End of training

all_results.json

@@ -1,13 +1,13 @@
 {
-    "epoch": 9.99,
-    "eval_accuracy": 0.7794067507671326,
-    "eval_loss": 0.45381179451942444,
-    "eval_runtime": 44.0429,
-    "eval_samples_per_second": 133.188,
-    "eval_steps_per_second": 4.178,
-    "total_flos": 1.3106102295963156e+19,
-    "train_loss": 0.3501510629954847,
-    "train_runtime": 8665.4247,
-    "train_samples_per_second": 60.918,
-    "train_steps_per_second": 0.475
+    "epoch": 3.0,
+    "eval_accuracy": 0.7774538386783285,
+    "eval_loss": 0.39165472984313965,
+    "eval_runtime": 52.9368,
+    "eval_samples_per_second": 136.068,
+    "eval_steps_per_second": 4.269,
+    "total_flos": 4.829589697691566e+18,
+    "train_loss": 0.5911478077470078,
+    "train_runtime": 3253.5777,
+    "train_samples_per_second": 59.774,
+    "train_steps_per_second": 0.467
 }

eval_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 9.99,
-    "eval_accuracy": 0.7794067507671326,
-    "eval_loss": 0.45381179451942444,
-    "eval_runtime": 44.0429,
-    "eval_samples_per_second": 133.188,
-    "eval_steps_per_second": 4.178
+    "epoch": 3.0,
+    "eval_accuracy": 0.7774538386783285,
+    "eval_loss": 0.39165472984313965,
+    "eval_runtime": 52.9368,
+    "eval_samples_per_second": 136.068,
+    "eval_steps_per_second": 4.269
 }

train_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 9.99,
-    "total_flos": 1.3106102295963156e+19,
-    "train_loss": 0.3501510629954847,
-    "train_runtime": 8665.4247,
-    "train_samples_per_second": 60.918,
-    "train_steps_per_second": 0.475
+    "epoch": 3.0,
+    "total_flos": 4.829589697691566e+18,
+    "train_loss": 0.5911478077470078,
+    "train_runtime": 3253.5777,
+    "train_samples_per_second": 59.774,
+    "train_steps_per_second": 0.467
 }

trainer_state.json

@@ -1,2590 +1,1873 @@
 {
-  "best_metric": 0.7794067507671326,
-  "best_model_checkpoint": "swin-tiny-patch4-window7-224-img_orientation/checkpoint-4120",
-  "epoch": 9.987878787878788,
+  "best_metric": 0.7774538386783285,
+  "best_model_checkpoint": "swin-tiny-patch4-window7-224-img_orientation/checkpoint-1518",
+  "epoch": 2.9970384995064165,
   "eval_steps": 500,
-  "global_step": 4120,
+  "global_step": 1518,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.02,
-      "learning_rate": 1.2135922330097087e-06,
-      "loss": 0.4514,
-      "step": 10
-    },
-    {
-      "epoch": 0.05,
-      "learning_rate": 2.4271844660194174e-06,
-      "loss": 0.4198,
-      "step": 20
-    },
-    {
-      "epoch": 0.07,
-      "learning_rate": 3.6407766990291263e-06,
-      "loss": 0.4392,
-      "step": 30
-    },
-    {
-      "epoch": 0.1,
-      "learning_rate": 4.854368932038835e-06,
-      "loss": 0.4007,
-      "step": 40
-    },
-    {
-      "epoch": 0.12,
-      "learning_rate": 6.067961165048544e-06,
-      "loss": 0.4043,
-      "step": 50
-    },
-    {
-      "epoch": 0.15,
-      "learning_rate": 7.281553398058253e-06,
-      "loss": 0.3604,
-      "step": 60
-    },
-    {
-      "epoch": 0.17,
-      "learning_rate": 8.495145631067963e-06,
-      "loss": 0.383,
-      "step": 70
-    },
-    {
-      "epoch": 0.19,
-      "learning_rate": 9.70873786407767e-06,
-      "loss": 0.3708,
-      "step": 80
-    },
-    {
-      "epoch": 0.22,
-      "learning_rate": 1.0922330097087378e-05,
-      "loss": 0.3486,
-      "step": 90
-    },
-    {
-      "epoch": 0.24,
-      "learning_rate": 1.2135922330097088e-05,
-      "loss": 0.367,
-      "step": 100
-    },
-    {
-      "epoch": 0.27,
-      "learning_rate": 1.3349514563106797e-05,
-      "loss": 0.3288,
-      "step": 110
-    },
-    {
-      "epoch": 0.29,
-      "learning_rate": 1.4563106796116505e-05,
-      "loss": 0.3155,
-      "step": 120
-    },
-    {
-      "epoch": 0.32,
-      "learning_rate": 1.5776699029126215e-05,
-      "loss": 0.3364,
-      "step": 130
-    },
-    {
-      "epoch": 0.34,
-      "learning_rate": 1.6990291262135926e-05,
-      "loss": 0.3258,
-      "step": 140
-    },
-    {
-      "epoch": 0.36,
-      "learning_rate": 1.8203883495145632e-05,
-      "loss": 0.3078,
-      "step": 150
-    },
-    {
-      "epoch": 0.39,
-      "learning_rate": 1.941747572815534e-05,
-      "loss": 0.293,
-      "step": 160
-    },
-    {
-      "epoch": 0.41,
-      "learning_rate": 2.063106796116505e-05,
-      "loss": 0.2877,
-      "step": 170
-    },
-    {
-      "epoch": 0.44,
-      "learning_rate": 2.1844660194174756e-05,
-      "loss": 0.3103,
-      "step": 180
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 2.305825242718447e-05,
-      "loss": 0.2944,
-      "step": 190
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 2.4271844660194176e-05,
-      "loss": 0.2986,
-      "step": 200
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 2.5485436893203883e-05,
-      "loss": 0.2752,
-      "step": 210
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 2.6699029126213593e-05,
-      "loss": 0.3086,
-      "step": 220
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 2.79126213592233e-05,
-      "loss": 0.313,
-      "step": 230
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 2.912621359223301e-05,
-      "loss": 0.2864,
-      "step": 240
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 3.0339805825242717e-05,
-      "loss": 0.3166,
-      "step": 250
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 3.155339805825243e-05,
-      "loss": 0.292,
-      "step": 260
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 3.276699029126214e-05,
-      "loss": 0.3025,
-      "step": 270
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 3.398058252427185e-05,
-      "loss": 0.2801,
-      "step": 280
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 3.519417475728156e-05,
-      "loss": 0.2705,
-      "step": 290
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 3.6407766990291265e-05,
-      "loss": 0.2895,
-      "step": 300
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 3.762135922330097e-05,
-      "loss": 0.3195,
-      "step": 310
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 3.883495145631068e-05,
-      "loss": 0.3023,
-      "step": 320
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 4.004854368932039e-05,
-      "loss": 0.3075,
-      "step": 330
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 4.12621359223301e-05,
-      "loss": 0.3065,
-      "step": 340
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 4.2475728155339805e-05,
-      "loss": 0.2892,
-      "step": 350
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 4.368932038834951e-05,
-      "loss": 0.2819,
-      "step": 360
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 4.4902912621359226e-05,
-      "loss": 0.35,
-      "step": 370
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 4.611650485436894e-05,
-      "loss": 0.3134,
-      "step": 380
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 4.7330097087378646e-05,
-      "loss": 0.3159,
-      "step": 390
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 4.854368932038835e-05,
-      "loss": 0.3149,
-      "step": 400
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 4.975728155339806e-05,
-      "loss": 0.3217,
-      "step": 410
-    },
-    {
-      "epoch": 1.0,
-      "eval_accuracy": 0.7533242413910671,
-      "eval_loss": 0.4921741783618927,
-      "eval_runtime": 44.32,
-      "eval_samples_per_second": 132.356,
-      "eval_steps_per_second": 4.152,
-      "step": 412
-    },
-    {
-      "epoch": 1.02,
-      "learning_rate": 4.9892125134843584e-05,
-      "loss": 0.2984,
-      "step": 420
-    },
-    {
-      "epoch": 1.04,
-      "learning_rate": 4.975728155339806e-05,
-      "loss": 0.2742,
-      "step": 430
-    },
-    {
-      "epoch": 1.07,
-      "learning_rate": 4.962243797195254e-05,
-      "loss": 0.2608,
-      "step": 440
-    },
-    {
-      "epoch": 1.09,
-      "learning_rate": 4.948759439050701e-05,
-      "loss": 0.2663,
-      "step": 450
-    },
-    {
-      "epoch": 1.12,
-      "learning_rate": 4.935275080906149e-05,
-      "loss": 0.3096,
-      "step": 460
-    },
-    {
-      "epoch": 1.14,
-      "learning_rate": 4.921790722761597e-05,
-      "loss": 0.3046,
-      "step": 470
-    },
-    {
-      "epoch": 1.16,
-      "learning_rate": 4.9083063646170444e-05,
-      "loss": 0.3038,
-      "step": 480
-    },
-    {
-      "epoch": 1.19,
-      "learning_rate": 4.894822006472492e-05,
-      "loss": 0.2892,
-      "step": 490
-    },
-    {
-      "epoch": 1.21,
-      "learning_rate": 4.88133764832794e-05,
-      "loss": 0.3028,
-      "step": 500
-    },
-    {
-      "epoch": 1.24,
-      "learning_rate": 4.867853290183387e-05,
-      "loss": 0.3175,
-      "step": 510
-    },
-    {
-      "epoch": 1.26,
-      "learning_rate": 4.854368932038835e-05,
-      "loss": 0.2881,
-      "step": 520
-    },
-    {
-      "epoch": 1.28,
-      "learning_rate": 4.840884573894283e-05,
-      "loss": 0.3002,
-      "step": 530
-    },
-    {
-      "epoch": 1.31,
-      "learning_rate": 4.8274002157497304e-05,
-      "loss": 0.297,
-      "step": 540
-    },
-    {
-      "epoch": 1.33,
-      "learning_rate": 4.8139158576051786e-05,
-      "loss": 0.3127,
-      "step": 550
-    },
-    {
-      "epoch": 1.36,
-      "learning_rate": 4.800431499460626e-05,
-      "loss": 0.2924,
-      "step": 560
-    },
-    {
-      "epoch": 1.38,
-      "learning_rate": 4.786947141316074e-05,
-      "loss": 0.2823,
-      "step": 570
-    },
-    {
-      "epoch": 1.41,
-      "learning_rate": 4.773462783171521e-05,
-      "loss": 0.2947,
-      "step": 580
-    },
-    {
-      "epoch": 1.43,
-      "learning_rate": 4.759978425026969e-05,
-      "loss": 0.3045,
-      "step": 590
-    },
-    {
-      "epoch": 1.45,
-      "learning_rate": 4.7464940668824164e-05,
-      "loss": 0.3217,
-      "step": 600
-    },
-    {
-      "epoch": 1.48,
-      "learning_rate": 4.7330097087378646e-05,
-      "loss": 0.2629,
-      "step": 610
-    },
-    {
-      "epoch": 1.5,
-      "learning_rate": 4.7195253505933115e-05,
-      "loss": 0.2978,
-      "step": 620
-    },
-    {
-      "epoch": 1.53,
-      "learning_rate": 4.70604099244876e-05,
-      "loss": 0.3395,
-      "step": 630
-    },
-    {
-      "epoch": 1.55,
-      "learning_rate": 4.692556634304207e-05,
-      "loss": 0.3125,
-      "step": 640
-    },
-    {
-      "epoch": 1.58,
-      "learning_rate": 4.679072276159655e-05,
-      "loss": 0.3282,
-      "step": 650
-    },
-    {
-      "epoch": 1.6,
-      "learning_rate": 4.665587918015103e-05,
-      "loss": 0.3308,
-      "step": 660
-    },
-    {
-      "epoch": 1.62,
-      "learning_rate": 4.6521035598705506e-05,
-      "loss": 0.3026,
-      "step": 670
-    },
-    {
-      "epoch": 1.65,
-      "learning_rate": 4.638619201725998e-05,
-      "loss": 0.3007,
-      "step": 680
-    },
-    {
-      "epoch": 1.67,
-      "learning_rate": 4.625134843581446e-05,
-      "loss": 0.3106,
-      "step": 690
-    },
-    {
-      "epoch": 1.7,
-      "learning_rate": 4.611650485436894e-05,
-      "loss": 0.3172,
-      "step": 700
+      "epoch": 0.01,
+      "learning_rate": 1.6447368421052632e-06,
+      "loss": 1.4583,
+      "step": 5
     },
     {
-      "epoch": 1.72,
-      "learning_rate": 4.598166127292341e-05,
-      "loss": 0.3423,
-      "step": 710
-    },
-    {
-      "epoch": 1.75,
-      "learning_rate": 4.584681769147789e-05,
-      "loss": 0.3383,
-      "step": 720
-    },
-    {
-      "epoch": 1.77,
-      "learning_rate": 4.5711974110032366e-05,
-      "loss": 0.3006,
-      "step": 730
-    },
-    {
-      "epoch": 1.79,
-      "learning_rate": 4.557713052858684e-05,
-      "loss": 0.3038,
-      "step": 740
-    },
-    {
-      "epoch": 1.82,
-      "learning_rate": 4.544228694714132e-05,
-      "loss": 0.3614,
-      "step": 750
-    },
-    {
-      "epoch": 1.84,
-      "learning_rate": 4.530744336569579e-05,
-      "loss": 0.3106,
-      "step": 760
-    },
-    {
-      "epoch": 1.87,
-      "learning_rate": 4.5172599784250275e-05,
-      "loss": 0.327,
-      "step": 770
-    },
-    {
-      "epoch": 1.89,
-      "learning_rate": 4.503775620280475e-05,
-      "loss": 0.3241,
-      "step": 780
-    },
-    {
-      "epoch": 1.92,
-      "learning_rate": 4.4902912621359226e-05,
-      "loss": 0.3476,
-      "step": 790
-    },
-    {
-      "epoch": 1.94,
-      "learning_rate": 4.47680690399137e-05,
-      "loss": 0.3401,
-      "step": 800
-    },
-    {
-      "epoch": 1.96,
-      "learning_rate": 4.4633225458468184e-05,
-      "loss": 0.3372,
-      "step": 810
-    },
-    {
-      "epoch": 1.99,
-      "learning_rate": 4.449838187702265e-05,
-      "loss": 0.3716,
-      "step": 820
-    },
-    {
-      "epoch": 2.0,
-      "eval_accuracy": 0.7601431980906921,
-      "eval_loss": 0.49858367443084717,
-      "eval_runtime": 44.1192,
-      "eval_samples_per_second": 132.958,
-      "eval_steps_per_second": 4.171,
-      "step": 825
-    },
-    {
-      "epoch": 2.01,
-      "learning_rate": 4.4363538295577135e-05,
-      "loss": 0.3225,
-      "step": 830
-    },
-    {
-      "epoch": 2.04,
-      "learning_rate": 4.422869471413161e-05,
-      "loss": 0.307,
-      "step": 840
-    },
-    {
-      "epoch": 2.06,
-      "learning_rate": 4.4093851132686086e-05,
-      "loss": 0.3115,
-      "step": 850
-    },
-    {
-      "epoch": 2.08,
-      "learning_rate": 4.395900755124056e-05,
-      "loss": 0.2849,
-      "step": 860
-    },
-    {
-      "epoch": 2.11,
-      "learning_rate": 4.3824163969795043e-05,
-      "loss": 0.3344,
-      "step": 870
-    },
-    {
-      "epoch": 2.13,
-      "learning_rate": 4.368932038834951e-05,
-      "loss": 0.3153,
-      "step": 880
-    },
-    {
-      "epoch": 2.16,
-      "learning_rate": 4.3554476806903994e-05,
-      "loss": 0.3125,
-      "step": 890
-    },
-    {
-      "epoch": 2.18,
-      "learning_rate": 4.341963322545847e-05,
-      "loss": 0.2922,
-      "step": 900
-    },
-    {
-      "epoch": 2.21,
-      "learning_rate": 4.3284789644012946e-05,
-      "loss": 0.3061,
-      "step": 910
-    },
-    {
-      "epoch": 2.23,
-      "learning_rate": 4.314994606256743e-05,
-      "loss": 0.3327,
-      "step": 920
-    },
-    {
-      "epoch": 2.25,
-      "learning_rate": 4.3015102481121897e-05,
-      "loss": 0.3186,
-      "step": 930
-    },
-    {
-      "epoch": 2.28,
-      "learning_rate": 4.288025889967638e-05,
-      "loss": 0.3247,
-      "step": 940
-    },
-    {
-      "epoch": 2.3,
-      "learning_rate": 4.2745415318230854e-05,
-      "loss": 0.3278,
-      "step": 950
-    },
-    {
-      "epoch": 2.33,
-      "learning_rate": 4.261057173678533e-05,
-      "loss": 0.3081,
-      "step": 960
-    },
-    {
-      "epoch": 2.35,
-      "learning_rate": 4.2475728155339805e-05,
-      "loss": 0.2979,
-      "step": 970
-    },
-    {
-      "epoch": 2.38,
-      "learning_rate": 4.234088457389429e-05,
-      "loss": 0.3394,
-      "step": 980
-    },
-    {
-      "epoch": 2.4,
-      "learning_rate": 4.2206040992448756e-05,
-      "loss": 0.3274,
-      "step": 990
-    },
-    {
-      "epoch": 2.42,
-      "learning_rate": 4.207119741100324e-05,
-      "loss": 0.3028,
-      "step": 1000
-    },
-    {
-      "epoch": 2.45,
-      "learning_rate": 4.1936353829557714e-05,
-      "loss": 0.3234,
-      "step": 1010
-    },
-    {
-      "epoch": 2.47,
-      "learning_rate": 4.180151024811219e-05,
-      "loss": 0.3177,
-      "step": 1020
-    },
-    {
-      "epoch": 2.5,
-      "learning_rate": 4.166666666666667e-05,
-      "loss": 0.3377,
-      "step": 1030
-    },
-    {
-      "epoch": 2.52,
-      "learning_rate": 4.153182308522115e-05,
-      "loss": 0.351,
-      "step": 1040
-    },
-    {
-      "epoch": 2.55,
-      "learning_rate": 4.139697950377562e-05,
-      "loss": 0.3438,
-      "step": 1050
-    },
-    {
-      "epoch": 2.57,
-      "learning_rate": 4.12621359223301e-05,
-      "loss": 0.3125,
-      "step": 1060
-    },
-    {
-      "epoch": 2.59,
-      "learning_rate": 4.1127292340884574e-05,
-      "loss": 0.3392,
-      "step": 1070
-    },
-    {
-      "epoch": 2.62,
-      "learning_rate": 4.099244875943905e-05,
-      "loss": 0.333,
-      "step": 1080
-    },
-    {
-      "epoch": 2.64,
-      "learning_rate": 4.085760517799353e-05,
-      "loss": 0.3557,
-      "step": 1090
-    },
-    {
-      "epoch": 2.67,
-      "learning_rate": 4.0722761596548e-05,
-      "loss": 0.3162,
-      "step": 1100
-    },
-    {
-      "epoch": 2.69,
-      "learning_rate": 4.058791801510248e-05,
-      "loss": 0.3393,
-      "step": 1110
-    },
-    {
-      "epoch": 2.72,
-      "learning_rate": 4.0453074433656965e-05,
-      "loss": 0.3408,
-      "step": 1120
-    },
-    {
-      "epoch": 2.74,
-      "learning_rate": 4.0318230852211434e-05,
-      "loss": 0.3223,
-      "step": 1130
-    },
-    {
-      "epoch": 2.76,
-      "learning_rate": 4.0183387270765916e-05,
-      "loss": 0.3516,
-      "step": 1140
-    },
-    {
-      "epoch": 2.79,
-      "learning_rate": 4.004854368932039e-05,
-      "loss": 0.348,
-      "step": 1150
-    },
-    {
-      "epoch": 2.81,
-      "learning_rate": 3.991370010787487e-05,
-      "loss": 0.3358,
-      "step": 1160
+      "epoch": 0.02,
+      "learning_rate": 3.2894736842105265e-06,
+      "loss": 1.4742,
+      "step": 10
     },
     {
-      "epoch": 2.84,
-      "learning_rate": 3.977885652642934e-05,
-      "loss": 0.3781,
-      "step": 1170
+      "epoch": 0.03,
+      "learning_rate": 4.9342105263157895e-06,
+      "loss": 1.4362,
+      "step": 15
     },
     {
-      "epoch": 2.86,
-      "learning_rate": 3.964401294498382e-05,
-      "loss": 0.358,
-      "step": 1180
+      "epoch": 0.04,
+      "learning_rate": 6.578947368421053e-06,
+      "loss": 1.458,
+      "step": 20
     },
     {
-      "epoch": 2.88,
-      "learning_rate": 3.9509169363538294e-05,
-      "loss": 0.3959,
-      "step": 1190
+      "epoch": 0.05,
+      "learning_rate": 8.223684210526317e-06,
+      "loss": 1.3734,
+      "step": 25
     },
     {
-      "epoch": 2.91,
-      "learning_rate": 3.9374325782092776e-05,
-      "loss": 0.4092,
-      "step": 1200
+      "epoch": 0.06,
+      "learning_rate": 9.868421052631579e-06,
+      "loss": 1.3655,
+      "step": 30
     },
     {
-      "epoch": 2.93,
-      "learning_rate": 3.923948220064725e-05,
-      "loss": 0.3731,
-      "step": 1210
+      "epoch": 0.07,
+      "learning_rate": 1.1513157894736843e-05,
+      "loss": 1.3442,
+      "step": 35
     },
     {
-      "epoch": 2.96,
-      "learning_rate": 3.910463861920173e-05,
-      "loss": 0.388,
-      "step": 1220
+      "epoch": 0.08,
+      "learning_rate": 1.3157894736842106e-05,
+      "loss": 1.3333,
+      "step": 40
     },
     {
-      "epoch": 2.98,
-      "learning_rate": 3.896979503775621e-05,
-      "loss": 0.3756,
-      "step": 1230
+      "epoch": 0.09,
+      "learning_rate": 1.4802631578947368e-05,
+      "loss": 1.2669,
+      "step": 45
     },
     {
-      "epoch": 3.0,
-      "eval_accuracy": 0.7761677463348108,
-      "eval_loss": 0.44274240732192993,
-      "eval_runtime": 43.5681,
-      "eval_samples_per_second": 134.64,
-      "eval_steps_per_second": 4.223,
-      "step": 1237
+      "epoch": 0.1,
+      "learning_rate": 1.6447368421052635e-05,
+      "loss": 1.2279,
+      "step": 50
     },
     {
-      "epoch": 3.01,
-      "learning_rate": 3.883495145631068e-05,
-      "loss": 0.3399,
-      "step": 1240
+      "epoch": 0.11,
+      "learning_rate": 1.8092105263157896e-05,
+      "loss": 1.176,
+      "step": 55
     },
     {
-      "epoch": 3.03,
-      "learning_rate": 3.870010787486516e-05,
-      "loss": 0.3321,
-      "step": 1250
+      "epoch": 0.12,
+      "learning_rate": 1.9736842105263158e-05,
+      "loss": 1.1198,
+      "step": 60
     },
     {
-      "epoch": 3.05,
-      "learning_rate": 3.8565264293419636e-05,
-      "loss": 0.3665,
-      "step": 1260
+      "epoch": 0.13,
+      "learning_rate": 2.1381578947368423e-05,
+      "loss": 1.0456,
+      "step": 65
     },
     {
-      "epoch": 3.08,
-      "learning_rate": 3.843042071197411e-05,
-      "loss": 0.3489,
-      "step": 1270
+      "epoch": 0.14,
+      "learning_rate": 2.3026315789473685e-05,
+      "loss": 0.9941,
+      "step": 70
     },
     {
-      "epoch": 3.1,
-      "learning_rate": 3.829557713052859e-05,
-      "loss": 0.3717,
-      "step": 1280
+      "epoch": 0.15,
+      "learning_rate": 2.4671052631578947e-05,
+      "loss": 0.991,
+      "step": 75
     },
     {
-      "epoch": 3.13,
-      "learning_rate": 3.816073354908307e-05,
-      "loss": 0.363,
-      "step": 1290
+      "epoch": 0.16,
+      "learning_rate": 2.6315789473684212e-05,
+      "loss": 0.9243,
+      "step": 80
     },
     {
-      "epoch": 3.15,
-      "learning_rate": 3.802588996763754e-05,
-      "loss": 0.3485,
-      "step": 1300
+      "epoch": 0.17,
+      "learning_rate": 2.7960526315789477e-05,
+      "loss": 0.9057,
+      "step": 85
     },
     {
-      "epoch": 3.18,
-      "learning_rate": 3.789104638619202e-05,
-      "loss": 0.4041,
-      "step": 1310
+      "epoch": 0.18,
+      "learning_rate": 2.9605263157894735e-05,
+      "loss": 0.8941,
+      "step": 90
     },
     {
-      "epoch": 3.2,
-      "learning_rate": 3.7756202804746496e-05,
-      "loss": 0.3738,
-      "step": 1320
+      "epoch": 0.19,
+      "learning_rate": 3.125e-05,
+      "loss": 0.8725,
+      "step": 95
     },
     {
-      "epoch": 3.22,
-      "learning_rate": 3.762135922330097e-05,
-      "loss": 0.3573,
-      "step": 1330
+      "epoch": 0.2,
+      "learning_rate": 3.289473684210527e-05,
+      "loss": 0.8545,
+      "step": 100
     },
     {
-      "epoch": 3.25,
-      "learning_rate": 3.7486515641855454e-05,
-      "loss": 0.3255,
-      "step": 1340
+      "epoch": 0.21,
+      "learning_rate": 3.4539473684210524e-05,
+      "loss": 0.8274,
+      "step": 105
     },
     {
-      "epoch": 3.27,
-      "learning_rate": 3.735167206040992e-05,
-      "loss": 0.3402,
-      "step": 1350
+      "epoch": 0.22,
+      "learning_rate": 3.618421052631579e-05,
+      "loss": 0.818,
+      "step": 110
     },
     {
-      "epoch": 3.3,
-      "learning_rate": 3.7216828478964405e-05,
-      "loss": 0.3754,
-      "step": 1360
+      "epoch": 0.23,
+      "learning_rate": 3.7828947368421054e-05,
+      "loss": 0.8329,
+      "step": 115
     },
     {
-      "epoch": 3.32,
-      "learning_rate": 3.708198489751888e-05,
-      "loss": 0.365,
-      "step": 1370
+      "epoch": 0.24,
+      "learning_rate": 3.9473684210526316e-05,
+      "loss": 0.8005,
+      "step": 120
     },
     {
-      "epoch": 3.35,
-      "learning_rate": 3.6947141316073356e-05,
-      "loss": 0.3738,
-      "step": 1380
+      "epoch": 0.25,
+      "learning_rate": 4.111842105263158e-05,
+      "loss": 0.7636,
+      "step": 125
     },
     {
-      "epoch": 3.37,
-      "learning_rate": 3.681229773462783e-05,
-      "loss": 0.3628,
-      "step": 1390
+      "epoch": 0.26,
+      "learning_rate": 4.2763157894736847e-05,
+      "loss": 0.8036,
+      "step": 130
     },
     {
-      "epoch": 3.39,
-      "learning_rate": 3.6677454153182314e-05,
-      "loss": 0.3674,
-      "step": 1400
+      "epoch": 0.27,
+      "learning_rate": 4.440789473684211e-05,
+      "loss": 0.7663,
+      "step": 135
     },
     {
-      "epoch": 3.42,
-      "learning_rate": 3.654261057173678e-05,
-      "loss": 0.3268,
-      "step": 1410
+      "epoch": 0.28,
+      "learning_rate": 4.605263157894737e-05,
+      "loss": 0.7535,
+      "step": 140
     },
     {
-      "epoch": 3.44,
-      "learning_rate": 3.6407766990291265e-05,
-      "loss": 0.3433,
-      "step": 1420
+      "epoch": 0.29,
+      "learning_rate": 4.769736842105263e-05,
+      "loss": 0.7418,
+      "step": 145
     },
     {
-      "epoch": 3.47,
-      "learning_rate": 3.627292340884574e-05,
-      "loss": 0.3665,
-      "step": 1430
+      "epoch": 0.3,
+      "learning_rate": 4.9342105263157894e-05,
+      "loss": 0.7471,
+      "step": 150
     },
     {
-      "epoch": 3.49,
-      "learning_rate": 3.6138079827400216e-05,
-      "loss": 0.3512,
-      "step": 1440
+      "epoch": 0.31,
+      "learning_rate": 4.9890190336749635e-05,
+      "loss": 0.7603,
+      "step": 155
     },
     {
-      "epoch": 3.52,
-      "learning_rate": 3.60032362459547e-05,
-      "loss": 0.3571,
-      "step": 1450
+      "epoch": 0.32,
+      "learning_rate": 4.970717423133236e-05,
+      "loss": 0.7101,
+      "step": 160
     },
     {
-      "epoch": 3.54,
-      "learning_rate": 3.5868392664509174e-05,
-      "loss": 0.3808,
-      "step": 1460
+      "epoch": 0.33,
+      "learning_rate": 4.952415812591508e-05,
+      "loss": 0.6832,
+      "step": 165
     },
     {
-      "epoch": 3.56,
-      "learning_rate": 3.573354908306365e-05,
-      "loss": 0.3969,
-      "step": 1470
+      "epoch": 0.34,
+      "learning_rate": 4.934114202049781e-05,
+      "loss": 0.6959,
+      "step": 170
     },
     {
-      "epoch": 3.59,
-      "learning_rate": 3.5598705501618125e-05,
-      "loss": 0.3663,
-      "step": 1480
+      "epoch": 0.35,
+      "learning_rate": 4.9158125915080526e-05,
+      "loss": 0.7345,
+      "step": 175
     },
     {
-      "epoch": 3.61,
-      "learning_rate": 3.54638619201726e-05,
-      "loss": 0.3562,
-      "step": 1490
+      "epoch": 0.36,
+      "learning_rate": 4.897510980966325e-05,
+      "loss": 0.7131,
+      "step": 180
     },
     {
-      "epoch": 3.64,
-      "learning_rate": 3.5329018338727076e-05,
-      "loss": 0.3686,
-      "step": 1500
+      "epoch": 0.37,
+      "learning_rate": 4.879209370424598e-05,
+      "loss": 0.7713,
+      "step": 185
     },
     {
-      "epoch": 3.66,
-      "learning_rate": 3.519417475728156e-05,
-      "loss": 0.3815,
-      "step": 1510
+      "epoch": 0.38,
+      "learning_rate": 4.86090775988287e-05,
+      "loss": 0.6945,
+      "step": 190
     },
     {
-      "epoch": 3.68,
-      "learning_rate": 3.505933117583603e-05,
-      "loss": 0.3892,
-      "step": 1520
+      "epoch": 0.38,
+      "learning_rate": 4.8426061493411424e-05,
+      "loss": 0.6644,
+      "step": 195
     },
     {
-      "epoch": 3.71,
-      "learning_rate": 3.492448759439051e-05,
-      "loss": 0.3847,
-      "step": 1530
+      "epoch": 0.39,
+      "learning_rate": 4.824304538799414e-05,
+      "loss": 0.679,
+      "step": 200
     },
     {
-      "epoch": 3.73,
-      "learning_rate": 3.4789644012944984e-05,
-      "loss": 0.3849,
-      "step": 1540
+      "epoch": 0.4,
+      "learning_rate": 4.806002928257687e-05,
+      "loss": 0.6657,
+      "step": 205
     },
     {
-      "epoch": 3.76,
-      "learning_rate": 3.465480043149946e-05,
-      "loss": 0.3964,
-      "step": 1550
+      "epoch": 0.41,
+      "learning_rate": 4.787701317715959e-05,
+      "loss": 0.685,
+      "step": 210
     },
     {
-      "epoch": 3.78,
-      "learning_rate": 3.451995685005394e-05,
-      "loss": 0.4121,
-      "step": 1560
+      "epoch": 0.42,
+      "learning_rate": 4.7693997071742315e-05,
+      "loss": 0.6931,
+      "step": 215
     },
     {
-      "epoch": 3.81,
-      "learning_rate": 3.438511326860842e-05,
-      "loss": 0.3863,
-      "step": 1570
+      "epoch": 0.43,
+      "learning_rate": 4.751098096632504e-05,
+      "loss": 0.6803,
+      "step": 220
     },
     {
-      "epoch": 3.83,
-      "learning_rate": 3.425026968716289e-05,
-      "loss": 0.3724,
-      "step": 1580
+      "epoch": 0.44,
+      "learning_rate": 4.732796486090777e-05,
+      "loss": 0.6415,
+      "step": 225
     },
     {
-      "epoch": 3.85,
-      "learning_rate": 3.411542610571737e-05,
-      "loss": 0.3631,
-      "step": 1590
+      "epoch": 0.45,
+      "learning_rate": 4.7144948755490486e-05,
+      "loss": 0.6753,
+      "step": 230
     },
     {
-      "epoch": 3.88,
-      "learning_rate": 3.398058252427185e-05,
-      "loss": 0.3825,
-      "step": 1600
+      "epoch": 0.46,
+      "learning_rate": 4.696193265007321e-05,
+      "loss": 0.6676,
+      "step": 235
     },
     {
-      "epoch": 3.9,
-      "learning_rate": 3.384573894282632e-05,
-      "loss": 0.4403,
-      "step": 1610
+      "epoch": 0.47,
+      "learning_rate": 4.677891654465593e-05,
+      "loss": 0.6445,
+      "step": 240
     },
     {
-      "epoch": 3.93,
-      "learning_rate": 3.37108953613808e-05,
-      "loss": 0.4074,
-      "step": 1620
+      "epoch": 0.48,
+      "learning_rate": 4.659590043923866e-05,
+      "loss": 0.7077,
+      "step": 245
     },
     {
-      "epoch": 3.95,
-      "learning_rate": 3.357605177993528e-05,
-      "loss": 0.4067,
-      "step": 1630
+      "epoch": 0.49,
+      "learning_rate": 4.641288433382138e-05,
+      "loss": 0.6417,
+      "step": 250
     },
     {
-      "epoch": 3.98,
-      "learning_rate": 3.344120819848975e-05,
-      "loss": 0.3598,
-      "step": 1640
+      "epoch": 0.5,
+      "learning_rate": 4.6229868228404096e-05,
+      "loss": 0.6383,
+      "step": 255
     },
     {
-      "epoch": 4.0,
-      "learning_rate": 3.330636461704423e-05,
-      "loss": 0.3926,
-      "step": 1650
+      "epoch": 0.51,
+      "learning_rate": 4.604685212298683e-05,
+      "loss": 0.6662,
+      "step": 260
     },
     {
-      "epoch": 4.0,
-      "eval_accuracy": 0.7732696897374701,
-      "eval_loss": 0.4515363872051239,
-      "eval_runtime": 43.3999,
-      "eval_samples_per_second": 135.162,
-      "eval_steps_per_second": 4.24,
-      "step": 1650
+      "epoch": 0.52,
+      "learning_rate": 4.586383601756955e-05,
+      "loss": 0.6741,
+      "step": 265
     },
     {
-      "epoch": 4.02,
-      "learning_rate": 3.3171521035598704e-05,
-      "loss": 0.3523,
-      "step": 1660
+      "epoch": 0.53,
+      "learning_rate": 4.5680819912152275e-05,
+      "loss": 0.6429,
+      "step": 270
     },
     {
-      "epoch": 4.05,
-      "learning_rate": 3.3036677454153186e-05,
-      "loss": 0.3657,
-      "step": 1670
+      "epoch": 0.54,
+      "learning_rate": 4.5497803806734994e-05,
+      "loss": 0.6361,
+      "step": 275
     },
     {
-      "epoch": 4.07,
-      "learning_rate": 3.290183387270766e-05,
-      "loss": 0.3677,
-      "step": 1680
+      "epoch": 0.55,
+      "learning_rate": 4.531478770131772e-05,
+      "loss": 0.6221,
+      "step": 280
     },
     {
-      "epoch": 4.1,
-      "learning_rate": 3.276699029126214e-05,
-      "loss": 0.4004,
-      "step": 1690
+      "epoch": 0.56,
+      "learning_rate": 4.513177159590044e-05,
+      "loss": 0.64,
+      "step": 285
     },
     {
-      "epoch": 4.12,
-      "learning_rate": 3.263214670981661e-05,
-      "loss": 0.3857,
-      "step": 1700
+      "epoch": 0.57,
+      "learning_rate": 4.4948755490483165e-05,
+      "loss": 0.6432,
+      "step": 290
     },
     {
-      "epoch": 4.15,
-      "learning_rate": 3.2497303128371095e-05,
-      "loss": 0.4001,
-      "step": 1710
+      "epoch": 0.58,
+      "learning_rate": 4.4765739385065885e-05,
+      "loss": 0.6621,
+      "step": 295
     },
     {
-      "epoch": 4.17,
-      "learning_rate": 3.2362459546925564e-05,
-      "loss": 0.3483,
-      "step": 1720
+      "epoch": 0.59,
+      "learning_rate": 4.458272327964861e-05,
+      "loss": 0.5987,
+      "step": 300
     },
     {
-      "epoch": 4.19,
-      "learning_rate": 3.2227615965480046e-05,
-      "loss": 0.3868,
-      "step": 1730
+      "epoch": 0.6,
+      "learning_rate": 4.439970717423134e-05,
+      "loss": 0.6368,
+      "step": 305
     },
     {
-      "epoch": 4.22,
-      "learning_rate": 3.209277238403452e-05,
-      "loss": 0.3811,
-      "step": 1740
+      "epoch": 0.61,
+      "learning_rate": 4.4216691068814056e-05,
+      "loss": 0.6285,
+      "step": 310
     },
     {
-      "epoch": 4.24,
-      "learning_rate": 3.1957928802589e-05,
-      "loss": 0.344,
-      "step": 1750
+      "epoch": 0.62,
+      "learning_rate": 4.403367496339678e-05,
+      "loss": 0.645,
+      "step": 315
     },
     {
-      "epoch": 4.27,
-      "learning_rate": 3.182308522114347e-05,
-      "loss": 0.4038,
-      "step": 1760
+      "epoch": 0.63,
+      "learning_rate": 4.38506588579795e-05,
+      "loss": 0.6416,
+      "step": 320
     },
     {
-      "epoch": 4.29,
-      "learning_rate": 3.1688241639697955e-05,
-      "loss": 0.3716,
-      "step": 1770
+      "epoch": 0.64,
+      "learning_rate": 4.366764275256223e-05,
+      "loss": 0.6513,
+      "step": 325
     },
     {
-      "epoch": 4.32,
-      "learning_rate": 3.155339805825243e-05,
-      "loss": 0.395,
-      "step": 1780
+      "epoch": 0.65,
+      "learning_rate": 4.348462664714495e-05,
+      "loss": 0.6353,
+      "step": 330
     },
     {
-      "epoch": 4.34,
-      "learning_rate": 3.1418554476806906e-05,
-      "loss": 0.3959,
-      "step": 1790
+      "epoch": 0.66,
+      "learning_rate": 4.330161054172767e-05,
+      "loss": 0.6323,
+      "step": 335
     },
     {
-      "epoch": 4.36,
-      "learning_rate": 3.128371089536138e-05,
-      "loss": 0.3983,
-      "step": 1800
+      "epoch": 0.67,
+      "learning_rate": 4.31185944363104e-05,
+      "loss": 0.6454,
+      "step": 340
     },
     {
-      "epoch": 4.39,
-      "learning_rate": 3.114886731391586e-05,
-      "loss": 0.4456,
-      "step": 1810
+      "epoch": 0.68,
+      "learning_rate": 4.2935578330893125e-05,
+      "loss": 0.6282,
+      "step": 345
     },
     {
-      "epoch": 4.41,
-      "learning_rate": 3.101402373247034e-05,
-      "loss": 0.3956,
-      "step": 1820
+      "epoch": 0.69,
+      "learning_rate": 4.2752562225475845e-05,
+      "loss": 0.6029,
+      "step": 350
     },
     {
-      "epoch": 4.44,
-      "learning_rate": 3.087918015102481e-05,
-      "loss": 0.3926,
-      "step": 1830
+      "epoch": 0.7,
+      "learning_rate": 4.256954612005857e-05,
+      "loss": 0.6069,
+      "step": 355
     },
     {
-      "epoch": 4.46,
-      "learning_rate": 3.074433656957929e-05,
-      "loss": 0.4182,
-      "step": 1840
+      "epoch": 0.71,
+      "learning_rate": 4.238653001464129e-05,
+      "loss": 0.5707,
+      "step": 360
     },
     {
-      "epoch": 4.48,
-      "learning_rate": 3.0609492988133766e-05,
-      "loss": 0.4022,
-      "step": 1850
+      "epoch": 0.72,
+      "learning_rate": 4.2203513909224016e-05,
+      "loss": 0.6161,
+      "step": 365
     },
     {
-      "epoch": 4.51,
-      "learning_rate": 3.0474649406688245e-05,
-      "loss": 0.4088,
-      "step": 1860
+      "epoch": 0.73,
+      "learning_rate": 4.2020497803806735e-05,
+      "loss": 0.6118,
+      "step": 370
     },
     {
-      "epoch": 4.53,
-      "learning_rate": 3.0339805825242717e-05,
-      "loss": 0.386,
-      "step": 1870
+      "epoch": 0.74,
+      "learning_rate": 4.1837481698389455e-05,
+      "loss": 0.613,
+      "step": 375
     },
     {
-      "epoch": 4.56,
-      "learning_rate": 3.0204962243797196e-05,
-      "loss": 0.411,
-      "step": 1880
+      "epoch": 0.75,
+      "learning_rate": 4.165446559297219e-05,
+      "loss": 0.6044,
+      "step": 380
     },
     {
-      "epoch": 4.58,
-      "learning_rate": 3.0070118662351675e-05,
-      "loss": 0.3774,
-      "step": 1890
+      "epoch": 0.76,
+      "learning_rate": 4.147144948755491e-05,
+      "loss": 0.5668,
+      "step": 385
     },
     {
-      "epoch": 4.61,
-      "learning_rate": 2.993527508090615e-05,
-      "loss": 0.4273,
-      "step": 1900
+      "epoch": 0.77,
+      "learning_rate": 4.128843338213763e-05,
+      "loss": 0.6372,
+      "step": 390
     },
     {
-      "epoch": 4.63,
-      "learning_rate": 2.980043149946063e-05,
-      "loss": 0.4008,
-      "step": 1910
+      "epoch": 0.78,
+      "learning_rate": 4.110541727672035e-05,
+      "loss": 0.5783,
+      "step": 395
     },
     {
-      "epoch": 4.65,
-      "learning_rate": 2.96655879180151e-05,
-      "loss": 0.4021,
-      "step": 1920
+      "epoch": 0.79,
+      "learning_rate": 4.092240117130308e-05,
+      "loss": 0.6007,
+      "step": 400
     },
     {
-      "epoch": 4.68,
-      "learning_rate": 2.9530744336569584e-05,
-      "loss": 0.4117,
-      "step": 1930
+      "epoch": 0.8,
+      "learning_rate": 4.07393850658858e-05,
+      "loss": 0.5667,
+      "step": 405
     },
     {
-      "epoch": 4.7,
-      "learning_rate": 2.9395900755124056e-05,
-      "loss": 0.4169,
-      "step": 1940
+      "epoch": 0.81,
+      "learning_rate": 4.0556368960468524e-05,
+      "loss": 0.5819,
+      "step": 410
     },
     {
-      "epoch": 4.73,
-      "learning_rate": 2.9261057173678535e-05,
-      "loss": 0.3996,
-      "step": 1950
+      "epoch": 0.82,
+      "learning_rate": 4.037335285505124e-05,
+      "loss": 0.5884,
+      "step": 415
     },
     {
-      "epoch": 4.75,
-      "learning_rate": 2.912621359223301e-05,
-      "loss": 0.4179,
-      "step": 1960
+      "epoch": 0.83,
+      "learning_rate": 4.019033674963397e-05,
+      "loss": 0.6358,
+      "step": 420
     },
     {
-      "epoch": 4.78,
-      "learning_rate": 2.899137001078749e-05,
-      "loss": 0.4037,
-      "step": 1970
+      "epoch": 0.84,
+      "learning_rate": 4.0007320644216695e-05,
+      "loss": 0.6365,
+      "step": 425
     },
     {
-      "epoch": 4.8,
-      "learning_rate": 2.885652642934196e-05,
-      "loss": 0.4045,
-      "step": 1980
+      "epoch": 0.85,
+      "learning_rate": 3.9824304538799415e-05,
+      "loss": 0.569,
+      "step": 430
     },
     {
-      "epoch": 4.82,
-      "learning_rate": 2.872168284789644e-05,
-      "loss": 0.4055,
-      "step": 1990
+      "epoch": 0.86,
+      "learning_rate": 3.964128843338214e-05,
+      "loss": 0.6281,
+      "step": 435
     },
     {
-      "epoch": 4.85,
-      "learning_rate": 2.8586839266450923e-05,
-      "loss": 0.4294,
-      "step": 2000
+      "epoch": 0.87,
+      "learning_rate": 3.945827232796486e-05,
+      "loss": 0.5812,
+      "step": 440
     },
     {
-      "epoch": 4.87,
-      "learning_rate": 2.8451995685005395e-05,
-      "loss": 0.4161,
-      "step": 2010
+      "epoch": 0.88,
+      "learning_rate": 3.9275256222547586e-05,
+      "loss": 0.604,
+      "step": 445
     },
     {
-      "epoch": 4.9,
-      "learning_rate": 2.8317152103559874e-05,
-      "loss": 0.4084,
-      "step": 2020
+      "epoch": 0.89,
+      "learning_rate": 3.9092240117130305e-05,
+      "loss": 0.5833,
+      "step": 450
     },
     {
-      "epoch": 4.92,
-      "learning_rate": 2.818230852211435e-05,
-      "loss": 0.4111,
-      "step": 2030
+      "epoch": 0.9,
+      "learning_rate": 3.890922401171303e-05,
+      "loss": 0.5247,
+      "step": 455
     },
     {
-      "epoch": 4.95,
-      "learning_rate": 2.8047464940668828e-05,
-      "loss": 0.3972,
-      "step": 2040
+      "epoch": 0.91,
+      "learning_rate": 3.872620790629576e-05,
+      "loss": 0.5513,
+      "step": 460
     },
     {
-      "epoch": 4.97,
-      "learning_rate": 2.79126213592233e-05,
-      "loss": 0.3997,
-      "step": 2050
+      "epoch": 0.92,
+      "learning_rate": 3.8543191800878484e-05,
+      "loss": 0.5885,
+      "step": 465
     },
     {
-      "epoch": 4.99,
-      "learning_rate": 2.777777777777778e-05,
-      "loss": 0.3859,
-      "step": 2060
+      "epoch": 0.93,
+      "learning_rate": 3.83601756954612e-05,
+      "loss": 0.5995,
+      "step": 470
     },
     {
-      "epoch": 5.0,
-      "eval_accuracy": 0.7717354244800545,
-      "eval_loss": 0.4393536448478699,
-      "eval_runtime": 44.2527,
-      "eval_samples_per_second": 132.557,
-      "eval_steps_per_second": 4.158,
-      "step": 2062
+      "epoch": 0.94,
+      "learning_rate": 3.817715959004393e-05,
+      "loss": 0.591,
+      "step": 475
     },
     {
-      "epoch": 5.02,
-      "learning_rate": 2.7642934196332255e-05,
-      "loss": 0.4082,
-      "step": 2070
+      "epoch": 0.95,
+      "learning_rate": 3.799414348462665e-05,
+      "loss": 0.6269,
+      "step": 480
     },
     {
-      "epoch": 5.04,
-      "learning_rate": 2.7508090614886734e-05,
-      "loss": 0.384,
-      "step": 2080
+      "epoch": 0.96,
+      "learning_rate": 3.7811127379209375e-05,
+      "loss": 0.5496,
+      "step": 485
     },
     {
-      "epoch": 5.07,
-      "learning_rate": 2.7373247033441206e-05,
-      "loss": 0.4004,
-      "step": 2090
+      "epoch": 0.97,
+      "learning_rate": 3.7628111273792094e-05,
+      "loss": 0.6007,
+      "step": 490
     },
     {
-      "epoch": 5.09,
-      "learning_rate": 2.7238403451995688e-05,
-      "loss": 0.4186,
-      "step": 2100
+      "epoch": 0.98,
+      "learning_rate": 3.744509516837481e-05,
+      "loss": 0.5894,
+      "step": 495
     },
     {
-      "epoch": 5.12,
-      "learning_rate": 2.7103559870550167e-05,
-      "loss": 0.385,
-      "step": 2110
+      "epoch": 0.99,
+      "learning_rate": 3.7262079062957546e-05,
+      "loss": 0.6417,
+      "step": 500
     },
     {
-      "epoch": 5.14,
-      "learning_rate": 2.696871628910464e-05,
-      "loss": 0.3718,
-      "step": 2120
+      "epoch": 1.0,
+      "learning_rate": 3.7079062957540265e-05,
+      "loss": 0.5654,
+      "step": 505
     },
     {
-      "epoch": 5.16,
-      "learning_rate": 2.6833872707659118e-05,
-      "loss": 0.3845,
-      "step": 2130
+      "epoch": 1.0,
+      "eval_accuracy": 0.7356656948493683,
+      "eval_loss": 0.46419134736061096,
+      "eval_runtime": 54.4631,
+      "eval_samples_per_second": 132.255,
+      "eval_steps_per_second": 4.15,
+      "step": 506
     },
     {
-      "epoch": 5.19,
-      "learning_rate": 2.6699029126213593e-05,
-      "loss": 0.3861,
-      "step": 2140
+      "epoch": 1.01,
+      "learning_rate": 3.689604685212299e-05,
+      "loss": 0.5932,
+      "step": 510
     },
     {
-      "epoch": 5.21,
-      "learning_rate": 2.6564185544768072e-05,
-      "loss": 0.3959,
-      "step": 2150
+      "epoch": 1.02,
+      "learning_rate": 3.671303074670571e-05,
+      "loss": 0.5732,
+      "step": 515
     },
     {
-      "epoch": 5.24,
-      "learning_rate": 2.6429341963322544e-05,
-      "loss": 0.411,
-      "step": 2160
+      "epoch": 1.03,
+      "learning_rate": 3.653001464128844e-05,
+      "loss": 0.5225,
+      "step": 520
     },
     {
-      "epoch": 5.26,
-      "learning_rate": 2.6294498381877027e-05,
-      "loss": 0.3849,
-      "step": 2170
+      "epoch": 1.04,
+      "learning_rate": 3.6346998535871156e-05,
+      "loss": 0.57,
+      "step": 525
     },
     {
-      "epoch": 5.28,
-      "learning_rate": 2.61596548004315e-05,
-      "loss": 0.3884,
-      "step": 2180
+      "epoch": 1.05,
+      "learning_rate": 3.616398243045388e-05,
+      "loss": 0.5341,
+      "step": 530
     },
     {
-      "epoch": 5.31,
-      "learning_rate": 2.6024811218985978e-05,
-      "loss": 0.3848,
-      "step": 2190
+      "epoch": 1.06,
+      "learning_rate": 3.59809663250366e-05,
+      "loss": 0.5654,
+      "step": 535
     },
     {
-      "epoch": 5.33,
-      "learning_rate": 2.5889967637540453e-05,
-      "loss": 0.3937,
-      "step": 2200
+      "epoch": 1.07,
+      "learning_rate": 3.579795021961933e-05,
+      "loss": 0.5749,
+      "step": 540
     },
     {
-      "epoch": 5.36,
-      "learning_rate": 2.5755124056094932e-05,
-      "loss": 0.4112,
-      "step": 2210
+      "epoch": 1.08,
+      "learning_rate": 3.5614934114202054e-05,
+      "loss": 0.5546,
+      "step": 545
     },
     {
-      "epoch": 5.38,
-      "learning_rate": 2.5620280474649404e-05,
-      "loss": 0.3902,
-      "step": 2220
+      "epoch": 1.09,
+      "learning_rate": 3.543191800878477e-05,
+      "loss": 0.5608,
+      "step": 550
     },
     {
-      "epoch": 5.41,
-      "learning_rate": 2.5485436893203883e-05,
-      "loss": 0.3899,
-      "step": 2230
+      "epoch": 1.1,
+      "learning_rate": 3.52489019033675e-05,
+      "loss": 0.5155,
+      "step": 555
     },
     {
-      "epoch": 5.43,
-      "learning_rate": 2.5350593311758362e-05,
-      "loss": 0.4006,
-      "step": 2240
+      "epoch": 1.11,
+      "learning_rate": 3.506588579795022e-05,
+      "loss": 0.541,
+      "step": 560
     },
     {
-      "epoch": 5.45,
-      "learning_rate": 2.5215749730312838e-05,
-      "loss": 0.4078,
-      "step": 2250
+      "epoch": 1.12,
+      "learning_rate": 3.4882869692532945e-05,
+      "loss": 0.5515,
+      "step": 565
     },
     {
-      "epoch": 5.48,
-      "learning_rate": 2.5080906148867317e-05,
-      "loss": 0.4113,
-      "step": 2260
+      "epoch": 1.13,
+      "learning_rate": 3.4699853587115664e-05,
+      "loss": 0.5686,
+      "step": 570
     },
     {
-      "epoch": 5.5,
-      "learning_rate": 2.4946062567421792e-05,
-      "loss": 0.404,
-      "step": 2270
+      "epoch": 1.14,
+      "learning_rate": 3.451683748169839e-05,
+      "loss": 0.566,
+      "step": 575
     },
     {
-      "epoch": 5.53,
-      "learning_rate": 2.481121898597627e-05,
-      "loss": 0.429,
-      "step": 2280
+      "epoch": 1.15,
+      "learning_rate": 3.4333821376281116e-05,
+      "loss": 0.5548,
+      "step": 580
     },
     {
-      "epoch": 5.55,
-      "learning_rate": 2.4676375404530747e-05,
-      "loss": 0.3862,
-      "step": 2290
+      "epoch": 1.15,
+      "learning_rate": 3.415080527086384e-05,
+      "loss": 0.5808,
+      "step": 585
     },
     {
-      "epoch": 5.58,
-      "learning_rate": 2.4541531823085222e-05,
-      "loss": 0.3846,
-      "step": 2300
+      "epoch": 1.16,
+      "learning_rate": 3.396778916544656e-05,
+      "loss": 0.5683,
+      "step": 590
     },
     {
-      "epoch": 5.6,
-      "learning_rate": 2.44066882416397e-05,
-      "loss": 0.3923,
-      "step": 2310
+      "epoch": 1.17,
+      "learning_rate": 3.378477306002929e-05,
+      "loss": 0.5428,
+      "step": 595
     },
     {
-      "epoch": 5.62,
-      "learning_rate": 2.4271844660194176e-05,
-      "loss": 0.3899,
-      "step": 2320
+      "epoch": 1.18,
+      "learning_rate": 3.360175695461201e-05,
+      "loss": 0.541,
+      "step": 600
     },
     {
-      "epoch": 5.65,
-      "learning_rate": 2.4137001078748652e-05,
-      "loss": 0.3964,
-      "step": 2330
+      "epoch": 1.19,
+      "learning_rate": 3.3418740849194726e-05,
+      "loss": 0.5082,
+      "step": 605
     },
     {
-      "epoch": 5.67,
-      "learning_rate": 2.400215749730313e-05,
-      "loss": 0.3773,
-      "step": 2340
+      "epoch": 1.2,
+      "learning_rate": 3.323572474377745e-05,
+      "loss": 0.5672,
+      "step": 610
     },
     {
-      "epoch": 5.7,
-      "learning_rate": 2.3867313915857606e-05,
-      "loss": 0.3735,
-      "step": 2350
+      "epoch": 1.21,
+      "learning_rate": 3.305270863836017e-05,
+      "loss": 0.5593,
+      "step": 615
     },
     {
-      "epoch": 5.72,
-      "learning_rate": 2.3732470334412082e-05,
-      "loss": 0.3679,
-      "step": 2360
+      "epoch": 1.22,
+      "learning_rate": 3.2869692532942905e-05,
+      "loss": 0.5492,
+      "step": 620
     },
     {
-      "epoch": 5.75,
-      "learning_rate": 2.3597626752966557e-05,
-      "loss": 0.3818,
-      "step": 2370
+      "epoch": 1.23,
+      "learning_rate": 3.2686676427525624e-05,
+      "loss": 0.553,
+      "step": 625
     },
     {
-      "epoch": 5.77,
-      "learning_rate": 2.3462783171521036e-05,
-      "loss": 0.4101,
-      "step": 2380
+      "epoch": 1.24,
+      "learning_rate": 3.250366032210835e-05,
+      "loss": 0.5812,
+      "step": 630
     },
     {
-      "epoch": 5.79,
-      "learning_rate": 2.3327939590075515e-05,
-      "loss": 0.4143,
-      "step": 2390
+      "epoch": 1.25,
+      "learning_rate": 3.232064421669107e-05,
+      "loss": 0.5467,
+      "step": 635
     },
     {
-      "epoch": 5.82,
-      "learning_rate": 2.319309600862999e-05,
-      "loss": 0.3931,
-      "step": 2400
+      "epoch": 1.26,
+      "learning_rate": 3.2137628111273795e-05,
+      "loss": 0.5604,
+      "step": 640
     },
     {
-      "epoch": 5.84,
-      "learning_rate": 2.305825242718447e-05,
-      "loss": 0.3645,
-      "step": 2410
+      "epoch": 1.27,
+      "learning_rate": 3.1954612005856515e-05,
+      "loss": 0.5613,
+      "step": 645
     },
     {
-      "epoch": 5.87,
-      "learning_rate": 2.2923408845738945e-05,
-      "loss": 0.3971,
-      "step": 2420
+      "epoch": 1.28,
+      "learning_rate": 3.177159590043924e-05,
+      "loss": 0.5247,
+      "step": 650
     },
     {
-      "epoch": 5.89,
-      "learning_rate": 2.278856526429342e-05,
-      "loss": 0.3883,
-      "step": 2430
+      "epoch": 1.29,
+      "learning_rate": 3.158857979502196e-05,
+      "loss": 0.5596,
+      "step": 655
     },
     {
-      "epoch": 5.92,
-      "learning_rate": 2.2653721682847896e-05,
-      "loss": 0.3859,
-      "step": 2440
+      "epoch": 1.3,
+      "learning_rate": 3.1405563689604686e-05,
+      "loss": 0.5211,
+      "step": 660
     },
     {
-      "epoch": 5.94,
-      "learning_rate": 2.2518878101402375e-05,
-      "loss": 0.3803,
-      "step": 2450
+      "epoch": 1.31,
+      "learning_rate": 3.122254758418741e-05,
+      "loss": 0.5338,
+      "step": 665
     },
     {
-      "epoch": 5.96,
-      "learning_rate": 2.238403451995685e-05,
-      "loss": 0.3686,
-      "step": 2460
+      "epoch": 1.32,
+      "learning_rate": 3.103953147877013e-05,
+      "loss": 0.5422,
+      "step": 670
     },
     {
-      "epoch": 5.99,
-      "learning_rate": 2.2249190938511326e-05,
-      "loss": 0.3822,
-      "step": 2470
+      "epoch": 1.33,
+      "learning_rate": 3.085651537335286e-05,
+      "loss": 0.5502,
+      "step": 675
     },
     {
-      "epoch": 6.0,
-      "eval_accuracy": 0.7783839072621889,
-      "eval_loss": 0.4409945607185364,
-      "eval_runtime": 44.1667,
-      "eval_samples_per_second": 132.815,
-      "eval_steps_per_second": 4.166,
-      "step": 2475
+      "epoch": 1.34,
+      "learning_rate": 3.067349926793558e-05,
+      "loss": 0.5305,
+      "step": 680
     },
     {
-      "epoch": 6.01,
-      "learning_rate": 2.2114347357065805e-05,
-      "loss": 0.402,
-      "step": 2480
+      "epoch": 1.35,
+      "learning_rate": 3.0490483162518303e-05,
+      "loss": 0.5415,
+      "step": 685
     },
     {
-      "epoch": 6.04,
-      "learning_rate": 2.197950377562028e-05,
-      "loss": 0.3558,
-      "step": 2490
+      "epoch": 1.36,
+      "learning_rate": 3.0307467057101026e-05,
+      "loss": 0.5065,
+      "step": 690
     },
     {
-      "epoch": 6.06,
-      "learning_rate": 2.1844660194174756e-05,
-      "loss": 0.3853,
-      "step": 2500
+      "epoch": 1.37,
+      "learning_rate": 3.012445095168375e-05,
+      "loss": 0.551,
+      "step": 695
     },
     {
-      "epoch": 6.08,
-      "learning_rate": 2.1709816612729235e-05,
-      "loss": 0.3455,
-      "step": 2510
+      "epoch": 1.38,
+      "learning_rate": 2.9941434846266475e-05,
+      "loss": 0.5345,
+      "step": 700
     },
     {
-      "epoch": 6.11,
-      "learning_rate": 2.1574973031283714e-05,
-      "loss": 0.3913,
-      "step": 2520
+      "epoch": 1.39,
+      "learning_rate": 2.9758418740849197e-05,
+      "loss": 0.514,
+      "step": 705
     },
     {
-      "epoch": 6.13,
-      "learning_rate": 2.144012944983819e-05,
-      "loss": 0.3874,
-      "step": 2530
+      "epoch": 1.4,
+      "learning_rate": 2.957540263543192e-05,
+      "loss": 0.5535,
+      "step": 710
     },
     {
-      "epoch": 6.16,
-      "learning_rate": 2.1305285868392665e-05,
-      "loss": 0.3646,
-      "step": 2540
+      "epoch": 1.41,
+      "learning_rate": 2.9392386530014643e-05,
+      "loss": 0.5537,
+      "step": 715
     },
     {
-      "epoch": 6.18,
-      "learning_rate": 2.1170442286947144e-05,
-      "loss": 0.3933,
-      "step": 2550
+      "epoch": 1.42,
+      "learning_rate": 2.9209370424597365e-05,
+      "loss": 0.5305,
+      "step": 720
     },
     {
-      "epoch": 6.21,
-      "learning_rate": 2.103559870550162e-05,
-      "loss": 0.3445,
-      "step": 2560
+      "epoch": 1.43,
+      "learning_rate": 2.9026354319180088e-05,
+      "loss": 0.5551,
+      "step": 725
     },
     {
-      "epoch": 6.23,
-      "learning_rate": 2.0900755124056095e-05,
-      "loss": 0.3621,
-      "step": 2570
+      "epoch": 1.44,
+      "learning_rate": 2.884333821376281e-05,
+      "loss": 0.5723,
+      "step": 730
     },
     {
-      "epoch": 6.25,
-      "learning_rate": 2.0765911542610574e-05,
-      "loss": 0.3742,
-      "step": 2580
+      "epoch": 1.45,
+      "learning_rate": 2.8660322108345534e-05,
+      "loss": 0.5722,
+      "step": 735
     },
     {
-      "epoch": 6.28,
-      "learning_rate": 2.063106796116505e-05,
-      "loss": 0.3861,
-      "step": 2590
+      "epoch": 1.46,
+      "learning_rate": 2.847730600292826e-05,
+      "loss": 0.5376,
+      "step": 740
     },
     {
-      "epoch": 6.3,
-      "learning_rate": 2.0496224379719525e-05,
-      "loss": 0.3535,
-      "step": 2600
+      "epoch": 1.47,
+      "learning_rate": 2.8294289897510982e-05,
+      "loss": 0.577,
+      "step": 745
     },
     {
-      "epoch": 6.33,
-      "learning_rate": 2.0361380798274e-05,
-      "loss": 0.3739,
-      "step": 2610
+      "epoch": 1.48,
+      "learning_rate": 2.8111273792093705e-05,
+      "loss": 0.5322,
+      "step": 750
     },
     {
-      "epoch": 6.35,
-      "learning_rate": 2.0226537216828483e-05,
-      "loss": 0.3711,
-      "step": 2620
+      "epoch": 1.49,
+      "learning_rate": 2.7928257686676428e-05,
+      "loss": 0.5505,
+      "step": 755
     },
     {
-      "epoch": 6.38,
-      "learning_rate": 2.0091693635382958e-05,
-      "loss": 0.3601,
-      "step": 2630
+      "epoch": 1.5,
+      "learning_rate": 2.774524158125915e-05,
+      "loss": 0.4995,
+      "step": 760
     },
     {
-      "epoch": 6.4,
-      "learning_rate": 1.9956850053937434e-05,
-      "loss": 0.3603,
-      "step": 2640
+      "epoch": 1.51,
+      "learning_rate": 2.7562225475841873e-05,
+      "loss": 0.5371,
+      "step": 765
     },
     {
-      "epoch": 6.42,
-      "learning_rate": 1.982200647249191e-05,
-      "loss": 0.3574,
-      "step": 2650
+      "epoch": 1.52,
+      "learning_rate": 2.7379209370424596e-05,
+      "loss": 0.5526,
+      "step": 770
     },
     {
-      "epoch": 6.45,
-      "learning_rate": 1.9687162891046388e-05,
-      "loss": 0.3706,
-      "step": 2660
+      "epoch": 1.53,
+      "learning_rate": 2.719619326500732e-05,
+      "loss": 0.5275,
+      "step": 775
     },
     {
-      "epoch": 6.47,
-      "learning_rate": 1.9552319309600864e-05,
-      "loss": 0.3829,
-      "step": 2670
+      "epoch": 1.54,
+      "learning_rate": 2.7013177159590048e-05,
+      "loss": 0.5073,
+      "step": 780
     },
     {
-      "epoch": 6.5,
-      "learning_rate": 1.941747572815534e-05,
-      "loss": 0.3904,
-      "step": 2680
+      "epoch": 1.55,
+      "learning_rate": 2.683016105417277e-05,
+      "loss": 0.5645,
+      "step": 785
     },
     {
-      "epoch": 6.52,
-      "learning_rate": 1.9282632146709818e-05,
-      "loss": 0.3595,
-      "step": 2690
+      "epoch": 1.56,
+      "learning_rate": 2.6647144948755493e-05,
+      "loss": 0.5253,
+      "step": 790
     },
     {
-      "epoch": 6.55,
-      "learning_rate": 1.9147788565264294e-05,
-      "loss": 0.3494,
-      "step": 2700
+      "epoch": 1.57,
+      "learning_rate": 2.6464128843338216e-05,
+      "loss": 0.5327,
+      "step": 795
     },
     {
-      "epoch": 6.57,
-      "learning_rate": 1.901294498381877e-05,
-      "loss": 0.386,
-      "step": 2710
+      "epoch": 1.58,
+      "learning_rate": 2.628111273792094e-05,
+      "loss": 0.5219,
+      "step": 800
     },
     {
-      "epoch": 6.59,
-      "learning_rate": 1.8878101402373248e-05,
-      "loss": 0.3716,
-      "step": 2720
+      "epoch": 1.59,
+      "learning_rate": 2.609809663250366e-05,
+      "loss": 0.5068,
+      "step": 805
     },
     {
-      "epoch": 6.62,
-      "learning_rate": 1.8743257820927727e-05,
-      "loss": 0.3508,
-      "step": 2730
+      "epoch": 1.6,
+      "learning_rate": 2.5915080527086384e-05,
+      "loss": 0.5378,
+      "step": 810
     },
     {
-      "epoch": 6.64,
-      "learning_rate": 1.8608414239482202e-05,
-      "loss": 0.3528,
-      "step": 2740
+      "epoch": 1.61,
+      "learning_rate": 2.5732064421669107e-05,
+      "loss": 0.5464,
+      "step": 815
     },
     {
-      "epoch": 6.67,
-      "learning_rate": 1.8473570658036678e-05,
-      "loss": 0.3731,
-      "step": 2750
+      "epoch": 1.62,
+      "learning_rate": 2.5549048316251833e-05,
+      "loss": 0.5195,
+      "step": 820
     },
     {
-      "epoch": 6.69,
-      "learning_rate": 1.8338727076591157e-05,
-      "loss": 0.3812,
-      "step": 2760
+      "epoch": 1.63,
+      "learning_rate": 2.5366032210834556e-05,
+      "loss": 0.5481,
+      "step": 825
     },
     {
-      "epoch": 6.72,
-      "learning_rate": 1.8203883495145632e-05,
-      "loss": 0.3745,
-      "step": 2770
+      "epoch": 1.64,
+      "learning_rate": 2.518301610541728e-05,
+      "loss": 0.512,
+      "step": 830
     },
     {
-      "epoch": 6.74,
-      "learning_rate": 1.8069039913700108e-05,
-      "loss": 0.3946,
-      "step": 2780
+      "epoch": 1.65,
+      "learning_rate": 2.5e-05,
+      "loss": 0.5342,
+      "step": 835
     },
     {
-      "epoch": 6.76,
-      "learning_rate": 1.7934196332254587e-05,
-      "loss": 0.3744,
-      "step": 2790
+      "epoch": 1.66,
+      "learning_rate": 2.4816983894582724e-05,
+      "loss": 0.5444,
+      "step": 840
     },
     {
-      "epoch": 6.79,
-      "learning_rate": 1.7799352750809062e-05,
-      "loss": 0.3496,
-      "step": 2800
+      "epoch": 1.67,
+      "learning_rate": 2.463396778916545e-05,
+      "loss": 0.533,
+      "step": 845
     },
     {
-      "epoch": 6.81,
-      "learning_rate": 1.7664509169363538e-05,
-      "loss": 0.359,
-      "step": 2810
+      "epoch": 1.68,
+      "learning_rate": 2.4450951683748173e-05,
+      "loss": 0.5249,
+      "step": 850
     },
     {
-      "epoch": 6.84,
-      "learning_rate": 1.7529665587918013e-05,
-      "loss": 0.3781,
-      "step": 2820
+      "epoch": 1.69,
+      "learning_rate": 2.4267935578330895e-05,
+      "loss": 0.5867,
+      "step": 855
     },
     {
-      "epoch": 6.86,
-      "learning_rate": 1.7394822006472492e-05,
-      "loss": 0.3524,
-      "step": 2830
+      "epoch": 1.7,
+      "learning_rate": 2.4084919472913618e-05,
+      "loss": 0.4845,
+      "step": 860
     },
     {
-      "epoch": 6.88,
-      "learning_rate": 1.725997842502697e-05,
-      "loss": 0.3977,
-      "step": 2840
+      "epoch": 1.71,
+      "learning_rate": 2.390190336749634e-05,
+      "loss": 0.5419,
+      "step": 865
     },
     {
-      "epoch": 6.91,
-      "learning_rate": 1.7125134843581447e-05,
-      "loss": 0.3668,
-      "step": 2850
+      "epoch": 1.72,
+      "learning_rate": 2.3718887262079064e-05,
+      "loss": 0.5173,
+      "step": 870
     },
     {
-      "epoch": 6.93,
-      "learning_rate": 1.6990291262135926e-05,
-      "loss": 0.3718,
-      "step": 2860
+      "epoch": 1.73,
+      "learning_rate": 2.3535871156661786e-05,
+      "loss": 0.5138,
+      "step": 875
     },
     {
-      "epoch": 6.96,
-      "learning_rate": 1.68554476806904e-05,
-      "loss": 0.3519,
-      "step": 2870
+      "epoch": 1.74,
+      "learning_rate": 2.335285505124451e-05,
+      "loss": 0.5291,
+      "step": 880
     },
     {
-      "epoch": 6.98,
-      "learning_rate": 1.6720604099244877e-05,
-      "loss": 0.3804,
-      "step": 2880
+      "epoch": 1.75,
+      "learning_rate": 2.3169838945827235e-05,
+      "loss": 0.5299,
+      "step": 885
     },
     {
-      "epoch": 7.0,
-      "eval_accuracy": 0.7766791680872827,
-      "eval_loss": 0.4561130106449127,
-      "eval_runtime": 44.252,
-      "eval_samples_per_second": 132.559,
-      "eval_steps_per_second": 4.158,
-      "step": 2887
+      "epoch": 1.76,
+      "learning_rate": 2.2986822840409958e-05,
+      "loss": 0.5249,
+      "step": 890
     },
     {
-      "epoch": 7.01,
-      "learning_rate": 1.6585760517799352e-05,
-      "loss": 0.3441,
-      "step": 2890
+      "epoch": 1.77,
+      "learning_rate": 2.280380673499268e-05,
+      "loss": 0.4676,
+      "step": 895
     },
     {
-      "epoch": 7.03,
-      "learning_rate": 1.645091693635383e-05,
-      "loss": 0.3618,
-      "step": 2900
+      "epoch": 1.78,
+      "learning_rate": 2.2620790629575403e-05,
+      "loss": 0.5323,
+      "step": 900
     },
     {
-      "epoch": 7.05,
-      "learning_rate": 1.6316073354908307e-05,
-      "loss": 0.3523,
-      "step": 2910
+      "epoch": 1.79,
+      "learning_rate": 2.243777452415813e-05,
+      "loss": 0.4845,
+      "step": 905
     },
     {
-      "epoch": 7.08,
-      "learning_rate": 1.6181229773462782e-05,
-      "loss": 0.3421,
-      "step": 2920
+      "epoch": 1.8,
+      "learning_rate": 2.2254758418740852e-05,
+      "loss": 0.5559,
+      "step": 910
     },
     {
-      "epoch": 7.1,
-      "learning_rate": 1.604638619201726e-05,
-      "loss": 0.342,
-      "step": 2930
+      "epoch": 1.81,
+      "learning_rate": 2.2071742313323575e-05,
+      "loss": 0.5064,
+      "step": 915
     },
     {
-      "epoch": 7.13,
-      "learning_rate": 1.5911542610571736e-05,
-      "loss": 0.3646,
-      "step": 2940
+      "epoch": 1.82,
+      "learning_rate": 2.1888726207906297e-05,
+      "loss": 0.5281,
+      "step": 920
     },
     {
-      "epoch": 7.15,
-      "learning_rate": 1.5776699029126215e-05,
-      "loss": 0.3115,
-      "step": 2950
+      "epoch": 1.83,
+      "learning_rate": 2.170571010248902e-05,
+      "loss": 0.5471,
+      "step": 925
     },
     {
-      "epoch": 7.18,
-      "learning_rate": 1.564185544768069e-05,
-      "loss": 0.3539,
-      "step": 2960
+      "epoch": 1.84,
+      "learning_rate": 2.1522693997071743e-05,
+      "loss": 0.4968,
+      "step": 930
     },
     {
-      "epoch": 7.2,
-      "learning_rate": 1.550701186623517e-05,
-      "loss": 0.3236,
-      "step": 2970
+      "epoch": 1.85,
+      "learning_rate": 2.1339677891654465e-05,
+      "loss": 0.481,
+      "step": 935
     },
     {
-      "epoch": 7.22,
-      "learning_rate": 1.5372168284789645e-05,
-      "loss": 0.3603,
-      "step": 2980
+      "epoch": 1.86,
+      "learning_rate": 2.1156661786237188e-05,
+      "loss": 0.5199,
+      "step": 940
     },
     {
-      "epoch": 7.25,
-      "learning_rate": 1.5237324703344123e-05,
-      "loss": 0.3449,
-      "step": 2990
+      "epoch": 1.87,
+      "learning_rate": 2.0973645680819914e-05,
+      "loss": 0.4974,
+      "step": 945
     },
     {
-      "epoch": 7.27,
-      "learning_rate": 1.5102481121898598e-05,
-      "loss": 0.3477,
-      "step": 3000
+      "epoch": 1.88,
+      "learning_rate": 2.0790629575402637e-05,
+      "loss": 0.5176,
+      "step": 950
     },
     {
-      "epoch": 7.3,
-      "learning_rate": 1.4967637540453075e-05,
-      "loss": 0.364,
-      "step": 3010
+      "epoch": 1.89,
+      "learning_rate": 2.060761346998536e-05,
+      "loss": 0.5054,
+      "step": 955
     },
     {
-      "epoch": 7.32,
-      "learning_rate": 1.483279395900755e-05,
-      "loss": 0.37,
-      "step": 3020
+      "epoch": 1.9,
+      "learning_rate": 2.0424597364568082e-05,
+      "loss": 0.5211,
+      "step": 960
     },
     {
-      "epoch": 7.35,
-      "learning_rate": 1.4697950377562028e-05,
-      "loss": 0.3719,
-      "step": 3030
+      "epoch": 1.91,
+      "learning_rate": 2.024158125915081e-05,
+      "loss": 0.4668,
+      "step": 965
     },
     {
-      "epoch": 7.37,
-      "learning_rate": 1.4563106796116505e-05,
-      "loss": 0.358,
-      "step": 3040
+      "epoch": 1.92,
+      "learning_rate": 2.005856515373353e-05,
+      "loss": 0.4834,
+      "step": 970
     },
     {
-      "epoch": 7.39,
-      "learning_rate": 1.442826321467098e-05,
-      "loss": 0.3529,
-      "step": 3050
+      "epoch": 1.92,
+      "learning_rate": 1.9875549048316254e-05,
+      "loss": 0.5393,
+      "step": 975
     },
     {
-      "epoch": 7.42,
-      "learning_rate": 1.4293419633225461e-05,
-      "loss": 0.3502,
-      "step": 3060
+      "epoch": 1.93,
+      "learning_rate": 1.9692532942898977e-05,
+      "loss": 0.5269,
+      "step": 980
     },
     {
-      "epoch": 7.44,
-      "learning_rate": 1.4158576051779937e-05,
-      "loss": 0.3708,
-      "step": 3070
+      "epoch": 1.94,
+      "learning_rate": 1.95095168374817e-05,
+      "loss": 0.5445,
+      "step": 985
     },
     {
-      "epoch": 7.47,
-      "learning_rate": 1.4023732470334414e-05,
-      "loss": 0.3221,
-      "step": 3080
+      "epoch": 1.95,
+      "learning_rate": 1.9326500732064422e-05,
+      "loss": 0.5473,
+      "step": 990
     },
     {
-      "epoch": 7.49,
-      "learning_rate": 1.388888888888889e-05,
-      "loss": 0.3511,
-      "step": 3090
+      "epoch": 1.96,
+      "learning_rate": 1.9143484626647145e-05,
+      "loss": 0.5002,
+      "step": 995
     },
     {
-      "epoch": 7.52,
-      "learning_rate": 1.3754045307443367e-05,
-      "loss": 0.333,
-      "step": 3100
+      "epoch": 1.97,
+      "learning_rate": 1.8960468521229867e-05,
+      "loss": 0.4839,
+      "step": 1000
     },
     {
-      "epoch": 7.54,
-      "learning_rate": 1.3619201725997844e-05,
-      "loss": 0.3606,
-      "step": 3110
+      "epoch": 1.98,
+      "learning_rate": 1.8777452415812594e-05,
+      "loss": 0.5268,
+      "step": 1005
     },
     {
-      "epoch": 7.56,
-      "learning_rate": 1.348435814455232e-05,
-      "loss": 0.3492,
-      "step": 3120
+      "epoch": 1.99,
+      "learning_rate": 1.8594436310395316e-05,
+      "loss": 0.5194,
+      "step": 1010
     },
     {
-      "epoch": 7.59,
-      "learning_rate": 1.3349514563106797e-05,
-      "loss": 0.3365,
-      "step": 3130
+      "epoch": 2.0,
+      "eval_accuracy": 0.7675968346522283,
+      "eval_loss": 0.4097585380077362,
+      "eval_runtime": 51.2327,
+      "eval_samples_per_second": 140.594,
+      "eval_steps_per_second": 4.411,
+      "step": 1013
     },
     {
-      "epoch": 7.61,
-      "learning_rate": 1.3214670981661272e-05,
-      "loss": 0.3551,
-      "step": 3140
+      "epoch": 2.0,
+      "learning_rate": 1.841142020497804e-05,
+      "loss": 0.4812,
+      "step": 1015
     },
     {
-      "epoch": 7.64,
-      "learning_rate": 1.307982740021575e-05,
-      "loss": 0.362,
-      "step": 3150
+      "epoch": 2.01,
+      "learning_rate": 1.822840409956076e-05,
+      "loss": 0.5085,
+      "step": 1020
     },
     {
-      "epoch": 7.66,
-      "learning_rate": 1.2944983818770227e-05,
-      "loss": 0.3274,
-      "step": 3160
+      "epoch": 2.02,
+      "learning_rate": 1.8045387994143488e-05,
+      "loss": 0.461,
+      "step": 1025
     },
     {
-      "epoch": 7.68,
-      "learning_rate": 1.2810140237324702e-05,
-      "loss": 0.329,
-      "step": 3170
+      "epoch": 2.03,
+      "learning_rate": 1.786237188872621e-05,
+      "loss": 0.5006,
+      "step": 1030
     },
     {
-      "epoch": 7.71,
-      "learning_rate": 1.2675296655879181e-05,
-      "loss": 0.352,
-      "step": 3180
+      "epoch": 2.04,
+      "learning_rate": 1.7679355783308933e-05,
+      "loss": 0.4821,
+      "step": 1035
     },
     {
-      "epoch": 7.73,
-      "learning_rate": 1.2540453074433658e-05,
-      "loss": 0.3475,
-      "step": 3190
+      "epoch": 2.05,
+      "learning_rate": 1.7496339677891656e-05,
+      "loss": 0.4414,
+      "step": 1040
     },
     {
-      "epoch": 7.76,
-      "learning_rate": 1.2405609492988135e-05,
-      "loss": 0.3418,
-      "step": 3200
+      "epoch": 2.06,
+      "learning_rate": 1.731332357247438e-05,
+      "loss": 0.4986,
+      "step": 1045
     },
     {
-      "epoch": 7.78,
-      "learning_rate": 1.2270765911542611e-05,
-      "loss": 0.3795,
-      "step": 3210
+      "epoch": 2.07,
+      "learning_rate": 1.71303074670571e-05,
+      "loss": 0.4747,
+      "step": 1050
     },
     {
-      "epoch": 7.81,
-      "learning_rate": 1.2135922330097088e-05,
-      "loss": 0.3549,
-      "step": 3220
+      "epoch": 2.08,
+      "learning_rate": 1.6947291361639824e-05,
+      "loss": 0.5198,
+      "step": 1055
     },
     {
-      "epoch": 7.83,
-      "learning_rate": 1.2001078748651565e-05,
-      "loss": 0.3421,
-      "step": 3230
+      "epoch": 2.09,
+      "learning_rate": 1.6764275256222547e-05,
+      "loss": 0.4425,
+      "step": 1060
     },
     {
-      "epoch": 7.85,
-      "learning_rate": 1.1866235167206041e-05,
-      "loss": 0.3171,
-      "step": 3240
+      "epoch": 2.1,
+      "learning_rate": 1.6581259150805273e-05,
+      "loss": 0.4593,
+      "step": 1065
     },
     {
-      "epoch": 7.88,
-      "learning_rate": 1.1731391585760518e-05,
-      "loss": 0.3513,
-      "step": 3250
+      "epoch": 2.11,
+      "learning_rate": 1.6398243045387995e-05,
+      "loss": 0.5233,
+      "step": 1070
     },
     {
-      "epoch": 7.9,
-      "learning_rate": 1.1596548004314995e-05,
-      "loss": 0.3514,
-      "step": 3260
+      "epoch": 2.12,
+      "learning_rate": 1.6215226939970718e-05,
+      "loss": 0.4698,
+      "step": 1075
     },
     {
-      "epoch": 7.93,
-      "learning_rate": 1.1461704422869473e-05,
-      "loss": 0.3622,
-      "step": 3270
+      "epoch": 2.13,
+      "learning_rate": 1.603221083455344e-05,
+      "loss": 0.5108,
+      "step": 1080
     },
     {
-      "epoch": 7.95,
-      "learning_rate": 1.1326860841423948e-05,
-      "loss": 0.3541,
-      "step": 3280
+      "epoch": 2.14,
+      "learning_rate": 1.5849194729136167e-05,
+      "loss": 0.5274,
+      "step": 1085
     },
     {
-      "epoch": 7.98,
-      "learning_rate": 1.1192017259978425e-05,
-      "loss": 0.3734,
-      "step": 3290
+      "epoch": 2.15,
+      "learning_rate": 1.566617862371889e-05,
+      "loss": 0.4443,
+      "step": 1090
     },
     {
-      "epoch": 8.0,
-      "learning_rate": 1.1057173678532903e-05,
-      "loss": 0.3231,
-      "step": 3300
+      "epoch": 2.16,
+      "learning_rate": 1.5483162518301612e-05,
+      "loss": 0.4591,
+      "step": 1095
     },
     {
-      "epoch": 8.0,
-      "eval_accuracy": 0.7783839072621889,
-      "eval_loss": 0.4504648447036743,
-      "eval_runtime": 43.9275,
-      "eval_samples_per_second": 133.538,
-      "eval_steps_per_second": 4.189,
-      "step": 3300
+      "epoch": 2.17,
+      "learning_rate": 1.5300146412884335e-05,
+      "loss": 0.5101,
+      "step": 1100
     },
     {
-      "epoch": 8.02,
-      "learning_rate": 1.0922330097087378e-05,
-      "loss": 0.3522,
-      "step": 3310
+      "epoch": 2.18,
+      "learning_rate": 1.511713030746706e-05,
+      "loss": 0.4938,
+      "step": 1105
     },
     {
-      "epoch": 8.05,
-      "learning_rate": 1.0787486515641857e-05,
-      "loss": 0.3337,
-      "step": 3320
+      "epoch": 2.19,
+      "learning_rate": 1.4934114202049782e-05,
+      "loss": 0.5465,
+      "step": 1110
     },
     {
-      "epoch": 8.07,
-      "learning_rate": 1.0652642934196332e-05,
-      "loss": 0.2974,
-      "step": 3330
+      "epoch": 2.2,
+      "learning_rate": 1.4751098096632505e-05,
+      "loss": 0.4086,
+      "step": 1115
     },
     {
-      "epoch": 8.1,
-      "learning_rate": 1.051779935275081e-05,
-      "loss": 0.3454,
-      "step": 3340
+      "epoch": 2.21,
+      "learning_rate": 1.4568081991215226e-05,
+      "loss": 0.4771,
+      "step": 1120
     },
     {
-      "epoch": 8.12,
-      "learning_rate": 1.0382955771305287e-05,
-      "loss": 0.3054,
-      "step": 3350
+      "epoch": 2.22,
+      "learning_rate": 1.4385065885797952e-05,
+      "loss": 0.5017,
+      "step": 1125
     },
     {
-      "epoch": 8.15,
-      "learning_rate": 1.0248112189859762e-05,
-      "loss": 0.3125,
-      "step": 3360
+      "epoch": 2.23,
+      "learning_rate": 1.4202049780380675e-05,
+      "loss": 0.4975,
+      "step": 1130
     },
     {
-      "epoch": 8.17,
-      "learning_rate": 1.0113268608414241e-05,
-      "loss": 0.3233,
-      "step": 3370
+      "epoch": 2.24,
+      "learning_rate": 1.4019033674963397e-05,
+      "loss": 0.573,
+      "step": 1135
     },
     {
-      "epoch": 8.19,
-      "learning_rate": 9.978425026968717e-06,
-      "loss": 0.3265,
-      "step": 3380
+      "epoch": 2.25,
+      "learning_rate": 1.383601756954612e-05,
+      "loss": 0.4946,
+      "step": 1140
     },
     {
-      "epoch": 8.22,
-      "learning_rate": 9.843581445523194e-06,
-      "loss": 0.3262,
-      "step": 3390
+      "epoch": 2.26,
+      "learning_rate": 1.3653001464128845e-05,
+      "loss": 0.4861,
+      "step": 1145
     },
     {
-      "epoch": 8.24,
-      "learning_rate": 9.70873786407767e-06,
-      "loss": 0.3339,
-      "step": 3400
+      "epoch": 2.27,
+      "learning_rate": 1.3469985358711567e-05,
+      "loss": 0.4579,
+      "step": 1150
     },
     {
-      "epoch": 8.27,
-      "learning_rate": 9.573894282632147e-06,
-      "loss": 0.3501,
-      "step": 3410
+      "epoch": 2.28,
+      "learning_rate": 1.328696925329429e-05,
+      "loss": 0.4574,
+      "step": 1155
     },
     {
-      "epoch": 8.29,
-      "learning_rate": 9.439050701186624e-06,
-      "loss": 0.3319,
-      "step": 3420
+      "epoch": 2.29,
+      "learning_rate": 1.3103953147877013e-05,
+      "loss": 0.5262,
+      "step": 1160
     },
     {
-      "epoch": 8.32,
-      "learning_rate": 9.304207119741101e-06,
-      "loss": 0.3193,
-      "step": 3430
+      "epoch": 2.3,
+      "learning_rate": 1.2920937042459739e-05,
+      "loss": 0.4991,
+      "step": 1165
     },
     {
-      "epoch": 8.34,
-      "learning_rate": 9.169363538295578e-06,
-      "loss": 0.3383,
-      "step": 3440
+      "epoch": 2.31,
+      "learning_rate": 1.2737920937042461e-05,
+      "loss": 0.5082,
+      "step": 1170
     },
     {
-      "epoch": 8.36,
-      "learning_rate": 9.034519956850054e-06,
-      "loss": 0.3369,
-      "step": 3450
+      "epoch": 2.32,
+      "learning_rate": 1.2554904831625182e-05,
+      "loss": 0.4557,
+      "step": 1175
     },
     {
-      "epoch": 8.39,
-      "learning_rate": 8.899676375404531e-06,
-      "loss": 0.3473,
-      "step": 3460
+      "epoch": 2.33,
+      "learning_rate": 1.2371888726207907e-05,
+      "loss": 0.5222,
+      "step": 1180
     },
     {
-      "epoch": 8.41,
-      "learning_rate": 8.764832793959007e-06,
-      "loss": 0.315,
-      "step": 3470
+      "epoch": 2.34,
+      "learning_rate": 1.218887262079063e-05,
+      "loss": 0.4846,
+      "step": 1185
     },
     {
-      "epoch": 8.44,
-      "learning_rate": 8.629989212513486e-06,
-      "loss": 0.3416,
-      "step": 3480
+      "epoch": 2.35,
+      "learning_rate": 1.2005856515373354e-05,
+      "loss": 0.4616,
+      "step": 1190
     },
     {
-      "epoch": 8.46,
-      "learning_rate": 8.495145631067963e-06,
-      "loss": 0.3297,
-      "step": 3490
+      "epoch": 2.36,
+      "learning_rate": 1.1822840409956077e-05,
+      "loss": 0.4609,
+      "step": 1195
     },
     {
-      "epoch": 8.48,
-      "learning_rate": 8.360302049622438e-06,
-      "loss": 0.3263,
-      "step": 3500
+      "epoch": 2.37,
+      "learning_rate": 1.1639824304538801e-05,
+      "loss": 0.501,
+      "step": 1200
     },
     {
-      "epoch": 8.51,
-      "learning_rate": 8.225458468176916e-06,
-      "loss": 0.3569,
-      "step": 3510
+      "epoch": 2.38,
+      "learning_rate": 1.1456808199121522e-05,
+      "loss": 0.4591,
+      "step": 1205
     },
     {
-      "epoch": 8.53,
-      "learning_rate": 8.090614886731391e-06,
-      "loss": 0.3299,
-      "step": 3520
+      "epoch": 2.39,
+      "learning_rate": 1.1273792093704246e-05,
+      "loss": 0.4996,
+      "step": 1210
     },
     {
-      "epoch": 8.56,
-      "learning_rate": 7.955771305285868e-06,
-      "loss": 0.3346,
-      "step": 3530
+      "epoch": 2.4,
+      "learning_rate": 1.109077598828697e-05,
+      "loss": 0.4629,
+      "step": 1215
     },
     {
-      "epoch": 8.58,
-      "learning_rate": 7.820927723840345e-06,
-      "loss": 0.3047,
-      "step": 3540
+      "epoch": 2.41,
+      "learning_rate": 1.0907759882869694e-05,
+      "loss": 0.4559,
+      "step": 1220
     },
     {
-      "epoch": 8.61,
-      "learning_rate": 7.686084142394823e-06,
-      "loss": 0.336,
-      "step": 3550
+      "epoch": 2.42,
+      "learning_rate": 1.0724743777452416e-05,
+      "loss": 0.4881,
+      "step": 1225
     },
     {
-      "epoch": 8.63,
-      "learning_rate": 7.551240560949299e-06,
-      "loss": 0.3207,
-      "step": 3560
+      "epoch": 2.43,
+      "learning_rate": 1.054172767203514e-05,
+      "loss": 0.4593,
+      "step": 1230
     },
     {
-      "epoch": 8.65,
-      "learning_rate": 7.416396979503775e-06,
-      "loss": 0.3171,
-      "step": 3570
+      "epoch": 2.44,
+      "learning_rate": 1.0358711566617862e-05,
+      "loss": 0.4847,
+      "step": 1235
     },
     {
-      "epoch": 8.68,
-      "learning_rate": 7.281553398058253e-06,
-      "loss": 0.3186,
-      "step": 3580
+      "epoch": 2.45,
+      "learning_rate": 1.0175695461200586e-05,
+      "loss": 0.4676,
+      "step": 1240
     },
     {
-      "epoch": 8.7,
-      "learning_rate": 7.146709816612731e-06,
-      "loss": 0.3307,
-      "step": 3590
+      "epoch": 2.46,
+      "learning_rate": 9.992679355783309e-06,
+      "loss": 0.4893,
+      "step": 1245
     },
     {
-      "epoch": 8.73,
-      "learning_rate": 7.011866235167207e-06,
-      "loss": 0.3409,
-      "step": 3600
+      "epoch": 2.47,
+      "learning_rate": 9.809663250366033e-06,
+      "loss": 0.4845,
+      "step": 1250
     },
     {
-      "epoch": 8.75,
-      "learning_rate": 6.877022653721683e-06,
-      "loss": 0.324,
-      "step": 3610
+      "epoch": 2.48,
+      "learning_rate": 9.626647144948756e-06,
+      "loss": 0.4829,
+      "step": 1255
     },
     {
-      "epoch": 8.78,
-      "learning_rate": 6.74217907227616e-06,
-      "loss": 0.3397,
-      "step": 3620
+      "epoch": 2.49,
+      "learning_rate": 9.44363103953148e-06,
+      "loss": 0.5043,
+      "step": 1260
     },
     {
-      "epoch": 8.8,
-      "learning_rate": 6.607335490830636e-06,
-      "loss": 0.3437,
-      "step": 3630
+      "epoch": 2.5,
+      "learning_rate": 9.260614934114201e-06,
+      "loss": 0.497,
+      "step": 1265
     },
     {
-      "epoch": 8.82,
-      "learning_rate": 6.472491909385113e-06,
-      "loss": 0.3441,
-      "step": 3640
+      "epoch": 2.51,
+      "learning_rate": 9.077598828696926e-06,
+      "loss": 0.5003,
+      "step": 1270
     },
     {
-      "epoch": 8.85,
-      "learning_rate": 6.3376483279395905e-06,
-      "loss": 0.3405,
-      "step": 3650
+      "epoch": 2.52,
+      "learning_rate": 8.894582723279648e-06,
+      "loss": 0.4462,
+      "step": 1275
     },
     {
-      "epoch": 8.87,
-      "learning_rate": 6.202804746494068e-06,
-      "loss": 0.3191,
-      "step": 3660
+      "epoch": 2.53,
+      "learning_rate": 8.711566617862373e-06,
+      "loss": 0.5062,
+      "step": 1280
     },
     {
-      "epoch": 8.9,
-      "learning_rate": 6.067961165048544e-06,
-      "loss": 0.334,
-      "step": 3670
+      "epoch": 2.54,
+      "learning_rate": 8.528550512445096e-06,
+      "loss": 0.4795,
+      "step": 1285
     },
     {
-      "epoch": 8.92,
-      "learning_rate": 5.9331175836030205e-06,
-      "loss": 0.3244,
-      "step": 3680
+      "epoch": 2.55,
+      "learning_rate": 8.34553440702782e-06,
+      "loss": 0.4773,
+      "step": 1290
     },
     {
-      "epoch": 8.95,
-      "learning_rate": 5.798274002157498e-06,
-      "loss": 0.3257,
-      "step": 3690
+      "epoch": 2.56,
+      "learning_rate": 8.162518301610541e-06,
+      "loss": 0.4667,
+      "step": 1295
     },
     {
-      "epoch": 8.97,
-      "learning_rate": 5.663430420711974e-06,
-      "loss": 0.3201,
-      "step": 3700
+      "epoch": 2.57,
+      "learning_rate": 7.979502196193265e-06,
+      "loss": 0.5078,
+      "step": 1300
     },
     {
-      "epoch": 8.99,
-      "learning_rate": 5.528586839266451e-06,
-      "loss": 0.3295,
-      "step": 3710
+      "epoch": 2.58,
+      "learning_rate": 7.796486090775988e-06,
+      "loss": 0.4586,
+      "step": 1305
     },
     {
-      "epoch": 9.0,
-      "eval_accuracy": 0.7788953290146607,
-      "eval_loss": 0.4606703519821167,
-      "eval_runtime": 43.9193,
-      "eval_samples_per_second": 133.563,
-      "eval_steps_per_second": 4.19,
-      "step": 3712
+      "epoch": 2.59,
+      "learning_rate": 7.613469985358712e-06,
+      "loss": 0.4328,
+      "step": 1310
     },
     {
-      "epoch": 9.02,
-      "learning_rate": 5.3937432578209285e-06,
-      "loss": 0.2991,
-      "step": 3720
+      "epoch": 2.6,
+      "learning_rate": 7.430453879941435e-06,
+      "loss": 0.443,
+      "step": 1315
     },
     {
-      "epoch": 9.04,
-      "learning_rate": 5.258899676375405e-06,
-      "loss": 0.2793,
-      "step": 3730
+      "epoch": 2.61,
+      "learning_rate": 7.247437774524159e-06,
+      "loss": 0.4597,
+      "step": 1320
     },
     {
-      "epoch": 9.07,
-      "learning_rate": 5.124056094929881e-06,
-      "loss": 0.3362,
-      "step": 3740
+      "epoch": 2.62,
+      "learning_rate": 7.064421669106881e-06,
+      "loss": 0.5333,
+      "step": 1325
     },
     {
-      "epoch": 9.09,
-      "learning_rate": 4.989212513484358e-06,
-      "loss": 0.3433,
-      "step": 3750
+      "epoch": 2.63,
+      "learning_rate": 6.881405563689605e-06,
+      "loss": 0.4654,
+      "step": 1330
     },
     {
-      "epoch": 9.12,
-      "learning_rate": 4.854368932038835e-06,
-      "loss": 0.2879,
-      "step": 3760
+      "epoch": 2.64,
+      "learning_rate": 6.698389458272328e-06,
+      "loss": 0.4383,
+      "step": 1335
     },
     {
-      "epoch": 9.14,
-      "learning_rate": 4.719525350593312e-06,
-      "loss": 0.3276,
-      "step": 3770
+      "epoch": 2.65,
+      "learning_rate": 6.515373352855052e-06,
+      "loss": 0.4828,
+      "step": 1340
     },
     {
-      "epoch": 9.16,
-      "learning_rate": 4.584681769147789e-06,
-      "loss": 0.3038,
-      "step": 3780
+      "epoch": 2.66,
+      "learning_rate": 6.332357247437774e-06,
+      "loss": 0.4849,
+      "step": 1345
     },
     {
-      "epoch": 9.19,
-      "learning_rate": 4.4498381877022656e-06,
-      "loss": 0.3275,
-      "step": 3790
+      "epoch": 2.67,
+      "learning_rate": 6.149341142020498e-06,
+      "loss": 0.465,
+      "step": 1350
     },
     {
-      "epoch": 9.21,
-      "learning_rate": 4.314994606256743e-06,
-      "loss": 0.3071,
-      "step": 3800
+      "epoch": 2.68,
+      "learning_rate": 5.966325036603222e-06,
+      "loss": 0.5227,
+      "step": 1355
     },
     {
-      "epoch": 9.24,
-      "learning_rate": 4.180151024811219e-06,
-      "loss": 0.3513,
-      "step": 3810
+      "epoch": 2.69,
+      "learning_rate": 5.7833089311859446e-06,
+      "loss": 0.5229,
+      "step": 1360
     },
     {
-      "epoch": 9.26,
-      "learning_rate": 4.0453074433656955e-06,
-      "loss": 0.3253,
-      "step": 3820
+      "epoch": 2.69,
+      "learning_rate": 5.600292825768668e-06,
+      "loss": 0.4592,
+      "step": 1365
     },
     {
-      "epoch": 9.28,
-      "learning_rate": 3.910463861920173e-06,
-      "loss": 0.3332,
-      "step": 3830
+      "epoch": 2.7,
+      "learning_rate": 5.417276720351392e-06,
+      "loss": 0.4769,
+      "step": 1370
     },
     {
-      "epoch": 9.31,
-      "learning_rate": 3.7756202804746495e-06,
-      "loss": 0.3215,
-      "step": 3840
+      "epoch": 2.71,
+      "learning_rate": 5.234260614934114e-06,
+      "loss": 0.4959,
+      "step": 1375
     },
     {
-      "epoch": 9.33,
-      "learning_rate": 3.6407766990291263e-06,
-      "loss": 0.3118,
-      "step": 3850
+      "epoch": 2.72,
+      "learning_rate": 5.051244509516838e-06,
+      "loss": 0.4731,
+      "step": 1380
     },
     {
-      "epoch": 9.36,
-      "learning_rate": 3.5059331175836035e-06,
-      "loss": 0.3036,
-      "step": 3860
+      "epoch": 2.73,
+      "learning_rate": 4.8682284040995615e-06,
+      "loss": 0.4718,
+      "step": 1385
     },
     {
-      "epoch": 9.38,
-      "learning_rate": 3.37108953613808e-06,
-      "loss": 0.3008,
-      "step": 3870
+      "epoch": 2.74,
+      "learning_rate": 4.685212298682284e-06,
+      "loss": 0.5007,
+      "step": 1390
     },
     {
-      "epoch": 9.41,
-      "learning_rate": 3.2362459546925567e-06,
-      "loss": 0.3381,
-      "step": 3880
+      "epoch": 2.75,
+      "learning_rate": 4.502196193265008e-06,
+      "loss": 0.437,
+      "step": 1395
     },
     {
-      "epoch": 9.43,
-      "learning_rate": 3.101402373247034e-06,
-      "loss": 0.3265,
-      "step": 3890
+      "epoch": 2.76,
+      "learning_rate": 4.319180087847731e-06,
+      "loss": 0.4732,
+      "step": 1400
     },
     {
-      "epoch": 9.45,
-      "learning_rate": 2.9665587918015102e-06,
-      "loss": 0.3314,
-      "step": 3900
+      "epoch": 2.77,
+      "learning_rate": 4.136163982430454e-06,
+      "loss": 0.4582,
+      "step": 1405
     },
     {
-      "epoch": 9.48,
-      "learning_rate": 2.831715210355987e-06,
-      "loss": 0.3242,
-      "step": 3910
+      "epoch": 2.78,
+      "learning_rate": 3.9531478770131775e-06,
+      "loss": 0.4903,
+      "step": 1410
     },
     {
-      "epoch": 9.5,
-      "learning_rate": 2.6968716289104642e-06,
-      "loss": 0.3089,
-      "step": 3920
+      "epoch": 2.79,
+      "learning_rate": 3.7701317715959007e-06,
+      "loss": 0.5049,
+      "step": 1415
     },
     {
-      "epoch": 9.53,
-      "learning_rate": 2.5620280474649406e-06,
-      "loss": 0.3296,
-      "step": 3930
+      "epoch": 2.8,
+      "learning_rate": 3.587115666178624e-06,
+      "loss": 0.457,
+      "step": 1420
     },
     {
-      "epoch": 9.55,
-      "learning_rate": 2.4271844660194174e-06,
-      "loss": 0.3257,
-      "step": 3940
+      "epoch": 2.81,
+      "learning_rate": 3.4040995607613473e-06,
+      "loss": 0.4516,
+      "step": 1425
     },
     {
-      "epoch": 9.58,
-      "learning_rate": 2.2923408845738946e-06,
-      "loss": 0.364,
-      "step": 3950
+      "epoch": 2.82,
+      "learning_rate": 3.2210834553440705e-06,
+      "loss": 0.5026,
+      "step": 1430
     },
     {
-      "epoch": 9.6,
-      "learning_rate": 2.1574973031283714e-06,
-      "loss": 0.3389,
-      "step": 3960
+      "epoch": 2.83,
+      "learning_rate": 3.0380673499267936e-06,
+      "loss": 0.5247,
+      "step": 1435
     },
     {
-      "epoch": 9.62,
-      "learning_rate": 2.0226537216828478e-06,
-      "loss": 0.3048,
-      "step": 3970
+      "epoch": 2.84,
+      "learning_rate": 2.855051244509517e-06,
+      "loss": 0.4605,
+      "step": 1440
     },
     {
-      "epoch": 9.65,
-      "learning_rate": 1.8878101402373248e-06,
-      "loss": 0.343,
-      "step": 3980
+      "epoch": 2.85,
+      "learning_rate": 2.6720351390922403e-06,
+      "loss": 0.4889,
+      "step": 1445
     },
     {
-      "epoch": 9.67,
-      "learning_rate": 1.7529665587918018e-06,
-      "loss": 0.3005,
-      "step": 3990
+      "epoch": 2.86,
+      "learning_rate": 2.4890190336749634e-06,
+      "loss": 0.4645,
+      "step": 1450
     },
     {
-      "epoch": 9.7,
-      "learning_rate": 1.6181229773462783e-06,
-      "loss": 0.3253,
-      "step": 4000
+      "epoch": 2.87,
+      "learning_rate": 2.306002928257687e-06,
+      "loss": 0.4493,
+      "step": 1455
     },
     {
-      "epoch": 9.72,
-      "learning_rate": 1.4832793959007551e-06,
-      "loss": 0.3117,
-      "step": 4010
+      "epoch": 2.88,
+      "learning_rate": 2.12298682284041e-06,
+      "loss": 0.492,
+      "step": 1460
     },
     {
-      "epoch": 9.75,
-      "learning_rate": 1.3484358144552321e-06,
-      "loss": 0.3244,
-      "step": 4020
+      "epoch": 2.89,
+      "learning_rate": 1.9399707174231332e-06,
+      "loss": 0.5066,
+      "step": 1465
     },
     {
-      "epoch": 9.77,
-      "learning_rate": 1.2135922330097087e-06,
-      "loss": 0.3004,
-      "step": 4030
+      "epoch": 2.9,
+      "learning_rate": 1.7569546120058566e-06,
+      "loss": 0.4364,
+      "step": 1470
     },
     {
-      "epoch": 9.79,
-      "learning_rate": 1.0787486515641857e-06,
-      "loss": 0.3204,
-      "step": 4040
+      "epoch": 2.91,
+      "learning_rate": 1.57393850658858e-06,
+      "loss": 0.4823,
+      "step": 1475
     },
     {
-      "epoch": 9.82,
-      "learning_rate": 9.439050701186624e-07,
-      "loss": 0.3185,
-      "step": 4050
+      "epoch": 2.92,
+      "learning_rate": 1.3909224011713032e-06,
+      "loss": 0.5101,
+      "step": 1480
     },
     {
-      "epoch": 9.84,
-      "learning_rate": 8.090614886731392e-07,
-      "loss": 0.3223,
-      "step": 4060
+      "epoch": 2.93,
+      "learning_rate": 1.2079062957540264e-06,
+      "loss": 0.5322,
+      "step": 1485
     },
     {
-      "epoch": 9.87,
-      "learning_rate": 6.742179072276161e-07,
-      "loss": 0.3153,
-      "step": 4070
+      "epoch": 2.94,
+      "learning_rate": 1.0248901903367497e-06,
+      "loss": 0.4745,
+      "step": 1490
     },
     {
-      "epoch": 9.89,
-      "learning_rate": 5.393743257820928e-07,
-      "loss": 0.3203,
-      "step": 4080
+      "epoch": 2.95,
+      "learning_rate": 8.41874084919473e-07,
+      "loss": 0.4728,
+      "step": 1495
     },
     {
-      "epoch": 9.92,
-      "learning_rate": 4.045307443365696e-07,
-      "loss": 0.3241,
-      "step": 4090
+      "epoch": 2.96,
+      "learning_rate": 6.588579795021963e-07,
+      "loss": 0.5055,
+      "step": 1500
     },
     {
-      "epoch": 9.94,
-      "learning_rate": 2.696871628910464e-07,
-      "loss": 0.2784,
-      "step": 4100
+      "epoch": 2.97,
+      "learning_rate": 4.758418740849195e-07,
+      "loss": 0.5092,
+      "step": 1505
     },
     {
-      "epoch": 9.96,
-      "learning_rate": 1.348435814455232e-07,
-      "loss": 0.328,
-      "step": 4110
+      "epoch": 2.98,
+      "learning_rate": 2.9282576866764276e-07,
+      "loss": 0.4562,
+      "step": 1510
     },
     {
-      "epoch": 9.99,
-      "learning_rate": 0.0,
-      "loss": 0.321,
-      "step": 4120
+      "epoch": 2.99,
+      "learning_rate": 1.0980966325036604e-07,
+      "loss": 0.4478,
+      "step": 1515
     },
     {
-      "epoch": 9.99,
-      "eval_accuracy": 0.7794067507671326,
-      "eval_loss": 0.45381179451942444,
-      "eval_runtime": 43.3941,
-      "eval_samples_per_second": 135.18,
-      "eval_steps_per_second": 4.24,
-      "step": 4120
+      "epoch": 3.0,
+      "eval_accuracy": 0.7774538386783285,
+      "eval_loss": 0.39165472984313965,
+      "eval_runtime": 56.6277,
+      "eval_samples_per_second": 127.199,
+      "eval_steps_per_second": 3.991,
+      "step": 1518
     },
     {
-      "epoch": 9.99,
-      "step": 4120,
-      "total_flos": 1.3106102295963156e+19,
-      "train_loss": 0.3501510629954847,
-      "train_runtime": 8665.4247,
-      "train_samples_per_second": 60.918,
-      "train_steps_per_second": 0.475
+      "epoch": 3.0,
+      "step": 1518,
+      "total_flos": 4.829589697691566e+18,
+      "train_loss": 0.5911478077470078,
+      "train_runtime": 3253.5777,
+      "train_samples_per_second": 59.774,
+      "train_steps_per_second": 0.467
     }
   ],
-  "logging_steps": 10,
-  "max_steps": 4120,
-  "num_train_epochs": 10,
+  "logging_steps": 5,
+  "max_steps": 1518,
+  "num_train_epochs": 3,
   "save_steps": 500,
-  "total_flos": 1.3106102295963156e+19,
+  "total_flos": 4.829589697691566e+18,
   "trial_name": null,
   "trial_params": null
 }