End of training

all_results.json

@@ -1,13 +1,13 @@
 {
-    "epoch": 4.99,
-    "eval_accuracy": 0.7797476986021139,
-    "eval_loss": 0.3965856432914734,
-    "eval_runtime": 44.3045,
-    "eval_samples_per_second": 132.402,
-    "eval_steps_per_second": 4.153,
-    "total_flos": 6.553200292249928e+18,
-    "train_loss": 0.4588474294514332,
-    "train_runtime": 4338.0094,
-    "train_samples_per_second": 60.844,
+    "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
 }

eval_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 4.99,
-    "eval_accuracy": 0.7797476986021139,
-    "eval_loss": 0.3965856432914734,
-    "eval_runtime": 44.3045,
-    "eval_samples_per_second": 132.402,
-    "eval_steps_per_second": 4.153
+    "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
 }

train_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 4.99,
-    "total_flos": 6.553200292249928e+18,
-    "train_loss": 0.4588474294514332,
-    "train_runtime": 4338.0094,
-    "train_samples_per_second": 60.844,
+    "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
 }

trainer_state.json

@@ -1,1309 +1,2590 @@
 {
-  "best_metric": 0.7797476986021139,
-  "best_model_checkpoint": "swin-tiny-patch4-window7-224-img_orientation/checkpoint-2060",
-  "epoch": 4.993939393939394,
+  "best_metric": 0.7794067507671326,
+  "best_model_checkpoint": "swin-tiny-patch4-window7-224-img_orientation/checkpoint-4120",
+  "epoch": 9.987878787878788,
   "eval_steps": 500,
-  "global_step": 2060,
+  "global_step": 4120,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
       "epoch": 0.02,
-      "learning_rate": 2.4271844660194174e-06,
-      "loss": 0.5485,
+      "learning_rate": 1.2135922330097087e-06,
+      "loss": 0.4514,
       "step": 10
     },
     {
       "epoch": 0.05,
-      "learning_rate": 4.854368932038835e-06,
-      "loss": 0.524,
+      "learning_rate": 2.4271844660194174e-06,
+      "loss": 0.4198,
       "step": 20
     },
     {
       "epoch": 0.07,
-      "learning_rate": 7.281553398058253e-06,
-      "loss": 0.5436,
+      "learning_rate": 3.6407766990291263e-06,
+      "loss": 0.4392,
       "step": 30
     },
     {
       "epoch": 0.1,
-      "learning_rate": 9.70873786407767e-06,
-      "loss": 0.5309,
+      "learning_rate": 4.854368932038835e-06,
+      "loss": 0.4007,
       "step": 40
     },
     {
       "epoch": 0.12,
-      "learning_rate": 1.2135922330097088e-05,
-      "loss": 0.5218,
+      "learning_rate": 6.067961165048544e-06,
+      "loss": 0.4043,
       "step": 50
     },
     {
       "epoch": 0.15,
-      "learning_rate": 1.4563106796116505e-05,
-      "loss": 0.4976,
+      "learning_rate": 7.281553398058253e-06,
+      "loss": 0.3604,
       "step": 60
     },
     {
       "epoch": 0.17,
-      "learning_rate": 1.6990291262135926e-05,
-      "loss": 0.5168,
+      "learning_rate": 8.495145631067963e-06,
+      "loss": 0.383,
       "step": 70
     },
     {
       "epoch": 0.19,
-      "learning_rate": 1.941747572815534e-05,
-      "loss": 0.5168,
+      "learning_rate": 9.70873786407767e-06,
+      "loss": 0.3708,
       "step": 80
     },
     {
       "epoch": 0.22,
-      "learning_rate": 2.1844660194174756e-05,
-      "loss": 0.4882,
+      "learning_rate": 1.0922330097087378e-05,
+      "loss": 0.3486,
       "step": 90
     },
     {
       "epoch": 0.24,
-      "learning_rate": 2.4271844660194176e-05,
-      "loss": 0.5174,
+      "learning_rate": 1.2135922330097088e-05,
+      "loss": 0.367,
       "step": 100
     },
     {
       "epoch": 0.27,
-      "learning_rate": 2.6699029126213593e-05,
-      "loss": 0.5012,
+      "learning_rate": 1.3349514563106797e-05,
+      "loss": 0.3288,
       "step": 110
     },
     {
       "epoch": 0.29,
-      "learning_rate": 2.912621359223301e-05,
-      "loss": 0.462,
+      "learning_rate": 1.4563106796116505e-05,
+      "loss": 0.3155,
       "step": 120
     },
     {
       "epoch": 0.32,
-      "learning_rate": 3.155339805825243e-05,
-      "loss": 0.4785,
+      "learning_rate": 1.5776699029126215e-05,
+      "loss": 0.3364,
       "step": 130
     },
     {
       "epoch": 0.34,
-      "learning_rate": 3.398058252427185e-05,
-      "loss": 0.47,
+      "learning_rate": 1.6990291262135926e-05,
+      "loss": 0.3258,
       "step": 140
     },
     {
       "epoch": 0.36,
-      "learning_rate": 3.6407766990291265e-05,
-      "loss": 0.4878,
+      "learning_rate": 1.8203883495145632e-05,
+      "loss": 0.3078,
       "step": 150
     },
     {
       "epoch": 0.39,
-      "learning_rate": 3.883495145631068e-05,
-      "loss": 0.4648,
+      "learning_rate": 1.941747572815534e-05,
+      "loss": 0.293,
       "step": 160
     },
     {
       "epoch": 0.41,
-      "learning_rate": 4.12621359223301e-05,
-      "loss": 0.4636,
+      "learning_rate": 2.063106796116505e-05,
+      "loss": 0.2877,
       "step": 170
     },
     {
       "epoch": 0.44,
-      "learning_rate": 4.368932038834951e-05,
-      "loss": 0.4862,
+      "learning_rate": 2.1844660194174756e-05,
+      "loss": 0.3103,
       "step": 180
     },
     {
       "epoch": 0.46,
-      "learning_rate": 4.611650485436894e-05,
-      "loss": 0.4594,
+      "learning_rate": 2.305825242718447e-05,
+      "loss": 0.2944,
       "step": 190
     },
     {
       "epoch": 0.48,
-      "learning_rate": 4.854368932038835e-05,
-      "loss": 0.4882,
+      "learning_rate": 2.4271844660194176e-05,
+      "loss": 0.2986,
       "step": 200
     },
     {
       "epoch": 0.51,
-      "learning_rate": 4.9892125134843584e-05,
-      "loss": 0.4429,
+      "learning_rate": 2.5485436893203883e-05,
+      "loss": 0.2752,
       "step": 210
     },
     {
       "epoch": 0.53,
-      "learning_rate": 4.962243797195254e-05,
-      "loss": 0.4683,
+      "learning_rate": 2.6699029126213593e-05,
+      "loss": 0.3086,
       "step": 220
     },
     {
       "epoch": 0.56,
-      "learning_rate": 4.935275080906149e-05,
-      "loss": 0.4689,
+      "learning_rate": 2.79126213592233e-05,
+      "loss": 0.313,
       "step": 230
     },
     {
       "epoch": 0.58,
-      "learning_rate": 4.9083063646170444e-05,
-      "loss": 0.4786,
+      "learning_rate": 2.912621359223301e-05,
+      "loss": 0.2864,
       "step": 240
     },
     {
       "epoch": 0.61,
-      "learning_rate": 4.88133764832794e-05,
-      "loss": 0.5182,
+      "learning_rate": 3.0339805825242717e-05,
+      "loss": 0.3166,
       "step": 250
     },
     {
       "epoch": 0.63,
-      "learning_rate": 4.854368932038835e-05,
-      "loss": 0.4701,
+      "learning_rate": 3.155339805825243e-05,
+      "loss": 0.292,
       "step": 260
     },
     {
       "epoch": 0.65,
-      "learning_rate": 4.8274002157497304e-05,
-      "loss": 0.4795,
+      "learning_rate": 3.276699029126214e-05,
+      "loss": 0.3025,
       "step": 270
     },
     {
       "epoch": 0.68,
-      "learning_rate": 4.800431499460626e-05,
-      "loss": 0.4394,
+      "learning_rate": 3.398058252427185e-05,
+      "loss": 0.2801,
       "step": 280
     },
     {
       "epoch": 0.7,
-      "learning_rate": 4.773462783171521e-05,
-      "loss": 0.4252,
+      "learning_rate": 3.519417475728156e-05,
+      "loss": 0.2705,
       "step": 290
     },
     {
       "epoch": 0.73,
-      "learning_rate": 4.7464940668824164e-05,
-      "loss": 0.4688,
+      "learning_rate": 3.6407766990291265e-05,
+      "loss": 0.2895,
       "step": 300
     },
     {
       "epoch": 0.75,
-      "learning_rate": 4.7195253505933115e-05,
-      "loss": 0.4879,
+      "learning_rate": 3.762135922330097e-05,
+      "loss": 0.3195,
       "step": 310
     },
     {
       "epoch": 0.78,
-      "learning_rate": 4.692556634304207e-05,
-      "loss": 0.4517,
+      "learning_rate": 3.883495145631068e-05,
+      "loss": 0.3023,
       "step": 320
     },
     {
       "epoch": 0.8,
-      "learning_rate": 4.665587918015103e-05,
-      "loss": 0.4492,
+      "learning_rate": 4.004854368932039e-05,
+      "loss": 0.3075,
       "step": 330
     },
     {
       "epoch": 0.82,
-      "learning_rate": 4.638619201725998e-05,
-      "loss": 0.4738,
+      "learning_rate": 4.12621359223301e-05,
+      "loss": 0.3065,
       "step": 340
     },
     {
       "epoch": 0.85,
-      "learning_rate": 4.611650485436894e-05,
-      "loss": 0.4579,
+      "learning_rate": 4.2475728155339805e-05,
+      "loss": 0.2892,
       "step": 350
     },
     {
       "epoch": 0.87,
-      "learning_rate": 4.584681769147789e-05,
-      "loss": 0.4413,
+      "learning_rate": 4.368932038834951e-05,
+      "loss": 0.2819,
       "step": 360
     },
     {
       "epoch": 0.9,
-      "learning_rate": 4.557713052858684e-05,
-      "loss": 0.508,
+      "learning_rate": 4.4902912621359226e-05,
+      "loss": 0.35,
       "step": 370
     },
     {
       "epoch": 0.92,
-      "learning_rate": 4.530744336569579e-05,
-      "loss": 0.4605,
+      "learning_rate": 4.611650485436894e-05,
+      "loss": 0.3134,
       "step": 380
     },
     {
       "epoch": 0.95,
-      "learning_rate": 4.503775620280475e-05,
-      "loss": 0.4459,
+      "learning_rate": 4.7330097087378646e-05,
+      "loss": 0.3159,
       "step": 390
     },
     {
       "epoch": 0.97,
-      "learning_rate": 4.47680690399137e-05,
-      "loss": 0.4568,
+      "learning_rate": 4.854368932038835e-05,
+      "loss": 0.3149,
       "step": 400
     },
     {
       "epoch": 0.99,
-      "learning_rate": 4.449838187702265e-05,
-      "loss": 0.4741,
+      "learning_rate": 4.975728155339806e-05,
+      "loss": 0.3217,
       "step": 410
     },
     {
       "epoch": 1.0,
-      "eval_accuracy": 0.7345721104670986,
-      "eval_loss": 0.47738537192344666,
-      "eval_runtime": 44.109,
-      "eval_samples_per_second": 132.989,
-      "eval_steps_per_second": 4.171,
+      "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": 4.422869471413161e-05,
-      "loss": 0.4517,
+      "learning_rate": 4.9892125134843584e-05,
+      "loss": 0.2984,
       "step": 420
     },
     {
       "epoch": 1.04,
-      "learning_rate": 4.395900755124056e-05,
-      "loss": 0.4245,
+      "learning_rate": 4.975728155339806e-05,
+      "loss": 0.2742,
       "step": 430
     },
     {
       "epoch": 1.07,
-      "learning_rate": 4.368932038834951e-05,
-      "loss": 0.4318,
+      "learning_rate": 4.962243797195254e-05,
+      "loss": 0.2608,
       "step": 440
     },
     {
       "epoch": 1.09,
-      "learning_rate": 4.341963322545847e-05,
-      "loss": 0.4309,
+      "learning_rate": 4.948759439050701e-05,
+      "loss": 0.2663,
       "step": 450
     },
     {
       "epoch": 1.12,
-      "learning_rate": 4.314994606256743e-05,
-      "loss": 0.4387,
+      "learning_rate": 4.935275080906149e-05,
+      "loss": 0.3096,
       "step": 460
     },
     {
       "epoch": 1.14,
-      "learning_rate": 4.288025889967638e-05,
-      "loss": 0.4605,
+      "learning_rate": 4.921790722761597e-05,
+      "loss": 0.3046,
       "step": 470
     },
     {
       "epoch": 1.16,
-      "learning_rate": 4.261057173678533e-05,
-      "loss": 0.4623,
+      "learning_rate": 4.9083063646170444e-05,
+      "loss": 0.3038,
       "step": 480
     },
     {
       "epoch": 1.19,
-      "learning_rate": 4.234088457389429e-05,
-      "loss": 0.4471,
+      "learning_rate": 4.894822006472492e-05,
+      "loss": 0.2892,
       "step": 490
     },
     {
       "epoch": 1.21,
-      "learning_rate": 4.207119741100324e-05,
-      "loss": 0.4685,
+      "learning_rate": 4.88133764832794e-05,
+      "loss": 0.3028,
       "step": 500
     },
     {
       "epoch": 1.24,
-      "learning_rate": 4.180151024811219e-05,
-      "loss": 0.4549,
+      "learning_rate": 4.867853290183387e-05,
+      "loss": 0.3175,
       "step": 510
     },
     {
       "epoch": 1.26,
-      "learning_rate": 4.153182308522115e-05,
-      "loss": 0.4197,
+      "learning_rate": 4.854368932038835e-05,
+      "loss": 0.2881,
       "step": 520
     },
     {
       "epoch": 1.28,
-      "learning_rate": 4.12621359223301e-05,
-      "loss": 0.4243,
+      "learning_rate": 4.840884573894283e-05,
+      "loss": 0.3002,
       "step": 530
     },
     {
       "epoch": 1.31,
-      "learning_rate": 4.099244875943905e-05,
-      "loss": 0.4701,
+      "learning_rate": 4.8274002157497304e-05,
+      "loss": 0.297,
       "step": 540
     },
     {
       "epoch": 1.33,
-      "learning_rate": 4.0722761596548e-05,
-      "loss": 0.4475,
+      "learning_rate": 4.8139158576051786e-05,
+      "loss": 0.3127,
       "step": 550
     },
     {
       "epoch": 1.36,
-      "learning_rate": 4.0453074433656965e-05,
-      "loss": 0.444,
+      "learning_rate": 4.800431499460626e-05,
+      "loss": 0.2924,
       "step": 560
     },
     {
       "epoch": 1.38,
-      "learning_rate": 4.0183387270765916e-05,
-      "loss": 0.4231,
+      "learning_rate": 4.786947141316074e-05,
+      "loss": 0.2823,
       "step": 570
     },
     {
       "epoch": 1.41,
-      "learning_rate": 3.991370010787487e-05,
-      "loss": 0.431,
+      "learning_rate": 4.773462783171521e-05,
+      "loss": 0.2947,
       "step": 580
     },
     {
       "epoch": 1.43,
-      "learning_rate": 3.964401294498382e-05,
-      "loss": 0.4441,
+      "learning_rate": 4.759978425026969e-05,
+      "loss": 0.3045,
       "step": 590
     },
     {
       "epoch": 1.45,
-      "learning_rate": 3.9374325782092776e-05,
-      "loss": 0.466,
+      "learning_rate": 4.7464940668824164e-05,
+      "loss": 0.3217,
       "step": 600
     },
     {
       "epoch": 1.48,
-      "learning_rate": 3.910463861920173e-05,
-      "loss": 0.4114,
+      "learning_rate": 4.7330097087378646e-05,
+      "loss": 0.2629,
       "step": 610
     },
     {
       "epoch": 1.5,
-      "learning_rate": 3.883495145631068e-05,
-      "loss": 0.4523,
+      "learning_rate": 4.7195253505933115e-05,
+      "loss": 0.2978,
       "step": 620
     },
     {
       "epoch": 1.53,
-      "learning_rate": 3.8565264293419636e-05,
-      "loss": 0.4632,
+      "learning_rate": 4.70604099244876e-05,
+      "loss": 0.3395,
       "step": 630
     },
     {
       "epoch": 1.55,
-      "learning_rate": 3.829557713052859e-05,
-      "loss": 0.4373,
+      "learning_rate": 4.692556634304207e-05,
+      "loss": 0.3125,
       "step": 640
     },
     {
       "epoch": 1.58,
-      "learning_rate": 3.802588996763754e-05,
-      "loss": 0.4487,
+      "learning_rate": 4.679072276159655e-05,
+      "loss": 0.3282,
       "step": 650
     },
     {
       "epoch": 1.6,
-      "learning_rate": 3.7756202804746496e-05,
-      "loss": 0.4818,
+      "learning_rate": 4.665587918015103e-05,
+      "loss": 0.3308,
       "step": 660
     },
     {
       "epoch": 1.62,
-      "learning_rate": 3.7486515641855454e-05,
-      "loss": 0.4461,
+      "learning_rate": 4.6521035598705506e-05,
+      "loss": 0.3026,
       "step": 670
     },
     {
       "epoch": 1.65,
-      "learning_rate": 3.7216828478964405e-05,
-      "loss": 0.4475,
+      "learning_rate": 4.638619201725998e-05,
+      "loss": 0.3007,
       "step": 680
     },
     {
       "epoch": 1.67,
-      "learning_rate": 3.6947141316073356e-05,
-      "loss": 0.469,
+      "learning_rate": 4.625134843581446e-05,
+      "loss": 0.3106,
       "step": 690
     },
     {
       "epoch": 1.7,
-      "learning_rate": 3.6677454153182314e-05,
-      "loss": 0.448,
+      "learning_rate": 4.611650485436894e-05,
+      "loss": 0.3172,
       "step": 700
     },
     {
       "epoch": 1.72,
-      "learning_rate": 3.6407766990291265e-05,
-      "loss": 0.4411,
+      "learning_rate": 4.598166127292341e-05,
+      "loss": 0.3423,
       "step": 710
     },
     {
       "epoch": 1.75,
-      "learning_rate": 3.6138079827400216e-05,
-      "loss": 0.4529,
+      "learning_rate": 4.584681769147789e-05,
+      "loss": 0.3383,
       "step": 720
     },
     {
       "epoch": 1.77,
-      "learning_rate": 3.5868392664509174e-05,
-      "loss": 0.4553,
+      "learning_rate": 4.5711974110032366e-05,
+      "loss": 0.3006,
       "step": 730
     },
     {
       "epoch": 1.79,
-      "learning_rate": 3.5598705501618125e-05,
-      "loss": 0.4558,
+      "learning_rate": 4.557713052858684e-05,
+      "loss": 0.3038,
       "step": 740
     },
     {
       "epoch": 1.82,
-      "learning_rate": 3.5329018338727076e-05,
-      "loss": 0.4976,
+      "learning_rate": 4.544228694714132e-05,
+      "loss": 0.3614,
       "step": 750
     },
     {
       "epoch": 1.84,
-      "learning_rate": 3.505933117583603e-05,
-      "loss": 0.4474,
+      "learning_rate": 4.530744336569579e-05,
+      "loss": 0.3106,
       "step": 760
     },
     {
       "epoch": 1.87,
-      "learning_rate": 3.4789644012944984e-05,
-      "loss": 0.4646,
+      "learning_rate": 4.5172599784250275e-05,
+      "loss": 0.327,
       "step": 770
     },
     {
       "epoch": 1.89,
-      "learning_rate": 3.451995685005394e-05,
-      "loss": 0.4537,
+      "learning_rate": 4.503775620280475e-05,
+      "loss": 0.3241,
       "step": 780
     },
     {
       "epoch": 1.92,
-      "learning_rate": 3.425026968716289e-05,
-      "loss": 0.4649,
+      "learning_rate": 4.4902912621359226e-05,
+      "loss": 0.3476,
       "step": 790
     },
     {
       "epoch": 1.94,
-      "learning_rate": 3.398058252427185e-05,
-      "loss": 0.4751,
+      "learning_rate": 4.47680690399137e-05,
+      "loss": 0.3401,
       "step": 800
     },
     {
       "epoch": 1.96,
-      "learning_rate": 3.37108953613808e-05,
-      "loss": 0.4954,
+      "learning_rate": 4.4633225458468184e-05,
+      "loss": 0.3372,
       "step": 810
     },
     {
       "epoch": 1.99,
-      "learning_rate": 3.344120819848975e-05,
-      "loss": 0.4958,
+      "learning_rate": 4.449838187702265e-05,
+      "loss": 0.3716,
       "step": 820
     },
     {
       "epoch": 2.0,
-      "eval_accuracy": 0.7569041936583703,
-      "eval_loss": 0.440324068069458,
-      "eval_runtime": 44.2872,
-      "eval_samples_per_second": 132.454,
-      "eval_steps_per_second": 4.155,
+      "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": 3.3171521035598704e-05,
-      "loss": 0.4511,
+      "learning_rate": 4.4363538295577135e-05,
+      "loss": 0.3225,
       "step": 830
     },
     {
       "epoch": 2.04,
-      "learning_rate": 3.290183387270766e-05,
-      "loss": 0.4175,
+      "learning_rate": 4.422869471413161e-05,
+      "loss": 0.307,
       "step": 840
     },
     {
       "epoch": 2.06,
-      "learning_rate": 3.263214670981661e-05,
-      "loss": 0.4511,
+      "learning_rate": 4.4093851132686086e-05,
+      "loss": 0.3115,
       "step": 850
     },
     {
       "epoch": 2.08,
-      "learning_rate": 3.2362459546925564e-05,
-      "loss": 0.4188,
+      "learning_rate": 4.395900755124056e-05,
+      "loss": 0.2849,
       "step": 860
     },
     {
       "epoch": 2.11,
-      "learning_rate": 3.209277238403452e-05,
-      "loss": 0.453,
+      "learning_rate": 4.3824163969795043e-05,
+      "loss": 0.3344,
       "step": 870
     },
     {
       "epoch": 2.13,
-      "learning_rate": 3.182308522114347e-05,
-      "loss": 0.4392,
+      "learning_rate": 4.368932038834951e-05,
+      "loss": 0.3153,
       "step": 880
     },
     {
       "epoch": 2.16,
-      "learning_rate": 3.155339805825243e-05,
-      "loss": 0.4585,
+      "learning_rate": 4.3554476806903994e-05,
+      "loss": 0.3125,
       "step": 890
     },
     {
       "epoch": 2.18,
-      "learning_rate": 3.128371089536138e-05,
-      "loss": 0.434,
+      "learning_rate": 4.341963322545847e-05,
+      "loss": 0.2922,
       "step": 900
     },
     {
       "epoch": 2.21,
-      "learning_rate": 3.101402373247034e-05,
-      "loss": 0.4336,
+      "learning_rate": 4.3284789644012946e-05,
+      "loss": 0.3061,
       "step": 910
     },
     {
       "epoch": 2.23,
-      "learning_rate": 3.074433656957929e-05,
-      "loss": 0.446,
+      "learning_rate": 4.314994606256743e-05,
+      "loss": 0.3327,
       "step": 920
     },
     {
       "epoch": 2.25,
-      "learning_rate": 3.0474649406688245e-05,
-      "loss": 0.4559,
+      "learning_rate": 4.3015102481121897e-05,
+      "loss": 0.3186,
       "step": 930
     },
     {
       "epoch": 2.28,
-      "learning_rate": 3.0204962243797196e-05,
-      "loss": 0.4529,
+      "learning_rate": 4.288025889967638e-05,
+      "loss": 0.3247,
       "step": 940
     },
     {
       "epoch": 2.3,
-      "learning_rate": 2.993527508090615e-05,
-      "loss": 0.4681,
+      "learning_rate": 4.2745415318230854e-05,
+      "loss": 0.3278,
       "step": 950
     },
     {
       "epoch": 2.33,
-      "learning_rate": 2.96655879180151e-05,
-      "loss": 0.4512,
+      "learning_rate": 4.261057173678533e-05,
+      "loss": 0.3081,
       "step": 960
     },
     {
       "epoch": 2.35,
-      "learning_rate": 2.9395900755124056e-05,
-      "loss": 0.4373,
+      "learning_rate": 4.2475728155339805e-05,
+      "loss": 0.2979,
       "step": 970
     },
     {
       "epoch": 2.38,
-      "learning_rate": 2.912621359223301e-05,
-      "loss": 0.4706,
+      "learning_rate": 4.234088457389429e-05,
+      "loss": 0.3394,
       "step": 980
     },
     {
       "epoch": 2.4,
-      "learning_rate": 2.885652642934196e-05,
-      "loss": 0.4582,
+      "learning_rate": 4.2206040992448756e-05,
+      "loss": 0.3274,
       "step": 990
     },
     {
       "epoch": 2.42,
-      "learning_rate": 2.8586839266450923e-05,
-      "loss": 0.4442,
+      "learning_rate": 4.207119741100324e-05,
+      "loss": 0.3028,
       "step": 1000
     },
     {
       "epoch": 2.45,
-      "learning_rate": 2.8317152103559874e-05,
-      "loss": 0.4593,
+      "learning_rate": 4.1936353829557714e-05,
+      "loss": 0.3234,
       "step": 1010
     },
     {
       "epoch": 2.47,
-      "learning_rate": 2.8047464940668828e-05,
-      "loss": 0.4525,
+      "learning_rate": 4.180151024811219e-05,
+      "loss": 0.3177,
       "step": 1020
     },
     {
       "epoch": 2.5,
-      "learning_rate": 2.777777777777778e-05,
-      "loss": 0.4512,
+      "learning_rate": 4.166666666666667e-05,
+      "loss": 0.3377,
       "step": 1030
     },
     {
       "epoch": 2.52,
-      "learning_rate": 2.7508090614886734e-05,
-      "loss": 0.4668,
+      "learning_rate": 4.153182308522115e-05,
+      "loss": 0.351,
       "step": 1040
     },
     {
       "epoch": 2.55,
-      "learning_rate": 2.7238403451995688e-05,
-      "loss": 0.4669,
+      "learning_rate": 4.139697950377562e-05,
+      "loss": 0.3438,
       "step": 1050
     },
     {
       "epoch": 2.57,
-      "learning_rate": 2.696871628910464e-05,
-      "loss": 0.4513,
+      "learning_rate": 4.12621359223301e-05,
+      "loss": 0.3125,
       "step": 1060
     },
     {
       "epoch": 2.59,
-      "learning_rate": 2.6699029126213593e-05,
-      "loss": 0.4727,
+      "learning_rate": 4.1127292340884574e-05,
+      "loss": 0.3392,
       "step": 1070
     },
     {
       "epoch": 2.62,
-      "learning_rate": 2.6429341963322544e-05,
-      "loss": 0.4638,
+      "learning_rate": 4.099244875943905e-05,
+      "loss": 0.333,
       "step": 1080
     },
     {
       "epoch": 2.64,
-      "learning_rate": 2.61596548004315e-05,
-      "loss": 0.473,
+      "learning_rate": 4.085760517799353e-05,
+      "loss": 0.3557,
       "step": 1090
     },
     {
       "epoch": 2.67,
-      "learning_rate": 2.5889967637540453e-05,
-      "loss": 0.4535,
+      "learning_rate": 4.0722761596548e-05,
+      "loss": 0.3162,
       "step": 1100
     },
     {
       "epoch": 2.69,
-      "learning_rate": 2.5620280474649404e-05,
-      "loss": 0.4874,
+      "learning_rate": 4.058791801510248e-05,
+      "loss": 0.3393,
       "step": 1110
     },
     {
       "epoch": 2.72,
-      "learning_rate": 2.5350593311758362e-05,
-      "loss": 0.4746,
+      "learning_rate": 4.0453074433656965e-05,
+      "loss": 0.3408,
       "step": 1120
     },
     {
       "epoch": 2.74,
-      "learning_rate": 2.5080906148867317e-05,
-      "loss": 0.4474,
+      "learning_rate": 4.0318230852211434e-05,
+      "loss": 0.3223,
       "step": 1130
     },
     {
       "epoch": 2.76,
-      "learning_rate": 2.481121898597627e-05,
-      "loss": 0.4717,
+      "learning_rate": 4.0183387270765916e-05,
+      "loss": 0.3516,
       "step": 1140
     },
     {
       "epoch": 2.79,
-      "learning_rate": 2.4541531823085222e-05,
-      "loss": 0.4634,
+      "learning_rate": 4.004854368932039e-05,
+      "loss": 0.348,
       "step": 1150
     },
     {
       "epoch": 2.81,
-      "learning_rate": 2.4271844660194176e-05,
-      "loss": 0.478,
+      "learning_rate": 3.991370010787487e-05,
+      "loss": 0.3358,
       "step": 1160
     },
     {
       "epoch": 2.84,
-      "learning_rate": 2.400215749730313e-05,
-      "loss": 0.5012,
+      "learning_rate": 3.977885652642934e-05,
+      "loss": 0.3781,
       "step": 1170
     },
     {
       "epoch": 2.86,
-      "learning_rate": 2.3732470334412082e-05,
-      "loss": 0.4592,
+      "learning_rate": 3.964401294498382e-05,
+      "loss": 0.358,
       "step": 1180
     },
     {
       "epoch": 2.88,
-      "learning_rate": 2.3462783171521036e-05,
-      "loss": 0.519,
+      "learning_rate": 3.9509169363538294e-05,
+      "loss": 0.3959,
       "step": 1190
     },
     {
       "epoch": 2.91,
-      "learning_rate": 2.319309600862999e-05,
-      "loss": 0.5202,
+      "learning_rate": 3.9374325782092776e-05,
+      "loss": 0.4092,
       "step": 1200
     },
     {
       "epoch": 2.93,
-      "learning_rate": 2.2923408845738945e-05,
-      "loss": 0.4819,
+      "learning_rate": 3.923948220064725e-05,
+      "loss": 0.3731,
       "step": 1210
     },
     {
       "epoch": 2.96,
-      "learning_rate": 2.2653721682847896e-05,
-      "loss": 0.4969,
+      "learning_rate": 3.910463861920173e-05,
+      "loss": 0.388,
       "step": 1220
     },
     {
       "epoch": 2.98,
-      "learning_rate": 2.238403451995685e-05,
-      "loss": 0.473,
+      "learning_rate": 3.896979503775621e-05,
+      "loss": 0.3756,
       "step": 1230
     },
     {
       "epoch": 3.0,
-      "eval_accuracy": 0.7659393112853733,
-      "eval_loss": 0.40928295254707336,
-      "eval_runtime": 44.1748,
-      "eval_samples_per_second": 132.79,
-      "eval_steps_per_second": 4.165,
+      "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": 3.01,
-      "learning_rate": 2.2114347357065805e-05,
-      "loss": 0.4556,
+      "learning_rate": 3.883495145631068e-05,
+      "loss": 0.3399,
       "step": 1240
     },
     {
       "epoch": 3.03,
-      "learning_rate": 2.1844660194174756e-05,
-      "loss": 0.4626,
+      "learning_rate": 3.870010787486516e-05,
+      "loss": 0.3321,
       "step": 1250
     },
     {
       "epoch": 3.05,
-      "learning_rate": 2.1574973031283714e-05,
-      "loss": 0.4897,
+      "learning_rate": 3.8565264293419636e-05,
+      "loss": 0.3665,
       "step": 1260
     },
     {
       "epoch": 3.08,
-      "learning_rate": 2.1305285868392665e-05,
-      "loss": 0.4705,
+      "learning_rate": 3.843042071197411e-05,
+      "loss": 0.3489,
       "step": 1270
     },
     {
       "epoch": 3.1,
-      "learning_rate": 2.103559870550162e-05,
-      "loss": 0.4876,
+      "learning_rate": 3.829557713052859e-05,
+      "loss": 0.3717,
       "step": 1280
     },
     {
       "epoch": 3.13,
-      "learning_rate": 2.0765911542610574e-05,
-      "loss": 0.4844,
+      "learning_rate": 3.816073354908307e-05,
+      "loss": 0.363,
       "step": 1290
     },
     {
       "epoch": 3.15,
-      "learning_rate": 2.0496224379719525e-05,
-      "loss": 0.4737,
+      "learning_rate": 3.802588996763754e-05,
+      "loss": 0.3485,
       "step": 1300
     },
     {
       "epoch": 3.18,
-      "learning_rate": 2.0226537216828483e-05,
-      "loss": 0.4978,
+      "learning_rate": 3.789104638619202e-05,
+      "loss": 0.4041,
       "step": 1310
     },
     {
       "epoch": 3.2,
-      "learning_rate": 1.9956850053937434e-05,
-      "loss": 0.4744,
+      "learning_rate": 3.7756202804746496e-05,
+      "loss": 0.3738,
       "step": 1320
     },
     {
       "epoch": 3.22,
-      "learning_rate": 1.9687162891046388e-05,
-      "loss": 0.4588,
+      "learning_rate": 3.762135922330097e-05,
+      "loss": 0.3573,
       "step": 1330
     },
     {
       "epoch": 3.25,
-      "learning_rate": 1.941747572815534e-05,
-      "loss": 0.4344,
+      "learning_rate": 3.7486515641855454e-05,
+      "loss": 0.3255,
       "step": 1340
     },
     {
       "epoch": 3.27,
-      "learning_rate": 1.9147788565264294e-05,
-      "loss": 0.4651,
+      "learning_rate": 3.735167206040992e-05,
+      "loss": 0.3402,
       "step": 1350
     },
     {
       "epoch": 3.3,
-      "learning_rate": 1.8878101402373248e-05,
-      "loss": 0.4813,
+      "learning_rate": 3.7216828478964405e-05,
+      "loss": 0.3754,
       "step": 1360
     },
     {
       "epoch": 3.32,
-      "learning_rate": 1.8608414239482202e-05,
-      "loss": 0.4614,
+      "learning_rate": 3.708198489751888e-05,
+      "loss": 0.365,
       "step": 1370
     },
     {
       "epoch": 3.35,
-      "learning_rate": 1.8338727076591157e-05,
-      "loss": 0.4687,
+      "learning_rate": 3.6947141316073356e-05,
+      "loss": 0.3738,
       "step": 1380
     },
     {
       "epoch": 3.37,
-      "learning_rate": 1.8069039913700108e-05,
-      "loss": 0.4735,
+      "learning_rate": 3.681229773462783e-05,
+      "loss": 0.3628,
       "step": 1390
     },
     {
       "epoch": 3.39,
-      "learning_rate": 1.7799352750809062e-05,
-      "loss": 0.4798,
+      "learning_rate": 3.6677454153182314e-05,
+      "loss": 0.3674,
       "step": 1400
     },
     {
       "epoch": 3.42,
-      "learning_rate": 1.7529665587918013e-05,
-      "loss": 0.4288,
+      "learning_rate": 3.654261057173678e-05,
+      "loss": 0.3268,
       "step": 1410
     },
     {
       "epoch": 3.44,
-      "learning_rate": 1.725997842502697e-05,
-      "loss": 0.4592,
+      "learning_rate": 3.6407766990291265e-05,
+      "loss": 0.3433,
       "step": 1420
     },
     {
       "epoch": 3.47,
-      "learning_rate": 1.6990291262135926e-05,
-      "loss": 0.4694,
+      "learning_rate": 3.627292340884574e-05,
+      "loss": 0.3665,
       "step": 1430
     },
     {
       "epoch": 3.49,
-      "learning_rate": 1.6720604099244877e-05,
-      "loss": 0.4409,
+      "learning_rate": 3.6138079827400216e-05,
+      "loss": 0.3512,
       "step": 1440
     },
     {
       "epoch": 3.52,
-      "learning_rate": 1.645091693635383e-05,
-      "loss": 0.4536,
+      "learning_rate": 3.60032362459547e-05,
+      "loss": 0.3571,
       "step": 1450
     },
     {
       "epoch": 3.54,
-      "learning_rate": 1.6181229773462782e-05,
-      "loss": 0.4635,
+      "learning_rate": 3.5868392664509174e-05,
+      "loss": 0.3808,
       "step": 1460
     },
     {
       "epoch": 3.56,
-      "learning_rate": 1.5911542610571736e-05,
-      "loss": 0.4554,
+      "learning_rate": 3.573354908306365e-05,
+      "loss": 0.3969,
       "step": 1470
     },
     {
       "epoch": 3.59,
-      "learning_rate": 1.564185544768069e-05,
-      "loss": 0.4504,
+      "learning_rate": 3.5598705501618125e-05,
+      "loss": 0.3663,
       "step": 1480
     },
     {
       "epoch": 3.61,
-      "learning_rate": 1.5372168284789645e-05,
-      "loss": 0.445,
+      "learning_rate": 3.54638619201726e-05,
+      "loss": 0.3562,
       "step": 1490
     },
     {
       "epoch": 3.64,
-      "learning_rate": 1.5102481121898598e-05,
-      "loss": 0.4476,
+      "learning_rate": 3.5329018338727076e-05,
+      "loss": 0.3686,
       "step": 1500
     },
     {
       "epoch": 3.66,
-      "learning_rate": 1.483279395900755e-05,
-      "loss": 0.4828,
+      "learning_rate": 3.519417475728156e-05,
+      "loss": 0.3815,
       "step": 1510
     },
     {
       "epoch": 3.68,
-      "learning_rate": 1.4563106796116505e-05,
-      "loss": 0.4698,
+      "learning_rate": 3.505933117583603e-05,
+      "loss": 0.3892,
       "step": 1520
     },
     {
       "epoch": 3.71,
-      "learning_rate": 1.4293419633225461e-05,
-      "loss": 0.4603,
+      "learning_rate": 3.492448759439051e-05,
+      "loss": 0.3847,
       "step": 1530
     },
     {
       "epoch": 3.73,
-      "learning_rate": 1.4023732470334414e-05,
-      "loss": 0.4558,
+      "learning_rate": 3.4789644012944984e-05,
+      "loss": 0.3849,
       "step": 1540
     },
     {
       "epoch": 3.76,
-      "learning_rate": 1.3754045307443367e-05,
-      "loss": 0.4632,
+      "learning_rate": 3.465480043149946e-05,
+      "loss": 0.3964,
       "step": 1550
     },
     {
       "epoch": 3.78,
-      "learning_rate": 1.348435814455232e-05,
-      "loss": 0.4785,
+      "learning_rate": 3.451995685005394e-05,
+      "loss": 0.4121,
       "step": 1560
     },
     {
       "epoch": 3.81,
-      "learning_rate": 1.3214670981661272e-05,
-      "loss": 0.4593,
+      "learning_rate": 3.438511326860842e-05,
+      "loss": 0.3863,
       "step": 1570
     },
     {
       "epoch": 3.83,
-      "learning_rate": 1.2944983818770227e-05,
-      "loss": 0.439,
+      "learning_rate": 3.425026968716289e-05,
+      "loss": 0.3724,
       "step": 1580
     },
     {
       "epoch": 3.85,
-      "learning_rate": 1.2675296655879181e-05,
-      "loss": 0.437,
+      "learning_rate": 3.411542610571737e-05,
+      "loss": 0.3631,
       "step": 1590
     },
     {
       "epoch": 3.88,
-      "learning_rate": 1.2405609492988135e-05,
-      "loss": 0.4649,
+      "learning_rate": 3.398058252427185e-05,
+      "loss": 0.3825,
       "step": 1600
     },
     {
       "epoch": 3.9,
-      "learning_rate": 1.2135922330097088e-05,
-      "loss": 0.4976,
+      "learning_rate": 3.384573894282632e-05,
+      "loss": 0.4403,
       "step": 1610
     },
     {
       "epoch": 3.93,
-      "learning_rate": 1.1866235167206041e-05,
-      "loss": 0.4667,
+      "learning_rate": 3.37108953613808e-05,
+      "loss": 0.4074,
       "step": 1620
     },
     {
       "epoch": 3.95,
-      "learning_rate": 1.1596548004314995e-05,
-      "loss": 0.4591,
+      "learning_rate": 3.357605177993528e-05,
+      "loss": 0.4067,
       "step": 1630
     },
     {
       "epoch": 3.98,
-      "learning_rate": 1.1326860841423948e-05,
-      "loss": 0.4113,
+      "learning_rate": 3.344120819848975e-05,
+      "loss": 0.3598,
       "step": 1640
     },
     {
       "epoch": 4.0,
-      "learning_rate": 1.1057173678532903e-05,
-      "loss": 0.4476,
+      "learning_rate": 3.330636461704423e-05,
+      "loss": 0.3926,
       "step": 1650
     },
     {
       "epoch": 4.0,
-      "eval_accuracy": 0.7734401636549608,
-      "eval_loss": 0.4010549485683441,
-      "eval_runtime": 43.1679,
-      "eval_samples_per_second": 135.888,
-      "eval_steps_per_second": 4.262,
+      "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": 4.02,
-      "learning_rate": 1.0787486515641857e-05,
-      "loss": 0.4337,
+      "learning_rate": 3.3171521035598704e-05,
+      "loss": 0.3523,
       "step": 1660
     },
     {
       "epoch": 4.05,
-      "learning_rate": 1.051779935275081e-05,
-      "loss": 0.429,
+      "learning_rate": 3.3036677454153186e-05,
+      "loss": 0.3657,
       "step": 1670
     },
     {
       "epoch": 4.07,
-      "learning_rate": 1.0248112189859762e-05,
-      "loss": 0.4329,
+      "learning_rate": 3.290183387270766e-05,
+      "loss": 0.3677,
       "step": 1680
     },
     {
       "epoch": 4.1,
-      "learning_rate": 9.978425026968717e-06,
-      "loss": 0.451,
+      "learning_rate": 3.276699029126214e-05,
+      "loss": 0.4004,
       "step": 1690
     },
     {
       "epoch": 4.12,
-      "learning_rate": 9.70873786407767e-06,
-      "loss": 0.4556,
+      "learning_rate": 3.263214670981661e-05,
+      "loss": 0.3857,
       "step": 1700
     },
     {
       "epoch": 4.15,
-      "learning_rate": 9.439050701186624e-06,
-      "loss": 0.466,
+      "learning_rate": 3.2497303128371095e-05,
+      "loss": 0.4001,
       "step": 1710
     },
     {
       "epoch": 4.17,
-      "learning_rate": 9.169363538295578e-06,
-      "loss": 0.4192,
+      "learning_rate": 3.2362459546925564e-05,
+      "loss": 0.3483,
       "step": 1720
     },
     {
       "epoch": 4.19,
-      "learning_rate": 8.899676375404531e-06,
-      "loss": 0.4304,
+      "learning_rate": 3.2227615965480046e-05,
+      "loss": 0.3868,
       "step": 1730
     },
     {
       "epoch": 4.22,
-      "learning_rate": 8.629989212513486e-06,
-      "loss": 0.4474,
+      "learning_rate": 3.209277238403452e-05,
+      "loss": 0.3811,
       "step": 1740
     },
     {
       "epoch": 4.24,
-      "learning_rate": 8.360302049622438e-06,
-      "loss": 0.4011,
+      "learning_rate": 3.1957928802589e-05,
+      "loss": 0.344,
       "step": 1750
     },
     {
       "epoch": 4.27,
-      "learning_rate": 8.090614886731391e-06,
-      "loss": 0.4566,
+      "learning_rate": 3.182308522114347e-05,
+      "loss": 0.4038,
       "step": 1760
     },
     {
       "epoch": 4.29,
-      "learning_rate": 7.820927723840345e-06,
-      "loss": 0.417,
+      "learning_rate": 3.1688241639697955e-05,
+      "loss": 0.3716,
       "step": 1770
     },
     {
       "epoch": 4.32,
-      "learning_rate": 7.551240560949299e-06,
-      "loss": 0.4359,
+      "learning_rate": 3.155339805825243e-05,
+      "loss": 0.395,
       "step": 1780
     },
     {
       "epoch": 4.34,
-      "learning_rate": 7.281553398058253e-06,
-      "loss": 0.4322,
+      "learning_rate": 3.1418554476806906e-05,
+      "loss": 0.3959,
       "step": 1790
     },
     {
       "epoch": 4.36,
-      "learning_rate": 7.011866235167207e-06,
-      "loss": 0.4457,
+      "learning_rate": 3.128371089536138e-05,
+      "loss": 0.3983,
       "step": 1800
     },
     {
       "epoch": 4.39,
-      "learning_rate": 6.74217907227616e-06,
-      "loss": 0.4896,
+      "learning_rate": 3.114886731391586e-05,
+      "loss": 0.4456,
       "step": 1810
     },
     {
       "epoch": 4.41,
-      "learning_rate": 6.472491909385113e-06,
-      "loss": 0.4384,
+      "learning_rate": 3.101402373247034e-05,
+      "loss": 0.3956,
       "step": 1820
     },
     {
       "epoch": 4.44,
-      "learning_rate": 6.202804746494068e-06,
-      "loss": 0.4464,
+      "learning_rate": 3.087918015102481e-05,
+      "loss": 0.3926,
       "step": 1830
     },
     {
       "epoch": 4.46,
-      "learning_rate": 5.9331175836030205e-06,
-      "loss": 0.4671,
+      "learning_rate": 3.074433656957929e-05,
+      "loss": 0.4182,
       "step": 1840
     },
     {
       "epoch": 4.48,
-      "learning_rate": 5.663430420711974e-06,
-      "loss": 0.4436,
+      "learning_rate": 3.0609492988133766e-05,
+      "loss": 0.4022,
       "step": 1850
     },
     {
       "epoch": 4.51,
-      "learning_rate": 5.3937432578209285e-06,
-      "loss": 0.4403,
+      "learning_rate": 3.0474649406688245e-05,
+      "loss": 0.4088,
       "step": 1860
     },
     {
       "epoch": 4.53,
-      "learning_rate": 5.124056094929881e-06,
-      "loss": 0.4329,
+      "learning_rate": 3.0339805825242717e-05,
+      "loss": 0.386,
       "step": 1870
     },
     {
       "epoch": 4.56,
-      "learning_rate": 4.854368932038835e-06,
-      "loss": 0.4651,
+      "learning_rate": 3.0204962243797196e-05,
+      "loss": 0.411,
       "step": 1880
     },
     {
       "epoch": 4.58,
-      "learning_rate": 4.584681769147789e-06,
-      "loss": 0.424,
+      "learning_rate": 3.0070118662351675e-05,
+      "loss": 0.3774,
       "step": 1890
     },
     {
       "epoch": 4.61,
-      "learning_rate": 4.314994606256743e-06,
-      "loss": 0.4539,
+      "learning_rate": 2.993527508090615e-05,
+      "loss": 0.4273,
       "step": 1900
     },
     {
       "epoch": 4.63,
-      "learning_rate": 4.0453074433656955e-06,
-      "loss": 0.4237,
+      "learning_rate": 2.980043149946063e-05,
+      "loss": 0.4008,
       "step": 1910
     },
     {
       "epoch": 4.65,
-      "learning_rate": 3.7756202804746495e-06,
-      "loss": 0.4337,
+      "learning_rate": 2.96655879180151e-05,
+      "loss": 0.4021,
       "step": 1920
     },
     {
       "epoch": 4.68,
-      "learning_rate": 3.5059331175836035e-06,
-      "loss": 0.4322,
+      "learning_rate": 2.9530744336569584e-05,
+      "loss": 0.4117,
       "step": 1930
     },
     {
       "epoch": 4.7,
-      "learning_rate": 3.2362459546925567e-06,
-      "loss": 0.441,
+      "learning_rate": 2.9395900755124056e-05,
+      "loss": 0.4169,
       "step": 1940
     },
     {
       "epoch": 4.73,
-      "learning_rate": 2.9665587918015102e-06,
-      "loss": 0.4151,
+      "learning_rate": 2.9261057173678535e-05,
+      "loss": 0.3996,
       "step": 1950
     },
     {
       "epoch": 4.75,
-      "learning_rate": 2.6968716289104642e-06,
-      "loss": 0.4437,
+      "learning_rate": 2.912621359223301e-05,
+      "loss": 0.4179,
       "step": 1960
     },
     {
       "epoch": 4.78,
-      "learning_rate": 2.4271844660194174e-06,
-      "loss": 0.4187,
+      "learning_rate": 2.899137001078749e-05,
+      "loss": 0.4037,
       "step": 1970
     },
     {
       "epoch": 4.8,
-      "learning_rate": 2.1574973031283714e-06,
-      "loss": 0.4216,
+      "learning_rate": 2.885652642934196e-05,
+      "loss": 0.4045,
       "step": 1980
     },
     {
       "epoch": 4.82,
-      "learning_rate": 1.8878101402373248e-06,
-      "loss": 0.4153,
+      "learning_rate": 2.872168284789644e-05,
+      "loss": 0.4055,
       "step": 1990
     },
     {
       "epoch": 4.85,
-      "learning_rate": 1.6181229773462783e-06,
-      "loss": 0.4495,
+      "learning_rate": 2.8586839266450923e-05,
+      "loss": 0.4294,
       "step": 2000
     },
     {
       "epoch": 4.87,
-      "learning_rate": 1.3484358144552321e-06,
-      "loss": 0.4395,
+      "learning_rate": 2.8451995685005395e-05,
+      "loss": 0.4161,
       "step": 2010
     },
     {
       "epoch": 4.9,
-      "learning_rate": 1.0787486515641857e-06,
-      "loss": 0.4226,
+      "learning_rate": 2.8317152103559874e-05,
+      "loss": 0.4084,
       "step": 2020
     },
     {
       "epoch": 4.92,
-      "learning_rate": 8.090614886731392e-07,
-      "loss": 0.4334,
+      "learning_rate": 2.818230852211435e-05,
+      "loss": 0.4111,
       "step": 2030
     },
     {
       "epoch": 4.95,
-      "learning_rate": 5.393743257820928e-07,
-      "loss": 0.4169,
+      "learning_rate": 2.8047464940668828e-05,
+      "loss": 0.3972,
       "step": 2040
     },
     {
       "epoch": 4.97,
-      "learning_rate": 2.696871628910464e-07,
-      "loss": 0.4196,
+      "learning_rate": 2.79126213592233e-05,
+      "loss": 0.3997,
       "step": 2050
     },
     {
       "epoch": 4.99,
-      "learning_rate": 0.0,
-      "loss": 0.4093,
+      "learning_rate": 2.777777777777778e-05,
+      "loss": 0.3859,
       "step": 2060
     },
     {
-      "epoch": 4.99,
-      "eval_accuracy": 0.7797476986021139,
-      "eval_loss": 0.3965856432914734,
-      "eval_runtime": 45.954,
-      "eval_samples_per_second": 127.649,
-      "eval_steps_per_second": 4.004,
-      "step": 2060
+      "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": 4.99,
-      "step": 2060,
-      "total_flos": 6.553200292249928e+18,
-      "train_loss": 0.4588474294514332,
-      "train_runtime": 4338.0094,
-      "train_samples_per_second": 60.844,
+      "epoch": 5.02,
+      "learning_rate": 2.7642934196332255e-05,
+      "loss": 0.4082,
+      "step": 2070
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 2.7508090614886734e-05,
+      "loss": 0.384,
+      "step": 2080
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 2.7373247033441206e-05,
+      "loss": 0.4004,
+      "step": 2090
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 2.7238403451995688e-05,
+      "loss": 0.4186,
+      "step": 2100
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 2.7103559870550167e-05,
+      "loss": 0.385,
+      "step": 2110
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 2.696871628910464e-05,
+      "loss": 0.3718,
+      "step": 2120
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 2.6833872707659118e-05,
+      "loss": 0.3845,
+      "step": 2130
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 2.6699029126213593e-05,
+      "loss": 0.3861,
+      "step": 2140
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 2.6564185544768072e-05,
+      "loss": 0.3959,
+      "step": 2150
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 2.6429341963322544e-05,
+      "loss": 0.411,
+      "step": 2160
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 2.6294498381877027e-05,
+      "loss": 0.3849,
+      "step": 2170
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 2.61596548004315e-05,
+      "loss": 0.3884,
+      "step": 2180
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 2.6024811218985978e-05,
+      "loss": 0.3848,
+      "step": 2190
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 2.5889967637540453e-05,
+      "loss": 0.3937,
+      "step": 2200
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 2.5755124056094932e-05,
+      "loss": 0.4112,
+      "step": 2210
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 2.5620280474649404e-05,
+      "loss": 0.3902,
+      "step": 2220
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 2.5485436893203883e-05,
+      "loss": 0.3899,
+      "step": 2230
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 2.5350593311758362e-05,
+      "loss": 0.4006,
+      "step": 2240
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 2.5215749730312838e-05,
+      "loss": 0.4078,
+      "step": 2250
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 2.5080906148867317e-05,
+      "loss": 0.4113,
+      "step": 2260
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 2.4946062567421792e-05,
+      "loss": 0.404,
+      "step": 2270
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 2.481121898597627e-05,
+      "loss": 0.429,
+      "step": 2280
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 2.4676375404530747e-05,
+      "loss": 0.3862,
+      "step": 2290
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 2.4541531823085222e-05,
+      "loss": 0.3846,
+      "step": 2300
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 2.44066882416397e-05,
+      "loss": 0.3923,
+      "step": 2310
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 2.4271844660194176e-05,
+      "loss": 0.3899,
+      "step": 2320
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 2.4137001078748652e-05,
+      "loss": 0.3964,
+      "step": 2330
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 2.400215749730313e-05,
+      "loss": 0.3773,
+      "step": 2340
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 2.3867313915857606e-05,
+      "loss": 0.3735,
+      "step": 2350
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 2.3732470334412082e-05,
+      "loss": 0.3679,
+      "step": 2360
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 2.3597626752966557e-05,
+      "loss": 0.3818,
+      "step": 2370
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 2.3462783171521036e-05,
+      "loss": 0.4101,
+      "step": 2380
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 2.3327939590075515e-05,
+      "loss": 0.4143,
+      "step": 2390
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 2.319309600862999e-05,
+      "loss": 0.3931,
+      "step": 2400
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 2.305825242718447e-05,
+      "loss": 0.3645,
+      "step": 2410
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 2.2923408845738945e-05,
+      "loss": 0.3971,
+      "step": 2420
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 2.278856526429342e-05,
+      "loss": 0.3883,
+      "step": 2430
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 2.2653721682847896e-05,
+      "loss": 0.3859,
+      "step": 2440
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 2.2518878101402375e-05,
+      "loss": 0.3803,
+      "step": 2450
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 2.238403451995685e-05,
+      "loss": 0.3686,
+      "step": 2460
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 2.2249190938511326e-05,
+      "loss": 0.3822,
+      "step": 2470
+    },
+    {
+      "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": 6.01,
+      "learning_rate": 2.2114347357065805e-05,
+      "loss": 0.402,
+      "step": 2480
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 2.197950377562028e-05,
+      "loss": 0.3558,
+      "step": 2490
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 2.1844660194174756e-05,
+      "loss": 0.3853,
+      "step": 2500
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 2.1709816612729235e-05,
+      "loss": 0.3455,
+      "step": 2510
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 2.1574973031283714e-05,
+      "loss": 0.3913,
+      "step": 2520
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 2.144012944983819e-05,
+      "loss": 0.3874,
+      "step": 2530
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 2.1305285868392665e-05,
+      "loss": 0.3646,
+      "step": 2540
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 2.1170442286947144e-05,
+      "loss": 0.3933,
+      "step": 2550
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 2.103559870550162e-05,
+      "loss": 0.3445,
+      "step": 2560
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 2.0900755124056095e-05,
+      "loss": 0.3621,
+      "step": 2570
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 2.0765911542610574e-05,
+      "loss": 0.3742,
+      "step": 2580
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 2.063106796116505e-05,
+      "loss": 0.3861,
+      "step": 2590
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 2.0496224379719525e-05,
+      "loss": 0.3535,
+      "step": 2600
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 2.0361380798274e-05,
+      "loss": 0.3739,
+      "step": 2610
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 2.0226537216828483e-05,
+      "loss": 0.3711,
+      "step": 2620
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 2.0091693635382958e-05,
+      "loss": 0.3601,
+      "step": 2630
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 1.9956850053937434e-05,
+      "loss": 0.3603,
+      "step": 2640
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.982200647249191e-05,
+      "loss": 0.3574,
+      "step": 2650
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1.9687162891046388e-05,
+      "loss": 0.3706,
+      "step": 2660
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.9552319309600864e-05,
+      "loss": 0.3829,
+      "step": 2670
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 1.941747572815534e-05,
+      "loss": 0.3904,
+      "step": 2680
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 1.9282632146709818e-05,
+      "loss": 0.3595,
+      "step": 2690
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 1.9147788565264294e-05,
+      "loss": 0.3494,
+      "step": 2700
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 1.901294498381877e-05,
+      "loss": 0.386,
+      "step": 2710
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1.8878101402373248e-05,
+      "loss": 0.3716,
+      "step": 2720
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 1.8743257820927727e-05,
+      "loss": 0.3508,
+      "step": 2730
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 1.8608414239482202e-05,
+      "loss": 0.3528,
+      "step": 2740
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.8473570658036678e-05,
+      "loss": 0.3731,
+      "step": 2750
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 1.8338727076591157e-05,
+      "loss": 0.3812,
+      "step": 2760
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 1.8203883495145632e-05,
+      "loss": 0.3745,
+      "step": 2770
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 1.8069039913700108e-05,
+      "loss": 0.3946,
+      "step": 2780
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 1.7934196332254587e-05,
+      "loss": 0.3744,
+      "step": 2790
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 1.7799352750809062e-05,
+      "loss": 0.3496,
+      "step": 2800
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 1.7664509169363538e-05,
+      "loss": 0.359,
+      "step": 2810
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 1.7529665587918013e-05,
+      "loss": 0.3781,
+      "step": 2820
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 1.7394822006472492e-05,
+      "loss": 0.3524,
+      "step": 2830
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 1.725997842502697e-05,
+      "loss": 0.3977,
+      "step": 2840
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 1.7125134843581447e-05,
+      "loss": 0.3668,
+      "step": 2850
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 1.6990291262135926e-05,
+      "loss": 0.3718,
+      "step": 2860
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 1.68554476806904e-05,
+      "loss": 0.3519,
+      "step": 2870
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 1.6720604099244877e-05,
+      "loss": 0.3804,
+      "step": 2880
+    },
+    {
+      "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": 7.01,
+      "learning_rate": 1.6585760517799352e-05,
+      "loss": 0.3441,
+      "step": 2890
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.645091693635383e-05,
+      "loss": 0.3618,
+      "step": 2900
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 1.6316073354908307e-05,
+      "loss": 0.3523,
+      "step": 2910
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 1.6181229773462782e-05,
+      "loss": 0.3421,
+      "step": 2920
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 1.604638619201726e-05,
+      "loss": 0.342,
+      "step": 2930
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 1.5911542610571736e-05,
+      "loss": 0.3646,
+      "step": 2940
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 1.5776699029126215e-05,
+      "loss": 0.3115,
+      "step": 2950
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 1.564185544768069e-05,
+      "loss": 0.3539,
+      "step": 2960
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 1.550701186623517e-05,
+      "loss": 0.3236,
+      "step": 2970
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 1.5372168284789645e-05,
+      "loss": 0.3603,
+      "step": 2980
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 1.5237324703344123e-05,
+      "loss": 0.3449,
+      "step": 2990
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 1.5102481121898598e-05,
+      "loss": 0.3477,
+      "step": 3000
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 1.4967637540453075e-05,
+      "loss": 0.364,
+      "step": 3010
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 1.483279395900755e-05,
+      "loss": 0.37,
+      "step": 3020
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 1.4697950377562028e-05,
+      "loss": 0.3719,
+      "step": 3030
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 1.4563106796116505e-05,
+      "loss": 0.358,
+      "step": 3040
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 1.442826321467098e-05,
+      "loss": 0.3529,
+      "step": 3050
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 1.4293419633225461e-05,
+      "loss": 0.3502,
+      "step": 3060
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 1.4158576051779937e-05,
+      "loss": 0.3708,
+      "step": 3070
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 1.4023732470334414e-05,
+      "loss": 0.3221,
+      "step": 3080
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 1.388888888888889e-05,
+      "loss": 0.3511,
+      "step": 3090
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 1.3754045307443367e-05,
+      "loss": 0.333,
+      "step": 3100
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 1.3619201725997844e-05,
+      "loss": 0.3606,
+      "step": 3110
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 1.348435814455232e-05,
+      "loss": 0.3492,
+      "step": 3120
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 1.3349514563106797e-05,
+      "loss": 0.3365,
+      "step": 3130
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 1.3214670981661272e-05,
+      "loss": 0.3551,
+      "step": 3140
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 1.307982740021575e-05,
+      "loss": 0.362,
+      "step": 3150
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 1.2944983818770227e-05,
+      "loss": 0.3274,
+      "step": 3160
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 1.2810140237324702e-05,
+      "loss": 0.329,
+      "step": 3170
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 1.2675296655879181e-05,
+      "loss": 0.352,
+      "step": 3180
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 1.2540453074433658e-05,
+      "loss": 0.3475,
+      "step": 3190
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 1.2405609492988135e-05,
+      "loss": 0.3418,
+      "step": 3200
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 1.2270765911542611e-05,
+      "loss": 0.3795,
+      "step": 3210
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 1.2135922330097088e-05,
+      "loss": 0.3549,
+      "step": 3220
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 1.2001078748651565e-05,
+      "loss": 0.3421,
+      "step": 3230
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.1866235167206041e-05,
+      "loss": 0.3171,
+      "step": 3240
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 1.1731391585760518e-05,
+      "loss": 0.3513,
+      "step": 3250
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 1.1596548004314995e-05,
+      "loss": 0.3514,
+      "step": 3260
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 1.1461704422869473e-05,
+      "loss": 0.3622,
+      "step": 3270
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 1.1326860841423948e-05,
+      "loss": 0.3541,
+      "step": 3280
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 1.1192017259978425e-05,
+      "loss": 0.3734,
+      "step": 3290
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 1.1057173678532903e-05,
+      "loss": 0.3231,
+      "step": 3300
+    },
+    {
+      "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": 8.02,
+      "learning_rate": 1.0922330097087378e-05,
+      "loss": 0.3522,
+      "step": 3310
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 1.0787486515641857e-05,
+      "loss": 0.3337,
+      "step": 3320
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 1.0652642934196332e-05,
+      "loss": 0.2974,
+      "step": 3330
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 1.051779935275081e-05,
+      "loss": 0.3454,
+      "step": 3340
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 1.0382955771305287e-05,
+      "loss": 0.3054,
+      "step": 3350
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 1.0248112189859762e-05,
+      "loss": 0.3125,
+      "step": 3360
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 1.0113268608414241e-05,
+      "loss": 0.3233,
+      "step": 3370
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 9.978425026968717e-06,
+      "loss": 0.3265,
+      "step": 3380
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 9.843581445523194e-06,
+      "loss": 0.3262,
+      "step": 3390
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 9.70873786407767e-06,
+      "loss": 0.3339,
+      "step": 3400
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 9.573894282632147e-06,
+      "loss": 0.3501,
+      "step": 3410
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 9.439050701186624e-06,
+      "loss": 0.3319,
+      "step": 3420
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 9.304207119741101e-06,
+      "loss": 0.3193,
+      "step": 3430
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 9.169363538295578e-06,
+      "loss": 0.3383,
+      "step": 3440
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 9.034519956850054e-06,
+      "loss": 0.3369,
+      "step": 3450
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 8.899676375404531e-06,
+      "loss": 0.3473,
+      "step": 3460
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 8.764832793959007e-06,
+      "loss": 0.315,
+      "step": 3470
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 8.629989212513486e-06,
+      "loss": 0.3416,
+      "step": 3480
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 8.495145631067963e-06,
+      "loss": 0.3297,
+      "step": 3490
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 8.360302049622438e-06,
+      "loss": 0.3263,
+      "step": 3500
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 8.225458468176916e-06,
+      "loss": 0.3569,
+      "step": 3510
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 8.090614886731391e-06,
+      "loss": 0.3299,
+      "step": 3520
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 7.955771305285868e-06,
+      "loss": 0.3346,
+      "step": 3530
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 7.820927723840345e-06,
+      "loss": 0.3047,
+      "step": 3540
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 7.686084142394823e-06,
+      "loss": 0.336,
+      "step": 3550
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 7.551240560949299e-06,
+      "loss": 0.3207,
+      "step": 3560
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 7.416396979503775e-06,
+      "loss": 0.3171,
+      "step": 3570
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 7.281553398058253e-06,
+      "loss": 0.3186,
+      "step": 3580
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 7.146709816612731e-06,
+      "loss": 0.3307,
+      "step": 3590
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 7.011866235167207e-06,
+      "loss": 0.3409,
+      "step": 3600
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 6.877022653721683e-06,
+      "loss": 0.324,
+      "step": 3610
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 6.74217907227616e-06,
+      "loss": 0.3397,
+      "step": 3620
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 6.607335490830636e-06,
+      "loss": 0.3437,
+      "step": 3630
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 6.472491909385113e-06,
+      "loss": 0.3441,
+      "step": 3640
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 6.3376483279395905e-06,
+      "loss": 0.3405,
+      "step": 3650
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 6.202804746494068e-06,
+      "loss": 0.3191,
+      "step": 3660
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 6.067961165048544e-06,
+      "loss": 0.334,
+      "step": 3670
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 5.9331175836030205e-06,
+      "loss": 0.3244,
+      "step": 3680
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 5.798274002157498e-06,
+      "loss": 0.3257,
+      "step": 3690
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 5.663430420711974e-06,
+      "loss": 0.3201,
+      "step": 3700
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 5.528586839266451e-06,
+      "loss": 0.3295,
+      "step": 3710
+    },
+    {
+      "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": 9.02,
+      "learning_rate": 5.3937432578209285e-06,
+      "loss": 0.2991,
+      "step": 3720
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 5.258899676375405e-06,
+      "loss": 0.2793,
+      "step": 3730
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 5.124056094929881e-06,
+      "loss": 0.3362,
+      "step": 3740
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 4.989212513484358e-06,
+      "loss": 0.3433,
+      "step": 3750
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 4.854368932038835e-06,
+      "loss": 0.2879,
+      "step": 3760
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 4.719525350593312e-06,
+      "loss": 0.3276,
+      "step": 3770
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 4.584681769147789e-06,
+      "loss": 0.3038,
+      "step": 3780
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 4.4498381877022656e-06,
+      "loss": 0.3275,
+      "step": 3790
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 4.314994606256743e-06,
+      "loss": 0.3071,
+      "step": 3800
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 4.180151024811219e-06,
+      "loss": 0.3513,
+      "step": 3810
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 4.0453074433656955e-06,
+      "loss": 0.3253,
+      "step": 3820
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 3.910463861920173e-06,
+      "loss": 0.3332,
+      "step": 3830
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 3.7756202804746495e-06,
+      "loss": 0.3215,
+      "step": 3840
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 3.6407766990291263e-06,
+      "loss": 0.3118,
+      "step": 3850
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 3.5059331175836035e-06,
+      "loss": 0.3036,
+      "step": 3860
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 3.37108953613808e-06,
+      "loss": 0.3008,
+      "step": 3870
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 3.2362459546925567e-06,
+      "loss": 0.3381,
+      "step": 3880
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 3.101402373247034e-06,
+      "loss": 0.3265,
+      "step": 3890
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 2.9665587918015102e-06,
+      "loss": 0.3314,
+      "step": 3900
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 2.831715210355987e-06,
+      "loss": 0.3242,
+      "step": 3910
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 2.6968716289104642e-06,
+      "loss": 0.3089,
+      "step": 3920
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 2.5620280474649406e-06,
+      "loss": 0.3296,
+      "step": 3930
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 2.4271844660194174e-06,
+      "loss": 0.3257,
+      "step": 3940
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 2.2923408845738946e-06,
+      "loss": 0.364,
+      "step": 3950
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 2.1574973031283714e-06,
+      "loss": 0.3389,
+      "step": 3960
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 2.0226537216828478e-06,
+      "loss": 0.3048,
+      "step": 3970
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.8878101402373248e-06,
+      "loss": 0.343,
+      "step": 3980
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.7529665587918018e-06,
+      "loss": 0.3005,
+      "step": 3990
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 1.6181229773462783e-06,
+      "loss": 0.3253,
+      "step": 4000
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 1.4832793959007551e-06,
+      "loss": 0.3117,
+      "step": 4010
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 1.3484358144552321e-06,
+      "loss": 0.3244,
+      "step": 4020
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 1.2135922330097087e-06,
+      "loss": 0.3004,
+      "step": 4030
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 1.0787486515641857e-06,
+      "loss": 0.3204,
+      "step": 4040
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 9.439050701186624e-07,
+      "loss": 0.3185,
+      "step": 4050
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 8.090614886731392e-07,
+      "loss": 0.3223,
+      "step": 4060
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 6.742179072276161e-07,
+      "loss": 0.3153,
+      "step": 4070
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 5.393743257820928e-07,
+      "loss": 0.3203,
+      "step": 4080
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 4.045307443365696e-07,
+      "loss": 0.3241,
+      "step": 4090
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 2.696871628910464e-07,
+      "loss": 0.2784,
+      "step": 4100
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.348435814455232e-07,
+      "loss": 0.328,
+      "step": 4110
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 0.0,
+      "loss": 0.321,
+      "step": 4120
+    },
+    {
+      "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": 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
     }
   ],
   "logging_steps": 10,
-  "max_steps": 2060,
-  "num_train_epochs": 5,
+  "max_steps": 4120,
+  "num_train_epochs": 10,
   "save_steps": 500,
-  "total_flos": 6.553200292249928e+18,
+  "total_flos": 1.3106102295963156e+19,
   "trial_name": null,
   "trial_params": null
 }