{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 7.0,
  "global_step": 160580,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.02,
      "learning_rate": 1.5566625155666254e-05,
      "loss": 9.0543,
      "step": 500
    },
    {
      "epoch": 0.04,
      "learning_rate": 3.113325031133251e-05,
      "loss": 6.7369,
      "step": 1000
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.6699875466998756e-05,
      "loss": 6.0998,
      "step": 1500
    },
    {
      "epoch": 0.09,
      "learning_rate": 6.226650062266502e-05,
      "loss": 5.4282,
      "step": 2000
    },
    {
      "epoch": 0.11,
      "learning_rate": 7.783312577833126e-05,
      "loss": 4.749,
      "step": 2500
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.339975093399751e-05,
      "loss": 4.3351,
      "step": 3000
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.990941915030131e-05,
      "loss": 4.0473,
      "step": 3500
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.975216073068553e-05,
      "loss": 3.8251,
      "step": 4000
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.959490231106974e-05,
      "loss": 3.6631,
      "step": 4500
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.943764389145395e-05,
      "loss": 3.532,
      "step": 5000
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.928038547183816e-05,
      "loss": 3.4285,
      "step": 5500
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.912312705222237e-05,
      "loss": 3.3414,
      "step": 6000
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.89658686326066e-05,
      "loss": 3.2622,
      "step": 6500
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.880861021299081e-05,
      "loss": 3.2048,
      "step": 7000
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.865135179337502e-05,
      "loss": 3.1463,
      "step": 7500
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.849409337375923e-05,
      "loss": 3.0932,
      "step": 8000
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.833683495414345e-05,
      "loss": 3.0433,
      "step": 8500
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.817957653452766e-05,
      "loss": 3.0056,
      "step": 9000
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.802231811491189e-05,
      "loss": 2.9611,
      "step": 9500
    },
    {
      "epoch": 0.44,
      "learning_rate": 9.78650596952961e-05,
      "loss": 2.9247,
      "step": 10000
    },
    {
      "epoch": 0.46,
      "learning_rate": 9.77078012756803e-05,
      "loss": 2.8874,
      "step": 10500
    },
    {
      "epoch": 0.48,
      "learning_rate": 9.755054285606452e-05,
      "loss": 2.8569,
      "step": 11000
    },
    {
      "epoch": 0.5,
      "learning_rate": 9.739328443644873e-05,
      "loss": 2.8276,
      "step": 11500
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.723602601683295e-05,
      "loss": 2.7987,
      "step": 12000
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.70790821140564e-05,
      "loss": 2.7716,
      "step": 12500
    },
    {
      "epoch": 0.57,
      "learning_rate": 9.69218236944406e-05,
      "loss": 2.745,
      "step": 13000
    },
    {
      "epoch": 0.59,
      "learning_rate": 9.676456527482482e-05,
      "loss": 2.7229,
      "step": 13500
    },
    {
      "epoch": 0.61,
      "learning_rate": 9.660730685520904e-05,
      "loss": 2.7021,
      "step": 14000
    },
    {
      "epoch": 0.63,
      "learning_rate": 9.645036295243249e-05,
      "loss": 2.679,
      "step": 14500
    },
    {
      "epoch": 0.65,
      "learning_rate": 9.62931045328167e-05,
      "loss": 2.6618,
      "step": 15000
    },
    {
      "epoch": 0.68,
      "learning_rate": 9.61358461132009e-05,
      "loss": 2.6444,
      "step": 15500
    },
    {
      "epoch": 0.7,
      "learning_rate": 9.597858769358512e-05,
      "loss": 2.6249,
      "step": 16000
    },
    {
      "epoch": 0.72,
      "learning_rate": 9.582164379080856e-05,
      "loss": 2.6169,
      "step": 16500
    },
    {
      "epoch": 0.74,
      "learning_rate": 9.566438537119277e-05,
      "loss": 2.5936,
      "step": 17000
    },
    {
      "epoch": 0.76,
      "learning_rate": 9.5507126951577e-05,
      "loss": 2.5854,
      "step": 17500
    },
    {
      "epoch": 0.78,
      "learning_rate": 9.535018304880043e-05,
      "loss": 2.5693,
      "step": 18000
    },
    {
      "epoch": 0.81,
      "learning_rate": 9.519292462918465e-05,
      "loss": 2.5525,
      "step": 18500
    },
    {
      "epoch": 0.83,
      "learning_rate": 9.503566620956886e-05,
      "loss": 2.539,
      "step": 19000
    },
    {
      "epoch": 0.85,
      "learning_rate": 9.487840778995309e-05,
      "loss": 2.5256,
      "step": 19500
    },
    {
      "epoch": 0.87,
      "learning_rate": 9.47211493703373e-05,
      "loss": 2.5207,
      "step": 20000
    },
    {
      "epoch": 0.89,
      "learning_rate": 9.45638909507215e-05,
      "loss": 2.5017,
      "step": 20500
    },
    {
      "epoch": 0.92,
      "learning_rate": 9.440663253110572e-05,
      "loss": 2.4895,
      "step": 21000
    },
    {
      "epoch": 0.94,
      "learning_rate": 9.424968862832916e-05,
      "loss": 2.4808,
      "step": 21500
    },
    {
      "epoch": 0.96,
      "learning_rate": 9.409243020871337e-05,
      "loss": 2.4713,
      "step": 22000
    },
    {
      "epoch": 0.98,
      "learning_rate": 9.39351717890976e-05,
      "loss": 2.4602,
      "step": 22500
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.5603946985569819,
      "eval_loss": 2.3326776027679443,
      "eval_runtime": 361.5158,
      "eval_samples_per_second": 492.238,
      "eval_steps_per_second": 30.765,
      "step": 22940
    },
    {
      "epoch": 1.0,
      "learning_rate": 9.377791336948181e-05,
      "loss": 2.4505,
      "step": 23000
    },
    {
      "epoch": 1.02,
      "learning_rate": 9.362096946670525e-05,
      "loss": 2.4409,
      "step": 23500
    },
    {
      "epoch": 1.05,
      "learning_rate": 9.346371104708946e-05,
      "loss": 2.4304,
      "step": 24000
    },
    {
      "epoch": 1.07,
      "learning_rate": 9.330645262747369e-05,
      "loss": 2.4249,
      "step": 24500
    },
    {
      "epoch": 1.09,
      "learning_rate": 9.31491942078579e-05,
      "loss": 2.4169,
      "step": 25000
    },
    {
      "epoch": 1.11,
      "learning_rate": 9.299225030508134e-05,
      "loss": 2.4096,
      "step": 25500
    },
    {
      "epoch": 1.13,
      "learning_rate": 9.283499188546555e-05,
      "loss": 2.3994,
      "step": 26000
    },
    {
      "epoch": 1.16,
      "learning_rate": 9.267773346584976e-05,
      "loss": 2.3931,
      "step": 26500
    },
    {
      "epoch": 1.18,
      "learning_rate": 9.252047504623397e-05,
      "loss": 2.3902,
      "step": 27000
    },
    {
      "epoch": 1.2,
      "learning_rate": 9.236353114345743e-05,
      "loss": 2.3863,
      "step": 27500
    },
    {
      "epoch": 1.22,
      "learning_rate": 9.220627272384163e-05,
      "loss": 2.3745,
      "step": 28000
    },
    {
      "epoch": 1.24,
      "learning_rate": 9.204901430422585e-05,
      "loss": 2.3619,
      "step": 28500
    },
    {
      "epoch": 1.26,
      "learning_rate": 9.189175588461006e-05,
      "loss": 2.3611,
      "step": 29000
    },
    {
      "epoch": 1.29,
      "learning_rate": 9.173481198183351e-05,
      "loss": 2.3515,
      "step": 29500
    },
    {
      "epoch": 1.31,
      "learning_rate": 9.157755356221772e-05,
      "loss": 2.3454,
      "step": 30000
    },
    {
      "epoch": 1.33,
      "learning_rate": 9.142029514260194e-05,
      "loss": 2.3421,
      "step": 30500
    },
    {
      "epoch": 1.35,
      "learning_rate": 9.126303672298615e-05,
      "loss": 2.3337,
      "step": 31000
    },
    {
      "epoch": 1.37,
      "learning_rate": 9.11060928202096e-05,
      "loss": 2.3294,
      "step": 31500
    },
    {
      "epoch": 1.39,
      "learning_rate": 9.094883440059381e-05,
      "loss": 2.3294,
      "step": 32000
    },
    {
      "epoch": 1.42,
      "learning_rate": 9.079157598097803e-05,
      "loss": 2.32,
      "step": 32500
    },
    {
      "epoch": 1.44,
      "learning_rate": 9.063431756136223e-05,
      "loss": 2.3175,
      "step": 33000
    },
    {
      "epoch": 1.46,
      "learning_rate": 9.047737365858569e-05,
      "loss": 2.3086,
      "step": 33500
    },
    {
      "epoch": 1.48,
      "learning_rate": 9.03201152389699e-05,
      "loss": 2.3043,
      "step": 34000
    },
    {
      "epoch": 1.5,
      "learning_rate": 9.016285681935411e-05,
      "loss": 2.2977,
      "step": 34500
    },
    {
      "epoch": 1.53,
      "learning_rate": 9.000559839973832e-05,
      "loss": 2.2956,
      "step": 35000
    },
    {
      "epoch": 1.55,
      "learning_rate": 8.984865449696177e-05,
      "loss": 2.2941,
      "step": 35500
    },
    {
      "epoch": 1.57,
      "learning_rate": 8.969139607734599e-05,
      "loss": 2.289,
      "step": 36000
    },
    {
      "epoch": 1.59,
      "learning_rate": 8.95341376577302e-05,
      "loss": 2.282,
      "step": 36500
    },
    {
      "epoch": 1.61,
      "learning_rate": 8.937687923811441e-05,
      "loss": 2.278,
      "step": 37000
    },
    {
      "epoch": 1.63,
      "learning_rate": 8.921993533533786e-05,
      "loss": 2.2772,
      "step": 37500
    },
    {
      "epoch": 1.66,
      "learning_rate": 8.906267691572208e-05,
      "loss": 2.2656,
      "step": 38000
    },
    {
      "epoch": 1.68,
      "learning_rate": 8.890541849610629e-05,
      "loss": 2.2639,
      "step": 38500
    },
    {
      "epoch": 1.7,
      "learning_rate": 8.87481600764905e-05,
      "loss": 2.256,
      "step": 39000
    },
    {
      "epoch": 1.72,
      "learning_rate": 8.859121617371395e-05,
      "loss": 2.2608,
      "step": 39500
    },
    {
      "epoch": 1.74,
      "learning_rate": 8.843395775409816e-05,
      "loss": 2.2586,
      "step": 40000
    },
    {
      "epoch": 1.77,
      "learning_rate": 8.827669933448237e-05,
      "loss": 2.2503,
      "step": 40500
    },
    {
      "epoch": 1.79,
      "learning_rate": 8.811944091486659e-05,
      "loss": 2.2452,
      "step": 41000
    },
    {
      "epoch": 1.81,
      "learning_rate": 8.796249701209002e-05,
      "loss": 2.246,
      "step": 41500
    },
    {
      "epoch": 1.83,
      "learning_rate": 8.780523859247425e-05,
      "loss": 2.2347,
      "step": 42000
    },
    {
      "epoch": 1.85,
      "learning_rate": 8.764798017285846e-05,
      "loss": 2.2385,
      "step": 42500
    },
    {
      "epoch": 1.87,
      "learning_rate": 8.749072175324268e-05,
      "loss": 2.2317,
      "step": 43000
    },
    {
      "epoch": 1.9,
      "learning_rate": 8.733377785046611e-05,
      "loss": 2.2273,
      "step": 43500
    },
    {
      "epoch": 1.92,
      "learning_rate": 8.717651943085034e-05,
      "loss": 2.2274,
      "step": 44000
    },
    {
      "epoch": 1.94,
      "learning_rate": 8.701926101123455e-05,
      "loss": 2.2239,
      "step": 44500
    },
    {
      "epoch": 1.96,
      "learning_rate": 8.686200259161877e-05,
      "loss": 2.2145,
      "step": 45000
    },
    {
      "epoch": 1.98,
      "learning_rate": 8.67050586888422e-05,
      "loss": 2.2189,
      "step": 45500
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.5917185537566337,
      "eval_loss": 2.0854098796844482,
      "eval_runtime": 362.7142,
      "eval_samples_per_second": 490.612,
      "eval_steps_per_second": 30.663,
      "step": 45880
    },
    {
      "epoch": 2.01,
      "learning_rate": 8.654780026922643e-05,
      "loss": 2.213,
      "step": 46000
    },
    {
      "epoch": 2.03,
      "learning_rate": 8.639054184961062e-05,
      "loss": 2.2002,
      "step": 46500
    },
    {
      "epoch": 2.05,
      "learning_rate": 8.623328342999485e-05,
      "loss": 2.1991,
      "step": 47000
    },
    {
      "epoch": 2.07,
      "learning_rate": 8.607633952721829e-05,
      "loss": 2.202,
      "step": 47500
    },
    {
      "epoch": 2.09,
      "learning_rate": 8.59190811076025e-05,
      "loss": 2.1981,
      "step": 48000
    },
    {
      "epoch": 2.11,
      "learning_rate": 8.576182268798671e-05,
      "loss": 2.1908,
      "step": 48500
    },
    {
      "epoch": 2.14,
      "learning_rate": 8.560456426837094e-05,
      "loss": 2.1923,
      "step": 49000
    },
    {
      "epoch": 2.16,
      "learning_rate": 8.544762036559438e-05,
      "loss": 2.1905,
      "step": 49500
    },
    {
      "epoch": 2.18,
      "learning_rate": 8.529036194597859e-05,
      "loss": 2.184,
      "step": 50000
    },
    {
      "epoch": 2.2,
      "learning_rate": 8.51331035263628e-05,
      "loss": 2.1787,
      "step": 50500
    },
    {
      "epoch": 2.22,
      "learning_rate": 8.497584510674703e-05,
      "loss": 2.1846,
      "step": 51000
    },
    {
      "epoch": 2.24,
      "learning_rate": 8.481890120397047e-05,
      "loss": 2.1789,
      "step": 51500
    },
    {
      "epoch": 2.27,
      "learning_rate": 8.466164278435468e-05,
      "loss": 2.1774,
      "step": 52000
    },
    {
      "epoch": 2.29,
      "learning_rate": 8.450438436473889e-05,
      "loss": 2.1738,
      "step": 52500
    },
    {
      "epoch": 2.31,
      "learning_rate": 8.43471259451231e-05,
      "loss": 2.1707,
      "step": 53000
    },
    {
      "epoch": 2.33,
      "learning_rate": 8.419018204234655e-05,
      "loss": 2.1695,
      "step": 53500
    },
    {
      "epoch": 2.35,
      "learning_rate": 8.403292362273076e-05,
      "loss": 2.1649,
      "step": 54000
    },
    {
      "epoch": 2.38,
      "learning_rate": 8.387566520311498e-05,
      "loss": 2.16,
      "step": 54500
    },
    {
      "epoch": 2.4,
      "learning_rate": 8.371840678349919e-05,
      "loss": 2.1559,
      "step": 55000
    },
    {
      "epoch": 2.42,
      "learning_rate": 8.356146288072264e-05,
      "loss": 2.158,
      "step": 55500
    },
    {
      "epoch": 2.44,
      "learning_rate": 8.340420446110685e-05,
      "loss": 2.1555,
      "step": 56000
    },
    {
      "epoch": 2.46,
      "learning_rate": 8.324694604149107e-05,
      "loss": 2.1527,
      "step": 56500
    },
    {
      "epoch": 2.48,
      "learning_rate": 8.308968762187528e-05,
      "loss": 2.1512,
      "step": 57000
    },
    {
      "epoch": 2.51,
      "learning_rate": 8.293274371909873e-05,
      "loss": 2.1523,
      "step": 57500
    },
    {
      "epoch": 2.53,
      "learning_rate": 8.277548529948294e-05,
      "loss": 2.1444,
      "step": 58000
    },
    {
      "epoch": 2.55,
      "learning_rate": 8.261822687986716e-05,
      "loss": 2.1453,
      "step": 58500
    },
    {
      "epoch": 2.57,
      "learning_rate": 8.246096846025136e-05,
      "loss": 2.143,
      "step": 59000
    },
    {
      "epoch": 2.59,
      "learning_rate": 8.230402455747482e-05,
      "loss": 2.1377,
      "step": 59500
    },
    {
      "epoch": 2.62,
      "learning_rate": 8.214676613785903e-05,
      "loss": 2.1342,
      "step": 60000
    },
    {
      "epoch": 2.64,
      "learning_rate": 8.198950771824324e-05,
      "loss": 2.1356,
      "step": 60500
    },
    {
      "epoch": 2.66,
      "learning_rate": 8.183224929862745e-05,
      "loss": 2.136,
      "step": 61000
    },
    {
      "epoch": 2.68,
      "learning_rate": 8.16753053958509e-05,
      "loss": 2.1317,
      "step": 61500
    },
    {
      "epoch": 2.7,
      "learning_rate": 8.15180469762351e-05,
      "loss": 2.1316,
      "step": 62000
    },
    {
      "epoch": 2.72,
      "learning_rate": 8.136078855661933e-05,
      "loss": 2.1297,
      "step": 62500
    },
    {
      "epoch": 2.75,
      "learning_rate": 8.120353013700354e-05,
      "loss": 2.1258,
      "step": 63000
    },
    {
      "epoch": 2.77,
      "learning_rate": 8.104627171738776e-05,
      "loss": 2.1227,
      "step": 63500
    },
    {
      "epoch": 2.79,
      "learning_rate": 8.08893278146112e-05,
      "loss": 2.1209,
      "step": 64000
    },
    {
      "epoch": 2.81,
      "learning_rate": 8.073206939499542e-05,
      "loss": 2.1214,
      "step": 64500
    },
    {
      "epoch": 2.83,
      "learning_rate": 8.057481097537963e-05,
      "loss": 2.1213,
      "step": 65000
    },
    {
      "epoch": 2.86,
      "learning_rate": 8.041755255576384e-05,
      "loss": 2.1189,
      "step": 65500
    },
    {
      "epoch": 2.88,
      "learning_rate": 8.026060865298728e-05,
      "loss": 2.1135,
      "step": 66000
    },
    {
      "epoch": 2.9,
      "learning_rate": 8.01033502333715e-05,
      "loss": 2.1103,
      "step": 66500
    },
    {
      "epoch": 2.92,
      "learning_rate": 7.99460918137557e-05,
      "loss": 2.1117,
      "step": 67000
    },
    {
      "epoch": 2.94,
      "learning_rate": 7.978883339413993e-05,
      "loss": 2.1109,
      "step": 67500
    },
    {
      "epoch": 2.96,
      "learning_rate": 7.963188949136337e-05,
      "loss": 2.1073,
      "step": 68000
    },
    {
      "epoch": 2.99,
      "learning_rate": 7.947463107174758e-05,
      "loss": 2.1068,
      "step": 68500
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.6074305807247455,
      "eval_loss": 1.9807746410369873,
      "eval_runtime": 358.1234,
      "eval_samples_per_second": 496.901,
      "eval_steps_per_second": 31.056,
      "step": 68820
    },
    {
      "epoch": 3.01,
      "learning_rate": 7.93173726521318e-05,
      "loss": 2.1035,
      "step": 69000
    },
    {
      "epoch": 3.03,
      "learning_rate": 7.916011423251602e-05,
      "loss": 2.1006,
      "step": 69500
    },
    {
      "epoch": 3.05,
      "learning_rate": 7.900317032973946e-05,
      "loss": 2.0974,
      "step": 70000
    },
    {
      "epoch": 3.07,
      "learning_rate": 7.884591191012367e-05,
      "loss": 2.0998,
      "step": 70500
    },
    {
      "epoch": 3.1,
      "learning_rate": 7.868865349050788e-05,
      "loss": 2.0908,
      "step": 71000
    },
    {
      "epoch": 3.12,
      "learning_rate": 7.85313950708921e-05,
      "loss": 2.0897,
      "step": 71500
    },
    {
      "epoch": 3.14,
      "learning_rate": 7.837445116811555e-05,
      "loss": 2.0927,
      "step": 72000
    },
    {
      "epoch": 3.16,
      "learning_rate": 7.821719274849975e-05,
      "loss": 2.0891,
      "step": 72500
    },
    {
      "epoch": 3.18,
      "learning_rate": 7.805993432888397e-05,
      "loss": 2.0876,
      "step": 73000
    },
    {
      "epoch": 3.2,
      "learning_rate": 7.790267590926819e-05,
      "loss": 2.0872,
      "step": 73500
    },
    {
      "epoch": 3.23,
      "learning_rate": 7.774573200649163e-05,
      "loss": 2.085,
      "step": 74000
    },
    {
      "epoch": 3.25,
      "learning_rate": 7.758847358687584e-05,
      "loss": 2.0806,
      "step": 74500
    },
    {
      "epoch": 3.27,
      "learning_rate": 7.743121516726006e-05,
      "loss": 2.0836,
      "step": 75000
    },
    {
      "epoch": 3.29,
      "learning_rate": 7.727395674764428e-05,
      "loss": 2.0802,
      "step": 75500
    },
    {
      "epoch": 3.31,
      "learning_rate": 7.711701284486772e-05,
      "loss": 2.0784,
      "step": 76000
    },
    {
      "epoch": 3.33,
      "learning_rate": 7.695975442525193e-05,
      "loss": 2.0796,
      "step": 76500
    },
    {
      "epoch": 3.36,
      "learning_rate": 7.680249600563615e-05,
      "loss": 2.0748,
      "step": 77000
    },
    {
      "epoch": 3.38,
      "learning_rate": 7.664523758602035e-05,
      "loss": 2.0778,
      "step": 77500
    },
    {
      "epoch": 3.4,
      "learning_rate": 7.648829368324381e-05,
      "loss": 2.0705,
      "step": 78000
    },
    {
      "epoch": 3.42,
      "learning_rate": 7.633103526362802e-05,
      "loss": 2.0743,
      "step": 78500
    },
    {
      "epoch": 3.44,
      "learning_rate": 7.617377684401223e-05,
      "loss": 2.0706,
      "step": 79000
    },
    {
      "epoch": 3.47,
      "learning_rate": 7.601651842439644e-05,
      "loss": 2.0684,
      "step": 79500
    },
    {
      "epoch": 3.49,
      "learning_rate": 7.585957452161989e-05,
      "loss": 2.0713,
      "step": 80000
    },
    {
      "epoch": 3.51,
      "learning_rate": 7.57023161020041e-05,
      "loss": 2.0681,
      "step": 80500
    },
    {
      "epoch": 3.53,
      "learning_rate": 7.554505768238832e-05,
      "loss": 2.0674,
      "step": 81000
    },
    {
      "epoch": 3.55,
      "learning_rate": 7.538779926277253e-05,
      "loss": 2.0669,
      "step": 81500
    },
    {
      "epoch": 3.57,
      "learning_rate": 7.523085535999598e-05,
      "loss": 2.0657,
      "step": 82000
    },
    {
      "epoch": 3.6,
      "learning_rate": 7.507359694038019e-05,
      "loss": 2.0613,
      "step": 82500
    },
    {
      "epoch": 3.62,
      "learning_rate": 7.491633852076441e-05,
      "loss": 2.0604,
      "step": 83000
    },
    {
      "epoch": 3.64,
      "learning_rate": 7.475908010114862e-05,
      "loss": 2.0623,
      "step": 83500
    },
    {
      "epoch": 3.66,
      "learning_rate": 7.460213619837207e-05,
      "loss": 2.0571,
      "step": 84000
    },
    {
      "epoch": 3.68,
      "learning_rate": 7.444487777875628e-05,
      "loss": 2.0554,
      "step": 84500
    },
    {
      "epoch": 3.71,
      "learning_rate": 7.428761935914049e-05,
      "loss": 2.0539,
      "step": 85000
    },
    {
      "epoch": 3.73,
      "learning_rate": 7.41303609395247e-05,
      "loss": 2.0514,
      "step": 85500
    },
    {
      "epoch": 3.75,
      "learning_rate": 7.397341703674814e-05,
      "loss": 2.054,
      "step": 86000
    },
    {
      "epoch": 3.77,
      "learning_rate": 7.381615861713237e-05,
      "loss": 2.0526,
      "step": 86500
    },
    {
      "epoch": 3.79,
      "learning_rate": 7.365890019751658e-05,
      "loss": 2.0502,
      "step": 87000
    },
    {
      "epoch": 3.81,
      "learning_rate": 7.350164177790079e-05,
      "loss": 2.0496,
      "step": 87500
    },
    {
      "epoch": 3.84,
      "learning_rate": 7.334469787512423e-05,
      "loss": 2.0481,
      "step": 88000
    },
    {
      "epoch": 3.86,
      "learning_rate": 7.318743945550846e-05,
      "loss": 2.0547,
      "step": 88500
    },
    {
      "epoch": 3.88,
      "learning_rate": 7.303018103589267e-05,
      "loss": 2.0447,
      "step": 89000
    },
    {
      "epoch": 3.9,
      "learning_rate": 7.287292261627688e-05,
      "loss": 2.0466,
      "step": 89500
    },
    {
      "epoch": 3.92,
      "learning_rate": 7.271566419666109e-05,
      "loss": 2.0442,
      "step": 90000
    },
    {
      "epoch": 3.95,
      "learning_rate": 7.255872029388455e-05,
      "loss": 2.0433,
      "step": 90500
    },
    {
      "epoch": 3.97,
      "learning_rate": 7.240146187426874e-05,
      "loss": 2.0438,
      "step": 91000
    },
    {
      "epoch": 3.99,
      "learning_rate": 7.224420345465297e-05,
      "loss": 2.0402,
      "step": 91500
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.615199760449867,
      "eval_loss": 1.9178446531295776,
      "eval_runtime": 357.8415,
      "eval_samples_per_second": 497.293,
      "eval_steps_per_second": 31.081,
      "step": 91760
    },
    {
      "epoch": 4.01,
      "learning_rate": 7.208694503503718e-05,
      "loss": 2.034,
      "step": 92000
    },
    {
      "epoch": 4.03,
      "learning_rate": 7.193000113226062e-05,
      "loss": 2.0301,
      "step": 92500
    },
    {
      "epoch": 4.05,
      "learning_rate": 7.177274271264483e-05,
      "loss": 2.0291,
      "step": 93000
    },
    {
      "epoch": 4.08,
      "learning_rate": 7.161548429302906e-05,
      "loss": 2.0322,
      "step": 93500
    },
    {
      "epoch": 4.1,
      "learning_rate": 7.145822587341327e-05,
      "loss": 2.0284,
      "step": 94000
    },
    {
      "epoch": 4.12,
      "learning_rate": 7.130128197063671e-05,
      "loss": 2.0297,
      "step": 94500
    },
    {
      "epoch": 4.14,
      "learning_rate": 7.114402355102092e-05,
      "loss": 2.0276,
      "step": 95000
    },
    {
      "epoch": 4.16,
      "learning_rate": 7.098676513140515e-05,
      "loss": 2.028,
      "step": 95500
    },
    {
      "epoch": 4.18,
      "learning_rate": 7.082950671178934e-05,
      "loss": 2.0207,
      "step": 96000
    },
    {
      "epoch": 4.21,
      "learning_rate": 7.067224829217357e-05,
      "loss": 2.0215,
      "step": 96500
    },
    {
      "epoch": 4.23,
      "learning_rate": 7.051530438939701e-05,
      "loss": 2.0271,
      "step": 97000
    },
    {
      "epoch": 4.25,
      "learning_rate": 7.035804596978122e-05,
      "loss": 2.0238,
      "step": 97500
    },
    {
      "epoch": 4.27,
      "learning_rate": 7.020078755016543e-05,
      "loss": 2.0258,
      "step": 98000
    },
    {
      "epoch": 4.29,
      "learning_rate": 7.004352913054966e-05,
      "loss": 2.0222,
      "step": 98500
    },
    {
      "epoch": 4.32,
      "learning_rate": 6.988658522777309e-05,
      "loss": 2.0208,
      "step": 99000
    },
    {
      "epoch": 4.34,
      "learning_rate": 6.972932680815731e-05,
      "loss": 2.0201,
      "step": 99500
    },
    {
      "epoch": 4.36,
      "learning_rate": 6.957206838854152e-05,
      "loss": 2.0209,
      "step": 100000
    },
    {
      "epoch": 4.38,
      "learning_rate": 6.941480996892575e-05,
      "loss": 2.0172,
      "step": 100500
    },
    {
      "epoch": 4.4,
      "learning_rate": 6.925786606614918e-05,
      "loss": 2.0228,
      "step": 101000
    },
    {
      "epoch": 4.42,
      "learning_rate": 6.91006076465334e-05,
      "loss": 2.0132,
      "step": 101500
    },
    {
      "epoch": 4.45,
      "learning_rate": 6.894334922691761e-05,
      "loss": 2.0141,
      "step": 102000
    },
    {
      "epoch": 4.47,
      "learning_rate": 6.878609080730182e-05,
      "loss": 2.0148,
      "step": 102500
    },
    {
      "epoch": 4.49,
      "learning_rate": 6.862914690452527e-05,
      "loss": 2.0111,
      "step": 103000
    },
    {
      "epoch": 4.51,
      "learning_rate": 6.847188848490948e-05,
      "loss": 2.012,
      "step": 103500
    },
    {
      "epoch": 4.53,
      "learning_rate": 6.831463006529369e-05,
      "loss": 2.0083,
      "step": 104000
    },
    {
      "epoch": 4.56,
      "learning_rate": 6.815737164567791e-05,
      "loss": 2.0135,
      "step": 104500
    },
    {
      "epoch": 4.58,
      "learning_rate": 6.800042774290136e-05,
      "loss": 2.0127,
      "step": 105000
    },
    {
      "epoch": 4.6,
      "learning_rate": 6.784316932328557e-05,
      "loss": 2.0064,
      "step": 105500
    },
    {
      "epoch": 4.62,
      "learning_rate": 6.768591090366978e-05,
      "loss": 2.0096,
      "step": 106000
    },
    {
      "epoch": 4.64,
      "learning_rate": 6.7528652484054e-05,
      "loss": 2.011,
      "step": 106500
    },
    {
      "epoch": 4.66,
      "learning_rate": 6.737170858127745e-05,
      "loss": 2.0083,
      "step": 107000
    },
    {
      "epoch": 4.69,
      "learning_rate": 6.721445016166166e-05,
      "loss": 2.0025,
      "step": 107500
    },
    {
      "epoch": 4.71,
      "learning_rate": 6.705719174204587e-05,
      "loss": 2.0069,
      "step": 108000
    },
    {
      "epoch": 4.73,
      "learning_rate": 6.689993332243008e-05,
      "loss": 2.0025,
      "step": 108500
    },
    {
      "epoch": 4.75,
      "learning_rate": 6.674298941965354e-05,
      "loss": 2.0028,
      "step": 109000
    },
    {
      "epoch": 4.77,
      "learning_rate": 6.658573100003774e-05,
      "loss": 2.0058,
      "step": 109500
    },
    {
      "epoch": 4.8,
      "learning_rate": 6.642847258042196e-05,
      "loss": 2.0015,
      "step": 110000
    },
    {
      "epoch": 4.82,
      "learning_rate": 6.627121416080617e-05,
      "loss": 2.0062,
      "step": 110500
    },
    {
      "epoch": 4.84,
      "learning_rate": 6.611427025802962e-05,
      "loss": 1.9985,
      "step": 111000
    },
    {
      "epoch": 4.86,
      "learning_rate": 6.595701183841383e-05,
      "loss": 1.9983,
      "step": 111500
    },
    {
      "epoch": 4.88,
      "learning_rate": 6.579975341879805e-05,
      "loss": 1.996,
      "step": 112000
    },
    {
      "epoch": 4.9,
      "learning_rate": 6.564249499918226e-05,
      "loss": 1.9952,
      "step": 112500
    },
    {
      "epoch": 4.93,
      "learning_rate": 6.54855510964057e-05,
      "loss": 1.9976,
      "step": 113000
    },
    {
      "epoch": 4.95,
      "learning_rate": 6.532829267678992e-05,
      "loss": 1.9986,
      "step": 113500
    },
    {
      "epoch": 4.97,
      "learning_rate": 6.517103425717414e-05,
      "loss": 1.9985,
      "step": 114000
    },
    {
      "epoch": 4.99,
      "learning_rate": 6.501377583755834e-05,
      "loss": 1.9928,
      "step": 114500
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.6206939460659205,
      "eval_loss": 1.877627968788147,
      "eval_runtime": 358.6359,
      "eval_samples_per_second": 496.191,
      "eval_steps_per_second": 31.012,
      "step": 114700
    },
    {
      "epoch": 5.01,
      "learning_rate": 6.485651741794256e-05,
      "loss": 1.9898,
      "step": 115000
    },
    {
      "epoch": 5.03,
      "learning_rate": 6.469957351516601e-05,
      "loss": 1.988,
      "step": 115500
    },
    {
      "epoch": 5.06,
      "learning_rate": 6.454231509555022e-05,
      "loss": 1.9881,
      "step": 116000
    },
    {
      "epoch": 5.08,
      "learning_rate": 6.438505667593443e-05,
      "loss": 1.9849,
      "step": 116500
    },
    {
      "epoch": 5.1,
      "learning_rate": 6.422779825631865e-05,
      "loss": 1.9872,
      "step": 117000
    },
    {
      "epoch": 5.12,
      "learning_rate": 6.407085435354208e-05,
      "loss": 1.9834,
      "step": 117500
    },
    {
      "epoch": 5.14,
      "learning_rate": 6.391359593392631e-05,
      "loss": 1.9814,
      "step": 118000
    },
    {
      "epoch": 5.17,
      "learning_rate": 6.375633751431052e-05,
      "loss": 1.978,
      "step": 118500
    },
    {
      "epoch": 5.19,
      "learning_rate": 6.359907909469474e-05,
      "loss": 1.9857,
      "step": 119000
    },
    {
      "epoch": 5.21,
      "learning_rate": 6.344213519191819e-05,
      "loss": 1.9816,
      "step": 119500
    },
    {
      "epoch": 5.23,
      "learning_rate": 6.32848767723024e-05,
      "loss": 1.9817,
      "step": 120000
    },
    {
      "epoch": 5.25,
      "learning_rate": 6.312761835268661e-05,
      "loss": 1.9815,
      "step": 120500
    },
    {
      "epoch": 5.27,
      "learning_rate": 6.297035993307082e-05,
      "loss": 1.9796,
      "step": 121000
    },
    {
      "epoch": 5.3,
      "learning_rate": 6.281341603029426e-05,
      "loss": 1.9842,
      "step": 121500
    },
    {
      "epoch": 5.32,
      "learning_rate": 6.265615761067847e-05,
      "loss": 1.9816,
      "step": 122000
    },
    {
      "epoch": 5.34,
      "learning_rate": 6.249889919106268e-05,
      "loss": 1.976,
      "step": 122500
    },
    {
      "epoch": 5.36,
      "learning_rate": 6.234164077144691e-05,
      "loss": 1.9798,
      "step": 123000
    },
    {
      "epoch": 5.38,
      "learning_rate": 6.218469686867035e-05,
      "loss": 1.9768,
      "step": 123500
    },
    {
      "epoch": 5.41,
      "learning_rate": 6.202743844905456e-05,
      "loss": 1.9786,
      "step": 124000
    },
    {
      "epoch": 5.43,
      "learning_rate": 6.187018002943879e-05,
      "loss": 1.9753,
      "step": 124500
    },
    {
      "epoch": 5.45,
      "learning_rate": 6.1712921609823e-05,
      "loss": 1.9755,
      "step": 125000
    },
    {
      "epoch": 5.47,
      "learning_rate": 6.155597770704644e-05,
      "loss": 1.9771,
      "step": 125500
    },
    {
      "epoch": 5.49,
      "learning_rate": 6.139871928743065e-05,
      "loss": 1.9736,
      "step": 126000
    },
    {
      "epoch": 5.51,
      "learning_rate": 6.124146086781488e-05,
      "loss": 1.9729,
      "step": 126500
    },
    {
      "epoch": 5.54,
      "learning_rate": 6.108420244819907e-05,
      "loss": 1.9692,
      "step": 127000
    },
    {
      "epoch": 5.56,
      "learning_rate": 6.0927258545422526e-05,
      "loss": 1.9702,
      "step": 127500
    },
    {
      "epoch": 5.58,
      "learning_rate": 6.077000012580674e-05,
      "loss": 1.9709,
      "step": 128000
    },
    {
      "epoch": 5.6,
      "learning_rate": 6.0612741706190954e-05,
      "loss": 1.9712,
      "step": 128500
    },
    {
      "epoch": 5.62,
      "learning_rate": 6.0455483286575164e-05,
      "loss": 1.9714,
      "step": 129000
    },
    {
      "epoch": 5.65,
      "learning_rate": 6.0298539383798616e-05,
      "loss": 1.9736,
      "step": 129500
    },
    {
      "epoch": 5.67,
      "learning_rate": 6.014128096418282e-05,
      "loss": 1.9677,
      "step": 130000
    },
    {
      "epoch": 5.69,
      "learning_rate": 5.9984022544567044e-05,
      "loss": 1.9626,
      "step": 130500
    },
    {
      "epoch": 5.71,
      "learning_rate": 5.9826764124951254e-05,
      "loss": 1.966,
      "step": 131000
    },
    {
      "epoch": 5.73,
      "learning_rate": 5.96698202221747e-05,
      "loss": 1.9626,
      "step": 131500
    },
    {
      "epoch": 5.75,
      "learning_rate": 5.951256180255891e-05,
      "loss": 1.9652,
      "step": 132000
    },
    {
      "epoch": 5.78,
      "learning_rate": 5.935530338294313e-05,
      "loss": 1.9652,
      "step": 132500
    },
    {
      "epoch": 5.8,
      "learning_rate": 5.919804496332734e-05,
      "loss": 1.9638,
      "step": 133000
    },
    {
      "epoch": 5.82,
      "learning_rate": 5.9040786543711554e-05,
      "loss": 1.9591,
      "step": 133500
    },
    {
      "epoch": 5.84,
      "learning_rate": 5.888384264093499e-05,
      "loss": 1.9648,
      "step": 134000
    },
    {
      "epoch": 5.86,
      "learning_rate": 5.872658422131922e-05,
      "loss": 1.9617,
      "step": 134500
    },
    {
      "epoch": 5.88,
      "learning_rate": 5.856932580170342e-05,
      "loss": 1.9626,
      "step": 135000
    },
    {
      "epoch": 5.91,
      "learning_rate": 5.8412067382087644e-05,
      "loss": 1.9597,
      "step": 135500
    },
    {
      "epoch": 5.93,
      "learning_rate": 5.825512347931108e-05,
      "loss": 1.9597,
      "step": 136000
    },
    {
      "epoch": 5.95,
      "learning_rate": 5.80978650596953e-05,
      "loss": 1.962,
      "step": 136500
    },
    {
      "epoch": 5.97,
      "learning_rate": 5.794060664007951e-05,
      "loss": 1.9557,
      "step": 137000
    },
    {
      "epoch": 5.99,
      "learning_rate": 5.778334822046373e-05,
      "loss": 1.9576,
      "step": 137500
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.6267606171222161,
      "eval_loss": 1.8356587886810303,
      "eval_runtime": 358.323,
      "eval_samples_per_second": 496.625,
      "eval_steps_per_second": 31.039,
      "step": 137640
    },
    {
      "epoch": 6.02,
      "learning_rate": 5.762640431768718e-05,
      "loss": 1.9502,
      "step": 138000
    },
    {
      "epoch": 6.04,
      "learning_rate": 5.746914589807138e-05,
      "loss": 1.9521,
      "step": 138500
    },
    {
      "epoch": 6.06,
      "learning_rate": 5.7311887478455593e-05,
      "loss": 1.9533,
      "step": 139000
    },
    {
      "epoch": 6.08,
      "learning_rate": 5.715462905883982e-05,
      "loss": 1.9533,
      "step": 139500
    },
    {
      "epoch": 6.1,
      "learning_rate": 5.699768515606326e-05,
      "loss": 1.9513,
      "step": 140000
    },
    {
      "epoch": 6.12,
      "learning_rate": 5.684042673644747e-05,
      "loss": 1.9491,
      "step": 140500
    },
    {
      "epoch": 6.15,
      "learning_rate": 5.668316831683168e-05,
      "loss": 1.9514,
      "step": 141000
    },
    {
      "epoch": 6.17,
      "learning_rate": 5.65259098972159e-05,
      "loss": 1.9492,
      "step": 141500
    },
    {
      "epoch": 6.19,
      "learning_rate": 5.6368965994439346e-05,
      "loss": 1.947,
      "step": 142000
    },
    {
      "epoch": 6.21,
      "learning_rate": 5.6211707574823556e-05,
      "loss": 1.945,
      "step": 142500
    },
    {
      "epoch": 6.23,
      "learning_rate": 5.605444915520778e-05,
      "loss": 1.9505,
      "step": 143000
    },
    {
      "epoch": 6.26,
      "learning_rate": 5.5897190735591984e-05,
      "loss": 1.9488,
      "step": 143500
    },
    {
      "epoch": 6.28,
      "learning_rate": 5.5740246832815436e-05,
      "loss": 1.9451,
      "step": 144000
    },
    {
      "epoch": 6.3,
      "learning_rate": 5.5582988413199646e-05,
      "loss": 1.9416,
      "step": 144500
    },
    {
      "epoch": 6.32,
      "learning_rate": 5.542572999358386e-05,
      "loss": 1.9487,
      "step": 145000
    },
    {
      "epoch": 6.34,
      "learning_rate": 5.5268471573968074e-05,
      "loss": 1.9394,
      "step": 145500
    },
    {
      "epoch": 6.36,
      "learning_rate": 5.511152767119152e-05,
      "loss": 1.9463,
      "step": 146000
    },
    {
      "epoch": 6.39,
      "learning_rate": 5.495426925157573e-05,
      "loss": 1.947,
      "step": 146500
    },
    {
      "epoch": 6.41,
      "learning_rate": 5.4797010831959947e-05,
      "loss": 1.9424,
      "step": 147000
    },
    {
      "epoch": 6.43,
      "learning_rate": 5.463975241234416e-05,
      "loss": 1.9433,
      "step": 147500
    },
    {
      "epoch": 6.45,
      "learning_rate": 5.448280850956761e-05,
      "loss": 1.9416,
      "step": 148000
    },
    {
      "epoch": 6.47,
      "learning_rate": 5.432555008995181e-05,
      "loss": 1.9443,
      "step": 148500
    },
    {
      "epoch": 6.5,
      "learning_rate": 5.4168291670336036e-05,
      "loss": 1.9422,
      "step": 149000
    },
    {
      "epoch": 6.52,
      "learning_rate": 5.401103325072025e-05,
      "loss": 1.9421,
      "step": 149500
    },
    {
      "epoch": 6.54,
      "learning_rate": 5.385408934794369e-05,
      "loss": 1.9412,
      "step": 150000
    },
    {
      "epoch": 6.56,
      "learning_rate": 5.36968309283279e-05,
      "loss": 1.9411,
      "step": 150500
    },
    {
      "epoch": 6.58,
      "learning_rate": 5.353957250871212e-05,
      "loss": 1.9375,
      "step": 151000
    },
    {
      "epoch": 6.6,
      "learning_rate": 5.338231408909633e-05,
      "loss": 1.9399,
      "step": 151500
    },
    {
      "epoch": 6.63,
      "learning_rate": 5.3225370186319776e-05,
      "loss": 1.9344,
      "step": 152000
    },
    {
      "epoch": 6.65,
      "learning_rate": 5.3068111766703986e-05,
      "loss": 1.9419,
      "step": 152500
    },
    {
      "epoch": 6.67,
      "learning_rate": 5.291085334708821e-05,
      "loss": 1.9353,
      "step": 153000
    },
    {
      "epoch": 6.69,
      "learning_rate": 5.275359492747241e-05,
      "loss": 1.9386,
      "step": 153500
    },
    {
      "epoch": 6.71,
      "learning_rate": 5.259633650785664e-05,
      "loss": 1.9403,
      "step": 154000
    },
    {
      "epoch": 6.73,
      "learning_rate": 5.2439392605080076e-05,
      "loss": 1.9336,
      "step": 154500
    },
    {
      "epoch": 6.76,
      "learning_rate": 5.228213418546429e-05,
      "loss": 1.934,
      "step": 155000
    },
    {
      "epoch": 6.78,
      "learning_rate": 5.21248757658485e-05,
      "loss": 1.9322,
      "step": 155500
    },
    {
      "epoch": 6.8,
      "learning_rate": 5.196761734623272e-05,
      "loss": 1.9316,
      "step": 156000
    },
    {
      "epoch": 6.82,
      "learning_rate": 5.181067344345617e-05,
      "loss": 1.9319,
      "step": 156500
    },
    {
      "epoch": 6.84,
      "learning_rate": 5.1653415023840376e-05,
      "loss": 1.937,
      "step": 157000
    },
    {
      "epoch": 6.87,
      "learning_rate": 5.1496156604224586e-05,
      "loss": 1.9324,
      "step": 157500
    },
    {
      "epoch": 6.89,
      "learning_rate": 5.133889818460881e-05,
      "loss": 1.9305,
      "step": 158000
    },
    {
      "epoch": 6.91,
      "learning_rate": 5.1181954281832256e-05,
      "loss": 1.932,
      "step": 158500
    },
    {
      "epoch": 6.93,
      "learning_rate": 5.1024695862216466e-05,
      "loss": 1.9298,
      "step": 159000
    },
    {
      "epoch": 6.95,
      "learning_rate": 5.0867437442600676e-05,
      "loss": 1.9289,
      "step": 159500
    },
    {
      "epoch": 6.97,
      "learning_rate": 5.071017902298489e-05,
      "loss": 1.9263,
      "step": 160000
    },
    {
      "epoch": 7.0,
      "learning_rate": 5.055323512020834e-05,
      "loss": 1.9313,
      "step": 160500
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.631738439030596,
      "eval_loss": 1.8091248273849487,
      "eval_runtime": 359.593,
      "eval_samples_per_second": 494.871,
      "eval_steps_per_second": 30.929,
      "step": 160580
    }
  ],
  "max_steps": 321160,
  "num_train_epochs": 14,
  "total_flos": 1.2150058886378496e+18,
  "trial_name": null,
  "trial_params": null
}