{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 8.788598574821853,
  "global_step": 18500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.1,
      "learning_rate": 1.980997624703088e-05,
      "loss": 0.4065,
      "step": 200
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.961995249406176e-05,
      "loss": 0.3019,
      "step": 400
    },
    {
      "epoch": 0.24,
      "eval_accuracy": 0.9059633027522935,
      "eval_loss": 0.26677730679512024,
      "eval_runtime": 5.719,
      "eval_samples_per_second": 152.475,
      "eval_steps_per_second": 4.896,
      "step": 500
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.9429928741092638e-05,
      "loss": 0.2873,
      "step": 600
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.9239904988123516e-05,
      "loss": 0.2756,
      "step": 800
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.9049881235154395e-05,
      "loss": 0.2514,
      "step": 1000
    },
    {
      "epoch": 0.48,
      "eval_accuracy": 0.9220183486238532,
      "eval_loss": 0.20395953953266144,
      "eval_runtime": 4.9214,
      "eval_samples_per_second": 177.184,
      "eval_steps_per_second": 5.689,
      "step": 1000
    },
    {
      "epoch": 0.57,
      "learning_rate": 1.8859857482185277e-05,
      "loss": 0.2292,
      "step": 1200
    },
    {
      "epoch": 0.67,
      "learning_rate": 1.8669833729216152e-05,
      "loss": 0.2223,
      "step": 1400
    },
    {
      "epoch": 0.71,
      "eval_accuracy": 0.9254587155963303,
      "eval_loss": 0.20957307517528534,
      "eval_runtime": 9.3475,
      "eval_samples_per_second": 93.287,
      "eval_steps_per_second": 2.995,
      "step": 1500
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.847980997624703e-05,
      "loss": 0.2206,
      "step": 1600
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.8289786223277913e-05,
      "loss": 0.2177,
      "step": 1800
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.809976247030879e-05,
      "loss": 0.2141,
      "step": 2000
    },
    {
      "epoch": 0.95,
      "eval_accuracy": 0.9197247706422018,
      "eval_loss": 0.21411468088626862,
      "eval_runtime": 5.5585,
      "eval_samples_per_second": 156.877,
      "eval_steps_per_second": 5.037,
      "step": 2000
    },
    {
      "epoch": 1.05,
      "learning_rate": 1.790973871733967e-05,
      "loss": 0.191,
      "step": 2200
    },
    {
      "epoch": 1.14,
      "learning_rate": 1.7719714964370545e-05,
      "loss": 0.1615,
      "step": 2400
    },
    {
      "epoch": 1.19,
      "eval_accuracy": 0.9277522935779816,
      "eval_loss": 0.23593279719352722,
      "eval_runtime": 6.3221,
      "eval_samples_per_second": 137.929,
      "eval_steps_per_second": 4.429,
      "step": 2500
    },
    {
      "epoch": 1.24,
      "learning_rate": 1.7529691211401427e-05,
      "loss": 0.1598,
      "step": 2600
    },
    {
      "epoch": 1.33,
      "learning_rate": 1.7339667458432306e-05,
      "loss": 0.1513,
      "step": 2800
    },
    {
      "epoch": 1.43,
      "learning_rate": 1.7149643705463184e-05,
      "loss": 0.1696,
      "step": 3000
    },
    {
      "epoch": 1.43,
      "eval_accuracy": 0.9277522935779816,
      "eval_loss": 0.22338584065437317,
      "eval_runtime": 4.5931,
      "eval_samples_per_second": 189.852,
      "eval_steps_per_second": 6.096,
      "step": 3000
    },
    {
      "epoch": 1.52,
      "learning_rate": 1.6959619952494063e-05,
      "loss": 0.1598,
      "step": 3200
    },
    {
      "epoch": 1.62,
      "learning_rate": 1.676959619952494e-05,
      "loss": 0.1704,
      "step": 3400
    },
    {
      "epoch": 1.66,
      "eval_accuracy": 0.926605504587156,
      "eval_loss": 0.20875951647758484,
      "eval_runtime": 5.0919,
      "eval_samples_per_second": 171.251,
      "eval_steps_per_second": 5.499,
      "step": 3500
    },
    {
      "epoch": 1.71,
      "learning_rate": 1.657957244655582e-05,
      "loss": 0.1584,
      "step": 3600
    },
    {
      "epoch": 1.81,
      "learning_rate": 1.63895486935867e-05,
      "loss": 0.171,
      "step": 3800
    },
    {
      "epoch": 1.9,
      "learning_rate": 1.6199524940617578e-05,
      "loss": 0.1443,
      "step": 4000
    },
    {
      "epoch": 1.9,
      "eval_accuracy": 0.9288990825688074,
      "eval_loss": 0.22995886206626892,
      "eval_runtime": 5.7731,
      "eval_samples_per_second": 151.046,
      "eval_steps_per_second": 4.85,
      "step": 4000
    },
    {
      "epoch": 2.0,
      "learning_rate": 1.6009501187648456e-05,
      "loss": 0.1498,
      "step": 4200
    },
    {
      "epoch": 2.09,
      "learning_rate": 1.5819477434679335e-05,
      "loss": 0.1212,
      "step": 4400
    },
    {
      "epoch": 2.14,
      "eval_accuracy": 0.9311926605504587,
      "eval_loss": 0.25349077582359314,
      "eval_runtime": 4.9282,
      "eval_samples_per_second": 176.941,
      "eval_steps_per_second": 5.682,
      "step": 4500
    },
    {
      "epoch": 2.19,
      "learning_rate": 1.5629453681710217e-05,
      "loss": 0.133,
      "step": 4600
    },
    {
      "epoch": 2.28,
      "learning_rate": 1.5439429928741095e-05,
      "loss": 0.1213,
      "step": 4800
    },
    {
      "epoch": 2.38,
      "learning_rate": 1.5249406175771972e-05,
      "loss": 0.1283,
      "step": 5000
    },
    {
      "epoch": 2.38,
      "eval_accuracy": 0.9277522935779816,
      "eval_loss": 0.2518787384033203,
      "eval_runtime": 6.251,
      "eval_samples_per_second": 139.497,
      "eval_steps_per_second": 4.479,
      "step": 5000
    },
    {
      "epoch": 2.47,
      "learning_rate": 1.5059382422802851e-05,
      "loss": 0.12,
      "step": 5200
    },
    {
      "epoch": 2.57,
      "learning_rate": 1.4869358669833731e-05,
      "loss": 0.1313,
      "step": 5400
    },
    {
      "epoch": 2.61,
      "eval_accuracy": 0.9094036697247706,
      "eval_loss": 0.3524125814437866,
      "eval_runtime": 5.7105,
      "eval_samples_per_second": 152.701,
      "eval_steps_per_second": 4.903,
      "step": 5500
    },
    {
      "epoch": 2.66,
      "learning_rate": 1.467933491686461e-05,
      "loss": 0.1196,
      "step": 5600
    },
    {
      "epoch": 2.76,
      "learning_rate": 1.4489311163895487e-05,
      "loss": 0.123,
      "step": 5800
    },
    {
      "epoch": 2.85,
      "learning_rate": 1.4299287410926367e-05,
      "loss": 0.1304,
      "step": 6000
    },
    {
      "epoch": 2.85,
      "eval_accuracy": 0.9243119266055045,
      "eval_loss": 0.24740684032440186,
      "eval_runtime": 11.1285,
      "eval_samples_per_second": 78.357,
      "eval_steps_per_second": 2.516,
      "step": 6000
    },
    {
      "epoch": 2.95,
      "learning_rate": 1.4109263657957246e-05,
      "loss": 0.1248,
      "step": 6200
    },
    {
      "epoch": 3.04,
      "learning_rate": 1.3919239904988124e-05,
      "loss": 0.1061,
      "step": 6400
    },
    {
      "epoch": 3.09,
      "eval_accuracy": 0.9231651376146789,
      "eval_loss": 0.2941964566707611,
      "eval_runtime": 8.2674,
      "eval_samples_per_second": 105.475,
      "eval_steps_per_second": 3.387,
      "step": 6500
    },
    {
      "epoch": 3.14,
      "learning_rate": 1.3729216152019003e-05,
      "loss": 0.0876,
      "step": 6600
    },
    {
      "epoch": 3.23,
      "learning_rate": 1.3539192399049883e-05,
      "loss": 0.0922,
      "step": 6800
    },
    {
      "epoch": 3.33,
      "learning_rate": 1.3349168646080762e-05,
      "loss": 0.0988,
      "step": 7000
    },
    {
      "epoch": 3.33,
      "eval_accuracy": 0.9311926605504587,
      "eval_loss": 0.2566107213497162,
      "eval_runtime": 6.2677,
      "eval_samples_per_second": 139.127,
      "eval_steps_per_second": 4.467,
      "step": 7000
    },
    {
      "epoch": 3.42,
      "learning_rate": 1.3159144893111639e-05,
      "loss": 0.1097,
      "step": 7200
    },
    {
      "epoch": 3.52,
      "learning_rate": 1.296912114014252e-05,
      "loss": 0.0938,
      "step": 7400
    },
    {
      "epoch": 3.56,
      "eval_accuracy": 0.930045871559633,
      "eval_loss": 0.31746456027030945,
      "eval_runtime": 4.983,
      "eval_samples_per_second": 174.994,
      "eval_steps_per_second": 5.619,
      "step": 7500
    },
    {
      "epoch": 3.61,
      "learning_rate": 1.2779097387173398e-05,
      "loss": 0.1079,
      "step": 7600
    },
    {
      "epoch": 3.71,
      "learning_rate": 1.2589073634204277e-05,
      "loss": 0.1023,
      "step": 7800
    },
    {
      "epoch": 3.8,
      "learning_rate": 1.2399049881235155e-05,
      "loss": 0.1015,
      "step": 8000
    },
    {
      "epoch": 3.8,
      "eval_accuracy": 0.9311926605504587,
      "eval_loss": 0.3047061264514923,
      "eval_runtime": 5.0268,
      "eval_samples_per_second": 173.471,
      "eval_steps_per_second": 5.57,
      "step": 8000
    },
    {
      "epoch": 3.9,
      "learning_rate": 1.2209026128266036e-05,
      "loss": 0.1076,
      "step": 8200
    },
    {
      "epoch": 3.99,
      "learning_rate": 1.2019002375296912e-05,
      "loss": 0.1103,
      "step": 8400
    },
    {
      "epoch": 4.04,
      "eval_accuracy": 0.9277522935779816,
      "eval_loss": 0.34547296166419983,
      "eval_runtime": 6.2445,
      "eval_samples_per_second": 139.642,
      "eval_steps_per_second": 4.484,
      "step": 8500
    },
    {
      "epoch": 4.09,
      "learning_rate": 1.1828978622327791e-05,
      "loss": 0.0766,
      "step": 8600
    },
    {
      "epoch": 4.18,
      "learning_rate": 1.1638954869358671e-05,
      "loss": 0.0778,
      "step": 8800
    },
    {
      "epoch": 4.28,
      "learning_rate": 1.144893111638955e-05,
      "loss": 0.0841,
      "step": 9000
    },
    {
      "epoch": 4.28,
      "eval_accuracy": 0.9185779816513762,
      "eval_loss": 0.37421032786369324,
      "eval_runtime": 5.4233,
      "eval_samples_per_second": 160.787,
      "eval_steps_per_second": 5.163,
      "step": 9000
    },
    {
      "epoch": 4.37,
      "learning_rate": 1.1258907363420429e-05,
      "loss": 0.0774,
      "step": 9200
    },
    {
      "epoch": 4.47,
      "learning_rate": 1.1068883610451307e-05,
      "loss": 0.0839,
      "step": 9400
    },
    {
      "epoch": 4.51,
      "eval_accuracy": 0.9220183486238532,
      "eval_loss": 0.3504032492637634,
      "eval_runtime": 6.6438,
      "eval_samples_per_second": 131.249,
      "eval_steps_per_second": 4.214,
      "step": 9500
    },
    {
      "epoch": 4.56,
      "learning_rate": 1.0878859857482188e-05,
      "loss": 0.0857,
      "step": 9600
    },
    {
      "epoch": 4.66,
      "learning_rate": 1.0688836104513065e-05,
      "loss": 0.091,
      "step": 9800
    },
    {
      "epoch": 4.75,
      "learning_rate": 1.0498812351543943e-05,
      "loss": 0.0858,
      "step": 10000
    },
    {
      "epoch": 4.75,
      "eval_accuracy": 0.9288990825688074,
      "eval_loss": 0.2815171778202057,
      "eval_runtime": 4.9169,
      "eval_samples_per_second": 177.347,
      "eval_steps_per_second": 5.695,
      "step": 10000
    },
    {
      "epoch": 4.85,
      "learning_rate": 1.0308788598574823e-05,
      "loss": 0.0852,
      "step": 10200
    },
    {
      "epoch": 4.94,
      "learning_rate": 1.0118764845605702e-05,
      "loss": 0.0897,
      "step": 10400
    },
    {
      "epoch": 4.99,
      "eval_accuracy": 0.9220183486238532,
      "eval_loss": 0.3205639123916626,
      "eval_runtime": 7.2765,
      "eval_samples_per_second": 119.839,
      "eval_steps_per_second": 3.848,
      "step": 10500
    },
    {
      "epoch": 5.04,
      "learning_rate": 9.92874109263658e-06,
      "loss": 0.076,
      "step": 10600
    },
    {
      "epoch": 5.13,
      "learning_rate": 9.73871733966746e-06,
      "loss": 0.067,
      "step": 10800
    },
    {
      "epoch": 5.23,
      "learning_rate": 9.548693586698338e-06,
      "loss": 0.0641,
      "step": 11000
    },
    {
      "epoch": 5.23,
      "eval_accuracy": 0.9334862385321101,
      "eval_loss": 0.32430100440979004,
      "eval_runtime": 5.6582,
      "eval_samples_per_second": 154.113,
      "eval_steps_per_second": 4.949,
      "step": 11000
    },
    {
      "epoch": 5.32,
      "learning_rate": 9.358669833729217e-06,
      "loss": 0.0657,
      "step": 11200
    },
    {
      "epoch": 5.42,
      "learning_rate": 9.168646080760095e-06,
      "loss": 0.0758,
      "step": 11400
    },
    {
      "epoch": 5.46,
      "eval_accuracy": 0.9311926605504587,
      "eval_loss": 0.29846712946891785,
      "eval_runtime": 4.9813,
      "eval_samples_per_second": 175.054,
      "eval_steps_per_second": 5.621,
      "step": 11500
    },
    {
      "epoch": 5.51,
      "learning_rate": 8.978622327790974e-06,
      "loss": 0.0705,
      "step": 11600
    },
    {
      "epoch": 5.61,
      "learning_rate": 8.788598574821854e-06,
      "loss": 0.065,
      "step": 11800
    },
    {
      "epoch": 5.7,
      "learning_rate": 8.598574821852733e-06,
      "loss": 0.0744,
      "step": 12000
    },
    {
      "epoch": 5.7,
      "eval_accuracy": 0.9151376146788991,
      "eval_loss": 0.35644832253456116,
      "eval_runtime": 5.3372,
      "eval_samples_per_second": 163.382,
      "eval_steps_per_second": 5.246,
      "step": 12000
    },
    {
      "epoch": 5.8,
      "learning_rate": 8.408551068883611e-06,
      "loss": 0.07,
      "step": 12200
    },
    {
      "epoch": 5.89,
      "learning_rate": 8.21852731591449e-06,
      "loss": 0.0735,
      "step": 12400
    },
    {
      "epoch": 5.94,
      "eval_accuracy": 0.9220183486238532,
      "eval_loss": 0.36781516671180725,
      "eval_runtime": 5.0223,
      "eval_samples_per_second": 173.627,
      "eval_steps_per_second": 5.575,
      "step": 12500
    },
    {
      "epoch": 5.99,
      "learning_rate": 8.028503562945369e-06,
      "loss": 0.0683,
      "step": 12600
    },
    {
      "epoch": 6.08,
      "learning_rate": 7.838479809976247e-06,
      "loss": 0.0519,
      "step": 12800
    },
    {
      "epoch": 6.18,
      "learning_rate": 7.648456057007126e-06,
      "loss": 0.0581,
      "step": 13000
    },
    {
      "epoch": 6.18,
      "eval_accuracy": 0.9323394495412844,
      "eval_loss": 0.3235052824020386,
      "eval_runtime": 5.3353,
      "eval_samples_per_second": 163.439,
      "eval_steps_per_second": 5.248,
      "step": 13000
    },
    {
      "epoch": 6.27,
      "learning_rate": 7.458432304038005e-06,
      "loss": 0.0575,
      "step": 13200
    },
    {
      "epoch": 6.37,
      "learning_rate": 7.268408551068884e-06,
      "loss": 0.0605,
      "step": 13400
    },
    {
      "epoch": 6.41,
      "eval_accuracy": 0.9254587155963303,
      "eval_loss": 0.3677641451358795,
      "eval_runtime": 4.1548,
      "eval_samples_per_second": 209.877,
      "eval_steps_per_second": 6.739,
      "step": 13500
    },
    {
      "epoch": 6.46,
      "learning_rate": 7.0783847980997635e-06,
      "loss": 0.0548,
      "step": 13600
    },
    {
      "epoch": 6.56,
      "learning_rate": 6.888361045130641e-06,
      "loss": 0.054,
      "step": 13800
    },
    {
      "epoch": 6.65,
      "learning_rate": 6.698337292161521e-06,
      "loss": 0.0608,
      "step": 14000
    },
    {
      "epoch": 6.65,
      "eval_accuracy": 0.9231651376146789,
      "eval_loss": 0.35537707805633545,
      "eval_runtime": 5.0168,
      "eval_samples_per_second": 173.816,
      "eval_steps_per_second": 5.581,
      "step": 14000
    },
    {
      "epoch": 6.75,
      "learning_rate": 6.508313539192399e-06,
      "loss": 0.0589,
      "step": 14200
    },
    {
      "epoch": 6.84,
      "learning_rate": 6.318289786223278e-06,
      "loss": 0.055,
      "step": 14400
    },
    {
      "epoch": 6.89,
      "eval_accuracy": 0.9277522935779816,
      "eval_loss": 0.3431534171104431,
      "eval_runtime": 5.9333,
      "eval_samples_per_second": 146.968,
      "eval_steps_per_second": 4.719,
      "step": 14500
    },
    {
      "epoch": 6.94,
      "learning_rate": 6.1282660332541575e-06,
      "loss": 0.062,
      "step": 14600
    },
    {
      "epoch": 7.03,
      "learning_rate": 5.938242280285036e-06,
      "loss": 0.0571,
      "step": 14800
    },
    {
      "epoch": 7.13,
      "learning_rate": 5.748218527315916e-06,
      "loss": 0.0498,
      "step": 15000
    },
    {
      "epoch": 7.13,
      "eval_accuracy": 0.9288990825688074,
      "eval_loss": 0.3633643090724945,
      "eval_runtime": 4.8896,
      "eval_samples_per_second": 178.339,
      "eval_steps_per_second": 5.726,
      "step": 15000
    },
    {
      "epoch": 7.22,
      "learning_rate": 5.558194774346793e-06,
      "loss": 0.0569,
      "step": 15200
    },
    {
      "epoch": 7.32,
      "learning_rate": 5.368171021377673e-06,
      "loss": 0.0531,
      "step": 15400
    },
    {
      "epoch": 7.36,
      "eval_accuracy": 0.9277522935779816,
      "eval_loss": 0.34541434049606323,
      "eval_runtime": 5.4179,
      "eval_samples_per_second": 160.949,
      "eval_steps_per_second": 5.168,
      "step": 15500
    },
    {
      "epoch": 7.41,
      "learning_rate": 5.178147268408551e-06,
      "loss": 0.0501,
      "step": 15600
    },
    {
      "epoch": 7.51,
      "learning_rate": 4.98812351543943e-06,
      "loss": 0.0501,
      "step": 15800
    },
    {
      "epoch": 7.6,
      "learning_rate": 4.798099762470309e-06,
      "loss": 0.0521,
      "step": 16000
    },
    {
      "epoch": 7.6,
      "eval_accuracy": 0.9243119266055045,
      "eval_loss": 0.3439008593559265,
      "eval_runtime": 4.9353,
      "eval_samples_per_second": 176.685,
      "eval_steps_per_second": 5.673,
      "step": 16000
    },
    {
      "epoch": 7.7,
      "learning_rate": 4.608076009501188e-06,
      "loss": 0.0452,
      "step": 16200
    },
    {
      "epoch": 7.79,
      "learning_rate": 4.418052256532067e-06,
      "loss": 0.0481,
      "step": 16400
    },
    {
      "epoch": 7.84,
      "eval_accuracy": 0.9254587155963303,
      "eval_loss": 0.36306628584861755,
      "eval_runtime": 5.8137,
      "eval_samples_per_second": 149.99,
      "eval_steps_per_second": 4.816,
      "step": 16500
    },
    {
      "epoch": 7.89,
      "learning_rate": 4.228028503562946e-06,
      "loss": 0.0514,
      "step": 16600
    },
    {
      "epoch": 7.98,
      "learning_rate": 4.038004750593825e-06,
      "loss": 0.0486,
      "step": 16800
    },
    {
      "epoch": 8.08,
      "learning_rate": 3.8479809976247036e-06,
      "loss": 0.0386,
      "step": 17000
    },
    {
      "epoch": 8.08,
      "eval_accuracy": 0.930045871559633,
      "eval_loss": 0.37118515372276306,
      "eval_runtime": 6.6001,
      "eval_samples_per_second": 132.12,
      "eval_steps_per_second": 4.242,
      "step": 17000
    },
    {
      "epoch": 8.17,
      "learning_rate": 3.657957244655582e-06,
      "loss": 0.036,
      "step": 17200
    },
    {
      "epoch": 8.27,
      "learning_rate": 3.467933491686461e-06,
      "loss": 0.0399,
      "step": 17400
    },
    {
      "epoch": 8.31,
      "eval_accuracy": 0.9346330275229358,
      "eval_loss": 0.3605540692806244,
      "eval_runtime": 5.1639,
      "eval_samples_per_second": 168.866,
      "eval_steps_per_second": 5.422,
      "step": 17500
    },
    {
      "epoch": 8.36,
      "learning_rate": 3.27790973871734e-06,
      "loss": 0.0436,
      "step": 17600
    },
    {
      "epoch": 8.46,
      "learning_rate": 3.0878859857482185e-06,
      "loss": 0.0417,
      "step": 17800
    },
    {
      "epoch": 8.55,
      "learning_rate": 2.897862232779098e-06,
      "loss": 0.048,
      "step": 18000
    },
    {
      "epoch": 8.55,
      "eval_accuracy": 0.9380733944954128,
      "eval_loss": 0.3584529757499695,
      "eval_runtime": 5.8003,
      "eval_samples_per_second": 150.337,
      "eval_steps_per_second": 4.827,
      "step": 18000
    },
    {
      "epoch": 8.65,
      "learning_rate": 2.7078384798099766e-06,
      "loss": 0.0384,
      "step": 18200
    },
    {
      "epoch": 8.74,
      "learning_rate": 2.5178147268408552e-06,
      "loss": 0.0522,
      "step": 18400
    },
    {
      "epoch": 8.79,
      "eval_accuracy": 0.9334862385321101,
      "eval_loss": 0.337113618850708,
      "eval_runtime": 5.4571,
      "eval_samples_per_second": 159.791,
      "eval_steps_per_second": 5.131,
      "step": 18500
    }
  ],
  "max_steps": 21050,
  "num_train_epochs": 10,
  "total_flos": 4.538296624748544e+16,
  "trial_name": null,
  "trial_params": null
}