{
  "best_metric": 0.22204521298408508,
  "best_model_checkpoint": "./vit-base-randaug/checkpoint-10990",
  "epoch": 10.0,
  "eval_steps": 500,
  "global_step": 10990,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.09,
      "grad_norm": 21.143863677978516,
      "learning_rate": 2.999387175598269e-05,
      "loss": 2.5983,
      "step": 100
    },
    {
      "epoch": 0.18,
      "grad_norm": 33.44860076904297,
      "learning_rate": 2.9975492031314045e-05,
      "loss": 1.4492,
      "step": 200
    },
    {
      "epoch": 0.27,
      "grad_norm": 22.861770629882812,
      "learning_rate": 2.994487584405244e-05,
      "loss": 1.0708,
      "step": 300
    },
    {
      "epoch": 0.36,
      "grad_norm": 22.55377769470215,
      "learning_rate": 2.990204821066006e-05,
      "loss": 0.8748,
      "step": 400
    },
    {
      "epoch": 0.45,
      "grad_norm": 19.91851043701172,
      "learning_rate": 2.984704412556199e-05,
      "loss": 0.8057,
      "step": 500
    },
    {
      "epoch": 0.55,
      "grad_norm": 16.450101852416992,
      "learning_rate": 2.977990853255228e-05,
      "loss": 0.6794,
      "step": 600
    },
    {
      "epoch": 0.64,
      "grad_norm": 19.716415405273438,
      "learning_rate": 2.970069628807043e-05,
      "loss": 0.7128,
      "step": 700
    },
    {
      "epoch": 0.73,
      "grad_norm": 27.150230407714844,
      "learning_rate": 2.9609472116378222e-05,
      "loss": 0.6514,
      "step": 800
    },
    {
      "epoch": 0.82,
      "grad_norm": 20.179523468017578,
      "learning_rate": 2.9506310556673573e-05,
      "loss": 0.5759,
      "step": 900
    },
    {
      "epoch": 0.91,
      "grad_norm": 23.87883758544922,
      "learning_rate": 2.9391295902184625e-05,
      "loss": 0.6247,
      "step": 1000
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.8982107355864811,
      "eval_loss": 0.3584350645542145,
      "eval_runtime": 107.0963,
      "eval_samples_per_second": 23.484,
      "eval_steps_per_second": 1.475,
      "step": 1099
    },
    {
      "epoch": 1.0,
      "grad_norm": 27.020755767822266,
      "learning_rate": 2.9264522131293818e-05,
      "loss": 0.6407,
      "step": 1100
    },
    {
      "epoch": 1.09,
      "grad_norm": 6.8938822746276855,
      "learning_rate": 2.9126092830748217e-05,
      "loss": 0.5433,
      "step": 1200
    },
    {
      "epoch": 1.18,
      "grad_norm": 16.615201950073242,
      "learning_rate": 2.897612111101888e-05,
      "loss": 0.4678,
      "step": 1300
    },
    {
      "epoch": 1.27,
      "grad_norm": 21.532333374023438,
      "learning_rate": 2.8814729513878365e-05,
      "loss": 0.4886,
      "step": 1400
    },
    {
      "epoch": 1.36,
      "grad_norm": 28.544614791870117,
      "learning_rate": 2.864204991227195e-05,
      "loss": 0.4525,
      "step": 1500
    },
    {
      "epoch": 1.46,
      "grad_norm": 22.7189884185791,
      "learning_rate": 2.8458223402564366e-05,
      "loss": 0.4353,
      "step": 1600
    },
    {
      "epoch": 1.55,
      "grad_norm": 41.8264045715332,
      "learning_rate": 2.826340018925006e-05,
      "loss": 0.4861,
      "step": 1700
    },
    {
      "epoch": 1.64,
      "grad_norm": 31.894323348999023,
      "learning_rate": 2.8057739462221215e-05,
      "loss": 0.4865,
      "step": 1800
    },
    {
      "epoch": 1.73,
      "grad_norm": 24.164323806762695,
      "learning_rate": 2.7841409266693838e-05,
      "loss": 0.4857,
      "step": 1900
    },
    {
      "epoch": 1.82,
      "grad_norm": 27.033830642700195,
      "learning_rate": 2.761458636589813e-05,
      "loss": 0.4192,
      "step": 2000
    },
    {
      "epoch": 1.91,
      "grad_norm": 17.30659294128418,
      "learning_rate": 2.7377456096645395e-05,
      "loss": 0.4589,
      "step": 2100
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.9228628230616303,
      "eval_loss": 0.27798059582710266,
      "eval_runtime": 106.6611,
      "eval_samples_per_second": 23.579,
      "eval_steps_per_second": 1.481,
      "step": 2198
    },
    {
      "epoch": 2.0,
      "grad_norm": 16.04237174987793,
      "learning_rate": 2.7130212217889484e-05,
      "loss": 0.446,
      "step": 2200
    },
    {
      "epoch": 2.09,
      "grad_norm": 16.05491828918457,
      "learning_rate": 2.6873056752406504e-05,
      "loss": 0.3563,
      "step": 2300
    },
    {
      "epoch": 2.18,
      "grad_norm": 28.44228744506836,
      "learning_rate": 2.6606199821722166e-05,
      "loss": 0.3843,
      "step": 2400
    },
    {
      "epoch": 2.27,
      "grad_norm": 17.263530731201172,
      "learning_rate": 2.632985947442167e-05,
      "loss": 0.3778,
      "step": 2500
    },
    {
      "epoch": 2.37,
      "grad_norm": 17.353647232055664,
      "learning_rate": 2.6044261507982356e-05,
      "loss": 0.3887,
      "step": 2600
    },
    {
      "epoch": 2.46,
      "grad_norm": 17.53342056274414,
      "learning_rate": 2.5749639284274782e-05,
      "loss": 0.3823,
      "step": 2700
    },
    {
      "epoch": 2.55,
      "grad_norm": 13.253349304199219,
      "learning_rate": 2.5446233538882924e-05,
      "loss": 0.3608,
      "step": 2800
    },
    {
      "epoch": 2.64,
      "grad_norm": 26.626955032348633,
      "learning_rate": 2.513429218439932e-05,
      "loss": 0.3996,
      "step": 2900
    },
    {
      "epoch": 2.73,
      "grad_norm": 22.196542739868164,
      "learning_rate": 2.4814070107855878e-05,
      "loss": 0.3541,
      "step": 3000
    },
    {
      "epoch": 2.82,
      "grad_norm": 22.010648727416992,
      "learning_rate": 2.448582896245591e-05,
      "loss": 0.3804,
      "step": 3100
    },
    {
      "epoch": 2.91,
      "grad_norm": 30.56482696533203,
      "learning_rate": 2.4149836953777488e-05,
      "loss": 0.3647,
      "step": 3200
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.9264413518886679,
      "eval_loss": 0.2550301253795624,
      "eval_runtime": 106.8539,
      "eval_samples_per_second": 23.537,
      "eval_steps_per_second": 1.479,
      "step": 3297
    },
    {
      "epoch": 3.0,
      "grad_norm": 18.120542526245117,
      "learning_rate": 2.3806368620622876e-05,
      "loss": 0.364,
      "step": 3300
    },
    {
      "epoch": 3.09,
      "grad_norm": 17.039871215820312,
      "learning_rate": 2.345570461069312e-05,
      "loss": 0.303,
      "step": 3400
    },
    {
      "epoch": 3.18,
      "grad_norm": 9.184852600097656,
      "learning_rate": 2.3098131451271016e-05,
      "loss": 0.313,
      "step": 3500
    },
    {
      "epoch": 3.28,
      "grad_norm": 7.604340076446533,
      "learning_rate": 2.2733941315099883e-05,
      "loss": 0.3364,
      "step": 3600
    },
    {
      "epoch": 3.37,
      "grad_norm": 33.884151458740234,
      "learning_rate": 2.2363431781649483e-05,
      "loss": 0.2789,
      "step": 3700
    },
    {
      "epoch": 3.46,
      "grad_norm": 12.377265930175781,
      "learning_rate": 2.1986905593964048e-05,
      "loss": 0.3083,
      "step": 3800
    },
    {
      "epoch": 3.55,
      "grad_norm": 17.730192184448242,
      "learning_rate": 2.1604670411291174e-05,
      "loss": 0.321,
      "step": 3900
    },
    {
      "epoch": 3.64,
      "grad_norm": 34.82160186767578,
      "learning_rate": 2.121703855769373e-05,
      "loss": 0.3169,
      "step": 4000
    },
    {
      "epoch": 3.73,
      "grad_norm": 19.860687255859375,
      "learning_rate": 2.0824326766850072e-05,
      "loss": 0.3463,
      "step": 4100
    },
    {
      "epoch": 3.82,
      "grad_norm": 4.745611667633057,
      "learning_rate": 2.042685592325123e-05,
      "loss": 0.3086,
      "step": 4200
    },
    {
      "epoch": 3.91,
      "grad_norm": 11.519020080566406,
      "learning_rate": 2.0024950800006463e-05,
      "loss": 0.3042,
      "step": 4300
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.9399602385685885,
      "eval_loss": 0.2381041944026947,
      "eval_runtime": 106.6676,
      "eval_samples_per_second": 23.578,
      "eval_steps_per_second": 1.481,
      "step": 4396
    },
    {
      "epoch": 4.0,
      "grad_norm": 23.076053619384766,
      "learning_rate": 1.961893979347137e-05,
      "loss": 0.3073,
      "step": 4400
    },
    {
      "epoch": 4.09,
      "grad_norm": 14.372575759887695,
      "learning_rate": 1.9209154654915524e-05,
      "loss": 0.2757,
      "step": 4500
    },
    {
      "epoch": 4.19,
      "grad_norm": 20.99867820739746,
      "learning_rate": 1.879593021944875e-05,
      "loss": 0.2735,
      "step": 4600
    },
    {
      "epoch": 4.28,
      "grad_norm": 16.90118980407715,
      "learning_rate": 1.837960413242765e-05,
      "loss": 0.2656,
      "step": 4700
    },
    {
      "epoch": 4.37,
      "grad_norm": 8.394341468811035,
      "learning_rate": 1.796051657356582e-05,
      "loss": 0.251,
      "step": 4800
    },
    {
      "epoch": 4.46,
      "grad_norm": 25.299718856811523,
      "learning_rate": 1.7539009978973312e-05,
      "loss": 0.2582,
      "step": 4900
    },
    {
      "epoch": 4.55,
      "grad_norm": 21.776456832885742,
      "learning_rate": 1.711542876135233e-05,
      "loss": 0.2634,
      "step": 5000
    },
    {
      "epoch": 4.64,
      "grad_norm": 15.235289573669434,
      "learning_rate": 1.669011902857791e-05,
      "loss": 0.239,
      "step": 5100
    },
    {
      "epoch": 4.73,
      "grad_norm": 29.831466674804688,
      "learning_rate": 1.6263428300893422e-05,
      "loss": 0.286,
      "step": 5200
    },
    {
      "epoch": 4.82,
      "grad_norm": 22.001750946044922,
      "learning_rate": 1.5835705226952112e-05,
      "loss": 0.2838,
      "step": 5300
    },
    {
      "epoch": 4.91,
      "grad_norm": 0.7380269169807434,
      "learning_rate": 1.540729929893649e-05,
      "loss": 0.2912,
      "step": 5400
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.941948310139165,
      "eval_loss": 0.23470723628997803,
      "eval_runtime": 106.0011,
      "eval_samples_per_second": 23.726,
      "eval_steps_per_second": 1.491,
      "step": 5495
    },
    {
      "epoch": 5.0,
      "grad_norm": 4.909294128417969,
      "learning_rate": 1.4978560566988603e-05,
      "loss": 0.2726,
      "step": 5500
    },
    {
      "epoch": 5.1,
      "grad_norm": 17.943002700805664,
      "learning_rate": 1.454983935318433e-05,
      "loss": 0.2115,
      "step": 5600
    },
    {
      "epoch": 5.19,
      "grad_norm": 2.650315523147583,
      "learning_rate": 1.4121485965285485e-05,
      "loss": 0.2121,
      "step": 5700
    },
    {
      "epoch": 5.28,
      "grad_norm": 8.513612747192383,
      "learning_rate": 1.3693850410503614e-05,
      "loss": 0.21,
      "step": 5800
    },
    {
      "epoch": 5.37,
      "grad_norm": 0.9585162997245789,
      "learning_rate": 1.326728210950942e-05,
      "loss": 0.1968,
      "step": 5900
    },
    {
      "epoch": 5.46,
      "grad_norm": 20.247549057006836,
      "learning_rate": 1.2842129610921378e-05,
      "loss": 0.2298,
      "step": 6000
    },
    {
      "epoch": 5.55,
      "grad_norm": 13.879007339477539,
      "learning_rate": 1.2418740306506923e-05,
      "loss": 0.2327,
      "step": 6100
    },
    {
      "epoch": 5.64,
      "grad_norm": 10.441729545593262,
      "learning_rate": 1.1997460147328984e-05,
      "loss": 0.2337,
      "step": 6200
    },
    {
      "epoch": 5.73,
      "grad_norm": 22.691776275634766,
      "learning_rate": 1.1578633361069559e-05,
      "loss": 0.2394,
      "step": 6300
    },
    {
      "epoch": 5.82,
      "grad_norm": 15.436910629272461,
      "learning_rate": 1.1162602170761611e-05,
      "loss": 0.2117,
      "step": 6400
    },
    {
      "epoch": 5.91,
      "grad_norm": 15.4258451461792,
      "learning_rate": 1.0749706515158863e-05,
      "loss": 0.2464,
      "step": 6500
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.945924453280318,
      "eval_loss": 0.2269454002380371,
      "eval_runtime": 105.9532,
      "eval_samples_per_second": 23.737,
      "eval_steps_per_second": 1.491,
      "step": 6594
    },
    {
      "epoch": 6.01,
      "grad_norm": 0.4867183268070221,
      "learning_rate": 1.0340283770972167e-05,
      "loss": 0.2218,
      "step": 6600
    },
    {
      "epoch": 6.1,
      "grad_norm": 21.317001342773438,
      "learning_rate": 9.93466847719919e-06,
      "loss": 0.2358,
      "step": 6700
    },
    {
      "epoch": 6.19,
      "grad_norm": 12.322629928588867,
      "learning_rate": 9.533192061772919e-06,
      "loss": 0.2132,
      "step": 6800
    },
    {
      "epoch": 6.28,
      "grad_norm": 0.27425938844680786,
      "learning_rate": 9.136182570752153e-06,
      "loss": 0.1987,
      "step": 6900
    },
    {
      "epoch": 6.37,
      "grad_norm": 5.367835521697998,
      "learning_rate": 8.743964400275304e-06,
      "loss": 0.1834,
      "step": 7000
    },
    {
      "epoch": 6.46,
      "grad_norm": 13.12271499633789,
      "learning_rate": 8.356858031496596e-06,
      "loss": 0.251,
      "step": 7100
    },
    {
      "epoch": 6.55,
      "grad_norm": 2.0102455615997314,
      "learning_rate": 7.975179768721187e-06,
      "loss": 0.1595,
      "step": 7200
    },
    {
      "epoch": 6.64,
      "grad_norm": 1.001270055770874,
      "learning_rate": 7.599241480953112e-06,
      "loss": 0.179,
      "step": 7300
    },
    {
      "epoch": 6.73,
      "grad_norm": 16.737274169921875,
      "learning_rate": 7.229350347067426e-06,
      "loss": 0.1667,
      "step": 7400
    },
    {
      "epoch": 6.82,
      "grad_norm": 11.864615440368652,
      "learning_rate": 6.865808604814564e-06,
      "loss": 0.1653,
      "step": 7500
    },
    {
      "epoch": 6.92,
      "grad_norm": 14.621787071228027,
      "learning_rate": 6.508913303862144e-06,
      "loss": 0.2132,
      "step": 7600
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.94831013916501,
      "eval_loss": 0.22579605877399445,
      "eval_runtime": 106.7179,
      "eval_samples_per_second": 23.567,
      "eval_steps_per_second": 1.481,
      "step": 7693
    },
    {
      "epoch": 7.01,
      "grad_norm": 16.592910766601562,
      "learning_rate": 6.1589560630758656e-06,
      "loss": 0.1783,
      "step": 7700
    },
    {
      "epoch": 7.1,
      "grad_norm": 0.33882206678390503,
      "learning_rate": 5.8162228322380155e-06,
      "loss": 0.1706,
      "step": 7800
    },
    {
      "epoch": 7.19,
      "grad_norm": 7.021131992340088,
      "learning_rate": 5.480993658398129e-06,
      "loss": 0.1708,
      "step": 7900
    },
    {
      "epoch": 7.28,
      "grad_norm": 1.4672304391860962,
      "learning_rate": 5.153542457046737e-06,
      "loss": 0.1709,
      "step": 8000
    },
    {
      "epoch": 7.37,
      "grad_norm": 1.1598705053329468,
      "learning_rate": 4.834136788299248e-06,
      "loss": 0.1732,
      "step": 8100
    },
    {
      "epoch": 7.46,
      "grad_norm": 22.20403289794922,
      "learning_rate": 4.523037638272822e-06,
      "loss": 0.1553,
      "step": 8200
    },
    {
      "epoch": 7.55,
      "grad_norm": 33.87173843383789,
      "learning_rate": 4.220499205834783e-06,
      "loss": 0.1816,
      "step": 8300
    },
    {
      "epoch": 7.64,
      "grad_norm": 5.8969807624816895,
      "learning_rate": 3.926768694896931e-06,
      "loss": 0.1802,
      "step": 8400
    },
    {
      "epoch": 7.73,
      "grad_norm": 11.662792205810547,
      "learning_rate": 3.6420861124254607e-06,
      "loss": 0.1622,
      "step": 8500
    },
    {
      "epoch": 7.83,
      "grad_norm": 15.271336555480957,
      "learning_rate": 3.3666840723314145e-06,
      "loss": 0.1686,
      "step": 8600
    },
    {
      "epoch": 7.92,
      "grad_norm": 0.5663726925849915,
      "learning_rate": 3.1007876054020724e-06,
      "loss": 0.1956,
      "step": 8700
    },
    {
      "epoch": 8.0,
      "eval_accuracy": 0.9495029821073558,
      "eval_loss": 0.22222255170345306,
      "eval_runtime": 106.0882,
      "eval_samples_per_second": 23.707,
      "eval_steps_per_second": 1.489,
      "step": 8792
    },
    {
      "epoch": 8.01,
      "grad_norm": 16.171127319335938,
      "learning_rate": 2.8446139754284486e-06,
      "loss": 0.2088,
      "step": 8800
    },
    {
      "epoch": 8.1,
      "grad_norm": 15.017475128173828,
      "learning_rate": 2.5983725016792574e-06,
      "loss": 0.1814,
      "step": 8900
    },
    {
      "epoch": 8.19,
      "grad_norm": 0.27975594997406006,
      "learning_rate": 2.36226438786627e-06,
      "loss": 0.1603,
      "step": 9000
    },
    {
      "epoch": 8.28,
      "grad_norm": 0.25504347681999207,
      "learning_rate": 2.1364825577409424e-06,
      "loss": 0.1322,
      "step": 9100
    },
    {
      "epoch": 8.37,
      "grad_norm": 1.7503337860107422,
      "learning_rate": 1.9212114974565664e-06,
      "loss": 0.1667,
      "step": 9200
    },
    {
      "epoch": 8.46,
      "grad_norm": 15.323562622070312,
      "learning_rate": 1.7166271048247796e-06,
      "loss": 0.1337,
      "step": 9300
    },
    {
      "epoch": 8.55,
      "grad_norm": 5.39154577255249,
      "learning_rate": 1.5228965455896054e-06,
      "loss": 0.1332,
      "step": 9400
    },
    {
      "epoch": 8.64,
      "grad_norm": 9.606640815734863,
      "learning_rate": 1.3401781168364591e-06,
      "loss": 0.1854,
      "step": 9500
    },
    {
      "epoch": 8.74,
      "grad_norm": 12.583927154541016,
      "learning_rate": 1.1686211176477208e-06,
      "loss": 0.148,
      "step": 9600
    },
    {
      "epoch": 8.83,
      "grad_norm": 7.303041458129883,
      "learning_rate": 1.00836572711058e-06,
      "loss": 0.1898,
      "step": 9700
    },
    {
      "epoch": 8.92,
      "grad_norm": 6.023931980133057,
      "learning_rate": 8.595428897768071e-07,
      "loss": 0.1723,
      "step": 9800
    },
    {
      "epoch": 9.0,
      "eval_accuracy": 0.9499005964214712,
      "eval_loss": 0.22232353687286377,
      "eval_runtime": 106.0307,
      "eval_samples_per_second": 23.72,
      "eval_steps_per_second": 1.49,
      "step": 9891
    },
    {
      "epoch": 9.01,
      "grad_norm": 24.670425415039062,
      "learning_rate": 7.222742086680756e-07,
      "loss": 0.174,
      "step": 9900
    },
    {
      "epoch": 9.1,
      "grad_norm": 0.7421072125434875,
      "learning_rate": 5.966718459142196e-07,
      "loss": 0.1787,
      "step": 10000
    },
    {
      "epoch": 9.19,
      "grad_norm": 34.601505279541016,
      "learning_rate": 4.82838431105655e-07,
      "loss": 0.2027,
      "step": 10100
    },
    {
      "epoch": 9.28,
      "grad_norm": 27.91753578186035,
      "learning_rate": 3.808669774348167e-07,
      "loss": 0.1434,
      "step": 10200
    },
    {
      "epoch": 9.37,
      "grad_norm": 22.817646026611328,
      "learning_rate": 2.908408056951578e-07,
      "loss": 0.1668,
      "step": 10300
    },
    {
      "epoch": 9.46,
      "grad_norm": 0.9755169749259949,
      "learning_rate": 2.1283347619979243e-07,
      "loss": 0.1511,
      "step": 10400
    },
    {
      "epoch": 9.55,
      "grad_norm": 4.755888938903809,
      "learning_rate": 1.4690872867542892e-07,
      "loss": 0.1821,
      "step": 10500
    },
    {
      "epoch": 9.65,
      "grad_norm": 0.373325377702713,
      "learning_rate": 9.312043018067762e-08,
      "loss": 0.2016,
      "step": 10600
    },
    {
      "epoch": 9.74,
      "grad_norm": 12.15149974822998,
      "learning_rate": 5.1512531091333914e-08,
      "loss": 0.1595,
      "step": 10700
    },
    {
      "epoch": 9.83,
      "grad_norm": 23.01797103881836,
      "learning_rate": 2.211902918855313e-08,
      "loss": 0.15,
      "step": 10800
    },
    {
      "epoch": 9.92,
      "grad_norm": 14.00448226928711,
      "learning_rate": 4.963941879295164e-09,
      "loss": 0.1558,
      "step": 10900
    },
    {
      "epoch": 10.0,
      "eval_accuracy": 0.9506958250497017,
      "eval_loss": 0.22204521298408508,
      "eval_runtime": 106.2562,
      "eval_samples_per_second": 23.669,
      "eval_steps_per_second": 1.487,
      "step": 10990
    },
    {
      "epoch": 10.0,
      "step": 10990,
      "total_flos": 4.09349935387607e+19,
      "train_loss": 0.33212552968755,
      "train_runtime": 19215.4665,
      "train_samples_per_second": 9.149,
      "train_steps_per_second": 0.572
    }
  ],
  "logging_steps": 100,
  "max_steps": 10990,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 10,
  "save_steps": 500,
  "total_flos": 4.09349935387607e+19,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}