
      "epoch": 0.5019171779141104,
      "grad_norm": 149.70127868652344,
      "learning_rate": 2.784405624201108e-05,
      "loss": 6.0527,
      "step": 1309
    },
    {
      "epoch": 0.5023006134969326,
      "grad_norm": 105.20108032226562,
      "learning_rate": 2.782275244993609e-05,
      "loss": 1.4551,
      "step": 1310
    },
    {
      "epoch": 0.5026840490797546,
      "grad_norm": 17.483646392822266,
      "learning_rate": 2.78014486578611e-05,
      "loss": 0.0961,
      "step": 1311
    },
    {
      "epoch": 0.5030674846625767,
      "grad_norm": 12.938019752502441,
      "learning_rate": 2.7780144865786113e-05,
      "loss": 0.4529,
      "step": 1312
    },
    {
      "epoch": 0.5034509202453987,
      "grad_norm": 19.03183364868164,
      "learning_rate": 2.7758841073711123e-05,
      "loss": 0.0877,
      "step": 1313
    },
    {
      "epoch": 0.5038343558282209,
      "grad_norm": 95.82111358642578,
      "learning_rate": 2.7737537281636132e-05,
      "loss": 1.2188,
      "step": 1314
    },
    {
      "epoch": 0.504217791411043,
      "grad_norm": 0.7974203824996948,
      "learning_rate": 2.7716233489561145e-05,
      "loss": 0.0022,
      "step": 1315
    },
    {
      "epoch": 0.504601226993865,
      "grad_norm": 21.486732482910156,
      "learning_rate": 2.7694929697486154e-05,
      "loss": 0.542,
      "step": 1316
    },
    {
      "epoch": 0.5049846625766872,
      "grad_norm": 12.884425163269043,
      "learning_rate": 2.7673625905411167e-05,
      "loss": 0.1489,
      "step": 1317
    },
    {
      "epoch": 0.5053680981595092,
      "grad_norm": 367.4250183105469,
      "learning_rate": 2.7652322113336176e-05,
      "loss": 1.3174,
      "step": 1318
    },
    {
      "epoch": 0.5057515337423313,
      "grad_norm": 6.886001110076904,
      "learning_rate": 2.763101832126119e-05,
      "loss": 0.0208,
      "step": 1319
    },
    {
      "epoch": 0.5061349693251533,
      "grad_norm": 4.112507343292236,
      "learning_rate": 2.7609714529186198e-05,
      "loss": 0.0295,
      "step": 1320
    },
    {
      "epoch": 0.5065184049079755,
      "grad_norm": 8.974945068359375,
      "learning_rate": 2.7588410737111207e-05,
      "loss": 0.0565,
      "step": 1321
    },
    {
      "epoch": 0.5069018404907976,
      "grad_norm": 5.564464092254639,
      "learning_rate": 2.756710694503622e-05,
      "loss": 0.4058,
      "step": 1322
    },
    {
      "epoch": 0.5072852760736196,
      "grad_norm": 8.100415229797363,
      "learning_rate": 2.754580315296123e-05,
      "loss": 0.4504,
      "step": 1323
    },
    {
      "epoch": 0.5076687116564417,
      "grad_norm": 0.20162974298000336,
      "learning_rate": 2.7524499360886242e-05,
      "loss": 0.0014,
      "step": 1324
    },
    {
      "epoch": 0.5080521472392638,
      "grad_norm": 4.389342784881592,
      "learning_rate": 2.750319556881125e-05,
      "loss": 0.012,
      "step": 1325
    },
    {
      "epoch": 0.5084355828220859,
      "grad_norm": 1.5896692276000977,
      "learning_rate": 2.7481891776736264e-05,
      "loss": 0.0022,
      "step": 1326
    },
    {
      "epoch": 0.508819018404908,
      "grad_norm": 100.52224731445312,
      "learning_rate": 2.7460587984661273e-05,
      "loss": 1.8311,
      "step": 1327
    },
    {
      "epoch": 0.50920245398773,
      "grad_norm": 3.928161382675171,
      "learning_rate": 2.7439284192586283e-05,
      "loss": 0.0204,
      "step": 1328
    },
    {
      "epoch": 0.5095858895705522,
      "grad_norm": 13.133651733398438,
      "learning_rate": 2.7417980400511295e-05,
      "loss": 0.0135,
      "step": 1329
    },
    {
      "epoch": 0.5099693251533742,
      "grad_norm": 32.4322624206543,
      "learning_rate": 2.73966766084363e-05,
      "loss": 0.5806,
      "step": 1330
    },
    {
      "epoch": 0.5103527607361963,
      "grad_norm": 0.5560774207115173,
      "learning_rate": 2.737537281636131e-05,
      "loss": 0.0034,
      "step": 1331
    },
    {
      "epoch": 0.5107361963190185,
      "grad_norm": 275.88140869140625,
      "learning_rate": 2.7354069024286323e-05,
      "loss": 5.6875,
      "step": 1332
    },
    {
      "epoch": 0.5111196319018405,
      "grad_norm": 10.551005363464355,
      "learning_rate": 2.7332765232211333e-05,
      "loss": 0.071,
      "step": 1333
    },
    {
      "epoch": 0.5115030674846626,
      "grad_norm": 128.31874084472656,
      "learning_rate": 2.7311461440136342e-05,
      "loss": 5.5439,
      "step": 1334
    },
    {
      "epoch": 0.5118865030674846,
      "grad_norm": 168.2147979736328,
      "learning_rate": 2.7290157648061355e-05,
      "loss": 1.3653,
      "step": 1335
    },
    {
      "epoch": 0.5122699386503068,
      "grad_norm": 3.8666961193084717,
      "learning_rate": 2.7268853855986364e-05,
      "loss": 0.0106,
      "step": 1336
    },
    {
      "epoch": 0.5126533742331288,
      "grad_norm": 1.494707703590393,
      "learning_rate": 2.7247550063911377e-05,
      "loss": 0.0083,
      "step": 1337
    },
    {
      "epoch": 0.5130368098159509,
      "grad_norm": 20.960519790649414,
      "learning_rate": 2.7226246271836386e-05,
      "loss": 0.4636,
      "step": 1338
    },
    {
      "epoch": 0.513420245398773,
      "grad_norm": 165.2379913330078,
      "learning_rate": 2.72049424797614e-05,
      "loss": 5.5386,
      "step": 1339
    },
    {
      "epoch": 0.5138036809815951,
      "grad_norm": 2.861499071121216,
      "learning_rate": 2.7183638687686408e-05,
      "loss": 0.0108,
      "step": 1340
    },
    {
      "epoch": 0.5141871165644172,
      "grad_norm": 23.677654266357422,
      "learning_rate": 2.7162334895611417e-05,
      "loss": 0.1123,
      "step": 1341
    },
    {
      "epoch": 0.5145705521472392,
      "grad_norm": 118.72830963134766,
      "learning_rate": 2.714103110353643e-05,
      "loss": 1.2725,
      "step": 1342
    },
    {
      "epoch": 0.5149539877300614,
      "grad_norm": 0.7612152099609375,
      "learning_rate": 2.711972731146144e-05,
      "loss": 0.005,
      "step": 1343
    },
    {
      "epoch": 0.5153374233128835,
      "grad_norm": 167.5882568359375,
      "learning_rate": 2.7098423519386452e-05,
      "loss": 2.1641,
      "step": 1344
    },
    {
      "epoch": 0.5157208588957055,
      "grad_norm": 5.0922956466674805,
      "learning_rate": 2.707711972731146e-05,
      "loss": 0.0371,
      "step": 1345
    },
    {
      "epoch": 0.5161042944785276,
      "grad_norm": 13.625322341918945,
      "learning_rate": 2.7055815935236474e-05,
      "loss": 0.5049,
      "step": 1346
    },
    {
      "epoch": 0.5164877300613497,
      "grad_norm": 1.7424520254135132,
      "learning_rate": 2.7034512143161483e-05,
      "loss": 0.005,
      "step": 1347
    },
    {
      "epoch": 0.5168711656441718,
      "grad_norm": 0.6367583274841309,
      "learning_rate": 2.7013208351086493e-05,
      "loss": 0.0036,
      "step": 1348
    },
    {
      "epoch": 0.5172546012269938,
      "grad_norm": 3.3440346717834473,
      "learning_rate": 2.6991904559011505e-05,
      "loss": 0.0164,
      "step": 1349
    },
    {
      "epoch": 0.5176380368098159,
      "grad_norm": 131.81895446777344,
      "learning_rate": 2.6970600766936515e-05,
      "loss": 6.2031,
      "step": 1350
    },
    {
      "epoch": 0.5180214723926381,
      "grad_norm": 32.145870208740234,
      "learning_rate": 2.6949296974861527e-05,
      "loss": 0.8281,
      "step": 1351
    },
    {
      "epoch": 0.5184049079754601,
      "grad_norm": 15.580379486083984,
      "learning_rate": 2.6927993182786537e-05,
      "loss": 0.1283,
      "step": 1352
    },
    {
      "epoch": 0.5187883435582822,
      "grad_norm": 8.98824405670166,
      "learning_rate": 2.690668939071155e-05,
      "loss": 0.1846,
      "step": 1353
    },
    {
      "epoch": 0.5191717791411042,
      "grad_norm": 18.761882781982422,
      "learning_rate": 2.688538559863656e-05,
      "loss": 0.5318,
      "step": 1354
    },
    {
      "epoch": 0.5195552147239264,
      "grad_norm": 7.901307106018066,
      "learning_rate": 2.686408180656157e-05,
      "loss": 0.4761,
      "step": 1355
    },
    {
      "epoch": 0.5199386503067485,
      "grad_norm": 183.93434143066406,
      "learning_rate": 2.684277801448658e-05,
      "loss": 1.3828,
      "step": 1356
    },
    {
      "epoch": 0.5203220858895705,
      "grad_norm": 32.8038444519043,
      "learning_rate": 2.682147422241159e-05,
      "loss": 0.6572,
      "step": 1357
    },
    {
      "epoch": 0.5207055214723927,
      "grad_norm": 3.186523675918579,
      "learning_rate": 2.6800170430336603e-05,
      "loss": 0.0134,
      "step": 1358
    },
    {
      "epoch": 0.5210889570552147,
      "grad_norm": 1.7899644374847412,
      "learning_rate": 2.6778866638261612e-05,
      "loss": 0.0081,
      "step": 1359
    },
    {
      "epoch": 0.5214723926380368,
      "grad_norm": 6.219476699829102,
      "learning_rate": 2.6757562846186625e-05,
      "loss": 0.0469,
      "step": 1360
    },
    {
      "epoch": 0.5218558282208589,
      "grad_norm": 3.5063388347625732,
      "learning_rate": 2.6736259054111634e-05,
      "loss": 0.3955,
      "step": 1361
    },
    {
      "epoch": 0.522239263803681,
      "grad_norm": 58.13272476196289,
      "learning_rate": 2.6714955262036647e-05,
      "loss": 0.7842,
      "step": 1362
    },
    {
      "epoch": 0.5226226993865031,
      "grad_norm": 216.25241088867188,
      "learning_rate": 2.6693651469961656e-05,
      "loss": 1.1358,
      "step": 1363
    },
    {
      "epoch": 0.5230061349693251,
      "grad_norm": 0.674191415309906,
      "learning_rate": 2.6672347677886665e-05,
      "loss": 0.0041,
      "step": 1364
    },
    {
      "epoch": 0.5233895705521472,
      "grad_norm": 36.996612548828125,
      "learning_rate": 2.6651043885811678e-05,
      "loss": 0.6734,
      "step": 1365
    },
    {
      "epoch": 0.5237730061349694,
      "grad_norm": 0.2353329062461853,
      "learning_rate": 2.6629740093736687e-05,
      "loss": 0.0021,
      "step": 1366
    },
    {
      "epoch": 0.5241564417177914,
      "grad_norm": 19.37664031982422,
      "learning_rate": 2.66084363016617e-05,
      "loss": 0.1535,
      "step": 1367
    },
    {
      "epoch": 0.5245398773006135,
      "grad_norm": 573.6376953125,
      "learning_rate": 2.658713250958671e-05,
      "loss": 1.7295,
      "step": 1368
    },
    {
      "epoch": 0.5249233128834356,
      "grad_norm": 153.51812744140625,
      "learning_rate": 2.6565828717511722e-05,
      "loss": 4.3696,
      "step": 1369
    },
    {
      "epoch": 0.5253067484662577,
      "grad_norm": 0.6202207803726196,
      "learning_rate": 2.654452492543673e-05,
      "loss": 0.003,
      "step": 1370
    },
    {
      "epoch": 0.5256901840490797,
      "grad_norm": 121.04489135742188,
      "learning_rate": 2.652322113336174e-05,
      "loss": 1.6329,
      "step": 1371
    },
    {
      "epoch": 0.5260736196319018,
      "grad_norm": 2.2874927520751953,
      "learning_rate": 2.6501917341286753e-05,
      "loss": 0.0081,
      "step": 1372
    },
    {
      "epoch": 0.526457055214724,
      "grad_norm": 0.22468091547489166,
      "learning_rate": 2.6480613549211763e-05,
      "loss": 0.0013,
      "step": 1373
    },
    {
      "epoch": 0.526840490797546,
      "grad_norm": 2.073918342590332,
      "learning_rate": 2.6459309757136775e-05,
      "loss": 0.0083,
      "step": 1374
    },
    {
      "epoch": 0.5272239263803681,
      "grad_norm": 5.445145130157471,
      "learning_rate": 2.6438005965061778e-05,
      "loss": 0.0729,
      "step": 1375
    },
    {
      "epoch": 0.5276073619631901,
      "grad_norm": 53.28449249267578,
      "learning_rate": 2.641670217298679e-05,
      "loss": 0.8887,
      "step": 1376
    },
    {
      "epoch": 0.5279907975460123,
      "grad_norm": 135.62582397460938,
      "learning_rate": 2.63953983809118e-05,
      "loss": 2.7295,
      "step": 1377
    },
    {
      "epoch": 0.5283742331288344,
      "grad_norm": 1.8507696390151978,
      "learning_rate": 2.6374094588836813e-05,
      "loss": 0.0066,
      "step": 1378
    },
    {
      "epoch": 0.5287576687116564,
      "grad_norm": 3.9919838905334473,
      "learning_rate": 2.6352790796761822e-05,
      "loss": 0.03,
      "step": 1379
    },
    {
      "epoch": 0.5291411042944786,
      "grad_norm": 10.276567459106445,
      "learning_rate": 2.6331487004686835e-05,
      "loss": 0.4624,
      "step": 1380
    },
    {
      "epoch": 0.5295245398773006,
      "grad_norm": 16.75113296508789,
      "learning_rate": 2.6310183212611844e-05,
      "loss": 0.1008,
      "step": 1381
    },
    {
      "epoch": 0.5299079754601227,
      "grad_norm": 5.067356109619141,
      "learning_rate": 2.6288879420536857e-05,
      "loss": 0.0235,
      "step": 1382
    },
    {
      "epoch": 0.5302914110429447,
      "grad_norm": 1.2853585481643677,
      "learning_rate": 2.6267575628461866e-05,
      "loss": 0.0039,
      "step": 1383
    },
    {
      "epoch": 0.5306748466257669,
      "grad_norm": 60.08701705932617,
      "learning_rate": 2.6246271836386875e-05,
      "loss": 0.8955,
      "step": 1384
    },
    {
      "epoch": 0.531058282208589,
      "grad_norm": 1.2853145599365234,
      "learning_rate": 2.6224968044311888e-05,
      "loss": 0.0051,
      "step": 1385
    },
    {
      "epoch": 0.531441717791411,
      "grad_norm": 63.4505615234375,
      "learning_rate": 2.6203664252236897e-05,
      "loss": 0.8331,
      "step": 1386
    },
    {
      "epoch": 0.5318251533742331,
      "grad_norm": 14.749077796936035,
      "learning_rate": 2.618236046016191e-05,
      "loss": 0.4832,
      "step": 1387
    },
    {
      "epoch": 0.5322085889570553,
      "grad_norm": 13.824406623840332,
      "learning_rate": 2.616105666808692e-05,
      "loss": 0.0809,
      "step": 1388
    },
    {
      "epoch": 0.5325920245398773,
      "grad_norm": 2.1554923057556152,
      "learning_rate": 2.6139752876011932e-05,
      "loss": 0.0092,
      "step": 1389
    },
    {
      "epoch": 0.5329754601226994,
      "grad_norm": 78.30574798583984,
      "learning_rate": 2.611844908393694e-05,
      "loss": 1.3371,
      "step": 1390
    },
    {
      "epoch": 0.5333588957055214,
      "grad_norm": 3.4257280826568604,
      "learning_rate": 2.609714529186195e-05,
      "loss": 0.0098,
      "step": 1391
    },
    {
      "epoch": 0.5337423312883436,
      "grad_norm": 16.018634796142578,
      "learning_rate": 2.6075841499786963e-05,
      "loss": 0.1732,
      "step": 1392
    },
    {
      "epoch": 0.5341257668711656,
      "grad_norm": 119.55533599853516,
      "learning_rate": 2.6054537707711973e-05,
      "loss": 0.752,
      "step": 1393
    },
    {
      "epoch": 0.5345092024539877,
      "grad_norm": 2.6413443088531494,
      "learning_rate": 2.6033233915636985e-05,
      "loss": 0.0117,
      "step": 1394
    },
    {
      "epoch": 0.5348926380368099,
      "grad_norm": 0.46682795882225037,
      "learning_rate": 2.6011930123561995e-05,
      "loss": 0.0039,
      "step": 1395
    },
    {
      "epoch": 0.5352760736196319,
      "grad_norm": 79.32655334472656,
      "learning_rate": 2.5990626331487007e-05,
      "loss": 0.8901,
      "step": 1396
    },
    {
      "epoch": 0.535659509202454,
      "grad_norm": 107.98719024658203,
      "learning_rate": 2.5969322539412017e-05,
      "loss": 1.2768,
      "step": 1397
    },
    {
      "epoch": 0.536042944785276,
      "grad_norm": 83.03223419189453,
      "learning_rate": 2.5948018747337026e-05,
      "loss": 1.0703,
      "step": 1398
    },
    {
      "epoch": 0.5364263803680982,
      "grad_norm": 2.588785171508789,
      "learning_rate": 2.592671495526204e-05,
      "loss": 0.0098,
      "step": 1399
    },
    {
      "epoch": 0.5368098159509203,
      "grad_norm": 0.832665741443634,
      "learning_rate": 2.5905411163187048e-05,
      "loss": 0.0029,
      "step": 1400
    },
    {
      "epoch": 0.5371932515337423,
      "grad_norm": 8.87395191192627,
      "learning_rate": 2.588410737111206e-05,
      "loss": 0.0303,
      "step": 1401
    },
    {
      "epoch": 0.5375766871165644,
      "grad_norm": 97.84419250488281,
      "learning_rate": 2.586280357903707e-05,
      "loss": 1.2237,
      "step": 1402
    },
    {
      "epoch": 0.5379601226993865,
      "grad_norm": 0.29744938015937805,
      "learning_rate": 2.5841499786962083e-05,
      "loss": 0.0024,
      "step": 1403
    },
    {
      "epoch": 0.5383435582822086,
      "grad_norm": 0.6852608919143677,
      "learning_rate": 2.5820195994887092e-05,
      "loss": 0.002,
      "step": 1404
    },
    {
      "epoch": 0.5387269938650306,
      "grad_norm": 53.20299530029297,
      "learning_rate": 2.57988922028121e-05,
      "loss": 0.9683,
      "step": 1405
    },
    {
      "epoch": 0.5391104294478528,
      "grad_norm": 2.4528932571411133,
      "learning_rate": 2.5777588410737114e-05,
      "loss": 0.0096,
      "step": 1406
    },
    {
      "epoch": 0.5394938650306749,
      "grad_norm": 0.36206158995628357,
      "learning_rate": 2.5756284618662123e-05,
      "loss": 0.0019,
      "step": 1407
    },
    {
      "epoch": 0.5398773006134969,
      "grad_norm": 0.45718395709991455,
      "learning_rate": 2.5734980826587136e-05,
      "loss": 0.0025,
      "step": 1408
    },
    {
      "epoch": 0.540260736196319,
      "grad_norm": 83.47513580322266,
      "learning_rate": 2.5713677034512145e-05,
      "loss": 6.1074,
      "step": 1409
    },
    {
      "epoch": 0.5406441717791411,
      "grad_norm": 199.39964294433594,
      "learning_rate": 2.5692373242437158e-05,
      "loss": 3.4777,
      "step": 1410
    },
    {
      "epoch": 0.5410276073619632,
      "grad_norm": 51.30986785888672,
      "learning_rate": 2.5671069450362167e-05,
      "loss": 0.8536,
      "step": 1411
    },
    {
      "epoch": 0.5414110429447853,
      "grad_norm": 7.736695766448975,
      "learning_rate": 2.5649765658287177e-05,
      "loss": 0.4246,
      "step": 1412
    },
    {
      "epoch": 0.5417944785276073,
      "grad_norm": 4.35651159286499,
      "learning_rate": 2.562846186621219e-05,
      "loss": 0.0247,
      "step": 1413
    },
    {
      "epoch": 0.5421779141104295,
      "grad_norm": 8.150458335876465,
      "learning_rate": 2.56071580741372e-05,
      "loss": 0.0403,
      "step": 1414
    },
    {
      "epoch": 0.5425613496932515,
      "grad_norm": 1.985188603401184,
      "learning_rate": 2.558585428206221e-05,
      "loss": 0.0082,
      "step": 1415
    },
    {
      "epoch": 0.5429447852760736,
      "grad_norm": 2.916771411895752,
      "learning_rate": 2.556455048998722e-05,
      "loss": 0.0088,
      "step": 1416
    },
    {
      "epoch": 0.5433282208588958,
      "grad_norm": 113.31708526611328,
      "learning_rate": 2.5543246697912233e-05,
      "loss": 1.5412,
      "step": 1417
    },
    {
      "epoch": 0.5437116564417178,
      "grad_norm": 138.85861206054688,
      "learning_rate": 2.5521942905837243e-05,
      "loss": 1.7139,
      "step": 1418
    },
    {
      "epoch": 0.5440950920245399,
      "grad_norm": 8.595303535461426,
      "learning_rate": 2.5500639113762252e-05,
      "loss": 0.0212,
      "step": 1419
    },
    {
      "epoch": 0.5444785276073619,
      "grad_norm": 245.66258239746094,
      "learning_rate": 2.5479335321687258e-05,
      "loss": 1.1719,
      "step": 1420
    },
    {
      "epoch": 0.5448619631901841,
      "grad_norm": 20.04214859008789,
      "learning_rate": 2.545803152961227e-05,
      "loss": 0.1265,
      "step": 1421
    },
    {
      "epoch": 0.5452453987730062,
      "grad_norm": 8.85911750793457,
      "learning_rate": 2.543672773753728e-05,
      "loss": 0.4402,
      "step": 1422
    },
    {
      "epoch": 0.5456288343558282,
      "grad_norm": 13.186134338378906,
      "learning_rate": 2.5415423945462293e-05,
      "loss": 0.0499,
      "step": 1423
    },
    {
      "epoch": 0.5460122699386503,
      "grad_norm": 55.921443939208984,
      "learning_rate": 2.5394120153387302e-05,
      "loss": 0.8443,
      "step": 1424
    },
    {
      "epoch": 0.5463957055214724,
      "grad_norm": 0.4546794295310974,
      "learning_rate": 2.537281636131231e-05,
      "loss": 0.0023,
      "step": 1425
    },
    {
      "epoch": 0.5467791411042945,
      "grad_norm": 288.71466064453125,
      "learning_rate": 2.5351512569237324e-05,
      "loss": 5.0381,
      "step": 1426
    },
    {
      "epoch": 0.5471625766871165,
      "grad_norm": 128.4453582763672,
      "learning_rate": 2.5330208777162333e-05,
      "loss": 5.7852,
      "step": 1427
    },
    {
      "epoch": 0.5475460122699386,
      "grad_norm": 3.204813003540039,
      "learning_rate": 2.5308904985087346e-05,
      "loss": 0.0066,
      "step": 1428
    },
    {
      "epoch": 0.5479294478527608,
      "grad_norm": 93.51945495605469,
      "learning_rate": 2.5287601193012355e-05,
      "loss": 1.4854,
      "step": 1429
    },
    {
      "epoch": 0.5483128834355828,
      "grad_norm": 300.50006103515625,
      "learning_rate": 2.5266297400937368e-05,
      "loss": 5.624,
      "step": 1430
    },
    {
      "epoch": 0.5486963190184049,
      "grad_norm": 52.6341438293457,
      "learning_rate": 2.5244993608862377e-05,
      "loss": 0.7945,
      "step": 1431
    },
    {
      "epoch": 0.549079754601227,
      "grad_norm": 1.61876380443573,
      "learning_rate": 2.5223689816787386e-05,
      "loss": 0.0069,
      "step": 1432
    },
    {
      "epoch": 0.5494631901840491,
      "grad_norm": 157.72666931152344,
      "learning_rate": 2.52023860247124e-05,
      "loss": 5.9502,
      "step": 1433
    },
    {
      "epoch": 0.5498466257668712,
      "grad_norm": 9.354452133178711,
      "learning_rate": 2.518108223263741e-05,
      "loss": 0.0192,
      "step": 1434
    },
    {
      "epoch": 0.5502300613496932,
      "grad_norm": 0.2993928790092468,
      "learning_rate": 2.515977844056242e-05,
      "loss": 0.0025,
      "step": 1435
    },
    {
      "epoch": 0.5506134969325154,
      "grad_norm": 136.23651123046875,
      "learning_rate": 2.513847464848743e-05,
      "loss": 1.8184,
      "step": 1436
    },
    {
      "epoch": 0.5509969325153374,
      "grad_norm": 1.0809216499328613,
      "learning_rate": 2.5117170856412443e-05,
      "loss": 0.0021,
      "step": 1437
    },
    {
      "epoch": 0.5513803680981595,
      "grad_norm": 39.01406478881836,
      "learning_rate": 2.5095867064337453e-05,
      "loss": 0.5596,
      "step": 1438
    },
    {
      "epoch": 0.5517638036809815,
      "grad_norm": 0.9668958187103271,
      "learning_rate": 2.5074563272262462e-05,
      "loss": 0.0034,
      "step": 1439
    },
    {
      "epoch": 0.5521472392638037,
      "grad_norm": 0.5503262877464294,
      "learning_rate": 2.5053259480187475e-05,
      "loss": 0.0023,
      "step": 1440
    },
    {
      "epoch": 0.5525306748466258,
      "grad_norm": 2.4875385761260986,
      "learning_rate": 2.5031955688112484e-05,
      "loss": 0.009,
      "step": 1441
    },
    {
      "epoch": 0.5529141104294478,
      "grad_norm": 0.723963737487793,
      "learning_rate": 2.5010651896037497e-05,
      "loss": 0.004,
      "step": 1442
    },
    {
      "epoch": 0.55329754601227,
      "grad_norm": 0.2790910005569458,
      "learning_rate": 2.4989348103962506e-05,
      "loss": 0.0018,
      "step": 1443
    },
    {
      "epoch": 0.553680981595092,
      "grad_norm": 1.9345526695251465,
      "learning_rate": 2.496804431188752e-05,
      "loss": 0.0034,
      "step": 1444
    },
    {
      "epoch": 0.5540644171779141,
      "grad_norm": 1.0085608959197998,
      "learning_rate": 2.4946740519812528e-05,
      "loss": 0.0036,
      "step": 1445
    },
    {
      "epoch": 0.5544478527607362,
      "grad_norm": 10.536890029907227,
      "learning_rate": 2.492543672773754e-05,
      "loss": 0.0603,
      "step": 1446
    },
    {
      "epoch": 0.5548312883435583,
      "grad_norm": 30.43031120300293,
      "learning_rate": 2.490413293566255e-05,
      "loss": 0.9595,
      "step": 1447
    },
    {
      "epoch": 0.5552147239263804,
      "grad_norm": 9.614569664001465,
      "learning_rate": 2.488282914358756e-05,
      "loss": 0.4419,
      "step": 1448
    },
    {
      "epoch": 0.5555981595092024,
      "grad_norm": 21.122318267822266,
      "learning_rate": 2.4861525351512572e-05,
      "loss": 0.5918,
      "step": 1449
    },
    {
      "epoch": 0.5559815950920245,
      "grad_norm": 174.86630249023438,
      "learning_rate": 2.484022155943758e-05,
      "loss": 5.9844,
      "step": 1450
    },
    {
      "epoch": 0.5563650306748467,
      "grad_norm": 123.41751861572266,
      "learning_rate": 2.4818917767362594e-05,
      "loss": 2.4866,
      "step": 1451
    },
    {
      "epoch": 0.5567484662576687,
      "grad_norm": 76.07979583740234,
      "learning_rate": 2.4797613975287603e-05,
      "loss": 0.8784,
      "step": 1452
    },
    {
      "epoch": 0.5571319018404908,
      "grad_norm": 20.474328994750977,
      "learning_rate": 2.4776310183212616e-05,
      "loss": 0.1867,
      "step": 1453
    },
    {
      "epoch": 0.5575153374233128,
      "grad_norm": 29.993284225463867,
      "learning_rate": 2.4755006391137622e-05,
      "loss": 0.5933,
      "step": 1454
    },
    {
      "epoch": 0.557898773006135,
      "grad_norm": 179.69223022460938,
      "learning_rate": 2.4733702599062635e-05,
      "loss": 5.0938,
      "step": 1455
    },
    {
      "epoch": 0.558282208588957,
      "grad_norm": 4.559651851654053,
      "learning_rate": 2.4712398806987644e-05,
      "loss": 0.0206,
      "step": 1456
    },
    {
      "epoch": 0.5586656441717791,
      "grad_norm": 56.0554084777832,
      "learning_rate": 2.4691095014912653e-05,
      "loss": 0.8277,
      "step": 1457
    },
    {
      "epoch": 0.5590490797546013,
      "grad_norm": 44.82016372680664,
      "learning_rate": 2.4669791222837666e-05,
      "loss": 0.9175,
      "step": 1458
    },
    {
      "epoch": 0.5594325153374233,
      "grad_norm": 22.089797973632812,
      "learning_rate": 2.4648487430762675e-05,
      "loss": 0.5376,
      "step": 1459
    },
    {
      "epoch": 0.5598159509202454,
      "grad_norm": 34.83670425415039,
      "learning_rate": 2.4627183638687688e-05,
      "loss": 0.5234,
      "step": 1460
    },
    {
      "epoch": 0.5601993865030674,
      "grad_norm": 63.4261589050293,
      "learning_rate": 2.4605879846612697e-05,
      "loss": 1.0977,
      "step": 1461
    },
    {
      "epoch": 0.5605828220858896,
      "grad_norm": 49.81941604614258,
      "learning_rate": 2.458457605453771e-05,
      "loss": 0.7837,
      "step": 1462
    },
    {
      "epoch": 0.5609662576687117,
      "grad_norm": 22.516889572143555,
      "learning_rate": 2.456327226246272e-05,
      "loss": 0.1992,
      "step": 1463
    },
    {
      "epoch": 0.5613496932515337,
      "grad_norm": 0.19163483381271362,
      "learning_rate": 2.454196847038773e-05,
      "loss": 0.0019,
      "step": 1464
    },
    {
      "epoch": 0.5617331288343558,
      "grad_norm": 0.478487104177475,
      "learning_rate": 2.452066467831274e-05,
      "loss": 0.0027,
      "step": 1465
    },
    {
      "epoch": 0.5621165644171779,
      "grad_norm": 0.2832421362400055,
      "learning_rate": 2.449936088623775e-05,
      "loss": 0.0017,
      "step": 1466
    },
    {
      "epoch": 0.5625,
      "grad_norm": 1.0812827348709106,
      "learning_rate": 2.4478057094162763e-05,
      "loss": 0.0055,
      "step": 1467
    },
    {
      "epoch": 0.5628834355828221,
      "grad_norm": 128.22760009765625,
      "learning_rate": 2.4456753302087773e-05,
      "loss": 2.5939,
      "step": 1468
    },
    {
      "epoch": 0.5632668711656442,
      "grad_norm": 87.38990020751953,
      "learning_rate": 2.4435449510012785e-05,
      "loss": 1.8282,
      "step": 1469
    },
    {
      "epoch": 0.5636503067484663,
      "grad_norm": 1.4836554527282715,
      "learning_rate": 2.4414145717937795e-05,
      "loss": 0.0054,
      "step": 1470
    },
    {
      "epoch": 0.5640337423312883,
      "grad_norm": 7.705040454864502,
      "learning_rate": 2.4392841925862804e-05,
      "loss": 0.0478,
      "step": 1471
    },
    {
      "epoch": 0.5644171779141104,
      "grad_norm": 50.90473556518555,
      "learning_rate": 2.4371538133787817e-05,
      "loss": 0.6255,
      "step": 1472
    },
    {
      "epoch": 0.5648006134969326,
      "grad_norm": 111.48176574707031,
      "learning_rate": 2.4350234341712826e-05,
      "loss": 2.2913,
      "step": 1473
    },
    {
      "epoch": 0.5651840490797546,
      "grad_norm": 140.29307556152344,
      "learning_rate": 2.432893054963784e-05,
      "loss": 3.7265,
      "step": 1474
    },
    {
      "epoch": 0.5655674846625767,
      "grad_norm": 1.5388623476028442,
      "learning_rate": 2.4307626757562848e-05,
      "loss": 0.006,
      "step": 1475
    },
    {
      "epoch": 0.5659509202453987,
      "grad_norm": 19.834774017333984,
      "learning_rate": 2.428632296548786e-05,
      "loss": 0.5942,
      "step": 1476
    },
    {
      "epoch": 0.5663343558282209,
      "grad_norm": 156.98837280273438,
      "learning_rate": 2.4265019173412866e-05,
      "loss": 6.3779,
      "step": 1477
    },
    {
      "epoch": 0.566717791411043,
      "grad_norm": 0.3030416965484619,
      "learning_rate": 2.424371538133788e-05,
      "loss": 0.0015,
      "step": 1478
    },
    {
      "epoch": 0.567101226993865,
      "grad_norm": 10.810173034667969,
      "learning_rate": 2.422241158926289e-05,
      "loss": 0.1487,
      "step": 1479
    },
    {
      "epoch": 0.5674846625766872,
      "grad_norm": 42.8617057800293,
      "learning_rate": 2.42011077971879e-05,
      "loss": 0.6021,
      "step": 1480
    },
    {
      "epoch": 0.5678680981595092,
      "grad_norm": 11.066485404968262,
      "learning_rate": 2.417980400511291e-05,
      "loss": 0.0384,
      "step": 1481
    },
    {
      "epoch": 0.5682515337423313,
      "grad_norm": 14.02347469329834,
      "learning_rate": 2.415850021303792e-05,
      "loss": 0.428,
      "step": 1482
    },
    {
      "epoch": 0.5686349693251533,
      "grad_norm": 105.50628662109375,
      "learning_rate": 2.4137196420962933e-05,
      "loss": 1.8076,
      "step": 1483
    },
    {
      "epoch": 0.5690184049079755,
      "grad_norm": 0.9760513305664062,
      "learning_rate": 2.4115892628887942e-05,
      "loss": 0.0036,
      "step": 1484
    },
    {
      "epoch": 0.5694018404907976,
      "grad_norm": 1.0313464403152466,
      "learning_rate": 2.4094588836812955e-05,
      "loss": 0.0047,
      "step": 1485
    },
    {
      "epoch": 0.5697852760736196,
      "grad_norm": 16.453495025634766,
      "learning_rate": 2.4073285044737964e-05,
      "loss": 0.4883,
      "step": 1486
    },
    {
      "epoch": 0.5701687116564417,
      "grad_norm": 0.4924452304840088,
      "learning_rate": 2.4051981252662977e-05,
      "loss": 0.0016,
      "step": 1487
    },
    {
      "epoch": 0.5705521472392638,
      "grad_norm": 30.859689712524414,
      "learning_rate": 2.4030677460587986e-05,
      "loss": 0.6582,
      "step": 1488
    },
    {
      "epoch": 0.5709355828220859,
      "grad_norm": 0.24691864848136902,
      "learning_rate": 2.4009373668512995e-05,
      "loss": 0.0016,
      "step": 1489
    },
    {
      "epoch": 0.571319018404908,
      "grad_norm": 5.21722412109375,
      "learning_rate": 2.3988069876438008e-05,
      "loss": 0.0255,
      "step": 1490
    },
    {
      "epoch": 0.57170245398773,
      "grad_norm": 13.898541450500488,
      "learning_rate": 2.3966766084363017e-05,
      "loss": 0.4402,
      "step": 1491
    },
    {
      "epoch": 0.5720858895705522,
      "grad_norm": 11.03413200378418,
      "learning_rate": 2.394546229228803e-05,
      "loss": 0.1442,
      "step": 1492
    },
    {
      "epoch": 0.5724693251533742,
      "grad_norm": 131.2899169921875,
      "learning_rate": 2.392415850021304e-05,
      "loss": 1.1788,
      "step": 1493
    },
    {
      "epoch": 0.5728527607361963,
      "grad_norm": 0.2610747218132019,
      "learning_rate": 2.3902854708138052e-05,
      "loss": 0.0022,
      "step": 1494
    },
    {
      "epoch": 0.5732361963190185,
      "grad_norm": 85.44371032714844,
      "learning_rate": 2.388155091606306e-05,
      "loss": 0.7159,
      "step": 1495
    },
    {
      "epoch": 0.5736196319018405,
      "grad_norm": 99.33592224121094,
      "learning_rate": 2.386024712398807e-05,
      "loss": 1.1123,
      "step": 1496
    },
    {
      "epoch": 0.5740030674846626,
      "grad_norm": 6.815636157989502,
      "learning_rate": 2.3838943331913083e-05,
      "loss": 0.2409,
      "step": 1497
    },
    {
      "epoch": 0.5743865030674846,
      "grad_norm": 61.178688049316406,
      "learning_rate": 2.3817639539838093e-05,
      "loss": 0.7031,
      "step": 1498
    },
    {
      "epoch": 0.5747699386503068,
      "grad_norm": 2.902374744415283,
      "learning_rate": 2.3796335747763102e-05,
      "loss": 0.0147,
      "step": 1499
    },
    {
      "epoch": 0.5751533742331288,
      "grad_norm": 3.590909481048584,
      "learning_rate": 2.377503195568811e-05,
      "loss": 0.0159,
      "step": 1500
    },
    {
      "epoch": 0.5755368098159509,
      "grad_norm": 0.2199438512325287,
      "learning_rate": 2.3753728163613124e-05,
      "loss": 0.0016,
      "step": 1501
    },
    {
      "epoch": 0.575920245398773,
      "grad_norm": 39.01709747314453,
      "learning_rate": 2.3732424371538133e-05,
      "loss": 0.6152,
      "step": 1502
    },
    {
      "epoch": 0.5763036809815951,
      "grad_norm": 4.145489692687988,
      "learning_rate": 2.3711120579463146e-05,
      "loss": 0.3848,
      "step": 1503
    },
    {
      "epoch": 0.5766871165644172,
      "grad_norm": 1.525215744972229,
      "learning_rate": 2.3689816787388155e-05,
      "loss": 0.0019,
      "step": 1504
    },
    {
      "epoch": 0.5770705521472392,
      "grad_norm": 36.02094650268555,
      "learning_rate": 2.3668512995313168e-05,
      "loss": 0.6113,
      "step": 1505
    },
    {
      "epoch": 0.5774539877300614,
      "grad_norm": 56.19095993041992,
      "learning_rate": 2.3647209203238177e-05,
      "loss": 0.5518,
      "step": 1506
    },
    {
      "epoch": 0.5778374233128835,
      "grad_norm": 1.4983949661254883,
      "learning_rate": 2.3625905411163186e-05,
      "loss": 0.0049,
      "step": 1507
    },
    {
      "epoch": 0.5782208588957055,
      "grad_norm": 58.191192626953125,
      "learning_rate": 2.36046016190882e-05,
      "loss": 0.9834,
      "step": 1508
    },
    {
      "epoch": 0.5786042944785276,
      "grad_norm": 9.700529098510742,
      "learning_rate": 2.358329782701321e-05,
      "loss": 0.0304,
      "step": 1509
    },
    {
      "epoch": 0.5789877300613497,
      "grad_norm": 75.30133056640625,
      "learning_rate": 2.356199403493822e-05,
      "loss": 0.7432,
      "step": 1510
    },
    {
      "epoch": 0.5793711656441718,
      "grad_norm": 16.457170486450195,
      "learning_rate": 2.354069024286323e-05,
      "loss": 0.1741,
      "step": 1511
    },
    {
      "epoch": 0.5797546012269938,
      "grad_norm": 12.593084335327148,
      "learning_rate": 2.3519386450788243e-05,
      "loss": 0.0602,
      "step": 1512
    },
    {
      "epoch": 0.5801380368098159,
      "grad_norm": 2.239633083343506,
      "learning_rate": 2.3498082658713253e-05,
      "loss": 0.0063,
      "step": 1513
    },
    {
      "epoch": 0.5805214723926381,
      "grad_norm": 69.36461639404297,
      "learning_rate": 2.3476778866638262e-05,
      "loss": 0.7149,
      "step": 1514
    },
    {
      "epoch": 0.5809049079754601,
      "grad_norm": 186.72915649414062,
      "learning_rate": 2.3455475074563275e-05,
      "loss": 6.3867,
      "step": 1515
    },
    {
      "epoch": 0.5812883435582822,
      "grad_norm": 205.23544311523438,
      "learning_rate": 2.3434171282488284e-05,
      "loss": 1.1739,
      "step": 1516
    },
    {
      "epoch": 0.5816717791411042,
      "grad_norm": 23.364177703857422,
      "learning_rate": 2.3412867490413297e-05,
      "loss": 0.071,
      "step": 1517
    },
    {
      "epoch": 0.5820552147239264,
      "grad_norm": 37.77939224243164,
      "learning_rate": 2.3391563698338306e-05,
      "loss": 0.7114,
      "step": 1518
    },
    {
      "epoch": 0.5824386503067485,
      "grad_norm": 14.460637092590332,
      "learning_rate": 2.337025990626332e-05,
      "loss": 0.0953,
      "step": 1519
    },
    {
      "epoch": 0.5828220858895705,
      "grad_norm": 12.150667190551758,
      "learning_rate": 2.3348956114188328e-05,
      "loss": 0.0507,
      "step": 1520
    },
    {
      "epoch": 0.5832055214723927,
      "grad_norm": 6.960376739501953,
      "learning_rate": 2.3327652322113337e-05,
      "loss": 0.0439,
      "step": 1521
    },
    {
      "epoch": 0.5835889570552147,
      "grad_norm": 2.1229488849639893,
      "learning_rate": 2.3306348530038346e-05,
      "loss": 0.0077,
      "step": 1522
    },
    {
      "epoch": 0.5839723926380368,
      "grad_norm": 0.3395184278488159,
      "learning_rate": 2.3285044737963356e-05,
      "loss": 0.0015,
      "step": 1523
    },
    {
      "epoch": 0.5843558282208589,
      "grad_norm": 0.09832275658845901,
      "learning_rate": 2.326374094588837e-05,
      "loss": 0.0009,
      "step": 1524
    },
    {
      "epoch": 0.584739263803681,
      "grad_norm": 0.6131349802017212,
      "learning_rate": 2.3242437153813378e-05,
      "loss": 0.0028,
      "step": 1525
    },
    {
      "epoch": 0.5851226993865031,
      "grad_norm": 0.6060605049133301,
      "learning_rate": 2.322113336173839e-05,
      "loss": 0.003,
      "step": 1526
    },
    {
      "epoch": 0.5855061349693251,
      "grad_norm": 17.108949661254883,
      "learning_rate": 2.31998295696634e-05,
      "loss": 0.1991,
      "step": 1527
    },
    {
      "epoch": 0.5858895705521472,
      "grad_norm": 1.113153338432312,
      "learning_rate": 2.3178525777588413e-05,
      "loss": 0.0061,
      "step": 1528
    },
    {
      "epoch": 0.5862730061349694,
      "grad_norm": 28.444774627685547,
      "learning_rate": 2.3157221985513422e-05,
      "loss": 0.522,
      "step": 1529
    },
    {
      "epoch": 0.5866564417177914,
      "grad_norm": 12.834977149963379,
      "learning_rate": 2.313591819343843e-05,
      "loss": 0.1884,
      "step": 1530
    },
    {
      "epoch": 0.5870398773006135,
      "grad_norm": 0.5456647872924805,
      "learning_rate": 2.3114614401363444e-05,
      "loss": 0.0032,
      "step": 1531
    },
    {
      "epoch": 0.5874233128834356,
      "grad_norm": 11.738081932067871,
      "learning_rate": 2.3093310609288453e-05,
      "loss": 0.4417,
      "step": 1532
    },
    {
      "epoch": 0.5878067484662577,
      "grad_norm": 16.152618408203125,
      "learning_rate": 2.3072006817213466e-05,
      "loss": 0.5591,
      "step": 1533
    },
    {
      "epoch": 0.5881901840490797,
      "grad_norm": 18.996780395507812,
      "learning_rate": 2.3050703025138475e-05,
      "loss": 0.1611,
      "step": 1534
    },
    {
      "epoch": 0.5885736196319018,
      "grad_norm": 0.45010751485824585,
      "learning_rate": 2.3029399233063488e-05,
      "loss": 0.0021,
      "step": 1535
    },
    {
      "epoch": 0.588957055214724,
      "grad_norm": 0.9772094488143921,
      "learning_rate": 2.3008095440988497e-05,
      "loss": 0.005,
      "step": 1536
    },
    {
      "epoch": 0.589340490797546,
      "grad_norm": 10.670825004577637,
      "learning_rate": 2.298679164891351e-05,
      "loss": 0.0729,
      "step": 1537
    },
    {
      "epoch": 0.5897239263803681,
      "grad_norm": 12.155583381652832,
      "learning_rate": 2.296548785683852e-05,
      "loss": 0.1005,
      "step": 1538
    },
    {
      "epoch": 0.5901073619631901,
      "grad_norm": 62.7037239074707,
      "learning_rate": 2.294418406476353e-05,
      "loss": 0.9976,
      "step": 1539
    },
    {
      "epoch": 0.5904907975460123,
      "grad_norm": 128.63902282714844,
      "learning_rate": 2.292288027268854e-05,
      "loss": 1.2705,
      "step": 1540
    },
    {
      "epoch": 0.5908742331288344,
      "grad_norm": 97.0986099243164,
      "learning_rate": 2.290157648061355e-05,
      "loss": 1.1673,
      "step": 1541
    },
    {
      "epoch": 0.5912576687116564,
      "grad_norm": 0.7080488801002502,
      "learning_rate": 2.2880272688538563e-05,
      "loss": 0.0027,
      "step": 1542
    },
    {
      "epoch": 0.5916411042944786,
      "grad_norm": 5.879316329956055,
      "learning_rate": 2.2858968896463573e-05,
      "loss": 0.0479,
      "step": 1543
    },
    {
      "epoch": 0.5920245398773006,
      "grad_norm": 187.53341674804688,
      "learning_rate": 2.2837665104388585e-05,
      "loss": 3.2312,
      "step": 1544
    },
    {
      "epoch": 0.5924079754601227,
      "grad_norm": 73.92506408691406,
      "learning_rate": 2.281636131231359e-05,
      "loss": 1.2617,
      "step": 1545
    },
    {
      "epoch": 0.5927914110429447,
      "grad_norm": 98.77815246582031,
      "learning_rate": 2.2795057520238604e-05,
      "loss": 1.798,
      "step": 1546
    },
    {
      "epoch": 0.5931748466257669,
      "grad_norm": 13.581368446350098,
      "learning_rate": 2.2773753728163613e-05,
      "loss": 0.0398,
      "step": 1547
    },
    {
      "epoch": 0.593558282208589,
      "grad_norm": 0.6549582481384277,
      "learning_rate": 2.2752449936088622e-05,
      "loss": 0.003,
      "step": 1548
    },
    {
      "epoch": 0.593941717791411,
      "grad_norm": 203.76071166992188,
      "learning_rate": 2.2731146144013635e-05,
      "loss": 2.1231,
      "step": 1549
    },
    {
      "epoch": 0.5943251533742331,
      "grad_norm": 0.31085968017578125,
      "learning_rate": 2.2709842351938644e-05,
      "loss": 0.0024,
      "step": 1550
    },
    {
      "epoch": 0.5947085889570553,
      "grad_norm": 0.24026542901992798,
      "learning_rate": 2.2688538559863657e-05,
      "loss": 0.0011,
      "step": 1551
    },
    {
      "epoch": 0.5950920245398773,
      "grad_norm": 134.61109924316406,
      "learning_rate": 2.2667234767788666e-05,
      "loss": 1.0635,
      "step": 1552
    },
    {
      "epoch": 0.5954754601226994,
      "grad_norm": 51.36643981933594,
      "learning_rate": 2.264593097571368e-05,
      "loss": 1.0771,
      "step": 1553
    },
    {
      "epoch": 0.5958588957055214,
      "grad_norm": 89.78507995605469,
      "learning_rate": 2.262462718363869e-05,
      "loss": 0.814,
      "step": 1554
    },
    {
      "epoch": 0.5962423312883436,
      "grad_norm": 152.59800720214844,
      "learning_rate": 2.2603323391563698e-05,
      "loss": 1.3302,
      "step": 1555
    },
    {
      "epoch": 0.5966257668711656,
      "grad_norm": 16.452253341674805,
      "learning_rate": 2.258201959948871e-05,
      "loss": 0.1604,
      "step": 1556
    },
    {
      "epoch": 0.5970092024539877,
      "grad_norm": 1.21446692943573,
      "learning_rate": 2.256071580741372e-05,
      "loss": 0.0063,
      "step": 1557
    },
    {
      "epoch": 0.5973926380368099,
      "grad_norm": 0.23559719324111938,
      "learning_rate": 2.2539412015338732e-05,
      "loss": 0.0012,
      "step": 1558
    },
    {
      "epoch": 0.5977760736196319,
      "grad_norm": 5.033271312713623,
      "learning_rate": 2.2518108223263742e-05,
      "loss": 0.0282,
      "step": 1559
    },
    {
      "epoch": 0.598159509202454,
      "grad_norm": 25.347471237182617,
      "learning_rate": 2.2496804431188755e-05,
      "loss": 0.5996,
      "step": 1560
    },
    {
      "epoch": 0.598542944785276,
      "grad_norm": 4.496798992156982,
      "learning_rate": 2.2475500639113764e-05,
      "loss": 0.0331,
      "step": 1561
    },
    {
      "epoch": 0.5989263803680982,
      "grad_norm": 74.6218490600586,
      "learning_rate": 2.2454196847038773e-05,
      "loss": 0.5161,
      "step": 1562
    },
    {
      "epoch": 0.5993098159509203,
      "grad_norm": 0.9332700967788696,
      "learning_rate": 2.2432893054963786e-05,
      "loss": 0.0054,
      "step": 1563
    },
    {
      "epoch": 0.5996932515337423,
      "grad_norm": 2.685960292816162,
      "learning_rate": 2.2411589262888795e-05,
      "loss": 0.0127,
      "step": 1564
    },
    {
      "epoch": 0.6000766871165644,
      "grad_norm": 3.394599676132202,
      "learning_rate": 2.2390285470813808e-05,
      "loss": 0.0126,
      "step": 1565
    },
    {
      "epoch": 0.6004601226993865,
      "grad_norm": 7.925353050231934,
      "learning_rate": 2.2368981678738817e-05,
      "loss": 0.0096,
      "step": 1566
    },
    {
      "epoch": 0.6008435582822086,
      "grad_norm": 65.36844635009766,
      "learning_rate": 2.2347677886663826e-05,
      "loss": 0.9502,
      "step": 1567
    },
    {
      "epoch": 0.6012269938650306,
      "grad_norm": 1.044142246246338,
      "learning_rate": 2.2326374094588836e-05,
      "loss": 0.0031,
      "step": 1568
    },
    {
      "epoch": 0.6016104294478528,
      "grad_norm": 1.1998786926269531,
      "learning_rate": 2.230507030251385e-05,
      "loss": 0.0049,
      "step": 1569
    },
    {
      "epoch": 0.6019938650306749,
      "grad_norm": 7.696629524230957,
      "learning_rate": 2.2283766510438858e-05,
      "loss": 0.0278,
      "step": 1570
    },
    {
      "epoch": 0.6023773006134969,
      "grad_norm": 2.3067355155944824,
      "learning_rate": 2.226246271836387e-05,
      "loss": 0.0082,
      "step": 1571
    },
    {
      "epoch": 0.602760736196319,
      "grad_norm": 8.440646171569824,
      "learning_rate": 2.224115892628888e-05,
      "loss": 0.2568,
      "step": 1572
    },
    {
      "epoch": 0.6031441717791411,
      "grad_norm": 23.990575790405273,
      "learning_rate": 2.221985513421389e-05,
      "loss": 0.5166,
      "step": 1573
    },
    {
      "epoch": 0.6035276073619632,
      "grad_norm": 72.5779037475586,
      "learning_rate": 2.2198551342138902e-05,
      "loss": 1.2823,
      "step": 1574
    },
    {
      "epoch": 0.6039110429447853,
      "grad_norm": 0.22823019325733185,
      "learning_rate": 2.217724755006391e-05,
      "loss": 0.0019,
      "step": 1575
    },
    {
      "epoch": 0.6042944785276073,
      "grad_norm": 18.47128677368164,
      "learning_rate": 2.2155943757988924e-05,
      "loss": 0.4653,
      "step": 1576
    },
    {
      "epoch": 0.6046779141104295,
      "grad_norm": 1.0686135292053223,
      "learning_rate": 2.2134639965913933e-05,
      "loss": 0.0038,
      "step": 1577
    },
    {
      "epoch": 0.6050613496932515,
      "grad_norm": 34.65068435668945,
      "learning_rate": 2.2113336173838946e-05,
      "loss": 0.6939,
      "step": 1578
    },
    {
      "epoch": 0.6054447852760736,
      "grad_norm": 21.57245635986328,
      "learning_rate": 2.2092032381763955e-05,
      "loss": 0.1171,
      "step": 1579
    },
    {
      "epoch": 0.6058282208588958,
      "grad_norm": 0.4185841381549835,
      "learning_rate": 2.2070728589688964e-05,
      "loss": 0.0025,
      "step": 1580
    },
    {
      "epoch": 0.6062116564417178,
      "grad_norm": 110.56187438964844,
      "learning_rate": 2.2049424797613977e-05,
      "loss": 1.5998,
      "step": 1581
    },
    {
      "epoch": 0.6065950920245399,
      "grad_norm": 38.00716781616211,
      "learning_rate": 2.2028121005538986e-05,
      "loss": 0.6504,
      "step": 1582
    },
    {
      "epoch": 0.6069785276073619,
      "grad_norm": 0.40848639607429504,
      "learning_rate": 2.2006817213464e-05,
      "loss": 0.0018,
      "step": 1583
    },
    {
      "epoch": 0.6073619631901841,
      "grad_norm": 0.4053003787994385,
      "learning_rate": 2.198551342138901e-05,
      "loss": 0.0019,
      "step": 1584
    },
    {
      "epoch": 0.6077453987730062,
      "grad_norm": 3.820105791091919,
      "learning_rate": 2.196420962931402e-05,
      "loss": 0.0084,
      "step": 1585
    },
    {
      "epoch": 0.6081288343558282,
      "grad_norm": 0.306547075510025,
      "learning_rate": 2.194290583723903e-05,
      "loss": 0.0018,
      "step": 1586
    },
    {
      "epoch": 0.6085122699386503,
      "grad_norm": 230.79635620117188,
      "learning_rate": 2.192160204516404e-05,
      "loss": 3.7719,
      "step": 1587
    },
    {
      "epoch": 0.6088957055214724,
      "grad_norm": 52.079227447509766,
      "learning_rate": 2.1900298253089052e-05,
      "loss": 0.7021,
      "step": 1588
    },
    {
      "epoch": 0.6092791411042945,
      "grad_norm": 1.4306854009628296,
      "learning_rate": 2.1878994461014062e-05,
      "loss": 0.0045,
      "step": 1589
    },
    {
      "epoch": 0.6096625766871165,
      "grad_norm": 29.657840728759766,
      "learning_rate": 2.185769066893907e-05,
      "loss": 0.7041,
      "step": 1590
    },
    {
      "epoch": 0.6100460122699386,
      "grad_norm": 0.2373381108045578,
      "learning_rate": 2.183638687686408e-05,
      "loss": 0.0016,
      "step": 1591
    },
    {
      "epoch": 0.6104294478527608,
      "grad_norm": 76.70428466796875,
      "learning_rate": 2.1815083084789093e-05,
      "loss": 2.3907,
      "step": 1592
    },
    {
      "epoch": 0.6108128834355828,
      "grad_norm": 25.90323829650879,
      "learning_rate": 2.1793779292714102e-05,
      "loss": 0.4868,
      "step": 1593
    },
    {
      "epoch": 0.6111963190184049,
      "grad_norm": 3.877763271331787,
      "learning_rate": 2.1772475500639115e-05,
      "loss": 0.0123,
      "step": 1594
    },
    {
      "epoch": 0.611579754601227,
      "grad_norm": 31.096912384033203,
      "learning_rate": 2.1751171708564124e-05,
      "loss": 0.7109,
      "step": 1595
    },
    {
      "epoch": 0.6119631901840491,
      "grad_norm": 306.73468017578125,
      "learning_rate": 2.1729867916489137e-05,
      "loss": 4.0736,
      "step": 1596
    },
    {
      "epoch": 0.6123466257668712,
      "grad_norm": 5.6847028732299805,
      "learning_rate": 2.1708564124414146e-05,
      "loss": 0.2793,
      "step": 1597
    },
    {
      "epoch": 0.6127300613496932,
      "grad_norm": 0.3299265205860138,
      "learning_rate": 2.1687260332339156e-05,
      "loss": 0.0015,
      "step": 1598
    },
    {
      "epoch": 0.6131134969325154,
      "grad_norm": 4.540491580963135,
      "learning_rate": 2.166595654026417e-05,
      "loss": 0.013,
      "step": 1599
    },
    {
      "epoch": 0.6134969325153374,
      "grad_norm": 13.582939147949219,
      "learning_rate": 2.1644652748189178e-05,
      "loss": 0.5747,
      "step": 1600
    },
    {
      "epoch": 0.6138803680981595,
      "grad_norm": 0.24872423708438873,
      "learning_rate": 2.162334895611419e-05,
      "loss": 0.002,
      "step": 1601
    },
    {
      "epoch": 0.6142638036809815,
      "grad_norm": 14.802531242370605,
      "learning_rate": 2.16020451640392e-05,
      "loss": 0.0799,
      "step": 1602
    },
    {
      "epoch": 0.6146472392638037,
      "grad_norm": 0.7813965678215027,
      "learning_rate": 2.1580741371964212e-05,
      "loss": 0.0032,
      "step": 1603
    },
    {
      "epoch": 0.6150306748466258,
      "grad_norm": 137.3957061767578,
      "learning_rate": 2.1559437579889222e-05,
      "loss": 1.9415,
      "step": 1604
    },
    {
      "epoch": 0.6154141104294478,
      "grad_norm": 0.37478020787239075,
      "learning_rate": 2.153813378781423e-05,
      "loss": 0.0012,
      "step": 1605
    },
    {
      "epoch": 0.61579754601227,
      "grad_norm": 313.3479309082031,
      "learning_rate": 2.1516829995739244e-05,
      "loss": 6.2529,
      "step": 1606
    },
    {
      "epoch": 0.616180981595092,
      "grad_norm": 0.25419455766677856,
      "learning_rate": 2.1495526203664253e-05,
      "loss": 0.0014,
      "step": 1607
    },
    {
      "epoch": 0.6165644171779141,
      "grad_norm": 4.846649646759033,
      "learning_rate": 2.1474222411589266e-05,
      "loss": 0.0219,
      "step": 1608
    },
    {
      "epoch": 0.6169478527607362,
      "grad_norm": 37.852149963378906,
      "learning_rate": 2.1452918619514275e-05,
      "loss": 0.5962,
      "step": 1609
    },
    {
      "epoch": 0.6173312883435583,
      "grad_norm": 1.2913243770599365,
      "learning_rate": 2.1431614827439288e-05,
      "loss": 0.003,
      "step": 1610
    },
    {
      "epoch": 0.6177147239263804,
      "grad_norm": 0.35508644580841064,
      "learning_rate": 2.1410311035364297e-05,
      "loss": 0.0011,
      "step": 1611
    },
    {
      "epoch": 0.6180981595092024,
      "grad_norm": 3.2686877250671387,
      "learning_rate": 2.1389007243289306e-05,
      "loss": 0.0142,
      "step": 1612
    },
    {
      "epoch": 0.6184815950920245,
      "grad_norm": 1.7194098234176636,
      "learning_rate": 2.1367703451214316e-05,
      "loss": 0.0058,
      "step": 1613
    },
    {
      "epoch": 0.6188650306748467,
      "grad_norm": 0.6371323466300964,
      "learning_rate": 2.1346399659139325e-05,
      "loss": 0.0014,
      "step": 1614
    },
    {
      "epoch": 0.6192484662576687,
      "grad_norm": 330.1398010253906,
      "learning_rate": 2.1325095867064338e-05,
      "loss": 3.541,
      "step": 1615
    },
    {
      "epoch": 0.6196319018404908,
      "grad_norm": 145.93212890625,
      "learning_rate": 2.1303792074989347e-05,
      "loss": 2.0606,
      "step": 1616
    },
    {
      "epoch": 0.6200153374233128,
      "grad_norm": 8.430885314941406,
      "learning_rate": 2.128248828291436e-05,
      "loss": 0.4036,
      "step": 1617
    },
    {
      "epoch": 0.620398773006135,
      "grad_norm": 22.9518985748291,
      "learning_rate": 2.126118449083937e-05,
      "loss": 0.5459,
      "step": 1618
    },
    {
      "epoch": 0.620782208588957,
      "grad_norm": 4.829712867736816,
      "learning_rate": 2.1239880698764382e-05,
      "loss": 0.031,
      "step": 1619
    },
    {
      "epoch": 0.6211656441717791,
      "grad_norm": 7.079444408416748,
      "learning_rate": 2.121857690668939e-05,
      "loss": 0.0195,
      "step": 1620
    },
    {
      "epoch": 0.6215490797546013,
      "grad_norm": 31.619905471801758,
      "learning_rate": 2.1197273114614404e-05,
      "loss": 0.71,
      "step": 1621
    },
    {
      "epoch": 0.6219325153374233,
      "grad_norm": 176.59059143066406,
      "learning_rate": 2.1175969322539413e-05,
      "loss": 1.6428,
      "step": 1622
    },
    {
      "epoch": 0.6223159509202454,
      "grad_norm": 2.9753220081329346,
      "learning_rate": 2.1154665530464422e-05,
      "loss": 0.0036,
      "step": 1623
    },
    {
      "epoch": 0.6226993865030674,
      "grad_norm": 14.047575950622559,
      "learning_rate": 2.1133361738389435e-05,
      "loss": 0.5679,
      "step": 1624
    },
    {
      "epoch": 0.6230828220858896,
      "grad_norm": 15.567380905151367,
      "learning_rate": 2.1112057946314444e-05,
      "loss": 0.0282,
      "step": 1625
    },
    {
      "epoch": 0.6234662576687117,
      "grad_norm": 0.5999993681907654,
      "learning_rate": 2.1090754154239457e-05,
      "loss": 0.0019,
      "step": 1626
    },
    {
      "epoch": 0.6238496932515337,
      "grad_norm": 48.12718963623047,
      "learning_rate": 2.1069450362164466e-05,
      "loss": 0.7349,
      "step": 1627
    },
    {
      "epoch": 0.6242331288343558,
      "grad_norm": 1.9771808385849,
      "learning_rate": 2.104814657008948e-05,
      "loss": 0.0071,
      "step": 1628
    },
    {
      "epoch": 0.6246165644171779,
      "grad_norm": 0.9278380870819092,
      "learning_rate": 2.102684277801449e-05,
      "loss": 0.0045,
      "step": 1629
    },
    {
      "epoch": 0.625,
      "grad_norm": 0.28498128056526184,
      "learning_rate": 2.1005538985939498e-05,
      "loss": 0.0013,
      "step": 1630
    },
    {
      "epoch": 0.6253834355828221,
      "grad_norm": 177.34898376464844,
      "learning_rate": 2.098423519386451e-05,
      "loss": 2.348,
      "step": 1631
    },
    {
      "epoch": 0.6257668711656442,
      "grad_norm": 60.71388244628906,
      "learning_rate": 2.096293140178952e-05,
      "loss": 0.7481,
      "step": 1632
    },
    {
      "epoch": 0.6261503067484663,
      "grad_norm": 13.655220031738281,
      "learning_rate": 2.0941627609714532e-05,
      "loss": 0.0885,
      "step": 1633
    },
    {
      "epoch": 0.6265337423312883,
      "grad_norm": 205.313232421875,
      "learning_rate": 2.0920323817639542e-05,
      "loss": 4.107,
      "step": 1634
    },
    {
      "epoch": 0.6269171779141104,
      "grad_norm": 2.9884133338928223,
      "learning_rate": 2.089902002556455e-05,
      "loss": 0.0125,
      "step": 1635
    },
    {
      "epoch": 0.6273006134969326,
      "grad_norm": 170.3273162841797,
      "learning_rate": 2.087771623348956e-05,
      "loss": 6.5684,
      "step": 1636
    },
    {
      "epoch": 0.6276840490797546,
      "grad_norm": 14.023516654968262,
      "learning_rate": 2.0856412441414573e-05,
      "loss": 0.1242,
      "step": 1637
    },
    {
      "epoch": 0.6280674846625767,
      "grad_norm": 17.409425735473633,
      "learning_rate": 2.0835108649339582e-05,
      "loss": 0.3997,
      "step": 1638
    },
    {
      "epoch": 0.6284509202453987,
      "grad_norm": 31.173315048217773,
      "learning_rate": 2.0813804857264592e-05,
      "loss": 0.6865,
      "step": 1639
    },
    {
      "epoch": 0.6288343558282209,
      "grad_norm": 1.5950876474380493,
      "learning_rate": 2.0792501065189604e-05,
      "loss": 0.0053,
      "step": 1640
    },
    {
      "epoch": 0.629217791411043,
      "grad_norm": 13.480043411254883,
      "learning_rate": 2.0771197273114614e-05,
      "loss": 0.0521,
      "step": 1641
    },
    {
      "epoch": 0.629601226993865,
      "grad_norm": 14.539470672607422,
      "learning_rate": 2.0749893481039626e-05,
      "loss": 0.1625,
      "step": 1642
    },
    {
      "epoch": 0.6299846625766872,
      "grad_norm": 26.961544036865234,
      "learning_rate": 2.0728589688964636e-05,
      "loss": 0.0735,
      "step": 1643
    },
    {
      "epoch": 0.6303680981595092,
      "grad_norm": 59.08195114135742,
      "learning_rate": 2.070728589688965e-05,
      "loss": 1.04,
      "step": 1644
    },
    {
      "epoch": 0.6307515337423313,
      "grad_norm": 0.13883554935455322,
      "learning_rate": 2.0685982104814658e-05,
      "loss": 0.0013,
      "step": 1645
    },
    {
      "epoch": 0.6311349693251533,
      "grad_norm": 54.861446380615234,
      "learning_rate": 2.0664678312739667e-05,
      "loss": 0.8756,
      "step": 1646
    },
    {
      "epoch": 0.6315184049079755,
      "grad_norm": 0.09707655757665634,
      "learning_rate": 2.064337452066468e-05,
      "loss": 0.0009,
      "step": 1647
    },
    {
      "epoch": 0.6319018404907976,
      "grad_norm": 5.110586643218994,
      "learning_rate": 2.062207072858969e-05,
      "loss": 0.0163,
      "step": 1648
    },
    {
      "epoch": 0.6322852760736196,
      "grad_norm": 17.940378189086914,
      "learning_rate": 2.0600766936514702e-05,
      "loss": 0.4658,
      "step": 1649
    },
    {
      "epoch": 0.6326687116564417,
      "grad_norm": 0.4684121012687683,
      "learning_rate": 2.057946314443971e-05,
      "loss": 0.0021,
      "step": 1650
    },
    {
      "epoch": 0.6330521472392638,
      "grad_norm": 2.591216564178467,
      "learning_rate": 2.0558159352364724e-05,
      "loss": 0.011,
      "step": 1651
    },
    {
      "epoch": 0.6334355828220859,
      "grad_norm": 1.9084800481796265,
      "learning_rate": 2.0536855560289733e-05,
      "loss": 0.0063,
      "step": 1652
    },
    {
      "epoch": 0.633819018404908,
      "grad_norm": 101.3776626586914,
      "learning_rate": 2.0515551768214742e-05,
      "loss": 1.0938,
      "step": 1653
    },
    {
      "epoch": 0.63420245398773,
      "grad_norm": 2.34978985786438,
      "learning_rate": 2.0494247976139755e-05,
      "loss": 0.0095,
      "step": 1654
    },
    {
      "epoch": 0.6345858895705522,
      "grad_norm": 116.3980941772461,
      "learning_rate": 2.0472944184064764e-05,
      "loss": 0.9522,
      "step": 1655
    },
    {
      "epoch": 0.6349693251533742,
      "grad_norm": 48.77947998046875,
      "learning_rate": 2.0451640391989777e-05,
      "loss": 0.6523,
      "step": 1656
    },
    {
      "epoch": 0.6353527607361963,
      "grad_norm": 91.40596771240234,
      "learning_rate": 2.0430336599914783e-05,
      "loss": 1.8175,
      "step": 1657
    },
    {
      "epoch": 0.6357361963190185,
      "grad_norm": 9.864049911499023,
      "learning_rate": 2.0409032807839796e-05,
      "loss": 0.1552,
      "step": 1658
    },
    {
      "epoch": 0.6361196319018405,
      "grad_norm": 226.86497497558594,
      "learning_rate": 2.0387729015764805e-05,
      "loss": 6.8369,
      "step": 1659
    },
    {
      "epoch": 0.6365030674846626,
      "grad_norm": 153.96597290039062,
      "learning_rate": 2.0366425223689818e-05,
      "loss": 4.1322,
      "step": 1660
    },
    {
      "epoch": 0.6368865030674846,
      "grad_norm": 154.2721710205078,
      "learning_rate": 2.0345121431614827e-05,
      "loss": 5.7915,
      "step": 1661
    },
    {
      "epoch": 0.6372699386503068,
      "grad_norm": 135.22979736328125,
      "learning_rate": 2.032381763953984e-05,
      "loss": 0.6553,
      "step": 1662
    },
    {
      "epoch": 0.6376533742331288,
      "grad_norm": 14.17731761932373,
      "learning_rate": 2.030251384746485e-05,
      "loss": 0.0685,
      "step": 1663
    },
    {
      "epoch": 0.6380368098159509,
      "grad_norm": 8.351577758789062,
      "learning_rate": 2.028121005538986e-05,
      "loss": 0.405,
      "step": 1664
    },
    {
      "epoch": 0.638420245398773,
      "grad_norm": 159.18130493164062,
      "learning_rate": 2.025990626331487e-05,
      "loss": 1.1095,
      "step": 1665
    },
    {
      "epoch": 0.6388036809815951,
      "grad_norm": 22.646175384521484,
      "learning_rate": 2.023860247123988e-05,
      "loss": 0.0969,
      "step": 1666
    },
    {
      "epoch": 0.6391871165644172,
      "grad_norm": 477.4721374511719,
      "learning_rate": 2.0217298679164893e-05,
      "loss": 1.6495,
      "step": 1667
    },
    {
      "epoch": 0.6395705521472392,
      "grad_norm": 0.9542309045791626,
      "learning_rate": 2.0195994887089902e-05,
      "loss": 0.0026,
      "step": 1668
    },
    {
      "epoch": 0.6399539877300614,
      "grad_norm": 2.2312052249908447,
      "learning_rate": 2.0174691095014915e-05,
      "loss": 0.0112,
      "step": 1669
    },
    {
      "epoch": 0.6403374233128835,
      "grad_norm": 2.4044041633605957,
      "learning_rate": 2.0153387302939924e-05,
      "loss": 0.0085,
      "step": 1670
    },
    {
      "epoch": 0.6407208588957055,
      "grad_norm": 2.774695873260498,
      "learning_rate": 2.0132083510864934e-05,
      "loss": 0.0101,
      "step": 1671
    },
    {
      "epoch": 0.6411042944785276,
      "grad_norm": 0.4862213134765625,
      "learning_rate": 2.0110779718789946e-05,
      "loss": 0.0024,
      "step": 1672
    },
    {
      "epoch": 0.6414877300613497,
      "grad_norm": 20.64777374267578,
      "learning_rate": 2.0089475926714956e-05,
      "loss": 0.4646,
      "step": 1673
    },
    {
      "epoch": 0.6418711656441718,
      "grad_norm": 79.48463439941406,
      "learning_rate": 2.006817213463997e-05,
      "loss": 1.1094,
      "step": 1674
    },
    {
      "epoch": 0.6422546012269938,
      "grad_norm": 6.128467082977295,
      "learning_rate": 2.0046868342564978e-05,
      "loss": 0.0144,
      "step": 1675
    },
    {
      "epoch": 0.6426380368098159,
      "grad_norm": 157.76174926757812,
      "learning_rate": 2.002556455048999e-05,
      "loss": 1.3321,
      "step": 1676
    },
    {
      "epoch": 0.6430214723926381,
      "grad_norm": 11.371692657470703,
      "learning_rate": 2.0004260758415e-05,
      "loss": 0.4241,
      "step": 1677
    },
    {
      "epoch": 0.6434049079754601,
      "grad_norm": 9.441778182983398,
      "learning_rate": 1.998295696634001e-05,
      "loss": 0.1427,
      "step": 1678
    },
    {
      "epoch": 0.6437883435582822,
      "grad_norm": 12.599998474121094,
      "learning_rate": 1.9961653174265022e-05,
      "loss": 0.1801,
      "step": 1679
    },
    {
      "epoch": 0.6441717791411042,
      "grad_norm": 1.525307059288025,
      "learning_rate": 1.994034938219003e-05,
      "loss": 0.0042,
      "step": 1680
    },
    {
      "epoch": 0.6445552147239264,
      "grad_norm": 0.44260019063949585,
      "learning_rate": 1.991904559011504e-05,
      "loss": 0.0021,
      "step": 1681
    },
    {
      "epoch": 0.6449386503067485,
      "grad_norm": 10.011754989624023,
      "learning_rate": 1.989774179804005e-05,
      "loss": 0.3962,
      "step": 1682
    },
    {
      "epoch": 0.6453220858895705,
      "grad_norm": 5.811450004577637,
      "learning_rate": 1.9876438005965062e-05,
      "loss": 0.3892,
      "step": 1683
    },
    {
      "epoch": 0.6457055214723927,
      "grad_norm": 0.6379611492156982,
      "learning_rate": 1.9855134213890072e-05,
      "loss": 0.0036,
      "step": 1684
    },
    {
      "epoch": 0.6460889570552147,
      "grad_norm": 0.11190050095319748,
      "learning_rate": 1.9833830421815084e-05,
      "loss": 0.0009,
      "step": 1685
    },
    {
      "epoch": 0.6464723926380368,
      "grad_norm": 4.65255069732666,
      "learning_rate": 1.9812526629740094e-05,
      "loss": 0.0162,
      "step": 1686
    },
    {
      "epoch": 0.6468558282208589,
      "grad_norm": 44.55350875854492,
      "learning_rate": 1.9791222837665106e-05,
      "loss": 0.8569,
      "step": 1687
    },
    {
      "epoch": 0.647239263803681,
      "grad_norm": 0.41566142439842224,
      "learning_rate": 1.9769919045590116e-05,
      "loss": 0.0015,
      "step": 1688
    },
    {
      "epoch": 0.6476226993865031,
      "grad_norm": 5.981119155883789,
      "learning_rate": 1.9748615253515125e-05,
      "loss": 0.3823,
      "step": 1689
    },
    {
      "epoch": 0.6480061349693251,
      "grad_norm": 2.8890583515167236,
      "learning_rate": 1.9727311461440138e-05,
      "loss": 0.0058,
      "step": 1690
    },
    {
      "epoch": 0.6483895705521472,
      "grad_norm": 180.98516845703125,
      "learning_rate": 1.9706007669365147e-05,
      "loss": 6.533,
      "step": 1691
    },
    {
      "epoch": 0.6487730061349694,
      "grad_norm": 0.5231181979179382,
      "learning_rate": 1.968470387729016e-05,
      "loss": 0.0025,
      "step": 1692
    },
    {
      "epoch": 0.6491564417177914,
      "grad_norm": 0.8823917508125305,
      "learning_rate": 1.966340008521517e-05,
      "loss": 0.0038,
      "step": 1693
    },
    {
      "epoch": 0.6495398773006135,
      "grad_norm": 9.24533462524414,
      "learning_rate": 1.9642096293140182e-05,
      "loss": 0.4578,
      "step": 1694
    },
    {
      "epoch": 0.6499233128834356,
      "grad_norm": 258.1440734863281,
      "learning_rate": 1.962079250106519e-05,
      "loss": 3.4919,
      "step": 1695
    },
    {
      "epoch": 0.6503067484662577,
      "grad_norm": 12.331808090209961,
      "learning_rate": 1.95994887089902e-05,
      "loss": 0.0584,
      "step": 1696
    },
    {
      "epoch": 0.6506901840490797,
      "grad_norm": 1.3217711448669434,
      "learning_rate": 1.9578184916915213e-05,
      "loss": 0.0049,
      "step": 1697
    },
    {
      "epoch": 0.6510736196319018,
      "grad_norm": 32.48533248901367,
      "learning_rate": 1.9556881124840222e-05,
      "loss": 0.6284,
      "step": 1698
    },
    {
      "epoch": 0.651457055214724,
      "grad_norm": 0.46559828519821167,
      "learning_rate": 1.9535577332765235e-05,
      "loss": 0.002,
      "step": 1699
    },
    {
      "epoch": 0.651840490797546,
      "grad_norm": 2.718585968017578,
      "learning_rate": 1.9514273540690244e-05,
      "loss": 0.0072,
      "step": 1700
    },
    {
      "epoch": 0.6522239263803681,
      "grad_norm": 5.244730472564697,
      "learning_rate": 1.9492969748615257e-05,
      "loss": 0.0048,
      "step": 1701
    },
    {
      "epoch": 0.6526073619631901,
      "grad_norm": 114.31527709960938,
      "learning_rate": 1.9471665956540266e-05,
      "loss": 0.8994,
      "step": 1702
    },
    {
      "epoch": 0.6529907975460123,
      "grad_norm": 28.290483474731445,
      "learning_rate": 1.9450362164465276e-05,
      "loss": 0.5547,
      "step": 1703
    },
    {
      "epoch": 0.6533742331288344,
      "grad_norm": 96.13031005859375,
      "learning_rate": 1.9429058372390285e-05,
      "loss": 1.3057,
      "step": 1704
    },
    {
      "epoch": 0.6537576687116564,
      "grad_norm": 0.8256839513778687,
      "learning_rate": 1.9407754580315294e-05,
      "loss": 0.0038,
      "step": 1705
    },
    {
      "epoch": 0.6541411042944786,
      "grad_norm": 69.43433380126953,
      "learning_rate": 1.9386450788240307e-05,
      "loss": 0.5801,
      "step": 1706
    },
    {
      "epoch": 0.6545245398773006,
      "grad_norm": 5.129863739013672,
      "learning_rate": 1.9365146996165316e-05,
      "loss": 0.0082,
      "step": 1707
    },
    {
      "epoch": 0.6549079754601227,
      "grad_norm": 290.7584533691406,
      "learning_rate": 1.934384320409033e-05,
      "loss": 3.4702,
      "step": 1708
    },
    {
      "epoch": 0.6552914110429447,
      "grad_norm": 15.134051322937012,
      "learning_rate": 1.932253941201534e-05,
      "loss": 0.4878,
      "step": 1709
    },
    {
      "epoch": 0.6556748466257669,
      "grad_norm": 7.622095108032227,
      "learning_rate": 1.930123561994035e-05,
      "loss": 0.0488,
      "step": 1710
    },
    {
      "epoch": 0.656058282208589,
      "grad_norm": 1.2107186317443848,
      "learning_rate": 1.927993182786536e-05,
      "loss": 0.0041,
      "step": 1711
    },
    {
      "epoch": 0.656441717791411,
      "grad_norm": 0.566555380821228,
      "learning_rate": 1.9258628035790373e-05,
      "loss": 0.0017,
      "step": 1712
    },
    {
      "epoch": 0.6568251533742331,
      "grad_norm": 0.2017064243555069,
      "learning_rate": 1.9237324243715382e-05,
      "loss": 0.0016,
      "step": 1713
    },
    {
      "epoch": 0.6572085889570553,
      "grad_norm": 42.69878387451172,
      "learning_rate": 1.9216020451640392e-05,
      "loss": 0.5176,
      "step": 1714
    },
    {
      "epoch": 0.6575920245398773,
      "grad_norm": 116.39155578613281,
      "learning_rate": 1.9194716659565404e-05,
      "loss": 1.8694,
      "step": 1715
    },
    {
      "epoch": 0.6579754601226994,
      "grad_norm": 9.295954704284668,
      "learning_rate": 1.9173412867490414e-05,
      "loss": 0.0294,
      "step": 1716
    },
    {
      "epoch": 0.6583588957055214,
      "grad_norm": 0.10093100368976593,
      "learning_rate": 1.9152109075415426e-05,
      "loss": 0.0007,
      "step": 1717
    },
    {
      "epoch": 0.6587423312883436,
      "grad_norm": 5.128528118133545,
      "learning_rate": 1.9130805283340436e-05,
      "loss": 0.0175,
      "step": 1718
    },
    {
      "epoch": 0.6591257668711656,
      "grad_norm": 46.20321273803711,
      "learning_rate": 1.910950149126545e-05,
      "loss": 0.7852,
      "step": 1719
    },
    {
      "epoch": 0.6595092024539877,
      "grad_norm": 96.1789779663086,
      "learning_rate": 1.9088197699190458e-05,
      "loss": 1.1777,
      "step": 1720
    },
    {
      "epoch": 0.6598926380368099,
      "grad_norm": 170.0279083251953,
      "learning_rate": 1.9066893907115467e-05,
      "loss": 5.8379,
      "step": 1721
    },
    {
      "epoch": 0.6602760736196319,
      "grad_norm": 15.705221176147461,
      "learning_rate": 1.904559011504048e-05,
      "loss": 0.4824,
      "step": 1722
    },
    {
      "epoch": 0.660659509202454,
      "grad_norm": 1.9702868461608887,
      "learning_rate": 1.902428632296549e-05,
      "loss": 0.0045,
      "step": 1723
    },
    {
      "epoch": 0.661042944785276,
      "grad_norm": 66.49562072753906,
      "learning_rate": 1.9002982530890502e-05,
      "loss": 1.2412,
      "step": 1724
    },
    {
      "epoch": 0.6614263803680982,
      "grad_norm": 9.27974796295166,
      "learning_rate": 1.8981678738815508e-05,
      "loss": 0.4216,
      "step": 1725
    },
    {
      "epoch": 0.6618098159509203,
      "grad_norm": 2.411372423171997,
      "learning_rate": 1.896037494674052e-05,
      "loss": 0.008,
      "step": 1726
    },
    {
      "epoch": 0.6621932515337423,
      "grad_norm": 32.30564880371094,
      "learning_rate": 1.893907115466553e-05,
      "loss": 0.1776,
      "step": 1727
    },
    {
      "epoch": 0.6625766871165644,
      "grad_norm": 2.3772690296173096,
      "learning_rate": 1.8917767362590542e-05,
      "loss": 0.0133,
      "step": 1728
    },
    {
      "epoch": 0.6629601226993865,
      "grad_norm": 39.052059173583984,
      "learning_rate": 1.8896463570515552e-05,
      "loss": 1.125,
      "step": 1729
    },
    {
      "epoch": 0.6633435582822086,
      "grad_norm": 0.2642825245857239,
      "learning_rate": 1.887515977844056e-05,
      "loss": 0.0021,
      "step": 1730
    },
    {
      "epoch": 0.6637269938650306,
      "grad_norm": 84.0382308959961,
      "learning_rate": 1.8853855986365574e-05,
      "loss": 1.0918,
      "step": 1731
    },
    {
      "epoch": 0.6641104294478528,
      "grad_norm": 45.17191696166992,
      "learning_rate": 1.8832552194290583e-05,
      "loss": 0.5645,
      "step": 1732
    },
    {
      "epoch": 0.6644938650306749,
      "grad_norm": 0.3179865777492523,
      "learning_rate": 1.8811248402215596e-05,
      "loss": 0.0017,
      "step": 1733
    },
    {
      "epoch": 0.6648773006134969,
      "grad_norm": 198.32339477539062,
      "learning_rate": 1.8789944610140605e-05,
      "loss": 6.1274,
      "step": 1734
    },
    {
      "epoch": 0.665260736196319,
      "grad_norm": 107.39860534667969,
      "learning_rate": 1.8768640818065618e-05,
      "loss": 0.8614,
      "step": 1735
    },
    {
      "epoch": 0.6656441717791411,
      "grad_norm": 2.4909374713897705,
      "learning_rate": 1.8747337025990627e-05,
      "loss": 0.014,
      "step": 1736
    },
    {
      "epoch": 0.6660276073619632,
      "grad_norm": 9.335639953613281,
      "learning_rate": 1.8726033233915636e-05,
      "loss": 0.0574,
      "step": 1737
    },
    {
      "epoch": 0.6664110429447853,
      "grad_norm": 14.218742370605469,
      "learning_rate": 1.870472944184065e-05,
      "loss": 0.5034,
      "step": 1738
    },
    {
      "epoch": 0.6667944785276073,
      "grad_norm": 0.4584377706050873,
      "learning_rate": 1.868342564976566e-05,
      "loss": 0.0018,
      "step": 1739
    },
    {
      "epoch": 0.6671779141104295,
      "grad_norm": 10.58724308013916,
      "learning_rate": 1.866212185769067e-05,
      "loss": 0.0159,
      "step": 1740
    },
    {
      "epoch": 0.6675613496932515,
      "grad_norm": 7.719966411590576,
      "learning_rate": 1.864081806561568e-05,
      "loss": 0.0215,
      "step": 1741
    },
    {
      "epoch": 0.6679447852760736,
      "grad_norm": 73.7376708984375,
      "learning_rate": 1.8619514273540693e-05,
      "loss": 0.8589,
      "step": 1742
    },
    {
      "epoch": 0.6683282208588958,
      "grad_norm": 4.980079174041748,
      "learning_rate": 1.8598210481465702e-05,
      "loss": 0.0174,
      "step": 1743
    },
    {
      "epoch": 0.6687116564417178,
      "grad_norm": 12.377878189086914,
      "learning_rate": 1.8576906689390715e-05,
      "loss": 0.4548,
      "step": 1744
    },
    {
      "epoch": 0.6690950920245399,
      "grad_norm": 134.05860900878906,
      "learning_rate": 1.8555602897315724e-05,
      "loss": 6.9141,
      "step": 1745
    },
    {
      "epoch": 0.6694785276073619,
      "grad_norm": 0.6603541374206543,
      "learning_rate": 1.8534299105240734e-05,
      "loss": 0.0032,
      "step": 1746
    },
    {
      "epoch": 0.6698619631901841,
      "grad_norm": 268.3301696777344,
      "learning_rate": 1.8512995313165746e-05,
      "loss": 5.8032,
      "step": 1747
    },
    {
      "epoch": 0.6702453987730062,
      "grad_norm": 12.854681968688965,
      "learning_rate": 1.8491691521090752e-05,
      "loss": 0.0681,
      "step": 1748
    },
    {
      "epoch": 0.6706288343558282,
      "grad_norm": 0.8064194321632385,
      "learning_rate": 1.8470387729015765e-05,
      "loss": 0.0032,
      "step": 1749
    },
    {
      "epoch": 0.6710122699386503,
      "grad_norm": 29.416452407836914,
      "learning_rate": 1.8449083936940774e-05,
      "loss": 0.5005,
      "step": 1750
    },
    {
      "epoch": 0.6713957055214724,
      "grad_norm": 204.04847717285156,
      "learning_rate": 1.8427780144865787e-05,
      "loss": 5.6729,
      "step": 1751
    },
    {
      "epoch": 0.6717791411042945,
      "grad_norm": 12.783069610595703,
      "learning_rate": 1.8406476352790796e-05,
      "loss": 0.5464,
      "step": 1752
    },
    {
      "epoch": 0.6721625766871165,
      "grad_norm": 193.79493713378906,
      "learning_rate": 1.838517256071581e-05,
      "loss": 2.2484,
      "step": 1753
    },
    {
      "epoch": 0.6725460122699386,
      "grad_norm": 4.909018516540527,
      "learning_rate": 1.836386876864082e-05,
      "loss": 0.0157,
      "step": 1754
    },
    {
      "epoch": 0.6729294478527608,
      "grad_norm": 124.53752136230469,
      "learning_rate": 1.8342564976565828e-05,
      "loss": 1.0596,
      "step": 1755
    },
    {
      "epoch": 0.6733128834355828,
      "grad_norm": 3.7626724243164062,
      "learning_rate": 1.832126118449084e-05,
      "loss": 0.0155,
      "step": 1756
    },
    {
      "epoch": 0.6736963190184049,
      "grad_norm": 18.932302474975586,
      "learning_rate": 1.829995739241585e-05,
      "loss": 0.4568,
      "step": 1757
    },
    {
      "epoch": 0.674079754601227,
      "grad_norm": 186.578857421875,
      "learning_rate": 1.8278653600340862e-05,
      "loss": 2.893,
      "step": 1758
    },
    {
      "epoch": 0.6744631901840491,
      "grad_norm": 195.05938720703125,
      "learning_rate": 1.8257349808265872e-05,
      "loss": 1.1065,
      "step": 1759
    },
    {
      "epoch": 0.6748466257668712,
      "grad_norm": 2.9090867042541504,
      "learning_rate": 1.8236046016190884e-05,
      "loss": 0.0057,
      "step": 1760
    },
    {
      "epoch": 0.6752300613496932,
      "grad_norm": 2.2006096839904785,
      "learning_rate": 1.8214742224115894e-05,
      "loss": 0.0067,
      "step": 1761
    },
    {
      "epoch": 0.6756134969325154,
      "grad_norm": 2.0305356979370117,
      "learning_rate": 1.8193438432040903e-05,
      "loss": 0.0048,
      "step": 1762
    },
    {
      "epoch": 0.6759969325153374,
      "grad_norm": 0.5942795872688293,
      "learning_rate": 1.8172134639965916e-05,
      "loss": 0.0025,
      "step": 1763
    },
    {
      "epoch": 0.6763803680981595,
      "grad_norm": 113.49565887451172,
      "learning_rate": 1.8150830847890925e-05,
      "loss": 0.855,
      "step": 1764
    },
    {
      "epoch": 0.6767638036809815,
      "grad_norm": 4.834312915802002,
      "learning_rate": 1.8129527055815938e-05,
      "loss": 0.0242,
      "step": 1765
    },
    {
      "epoch": 0.6771472392638037,
      "grad_norm": 15.199250221252441,
      "learning_rate": 1.8108223263740947e-05,
      "loss": 0.0306,
      "step": 1766
    },
    {
      "epoch": 0.6775306748466258,
      "grad_norm": 6.489785671234131,
      "learning_rate": 1.808691947166596e-05,
      "loss": 0.3616,
      "step": 1767
    },
    {
      "epoch": 0.6779141104294478,
      "grad_norm": 3.202645778656006,
      "learning_rate": 1.806561567959097e-05,
      "loss": 0.0137,
      "step": 1768
    },
    {
      "epoch": 0.67829754601227,
      "grad_norm": 49.37078857421875,
      "learning_rate": 1.804431188751598e-05,
      "loss": 0.5967,
      "step": 1769
    },
    {
      "epoch": 0.678680981595092,
      "grad_norm": 28.319705963134766,
      "learning_rate": 1.8023008095440988e-05,
      "loss": 0.4783,
      "step": 1770
    },
    {
      "epoch": 0.6790644171779141,
      "grad_norm": 8.28916072845459,
      "learning_rate": 1.8001704303366e-05,
      "loss": 0.0615,
      "step": 1771
    },
    {
      "epoch": 0.6794478527607362,
      "grad_norm": 6.079871654510498,
      "learning_rate": 1.798040051129101e-05,
      "loss": 0.0111,
      "step": 1772
    },
    {
      "epoch": 0.6798312883435583,
      "grad_norm": 188.42694091796875,
      "learning_rate": 1.795909671921602e-05,
      "loss": 1.5587,
      "step": 1773
    },
    {
      "epoch": 0.6802147239263804,
      "grad_norm": 20.358118057250977,
      "learning_rate": 1.7937792927141032e-05,
      "loss": 0.5034,
      "step": 1774
    },
    {
      "epoch": 0.6805981595092024,
      "grad_norm": 56.0653190612793,
      "learning_rate": 1.791648913506604e-05,
      "loss": 0.7251,
      "step": 1775
    },
    {
      "epoch": 0.6809815950920245,
      "grad_norm": 5.859439373016357,
      "learning_rate": 1.7895185342991054e-05,
      "loss": 0.0251,
      "step": 1776
    },
    {
      "epoch": 0.6813650306748467,
      "grad_norm": 13.532238006591797,
      "learning_rate": 1.7873881550916063e-05,
      "loss": 0.5747,
      "step": 1777
    },
    {
      "epoch": 0.6817484662576687,
      "grad_norm": 104.95072937011719,
      "learning_rate": 1.7852577758841076e-05,
      "loss": 1.2217,
      "step": 1778
    },
    {
      "epoch": 0.6821319018404908,
      "grad_norm": 84.7751235961914,
      "learning_rate": 1.7831273966766085e-05,
      "loss": 5.8945,
      "step": 1779
    },
    {
      "epoch": 0.6825153374233128,
      "grad_norm": 0.8566222786903381,
      "learning_rate": 1.7809970174691094e-05,
      "loss": 0.0023,
      "step": 1780
    },
    {
      "epoch": 0.682898773006135,
      "grad_norm": 5.526364803314209,
      "learning_rate": 1.7788666382616107e-05,
      "loss": 0.0271,
      "step": 1781
    },
    {
      "epoch": 0.683282208588957,
      "grad_norm": 65.81211853027344,
      "learning_rate": 1.7767362590541116e-05,
      "loss": 1.0322,
      "step": 1782
    },
    {
      "epoch": 0.6836656441717791,
      "grad_norm": 6.39776611328125,
      "learning_rate": 1.774605879846613e-05,
      "loss": 0.0161,
      "step": 1783
    },
    {
      "epoch": 0.6840490797546013,
      "grad_norm": 2.077016592025757,
      "learning_rate": 1.772475500639114e-05,
      "loss": 0.0071,
      "step": 1784
    },
    {
      "epoch": 0.6844325153374233,
      "grad_norm": 183.04539489746094,
      "learning_rate": 1.770345121431615e-05,
      "loss": 2.599,
      "step": 1785
    },
    {
      "epoch": 0.6848159509202454,
      "grad_norm": 165.1880340576172,
      "learning_rate": 1.768214742224116e-05,
      "loss": 1.0938,
      "step": 1786
    },
    {
      "epoch": 0.6851993865030674,
      "grad_norm": 181.1474151611328,
      "learning_rate": 1.766084363016617e-05,
      "loss": 3.0854,
      "step": 1787
    },
    {
      "epoch": 0.6855828220858896,
      "grad_norm": 0.4776207506656647,
      "learning_rate": 1.7639539838091182e-05,
      "loss": 0.0018,
      "step": 1788
    },
    {
      "epoch": 0.6859662576687117,
      "grad_norm": 0.3010045289993286,
      "learning_rate": 1.7618236046016192e-05,
      "loss": 0.0013,
      "step": 1789
    },
    {
      "epoch": 0.6863496932515337,
      "grad_norm": 4.359722137451172,
      "learning_rate": 1.7596932253941204e-05,
      "loss": 0.0199,
      "step": 1790
    },
    {
      "epoch": 0.6867331288343558,
      "grad_norm": 5.467195510864258,
      "learning_rate": 1.7575628461866214e-05,
      "loss": 0.021,
      "step": 1791
    },
    {
      "epoch": 0.6871165644171779,
      "grad_norm": 126.13877868652344,
      "learning_rate": 1.7554324669791226e-05,
      "loss": 1.8438,
      "step": 1792
    },
    {
      "epoch": 0.6875,
      "grad_norm": 39.74203872680664,
      "learning_rate": 1.7533020877716232e-05,
      "loss": 0.0711,
      "step": 1793
    },
    {
      "epoch": 0.6878834355828221,
      "grad_norm": 1.5761584043502808,
      "learning_rate": 1.7511717085641245e-05,
      "loss": 0.0063,
      "step": 1794
    },
    {
      "epoch": 0.6882668711656442,
      "grad_norm": 6.92527961730957,
      "learning_rate": 1.7490413293566254e-05,
      "loss": 0.0504,
      "step": 1795
    },
    {
      "epoch": 0.6886503067484663,
      "grad_norm": 14.018022537231445,
      "learning_rate": 1.7469109501491264e-05,
      "loss": 0.0411,
      "step": 1796
    },
    {
      "epoch": 0.6890337423312883,
      "grad_norm": 45.85357666015625,
      "learning_rate": 1.7447805709416276e-05,
      "loss": 0.6416,
      "step": 1797
    },
    {
      "epoch": 0.6894171779141104,
      "grad_norm": 7.583892345428467,
      "learning_rate": 1.7426501917341286e-05,
      "loss": 0.4094,
      "step": 1798
    },
    {
      "epoch": 0.6898006134969326,
      "grad_norm": 9.997591972351074,
      "learning_rate": 1.74051981252663e-05,
      "loss": 0.021,
      "step": 1799
    },
    {
      "epoch": 0.6901840490797546,
      "grad_norm": 37.529869079589844,
      "learning_rate": 1.7383894333191308e-05,
      "loss": 0.207,
      "step": 1800
    },
    {
      "epoch": 0.6905674846625767,
      "grad_norm": 0.7633021473884583,
      "learning_rate": 1.736259054111632e-05,
      "loss": 0.0027,
      "step": 1801
    },
    {
      "epoch": 0.6909509202453987,
      "grad_norm": 31.646526336669922,
      "learning_rate": 1.734128674904133e-05,
      "loss": 0.1269,
      "step": 1802
    },
    {
      "epoch": 0.6913343558282209,
      "grad_norm": 0.8677123188972473,
      "learning_rate": 1.7319982956966342e-05,
      "loss": 0.0021,
      "step": 1803
    },
    {
      "epoch": 0.691717791411043,
      "grad_norm": 94.05793762207031,
      "learning_rate": 1.7298679164891352e-05,
      "loss": 0.8174,
      "step": 1804
    },
    {
      "epoch": 0.692101226993865,
      "grad_norm": 0.7277902364730835,
      "learning_rate": 1.727737537281636e-05,
      "loss": 0.0035,
      "step": 1805
    },
    {
      "epoch": 0.6924846625766872,
      "grad_norm": 1.0158634185791016,
      "learning_rate": 1.7256071580741374e-05,
      "loss": 0.0061,
      "step": 1806
    },
    {
      "epoch": 0.6928680981595092,
      "grad_norm": 8.044918060302734,
      "learning_rate": 1.7234767788666383e-05,
      "loss": 0.4128,
      "step": 1807
    },
    {
      "epoch": 0.6932515337423313,
      "grad_norm": 1.4011287689208984,
      "learning_rate": 1.7213463996591396e-05,
      "loss": 0.0076,
      "step": 1808
    },
    {
      "epoch": 0.6936349693251533,
      "grad_norm": 11.000798225402832,
      "learning_rate": 1.7192160204516405e-05,
      "loss": 0.1869,
      "step": 1809
    },
    {
      "epoch": 0.6940184049079755,
      "grad_norm": 0.7866613864898682,
      "learning_rate": 1.7170856412441418e-05,
      "loss": 0.0037,
      "step": 1810
    },
    {
      "epoch": 0.6944018404907976,
      "grad_norm": 76.51680755615234,
      "learning_rate": 1.7149552620366427e-05,
      "loss": 0.9775,
      "step": 1811
    },
    {
      "epoch": 0.6947852760736196,
      "grad_norm": 20.117338180541992,
      "learning_rate": 1.7128248828291436e-05,
      "loss": 0.1213,
      "step": 1812
    },
    {
      "epoch": 0.6951687116564417,
      "grad_norm": 115.73905181884766,
      "learning_rate": 1.710694503621645e-05,
      "loss": 2.1172,
      "step": 1813
    },
    {
      "epoch": 0.6955521472392638,
      "grad_norm": 80.10689544677734,
      "learning_rate": 1.708564124414146e-05,
      "loss": 0.5376,
      "step": 1814
    },
    {
      "epoch": 0.6959355828220859,
      "grad_norm": 8.569589614868164,
      "learning_rate": 1.7064337452066468e-05,
      "loss": 0.4336,
      "step": 1815
    },
    {
      "epoch": 0.696319018404908,
      "grad_norm": 337.0047607421875,
      "learning_rate": 1.7043033659991477e-05,
      "loss": 3.3758,
      "step": 1816
    },
    {
      "epoch": 0.69670245398773,
      "grad_norm": 117.974853515625,
      "learning_rate": 1.702172986791649e-05,
      "loss": 6.3911,
      "step": 1817
    },
    {
      "epoch": 0.6970858895705522,
      "grad_norm": 198.02735900878906,
      "learning_rate": 1.70004260758415e-05,
      "loss": 5.3398,
      "step": 1818
    },
    {
      "epoch": 0.6974693251533742,
      "grad_norm": 18.306970596313477,
      "learning_rate": 1.6979122283766512e-05,
      "loss": 0.1182,
      "step": 1819
    },
    {
      "epoch": 0.6978527607361963,
      "grad_norm": 1.6636066436767578,
      "learning_rate": 1.695781849169152e-05,
      "loss": 0.0093,
      "step": 1820
    },
    {
      "epoch": 0.6982361963190185,
      "grad_norm": 38.072540283203125,
      "learning_rate": 1.693651469961653e-05,
      "loss": 0.9209,
      "step": 1821
    },
    {
      "epoch": 0.6986196319018405,
      "grad_norm": 38.3408317565918,
      "learning_rate": 1.6915210907541543e-05,
      "loss": 0.6152,
      "step": 1822
    },
    {
      "epoch": 0.6990030674846626,
      "grad_norm": 12.642842292785645,
      "learning_rate": 1.6893907115466552e-05,
      "loss": 0.0765,
      "step": 1823
    },
    {
      "epoch": 0.6993865030674846,
      "grad_norm": 7.309633255004883,
      "learning_rate": 1.6872603323391565e-05,
      "loss": 0.0517,
      "step": 1824
    },
    {
      "epoch": 0.6997699386503068,
      "grad_norm": 133.3367919921875,
      "learning_rate": 1.6851299531316574e-05,
      "loss": 1.2656,
      "step": 1825
    },
    {
      "epoch": 0.7001533742331288,
      "grad_norm": 4.016531467437744,
      "learning_rate": 1.6829995739241587e-05,
      "loss": 0.0179,
      "step": 1826
    },
    {
      "epoch": 0.7005368098159509,
      "grad_norm": 63.89706039428711,
      "learning_rate": 1.6808691947166596e-05,
      "loss": 0.9761,
      "step": 1827
    },
    {
      "epoch": 0.700920245398773,
      "grad_norm": 3.106642961502075,
      "learning_rate": 1.6787388155091606e-05,
      "loss": 0.0108,
      "step": 1828
    },
    {
      "epoch": 0.7013036809815951,
      "grad_norm": 199.66604614257812,
      "learning_rate": 1.676608436301662e-05,
      "loss": 5.8223,
      "step": 1829
    },
    {
      "epoch": 0.7016871165644172,
      "grad_norm": 7.568203926086426,
      "learning_rate": 1.6744780570941628e-05,
      "loss": 0.0156,
      "step": 1830
    },
    {
      "epoch": 0.7020705521472392,
      "grad_norm": 3.329444408416748,
      "learning_rate": 1.672347677886664e-05,
      "loss": 0.0109,
      "step": 1831
    },
    {
      "epoch": 0.7024539877300614,
      "grad_norm": 110.65080261230469,
      "learning_rate": 1.670217298679165e-05,
      "loss": 2.1547,
      "step": 1832
    },
    {
      "epoch": 0.7028374233128835,
      "grad_norm": 33.216392517089844,
      "learning_rate": 1.6680869194716662e-05,
      "loss": 0.7285,
      "step": 1833
    },
    {
      "epoch": 0.7032208588957055,
      "grad_norm": 20.46763801574707,
      "learning_rate": 1.6659565402641672e-05,
      "loss": 0.1825,
      "step": 1834
    },
    {
      "epoch": 0.7036042944785276,
      "grad_norm": 207.61228942871094,
      "learning_rate": 1.6638261610566684e-05,
      "loss": 5.3691,
      "step": 1835
    },
    {
      "epoch": 0.7039877300613497,
      "grad_norm": 10.169426918029785,
      "learning_rate": 1.6616957818491694e-05,
      "loss": 0.0364,
      "step": 1836
    },
    {
      "epoch": 0.7043711656441718,
      "grad_norm": 103.93666076660156,
      "learning_rate": 1.6595654026416703e-05,
      "loss": 1.1836,
      "step": 1837
    },
    {
      "epoch": 0.7047546012269938,
      "grad_norm": 83.81804656982422,
      "learning_rate": 1.6574350234341712e-05,
      "loss": 1.1963,
      "step": 1838
    },
    {
      "epoch": 0.7051380368098159,
      "grad_norm": 213.241943359375,
      "learning_rate": 1.655304644226672e-05,
      "loss": 2.1487,
      "step": 1839
    },
    {
      "epoch": 0.7055214723926381,
      "grad_norm": 1.7387597560882568,
      "learning_rate": 1.6531742650191734e-05,
      "loss": 0.0067,
      "step": 1840
    },
    {
      "epoch": 0.7059049079754601,
      "grad_norm": 183.66021728515625,
      "learning_rate": 1.6510438858116744e-05,
      "loss": 2.3087,
      "step": 1841
    },
    {
      "epoch": 0.7062883435582822,
      "grad_norm": 0.41703689098358154,
      "learning_rate": 1.6489135066041756e-05,
      "loss": 0.0019,
      "step": 1842
    },
    {
      "epoch": 0.7066717791411042,
      "grad_norm": 1.1585429906845093,
      "learning_rate": 1.6467831273966766e-05,
      "loss": 0.0027,
      "step": 1843
    },
    {
      "epoch": 0.7070552147239264,
      "grad_norm": 4.839847564697266,
      "learning_rate": 1.644652748189178e-05,
      "loss": 0.011,
      "step": 1844
    },
    {
      "epoch": 0.7074386503067485,
      "grad_norm": 87.01338195800781,
      "learning_rate": 1.6425223689816788e-05,
      "loss": 1.1348,
      "step": 1845
    },
    {
      "epoch": 0.7078220858895705,
      "grad_norm": 138.19476318359375,
      "learning_rate": 1.6403919897741797e-05,
      "loss": 2.0764,
      "step": 1846
    },
    {
      "epoch": 0.7082055214723927,
      "grad_norm": 0.5520815253257751,
      "learning_rate": 1.638261610566681e-05,
      "loss": 0.0023,
      "step": 1847
    },
    {
      "epoch": 0.7085889570552147,
      "grad_norm": 0.19649668037891388,
      "learning_rate": 1.636131231359182e-05,
      "loss": 0.0028,
      "step": 1848
    },
    {
      "epoch": 0.7089723926380368,
      "grad_norm": 12.944412231445312,
      "learning_rate": 1.6340008521516832e-05,
      "loss": 0.0554,
      "step": 1849
    },
    {
      "epoch": 0.7093558282208589,
      "grad_norm": 8.3750581741333,
      "learning_rate": 1.631870472944184e-05,
      "loss": 0.066,
      "step": 1850
    },
    {
      "epoch": 0.709739263803681,
      "grad_norm": 4.875927925109863,
      "learning_rate": 1.6297400937366854e-05,
      "loss": 0.0098,
      "step": 1851
    },
    {
      "epoch": 0.7101226993865031,
      "grad_norm": 6.181133270263672,
      "learning_rate": 1.6276097145291863e-05,
      "loss": 0.0357,
      "step": 1852
    },
    {
      "epoch": 0.7105061349693251,
      "grad_norm": 1.3159886598587036,
      "learning_rate": 1.6254793353216872e-05,
      "loss": 0.0032,
      "step": 1853
    },
    {
      "epoch": 0.7108895705521472,
      "grad_norm": 12.367630004882812,
      "learning_rate": 1.6233489561141885e-05,
      "loss": 0.0281,
      "step": 1854
    },
    {
      "epoch": 0.7112730061349694,
      "grad_norm": 0.16365942358970642,
      "learning_rate": 1.6212185769066894e-05,
      "loss": 0.0015,
      "step": 1855
    },
    {
      "epoch": 0.7116564417177914,
      "grad_norm": 0.538750946521759,
      "learning_rate": 1.6190881976991907e-05,
      "loss": 0.0025,
      "step": 1856
    },
    {
      "epoch": 0.7120398773006135,
      "grad_norm": 15.509056091308594,
      "learning_rate": 1.6169578184916916e-05,
      "loss": 0.5142,
      "step": 1857
    },
    {
      "epoch": 0.7124233128834356,
      "grad_norm": 0.3068455159664154,
      "learning_rate": 1.614827439284193e-05,
      "loss": 0.0024,
      "step": 1858
    },
    {
      "epoch": 0.7128067484662577,
      "grad_norm": 17.366540908813477,
      "learning_rate": 1.612697060076694e-05,
      "loss": 0.0475,
      "step": 1859
    },
    {
      "epoch": 0.7131901840490797,
      "grad_norm": 19.56414031982422,
      "learning_rate": 1.6105666808691948e-05,
      "loss": 0.4644,
      "step": 1860
    },
    {
      "epoch": 0.7135736196319018,
      "grad_norm": 138.43270874023438,
      "learning_rate": 1.6084363016616957e-05,
      "loss": 1.3018,
      "step": 1861
    },
    {
      "epoch": 0.713957055214724,
      "grad_norm": 4.688930511474609,
      "learning_rate": 1.606305922454197e-05,
      "loss": 0.3123,
      "step": 1862
    },
    {
      "epoch": 0.714340490797546,
      "grad_norm": 0.8955931067466736,
      "learning_rate": 1.604175543246698e-05,
      "loss": 0.0038,
      "step": 1863
    },
    {
      "epoch": 0.7147239263803681,
      "grad_norm": 0.4238760471343994,
      "learning_rate": 1.6020451640391988e-05,
      "loss": 0.0015,
      "step": 1864
    },
    {
      "epoch": 0.7151073619631901,
      "grad_norm": 37.68019485473633,
      "learning_rate": 1.5999147848317e-05,
      "loss": 0.5611,
      "step": 1865
    },
    {
      "epoch": 0.7154907975460123,
      "grad_norm": 1.6976826190948486,
      "learning_rate": 1.597784405624201e-05,
      "loss": 0.005,
      "step": 1866
    },
    {
      "epoch": 0.7158742331288344,
      "grad_norm": 257.343994140625,
      "learning_rate": 1.5956540264167023e-05,
      "loss": 1.7706,
      "step": 1867
    },
    {
      "epoch": 0.7162576687116564,
      "grad_norm": 11.13989543914795,
      "learning_rate": 1.5935236472092032e-05,
      "loss": 0.039,
      "step": 1868
    },
    {
      "epoch": 0.7166411042944786,
      "grad_norm": 60.56459426879883,
      "learning_rate": 1.5913932680017045e-05,
      "loss": 0.7422,
      "step": 1869
    },
    {
      "epoch": 0.7170245398773006,
      "grad_norm": 3.938284397125244,
      "learning_rate": 1.5892628887942054e-05,
      "loss": 0.0161,
      "step": 1870
    },
    {
      "epoch": 0.7174079754601227,
      "grad_norm": 0.5595266222953796,
      "learning_rate": 1.5871325095867064e-05,
      "loss": 0.0023,
      "step": 1871
    },
    {
      "epoch": 0.7177914110429447,
      "grad_norm": 18.529457092285156,
      "learning_rate": 1.5850021303792076e-05,
      "loss": 0.4453,
      "step": 1872
    },
    {
      "epoch": 0.7181748466257669,
      "grad_norm": 36.88288116455078,
      "learning_rate": 1.5828717511717086e-05,
      "loss": 0.6768,
      "step": 1873
    },
    {
      "epoch": 0.718558282208589,
      "grad_norm": 28.450756072998047,
      "learning_rate": 1.58074137196421e-05,
      "loss": 0.5103,
      "step": 1874
    },
    {
      "epoch": 0.718941717791411,
      "grad_norm": 11.081512451171875,
      "learning_rate": 1.5786109927567108e-05,
      "loss": 0.2457,
      "step": 1875
    },
    {
      "epoch": 0.7193251533742331,
      "grad_norm": 185.0625,
      "learning_rate": 1.576480613549212e-05,
      "loss": 6.2939,
      "step": 1876
    },
    {
      "epoch": 0.7197085889570553,
      "grad_norm": 11.565069198608398,
      "learning_rate": 1.574350234341713e-05,
      "loss": 0.2269,
      "step": 1877
    },
    {
      "epoch": 0.7200920245398773,
      "grad_norm": 1.4721031188964844,
      "learning_rate": 1.572219855134214e-05,
      "loss": 0.0078,
      "step": 1878
    },
    {
      "epoch": 0.7204754601226994,
      "grad_norm": 0.3564552366733551,
      "learning_rate": 1.570089475926715e-05,
      "loss": 0.0016,
      "step": 1879
    },
    {
      "epoch": 0.7208588957055214,
      "grad_norm": 1.6091878414154053,
      "learning_rate": 1.567959096719216e-05,
      "loss": 0.0045,
      "step": 1880
    },
    {
      "epoch": 0.7212423312883436,
      "grad_norm": 24.424392700195312,
      "learning_rate": 1.5658287175117174e-05,
      "loss": 0.563,
      "step": 1881
    },
    {
      "epoch": 0.7216257668711656,
      "grad_norm": 1.67574942111969,
      "learning_rate": 1.5636983383042183e-05,
      "loss": 0.0074,
      "step": 1882
    },
    {
      "epoch": 0.7220092024539877,
      "grad_norm": 12.126558303833008,
      "learning_rate": 1.5615679590967192e-05,
      "loss": 0.3853,
      "step": 1883
    },
    {
      "epoch": 0.7223926380368099,
      "grad_norm": 53.929473876953125,
      "learning_rate": 1.55943757988922e-05,
      "loss": 0.5855,
      "step": 1884
    },
    {
      "epoch": 0.7227760736196319,
      "grad_norm": 0.36175084114074707,
      "learning_rate": 1.5573072006817214e-05,
      "loss": 0.002,
      "step": 1885
    },
    {
      "epoch": 0.723159509202454,
      "grad_norm": 7.037576675415039,
      "learning_rate": 1.5551768214742224e-05,
      "loss": 0.0256,
      "step": 1886
    },
    {
      "epoch": 0.723542944785276,
      "grad_norm": 42.27104949951172,
      "learning_rate": 1.5530464422667236e-05,
      "loss": 0.5713,
      "step": 1887
    },
    {
      "epoch": 0.7239263803680982,
      "grad_norm": 7.669439315795898,
      "learning_rate": 1.5509160630592246e-05,
      "loss": 0.3967,
      "step": 1888
    },
    {
      "epoch": 0.7243098159509203,
      "grad_norm": 9.883378028869629,
      "learning_rate": 1.5487856838517255e-05,
      "loss": 0.1054,
      "step": 1889
    },
    {
      "epoch": 0.7246932515337423,
      "grad_norm": 5.921607494354248,
      "learning_rate": 1.5466553046442268e-05,
      "loss": 0.0218,
      "step": 1890
    },
    {
      "epoch": 0.7250766871165644,
      "grad_norm": 0.9740287065505981,
      "learning_rate": 1.5445249254367277e-05,
      "loss": 0.0042,
      "step": 1891
    },
    {
      "epoch": 0.7254601226993865,
      "grad_norm": 2.2117183208465576,
      "learning_rate": 1.542394546229229e-05,
      "loss": 0.007,
      "step": 1892
    },
    {
      "epoch": 0.7258435582822086,
      "grad_norm": 6.486509323120117,
      "learning_rate": 1.54026416702173e-05,
      "loss": 0.0383,
      "step": 1893
    },
    {
      "epoch": 0.7262269938650306,
      "grad_norm": 101.74858856201172,
      "learning_rate": 1.538133787814231e-05,
      "loss": 1.1943,
      "step": 1894
    },
    {
      "epoch": 0.7266104294478528,
      "grad_norm": 17.435060501098633,
      "learning_rate": 1.536003408606732e-05,
      "loss": 0.1314,
      "step": 1895
    },
    {
      "epoch": 0.7269938650306749,
      "grad_norm": 11.206552505493164,
      "learning_rate": 1.533873029399233e-05,
      "loss": 0.0237,
      "step": 1896
    },
    {
      "epoch": 0.7273773006134969,
      "grad_norm": 86.99957275390625,
      "learning_rate": 1.5317426501917343e-05,
      "loss": 0.5923,
      "step": 1897
    },
    {
      "epoch": 0.727760736196319,
      "grad_norm": 0.337234765291214,
      "learning_rate": 1.5296122709842352e-05,
      "loss": 0.0026,
      "step": 1898
    },
    {
      "epoch": 0.7281441717791411,
      "grad_norm": 5.304370403289795,
      "learning_rate": 1.5274818917767365e-05,
      "loss": 0.0127,
      "step": 1899
    },
    {
      "epoch": 0.7285276073619632,
      "grad_norm": 0.41007453203201294,
      "learning_rate": 1.5253515125692374e-05,
      "loss": 0.0018,
      "step": 1900
    },
    {
      "epoch": 0.7289110429447853,
      "grad_norm": 1.1569985151290894,
      "learning_rate": 1.5232211333617385e-05,
      "loss": 0.0064,
      "step": 1901
    },
    {
      "epoch": 0.7292944785276073,
      "grad_norm": 321.8655700683594,
      "learning_rate": 1.5210907541542396e-05,
      "loss": 6.2959,
      "step": 1902
    },
    {
      "epoch": 0.7296779141104295,
      "grad_norm": 1.9195574522018433,
      "learning_rate": 1.5189603749467407e-05,
      "loss": 0.0049,
      "step": 1903
    },
    {
      "epoch": 0.7300613496932515,
      "grad_norm": 17.68069076538086,
      "learning_rate": 1.5168299957392418e-05,
      "loss": 0.2111,
      "step": 1904
    },
    {
      "epoch": 0.7304447852760736,
      "grad_norm": 0.6155018210411072,
      "learning_rate": 1.514699616531743e-05,
      "loss": 0.0021,
      "step": 1905
    },
    {
      "epoch": 0.7308282208588958,
      "grad_norm": 5.575611114501953,
      "learning_rate": 1.5125692373242437e-05,
      "loss": 0.0278,
      "step": 1906
    },
    {
      "epoch": 0.7312116564417178,
      "grad_norm": 78.27565002441406,
      "learning_rate": 1.5104388581167448e-05,
      "loss": 0.8199,
      "step": 1907
    },
    {
      "epoch": 0.7315950920245399,
      "grad_norm": 0.3089018762111664,
      "learning_rate": 1.5083084789092459e-05,
      "loss": 0.0022,
      "step": 1908
    },
    {
      "epoch": 0.7319785276073619,
      "grad_norm": 6.002476692199707,
      "learning_rate": 1.5061780997017468e-05,
      "loss": 0.4094,
      "step": 1909
    },
    {
      "epoch": 0.7323619631901841,
      "grad_norm": 198.84535217285156,
      "learning_rate": 1.504047720494248e-05,
      "loss": 3.6476,
      "step": 1910
    },
    {
      "epoch": 0.7327453987730062,
      "grad_norm": 10.360122680664062,
      "learning_rate": 1.501917341286749e-05,
      "loss": 0.4385,
      "step": 1911
    },
    {
      "epoch": 0.7331288343558282,
      "grad_norm": 19.18494987487793,
      "learning_rate": 1.4997869620792501e-05,
      "loss": 0.0489,
      "step": 1912
    },
    {
      "epoch": 0.7335122699386503,
      "grad_norm": 0.40411585569381714,
      "learning_rate": 1.4976565828717512e-05,
      "loss": 0.0019,
      "step": 1913
    },
    {
      "epoch": 0.7338957055214724,
      "grad_norm": 5.745697021484375,
      "learning_rate": 1.4955262036642523e-05,
      "loss": 0.0181,
      "step": 1914
    },
    {
      "epoch": 0.7342791411042945,
      "grad_norm": 5.817564487457275,
      "learning_rate": 1.4933958244567534e-05,
      "loss": 0.0186,
      "step": 1915
    },
    {
      "epoch": 0.7346625766871165,
      "grad_norm": 0.47785699367523193,
      "learning_rate": 1.4912654452492544e-05,
      "loss": 0.0024,
      "step": 1916
    },
    {
      "epoch": 0.7350460122699386,
      "grad_norm": 0.995089590549469,
      "learning_rate": 1.4891350660417555e-05,
      "loss": 0.0047,
      "step": 1917
    },
    {
      "epoch": 0.7354294478527608,
      "grad_norm": 4.376603126525879,
      "learning_rate": 1.4870046868342566e-05,
      "loss": 0.3303,
      "step": 1918
    },
    {
      "epoch": 0.7358128834355828,
      "grad_norm": 2.5001165866851807,
      "learning_rate": 1.4848743076267577e-05,
      "loss": 0.0135,
      "step": 1919
    },
    {
      "epoch": 0.7361963190184049,
      "grad_norm": 1.037470817565918,
      "learning_rate": 1.4827439284192588e-05,
      "loss": 0.0028,
      "step": 1920
    },
    {
      "epoch": 0.736579754601227,
      "grad_norm": 33.8574333190918,
      "learning_rate": 1.4806135492117599e-05,
      "loss": 0.6431,
      "step": 1921
    },
    {
      "epoch": 0.7369631901840491,
      "grad_norm": 11.763459205627441,
      "learning_rate": 1.478483170004261e-05,
      "loss": 0.0238,
      "step": 1922
    },
    {
      "epoch": 0.7373466257668712,
      "grad_norm": 8.176692962646484,
      "learning_rate": 1.4763527907967619e-05,
      "loss": 0.0393,
      "step": 1923
    },
    {
      "epoch": 0.7377300613496932,
      "grad_norm": 95.25994873046875,
      "learning_rate": 1.474222411589263e-05,
      "loss": 1.3507,
      "step": 1924
    },
    {
      "epoch": 0.7381134969325154,
      "grad_norm": 6.803265571594238,
      "learning_rate": 1.4720920323817641e-05,
      "loss": 0.0887,
      "step": 1925
    },
    {
      "epoch": 0.7384969325153374,
      "grad_norm": 235.39866638183594,
      "learning_rate": 1.4699616531742652e-05,
      "loss": 2.6448,
      "step": 1926
    },
    {
      "epoch": 0.7388803680981595,
      "grad_norm": 157.25706481933594,
      "learning_rate": 1.4678312739667663e-05,
      "loss": 2.5195,
      "step": 1927
    },
    {
      "epoch": 0.7392638036809815,
      "grad_norm": 105.20527648925781,
      "learning_rate": 1.465700894759267e-05,
      "loss": 6.0176,
      "step": 1928
    },
    {
      "epoch": 0.7396472392638037,
      "grad_norm": 101.31459045410156,
      "learning_rate": 1.4635705155517682e-05,
      "loss": 0.9268,
      "step": 1929
    },
    {
      "epoch": 0.7400306748466258,
      "grad_norm": 74.20117950439453,
      "learning_rate": 1.4614401363442693e-05,
      "loss": 1.3682,
      "step": 1930
    },
    {
      "epoch": 0.7404141104294478,
      "grad_norm": 1.142648458480835,
      "learning_rate": 1.4593097571367704e-05,
      "loss": 0.003,
      "step": 1931
    },
    {
      "epoch": 0.74079754601227,
      "grad_norm": 0.8568863868713379,
      "learning_rate": 1.4571793779292715e-05,
      "loss": 0.0027,
      "step": 1932
    },
    {
      "epoch": 0.741180981595092,
      "grad_norm": 2.354698896408081,
      "learning_rate": 1.4550489987217724e-05,
      "loss": 0.0083,
      "step": 1933
    },
    {
      "epoch": 0.7415644171779141,
      "grad_norm": 137.51414489746094,
      "learning_rate": 1.4529186195142735e-05,
      "loss": 6.1094,
      "step": 1934
    },
    {
      "epoch": 0.7419478527607362,
      "grad_norm": 0.9444566369056702,
      "learning_rate": 1.4507882403067746e-05,
      "loss": 0.0025,
      "step": 1935
    },
    {
      "epoch": 0.7423312883435583,
      "grad_norm": 0.28314322233200073,
      "learning_rate": 1.4486578610992757e-05,
      "loss": 0.0026,
      "step": 1936
    },
    {
      "epoch": 0.7427147239263804,
      "grad_norm": 24.82479476928711,
      "learning_rate": 1.4465274818917768e-05,
      "loss": 0.5044,
      "step": 1937
    },
    {
      "epoch": 0.7430981595092024,
      "grad_norm": 12.466592788696289,
      "learning_rate": 1.4443971026842779e-05,
      "loss": 0.4556,
      "step": 1938
    },
    {
      "epoch": 0.7434815950920245,
      "grad_norm": 13.750520706176758,
      "learning_rate": 1.442266723476779e-05,
      "loss": 0.1217,
      "step": 1939
    },
    {
      "epoch": 0.7438650306748467,
      "grad_norm": 0.13975299894809723,
      "learning_rate": 1.44013634426928e-05,
      "loss": 0.0011,
      "step": 1940
    },
    {
      "epoch": 0.7442484662576687,
      "grad_norm": 6.503437519073486,
      "learning_rate": 1.438005965061781e-05,
      "loss": 0.0409,
      "step": 1941
    },
    {
      "epoch": 0.7446319018404908,
      "grad_norm": 9.731389999389648,
      "learning_rate": 1.4358755858542821e-05,
      "loss": 0.0243,
      "step": 1942
    },
    {
      "epoch": 0.7450153374233128,
      "grad_norm": 13.670307159423828,
      "learning_rate": 1.4337452066467832e-05,
      "loss": 0.1563,
      "step": 1943
    },
    {
      "epoch": 0.745398773006135,
      "grad_norm": 1.3887224197387695,
      "learning_rate": 1.4316148274392843e-05,
      "loss": 0.006,
      "step": 1944
    },
    {
      "epoch": 0.745782208588957,
      "grad_norm": 19.59745216369629,
      "learning_rate": 1.4294844482317854e-05,
      "loss": 0.6211,
      "step": 1945
    },
    {
      "epoch": 0.7461656441717791,
      "grad_norm": 0.4769805371761322,
      "learning_rate": 1.4273540690242865e-05,
      "loss": 0.0017,
      "step": 1946
    },
    {
      "epoch": 0.7465490797546013,
      "grad_norm": 5.039165019989014,
      "learning_rate": 1.4252236898167876e-05,
      "loss": 0.0082,
      "step": 1947
    },
    {
      "epoch": 0.7469325153374233,
      "grad_norm": 1.004677414894104,
      "learning_rate": 1.4230933106092886e-05,
      "loss": 0.0035,
      "step": 1948
    },
    {
      "epoch": 0.7473159509202454,
      "grad_norm": 0.9467623233795166,
      "learning_rate": 1.4209629314017897e-05,
      "loss": 0.0019,
      "step": 1949
    },
    {
      "epoch": 0.7476993865030674,
      "grad_norm": 8.143989562988281,
      "learning_rate": 1.4188325521942908e-05,
      "loss": 0.0665,
      "step": 1950
    },
    {
      "epoch": 0.7480828220858896,
      "grad_norm": 0.25524386763572693,
      "learning_rate": 1.4167021729867915e-05,
      "loss": 0.0008,
      "step": 1951
    },
    {
      "epoch": 0.7484662576687117,
      "grad_norm": 223.53726196289062,
      "learning_rate": 1.4145717937792926e-05,
      "loss": 1.3477,
      "step": 1952
    },
    {
      "epoch": 0.7488496932515337,
      "grad_norm": 214.31077575683594,
      "learning_rate": 1.4124414145717937e-05,
      "loss": 2.3165,
      "step": 1953
    },
    {
      "epoch": 0.7492331288343558,
      "grad_norm": 15.331868171691895,
      "learning_rate": 1.4103110353642948e-05,
      "loss": 0.4204,
      "step": 1954
    },
    {
      "epoch": 0.7496165644171779,
      "grad_norm": 139.1454315185547,
      "learning_rate": 1.408180656156796e-05,
      "loss": 6.4492,
      "step": 1955
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.3760151863098145,
      "learning_rate": 1.406050276949297e-05,
      "loss": 0.0166,
      "step": 1956
    },
    {
      "epoch": 0.7503834355828221,
      "grad_norm": 2.1321780681610107,
      "learning_rate": 1.4039198977417981e-05,
      "loss": 0.0044,
      "step": 1957
    },
    {
      "epoch": 0.7507668711656442,
      "grad_norm": 76.46648406982422,
      "learning_rate": 1.401789518534299e-05,
      "loss": 0.9727,
      "step": 1958
    },
    {
      "epoch": 0.7511503067484663,
      "grad_norm": 151.1408233642578,
      "learning_rate": 1.3996591393268002e-05,
      "loss": 1.4248,
      "step": 1959
    },
    {
      "epoch": 0.7515337423312883,
      "grad_norm": 24.7465877532959,
      "learning_rate": 1.3975287601193013e-05,
      "loss": 0.4456,
      "step": 1960
    },
    {
      "epoch": 0.7519171779141104,
      "grad_norm": 11.28043270111084,
      "learning_rate": 1.3953983809118024e-05,
      "loss": 0.1874,
      "step": 1961
    },
    {
      "epoch": 0.7523006134969326,
      "grad_norm": 226.95289611816406,
      "learning_rate": 1.3932680017043035e-05,
      "loss": 1.3262,
      "step": 1962
    },
    {
      "epoch": 0.7526840490797546,
      "grad_norm": 0.2464362531900406,
      "learning_rate": 1.3911376224968046e-05,
      "loss": 0.0022,
      "step": 1963
    },
    {
      "epoch": 0.7530674846625767,
      "grad_norm": 12.830216407775879,
      "learning_rate": 1.3890072432893057e-05,
      "loss": 0.0542,
      "step": 1964
    },
    {
      "epoch": 0.7534509202453987,
      "grad_norm": 149.88232421875,
      "learning_rate": 1.3868768640818066e-05,
      "loss": 2.9808,
      "step": 1965
    },
    {
      "epoch": 0.7538343558282209,
      "grad_norm": 18.341276168823242,
      "learning_rate": 1.3847464848743077e-05,
      "loss": 0.1652,
      "step": 1966
    },
    {
      "epoch": 0.754217791411043,
      "grad_norm": 29.104957580566406,
      "learning_rate": 1.3826161056668088e-05,
      "loss": 0.0528,
      "step": 1967
    },
    {
      "epoch": 0.754601226993865,
      "grad_norm": 78.68212890625,
      "learning_rate": 1.3804857264593099e-05,
      "loss": 0.9668,
      "step": 1968
    },
    {
      "epoch": 0.7549846625766872,
      "grad_norm": 15.954132080078125,
      "learning_rate": 1.378355347251811e-05,
      "loss": 0.0438,
      "step": 1969
    },
    {
      "epoch": 0.7553680981595092,
      "grad_norm": 7.746146202087402,
      "learning_rate": 1.3762249680443121e-05,
      "loss": 0.0301,
      "step": 1970
    },
    {
      "epoch": 0.7557515337423313,
      "grad_norm": 0.9864992499351501,
      "learning_rate": 1.3740945888368132e-05,
      "loss": 0.0029,
      "step": 1971
    },
    {
      "epoch": 0.7561349693251533,
      "grad_norm": 197.49974060058594,
      "learning_rate": 1.3719642096293141e-05,
      "loss": 5.2236,
      "step": 1972
    },
    {
      "epoch": 0.7565184049079755,
      "grad_norm": 0.28314509987831116,
      "learning_rate": 1.369833830421815e-05,
      "loss": 0.0017,
      "step": 1973
    },
    {
      "epoch": 0.7569018404907976,
      "grad_norm": 1.3988046646118164,
      "learning_rate": 1.3677034512143162e-05,
      "loss": 0.0048,
      "step": 1974
    },
    {
      "epoch": 0.7572852760736196,
      "grad_norm": 9.20186996459961,
      "learning_rate": 1.3655730720068171e-05,
      "loss": 0.0245,
      "step": 1975
    },
    {
      "epoch": 0.7576687116564417,
      "grad_norm": 11.838817596435547,
      "learning_rate": 1.3634426927993182e-05,
      "loss": 0.0832,
      "step": 1976
    },
    {
      "epoch": 0.7580521472392638,
      "grad_norm": 138.5713348388672,
      "learning_rate": 1.3613123135918193e-05,
      "loss": 6.6934,
      "step": 1977
    },
    {
      "epoch": 0.7584355828220859,
      "grad_norm": 3.0601112842559814,
      "learning_rate": 1.3591819343843204e-05,
      "loss": 0.0071,
      "step": 1978
    },
    {
      "epoch": 0.758819018404908,
      "grad_norm": 0.6360490918159485,
      "learning_rate": 1.3570515551768215e-05,
      "loss": 0.0027,
      "step": 1979
    },
    {
      "epoch": 0.75920245398773,
      "grad_norm": 162.20603942871094,
      "learning_rate": 1.3549211759693226e-05,
      "loss": 5.6367,
      "step": 1980
    },
    {
      "epoch": 0.7595858895705522,
      "grad_norm": 11.104833602905273,
      "learning_rate": 1.3527907967618237e-05,
      "loss": 0.1571,
      "step": 1981
    },
    {
      "epoch": 0.7599693251533742,
      "grad_norm": 12.747593879699707,
      "learning_rate": 1.3506604175543246e-05,
      "loss": 0.1888,
      "step": 1982
    },
    {
      "epoch": 0.7603527607361963,
      "grad_norm": 199.33963012695312,
      "learning_rate": 1.3485300383468257e-05,
      "loss": 2.0549,
      "step": 1983
    },
    {
      "epoch": 0.7607361963190185,
      "grad_norm": 43.54447555541992,
      "learning_rate": 1.3463996591393268e-05,
      "loss": 0.8364,
      "step": 1984
    },
    {
      "epoch": 0.7611196319018405,
      "grad_norm": 2.030282974243164,
      "learning_rate": 1.344269279931828e-05,
      "loss": 0.0062,
      "step": 1985
    },
    {
      "epoch": 0.7615030674846626,
      "grad_norm": 21.328853607177734,
      "learning_rate": 1.342138900724329e-05,
      "loss": 0.4646,
      "step": 1986
    },
    {
      "epoch": 0.7618865030674846,
      "grad_norm": 194.71197509765625,
      "learning_rate": 1.3400085215168301e-05,
      "loss": 6.418,
      "step": 1987
    },
    {
      "epoch": 0.7622699386503068,
      "grad_norm": 1.1269176006317139,
      "learning_rate": 1.3378781423093312e-05,
      "loss": 0.0029,
      "step": 1988
    },
    {
      "epoch": 0.7626533742331288,
      "grad_norm": 15.9646635055542,
      "learning_rate": 1.3357477631018323e-05,
      "loss": 0.4321,
      "step": 1989
    },
    {
      "epoch": 0.7630368098159509,
      "grad_norm": 7.329847812652588,
      "learning_rate": 1.3336173838943333e-05,
      "loss": 0.027,
      "step": 1990
    },
    {
      "epoch": 0.763420245398773,
      "grad_norm": 71.15471649169922,
      "learning_rate": 1.3314870046868344e-05,
      "loss": 0.6953,
      "step": 1991
    },
    {
      "epoch": 0.7638036809815951,
      "grad_norm": 7.559596538543701,
      "learning_rate": 1.3293566254793355e-05,
      "loss": 0.0185,
      "step": 1992
    },
    {
      "epoch": 0.7641871165644172,
      "grad_norm": 122.88265228271484,
      "learning_rate": 1.3272262462718366e-05,
      "loss": 6.1523,
      "step": 1993
    },
    {
      "epoch": 0.7645705521472392,
      "grad_norm": 0.2372620701789856,
      "learning_rate": 1.3250958670643377e-05,
      "loss": 0.0013,
      "step": 1994
    },
    {
      "epoch": 0.7649539877300614,
      "grad_norm": 1.6263588666915894,
      "learning_rate": 1.3229654878568388e-05,
      "loss": 0.0038,
      "step": 1995
    },
    {
      "epoch": 0.7653374233128835,
      "grad_norm": 27.143592834472656,
      "learning_rate": 1.3208351086493395e-05,
      "loss": 0.429,
      "step": 1996
    },
    {
      "epoch": 0.7657208588957055,
      "grad_norm": 3.7754852771759033,
      "learning_rate": 1.3187047294418406e-05,
      "loss": 0.0113,
      "step": 1997
    },
    {
      "epoch": 0.7661042944785276,
      "grad_norm": 12.764897346496582,
      "learning_rate": 1.3165743502343417e-05,
      "loss": 0.0658,
      "step": 1998
    },
    {
      "epoch": 0.7664877300613497,
      "grad_norm": 1.2359267473220825,
      "learning_rate": 1.3144439710268428e-05,
      "loss": 0.0053,
      "step": 1999
    },
    {
      "epoch": 0.7668711656441718,
      "grad_norm": 49.90631103515625,
      "learning_rate": 1.3123135918193438e-05,
      "loss": 0.5952,
      "step": 2000
    },
    {
      "epoch": 0.7672546012269938,
      "grad_norm": 253.21788024902344,
      "learning_rate": 1.3101832126118449e-05,
      "loss": 2.6388,
      "step": 2001
    },
    {
      "epoch": 0.7676380368098159,
      "grad_norm": 11.39282512664795,
      "learning_rate": 1.308052833404346e-05,
      "loss": 0.0591,
      "step": 2002
    },
    {
      "epoch": 0.7680214723926381,
      "grad_norm": 14.903972625732422,
      "learning_rate": 1.305922454196847e-05,
      "loss": 0.0207,
      "step": 2003
    },
    {
      "epoch": 0.7684049079754601,
      "grad_norm": 130.27838134765625,
      "learning_rate": 1.3037920749893482e-05,
      "loss": 6.0723,
      "step": 2004
    },
    {
      "epoch": 0.7687883435582822,
      "grad_norm": 20.10736083984375,
      "learning_rate": 1.3016616957818493e-05,
      "loss": 0.542,
      "step": 2005
    },
    {
      "epoch": 0.7691717791411042,
      "grad_norm": 133.43751525878906,
      "learning_rate": 1.2995313165743504e-05,
      "loss": 6.8203,
      "step": 2006
    },
    {
      "epoch": 0.7695552147239264,
      "grad_norm": 37.334869384765625,
      "learning_rate": 1.2974009373668513e-05,
      "loss": 0.6221,
      "step": 2007
    },
    {
      "epoch": 0.7699386503067485,
      "grad_norm": 2.647223949432373,
      "learning_rate": 1.2952705581593524e-05,
      "loss": 0.0089,
      "step": 2008
    },
    {
      "epoch": 0.7703220858895705,
      "grad_norm": 172.29574584960938,
      "learning_rate": 1.2931401789518535e-05,
      "loss": 6.0957,
      "step": 2009
    },
    {
      "epoch": 0.7707055214723927,
      "grad_norm": 2.514612913131714,
      "learning_rate": 1.2910097997443546e-05,
      "loss": 0.0052,
      "step": 2010
    },
    {
      "epoch": 0.7710889570552147,
      "grad_norm": 13.704341888427734,
      "learning_rate": 1.2888794205368557e-05,
      "loss": 0.4009,
      "step": 2011
    },
    {
      "epoch": 0.7714723926380368,
      "grad_norm": 59.756534576416016,
      "learning_rate": 1.2867490413293568e-05,
      "loss": 0.5493,
      "step": 2012
    },
    {
      "epoch": 0.7718558282208589,
      "grad_norm": 8.976195335388184,
      "learning_rate": 1.2846186621218579e-05,
      "loss": 0.0307,
      "step": 2013
    },
    {
      "epoch": 0.772239263803681,
      "grad_norm": 17.59994125366211,
      "learning_rate": 1.2824882829143588e-05,
      "loss": 0.5464,
      "step": 2014
    },
    {
      "epoch": 0.7726226993865031,
      "grad_norm": 5.109958648681641,
      "learning_rate": 1.28035790370686e-05,
      "loss": 0.0102,
      "step": 2015
    },
    {
      "epoch": 0.7730061349693251,
      "grad_norm": 134.91262817382812,
      "learning_rate": 1.278227524499361e-05,
      "loss": 6.0898,
      "step": 2016
    },
    {
      "epoch": 0.7733895705521472,
      "grad_norm": 3.405291795730591,
      "learning_rate": 1.2760971452918621e-05,
      "loss": 0.0038,
      "step": 2017
    },
    {
      "epoch": 0.7737730061349694,
      "grad_norm": 69.19807434082031,
      "learning_rate": 1.2739667660843629e-05,
      "loss": 0.667,
      "step": 2018
    },
    {
      "epoch": 0.7741564417177914,
      "grad_norm": 0.4252760112285614,
      "learning_rate": 1.271836386876864e-05,
      "loss": 0.0025,
      "step": 2019
    },
    {
      "epoch": 0.7745398773006135,
      "grad_norm": 20.130861282348633,
      "learning_rate": 1.2697060076693651e-05,
      "loss": 0.1615,
      "step": 2020
    },
    {
      "epoch": 0.7749233128834356,
      "grad_norm": 0.6197576522827148,
      "learning_rate": 1.2675756284618662e-05,
      "loss": 0.0023,
      "step": 2021
    },
    {
      "epoch": 0.7753067484662577,
      "grad_norm": 8.245320320129395,
      "learning_rate": 1.2654452492543673e-05,
      "loss": 0.0094,
      "step": 2022
    },
    {
      "epoch": 0.7756901840490797,
      "grad_norm": 3.970463752746582,
      "learning_rate": 1.2633148700468684e-05,
      "loss": 0.0144,
      "step": 2023
    },
    {
      "epoch": 0.7760736196319018,
      "grad_norm": 206.5851287841797,
      "learning_rate": 1.2611844908393693e-05,
      "loss": 2.6232,
      "step": 2024
    },
    {
      "epoch": 0.776457055214724,
      "grad_norm": 62.07779312133789,
      "learning_rate": 1.2590541116318704e-05,
      "loss": 0.8525,
      "step": 2025
    },
    {
      "epoch": 0.776840490797546,
      "grad_norm": 98.0234375,
      "learning_rate": 1.2569237324243715e-05,
      "loss": 1.0723,
      "step": 2026
    },
    {
      "epoch": 0.7772239263803681,
      "grad_norm": 24.69891357421875,
      "learning_rate": 1.2547933532168726e-05,
      "loss": 0.5298,
      "step": 2027
    },
    {
      "epoch": 0.7776073619631901,
      "grad_norm": 37.78566360473633,
      "learning_rate": 1.2526629740093737e-05,
      "loss": 0.8819,
      "step": 2028
    },
    {
      "epoch": 0.7779907975460123,
      "grad_norm": 7.116847038269043,
      "learning_rate": 1.2505325948018748e-05,
      "loss": 0.3687,
      "step": 2029
    },
    {
      "epoch": 0.7783742331288344,
      "grad_norm": 8.491434097290039,
      "learning_rate": 1.248402215594376e-05,
      "loss": 0.0429,
      "step": 2030
    },
    {
      "epoch": 0.7787576687116564,
      "grad_norm": 1.4491527080535889,
      "learning_rate": 1.246271836386877e-05,
      "loss": 0.0033,
      "step": 2031
    },
    {
      "epoch": 0.7791411042944786,
      "grad_norm": 2.5249433517456055,
      "learning_rate": 1.244141457179378e-05,
      "loss": 0.0051,
      "step": 2032
    },
    {
      "epoch": 0.7795245398773006,
      "grad_norm": 57.93564224243164,
      "learning_rate": 1.242011077971879e-05,
      "loss": 0.6123,
      "step": 2033
    },
    {
      "epoch": 0.7799079754601227,
      "grad_norm": 1.0227478742599487,
      "learning_rate": 1.2398806987643802e-05,
      "loss": 0.0024,
      "step": 2034
    },
    {
      "epoch": 0.7802914110429447,
      "grad_norm": 0.39101430773735046,
      "learning_rate": 1.2377503195568811e-05,
      "loss": 0.0028,
      "step": 2035
    },
    {
      "epoch": 0.7806748466257669,
      "grad_norm": 38.51777267456055,
      "learning_rate": 1.2356199403493822e-05,
      "loss": 0.7461,
      "step": 2036
    },
    {
      "epoch": 0.781058282208589,
      "grad_norm": 2.0663833618164062,
      "learning_rate": 1.2334895611418833e-05,
      "loss": 0.0059,
      "step": 2037
    },
    {
      "epoch": 0.781441717791411,
      "grad_norm": 0.12430649995803833,
      "learning_rate": 1.2313591819343844e-05,
      "loss": 0.0014,
      "step": 2038
    },
    {
      "epoch": 0.7818251533742331,
      "grad_norm": 1.2589157819747925,
      "learning_rate": 1.2292288027268855e-05,
      "loss": 0.0031,
      "step": 2039
    },
    {
      "epoch": 0.7822085889570553,
      "grad_norm": 3.417670726776123,
      "learning_rate": 1.2270984235193864e-05,
      "loss": 0.0119,
      "step": 2040
    },
    {
      "epoch": 0.7825920245398773,
      "grad_norm": 16.39604949951172,
      "learning_rate": 1.2249680443118875e-05,
      "loss": 0.1658,
      "step": 2041
    },
    {
      "epoch": 0.7829754601226994,
      "grad_norm": 0.9449207782745361,
      "learning_rate": 1.2228376651043886e-05,
      "loss": 0.0032,
      "step": 2042
    },
    {
      "epoch": 0.7833588957055214,
      "grad_norm": 106.27434539794922,
      "learning_rate": 1.2207072858968897e-05,
      "loss": 0.7173,
      "step": 2043
    },
    {
      "epoch": 0.7837423312883436,
      "grad_norm": 12.998677253723145,
      "learning_rate": 1.2185769066893908e-05,
      "loss": 0.0501,
      "step": 2044
    },
    {
      "epoch": 0.7841257668711656,
      "grad_norm": 2.242809534072876,
      "learning_rate": 1.216446527481892e-05,
      "loss": 0.0047,
      "step": 2045
    },
    {
      "epoch": 0.7845092024539877,
      "grad_norm": 299.3255615234375,
      "learning_rate": 1.214316148274393e-05,
      "loss": 5.7402,
      "step": 2046
    },
    {
      "epoch": 0.7848926380368099,
      "grad_norm": 0.0981670394539833,
      "learning_rate": 1.212185769066894e-05,
      "loss": 0.0011,
      "step": 2047
    },
    {
      "epoch": 0.7852760736196319,
      "grad_norm": 109.51537322998047,
      "learning_rate": 1.210055389859395e-05,
      "loss": 1.4248,
      "step": 2048
    },
    {
      "epoch": 0.785659509202454,
      "grad_norm": 243.7086944580078,
      "learning_rate": 1.207925010651896e-05,
      "loss": 2.7435,
      "step": 2049
    },
    {
      "epoch": 0.786042944785276,
      "grad_norm": 1.342392086982727,
      "learning_rate": 1.2057946314443971e-05,
      "loss": 0.0056,
      "step": 2050
    },
    {
      "epoch": 0.7864263803680982,
      "grad_norm": 12.527939796447754,
      "learning_rate": 1.2036642522368982e-05,
      "loss": 0.406,
      "step": 2051
    },
    {
      "epoch": 0.7868098159509203,
      "grad_norm": 149.59002685546875,
      "learning_rate": 1.2015338730293993e-05,
      "loss": 1.8253,
      "step": 2052
    },
    {
      "epoch": 0.7871932515337423,
      "grad_norm": 2.531747817993164,
      "learning_rate": 1.1994034938219004e-05,
      "loss": 0.0104,
      "step": 2053
    },
    {
      "epoch": 0.7875766871165644,
      "grad_norm": 0.16375118494033813,
      "learning_rate": 1.1972731146144015e-05,
      "loss": 0.0012,
      "step": 2054
    },
    {
      "epoch": 0.7879601226993865,
      "grad_norm": 7.788699150085449,
      "learning_rate": 1.1951427354069026e-05,
      "loss": 0.3816,
      "step": 2055
    },
    {
      "epoch": 0.7883435582822086,
      "grad_norm": 36.510345458984375,
      "learning_rate": 1.1930123561994035e-05,
      "loss": 0.1997,
      "step": 2056
    },
    {
      "epoch": 0.7887269938650306,
      "grad_norm": 3.886620044708252,
      "learning_rate": 1.1908819769919046e-05,
      "loss": 0.0128,
      "step": 2057
    },
    {
      "epoch": 0.7891104294478528,
      "grad_norm": 0.6223607063293457,
      "learning_rate": 1.1887515977844056e-05,
      "loss": 0.0023,
      "step": 2058
    },
    {
      "epoch": 0.7894938650306749,
      "grad_norm": 1.6074047088623047,
      "learning_rate": 1.1866212185769067e-05,
      "loss": 0.0034,
      "step": 2059
    },
    {
      "epoch": 0.7898773006134969,
      "grad_norm": 19.636877059936523,
      "learning_rate": 1.1844908393694078e-05,
      "loss": 0.0944,
      "step": 2060
    },
    {
      "epoch": 0.790260736196319,
      "grad_norm": 7.295237064361572,
      "learning_rate": 1.1823604601619089e-05,
      "loss": 0.0228,
      "step": 2061
    },
    {
      "epoch": 0.7906441717791411,
      "grad_norm": 15.375859260559082,
      "learning_rate": 1.18023008095441e-05,
      "loss": 0.0322,
      "step": 2062
    },
    {
      "epoch": 0.7910276073619632,
      "grad_norm": 116.54905700683594,
      "learning_rate": 1.178099701746911e-05,
      "loss": 6.8379,
      "step": 2063
    },
    {
      "epoch": 0.7914110429447853,
      "grad_norm": 4.101083755493164,
      "learning_rate": 1.1759693225394122e-05,
      "loss": 0.016,
      "step": 2064
    },
    {
      "epoch": 0.7917944785276073,
      "grad_norm": 290.10479736328125,
      "learning_rate": 1.1738389433319131e-05,
      "loss": 5.3223,
      "step": 2065
    },
    {
      "epoch": 0.7921779141104295,
      "grad_norm": 10.354349136352539,
      "learning_rate": 1.1717085641244142e-05,
      "loss": 0.0312,
      "step": 2066
    },
    {
      "epoch": 0.7925613496932515,
      "grad_norm": 55.29102325439453,
      "learning_rate": 1.1695781849169153e-05,
      "loss": 0.6509,
      "step": 2067
    },
    {
      "epoch": 0.7929447852760736,
      "grad_norm": 37.20539855957031,
      "learning_rate": 1.1674478057094164e-05,
      "loss": 0.6289,
      "step": 2068
    },
    {
      "epoch": 0.7933282208588958,
      "grad_norm": 19.556257247924805,
      "learning_rate": 1.1653174265019173e-05,
      "loss": 0.1942,
      "step": 2069
    },
    {
      "epoch": 0.7937116564417178,
      "grad_norm": 76.98854064941406,
      "learning_rate": 1.1631870472944184e-05,
      "loss": 0.8726,
      "step": 2070
    },
    {
      "epoch": 0.7940950920245399,
      "grad_norm": 10.885735511779785,
      "learning_rate": 1.1610566680869195e-05,
      "loss": 0.2725,
      "step": 2071
    },
    {
      "epoch": 0.7944785276073619,
      "grad_norm": 259.9325256347656,
      "learning_rate": 1.1589262888794206e-05,
      "loss": 1.6543,
      "step": 2072
    },
    {
      "epoch": 0.7948619631901841,
      "grad_norm": 23.60029411315918,
      "learning_rate": 1.1567959096719216e-05,
      "loss": 0.0136,
      "step": 2073
    },
    {
      "epoch": 0.7952453987730062,
      "grad_norm": 11.054368019104004,
      "learning_rate": 1.1546655304644227e-05,
      "loss": 0.428,
      "step": 2074
    },
    {
      "epoch": 0.7956288343558282,
      "grad_norm": 6.6344451904296875,
      "learning_rate": 1.1525351512569238e-05,
      "loss": 0.0208,
      "step": 2075
    },
    {
      "epoch": 0.7960122699386503,
      "grad_norm": 131.8979034423828,
      "learning_rate": 1.1504047720494249e-05,
      "loss": 2.4476,
      "step": 2076
    },
    {
      "epoch": 0.7963957055214724,
      "grad_norm": 138.93003845214844,
      "learning_rate": 1.148274392841926e-05,
      "loss": 6.1064,
      "step": 2077
    },
    {
      "epoch": 0.7967791411042945,
      "grad_norm": 72.7630844116211,
      "learning_rate": 1.146144013634427e-05,
      "loss": 1.6818,
      "step": 2078
    },
    {
      "epoch": 0.7971625766871165,
      "grad_norm": 0.30345648527145386,
      "learning_rate": 1.1440136344269282e-05,
      "loss": 0.0009,
      "step": 2079
    },
    {
      "epoch": 0.7975460122699386,
      "grad_norm": 80.4941635131836,
      "learning_rate": 1.1418832552194293e-05,
      "loss": 0.544,
      "step": 2080
    },
    {
      "epoch": 0.7979294478527608,
      "grad_norm": 21.229745864868164,
      "learning_rate": 1.1397528760119302e-05,
      "loss": 0.4605,
      "step": 2081
    },
    {
      "epoch": 0.7983128834355828,
      "grad_norm": 72.00467681884766,
      "learning_rate": 1.1376224968044311e-05,
      "loss": 0.8608,
      "step": 2082
    },
    {
      "epoch": 0.7986963190184049,
      "grad_norm": 57.26301193237305,
      "learning_rate": 1.1354921175969322e-05,
      "loss": 0.5435,
      "step": 2083
    },
    {
      "epoch": 0.799079754601227,
      "grad_norm": 107.73514556884766,
      "learning_rate": 1.1333617383894333e-05,
      "loss": 1.4268,
      "step": 2084
    },
    {
      "epoch": 0.7994631901840491,
      "grad_norm": 91.19815826416016,
      "learning_rate": 1.1312313591819344e-05,
      "loss": 0.7488,
      "step": 2085
    },
    {
      "epoch": 0.7998466257668712,
      "grad_norm": 1.1510038375854492,
      "learning_rate": 1.1291009799744355e-05,
      "loss": 0.005,
      "step": 2086
    },
    {
      "epoch": 0.8002300613496932,
      "grad_norm": 2.4116382598876953,
      "learning_rate": 1.1269706007669366e-05,
      "loss": 0.0055,
      "step": 2087
    },
    {
      "epoch": 0.8006134969325154,
      "grad_norm": 0.2294168919324875,
      "learning_rate": 1.1248402215594377e-05,
      "loss": 0.0013,
      "step": 2088
    },
    {
      "epoch": 0.8009969325153374,
      "grad_norm": 0.37558597326278687,
      "learning_rate": 1.1227098423519387e-05,
      "loss": 0.002,
      "step": 2089
    },
    {
      "epoch": 0.8013803680981595,
      "grad_norm": 325.6514892578125,
      "learning_rate": 1.1205794631444398e-05,
      "loss": 5.7324,
      "step": 2090
    },
    {
      "epoch": 0.8017638036809815,
      "grad_norm": 8.776749610900879,
      "learning_rate": 1.1184490839369409e-05,
      "loss": 0.0286,
      "step": 2091
    },
    {
      "epoch": 0.8021472392638037,
      "grad_norm": 78.64884185791016,
      "learning_rate": 1.1163187047294418e-05,
      "loss": 0.8096,
      "step": 2092
    },
    {
      "epoch": 0.8025306748466258,
      "grad_norm": 48.145835876464844,
      "learning_rate": 1.1141883255219429e-05,
      "loss": 0.8326,
      "step": 2093
    },
    {
      "epoch": 0.8029141104294478,
      "grad_norm": 159.64549255371094,
      "learning_rate": 1.112057946314444e-05,
      "loss": 6.3145,
      "step": 2094
    },
    {
      "epoch": 0.80329754601227,
      "grad_norm": 4.064866065979004,
      "learning_rate": 1.1099275671069451e-05,
      "loss": 0.0089,
      "step": 2095
    },
    {
      "epoch": 0.803680981595092,
      "grad_norm": 5.229132652282715,
      "learning_rate": 1.1077971878994462e-05,
      "loss": 0.0324,
      "step": 2096
    },
    {
      "epoch": 0.8040644171779141,
      "grad_norm": 0.9916634559631348,
      "learning_rate": 1.1056668086919473e-05,
      "loss": 0.0044,
      "step": 2097
    },
    {
      "epoch": 0.8044478527607362,
      "grad_norm": 4.942611217498779,
      "learning_rate": 1.1035364294844482e-05,
      "loss": 0.0221,
      "step": 2098
    },
    {
      "epoch": 0.8048312883435583,
      "grad_norm": 45.87492370605469,
      "learning_rate": 1.1014060502769493e-05,
      "loss": 0.7393,
      "step": 2099
    },
    {
      "epoch": 0.8052147239263804,
      "grad_norm": 90.27005767822266,
      "learning_rate": 1.0992756710694504e-05,
      "loss": 1.5313,
      "step": 2100
    },
    {
      "epoch": 0.8055981595092024,
      "grad_norm": 0.21876457333564758,
      "learning_rate": 1.0971452918619515e-05,
      "loss": 0.0013,
      "step": 2101
    },
    {
      "epoch": 0.8059815950920245,
      "grad_norm": 0.13131088018417358,
      "learning_rate": 1.0950149126544526e-05,
      "loss": 0.0011,
      "step": 2102
    },
    {
      "epoch": 0.8063650306748467,
      "grad_norm": 1.5603348016738892,
      "learning_rate": 1.0928845334469536e-05,
      "loss": 0.0047,
      "step": 2103
    },
    {
      "epoch": 0.8067484662576687,
      "grad_norm": 0.4653047025203705,
      "learning_rate": 1.0907541542394547e-05,
      "loss": 0.0027,
      "step": 2104
    },
    {
      "epoch": 0.8071319018404908,
      "grad_norm": 0.2466188371181488,
      "learning_rate": 1.0886237750319558e-05,
      "loss": 0.001,
      "step": 2105
    },
    {
      "epoch": 0.8075153374233128,
      "grad_norm": 2.751034736633301,
      "learning_rate": 1.0864933958244569e-05,
      "loss": 0.0074,
      "step": 2106
    },
    {
      "epoch": 0.807898773006135,
      "grad_norm": 16.325443267822266,
      "learning_rate": 1.0843630166169578e-05,
      "loss": 0.0557,
      "step": 2107
    },
    {
      "epoch": 0.808282208588957,
      "grad_norm": 1.1431059837341309,
      "learning_rate": 1.0822326374094589e-05,
      "loss": 0.0033,
      "step": 2108
    },
    {
      "epoch": 0.8086656441717791,
      "grad_norm": 5.046448707580566,
      "learning_rate": 1.08010225820196e-05,
      "loss": 0.0111,
      "step": 2109
    },
    {
      "epoch": 0.8090490797546013,
      "grad_norm": 0.3879411518573761,
      "learning_rate": 1.0779718789944611e-05,
      "loss": 0.0015,
      "step": 2110
    },
    {
      "epoch": 0.8094325153374233,
      "grad_norm": 7.425095081329346,
      "learning_rate": 1.0758414997869622e-05,
      "loss": 0.3953,
      "step": 2111
    },
    {
      "epoch": 0.8098159509202454,
      "grad_norm": 15.652329444885254,
      "learning_rate": 1.0737111205794633e-05,
      "loss": 0.0942,
      "step": 2112
    },
    {
      "epoch": 0.8101993865030674,
      "grad_norm": 9.081522941589355,
      "learning_rate": 1.0715807413719644e-05,
      "loss": 0.0416,
      "step": 2113
    },
    {
      "epoch": 0.8105828220858896,
      "grad_norm": 1.3029454946517944,
      "learning_rate": 1.0694503621644653e-05,
      "loss": 0.0029,
      "step": 2114
    },
    {
      "epoch": 0.8109662576687117,
      "grad_norm": 8.73009967803955,
      "learning_rate": 1.0673199829569663e-05,
      "loss": 0.0344,
      "step": 2115
    },
    {
      "epoch": 0.8113496932515337,
      "grad_norm": 1.926862359046936,
      "learning_rate": 1.0651896037494674e-05,
      "loss": 0.0036,
      "step": 2116
    },
    {
      "epoch": 0.8117331288343558,
      "grad_norm": 1.280436396598816,
      "learning_rate": 1.0630592245419685e-05,
      "loss": 0.0032,
      "step": 2117
    },
    {
      "epoch": 0.8121165644171779,
      "grad_norm": 0.5277543067932129,
      "learning_rate": 1.0609288453344696e-05,
      "loss": 0.0021,
      "step": 2118
    },
    {
      "epoch": 0.8125,
      "grad_norm": 21.588167190551758,
      "learning_rate": 1.0587984661269707e-05,
      "loss": 0.1061,
      "step": 2119
    },
    {
      "epoch": 0.8128834355828221,
      "grad_norm": 3.551272392272949,
      "learning_rate": 1.0566680869194718e-05,
      "loss": 0.0067,
      "step": 2120
    },
    {
      "epoch": 0.8132668711656442,
      "grad_norm": 0.41435983777046204,
      "learning_rate": 1.0545377077119729e-05,
      "loss": 0.0031,
      "step": 2121
    },
    {
      "epoch": 0.8136503067484663,
      "grad_norm": 301.73583984375,
      "learning_rate": 1.052407328504474e-05,
      "loss": 7.0078,
      "step": 2122
    },
    {
      "epoch": 0.8140337423312883,
      "grad_norm": 9.311138153076172,
      "learning_rate": 1.0502769492969749e-05,
      "loss": 0.0557,
      "step": 2123
    },
    {
      "epoch": 0.8144171779141104,
      "grad_norm": 0.23725460469722748,
      "learning_rate": 1.048146570089476e-05,
      "loss": 0.002,
      "step": 2124
    },
    {
      "epoch": 0.8148006134969326,
      "grad_norm": 7.481473922729492,
      "learning_rate": 1.0460161908819771e-05,
      "loss": 0.0067,
      "step": 2125
    },
    {
      "epoch": 0.8151840490797546,
      "grad_norm": 0.6002553105354309,
      "learning_rate": 1.043885811674478e-05,
      "loss": 0.0023,
      "step": 2126
    },
    {
      "epoch": 0.8155674846625767,
      "grad_norm": 93.15229797363281,
      "learning_rate": 1.0417554324669791e-05,
      "loss": 1.0733,
      "step": 2127
    },
    {
      "epoch": 0.8159509202453987,
      "grad_norm": 11.985151290893555,
      "learning_rate": 1.0396250532594802e-05,
      "loss": 0.0348,
      "step": 2128
    },
    {
      "epoch": 0.8163343558282209,
      "grad_norm": 11.536222457885742,
      "learning_rate": 1.0374946740519813e-05,
      "loss": 0.4173,
      "step": 2129
    },
    {
      "epoch": 0.816717791411043,
      "grad_norm": 8.33538818359375,
      "learning_rate": 1.0353642948444824e-05,
      "loss": 0.1705,
      "step": 2130
    },
    {
      "epoch": 0.817101226993865,
      "grad_norm": 3.2708263397216797,
      "learning_rate": 1.0332339156369834e-05,
      "loss": 0.375,
      "step": 2131
    },
    {
      "epoch": 0.8174846625766872,
      "grad_norm": 148.76092529296875,
      "learning_rate": 1.0311035364294845e-05,
      "loss": 6.4873,
      "step": 2132
    },
    {
      "epoch": 0.8178680981595092,
      "grad_norm": 1.8550126552581787,
      "learning_rate": 1.0289731572219856e-05,
      "loss": 0.0058,
      "step": 2133
    },
    {
      "epoch": 0.8182515337423313,
      "grad_norm": 25.314233779907227,
      "learning_rate": 1.0268427780144867e-05,
      "loss": 0.6489,
      "step": 2134
    },
    {
      "epoch": 0.8186349693251533,
      "grad_norm": 19.60906219482422,
      "learning_rate": 1.0247123988069878e-05,
      "loss": 0.4819,
      "step": 2135
    },
    {
      "epoch": 0.8190184049079755,
      "grad_norm": 14.061655044555664,
      "learning_rate": 1.0225820195994889e-05,
      "loss": 0.0454,
      "step": 2136
    },
    {
      "epoch": 0.8194018404907976,
      "grad_norm": 5.1389923095703125,
      "learning_rate": 1.0204516403919898e-05,
      "loss": 0.0107,
      "step": 2137
    },
    {
      "epoch": 0.8197852760736196,
      "grad_norm": 1.9320876598358154,
      "learning_rate": 1.0183212611844909e-05,
      "loss": 0.0059,
      "step": 2138
    },
    {
      "epoch": 0.8201687116564417,
      "grad_norm": 0.8540799021720886,
      "learning_rate": 1.016190881976992e-05,
      "loss": 0.0031,
      "step": 2139
    },
    {
      "epoch": 0.8205521472392638,
      "grad_norm": 11.2532377243042,
      "learning_rate": 1.014060502769493e-05,
      "loss": 0.4314,
      "step": 2140
    },
    {
      "epoch": 0.8209355828220859,
      "grad_norm": 146.77346801757812,
      "learning_rate": 1.011930123561994e-05,
      "loss": 4.4738,
      "step": 2141
    },
    {
      "epoch": 0.821319018404908,
      "grad_norm": 0.3676358163356781,
      "learning_rate": 1.0097997443544951e-05,
      "loss": 0.0019,
      "step": 2142
    },
    {
      "epoch": 0.82170245398773,
      "grad_norm": 101.42230224609375,
      "learning_rate": 1.0076693651469962e-05,
      "loss": 1.8779,
      "step": 2143
    },
    {
      "epoch": 0.8220858895705522,
      "grad_norm": 0.34330686926841736,
      "learning_rate": 1.0055389859394973e-05,
      "loss": 0.0016,
      "step": 2144
    },
    {
      "epoch": 0.8224693251533742,
      "grad_norm": 178.65760803222656,
      "learning_rate": 1.0034086067319984e-05,
      "loss": 1.7715,
      "step": 2145
    },
    {
      "epoch": 0.8228527607361963,
      "grad_norm": 485.91131591796875,
      "learning_rate": 1.0012782275244995e-05,
      "loss": 5.7656,
      "step": 2146
    },
    {
      "epoch": 0.8232361963190185,
      "grad_norm": 7.863787651062012,
      "learning_rate": 9.991478483170005e-06,
      "loss": 0.3987,
      "step": 2147
    },
    {
      "epoch": 0.8236196319018405,
      "grad_norm": 34.286102294921875,
      "learning_rate": 9.970174691095016e-06,
      "loss": 0.625,
      "step": 2148
    },
    {
      "epoch": 0.8240030674846626,
      "grad_norm": 48.19886016845703,
      "learning_rate": 9.948870899020025e-06,
      "loss": 0.6436,
      "step": 2149
    },
    {
      "epoch": 0.8243865030674846,
      "grad_norm": 202.91395568847656,
      "learning_rate": 9.927567106945036e-06,
      "loss": 3.4361,
      "step": 2150
    },
    {
      "epoch": 0.8247699386503068,
      "grad_norm": 14.648388862609863,
      "learning_rate": 9.906263314870047e-06,
      "loss": 0.4126,
      "step": 2151
    },
    {
      "epoch": 0.8251533742331288,
      "grad_norm": 16.33935546875,
      "learning_rate": 9.884959522795058e-06,
      "loss": 0.0564,
      "step": 2152
    },
    {
      "epoch": 0.8255368098159509,
      "grad_norm": 86.1637954711914,
      "learning_rate": 9.863655730720069e-06,
      "loss": 1.294,
      "step": 2153
    },
    {
      "epoch": 0.825920245398773,
      "grad_norm": 3.0902631282806396,
      "learning_rate": 9.84235193864508e-06,
      "loss": 0.0195,
      "step": 2154
    },
    {
      "epoch": 0.8263036809815951,
      "grad_norm": 0.3026873767375946,
      "learning_rate": 9.821048146570091e-06,
      "loss": 0.0021,
      "step": 2155
    },
    {
      "epoch": 0.8266871165644172,
      "grad_norm": 198.00852966308594,
      "learning_rate": 9.7997443544951e-06,
      "loss": 3.0872,
      "step": 2156
    },
    {
      "epoch": 0.8270705521472392,
      "grad_norm": 261.3690490722656,
      "learning_rate": 9.778440562420111e-06,
      "loss": 1.6778,
      "step": 2157
    },
    {
      "epoch": 0.8274539877300614,
      "grad_norm": 4.099433422088623,
      "learning_rate": 9.757136770345122e-06,
      "loss": 0.0214,
      "step": 2158
    },
    {
      "epoch": 0.8278374233128835,
      "grad_norm": 1.4283417463302612,
      "learning_rate": 9.735832978270133e-06,
      "loss": 0.0041,
      "step": 2159
    },
    {
      "epoch": 0.8282208588957055,
      "grad_norm": 181.2616424560547,
      "learning_rate": 9.714529186195143e-06,
      "loss": 1.8028,
      "step": 2160
    },
    {
      "epoch": 0.8286042944785276,
      "grad_norm": 7.175016403198242,
      "learning_rate": 9.693225394120154e-06,
      "loss": 0.0357,
      "step": 2161
    },
    {
      "epoch": 0.8289877300613497,
      "grad_norm": 1.671427845954895,
      "learning_rate": 9.671921602045165e-06,
      "loss": 0.0034,
      "step": 2162
    },
    {
      "epoch": 0.8293711656441718,
      "grad_norm": 16.254331588745117,
      "learning_rate": 9.650617809970176e-06,
      "loss": 0.42,
      "step": 2163
    },
    {
      "epoch": 0.8297546012269938,
      "grad_norm": 213.69357299804688,
      "learning_rate": 9.629314017895187e-06,
      "loss": 6.9487,
      "step": 2164
    },
    {
      "epoch": 0.8301380368098159,
      "grad_norm": 4.822175025939941,
      "learning_rate": 9.608010225820196e-06,
      "loss": 0.017,
      "step": 2165
    },
    {
      "epoch": 0.8305214723926381,
      "grad_norm": 5.070052146911621,
      "learning_rate": 9.586706433745207e-06,
      "loss": 0.0058,
      "step": 2166
    },
    {
      "epoch": 0.8309049079754601,
      "grad_norm": 3.2388699054718018,
      "learning_rate": 9.565402641670218e-06,
      "loss": 0.0139,
      "step": 2167
    },
    {
      "epoch": 0.8312883435582822,
      "grad_norm": 15.923335075378418,
      "learning_rate": 9.544098849595229e-06,
      "loss": 0.0522,
      "step": 2168
    },
    {
      "epoch": 0.8316717791411042,
      "grad_norm": 0.178175151348114,
      "learning_rate": 9.52279505752024e-06,
      "loss": 0.0014,
      "step": 2169
    },
    {
      "epoch": 0.8320552147239264,
      "grad_norm": 9.746631622314453,
      "learning_rate": 9.501491265445251e-06,
      "loss": 0.0461,
      "step": 2170
    },
    {
      "epoch": 0.8324386503067485,
      "grad_norm": 0.3492185175418854,
      "learning_rate": 9.48018747337026e-06,
      "loss": 0.0013,
      "step": 2171
    },
    {
      "epoch": 0.8328220858895705,
      "grad_norm": 22.301040649414062,
      "learning_rate": 9.458883681295271e-06,
      "loss": 0.141,
      "step": 2172
    },
    {
      "epoch": 0.8332055214723927,
      "grad_norm": 211.20962524414062,
      "learning_rate": 9.43757988922028e-06,
      "loss": 2.256,
      "step": 2173
    },
    {
      "epoch": 0.8335889570552147,
      "grad_norm": 27.95379066467285,
      "learning_rate": 9.416276097145292e-06,
      "loss": 0.4873,
      "step": 2174
    },
    {
      "epoch": 0.8339723926380368,
      "grad_norm": 224.555419921875,
      "learning_rate": 9.394972305070303e-06,
      "loss": 1.7041,
      "step": 2175
    },
    {
      "epoch": 0.8343558282208589,
      "grad_norm": 68.85265350341797,
      "learning_rate": 9.373668512995314e-06,
      "loss": 0.9019,
      "step": 2176
    },
    {
      "epoch": 0.834739263803681,
      "grad_norm": 118.31787872314453,
      "learning_rate": 9.352364720920325e-06,
      "loss": 1.0137,
      "step": 2177
    },
    {
      "epoch": 0.8351226993865031,
      "grad_norm": 33.278621673583984,
      "learning_rate": 9.331060928845336e-06,
      "loss": 0.6021,
      "step": 2178
    },
    {
      "epoch": 0.8355061349693251,
      "grad_norm": 0.8133134245872498,
      "learning_rate": 9.309757136770347e-06,
      "loss": 0.0033,
      "step": 2179
    },
    {
      "epoch": 0.8358895705521472,
      "grad_norm": 0.19997651875019073,
      "learning_rate": 9.288453344695358e-06,
      "loss": 0.0019,
      "step": 2180
    },
    {
      "epoch": 0.8362730061349694,
      "grad_norm": 35.009544372558594,
      "learning_rate": 9.267149552620367e-06,
      "loss": 0.2026,
      "step": 2181
    },
    {
      "epoch": 0.8366564417177914,
      "grad_norm": 11.999231338500977,
      "learning_rate": 9.245845760545376e-06,
      "loss": 0.0838,
      "step": 2182
    },
    {
      "epoch": 0.8370398773006135,
      "grad_norm": 12.60776138305664,
      "learning_rate": 9.224541968470387e-06,
      "loss": 0.0487,
      "step": 2183
    },
    {
      "epoch": 0.8374233128834356,
      "grad_norm": 0.29287654161453247,
      "learning_rate": 9.203238176395398e-06,
      "loss": 0.0014,
      "step": 2184
    },
    {
      "epoch": 0.8378067484662577,
      "grad_norm": 74.42601013183594,
      "learning_rate": 9.18193438432041e-06,
      "loss": 1.1533,
      "step": 2185
    },
    {
      "epoch": 0.8381901840490797,
      "grad_norm": 8.74606704711914,
      "learning_rate": 9.16063059224542e-06,
      "loss": 0.0355,
      "step": 2186
    },
    {
      "epoch": 0.8385736196319018,
      "grad_norm": 47.63106155395508,
      "learning_rate": 9.139326800170431e-06,
      "loss": 0.5376,
      "step": 2187
    },
    {
      "epoch": 0.838957055214724,
      "grad_norm": 0.6941925287246704,
      "learning_rate": 9.118023008095442e-06,
      "loss": 0.0024,
      "step": 2188
    },
    {
      "epoch": 0.839340490797546,
      "grad_norm": 7.404911994934082,
      "learning_rate": 9.096719216020452e-06,
      "loss": 0.3897,
      "step": 2189
    },
    {
      "epoch": 0.8397239263803681,
      "grad_norm": 0.8754544854164124,
      "learning_rate": 9.075415423945463e-06,
      "loss": 0.0029,
      "step": 2190
    },
    {
      "epoch": 0.8401073619631901,
      "grad_norm": 134.01356506347656,
      "learning_rate": 9.054111631870474e-06,
      "loss": 6.3232,
      "step": 2191
    },
    {
      "epoch": 0.8404907975460123,
      "grad_norm": 0.8344331383705139,
      "learning_rate": 9.032807839795485e-06,
      "loss": 0.0023,
      "step": 2192
    },
    {
      "epoch": 0.8408742331288344,
      "grad_norm": 0.300471693277359,
      "learning_rate": 9.011504047720494e-06,
      "loss": 0.0011,
      "step": 2193
    },
    {
      "epoch": 0.8412576687116564,
      "grad_norm": 585.5133056640625,
      "learning_rate": 8.990200255645505e-06,
      "loss": 5.4443,
      "step": 2194
    },
    {
      "epoch": 0.8416411042944786,
      "grad_norm": 6.138875484466553,
      "learning_rate": 8.968896463570516e-06,
      "loss": 0.0332,
      "step": 2195
    },
    {
      "epoch": 0.8420245398773006,
      "grad_norm": 4.308825492858887,
      "learning_rate": 8.947592671495527e-06,
      "loss": 0.0201,
      "step": 2196
    },
    {
      "epoch": 0.8424079754601227,
      "grad_norm": 64.83824157714844,
      "learning_rate": 8.926288879420538e-06,
      "loss": 0.7031,
      "step": 2197
    },
    {
      "epoch": 0.8427914110429447,
      "grad_norm": 45.091739654541016,
      "learning_rate": 8.904985087345547e-06,
      "loss": 0.5522,
      "step": 2198
    },
    {
      "epoch": 0.8431748466257669,
      "grad_norm": 15.416338920593262,
      "learning_rate": 8.883681295270558e-06,
      "loss": 0.4653,
      "step": 2199
    },
    {
      "epoch": 0.843558282208589,
      "grad_norm": 74.4372787475586,
      "learning_rate": 8.86237750319557e-06,
      "loss": 1.0098,
      "step": 2200
    },
    {
      "epoch": 0.843941717791411,
      "grad_norm": 36.97453308105469,
      "learning_rate": 8.84107371112058e-06,
      "loss": 0.5796,
      "step": 2201
    },
    {
      "epoch": 0.8443251533742331,
      "grad_norm": 29.4422607421875,
      "learning_rate": 8.819769919045591e-06,
      "loss": 0.5654,
      "step": 2202
    },
    {
      "epoch": 0.8447085889570553,
      "grad_norm": 4.072035312652588,
      "learning_rate": 8.798466126970602e-06,
      "loss": 0.0087,
      "step": 2203
    },
    {
      "epoch": 0.8450920245398773,
      "grad_norm": 0.17604690790176392,
      "learning_rate": 8.777162334895613e-06,
      "loss": 0.0011,
      "step": 2204
    },
    {
      "epoch": 0.8454754601226994,
      "grad_norm": 0.0971691682934761,
      "learning_rate": 8.755858542820623e-06,
      "loss": 0.0007,
      "step": 2205
    },
    {
      "epoch": 0.8458588957055214,
      "grad_norm": 0.4734111726284027,
      "learning_rate": 8.734554750745632e-06,
      "loss": 0.0015,
      "step": 2206
    },
    {
      "epoch": 0.8462423312883436,
      "grad_norm": 0.7215316295623779,
      "learning_rate": 8.713250958670643e-06,
      "loss": 0.0026,
      "step": 2207
    },
    {
      "epoch": 0.8466257668711656,
      "grad_norm": 11.359012603759766,
      "learning_rate": 8.691947166595654e-06,
      "loss": 0.0755,
      "step": 2208
    },
    {
      "epoch": 0.8470092024539877,
      "grad_norm": 36.86509323120117,
      "learning_rate": 8.670643374520665e-06,
      "loss": 0.2292,
      "step": 2209
    },
    {
      "epoch": 0.8473926380368099,
      "grad_norm": 14.986333847045898,
      "learning_rate": 8.649339582445676e-06,
      "loss": 0.2059,
      "step": 2210
    },
    {
      "epoch": 0.8477760736196319,
      "grad_norm": 14.543253898620605,
      "learning_rate": 8.628035790370687e-06,
      "loss": 0.0939,
      "step": 2211
    },
    {
      "epoch": 0.848159509202454,
      "grad_norm": 133.95114135742188,
      "learning_rate": 8.606731998295698e-06,
      "loss": 1.671,
      "step": 2212
    },
    {
      "epoch": 0.848542944785276,
      "grad_norm": 9.932710647583008,
      "learning_rate": 8.585428206220709e-06,
      "loss": 0.0348,
      "step": 2213
    },
    {
      "epoch": 0.8489263803680982,
      "grad_norm": 0.16129763424396515,
      "learning_rate": 8.564124414145718e-06,
      "loss": 0.0012,
      "step": 2214
    },
    {
      "epoch": 0.8493098159509203,
      "grad_norm": 0.8886722922325134,
      "learning_rate": 8.54282062207073e-06,
      "loss": 0.0034,
      "step": 2215
    },
    {
      "epoch": 0.8496932515337423,
      "grad_norm": 82.60255432128906,
      "learning_rate": 8.521516829995738e-06,
      "loss": 1.5283,
      "step": 2216
    },
    {
      "epoch": 0.8500766871165644,
      "grad_norm": 17.767559051513672,
      "learning_rate": 8.50021303792075e-06,
      "loss": 0.4824,
      "step": 2217
    },
    {
      "epoch": 0.8504601226993865,
      "grad_norm": 110.89055633544922,
      "learning_rate": 8.47890924584576e-06,
      "loss": 5.498,
      "step": 2218
    },
    {
      "epoch": 0.8508435582822086,
      "grad_norm": 1.0662866830825806,
      "learning_rate": 8.457605453770772e-06,
      "loss": 0.0029,
      "step": 2219
    },
    {
      "epoch": 0.8512269938650306,
      "grad_norm": 2.306121587753296,
      "learning_rate": 8.436301661695783e-06,
      "loss": 0.0028,
      "step": 2220
    },
    {
      "epoch": 0.8516104294478528,
      "grad_norm": 30.38901710510254,
      "learning_rate": 8.414997869620794e-06,
      "loss": 0.6836,
      "step": 2221
    },
    {
      "epoch": 0.8519938650306749,
      "grad_norm": 9.880842208862305,
      "learning_rate": 8.393694077545803e-06,
      "loss": 0.2307,
      "step": 2222
    },
    {
      "epoch": 0.8523773006134969,
      "grad_norm": 0.7096598148345947,
      "learning_rate": 8.372390285470814e-06,
      "loss": 0.0046,
      "step": 2223
    },
    {
      "epoch": 0.852760736196319,
      "grad_norm": 19.66064453125,
      "learning_rate": 8.351086493395825e-06,
      "loss": 0.4373,
      "step": 2224
    },
    {
      "epoch": 0.8531441717791411,
      "grad_norm": 40.124759674072266,
      "learning_rate": 8.329782701320836e-06,
      "loss": 0.4824,
      "step": 2225
    },
    {
      "epoch": 0.8535276073619632,
      "grad_norm": 1.8273738622665405,
      "learning_rate": 8.308478909245847e-06,
      "loss": 0.0027,
      "step": 2226
    },
    {
      "epoch": 0.8539110429447853,
      "grad_norm": 0.40088343620300293,
      "learning_rate": 8.287175117170856e-06,
      "loss": 0.0023,
      "step": 2227
    },
    {
      "epoch": 0.8542944785276073,
      "grad_norm": 1.148898959159851,
      "learning_rate": 8.265871325095867e-06,
      "loss": 0.0057,
      "step": 2228
    },
    {
      "epoch": 0.8546779141104295,
      "grad_norm": 6.235866546630859,
      "learning_rate": 8.244567533020878e-06,
      "loss": 0.398,
      "step": 2229
    },
    {
      "epoch": 0.8550613496932515,
      "grad_norm": 1.2773715257644653,
      "learning_rate": 8.22326374094589e-06,
      "loss": 0.0041,
      "step": 2230
    },
    {
      "epoch": 0.8554447852760736,
      "grad_norm": 5.057631015777588,
      "learning_rate": 8.201959948870898e-06,
      "loss": 0.0131,
      "step": 2231
    },
    {
      "epoch": 0.8558282208588958,
      "grad_norm": 1.29682195186615,
      "learning_rate": 8.18065615679591e-06,
      "loss": 0.0036,
      "step": 2232
    },
    {
      "epoch": 0.8562116564417178,
      "grad_norm": 166.40093994140625,
      "learning_rate": 8.15935236472092e-06,
      "loss": 1.7452,
      "step": 2233
    },
    {
      "epoch": 0.8565950920245399,
      "grad_norm": 15.131058692932129,
      "learning_rate": 8.138048572645932e-06,
      "loss": 0.1754,
      "step": 2234
    },
    {
      "epoch": 0.8569785276073619,
      "grad_norm": 63.33060836791992,
      "learning_rate": 8.116744780570943e-06,
      "loss": 0.9424,
      "step": 2235
    },
    {
      "epoch": 0.8573619631901841,
      "grad_norm": 34.5869140625,
      "learning_rate": 8.095440988495954e-06,
      "loss": 0.2004,
      "step": 2236
    },
    {
      "epoch": 0.8577453987730062,
      "grad_norm": 0.4424964189529419,
      "learning_rate": 8.074137196420965e-06,
      "loss": 0.0014,
      "step": 2237
    },
    {
      "epoch": 0.8581288343558282,
      "grad_norm": 0.6975522637367249,
      "learning_rate": 8.052833404345974e-06,
      "loss": 0.0038,
      "step": 2238
    },
    {
      "epoch": 0.8585122699386503,
      "grad_norm": 1.3421223163604736,
      "learning_rate": 8.031529612270985e-06,
      "loss": 0.0017,
      "step": 2239
    },
    {
      "epoch": 0.8588957055214724,
      "grad_norm": 1.981398344039917,
      "learning_rate": 8.010225820195994e-06,
      "loss": 0.0048,
      "step": 2240
    },
    {
      "epoch": 0.8592791411042945,
      "grad_norm": 130.21043395996094,
      "learning_rate": 7.988922028121005e-06,
      "loss": 6.0957,
      "step": 2241
    },
    {
      "epoch": 0.8596625766871165,
      "grad_norm": 132.2084197998047,
      "learning_rate": 7.967618236046016e-06,
      "loss": 1.5684,
      "step": 2242
    },
    {
      "epoch": 0.8600460122699386,
      "grad_norm": 120.15467834472656,
      "learning_rate": 7.946314443971027e-06,
      "loss": 6.2344,
      "step": 2243
    },
    {
      "epoch": 0.8604294478527608,
      "grad_norm": 30.702177047729492,
      "learning_rate": 7.925010651896038e-06,
      "loss": 0.1836,
      "step": 2244
    },
    {
      "epoch": 0.8608128834355828,
      "grad_norm": 5.360868453979492,
      "learning_rate": 7.90370685982105e-06,
      "loss": 0.0084,
      "step": 2245
    },
    {
      "epoch": 0.8611963190184049,
      "grad_norm": 284.1966247558594,
      "learning_rate": 7.88240306774606e-06,
      "loss": 2.4767,
      "step": 2246
    },
    {
      "epoch": 0.861579754601227,
      "grad_norm": 0.34775063395500183,
      "learning_rate": 7.86109927567107e-06,
      "loss": 0.0014,
      "step": 2247
    },
    {
      "epoch": 0.8619631901840491,
      "grad_norm": 27.3027286529541,
      "learning_rate": 7.83979548359608e-06,
      "loss": 0.1621,
      "step": 2248
    },
    {
      "epoch": 0.8623466257668712,
      "grad_norm": 71.96947479248047,
      "learning_rate": 7.818491691521092e-06,
      "loss": 0.6729,
      "step": 2249
    },
    {
      "epoch": 0.8627300613496932,
      "grad_norm": 172.47738647460938,
      "learning_rate": 7.7971878994461e-06,
      "loss": 3.5972,
      "step": 2250
    },
    {
      "epoch": 0.8631134969325154,
      "grad_norm": 0.444509893655777,
      "learning_rate": 7.775884107371112e-06,
      "loss": 0.0018,
      "step": 2251
    },
    {
      "epoch": 0.8634969325153374,
      "grad_norm": 102.0246810913086,
      "learning_rate": 7.754580315296123e-06,
      "loss": 1.4697,
      "step": 2252
    },
    {
      "epoch": 0.8638803680981595,
      "grad_norm": 4.7411394119262695,
      "learning_rate": 7.733276523221134e-06,
      "loss": 0.0111,
      "step": 2253
    },
    {
      "epoch": 0.8642638036809815,
      "grad_norm": 25.057506561279297,
      "learning_rate": 7.711972731146145e-06,
      "loss": 0.4507,
      "step": 2254
    },
    {
      "epoch": 0.8646472392638037,
      "grad_norm": 26.792598724365234,
      "learning_rate": 7.690668939071156e-06,
      "loss": 0.4717,
      "step": 2255
    },
    {
      "epoch": 0.8650306748466258,
      "grad_norm": 32.75210189819336,
      "learning_rate": 7.669365146996165e-06,
      "loss": 0.6104,
      "step": 2256
    },
    {
      "epoch": 0.8654141104294478,
      "grad_norm": 291.1597900390625,
      "learning_rate": 7.648061354921176e-06,
      "loss": 7.042,
      "step": 2257
    },
    {
      "epoch": 0.86579754601227,
      "grad_norm": 0.7177117466926575,
      "learning_rate": 7.626757562846187e-06,
      "loss": 0.0017,
      "step": 2258
    },
    {
      "epoch": 0.866180981595092,
      "grad_norm": 0.7613211870193481,
      "learning_rate": 7.605453770771198e-06,
      "loss": 0.0028,
      "step": 2259
    },
    {
      "epoch": 0.8665644171779141,
      "grad_norm": 0.4335547983646393,
      "learning_rate": 7.584149978696209e-06,
      "loss": 0.002,
      "step": 2260
    },
    {
      "epoch": 0.8669478527607362,
      "grad_norm": 0.32038646936416626,
      "learning_rate": 7.5628461866212185e-06,
      "loss": 0.0015,
      "step": 2261
    },
    {
      "epoch": 0.8673312883435583,
      "grad_norm": 84.02080535888672,
      "learning_rate": 7.5415423945462295e-06,
      "loss": 0.9361,
      "step": 2262
    },
    {
      "epoch": 0.8677147239263804,
      "grad_norm": 2.1479883193969727,
      "learning_rate": 7.52023860247124e-06,
      "loss": 0.0058,
      "step": 2263
    },
    {
      "epoch": 0.8680981595092024,
      "grad_norm": 87.77676391601562,
      "learning_rate": 7.498934810396251e-06,
      "loss": 0.9238,
      "step": 2264
    },
    {
      "epoch": 0.8684815950920245,
      "grad_norm": 38.867977142333984,
      "learning_rate": 7.477631018321262e-06,
      "loss": 0.6719,
      "step": 2265
    },
    {
      "epoch": 0.8688650306748467,
      "grad_norm": 4.981602191925049,
      "learning_rate": 7.456327226246272e-06,
      "loss": 0.0198,
      "step": 2266
    },
    {
      "epoch": 0.8692484662576687,
      "grad_norm": 115.20155334472656,
      "learning_rate": 7.435023434171283e-06,
      "loss": 1.0859,
      "step": 2267
    },
    {
      "epoch": 0.8696319018404908,
      "grad_norm": 81.47464752197266,
      "learning_rate": 7.413719642096294e-06,
      "loss": 0.7612,
      "step": 2268
    },
    {
      "epoch": 0.8700153374233128,
      "grad_norm": 1.5404905080795288,
      "learning_rate": 7.392415850021305e-06,
      "loss": 0.0041,
      "step": 2269
    },
    {
      "epoch": 0.870398773006135,
      "grad_norm": 18.807405471801758,
      "learning_rate": 7.371112057946315e-06,
      "loss": 0.469,
      "step": 2270
    },
    {
      "epoch": 0.870782208588957,
      "grad_norm": 3.0587081909179688,
      "learning_rate": 7.349808265871326e-06,
      "loss": 0.0128,
      "step": 2271
    },
    {
      "epoch": 0.8711656441717791,
      "grad_norm": 4.244951248168945,
      "learning_rate": 7.328504473796335e-06,
      "loss": 0.363,
      "step": 2272
    },
    {
      "epoch": 0.8715490797546013,
      "grad_norm": 176.03610229492188,
      "learning_rate": 7.307200681721346e-06,
      "loss": 0.9551,
      "step": 2273
    },
    {
      "epoch": 0.8719325153374233,
      "grad_norm": 8.542078971862793,
      "learning_rate": 7.285896889646357e-06,
      "loss": 0.0167,
      "step": 2274
    },
    {
      "epoch": 0.8723159509202454,
      "grad_norm": 0.5152387619018555,
      "learning_rate": 7.2645930975713675e-06,
      "loss": 0.0023,
      "step": 2275
    },
    {
      "epoch": 0.8726993865030674,
      "grad_norm": 1.4831758737564087,
      "learning_rate": 7.2432893054963785e-06,
      "loss": 0.0041,
      "step": 2276
    },
    {
      "epoch": 0.8730828220858896,
      "grad_norm": 43.16259765625,
      "learning_rate": 7.2219855134213895e-06,
      "loss": 1.0479,
      "step": 2277
    },
    {
      "epoch": 0.8734662576687117,
      "grad_norm": 145.6817169189453,
      "learning_rate": 7.2006817213464e-06,
      "loss": 6.7085,
      "step": 2278
    },
    {
      "epoch": 0.8738496932515337,
      "grad_norm": 64.62080383300781,
      "learning_rate": 7.179377929271411e-06,
      "loss": 0.6709,
      "step": 2279
    },
    {
      "epoch": 0.8742331288343558,
      "grad_norm": 164.92628479003906,
      "learning_rate": 7.158074137196422e-06,
      "loss": 2.2658,
      "step": 2280
    },
    {
      "epoch": 0.8746165644171779,
      "grad_norm": 1.5775574445724487,
      "learning_rate": 7.136770345121433e-06,
      "loss": 0.0039,
      "step": 2281
    },
    {
      "epoch": 0.875,
      "grad_norm": 84.53050994873047,
      "learning_rate": 7.115466553046443e-06,
      "loss": 1.6895,
      "step": 2282
    },
    {
      "epoch": 0.8753834355828221,
      "grad_norm": 109.07520294189453,
      "learning_rate": 7.094162760971454e-06,
      "loss": 1.4405,
      "step": 2283
    },
    {
      "epoch": 0.8757668711656442,
      "grad_norm": 22.942350387573242,
      "learning_rate": 7.072858968896463e-06,
      "loss": 0.5737,
      "step": 2284
    },
    {
      "epoch": 0.8761503067484663,
      "grad_norm": 17.40251350402832,
      "learning_rate": 7.051555176821474e-06,
      "loss": 0.2432,
      "step": 2285
    },
    {
      "epoch": 0.8765337423312883,
      "grad_norm": 84.14927673339844,
      "learning_rate": 7.030251384746485e-06,
      "loss": 0.9087,
      "step": 2286
    },
    {
      "epoch": 0.8769171779141104,
      "grad_norm": 11.85714340209961,
      "learning_rate": 7.008947592671495e-06,
      "loss": 0.2301,
      "step": 2287
    },
    {
      "epoch": 0.8773006134969326,
      "grad_norm": 125.55137634277344,
      "learning_rate": 6.987643800596506e-06,
      "loss": 6.3281,
      "step": 2288
    },
    {
      "epoch": 0.8776840490797546,
      "grad_norm": 231.6448211669922,
      "learning_rate": 6.966340008521517e-06,
      "loss": 3.7931,
      "step": 2289
    },
    {
      "epoch": 0.8780674846625767,
      "grad_norm": 0.2826296389102936,
      "learning_rate": 6.945036216446528e-06,
      "loss": 0.0011,
      "step": 2290
    },
    {
      "epoch": 0.8784509202453987,
      "grad_norm": 20.809093475341797,
      "learning_rate": 6.9237324243715385e-06,
      "loss": 0.4322,
      "step": 2291
    },
    {
      "epoch": 0.8788343558282209,
      "grad_norm": 0.10539169609546661,
      "learning_rate": 6.9024286322965495e-06,
      "loss": 0.0009,
      "step": 2292
    },
    {
      "epoch": 0.879217791411043,
      "grad_norm": 2.3163046836853027,
      "learning_rate": 6.8811248402215605e-06,
      "loss": 0.0087,
      "step": 2293
    },
    {
      "epoch": 0.879601226993865,
      "grad_norm": 48.467830657958984,
      "learning_rate": 6.859821048146571e-06,
      "loss": 1.0156,
      "step": 2294
    },
    {
      "epoch": 0.8799846625766872,
      "grad_norm": 2.4319138526916504,
      "learning_rate": 6.838517256071581e-06,
      "loss": 0.0056,
      "step": 2295
    },
    {
      "epoch": 0.8803680981595092,
      "grad_norm": 28.968769073486328,
      "learning_rate": 6.817213463996591e-06,
      "loss": 0.4573,
      "step": 2296
    },
    {
      "epoch": 0.8807515337423313,
      "grad_norm": 2.2300775051116943,
      "learning_rate": 6.795909671921602e-06,
      "loss": 0.0071,
      "step": 2297
    },
    {
      "epoch": 0.8811349693251533,
      "grad_norm": 0.5528029799461365,
      "learning_rate": 6.774605879846613e-06,
      "loss": 0.0016,
      "step": 2298
    },
    {
      "epoch": 0.8815184049079755,
      "grad_norm": 2.7337145805358887,
      "learning_rate": 6.753302087771623e-06,
      "loss": 0.0087,
      "step": 2299
    },
    {
      "epoch": 0.8819018404907976,
      "grad_norm": 0.2805922031402588,
      "learning_rate": 6.731998295696634e-06,
      "loss": 0.0018,
      "step": 2300
    },
    {
      "epoch": 0.8822852760736196,
      "grad_norm": 94.86458587646484,
      "learning_rate": 6.710694503621645e-06,
      "loss": 1.3252,
      "step": 2301
    },
    {
      "epoch": 0.8826687116564417,
      "grad_norm": 12.665270805358887,
      "learning_rate": 6.689390711546656e-06,
      "loss": 0.0423,
      "step": 2302
    },
    {
      "epoch": 0.8830521472392638,
      "grad_norm": 91.64434051513672,
      "learning_rate": 6.668086919471666e-06,
      "loss": 1.9256,
      "step": 2303
    },
    {
      "epoch": 0.8834355828220859,
      "grad_norm": 29.66964340209961,
      "learning_rate": 6.646783127396677e-06,
      "loss": 0.0693,
      "step": 2304
    },
    {
      "epoch": 0.883819018404908,
      "grad_norm": 56.94545364379883,
      "learning_rate": 6.625479335321688e-06,
      "loss": 0.8106,
      "step": 2305
    },
    {
      "epoch": 0.88420245398773,
      "grad_norm": 0.6294219493865967,
      "learning_rate": 6.604175543246698e-06,
      "loss": 0.0022,
      "step": 2306
    },
    {
      "epoch": 0.8845858895705522,
      "grad_norm": 4.835546493530273,
      "learning_rate": 6.582871751171709e-06,
      "loss": 0.3618,
      "step": 2307
    },
    {
      "epoch": 0.8849693251533742,
      "grad_norm": 21.996061325073242,
      "learning_rate": 6.561567959096719e-06,
      "loss": 0.4268,
      "step": 2308
    },
    {
      "epoch": 0.8853527607361963,
      "grad_norm": 189.7275848388672,
      "learning_rate": 6.54026416702173e-06,
      "loss": 5.3691,
      "step": 2309
    },
    {
      "epoch": 0.8857361963190185,
      "grad_norm": 368.8897399902344,
      "learning_rate": 6.518960374946741e-06,
      "loss": 3.4177,
      "step": 2310
    },
    {
      "epoch": 0.8861196319018405,
      "grad_norm": 18.67096519470215,
      "learning_rate": 6.497656582871752e-06,
      "loss": 0.543,
      "step": 2311
    },
    {
      "epoch": 0.8865030674846626,
      "grad_norm": 0.8498499989509583,
      "learning_rate": 6.476352790796762e-06,
      "loss": 0.0019,
      "step": 2312
    },
    {
      "epoch": 0.8868865030674846,
      "grad_norm": 61.36403274536133,
      "learning_rate": 6.455048998721773e-06,
      "loss": 0.5713,
      "step": 2313
    },
    {
      "epoch": 0.8872699386503068,
      "grad_norm": 92.98079681396484,
      "learning_rate": 6.433745206646784e-06,
      "loss": 7.3711,
      "step": 2314
    },
    {
      "epoch": 0.8876533742331288,
      "grad_norm": 0.798412561416626,
      "learning_rate": 6.412441414571794e-06,
      "loss": 0.002,
      "step": 2315
    },
    {
      "epoch": 0.8880368098159509,
      "grad_norm": 16.507095336914062,
      "learning_rate": 6.391137622496805e-06,
      "loss": 0.4097,
      "step": 2316
    },
    {
      "epoch": 0.888420245398773,
      "grad_norm": 5.5403289794921875,
      "learning_rate": 6.3698338304218145e-06,
      "loss": 0.0322,
      "step": 2317
    },
    {
      "epoch": 0.8888036809815951,
      "grad_norm": 92.69943237304688,
      "learning_rate": 6.3485300383468255e-06,
      "loss": 0.5801,
      "step": 2318
    },
    {
      "epoch": 0.8891871165644172,
      "grad_norm": 103.48361206054688,
      "learning_rate": 6.3272262462718365e-06,
      "loss": 6.1143,
      "step": 2319
    },
    {
      "epoch": 0.8895705521472392,
      "grad_norm": 0.7315687537193298,
      "learning_rate": 6.305922454196847e-06,
      "loss": 0.002,
      "step": 2320
    },
    {
      "epoch": 0.8899539877300614,
      "grad_norm": 69.4375,
      "learning_rate": 6.284618662121858e-06,
      "loss": 0.6387,
      "step": 2321
    },
    {
      "epoch": 0.8903374233128835,
      "grad_norm": 1.154506802558899,
      "learning_rate": 6.263314870046869e-06,
      "loss": 0.004,
      "step": 2322
    },
    {
      "epoch": 0.8907208588957055,
      "grad_norm": 125.8946304321289,
      "learning_rate": 6.24201107797188e-06,
      "loss": 2.3558,
      "step": 2323
    },
    {
      "epoch": 0.8911042944785276,
      "grad_norm": 0.334873229265213,
      "learning_rate": 6.22070728589689e-06,
      "loss": 0.0023,
      "step": 2324
    },
    {
      "epoch": 0.8914877300613497,
      "grad_norm": 115.25357055664062,
      "learning_rate": 6.199403493821901e-06,
      "loss": 1.4082,
      "step": 2325
    },
    {
      "epoch": 0.8918711656441718,
      "grad_norm": 113.96891784667969,
      "learning_rate": 6.178099701746911e-06,
      "loss": 0.8809,
      "step": 2326
    },
    {
      "epoch": 0.8922546012269938,
      "grad_norm": 45.48662567138672,
      "learning_rate": 6.156795909671922e-06,
      "loss": 0.5483,
      "step": 2327
    },
    {
      "epoch": 0.8926380368098159,
      "grad_norm": 121.83149719238281,
      "learning_rate": 6.135492117596932e-06,
      "loss": 3.0171,
      "step": 2328
    },
    {
      "epoch": 0.8930214723926381,
      "grad_norm": 4.581501483917236,
      "learning_rate": 6.114188325521943e-06,
      "loss": 0.0171,
      "step": 2329
    },
    {
      "epoch": 0.8934049079754601,
      "grad_norm": 9.469815254211426,
      "learning_rate": 6.092884533446954e-06,
      "loss": 0.0296,
      "step": 2330
    },
    {
      "epoch": 0.8937883435582822,
      "grad_norm": 0.4207109212875366,
      "learning_rate": 6.071580741371965e-06,
      "loss": 0.0019,
      "step": 2331
    },
    {
      "epoch": 0.8941717791411042,
      "grad_norm": 6.05441427230835,
      "learning_rate": 6.050276949296975e-06,
      "loss": 0.0243,
      "step": 2332
    },
    {
      "epoch": 0.8945552147239264,
      "grad_norm": 15.64470100402832,
      "learning_rate": 6.0289731572219855e-06,
      "loss": 0.4124,
      "step": 2333
    },
    {
      "epoch": 0.8949386503067485,
      "grad_norm": 118.96612548828125,
      "learning_rate": 6.0076693651469965e-06,
      "loss": 0.9951,
      "step": 2334
    },
    {
      "epoch": 0.8953220858895705,
      "grad_norm": 395.6971740722656,
      "learning_rate": 5.9863655730720075e-06,
      "loss": 2.096,
      "step": 2335
    },
    {
      "epoch": 0.8957055214723927,
      "grad_norm": 352.2525939941406,
      "learning_rate": 5.965061780997018e-06,
      "loss": 5.1981,
      "step": 2336
    },
    {
      "epoch": 0.8960889570552147,
      "grad_norm": 0.6346721649169922,
      "learning_rate": 5.943757988922028e-06,
      "loss": 0.002,
      "step": 2337
    },
    {
      "epoch": 0.8964723926380368,
      "grad_norm": 1.3291672468185425,
      "learning_rate": 5.922454196847039e-06,
      "loss": 0.0031,
      "step": 2338
    },
    {
      "epoch": 0.8968558282208589,
      "grad_norm": 206.52493286132812,
      "learning_rate": 5.90115040477205e-06,
      "loss": 5.8926,
      "step": 2339
    },
    {
      "epoch": 0.897239263803681,
      "grad_norm": 389.3257141113281,
      "learning_rate": 5.879846612697061e-06,
      "loss": 5.8574,
      "step": 2340
    },
    {
      "epoch": 0.8976226993865031,
      "grad_norm": 1.5397624969482422,
      "learning_rate": 5.858542820622071e-06,
      "loss": 0.0041,
      "step": 2341
    },
    {
      "epoch": 0.8980061349693251,
      "grad_norm": 0.9912972450256348,
      "learning_rate": 5.837239028547082e-06,
      "loss": 0.0024,
      "step": 2342
    },
    {
      "epoch": 0.8983895705521472,
      "grad_norm": 0.4512677490711212,
      "learning_rate": 5.815935236472092e-06,
      "loss": 0.0024,
      "step": 2343
    },
    {
      "epoch": 0.8987730061349694,
      "grad_norm": 115.8443603515625,
      "learning_rate": 5.794631444397103e-06,
      "loss": 6.1021,
      "step": 2344
    },
    {
      "epoch": 0.8991564417177914,
      "grad_norm": 0.5459240674972534,
      "learning_rate": 5.773327652322113e-06,
      "loss": 0.0024,
      "step": 2345
    },
    {
      "epoch": 0.8995398773006135,
      "grad_norm": 98.77571105957031,
      "learning_rate": 5.752023860247124e-06,
      "loss": 0.94,
      "step": 2346
    },
    {
      "epoch": 0.8999233128834356,
      "grad_norm": 26.973379135131836,
      "learning_rate": 5.730720068172135e-06,
      "loss": 0.4268,
      "step": 2347
    },
    {
      "epoch": 0.9003067484662577,
      "grad_norm": 76.31385803222656,
      "learning_rate": 5.709416276097146e-06,
      "loss": 0.8687,
      "step": 2348
    },
    {
      "epoch": 0.9006901840490797,
      "grad_norm": 6.279634952545166,
      "learning_rate": 5.688112484022156e-06,
      "loss": 0.3914,
      "step": 2349
    },
    {
      "epoch": 0.9010736196319018,
      "grad_norm": 32.16206359863281,
      "learning_rate": 5.666808691947167e-06,
      "loss": 0.5913,
      "step": 2350
    },
    {
      "epoch": 0.901457055214724,
      "grad_norm": 84.63687896728516,
      "learning_rate": 5.645504899872178e-06,
      "loss": 0.9142,
      "step": 2351
    },
    {
      "epoch": 0.901840490797546,
      "grad_norm": 9.987971305847168,
      "learning_rate": 5.624201107797189e-06,
      "loss": 0.3919,
      "step": 2352
    },
    {
      "epoch": 0.9022239263803681,
      "grad_norm": 100.80982208251953,
      "learning_rate": 5.602897315722199e-06,
      "loss": 6.0352,
      "step": 2353
    },
    {
      "epoch": 0.9026073619631901,
      "grad_norm": 211.78219604492188,
      "learning_rate": 5.581593523647209e-06,
      "loss": 5.8379,
      "step": 2354
    },
    {
      "epoch": 0.9029907975460123,
      "grad_norm": 411.9369201660156,
      "learning_rate": 5.56028973157222e-06,
      "loss": 5.6138,
      "step": 2355
    },
    {
      "epoch": 0.9033742331288344,
      "grad_norm": 14.870763778686523,
      "learning_rate": 5.538985939497231e-06,
      "loss": 0.0817,
      "step": 2356
    },
    {
      "epoch": 0.9037576687116564,
      "grad_norm": 52.021644592285156,
      "learning_rate": 5.517682147422241e-06,
      "loss": 0.1079,
      "step": 2357
    },
    {
      "epoch": 0.9041411042944786,
      "grad_norm": 0.7164086699485779,
      "learning_rate": 5.496378355347252e-06,
      "loss": 0.0031,
      "step": 2358
    },
    {
      "epoch": 0.9045245398773006,
      "grad_norm": 5.322601318359375,
      "learning_rate": 5.475074563272263e-06,
      "loss": 0.0064,
      "step": 2359
    },
    {
      "epoch": 0.9049079754601227,
      "grad_norm": 5.817986488342285,
      "learning_rate": 5.453770771197273e-06,
      "loss": 0.3792,
      "step": 2360
    },
    {
      "epoch": 0.9052914110429447,
      "grad_norm": 1.5054324865341187,
      "learning_rate": 5.432466979122284e-06,
      "loss": 0.0029,
      "step": 2361
    },
    {
      "epoch": 0.9056748466257669,
      "grad_norm": 0.9984419345855713,
      "learning_rate": 5.4111631870472944e-06,
      "loss": 0.0023,
      "step": 2362
    },
    {
      "epoch": 0.906058282208589,
      "grad_norm": 17.932889938354492,
      "learning_rate": 5.3898593949723054e-06,
      "loss": 0.1219,
      "step": 2363
    },
    {
      "epoch": 0.906441717791411,
      "grad_norm": 0.13186410069465637,
      "learning_rate": 5.3685556028973165e-06,
      "loss": 0.0011,
      "step": 2364
    },
    {
      "epoch": 0.9068251533742331,
      "grad_norm": 15.44273853302002,
      "learning_rate": 5.347251810822327e-06,
      "loss": 0.2459,
      "step": 2365
    },
    {
      "epoch": 0.9072085889570553,
      "grad_norm": 152.20364379882812,
      "learning_rate": 5.325948018747337e-06,
      "loss": 1.9581,
      "step": 2366
    },
    {
      "epoch": 0.9075920245398773,
      "grad_norm": 282.3381042480469,
      "learning_rate": 5.304644226672348e-06,
      "loss": 1.8594,
      "step": 2367
    },
    {
      "epoch": 0.9079754601226994,
      "grad_norm": 9.226381301879883,
      "learning_rate": 5.283340434597359e-06,
      "loss": 0.0276,
      "step": 2368
    },
    {
      "epoch": 0.9083588957055214,
      "grad_norm": 38.01323318481445,
      "learning_rate": 5.26203664252237e-06,
      "loss": 0.5596,
      "step": 2369
    },
    {
      "epoch": 0.9087423312883436,
      "grad_norm": 11.03457260131836,
      "learning_rate": 5.24073285044738e-06,
      "loss": 0.0641,
      "step": 2370
    },
    {
      "epoch": 0.9091257668711656,
      "grad_norm": 4.320954322814941,
      "learning_rate": 5.21942905837239e-06,
      "loss": 0.0196,
      "step": 2371
    },
    {
      "epoch": 0.9095092024539877,
      "grad_norm": 3.1540722846984863,
      "learning_rate": 5.198125266297401e-06,
      "loss": 0.0105,
      "step": 2372
    },
    {
      "epoch": 0.9098926380368099,
      "grad_norm": 13.682193756103516,
      "learning_rate": 5.176821474222412e-06,
      "loss": 0.0806,
      "step": 2373
    },
    {
      "epoch": 0.9102760736196319,
      "grad_norm": 50.47124099731445,
      "learning_rate": 5.155517682147422e-06,
      "loss": 0.9649,
      "step": 2374
    },
    {
      "epoch": 0.910659509202454,
      "grad_norm": 7.538661003112793,
      "learning_rate": 5.134213890072433e-06,
      "loss": 0.0341,
      "step": 2375
    },
    {
      "epoch": 0.911042944785276,
      "grad_norm": 17.855735778808594,
      "learning_rate": 5.112910097997444e-06,
      "loss": 0.4041,
      "step": 2376
    },
    {
      "epoch": 0.9114263803680982,
      "grad_norm": 145.78993225097656,
      "learning_rate": 5.0916063059224544e-06,
      "loss": 0.8418,
      "step": 2377
    },
    {
      "epoch": 0.9118098159509203,
      "grad_norm": 37.60247039794922,
      "learning_rate": 5.070302513847465e-06,
      "loss": 0.6544,
      "step": 2378
    },
    {
      "epoch": 0.9121932515337423,
      "grad_norm": 109.55523681640625,
      "learning_rate": 5.048998721772476e-06,
      "loss": 2.5608,
      "step": 2379
    },
    {
      "epoch": 0.9125766871165644,
      "grad_norm": 176.50656127929688,
      "learning_rate": 5.027694929697487e-06,
      "loss": 6.3423,
      "step": 2380
    },
    {
      "epoch": 0.9129601226993865,
      "grad_norm": 0.9892469048500061,
      "learning_rate": 5.006391137622498e-06,
      "loss": 0.0066,
      "step": 2381
    },
    {
      "epoch": 0.9133435582822086,
      "grad_norm": 21.42610740661621,
      "learning_rate": 4.985087345547508e-06,
      "loss": 0.47,
      "step": 2382
    },
    {
      "epoch": 0.9137269938650306,
      "grad_norm": 16.929384231567383,
      "learning_rate": 4.963783553472518e-06,
      "loss": 0.5884,
      "step": 2383
    },
    {
      "epoch": 0.9141104294478528,
      "grad_norm": 33.386165618896484,
      "learning_rate": 4.942479761397529e-06,
      "loss": 0.6704,
      "step": 2384
    },
    {
      "epoch": 0.9144938650306749,
      "grad_norm": 0.8747240304946899,
      "learning_rate": 4.92117596932254e-06,
      "loss": 0.0036,
      "step": 2385
    },
    {
      "epoch": 0.9148773006134969,
      "grad_norm": 173.33103942871094,
      "learning_rate": 4.89987217724755e-06,
      "loss": 6.1895,
      "step": 2386
    },
    {
      "epoch": 0.915260736196319,
      "grad_norm": 142.9227294921875,
      "learning_rate": 4.878568385172561e-06,
      "loss": 1.5088,
      "step": 2387
    },
    {
      "epoch": 0.9156441717791411,
      "grad_norm": 0.28370898962020874,
      "learning_rate": 4.857264593097571e-06,
      "loss": 0.0019,
      "step": 2388
    },
    {
      "epoch": 0.9160276073619632,
      "grad_norm": 5.686431407928467,
      "learning_rate": 4.835960801022582e-06,
      "loss": 0.0222,
      "step": 2389
    },
    {
      "epoch": 0.9164110429447853,
      "grad_norm": 37.061012268066406,
      "learning_rate": 4.814657008947593e-06,
      "loss": 0.5405,
      "step": 2390
    },
    {
      "epoch": 0.9167944785276073,
      "grad_norm": 0.753302276134491,
      "learning_rate": 4.7933532168726034e-06,
      "loss": 0.0015,
      "step": 2391
    },
    {
      "epoch": 0.9171779141104295,
      "grad_norm": 11.971282005310059,
      "learning_rate": 4.7720494247976144e-06,
      "loss": 0.0361,
      "step": 2392
    },
    {
      "epoch": 0.9175613496932515,
      "grad_norm": 4.824380874633789,
      "learning_rate": 4.7507456327226254e-06,
      "loss": 0.0121,
      "step": 2393
    },
    {
      "epoch": 0.9179447852760736,
      "grad_norm": 0.9249712228775024,
      "learning_rate": 4.729441840647636e-06,
      "loss": 0.0036,
      "step": 2394
    },
    {
      "epoch": 0.9183282208588958,
      "grad_norm": 62.19675827026367,
      "learning_rate": 4.708138048572646e-06,
      "loss": 0.7403,
      "step": 2395
    },
    {
      "epoch": 0.9187116564417178,
      "grad_norm": 1.465177059173584,
      "learning_rate": 4.686834256497657e-06,
      "loss": 0.0041,
      "step": 2396
    },
    {
      "epoch": 0.9190950920245399,
      "grad_norm": 214.34622192382812,
      "learning_rate": 4.665530464422668e-06,
      "loss": 1.5811,
      "step": 2397
    },
    {
      "epoch": 0.9194785276073619,
      "grad_norm": 84.20104217529297,
      "learning_rate": 4.644226672347679e-06,
      "loss": 0.9165,
      "step": 2398
    },
    {
      "epoch": 0.9198619631901841,
      "grad_norm": 13.95626163482666,
      "learning_rate": 4.622922880272688e-06,
      "loss": 0.4883,
      "step": 2399
    },
    {
      "epoch": 0.9202453987730062,
      "grad_norm": 102.75283813476562,
      "learning_rate": 4.601619088197699e-06,
      "loss": 1.0986,
      "step": 2400
    },
    {
      "epoch": 0.9206288343558282,
      "grad_norm": 237.07943725585938,
      "learning_rate": 4.58031529612271e-06,
      "loss": 3.1571,
      "step": 2401
    },
    {
      "epoch": 0.9210122699386503,
      "grad_norm": 250.6201934814453,
      "learning_rate": 4.559011504047721e-06,
      "loss": 6.2051,
      "step": 2402
    },
    {
      "epoch": 0.9213957055214724,
      "grad_norm": 0.19728828966617584,
      "learning_rate": 4.537707711972731e-06,
      "loss": 0.0014,
      "step": 2403
    },
    {
      "epoch": 0.9217791411042945,
      "grad_norm": 15.691020965576172,
      "learning_rate": 4.516403919897742e-06,
      "loss": 0.04,
      "step": 2404
    },
    {
      "epoch": 0.9221625766871165,
      "grad_norm": 78.75389862060547,
      "learning_rate": 4.4951001278227524e-06,
      "loss": 1.0703,
      "step": 2405
    },
    {
      "epoch": 0.9225460122699386,
      "grad_norm": 128.8955078125,
      "learning_rate": 4.4737963357477634e-06,
      "loss": 1.9014,
      "step": 2406
    },
    {
      "epoch": 0.9229294478527608,
      "grad_norm": 120.89571380615234,
      "learning_rate": 4.452492543672774e-06,
      "loss": 1.8448,
      "step": 2407
    },
    {
      "epoch": 0.9233128834355828,
      "grad_norm": 3.4830448627471924,
      "learning_rate": 4.431188751597785e-06,
      "loss": 0.0076,
      "step": 2408
    },
    {
      "epoch": 0.9236963190184049,
      "grad_norm": 3.8731465339660645,
      "learning_rate": 4.409884959522796e-06,
      "loss": 0.3518,
      "step": 2409
    },
    {
      "epoch": 0.924079754601227,
      "grad_norm": 190.44200134277344,
      "learning_rate": 4.388581167447807e-06,
      "loss": 6.1836,
      "step": 2410
    },
    {
      "epoch": 0.9244631901840491,
      "grad_norm": 8.610749244689941,
      "learning_rate": 4.367277375372816e-06,
      "loss": 0.2986,
      "step": 2411
    },
    {
      "epoch": 0.9248466257668712,
      "grad_norm": 0.6672891974449158,
      "learning_rate": 4.345973583297827e-06,
      "loss": 0.0024,
      "step": 2412
    },
    {
      "epoch": 0.9252300613496932,
      "grad_norm": 2.1213881969451904,
      "learning_rate": 4.324669791222838e-06,
      "loss": 0.0089,
      "step": 2413
    },
    {
      "epoch": 0.9256134969325154,
      "grad_norm": 97.69578552246094,
      "learning_rate": 4.303365999147849e-06,
      "loss": 1.3272,
      "step": 2414
    },
    {
      "epoch": 0.9259969325153374,
      "grad_norm": 2.741866111755371,
      "learning_rate": 4.282062207072859e-06,
      "loss": 0.0073,
      "step": 2415
    },
    {
      "epoch": 0.9263803680981595,
      "grad_norm": 165.419921875,
      "learning_rate": 4.260758414997869e-06,
      "loss": 1.989,
      "step": 2416
    },
    {
      "epoch": 0.9267638036809815,
      "grad_norm": 1.4724115133285522,
      "learning_rate": 4.23945462292288e-06,
      "loss": 0.0049,
      "step": 2417
    },
    {
      "epoch": 0.9271472392638037,
      "grad_norm": 141.52371215820312,
      "learning_rate": 4.218150830847891e-06,
      "loss": 6.0625,
      "step": 2418
    },
    {
      "epoch": 0.9275306748466258,
      "grad_norm": 8.115018844604492,
      "learning_rate": 4.196847038772901e-06,
      "loss": 0.0194,
      "step": 2419
    },
    {
      "epoch": 0.9279141104294478,
      "grad_norm": 30.14399528503418,
      "learning_rate": 4.175543246697912e-06,
      "loss": 0.4922,
      "step": 2420
    },
    {
      "epoch": 0.92829754601227,
      "grad_norm": 117.81703186035156,
      "learning_rate": 4.1542394546229234e-06,
      "loss": 6.6914,
      "step": 2421
    },
    {
      "epoch": 0.928680981595092,
      "grad_norm": 245.2737579345703,
      "learning_rate": 4.132935662547934e-06,
      "loss": 1.753,
      "step": 2422
    },
    {
      "epoch": 0.9290644171779141,
      "grad_norm": 1.8317362070083618,
      "learning_rate": 4.111631870472945e-06,
      "loss": 0.0031,
      "step": 2423
    },
    {
      "epoch": 0.9294478527607362,
      "grad_norm": 16.254554748535156,
      "learning_rate": 4.090328078397955e-06,
      "loss": 0.0554,
      "step": 2424
    },
    {
      "epoch": 0.9298312883435583,
      "grad_norm": 3.539659023284912,
      "learning_rate": 4.069024286322966e-06,
      "loss": 0.0079,
      "step": 2425
    },
    {
      "epoch": 0.9302147239263804,
      "grad_norm": 25.596466064453125,
      "learning_rate": 4.047720494247977e-06,
      "loss": 0.0656,
      "step": 2426
    },
    {
      "epoch": 0.9305981595092024,
      "grad_norm": 10.608080863952637,
      "learning_rate": 4.026416702172987e-06,
      "loss": 0.0469,
      "step": 2427
    },
    {
      "epoch": 0.9309815950920245,
      "grad_norm": 1.3304612636566162,
      "learning_rate": 4.005112910097997e-06,
      "loss": 0.004,
      "step": 2428
    },
    {
      "epoch": 0.9313650306748467,
      "grad_norm": 100.94953155517578,
      "learning_rate": 3.983809118023008e-06,
      "loss": 6.1836,
      "step": 2429
    },
    {
      "epoch": 0.9317484662576687,
      "grad_norm": 9.298829078674316,
      "learning_rate": 3.962505325948019e-06,
      "loss": 0.0185,
      "step": 2430
    },
    {
      "epoch": 0.9321319018404908,
      "grad_norm": 1.1112534999847412,
      "learning_rate": 3.94120153387303e-06,
      "loss": 0.0054,
      "step": 2431
    },
    {
      "epoch": 0.9325153374233128,
      "grad_norm": 2.8512754440307617,
      "learning_rate": 3.91989774179804e-06,
      "loss": 0.0169,
      "step": 2432
    },
    {
      "epoch": 0.932898773006135,
      "grad_norm": 174.24310302734375,
      "learning_rate": 3.89859394972305e-06,
      "loss": 2.583,
      "step": 2433
    },
    {
      "epoch": 0.933282208588957,
      "grad_norm": 146.72923278808594,
      "learning_rate": 3.877290157648061e-06,
      "loss": 1.752,
      "step": 2434
    },
    {
      "epoch": 0.9336656441717791,
      "grad_norm": 30.88206672668457,
      "learning_rate": 3.855986365573072e-06,
      "loss": 0.1167,
      "step": 2435
    },
    {
      "epoch": 0.9340490797546013,
      "grad_norm": 132.73995971679688,
      "learning_rate": 3.834682573498083e-06,
      "loss": 5.4766,
      "step": 2436
    },
    {
      "epoch": 0.9344325153374233,
      "grad_norm": 3.126117706298828,
      "learning_rate": 3.8133787814230936e-06,
      "loss": 0.0162,
      "step": 2437
    },
    {
      "epoch": 0.9348159509202454,
      "grad_norm": 20.657733917236328,
      "learning_rate": 3.7920749893481046e-06,
      "loss": 0.4814,
      "step": 2438
    },
    {
      "epoch": 0.9351993865030674,
      "grad_norm": 202.00848388671875,
      "learning_rate": 3.7707711972731147e-06,
      "loss": 2.4808,
      "step": 2439
    },
    {
      "epoch": 0.9355828220858896,
      "grad_norm": 2.9972264766693115,
      "learning_rate": 3.7494674051981253e-06,
      "loss": 0.0074,
      "step": 2440
    },
    {
      "epoch": 0.9359662576687117,
      "grad_norm": 0.44538336992263794,
      "learning_rate": 3.728163613123136e-06,
      "loss": 0.0021,
      "step": 2441
    },
    {
      "epoch": 0.9363496932515337,
      "grad_norm": 6.419764518737793,
      "learning_rate": 3.706859821048147e-06,
      "loss": 0.0383,
      "step": 2442
    },
    {
      "epoch": 0.9367331288343558,
      "grad_norm": 205.5599365234375,
      "learning_rate": 3.6855560289731575e-06,
      "loss": 2.389,
      "step": 2443
    },
    {
      "epoch": 0.9371165644171779,
      "grad_norm": 1.176452398300171,
      "learning_rate": 3.6642522368981677e-06,
      "loss": 0.0046,
      "step": 2444
    },
    {
      "epoch": 0.9375,
      "grad_norm": 144.3276824951172,
      "learning_rate": 3.6429484448231787e-06,
      "loss": 1.5157,
      "step": 2445
    },
    {
      "epoch": 0.9378834355828221,
      "grad_norm": 7.984076499938965,
      "learning_rate": 3.6216446527481892e-06,
      "loss": 0.3045,
      "step": 2446
    },
    {
      "epoch": 0.9382668711656442,
      "grad_norm": 26.80510902404785,
      "learning_rate": 3.6003408606732e-06,
      "loss": 0.502,
      "step": 2447
    },
    {
      "epoch": 0.9386503067484663,
      "grad_norm": 183.5045623779297,
      "learning_rate": 3.579037068598211e-06,
      "loss": 4.8023,
      "step": 2448
    },
    {
      "epoch": 0.9390337423312883,
      "grad_norm": 2.254589080810547,
      "learning_rate": 3.5577332765232214e-06,
      "loss": 0.0066,
      "step": 2449
    },
    {
      "epoch": 0.9394171779141104,
      "grad_norm": 2.363844156265259,
      "learning_rate": 3.5364294844482316e-06,
      "loss": 0.0128,
      "step": 2450
    },
    {
      "epoch": 0.9398006134969326,
      "grad_norm": 0.25697576999664307,
      "learning_rate": 3.5151256923732426e-06,
      "loss": 0.0016,
      "step": 2451
    },
    {
      "epoch": 0.9401840490797546,
      "grad_norm": 145.4252471923828,
      "learning_rate": 3.493821900298253e-06,
      "loss": 6.5972,
      "step": 2452
    },
    {
      "epoch": 0.9405674846625767,
      "grad_norm": 25.66444969177246,
      "learning_rate": 3.472518108223264e-06,
      "loss": 0.4866,
      "step": 2453
    },
    {
      "epoch": 0.9409509202453987,
      "grad_norm": 14.369718551635742,
      "learning_rate": 3.4512143161482747e-06,
      "loss": 0.0793,
      "step": 2454
    },
    {
      "epoch": 0.9413343558282209,
      "grad_norm": 4.101546287536621,
      "learning_rate": 3.4299105240732853e-06,
      "loss": 0.0145,
      "step": 2455
    },
    {
      "epoch": 0.941717791411043,
      "grad_norm": 15.712298393249512,
      "learning_rate": 3.4086067319982955e-06,
      "loss": 0.4119,
      "step": 2456
    },
    {
      "epoch": 0.942101226993865,
      "grad_norm": 3.878542423248291,
      "learning_rate": 3.3873029399233065e-06,
      "loss": 0.0136,
      "step": 2457
    },
    {
      "epoch": 0.9424846625766872,
      "grad_norm": 9.528055191040039,
      "learning_rate": 3.365999147848317e-06,
      "loss": 0.3865,
      "step": 2458
    },
    {
      "epoch": 0.9428680981595092,
      "grad_norm": 0.7963228225708008,
      "learning_rate": 3.344695355773328e-06,
      "loss": 0.0031,
      "step": 2459
    },
    {
      "epoch": 0.9432515337423313,
      "grad_norm": 1.9365051984786987,
      "learning_rate": 3.3233915636983387e-06,
      "loss": 0.0049,
      "step": 2460
    },
    {
      "epoch": 0.9436349693251533,
      "grad_norm": 13.211302757263184,
      "learning_rate": 3.302087771623349e-06,
      "loss": 0.4617,
      "step": 2461
    },
    {
      "epoch": 0.9440184049079755,
      "grad_norm": 2.683659315109253,
      "learning_rate": 3.2807839795483594e-06,
      "loss": 0.0094,
      "step": 2462
    },
    {
      "epoch": 0.9444018404907976,
      "grad_norm": 9.588644981384277,
      "learning_rate": 3.2594801874733704e-06,
      "loss": 0.405,
      "step": 2463
    },
    {
      "epoch": 0.9447852760736196,
      "grad_norm": 2.5389938354492188,
      "learning_rate": 3.238176395398381e-06,
      "loss": 0.01,
      "step": 2464
    },
    {
      "epoch": 0.9451687116564417,
      "grad_norm": 66.77586364746094,
      "learning_rate": 3.216872603323392e-06,
      "loss": 0.6187,
      "step": 2465
    },
    {
      "epoch": 0.9455521472392638,
      "grad_norm": 7.38784122467041,
      "learning_rate": 3.1955688112484026e-06,
      "loss": 0.0334,
      "step": 2466
    },
    {
      "epoch": 0.9459355828220859,
      "grad_norm": 139.38377380371094,
      "learning_rate": 3.1742650191734127e-06,
      "loss": 2.5042,
      "step": 2467
    },
    {
      "epoch": 0.946319018404908,
      "grad_norm": 79.56858825683594,
      "learning_rate": 3.1529612270984233e-06,
      "loss": 0.8926,
      "step": 2468
    },
    {
      "epoch": 0.94670245398773,
      "grad_norm": 0.7099560499191284,
      "learning_rate": 3.1316574350234343e-06,
      "loss": 0.0025,
      "step": 2469
    },
    {
      "epoch": 0.9470858895705522,
      "grad_norm": 39.421669006347656,
      "learning_rate": 3.110353642948445e-06,
      "loss": 0.6362,
      "step": 2470
    },
    {
      "epoch": 0.9474693251533742,
      "grad_norm": 20.225257873535156,
      "learning_rate": 3.0890498508734555e-06,
      "loss": 0.1151,
      "step": 2471
    },
    {
      "epoch": 0.9478527607361963,
      "grad_norm": 134.07919311523438,
      "learning_rate": 3.067746058798466e-06,
      "loss": 4.6895,
      "step": 2472
    },
    {
      "epoch": 0.9482361963190185,
      "grad_norm": 1.9007248878479004,
      "learning_rate": 3.046442266723477e-06,
      "loss": 0.0071,
      "step": 2473
    },
    {
      "epoch": 0.9486196319018405,
      "grad_norm": 139.70526123046875,
      "learning_rate": 3.0251384746484876e-06,
      "loss": 1.2463,
      "step": 2474
    },
    {
      "epoch": 0.9490030674846626,
      "grad_norm": 2.7501935958862305,
      "learning_rate": 3.0038346825734982e-06,
      "loss": 0.0124,
      "step": 2475
    },
    {
      "epoch": 0.9493865030674846,
      "grad_norm": 14.394262313842773,
      "learning_rate": 2.982530890498509e-06,
      "loss": 0.1665,
      "step": 2476
    },
    {
      "epoch": 0.9497699386503068,
      "grad_norm": 171.37869262695312,
      "learning_rate": 2.9612270984235194e-06,
      "loss": 5.6914,
      "step": 2477
    },
    {
      "epoch": 0.9501533742331288,
      "grad_norm": 187.74874877929688,
      "learning_rate": 2.9399233063485304e-06,
      "loss": 2.7542,
      "step": 2478
    },
    {
      "epoch": 0.9505368098159509,
      "grad_norm": 140.52798461914062,
      "learning_rate": 2.918619514273541e-06,
      "loss": 6.0078,
      "step": 2479
    },
    {
      "epoch": 0.950920245398773,
      "grad_norm": 32.75605773925781,
      "learning_rate": 2.8973157221985516e-06,
      "loss": 0.1206,
      "step": 2480
    },
    {
      "epoch": 0.9513036809815951,
      "grad_norm": 6.467891693115234,
      "learning_rate": 2.876011930123562e-06,
      "loss": 0.0302,
      "step": 2481
    },
    {
      "epoch": 0.9516871165644172,
      "grad_norm": 83.42328643798828,
      "learning_rate": 2.854708138048573e-06,
      "loss": 0.9303,
      "step": 2482
    },
    {
      "epoch": 0.9520705521472392,
      "grad_norm": 204.3867950439453,
      "learning_rate": 2.8334043459735833e-06,
      "loss": 5.5215,
      "step": 2483
    },
    {
      "epoch": 0.9524539877300614,
      "grad_norm": 4.0695905685424805,
      "learning_rate": 2.8121005538985943e-06,
      "loss": 0.0194,
      "step": 2484
    },
    {
      "epoch": 0.9528374233128835,
      "grad_norm": 132.18507385253906,
      "learning_rate": 2.7907967618236045e-06,
      "loss": 5.9141,
      "step": 2485
    },
    {
      "epoch": 0.9532208588957055,
      "grad_norm": 32.51949691772461,
      "learning_rate": 2.7694929697486155e-06,
      "loss": 0.1285,
      "step": 2486
    },
    {
      "epoch": 0.9536042944785276,
      "grad_norm": 0.47829169034957886,
      "learning_rate": 2.748189177673626e-06,
      "loss": 0.0027,
      "step": 2487
    },
    {
      "epoch": 0.9539877300613497,
      "grad_norm": 3.1473779678344727,
      "learning_rate": 2.7268853855986366e-06,
      "loss": 0.0097,
      "step": 2488
    },
    {
      "epoch": 0.9543711656441718,
      "grad_norm": 74.50848388671875,
      "learning_rate": 2.7055815935236472e-06,
      "loss": 1.0918,
      "step": 2489
    },
    {
      "epoch": 0.9547546012269938,
      "grad_norm": 102.51834106445312,
      "learning_rate": 2.6842778014486582e-06,
      "loss": 0.9526,
      "step": 2490
    },
    {
      "epoch": 0.9551380368098159,
      "grad_norm": 0.42178601026535034,
      "learning_rate": 2.6629740093736684e-06,
      "loss": 0.0024,
      "step": 2491
    },
    {
      "epoch": 0.9555214723926381,
      "grad_norm": 2.300245523452759,
      "learning_rate": 2.6416702172986794e-06,
      "loss": 0.0092,
      "step": 2492
    },
    {
      "epoch": 0.9559049079754601,
      "grad_norm": 24.434701919555664,
      "learning_rate": 2.62036642522369e-06,
      "loss": 0.0971,
      "step": 2493
    },
    {
      "epoch": 0.9562883435582822,
      "grad_norm": 0.3528355658054352,
      "learning_rate": 2.5990626331487006e-06,
      "loss": 0.0016,
      "step": 2494
    },
    {
      "epoch": 0.9566717791411042,
      "grad_norm": 15.437116622924805,
      "learning_rate": 2.577758841073711e-06,
      "loss": 0.1912,
      "step": 2495
    },
    {
      "epoch": 0.9570552147239264,
      "grad_norm": 107.94813537597656,
      "learning_rate": 2.556455048998722e-06,
      "loss": 1.1475,
      "step": 2496
    },
    {
      "epoch": 0.9574386503067485,
      "grad_norm": 1.5852389335632324,
      "learning_rate": 2.5351512569237323e-06,
      "loss": 0.0063,
      "step": 2497
    },
    {
      "epoch": 0.9578220858895705,
      "grad_norm": 41.83260726928711,
      "learning_rate": 2.5138474648487433e-06,
      "loss": 0.5923,
      "step": 2498
    },
    {
      "epoch": 0.9582055214723927,
      "grad_norm": 0.9283663630485535,
      "learning_rate": 2.492543672773754e-06,
      "loss": 0.0036,
      "step": 2499
    },
    {
      "epoch": 0.9585889570552147,
      "grad_norm": 50.718238830566406,
      "learning_rate": 2.4712398806987645e-06,
      "loss": 0.7407,
      "step": 2500
    },
    {
      "epoch": 0.9589723926380368,
      "grad_norm": 3.917775869369507,
      "learning_rate": 2.449936088623775e-06,
      "loss": 0.3538,
      "step": 2501
    },
    {
      "epoch": 0.9593558282208589,
      "grad_norm": 1.7006443738937378,
      "learning_rate": 2.4286322965487856e-06,
      "loss": 0.0052,
      "step": 2502
    },
    {
      "epoch": 0.959739263803681,
      "grad_norm": 181.96173095703125,
      "learning_rate": 2.4073285044737966e-06,
      "loss": 1.8067,
      "step": 2503
    },
    {
      "epoch": 0.9601226993865031,
      "grad_norm": 96.75506591796875,
      "learning_rate": 2.3860247123988072e-06,
      "loss": 0.6309,
      "step": 2504
    },
    {
      "epoch": 0.9605061349693251,
      "grad_norm": 0.08275123685598373,
      "learning_rate": 2.364720920323818e-06,
      "loss": 0.0006,
      "step": 2505
    },
    {
      "epoch": 0.9608895705521472,
      "grad_norm": 65.58185577392578,
      "learning_rate": 2.3434171282488284e-06,
      "loss": 1.1084,
      "step": 2506
    },
    {
      "epoch": 0.9612730061349694,
      "grad_norm": 19.927858352661133,
      "learning_rate": 2.3221133361738394e-06,
      "loss": 0.1521,
      "step": 2507
    },
    {
      "epoch": 0.9616564417177914,
      "grad_norm": 30.809091567993164,
      "learning_rate": 2.3008095440988495e-06,
      "loss": 0.5117,
      "step": 2508
    },
    {
      "epoch": 0.9620398773006135,
      "grad_norm": 117.1274642944336,
      "learning_rate": 2.2795057520238606e-06,
      "loss": 0.8424,
      "step": 2509
    },
    {
      "epoch": 0.9624233128834356,
      "grad_norm": 1.3715885877609253,
      "learning_rate": 2.258201959948871e-06,
      "loss": 0.0061,
      "step": 2510
    },
    {
      "epoch": 0.9628067484662577,
      "grad_norm": 0.5358007550239563,
      "learning_rate": 2.2368981678738817e-06,
      "loss": 0.0028,
      "step": 2511
    },
    {
      "epoch": 0.9631901840490797,
      "grad_norm": 7.49700403213501,
      "learning_rate": 2.2155943757988923e-06,
      "loss": 0.2952,
      "step": 2512
    },
    {
      "epoch": 0.9635736196319018,
      "grad_norm": 0.17554740607738495,
      "learning_rate": 2.1942905837239033e-06,
      "loss": 0.0012,
      "step": 2513
    },
    {
      "epoch": 0.963957055214724,
      "grad_norm": 86.48233795166016,
      "learning_rate": 2.1729867916489135e-06,
      "loss": 0.9214,
      "step": 2514
    },
    {
      "epoch": 0.964340490797546,
      "grad_norm": 1.360700011253357,
      "learning_rate": 2.1516829995739245e-06,
      "loss": 0.0034,
      "step": 2515
    },
    {
      "epoch": 0.9647239263803681,
      "grad_norm": 0.373241662979126,
      "learning_rate": 2.1303792074989346e-06,
      "loss": 0.0017,
      "step": 2516
    },
    {
      "epoch": 0.9651073619631901,
      "grad_norm": 0.5386194586753845,
      "learning_rate": 2.1090754154239456e-06,
      "loss": 0.0038,
      "step": 2517
    },
    {
      "epoch": 0.9654907975460123,
      "grad_norm": 0.5087804198265076,
      "learning_rate": 2.087771623348956e-06,
      "loss": 0.0014,
      "step": 2518
    },
    {
      "epoch": 0.9658742331288344,
      "grad_norm": 2.500549077987671,
      "learning_rate": 2.066467831273967e-06,
      "loss": 0.0079,
      "step": 2519
    },
    {
      "epoch": 0.9662576687116564,
      "grad_norm": 16.712366104125977,
      "learning_rate": 2.0451640391989774e-06,
      "loss": 0.4468,
      "step": 2520
    },
    {
      "epoch": 0.9666411042944786,
      "grad_norm": 29.929351806640625,
      "learning_rate": 2.0238602471239884e-06,
      "loss": 0.0795,
      "step": 2521
    },
    {
      "epoch": 0.9670245398773006,
      "grad_norm": 121.40982055664062,
      "learning_rate": 2.0025564550489985e-06,
      "loss": 1.1143,
      "step": 2522
    },
    {
      "epoch": 0.9674079754601227,
      "grad_norm": 1.408011555671692,
      "learning_rate": 1.9812526629740095e-06,
      "loss": 0.0031,
      "step": 2523
    },
    {
      "epoch": 0.9677914110429447,
      "grad_norm": 136.08197021484375,
      "learning_rate": 1.95994887089902e-06,
      "loss": 2.3304,
      "step": 2524
    },
    {
      "epoch": 0.9681748466257669,
      "grad_norm": 1.030416488647461,
      "learning_rate": 1.9386450788240307e-06,
      "loss": 0.0036,
      "step": 2525
    },
    {
      "epoch": 0.968558282208589,
      "grad_norm": 10.150887489318848,
      "learning_rate": 1.9173412867490413e-06,
      "loss": 0.4077,
      "step": 2526
    },
    {
      "epoch": 0.968941717791411,
      "grad_norm": 2.4118311405181885,
      "learning_rate": 1.8960374946740523e-06,
      "loss": 0.0071,
      "step": 2527
    },
    {
      "epoch": 0.9693251533742331,
      "grad_norm": 244.69618225097656,
      "learning_rate": 1.8747337025990627e-06,
      "loss": 2.379,
      "step": 2528
    },
    {
      "epoch": 0.9697085889570553,
      "grad_norm": 36.92804718017578,
      "learning_rate": 1.8534299105240735e-06,
      "loss": 0.6929,
      "step": 2529
    },
    {
      "epoch": 0.9700920245398773,
      "grad_norm": 36.10527420043945,
      "learning_rate": 1.8321261184490838e-06,
      "loss": 0.0977,
      "step": 2530
    },
    {
      "epoch": 0.9704754601226994,
      "grad_norm": 71.33319854736328,
      "learning_rate": 1.8108223263740946e-06,
      "loss": 0.8179,
      "step": 2531
    },
    {
      "epoch": 0.9708588957055214,
      "grad_norm": 1.6740525960922241,
      "learning_rate": 1.7895185342991054e-06,
      "loss": 0.0041,
      "step": 2532
    },
    {
      "epoch": 0.9712423312883436,
      "grad_norm": 5.817898273468018,
      "learning_rate": 1.7682147422241158e-06,
      "loss": 0.3701,
      "step": 2533
    },
    {
      "epoch": 0.9716257668711656,
      "grad_norm": 38.433372497558594,
      "learning_rate": 1.7469109501491266e-06,
      "loss": 0.5977,
      "step": 2534
    },
    {
      "epoch": 0.9720092024539877,
      "grad_norm": 8.080559730529785,
      "learning_rate": 1.7256071580741374e-06,
      "loss": 0.0315,
      "step": 2535
    },
    {
      "epoch": 0.9723926380368099,
      "grad_norm": 40.872737884521484,
      "learning_rate": 1.7043033659991477e-06,
      "loss": 0.7588,
      "step": 2536
    },
    {
      "epoch": 0.9727760736196319,
      "grad_norm": 123.65471649169922,
      "learning_rate": 1.6829995739241585e-06,
      "loss": 5.457,
      "step": 2537
    },
    {
      "epoch": 0.973159509202454,
      "grad_norm": 1.4213929176330566,
      "learning_rate": 1.6616957818491693e-06,
      "loss": 0.0041,
      "step": 2538
    },
    {
      "epoch": 0.973542944785276,
      "grad_norm": 17.070253372192383,
      "learning_rate": 1.6403919897741797e-06,
      "loss": 0.027,
      "step": 2539
    },
    {
      "epoch": 0.9739263803680982,
      "grad_norm": 124.78246307373047,
      "learning_rate": 1.6190881976991905e-06,
      "loss": 1.8664,
      "step": 2540
    },
    {
      "epoch": 0.9743098159509203,
      "grad_norm": 182.86981201171875,
      "learning_rate": 1.5977844056242013e-06,
      "loss": 6.3516,
      "step": 2541
    },
    {
      "epoch": 0.9746932515337423,
      "grad_norm": 1.20124089717865,
      "learning_rate": 1.5764806135492117e-06,
      "loss": 0.0052,
      "step": 2542
    },
    {
      "epoch": 0.9750766871165644,
      "grad_norm": 36.85075759887695,
      "learning_rate": 1.5551768214742224e-06,
      "loss": 0.4468,
      "step": 2543
    },
    {
      "epoch": 0.9754601226993865,
      "grad_norm": 5.178058624267578,
      "learning_rate": 1.533873029399233e-06,
      "loss": 0.0122,
      "step": 2544
    },
    {
      "epoch": 0.9758435582822086,
      "grad_norm": 0.5216243863105774,
      "learning_rate": 1.5125692373242438e-06,
      "loss": 0.0021,
      "step": 2545
    },
    {
      "epoch": 0.9762269938650306,
      "grad_norm": 1.5058411359786987,
      "learning_rate": 1.4912654452492544e-06,
      "loss": 0.0042,
      "step": 2546
    },
    {
      "epoch": 0.9766104294478528,
      "grad_norm": 1.4786838293075562,
      "learning_rate": 1.4699616531742652e-06,
      "loss": 0.0051,
      "step": 2547
    },
    {
      "epoch": 0.9769938650306749,
      "grad_norm": 0.48048126697540283,
      "learning_rate": 1.4486578610992758e-06,
      "loss": 0.0025,
      "step": 2548
    },
    {
      "epoch": 0.9773773006134969,
      "grad_norm": 0.6981642842292786,
      "learning_rate": 1.4273540690242866e-06,
      "loss": 0.0047,
      "step": 2549
    },
    {
      "epoch": 0.977760736196319,
      "grad_norm": 0.47611889243125916,
      "learning_rate": 1.4060502769492972e-06,
      "loss": 0.0037,
      "step": 2550
    },
    {
      "epoch": 0.9781441717791411,
      "grad_norm": 1.5145431756973267,
      "learning_rate": 1.3847464848743077e-06,
      "loss": 0.0091,
      "step": 2551
    },
    {
      "epoch": 0.9785276073619632,
      "grad_norm": 4.604262351989746,
      "learning_rate": 1.3634426927993183e-06,
      "loss": 0.0241,
      "step": 2552
    },
    {
      "epoch": 0.9789110429447853,
      "grad_norm": 221.05865478515625,
      "learning_rate": 1.3421389007243291e-06,
      "loss": 1.8107,
      "step": 2553
    },
    {
      "epoch": 0.9792944785276073,
      "grad_norm": 97.07608032226562,
      "learning_rate": 1.3208351086493397e-06,
      "loss": 0.9634,
      "step": 2554
    },
    {
      "epoch": 0.9796779141104295,
      "grad_norm": 16.28598976135254,
      "learning_rate": 1.2995313165743503e-06,
      "loss": 0.5479,
      "step": 2555
    },
    {
      "epoch": 0.9800613496932515,
      "grad_norm": 204.9497528076172,
      "learning_rate": 1.278227524499361e-06,
      "loss": 2.7604,
      "step": 2556
    },
    {
      "epoch": 0.9804447852760736,
      "grad_norm": 215.8769073486328,
      "learning_rate": 1.2569237324243717e-06,
      "loss": 6.1328,
      "step": 2557
    },
    {
      "epoch": 0.9808282208588958,
      "grad_norm": 22.42989730834961,
      "learning_rate": 1.2356199403493822e-06,
      "loss": 0.5039,
      "step": 2558
    },
    {
      "epoch": 0.9812116564417178,
      "grad_norm": 107.1098861694336,
      "learning_rate": 1.2143161482743928e-06,
      "loss": 2.715,
      "step": 2559
    },
    {
      "epoch": 0.9815950920245399,
      "grad_norm": 25.10755729675293,
      "learning_rate": 1.1930123561994036e-06,
      "loss": 0.1469,
      "step": 2560
    },
    {
      "epoch": 0.9819785276073619,
      "grad_norm": 12.362627983093262,
      "learning_rate": 1.1717085641244142e-06,
      "loss": 0.4625,
      "step": 2561
    },
    {
      "epoch": 0.9823619631901841,
      "grad_norm": 0.613105833530426,
      "learning_rate": 1.1504047720494248e-06,
      "loss": 0.0022,
      "step": 2562
    },
    {
      "epoch": 0.9827453987730062,
      "grad_norm": 155.92428588867188,
      "learning_rate": 1.1291009799744356e-06,
      "loss": 0.8775,
      "step": 2563
    },
    {
      "epoch": 0.9831288343558282,
      "grad_norm": 71.01866149902344,
      "learning_rate": 1.1077971878994461e-06,
      "loss": 0.8364,
      "step": 2564
    },
    {
      "epoch": 0.9835122699386503,
      "grad_norm": 5.93011474609375,
      "learning_rate": 1.0864933958244567e-06,
      "loss": 0.0209,
      "step": 2565
    },
    {
      "epoch": 0.9838957055214724,
      "grad_norm": 237.0297393798828,
      "learning_rate": 1.0651896037494673e-06,
      "loss": 2.2697,
      "step": 2566
    },
    {
      "epoch": 0.9842791411042945,
      "grad_norm": 1.2019493579864502,
      "learning_rate": 1.043885811674478e-06,
      "loss": 0.0051,
      "step": 2567
    },
    {
      "epoch": 0.9846625766871165,
      "grad_norm": 0.4159267544746399,
      "learning_rate": 1.0225820195994887e-06,
      "loss": 0.0015,
      "step": 2568
    },
    {
      "epoch": 0.9850460122699386,
      "grad_norm": 9.824392318725586,
      "learning_rate": 1.0012782275244993e-06,
      "loss": 0.397,
      "step": 2569
    },
    {
      "epoch": 0.9854294478527608,
      "grad_norm": 153.7998809814453,
      "learning_rate": 9.7997443544951e-07,
      "loss": 1.3145,
      "step": 2570
    },
    {
      "epoch": 0.9858128834355828,
      "grad_norm": 135.33055114746094,
      "learning_rate": 9.586706433745206e-07,
      "loss": 0.7295,
      "step": 2571
    },
    {
      "epoch": 0.9861963190184049,
      "grad_norm": 16.33224868774414,
      "learning_rate": 9.373668512995313e-07,
      "loss": 0.1193,
      "step": 2572
    },
    {
      "epoch": 0.986579754601227,
      "grad_norm": 20.235445022583008,
      "learning_rate": 9.160630592245419e-07,
      "loss": 0.4543,
      "step": 2573
    },
    {
      "epoch": 0.9869631901840491,
      "grad_norm": 6.205273151397705,
      "learning_rate": 8.947592671495527e-07,
      "loss": 0.0212,
      "step": 2574
    },
    {
      "epoch": 0.9873466257668712,
      "grad_norm": 56.99029541015625,
      "learning_rate": 8.734554750745633e-07,
      "loss": 0.8604,
      "step": 2575
    },
    {
      "epoch": 0.9877300613496932,
      "grad_norm": 482.4648742675781,
      "learning_rate": 8.521516829995739e-07,
      "loss": 2.7152,
      "step": 2576
    },
    {
      "epoch": 0.9881134969325154,
      "grad_norm": 102.6148910522461,
      "learning_rate": 8.308478909245847e-07,
      "loss": 2.2579,
      "step": 2577
    },
    {
      "epoch": 0.9884969325153374,
      "grad_norm": 100.00586700439453,
      "learning_rate": 8.095440988495952e-07,
      "loss": 1.3702,
      "step": 2578
    },
    {
      "epoch": 0.9888803680981595,
      "grad_norm": 0.8538267016410828,
      "learning_rate": 7.882403067746058e-07,
      "loss": 0.002,
      "step": 2579
    },
    {
      "epoch": 0.9892638036809815,
      "grad_norm": 14.852790832519531,
      "learning_rate": 7.669365146996165e-07,
      "loss": 0.4805,
      "step": 2580
    },
    {
      "epoch": 0.9896472392638037,
      "grad_norm": 0.8749407529830933,
      "learning_rate": 7.456327226246272e-07,
      "loss": 0.0018,
      "step": 2581
    },
    {
      "epoch": 0.9900306748466258,
      "grad_norm": 183.30038452148438,
      "learning_rate": 7.243289305496379e-07,
      "loss": 0.9742,
      "step": 2582
    },
    {
      "epoch": 0.9904141104294478,
      "grad_norm": 87.85253143310547,
      "learning_rate": 7.030251384746486e-07,
      "loss": 0.7261,
      "step": 2583
    },
    {
      "epoch": 0.99079754601227,
      "grad_norm": 53.84426498413086,
      "learning_rate": 6.817213463996592e-07,
      "loss": 0.9238,
      "step": 2584
    },
    {
      "epoch": 0.991180981595092,
      "grad_norm": 12.457634925842285,
      "learning_rate": 6.604175543246698e-07,
      "loss": 0.0502,
      "step": 2585
    },
    {
      "epoch": 0.9915644171779141,
      "grad_norm": 2.8092494010925293,
      "learning_rate": 6.391137622496805e-07,
      "loss": 0.0063,
      "step": 2586
    },
    {
      "epoch": 0.9919478527607362,
      "grad_norm": 24.783527374267578,
      "learning_rate": 6.178099701746911e-07,
      "loss": 0.0685,
      "step": 2587
    },
    {
      "epoch": 0.9923312883435583,
      "grad_norm": 13.74468994140625,
      "learning_rate": 5.965061780997018e-07,
      "loss": 0.1213,
      "step": 2588
    },
    {
      "epoch": 0.9927147239263804,
      "grad_norm": 1.3093582391738892,
      "learning_rate": 5.752023860247124e-07,
      "loss": 0.0043,
      "step": 2589
    },
    {
      "epoch": 0.9930981595092024,
      "grad_norm": 152.93663024902344,
      "learning_rate": 5.538985939497231e-07,
      "loss": 3.19,
      "step": 2590
    },
    {
      "epoch": 0.9934815950920245,
      "grad_norm": 31.667722702026367,
      "learning_rate": 5.325948018747337e-07,
      "loss": 0.4617,
      "step": 2591
    },
    {
      "epoch": 0.9938650306748467,
      "grad_norm": 46.76748275756836,
      "learning_rate": 5.112910097997443e-07,
      "loss": 0.5439,
      "step": 2592
    },
    {
      "epoch": 0.9942484662576687,
      "grad_norm": 2.7833364009857178,
      "learning_rate": 4.89987217724755e-07,
      "loss": 0.0067,
      "step": 2593
    },
    {
      "epoch": 0.9946319018404908,
      "grad_norm": 49.33405685424805,
      "learning_rate": 4.6868342564976567e-07,
      "loss": 0.0971,
      "step": 2594
    },
    {
      "epoch": 0.9950153374233128,
      "grad_norm": 137.8705596923828,
      "learning_rate": 4.4737963357477635e-07,
      "loss": 2.4572,
      "step": 2595
    },
    {
      "epoch": 0.995398773006135,
      "grad_norm": 160.5074005126953,
      "learning_rate": 4.2607584149978694e-07,
      "loss": 6.1523,
      "step": 2596
    },
    {
      "epoch": 0.995782208588957,
      "grad_norm": 16.433561325073242,
      "learning_rate": 4.047720494247976e-07,
      "loss": 0.0363,
      "step": 2597
    },
    {
      "epoch": 0.9961656441717791,
      "grad_norm": 3.9944512844085693,
      "learning_rate": 3.8346825734980826e-07,
      "loss": 0.0101,
      "step": 2598
    },
    {
      "epoch": 0.9965490797546013,
      "grad_norm": 0.3866381347179413,
      "learning_rate": 3.6216446527481895e-07,
      "loss": 0.0019,
      "step": 2599
    },
    {
      "epoch": 0.9969325153374233,
      "grad_norm": 0.9906994700431824,
      "learning_rate": 3.408606731998296e-07,
      "loss": 0.0035,
      "step": 2600
    },
    {
      "epoch": 0.9973159509202454,
      "grad_norm": 0.24632364511489868,
      "learning_rate": 3.1955688112484027e-07,
      "loss": 0.0017,
      "step": 2601
    },
    {
      "epoch": 0.9976993865030674,
      "grad_norm": 2.1847798824310303,
      "learning_rate": 2.982530890498509e-07,
      "loss": 0.0062,
      "step": 2602
    },
    {
      "epoch": 0.9980828220858896,
      "grad_norm": 22.774734497070312,
      "learning_rate": 2.7694929697486154e-07,
      "loss": 0.1135,
      "step": 2603
    },
    {
      "epoch": 0.9984662576687117,
      "grad_norm": 13.302017211914062,
      "learning_rate": 2.5564550489987217e-07,
      "loss": 0.0401,
      "step": 2604
    },
    {
      "epoch": 0.9988496932515337,
      "grad_norm": 3.474975109100342,
      "learning_rate": 2.3434171282488283e-07,
      "loss": 0.0167,
      "step": 2605
    },
    {
      "epoch": 0.9992331288343558,
      "grad_norm": 64.99967956542969,
      "learning_rate": 2.1303792074989347e-07,
      "loss": 0.8428,
      "step": 2606
    },
    {
      "epoch": 0.9996165644171779,
      "grad_norm": 15.43187427520752,
      "learning_rate": 1.9173412867490413e-07,
      "loss": 0.1441,
      "step": 2607
    },
    {
      "epoch": 1.0,
      "grad_norm": 18.857643127441406,
      "learning_rate": 1.704303365999148e-07,
      "loss": 0.0876,
      "step": 2608
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.5,
      "eval_auc": 0.8125,
      "eval_f1": 0.6666666666666666,
      "eval_loss": 1.8160488605499268,
      "eval_precision": 0.5,
      "eval_recall": 1.0,
      "eval_runtime": 0.3919,
      "eval_samples_per_second": 40.827,
      "eval_steps_per_second": 10.207,
      "step": 2608
    }
  ],
  "logging_steps": 1,
  "max_steps": 2608,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 500,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.01
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.1076386427961344e+17,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}