diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -1,4908 +1,2037 @@
 {
-  "best_metric": 0.8571428571428571,
-  "best_model_checkpoint": "vit-base-patch16-224-in21k-finetuned/checkpoint-7980",
-  "epoch": 10.0,
+  "best_metric": 0.7444126074498567,
+  "best_model_checkpoint": "vit-Facial-Expression-Recognition/checkpoint-1962",
+  "epoch": 4.9980894153611,
   "eval_steps": 500,
-  "global_step": 7980,
+  "global_step": 3270,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.01,
-      "learning_rate": 1.5625e-06,
-      "loss": 1.5128,
+      "epoch": 0.02,
+      "learning_rate": 1.5290519877675841e-06,
+      "loss": 1.9371,
       "step": 10
     },
     {
       "epoch": 0.03,
-      "learning_rate": 3.125e-06,
-      "loss": 1.3411,
+      "learning_rate": 3.0581039755351682e-06,
+      "loss": 1.8225,
       "step": 20
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 4.6875000000000004e-06,
-      "loss": 1.2153,
+      "epoch": 0.05,
+      "learning_rate": 4.587155963302753e-06,
+      "loss": 1.4076,
       "step": 30
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 6.25e-06,
-      "loss": 1.2697,
+      "epoch": 0.06,
+      "learning_rate": 6.1162079510703365e-06,
+      "loss": 1.5662,
       "step": 40
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 7.8125e-06,
-      "loss": 1.1394,
+      "epoch": 0.08,
+      "learning_rate": 7.645259938837921e-06,
+      "loss": 1.3985,
       "step": 50
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 9.375000000000001e-06,
-      "loss": 1.0215,
+      "epoch": 0.09,
+      "learning_rate": 9.174311926605506e-06,
+      "loss": 1.1727,
       "step": 60
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.09375e-05,
-      "loss": 0.9762,
+      "epoch": 0.11,
+      "learning_rate": 1.070336391437309e-05,
+      "loss": 1.1874,
       "step": 70
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.25e-05,
-      "loss": 0.8893,
+      "epoch": 0.12,
+      "learning_rate": 1.2232415902140673e-05,
+      "loss": 1.1256,
       "step": 80
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.4062500000000001e-05,
-      "loss": 0.9721,
+      "epoch": 0.14,
+      "learning_rate": 1.3761467889908258e-05,
+      "loss": 1.0438,
       "step": 90
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 1.5625e-05,
-      "loss": 0.9281,
+      "epoch": 0.15,
+      "learning_rate": 1.5290519877675842e-05,
+      "loss": 1.1966,
       "step": 100
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 1.71875e-05,
-      "loss": 0.9426,
+      "epoch": 0.17,
+      "learning_rate": 1.6819571865443427e-05,
+      "loss": 1.0475,
       "step": 110
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 1.8750000000000002e-05,
-      "loss": 0.8554,
+      "epoch": 0.18,
+      "learning_rate": 1.834862385321101e-05,
+      "loss": 0.9658,
       "step": 120
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 2.0312500000000002e-05,
-      "loss": 0.808,
+      "epoch": 0.2,
+      "learning_rate": 1.9877675840978592e-05,
+      "loss": 0.959,
       "step": 130
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 2.1875e-05,
-      "loss": 0.8681,
+      "epoch": 0.21,
+      "learning_rate": 2.140672782874618e-05,
+      "loss": 0.7963,
       "step": 140
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 2.34375e-05,
-      "loss": 0.897,
+      "epoch": 0.23,
+      "learning_rate": 2.2935779816513765e-05,
+      "loss": 0.9615,
       "step": 150
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 2.5e-05,
-      "loss": 0.861,
+      "epoch": 0.24,
+      "learning_rate": 2.4464831804281346e-05,
+      "loss": 0.8296,
       "step": 160
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 2.6562500000000002e-05,
-      "loss": 0.9079,
+      "epoch": 0.26,
+      "learning_rate": 2.5993883792048927e-05,
+      "loss": 0.8144,
       "step": 170
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 2.8125000000000003e-05,
-      "loss": 0.9046,
+      "epoch": 0.28,
+      "learning_rate": 2.7522935779816515e-05,
+      "loss": 0.9023,
       "step": 180
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 2.96875e-05,
-      "loss": 0.8711,
+      "epoch": 0.29,
+      "learning_rate": 2.90519877675841e-05,
+      "loss": 0.9142,
       "step": 190
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 3.125e-05,
-      "loss": 0.8316,
+      "epoch": 0.31,
+      "learning_rate": 3.0581039755351684e-05,
+      "loss": 0.9456,
       "step": 200
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 3.2812500000000005e-05,
-      "loss": 0.829,
+      "epoch": 0.32,
+      "learning_rate": 3.211009174311927e-05,
+      "loss": 0.8358,
       "step": 210
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 3.4375e-05,
-      "loss": 0.8234,
+      "epoch": 0.34,
+      "learning_rate": 3.363914373088685e-05,
+      "loss": 0.7678,
       "step": 220
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 3.59375e-05,
-      "loss": 0.8658,
+      "epoch": 0.35,
+      "learning_rate": 3.516819571865443e-05,
+      "loss": 0.8491,
       "step": 230
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 3.7500000000000003e-05,
-      "loss": 0.8653,
+      "epoch": 0.37,
+      "learning_rate": 3.669724770642202e-05,
+      "loss": 0.7184,
       "step": 240
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 3.90625e-05,
-      "loss": 0.8014,
+      "epoch": 0.38,
+      "learning_rate": 3.822629969418961e-05,
+      "loss": 0.767,
       "step": 250
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 4.0625000000000005e-05,
-      "loss": 0.7934,
+      "epoch": 0.4,
+      "learning_rate": 3.9755351681957185e-05,
+      "loss": 0.7829,
       "step": 260
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 4.21875e-05,
-      "loss": 0.8023,
+      "epoch": 0.41,
+      "learning_rate": 4.1284403669724776e-05,
+      "loss": 0.76,
       "step": 270
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 4.375e-05,
-      "loss": 0.8294,
+      "epoch": 0.43,
+      "learning_rate": 4.281345565749236e-05,
+      "loss": 0.8151,
       "step": 280
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 4.5312500000000004e-05,
-      "loss": 0.7873,
+      "epoch": 0.44,
+      "learning_rate": 4.434250764525994e-05,
+      "loss": 0.7488,
       "step": 290
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 4.6875e-05,
-      "loss": 0.8071,
+      "epoch": 0.46,
+      "learning_rate": 4.587155963302753e-05,
+      "loss": 0.6966,
       "step": 300
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 4.8437500000000005e-05,
-      "loss": 0.8086,
+      "epoch": 0.47,
+      "learning_rate": 4.740061162079511e-05,
+      "loss": 0.9042,
       "step": 310
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 5e-05,
-      "loss": 0.8296,
+      "epoch": 0.49,
+      "learning_rate": 4.892966360856269e-05,
+      "loss": 0.7496,
       "step": 320
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 4.9825905292479115e-05,
-      "loss": 0.8096,
+      "epoch": 0.5,
+      "learning_rate": 4.994903160040775e-05,
+      "loss": 0.7599,
       "step": 330
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 4.965181058495822e-05,
-      "loss": 0.861,
+      "epoch": 0.52,
+      "learning_rate": 4.977913693510024e-05,
+      "loss": 0.8631,
       "step": 340
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 4.9477715877437334e-05,
-      "loss": 0.8047,
+      "epoch": 0.53,
+      "learning_rate": 4.9609242269792734e-05,
+      "loss": 0.8586,
       "step": 350
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 4.930362116991643e-05,
-      "loss": 0.8182,
+      "epoch": 0.55,
+      "learning_rate": 4.943934760448522e-05,
+      "loss": 0.7269,
       "step": 360
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 4.9129526462395546e-05,
-      "loss": 0.8412,
+      "epoch": 0.57,
+      "learning_rate": 4.926945293917771e-05,
+      "loss": 0.8573,
       "step": 370
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 4.895543175487465e-05,
-      "loss": 0.8438,
+      "epoch": 0.58,
+      "learning_rate": 4.9099558273870206e-05,
+      "loss": 0.7636,
       "step": 380
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 4.8781337047353764e-05,
-      "loss": 0.7776,
+      "epoch": 0.6,
+      "learning_rate": 4.892966360856269e-05,
+      "loss": 0.7227,
       "step": 390
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 4.860724233983287e-05,
-      "loss": 0.8141,
+      "epoch": 0.61,
+      "learning_rate": 4.8759768943255185e-05,
+      "loss": 0.7689,
       "step": 400
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 4.843314763231198e-05,
-      "loss": 0.9028,
+      "epoch": 0.63,
+      "learning_rate": 4.858987427794767e-05,
+      "loss": 0.6832,
       "step": 410
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 4.825905292479109e-05,
-      "loss": 0.8347,
+      "epoch": 0.64,
+      "learning_rate": 4.8419979612640164e-05,
+      "loss": 0.7225,
       "step": 420
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 4.80849582172702e-05,
-      "loss": 0.8383,
+      "epoch": 0.66,
+      "learning_rate": 4.8250084947332657e-05,
+      "loss": 0.8064,
       "step": 430
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 4.79108635097493e-05,
-      "loss": 0.8052,
+      "epoch": 0.67,
+      "learning_rate": 4.808019028202514e-05,
+      "loss": 0.6557,
       "step": 440
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 4.7736768802228414e-05,
-      "loss": 0.8401,
+      "epoch": 0.69,
+      "learning_rate": 4.7910295616717635e-05,
+      "loss": 0.7788,
       "step": 450
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 4.756267409470752e-05,
-      "loss": 0.7737,
+      "epoch": 0.7,
+      "learning_rate": 4.774040095141013e-05,
+      "loss": 0.7525,
       "step": 460
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 4.738857938718663e-05,
-      "loss": 0.7741,
+      "epoch": 0.72,
+      "learning_rate": 4.7570506286102614e-05,
+      "loss": 0.7951,
       "step": 470
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 4.721448467966574e-05,
-      "loss": 0.8272,
+      "epoch": 0.73,
+      "learning_rate": 4.740061162079511e-05,
+      "loss": 0.8923,
       "step": 480
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 4.704038997214485e-05,
-      "loss": 0.7843,
+      "epoch": 0.75,
+      "learning_rate": 4.72307169554876e-05,
+      "loss": 0.7103,
       "step": 490
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 4.686629526462396e-05,
-      "loss": 0.8069,
+      "epoch": 0.76,
+      "learning_rate": 4.706082229018009e-05,
+      "loss": 0.6821,
       "step": 500
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 4.669220055710306e-05,
-      "loss": 0.7725,
+      "epoch": 0.78,
+      "learning_rate": 4.6890927624872586e-05,
+      "loss": 0.7814,
       "step": 510
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 4.6518105849582176e-05,
-      "loss": 0.8299,
+      "epoch": 0.79,
+      "learning_rate": 4.672103295956507e-05,
+      "loss": 0.7257,
       "step": 520
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 4.634401114206128e-05,
-      "loss": 0.7864,
+      "epoch": 0.81,
+      "learning_rate": 4.6551138294257565e-05,
+      "loss": 0.7441,
       "step": 530
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 4.6169916434540394e-05,
-      "loss": 0.8226,
+      "epoch": 0.83,
+      "learning_rate": 4.638124362895006e-05,
+      "loss": 0.7044,
       "step": 540
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 4.59958217270195e-05,
-      "loss": 0.785,
+      "epoch": 0.84,
+      "learning_rate": 4.6211348963642544e-05,
+      "loss": 0.779,
       "step": 550
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 4.582172701949861e-05,
-      "loss": 0.8298,
+      "epoch": 0.86,
+      "learning_rate": 4.604145429833504e-05,
+      "loss": 0.8438,
       "step": 560
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 4.564763231197772e-05,
-      "loss": 0.7736,
+      "epoch": 0.87,
+      "learning_rate": 4.587155963302753e-05,
+      "loss": 0.7342,
       "step": 570
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 4.547353760445683e-05,
-      "loss": 0.8041,
+      "epoch": 0.89,
+      "learning_rate": 4.5701664967720016e-05,
+      "loss": 0.6886,
       "step": 580
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 4.529944289693593e-05,
-      "loss": 0.7805,
+      "epoch": 0.9,
+      "learning_rate": 4.553177030241251e-05,
+      "loss": 0.7733,
       "step": 590
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 4.5125348189415044e-05,
-      "loss": 0.8247,
+      "epoch": 0.92,
+      "learning_rate": 4.5361875637104995e-05,
+      "loss": 0.7655,
       "step": 600
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 4.495125348189415e-05,
-      "loss": 0.7941,
+      "epoch": 0.93,
+      "learning_rate": 4.519198097179749e-05,
+      "loss": 0.6929,
       "step": 610
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 4.477715877437326e-05,
-      "loss": 0.7644,
+      "epoch": 0.95,
+      "learning_rate": 4.502208630648998e-05,
+      "loss": 0.788,
       "step": 620
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 4.460306406685237e-05,
-      "loss": 0.77,
+      "epoch": 0.96,
+      "learning_rate": 4.4852191641182466e-05,
+      "loss": 0.6691,
       "step": 630
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 4.442896935933148e-05,
-      "loss": 0.778,
+      "epoch": 0.98,
+      "learning_rate": 4.468229697587496e-05,
+      "loss": 0.6881,
       "step": 640
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 4.425487465181059e-05,
-      "loss": 0.8061,
+      "epoch": 0.99,
+      "learning_rate": 4.451240231056745e-05,
+      "loss": 0.7175,
       "step": 650
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 4.40807799442897e-05,
-      "loss": 0.7544,
+      "epoch": 1.0,
+      "eval_accuracy": 0.7308500477554919,
+      "eval_loss": 0.7081200480461121,
+      "eval_runtime": 48.365,
+      "eval_samples_per_second": 108.24,
+      "eval_steps_per_second": 13.543,
+      "step": 654
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.434250764525994e-05,
+      "loss": 0.6844,
       "step": 660
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 4.39066852367688e-05,
-      "loss": 0.8129,
+      "epoch": 1.02,
+      "learning_rate": 4.417261297995243e-05,
+      "loss": 0.7542,
       "step": 670
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 4.373259052924791e-05,
-      "loss": 0.7776,
+      "epoch": 1.04,
+      "learning_rate": 4.400271831464492e-05,
+      "loss": 0.6111,
       "step": 680
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 4.355849582172702e-05,
-      "loss": 0.8078,
+      "epoch": 1.05,
+      "learning_rate": 4.383282364933741e-05,
+      "loss": 0.6169,
       "step": 690
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 4.338440111420613e-05,
-      "loss": 0.7845,
+      "epoch": 1.07,
+      "learning_rate": 4.36629289840299e-05,
+      "loss": 0.7111,
       "step": 700
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 4.321030640668524e-05,
-      "loss": 0.7826,
+      "epoch": 1.09,
+      "learning_rate": 4.349303431872239e-05,
+      "loss": 0.732,
       "step": 710
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 4.303621169916435e-05,
-      "loss": 0.7844,
+      "epoch": 1.1,
+      "learning_rate": 4.332313965341488e-05,
+      "loss": 0.623,
       "step": 720
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 4.286211699164346e-05,
-      "loss": 0.7905,
+      "epoch": 1.12,
+      "learning_rate": 4.3153244988107375e-05,
+      "loss": 0.6272,
       "step": 730
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 4.268802228412256e-05,
-      "loss": 0.7858,
+      "epoch": 1.13,
+      "learning_rate": 4.298335032279987e-05,
+      "loss": 0.6331,
       "step": 740
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 4.2513927576601673e-05,
-      "loss": 0.822,
+      "epoch": 1.15,
+      "learning_rate": 4.281345565749236e-05,
+      "loss": 0.6036,
       "step": 750
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 4.233983286908078e-05,
-      "loss": 0.7246,
+      "epoch": 1.16,
+      "learning_rate": 4.264356099218485e-05,
+      "loss": 0.6559,
       "step": 760
     },
     {
-      "epoch": 0.96,
-      "learning_rate": 4.216573816155989e-05,
-      "loss": 0.797,
+      "epoch": 1.18,
+      "learning_rate": 4.247366632687734e-05,
+      "loss": 0.5873,
       "step": 770
     },
     {
-      "epoch": 0.98,
-      "learning_rate": 4.1991643454039e-05,
-      "loss": 0.7749,
+      "epoch": 1.19,
+      "learning_rate": 4.230377166156983e-05,
+      "loss": 0.6016,
       "step": 780
     },
     {
-      "epoch": 0.99,
-      "learning_rate": 4.181754874651811e-05,
-      "loss": 0.7964,
+      "epoch": 1.21,
+      "learning_rate": 4.213387699626232e-05,
+      "loss": 0.6399,
       "step": 790
     },
     {
-      "epoch": 1.0,
-      "eval_accuracy": 0.7270911654135338,
-      "eval_loss": 0.7869051694869995,
-      "eval_runtime": 240.7482,
-      "eval_samples_per_second": 106.069,
-      "eval_steps_per_second": 3.315,
-      "step": 798
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 4.164345403899722e-05,
-      "loss": 0.7452,
+      "epoch": 1.22,
+      "learning_rate": 4.196398233095481e-05,
+      "loss": 0.6719,
       "step": 800
     },
     {
-      "epoch": 1.02,
-      "learning_rate": 4.146935933147633e-05,
-      "loss": 0.6894,
+      "epoch": 1.24,
+      "learning_rate": 4.1794087665647304e-05,
+      "loss": 0.5458,
       "step": 810
     },
     {
-      "epoch": 1.03,
-      "learning_rate": 4.129526462395543e-05,
-      "loss": 0.7115,
+      "epoch": 1.25,
+      "learning_rate": 4.162419300033979e-05,
+      "loss": 0.6167,
       "step": 820
     },
     {
-      "epoch": 1.04,
-      "learning_rate": 4.112116991643454e-05,
-      "loss": 0.6841,
+      "epoch": 1.27,
+      "learning_rate": 4.145429833503228e-05,
+      "loss": 0.6865,
       "step": 830
     },
     {
-      "epoch": 1.05,
-      "learning_rate": 4.094707520891365e-05,
-      "loss": 0.6697,
+      "epoch": 1.28,
+      "learning_rate": 4.1284403669724776e-05,
+      "loss": 0.6527,
       "step": 840
     },
     {
-      "epoch": 1.07,
-      "learning_rate": 4.077298050139276e-05,
-      "loss": 0.6796,
+      "epoch": 1.3,
+      "learning_rate": 4.111450900441726e-05,
+      "loss": 0.5855,
       "step": 850
     },
     {
-      "epoch": 1.08,
-      "learning_rate": 4.0598885793871866e-05,
-      "loss": 0.7095,
+      "epoch": 1.31,
+      "learning_rate": 4.0944614339109755e-05,
+      "loss": 0.5976,
       "step": 860
     },
     {
-      "epoch": 1.09,
-      "learning_rate": 4.042479108635098e-05,
-      "loss": 0.667,
+      "epoch": 1.33,
+      "learning_rate": 4.077471967380224e-05,
+      "loss": 0.6886,
       "step": 870
     },
     {
-      "epoch": 1.1,
-      "learning_rate": 4.0250696378830085e-05,
-      "loss": 0.678,
+      "epoch": 1.35,
+      "learning_rate": 4.0604825008494734e-05,
+      "loss": 0.6992,
       "step": 880
     },
     {
-      "epoch": 1.12,
-      "learning_rate": 4.007660167130919e-05,
-      "loss": 0.6771,
+      "epoch": 1.36,
+      "learning_rate": 4.043493034318723e-05,
+      "loss": 0.6413,
       "step": 890
     },
     {
-      "epoch": 1.13,
-      "learning_rate": 3.9902506963788303e-05,
-      "loss": 0.7135,
+      "epoch": 1.38,
+      "learning_rate": 4.026503567787971e-05,
+      "loss": 0.6227,
       "step": 900
     },
     {
-      "epoch": 1.14,
-      "learning_rate": 3.972841225626741e-05,
-      "loss": 0.6982,
+      "epoch": 1.39,
+      "learning_rate": 4.0095141012572206e-05,
+      "loss": 0.611,
       "step": 910
     },
     {
-      "epoch": 1.15,
-      "learning_rate": 3.955431754874652e-05,
-      "loss": 0.7658,
+      "epoch": 1.41,
+      "learning_rate": 3.99252463472647e-05,
+      "loss": 0.5908,
       "step": 920
     },
     {
-      "epoch": 1.17,
-      "learning_rate": 3.938022284122563e-05,
-      "loss": 0.6841,
+      "epoch": 1.42,
+      "learning_rate": 3.9755351681957185e-05,
+      "loss": 0.6177,
       "step": 930
     },
     {
-      "epoch": 1.18,
-      "learning_rate": 3.920612813370474e-05,
-      "loss": 0.7158,
+      "epoch": 1.44,
+      "learning_rate": 3.958545701664968e-05,
+      "loss": 0.6077,
       "step": 940
     },
     {
-      "epoch": 1.19,
-      "learning_rate": 3.903203342618385e-05,
-      "loss": 0.7176,
+      "epoch": 1.45,
+      "learning_rate": 3.941556235134217e-05,
+      "loss": 0.6321,
       "step": 950
     },
     {
-      "epoch": 1.2,
-      "learning_rate": 3.885793871866296e-05,
-      "loss": 0.7032,
+      "epoch": 1.47,
+      "learning_rate": 3.9245667686034656e-05,
+      "loss": 0.5838,
       "step": 960
     },
     {
-      "epoch": 1.22,
-      "learning_rate": 3.868384401114206e-05,
-      "loss": 0.7131,
+      "epoch": 1.48,
+      "learning_rate": 3.907577302072715e-05,
+      "loss": 0.6619,
       "step": 970
     },
     {
-      "epoch": 1.23,
-      "learning_rate": 3.850974930362117e-05,
-      "loss": 0.7125,
+      "epoch": 1.5,
+      "learning_rate": 3.890587835541964e-05,
+      "loss": 0.5823,
       "step": 980
     },
     {
-      "epoch": 1.24,
-      "learning_rate": 3.833565459610028e-05,
-      "loss": 0.7089,
+      "epoch": 1.51,
+      "learning_rate": 3.8735983690112135e-05,
+      "loss": 0.5496,
       "step": 990
     },
     {
-      "epoch": 1.25,
-      "learning_rate": 3.816155988857939e-05,
-      "loss": 0.7246,
+      "epoch": 1.53,
+      "learning_rate": 3.856608902480463e-05,
+      "loss": 0.6632,
       "step": 1000
     },
     {
-      "epoch": 1.27,
-      "learning_rate": 3.7987465181058496e-05,
-      "loss": 0.6763,
+      "epoch": 1.54,
+      "learning_rate": 3.8396194359497114e-05,
+      "loss": 0.5744,
       "step": 1010
     },
     {
-      "epoch": 1.28,
-      "learning_rate": 3.781337047353761e-05,
-      "loss": 0.6693,
+      "epoch": 1.56,
+      "learning_rate": 3.822629969418961e-05,
+      "loss": 0.6259,
       "step": 1020
     },
     {
-      "epoch": 1.29,
-      "learning_rate": 3.7639275766016715e-05,
-      "loss": 0.6912,
+      "epoch": 1.57,
+      "learning_rate": 3.80564050288821e-05,
+      "loss": 0.6156,
       "step": 1030
     },
     {
-      "epoch": 1.3,
-      "learning_rate": 3.746518105849583e-05,
-      "loss": 0.6866,
+      "epoch": 1.59,
+      "learning_rate": 3.7886510363574586e-05,
+      "loss": 0.5593,
       "step": 1040
     },
     {
-      "epoch": 1.32,
-      "learning_rate": 3.7291086350974927e-05,
-      "loss": 0.6532,
+      "epoch": 1.6,
+      "learning_rate": 3.771661569826708e-05,
+      "loss": 0.6951,
       "step": 1050
     },
     {
-      "epoch": 1.33,
-      "learning_rate": 3.711699164345404e-05,
-      "loss": 0.7283,
+      "epoch": 1.62,
+      "learning_rate": 3.7546721032959565e-05,
+      "loss": 0.6099,
       "step": 1060
     },
     {
-      "epoch": 1.34,
-      "learning_rate": 3.6942896935933145e-05,
-      "loss": 0.7032,
+      "epoch": 1.64,
+      "learning_rate": 3.737682636765206e-05,
+      "loss": 0.6019,
       "step": 1070
     },
     {
-      "epoch": 1.35,
-      "learning_rate": 3.676880222841226e-05,
-      "loss": 0.6854,
+      "epoch": 1.65,
+      "learning_rate": 3.720693170234455e-05,
+      "loss": 0.5226,
       "step": 1080
     },
     {
-      "epoch": 1.37,
-      "learning_rate": 3.659470752089137e-05,
-      "loss": 0.6769,
+      "epoch": 1.67,
+      "learning_rate": 3.7037037037037037e-05,
+      "loss": 0.6539,
       "step": 1090
     },
     {
-      "epoch": 1.38,
-      "learning_rate": 3.642061281337048e-05,
-      "loss": 0.7349,
+      "epoch": 1.68,
+      "learning_rate": 3.686714237172953e-05,
+      "loss": 0.6031,
       "step": 1100
     },
     {
-      "epoch": 1.39,
-      "learning_rate": 3.624651810584959e-05,
-      "loss": 0.7274,
+      "epoch": 1.7,
+      "learning_rate": 3.669724770642202e-05,
+      "loss": 0.6356,
       "step": 1110
     },
     {
-      "epoch": 1.4,
-      "learning_rate": 3.607242339832869e-05,
-      "loss": 0.6858,
+      "epoch": 1.71,
+      "learning_rate": 3.652735304111451e-05,
+      "loss": 0.6456,
       "step": 1120
     },
     {
-      "epoch": 1.42,
-      "learning_rate": 3.58983286908078e-05,
-      "loss": 0.7229,
+      "epoch": 1.73,
+      "learning_rate": 3.6357458375807e-05,
+      "loss": 0.5222,
       "step": 1130
     },
     {
-      "epoch": 1.43,
-      "learning_rate": 3.572423398328691e-05,
-      "loss": 0.6924,
+      "epoch": 1.74,
+      "learning_rate": 3.6187563710499494e-05,
+      "loss": 0.5715,
       "step": 1140
     },
     {
-      "epoch": 1.44,
-      "learning_rate": 3.555013927576602e-05,
-      "loss": 0.686,
+      "epoch": 1.76,
+      "learning_rate": 3.601766904519198e-05,
+      "loss": 0.6142,
       "step": 1150
     },
     {
-      "epoch": 1.45,
-      "learning_rate": 3.5376044568245126e-05,
-      "loss": 0.7268,
+      "epoch": 1.77,
+      "learning_rate": 3.584777437988447e-05,
+      "loss": 0.6794,
       "step": 1160
     },
     {
-      "epoch": 1.47,
-      "learning_rate": 3.520194986072424e-05,
-      "loss": 0.7107,
+      "epoch": 1.79,
+      "learning_rate": 3.567787971457696e-05,
+      "loss": 0.7228,
       "step": 1170
     },
     {
-      "epoch": 1.48,
-      "learning_rate": 3.5027855153203345e-05,
-      "loss": 0.7095,
+      "epoch": 1.8,
+      "learning_rate": 3.550798504926945e-05,
+      "loss": 0.6374,
       "step": 1180
     },
     {
-      "epoch": 1.49,
-      "learning_rate": 3.485376044568246e-05,
-      "loss": 0.6764,
+      "epoch": 1.82,
+      "learning_rate": 3.5338090383961945e-05,
+      "loss": 0.6078,
       "step": 1190
     },
     {
-      "epoch": 1.5,
-      "learning_rate": 3.4679665738161556e-05,
-      "loss": 0.7076,
+      "epoch": 1.83,
+      "learning_rate": 3.516819571865443e-05,
+      "loss": 0.6115,
       "step": 1200
     },
     {
-      "epoch": 1.52,
-      "learning_rate": 3.450557103064067e-05,
-      "loss": 0.7194,
+      "epoch": 1.85,
+      "learning_rate": 3.4998301053346924e-05,
+      "loss": 0.6853,
       "step": 1210
     },
     {
-      "epoch": 1.53,
-      "learning_rate": 3.4331476323119775e-05,
-      "loss": 0.7164,
+      "epoch": 1.86,
+      "learning_rate": 3.482840638803942e-05,
+      "loss": 0.5883,
       "step": 1220
     },
     {
-      "epoch": 1.54,
-      "learning_rate": 3.415738161559889e-05,
-      "loss": 0.6958,
+      "epoch": 1.88,
+      "learning_rate": 3.465851172273191e-05,
+      "loss": 0.6143,
       "step": 1230
     },
     {
-      "epoch": 1.55,
-      "learning_rate": 3.3983286908077994e-05,
-      "loss": 0.662,
+      "epoch": 1.9,
+      "learning_rate": 3.44886170574244e-05,
+      "loss": 0.6067,
       "step": 1240
     },
     {
-      "epoch": 1.57,
-      "learning_rate": 3.3809192200557107e-05,
-      "loss": 0.6475,
+      "epoch": 1.91,
+      "learning_rate": 3.431872239211689e-05,
+      "loss": 0.64,
       "step": 1250
     },
     {
-      "epoch": 1.58,
-      "learning_rate": 3.363509749303621e-05,
-      "loss": 0.7169,
+      "epoch": 1.93,
+      "learning_rate": 3.414882772680938e-05,
+      "loss": 0.6043,
       "step": 1260
     },
     {
-      "epoch": 1.59,
-      "learning_rate": 3.346100278551532e-05,
-      "loss": 0.6948,
+      "epoch": 1.94,
+      "learning_rate": 3.3978933061501874e-05,
+      "loss": 0.5755,
       "step": 1270
     },
     {
-      "epoch": 1.6,
-      "learning_rate": 3.328690807799443e-05,
-      "loss": 0.7413,
+      "epoch": 1.96,
+      "learning_rate": 3.380903839619436e-05,
+      "loss": 0.6534,
       "step": 1280
     },
     {
-      "epoch": 1.62,
-      "learning_rate": 3.311281337047354e-05,
-      "loss": 0.6824,
+      "epoch": 1.97,
+      "learning_rate": 3.363914373088685e-05,
+      "loss": 0.5583,
       "step": 1290
     },
     {
-      "epoch": 1.63,
-      "learning_rate": 3.293871866295265e-05,
-      "loss": 0.7024,
+      "epoch": 1.99,
+      "learning_rate": 3.3469249065579346e-05,
+      "loss": 0.6952,
       "step": 1300
     },
     {
-      "epoch": 1.64,
-      "learning_rate": 3.2764623955431756e-05,
-      "loss": 0.6748,
+      "epoch": 2.0,
+      "eval_accuracy": 0.7379178605539637,
+      "eval_loss": 0.6931169629096985,
+      "eval_runtime": 48.247,
+      "eval_samples_per_second": 108.504,
+      "eval_steps_per_second": 13.576,
+      "step": 1308
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.329935440027183e-05,
+      "loss": 0.5479,
       "step": 1310
     },
     {
-      "epoch": 1.65,
-      "learning_rate": 3.259052924791087e-05,
-      "loss": 0.7112,
+      "epoch": 2.02,
+      "learning_rate": 3.3129459734964325e-05,
+      "loss": 0.523,
       "step": 1320
     },
     {
-      "epoch": 1.67,
-      "learning_rate": 3.2416434540389975e-05,
-      "loss": 0.6878,
+      "epoch": 2.03,
+      "learning_rate": 3.295956506965682e-05,
+      "loss": 0.4421,
       "step": 1330
     },
     {
-      "epoch": 1.68,
-      "learning_rate": 3.224233983286909e-05,
-      "loss": 0.7114,
+      "epoch": 2.05,
+      "learning_rate": 3.2789670404349304e-05,
+      "loss": 0.4887,
       "step": 1340
     },
     {
-      "epoch": 1.69,
-      "learning_rate": 3.2068245125348186e-05,
-      "loss": 0.6924,
+      "epoch": 2.06,
+      "learning_rate": 3.26197757390418e-05,
+      "loss": 0.4665,
       "step": 1350
     },
     {
-      "epoch": 1.7,
-      "learning_rate": 3.18941504178273e-05,
-      "loss": 0.7354,
+      "epoch": 2.08,
+      "learning_rate": 3.244988107373428e-05,
+      "loss": 0.5767,
       "step": 1360
     },
     {
-      "epoch": 1.72,
-      "learning_rate": 3.1720055710306405e-05,
-      "loss": 0.6757,
+      "epoch": 2.09,
+      "learning_rate": 3.2279986408426776e-05,
+      "loss": 0.4944,
       "step": 1370
     },
     {
-      "epoch": 1.73,
-      "learning_rate": 3.154596100278552e-05,
-      "loss": 0.6675,
+      "epoch": 2.11,
+      "learning_rate": 3.211009174311927e-05,
+      "loss": 0.4159,
       "step": 1380
     },
     {
-      "epoch": 1.74,
-      "learning_rate": 3.1371866295264624e-05,
-      "loss": 0.6923,
+      "epoch": 2.12,
+      "learning_rate": 3.1940197077811755e-05,
+      "loss": 0.3825,
       "step": 1390
     },
     {
-      "epoch": 1.75,
-      "learning_rate": 3.1197771587743737e-05,
-      "loss": 0.665,
+      "epoch": 2.14,
+      "learning_rate": 3.177030241250425e-05,
+      "loss": 0.46,
       "step": 1400
     },
     {
-      "epoch": 1.77,
-      "learning_rate": 3.102367688022284e-05,
-      "loss": 0.6473,
+      "epoch": 2.16,
+      "learning_rate": 3.160040774719674e-05,
+      "loss": 0.4904,
       "step": 1410
     },
     {
-      "epoch": 1.78,
-      "learning_rate": 3.0849582172701955e-05,
-      "loss": 0.6194,
+      "epoch": 2.17,
+      "learning_rate": 3.1430513081889227e-05,
+      "loss": 0.5993,
       "step": 1420
     },
     {
-      "epoch": 1.79,
-      "learning_rate": 3.0675487465181054e-05,
-      "loss": 0.6535,
+      "epoch": 2.19,
+      "learning_rate": 3.126061841658172e-05,
+      "loss": 0.5107,
       "step": 1430
     },
     {
-      "epoch": 1.8,
-      "learning_rate": 3.050139275766017e-05,
-      "loss": 0.6704,
+      "epoch": 2.2,
+      "learning_rate": 3.1090723751274206e-05,
+      "loss": 0.4616,
       "step": 1440
     },
     {
-      "epoch": 1.82,
-      "learning_rate": 3.0327298050139273e-05,
-      "loss": 0.713,
+      "epoch": 2.22,
+      "learning_rate": 3.09208290859667e-05,
+      "loss": 0.4318,
       "step": 1450
     },
     {
-      "epoch": 1.83,
-      "learning_rate": 3.0153203342618386e-05,
-      "loss": 0.7435,
+      "epoch": 2.23,
+      "learning_rate": 3.075093442065919e-05,
+      "loss": 0.4934,
       "step": 1460
     },
     {
-      "epoch": 1.84,
-      "learning_rate": 2.99791086350975e-05,
-      "loss": 0.7073,
+      "epoch": 2.25,
+      "learning_rate": 3.0581039755351684e-05,
+      "loss": 0.4119,
       "step": 1470
     },
     {
-      "epoch": 1.85,
-      "learning_rate": 2.98050139275766e-05,
-      "loss": 0.6691,
+      "epoch": 2.26,
+      "learning_rate": 3.0411145090044174e-05,
+      "loss": 0.5154,
       "step": 1480
     },
     {
-      "epoch": 1.87,
-      "learning_rate": 2.9630919220055714e-05,
-      "loss": 0.6859,
+      "epoch": 2.28,
+      "learning_rate": 3.0241250424736666e-05,
+      "loss": 0.5126,
       "step": 1490
     },
     {
-      "epoch": 1.88,
-      "learning_rate": 2.945682451253482e-05,
-      "loss": 0.7689,
+      "epoch": 2.29,
+      "learning_rate": 3.0071355759429153e-05,
+      "loss": 0.429,
       "step": 1500
     },
     {
-      "epoch": 1.89,
-      "learning_rate": 2.9282729805013932e-05,
-      "loss": 0.6777,
+      "epoch": 2.31,
+      "learning_rate": 2.9901461094121645e-05,
+      "loss": 0.5008,
       "step": 1510
     },
     {
-      "epoch": 1.9,
-      "learning_rate": 2.9108635097493035e-05,
-      "loss": 0.6909,
+      "epoch": 2.32,
+      "learning_rate": 2.9731566428814138e-05,
+      "loss": 0.4074,
       "step": 1520
     },
     {
-      "epoch": 1.92,
-      "learning_rate": 2.8934540389972148e-05,
-      "loss": 0.7016,
+      "epoch": 2.34,
+      "learning_rate": 2.9561671763506628e-05,
+      "loss": 0.4563,
       "step": 1530
     },
     {
-      "epoch": 1.93,
-      "learning_rate": 2.8760445682451254e-05,
-      "loss": 0.701,
+      "epoch": 2.35,
+      "learning_rate": 2.939177709819912e-05,
+      "loss": 0.5493,
       "step": 1540
     },
     {
-      "epoch": 1.94,
-      "learning_rate": 2.8586350974930366e-05,
-      "loss": 0.7016,
+      "epoch": 2.37,
+      "learning_rate": 2.9221882432891607e-05,
+      "loss": 0.5255,
       "step": 1550
     },
     {
-      "epoch": 1.95,
-      "learning_rate": 2.841225626740947e-05,
-      "loss": 0.6674,
+      "epoch": 2.38,
+      "learning_rate": 2.90519877675841e-05,
+      "loss": 0.4812,
       "step": 1560
     },
     {
-      "epoch": 1.97,
-      "learning_rate": 2.8238161559888582e-05,
-      "loss": 0.6663,
+      "epoch": 2.4,
+      "learning_rate": 2.8882093102276592e-05,
+      "loss": 0.4576,
       "step": 1570
     },
     {
-      "epoch": 1.98,
-      "learning_rate": 2.8064066852367688e-05,
-      "loss": 0.6621,
+      "epoch": 2.41,
+      "learning_rate": 2.871219843696908e-05,
+      "loss": 0.5143,
       "step": 1580
     },
     {
-      "epoch": 1.99,
-      "learning_rate": 2.78899721448468e-05,
-      "loss": 0.6567,
+      "epoch": 2.43,
+      "learning_rate": 2.854230377166157e-05,
+      "loss": 0.4439,
       "step": 1590
     },
     {
-      "epoch": 2.0,
-      "eval_accuracy": 0.738016917293233,
-      "eval_loss": 0.7538765668869019,
-      "eval_runtime": 243.4161,
-      "eval_samples_per_second": 104.907,
-      "eval_steps_per_second": 3.278,
-      "step": 1596
-    },
-    {
-      "epoch": 2.01,
-      "learning_rate": 3.69923416105825e-05,
-      "loss": 0.651,
+      "epoch": 2.45,
+      "learning_rate": 2.8372409106354064e-05,
+      "loss": 0.4637,
       "step": 1600
     },
     {
-      "epoch": 2.02,
-      "learning_rate": 3.687630540728708e-05,
-      "loss": 0.5874,
+      "epoch": 2.46,
+      "learning_rate": 2.820251444104655e-05,
+      "loss": 0.4575,
       "step": 1610
     },
     {
-      "epoch": 2.03,
-      "learning_rate": 3.6760269203991646e-05,
-      "loss": 0.6426,
+      "epoch": 2.48,
+      "learning_rate": 2.8032619775739043e-05,
+      "loss": 0.4955,
       "step": 1620
     },
     {
-      "epoch": 2.04,
-      "learning_rate": 3.664423300069622e-05,
-      "loss": 0.6418,
+      "epoch": 2.49,
+      "learning_rate": 2.7862725110431533e-05,
+      "loss": 0.446,
       "step": 1630
     },
     {
-      "epoch": 2.06,
-      "learning_rate": 3.652819679740079e-05,
-      "loss": 0.6023,
+      "epoch": 2.51,
+      "learning_rate": 2.7692830445124026e-05,
+      "loss": 0.5241,
       "step": 1640
     },
     {
-      "epoch": 2.07,
-      "learning_rate": 3.641216059410536e-05,
-      "loss": 0.6012,
+      "epoch": 2.52,
+      "learning_rate": 2.7522935779816515e-05,
+      "loss": 0.4872,
       "step": 1650
     },
     {
-      "epoch": 2.08,
-      "learning_rate": 3.6296124390809935e-05,
-      "loss": 0.6276,
+      "epoch": 2.54,
+      "learning_rate": 2.7353041114509004e-05,
+      "loss": 0.4526,
       "step": 1660
     },
     {
-      "epoch": 2.09,
-      "learning_rate": 3.6180088187514504e-05,
-      "loss": 0.5936,
+      "epoch": 2.55,
+      "learning_rate": 2.7183146449201497e-05,
+      "loss": 0.513,
       "step": 1670
     },
     {
-      "epoch": 2.11,
-      "learning_rate": 3.606405198421908e-05,
-      "loss": 0.6586,
+      "epoch": 2.57,
+      "learning_rate": 2.701325178389399e-05,
+      "loss": 0.5036,
       "step": 1680
     },
     {
-      "epoch": 2.12,
-      "learning_rate": 3.594801578092365e-05,
-      "loss": 0.6888,
+      "epoch": 2.58,
+      "learning_rate": 2.6843357118586476e-05,
+      "loss": 0.4714,
       "step": 1690
     },
     {
-      "epoch": 2.13,
-      "learning_rate": 3.5831979577628224e-05,
-      "loss": 0.6365,
+      "epoch": 2.6,
+      "learning_rate": 2.667346245327897e-05,
+      "loss": 0.4228,
       "step": 1700
     },
     {
-      "epoch": 2.14,
-      "learning_rate": 3.571594337433279e-05,
-      "loss": 0.627,
+      "epoch": 2.61,
+      "learning_rate": 2.6503567787971462e-05,
+      "loss": 0.4916,
       "step": 1710
     },
     {
-      "epoch": 2.16,
-      "learning_rate": 3.559990717103737e-05,
-      "loss": 0.6088,
+      "epoch": 2.63,
+      "learning_rate": 2.6333673122663948e-05,
+      "loss": 0.5439,
       "step": 1720
     },
     {
-      "epoch": 2.17,
-      "learning_rate": 3.548387096774194e-05,
-      "loss": 0.6739,
+      "epoch": 2.64,
+      "learning_rate": 2.616377845735644e-05,
+      "loss": 0.4702,
       "step": 1730
     },
     {
-      "epoch": 2.18,
-      "learning_rate": 3.536783476444651e-05,
-      "loss": 0.6618,
+      "epoch": 2.66,
+      "learning_rate": 2.5993883792048927e-05,
+      "loss": 0.5041,
       "step": 1740
     },
     {
-      "epoch": 2.19,
-      "learning_rate": 3.5251798561151075e-05,
-      "loss": 0.5868,
+      "epoch": 2.67,
+      "learning_rate": 2.582398912674142e-05,
+      "loss": 0.4341,
       "step": 1750
     },
     {
-      "epoch": 2.21,
-      "learning_rate": 3.513576235785565e-05,
-      "loss": 0.6162,
+      "epoch": 2.69,
+      "learning_rate": 2.5654094461433913e-05,
+      "loss": 0.4781,
       "step": 1760
     },
     {
-      "epoch": 2.22,
-      "learning_rate": 3.501972615456022e-05,
-      "loss": 0.6023,
+      "epoch": 2.71,
+      "learning_rate": 2.5484199796126402e-05,
+      "loss": 0.4447,
       "step": 1770
     },
     {
-      "epoch": 2.23,
-      "learning_rate": 3.4903689951264796e-05,
-      "loss": 0.6005,
+      "epoch": 2.72,
+      "learning_rate": 2.5314305130818895e-05,
+      "loss": 0.424,
       "step": 1780
     },
     {
-      "epoch": 2.24,
-      "learning_rate": 3.4787653747969365e-05,
-      "loss": 0.6084,
+      "epoch": 2.74,
+      "learning_rate": 2.5144410465511388e-05,
+      "loss": 0.5021,
       "step": 1790
     },
     {
-      "epoch": 2.26,
-      "learning_rate": 3.467161754467394e-05,
-      "loss": 0.5963,
+      "epoch": 2.75,
+      "learning_rate": 2.4974515800203874e-05,
+      "loss": 0.4491,
       "step": 1800
     },
     {
-      "epoch": 2.27,
-      "learning_rate": 3.455558134137851e-05,
-      "loss": 0.6816,
+      "epoch": 2.77,
+      "learning_rate": 2.4804621134896367e-05,
+      "loss": 0.47,
       "step": 1810
     },
     {
-      "epoch": 2.28,
-      "learning_rate": 3.4439545138083085e-05,
-      "loss": 0.6564,
+      "epoch": 2.78,
+      "learning_rate": 2.4634726469588856e-05,
+      "loss": 0.4573,
       "step": 1820
     },
     {
-      "epoch": 2.29,
-      "learning_rate": 3.4323508934787654e-05,
-      "loss": 0.6332,
+      "epoch": 2.8,
+      "learning_rate": 2.4464831804281346e-05,
+      "loss": 0.4301,
       "step": 1830
     },
     {
-      "epoch": 2.31,
-      "learning_rate": 3.420747273149223e-05,
-      "loss": 0.6295,
+      "epoch": 2.81,
+      "learning_rate": 2.4294937138973835e-05,
+      "loss": 0.5122,
       "step": 1840
     },
     {
-      "epoch": 2.32,
-      "learning_rate": 3.40914365281968e-05,
-      "loss": 0.6467,
+      "epoch": 2.83,
+      "learning_rate": 2.4125042473666328e-05,
+      "loss": 0.4943,
       "step": 1850
     },
     {
-      "epoch": 2.33,
-      "learning_rate": 3.397540032490137e-05,
-      "loss": 0.622,
+      "epoch": 2.84,
+      "learning_rate": 2.3955147808358818e-05,
+      "loss": 0.4656,
       "step": 1860
     },
     {
-      "epoch": 2.34,
-      "learning_rate": 3.385936412160594e-05,
-      "loss": 0.662,
+      "epoch": 2.86,
+      "learning_rate": 2.3785253143051307e-05,
+      "loss": 0.4625,
       "step": 1870
     },
     {
-      "epoch": 2.36,
-      "learning_rate": 3.374332791831051e-05,
-      "loss": 0.6514,
+      "epoch": 2.87,
+      "learning_rate": 2.36153584777438e-05,
+      "loss": 0.5568,
       "step": 1880
     },
     {
-      "epoch": 2.37,
-      "learning_rate": 3.362729171501509e-05,
-      "loss": 0.6493,
+      "epoch": 2.89,
+      "learning_rate": 2.3445463812436293e-05,
+      "loss": 0.3586,
       "step": 1890
     },
     {
-      "epoch": 2.38,
-      "learning_rate": 3.351125551171966e-05,
-      "loss": 0.6167,
+      "epoch": 2.9,
+      "learning_rate": 2.3275569147128782e-05,
+      "loss": 0.5519,
       "step": 1900
     },
     {
-      "epoch": 2.39,
-      "learning_rate": 3.339521930842423e-05,
-      "loss": 0.6575,
+      "epoch": 2.92,
+      "learning_rate": 2.3105674481821272e-05,
+      "loss": 0.5399,
       "step": 1910
     },
     {
-      "epoch": 2.41,
-      "learning_rate": 3.32791831051288e-05,
-      "loss": 0.5828,
+      "epoch": 2.93,
+      "learning_rate": 2.2935779816513765e-05,
+      "loss": 0.4763,
       "step": 1920
     },
     {
-      "epoch": 2.42,
-      "learning_rate": 3.316314690183338e-05,
-      "loss": 0.6323,
+      "epoch": 2.95,
+      "learning_rate": 2.2765885151206254e-05,
+      "loss": 0.5894,
       "step": 1930
     },
     {
-      "epoch": 2.43,
-      "learning_rate": 3.3047110698537946e-05,
-      "loss": 0.6285,
+      "epoch": 2.97,
+      "learning_rate": 2.2595990485898744e-05,
+      "loss": 0.5012,
       "step": 1940
     },
     {
-      "epoch": 2.44,
-      "learning_rate": 3.293107449524252e-05,
-      "loss": 0.5822,
+      "epoch": 2.98,
+      "learning_rate": 2.2426095820591233e-05,
+      "loss": 0.419,
       "step": 1950
     },
     {
-      "epoch": 2.46,
-      "learning_rate": 3.2815038291947084e-05,
-      "loss": 0.6441,
+      "epoch": 3.0,
+      "learning_rate": 2.2256201155283726e-05,
+      "loss": 0.5041,
       "step": 1960
     },
     {
-      "epoch": 2.47,
-      "learning_rate": 3.269900208865166e-05,
-      "loss": 0.6513,
+      "epoch": 3.0,
+      "eval_accuracy": 0.7444126074498567,
+      "eval_loss": 0.7038247585296631,
+      "eval_runtime": 52.7062,
+      "eval_samples_per_second": 99.324,
+      "eval_steps_per_second": 12.427,
+      "step": 1962
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.2086306489976216e-05,
+      "loss": 0.4415,
       "step": 1970
     },
     {
-      "epoch": 2.48,
-      "learning_rate": 3.258296588535623e-05,
-      "loss": 0.6371,
+      "epoch": 3.03,
+      "learning_rate": 2.1916411824668705e-05,
+      "loss": 0.2787,
       "step": 1980
     },
     {
-      "epoch": 2.49,
-      "learning_rate": 3.2466929682060804e-05,
-      "loss": 0.6149,
+      "epoch": 3.04,
+      "learning_rate": 2.1746517159361194e-05,
+      "loss": 0.4013,
       "step": 1990
     },
     {
-      "epoch": 2.51,
-      "learning_rate": 3.235089347876537e-05,
-      "loss": 0.6348,
+      "epoch": 3.06,
+      "learning_rate": 2.1576622494053687e-05,
+      "loss": 0.3279,
       "step": 2000
     },
     {
-      "epoch": 2.52,
-      "learning_rate": 3.223485727546995e-05,
-      "loss": 0.6046,
+      "epoch": 3.07,
+      "learning_rate": 2.140672782874618e-05,
+      "loss": 0.3274,
       "step": 2010
     },
     {
-      "epoch": 2.53,
-      "learning_rate": 3.211882107217452e-05,
-      "loss": 0.6119,
+      "epoch": 3.09,
+      "learning_rate": 2.123683316343867e-05,
+      "loss": 0.3664,
       "step": 2020
     },
     {
-      "epoch": 2.54,
-      "learning_rate": 3.200278486887909e-05,
-      "loss": 0.6382,
+      "epoch": 3.1,
+      "learning_rate": 2.106693849813116e-05,
+      "loss": 0.3076,
       "step": 2030
     },
     {
-      "epoch": 2.56,
-      "learning_rate": 3.188674866558366e-05,
-      "loss": 0.6272,
+      "epoch": 3.12,
+      "learning_rate": 2.0897043832823652e-05,
+      "loss": 0.4652,
       "step": 2040
     },
     {
-      "epoch": 2.57,
-      "learning_rate": 3.177071246228824e-05,
-      "loss": 0.6405,
+      "epoch": 3.13,
+      "learning_rate": 2.072714916751614e-05,
+      "loss": 0.358,
       "step": 2050
     },
     {
-      "epoch": 2.58,
-      "learning_rate": 3.165467625899281e-05,
-      "loss": 0.6679,
+      "epoch": 3.15,
+      "learning_rate": 2.055725450220863e-05,
+      "loss": 0.3003,
       "step": 2060
     },
     {
-      "epoch": 2.59,
-      "learning_rate": 3.1538640055697376e-05,
-      "loss": 0.6367,
+      "epoch": 3.16,
+      "learning_rate": 2.038735983690112e-05,
+      "loss": 0.3643,
       "step": 2070
     },
     {
-      "epoch": 2.61,
-      "learning_rate": 3.142260385240195e-05,
-      "loss": 0.6656,
+      "epoch": 3.18,
+      "learning_rate": 2.0217465171593613e-05,
+      "loss": 0.3784,
       "step": 2080
     },
     {
-      "epoch": 2.62,
-      "learning_rate": 3.130656764910652e-05,
-      "loss": 0.6427,
+      "epoch": 3.19,
+      "learning_rate": 2.0047570506286103e-05,
+      "loss": 0.3069,
       "step": 2090
     },
     {
-      "epoch": 2.63,
-      "learning_rate": 3.1190531445811096e-05,
-      "loss": 0.6316,
+      "epoch": 3.21,
+      "learning_rate": 1.9877675840978592e-05,
+      "loss": 0.3309,
       "step": 2100
     },
     {
-      "epoch": 2.64,
-      "learning_rate": 3.1074495242515665e-05,
-      "loss": 0.6047,
+      "epoch": 3.23,
+      "learning_rate": 1.9707781175671085e-05,
+      "loss": 0.3504,
       "step": 2110
     },
     {
-      "epoch": 2.66,
-      "learning_rate": 3.095845903922024e-05,
-      "loss": 0.6562,
+      "epoch": 3.24,
+      "learning_rate": 1.9537886510363575e-05,
+      "loss": 0.3749,
       "step": 2120
     },
     {
-      "epoch": 2.67,
-      "learning_rate": 3.084242283592481e-05,
-      "loss": 0.6164,
+      "epoch": 3.26,
+      "learning_rate": 1.9367991845056068e-05,
+      "loss": 0.3634,
       "step": 2130
     },
     {
-      "epoch": 2.68,
-      "learning_rate": 3.0726386632629385e-05,
-      "loss": 0.638,
+      "epoch": 3.27,
+      "learning_rate": 1.9198097179748557e-05,
+      "loss": 0.3223,
       "step": 2140
     },
     {
-      "epoch": 2.69,
-      "learning_rate": 3.0610350429333954e-05,
-      "loss": 0.6587,
+      "epoch": 3.29,
+      "learning_rate": 1.902820251444105e-05,
+      "loss": 0.4309,
       "step": 2150
     },
     {
-      "epoch": 2.71,
-      "learning_rate": 3.0494314226038523e-05,
-      "loss": 0.6784,
+      "epoch": 3.3,
+      "learning_rate": 1.885830784913354e-05,
+      "loss": 0.3728,
       "step": 2160
     },
     {
-      "epoch": 2.72,
-      "learning_rate": 3.0378278022743096e-05,
-      "loss": 0.6727,
+      "epoch": 3.32,
+      "learning_rate": 1.868841318382603e-05,
+      "loss": 0.2607,
       "step": 2170
     },
     {
-      "epoch": 2.73,
-      "learning_rate": 3.0262241819447668e-05,
-      "loss": 0.6163,
+      "epoch": 3.33,
+      "learning_rate": 1.8518518518518518e-05,
+      "loss": 0.3335,
       "step": 2180
     },
     {
-      "epoch": 2.74,
-      "learning_rate": 3.014620561615224e-05,
-      "loss": 0.6304,
+      "epoch": 3.35,
+      "learning_rate": 1.834862385321101e-05,
+      "loss": 0.2907,
       "step": 2190
     },
     {
-      "epoch": 2.76,
-      "learning_rate": 3.0030169412856812e-05,
-      "loss": 0.6053,
+      "epoch": 3.36,
+      "learning_rate": 1.81787291879035e-05,
+      "loss": 0.3103,
       "step": 2200
     },
     {
-      "epoch": 2.77,
-      "learning_rate": 2.9914133209561385e-05,
-      "loss": 0.6296,
+      "epoch": 3.38,
+      "learning_rate": 1.800883452259599e-05,
+      "loss": 0.3538,
       "step": 2210
     },
     {
-      "epoch": 2.78,
-      "learning_rate": 2.9798097006265957e-05,
-      "loss": 0.643,
+      "epoch": 3.39,
+      "learning_rate": 1.783893985728848e-05,
+      "loss": 0.4066,
       "step": 2220
     },
     {
-      "epoch": 2.79,
-      "learning_rate": 2.968206080297053e-05,
-      "loss": 0.6033,
+      "epoch": 3.41,
+      "learning_rate": 1.7669045191980972e-05,
+      "loss": 0.3397,
       "step": 2230
     },
     {
-      "epoch": 2.81,
-      "learning_rate": 2.95660245996751e-05,
-      "loss": 0.6418,
+      "epoch": 3.42,
+      "learning_rate": 1.7499150526673462e-05,
+      "loss": 0.3611,
       "step": 2240
     },
     {
-      "epoch": 2.82,
-      "learning_rate": 2.9449988396379674e-05,
-      "loss": 0.6206,
+      "epoch": 3.44,
+      "learning_rate": 1.7329255861365955e-05,
+      "loss": 0.3621,
       "step": 2250
     },
     {
-      "epoch": 2.83,
-      "learning_rate": 2.933395219308424e-05,
-      "loss": 0.6655,
+      "epoch": 3.45,
+      "learning_rate": 1.7159361196058444e-05,
+      "loss": 0.2716,
       "step": 2260
     },
     {
-      "epoch": 2.84,
-      "learning_rate": 2.9217915989788812e-05,
-      "loss": 0.5923,
+      "epoch": 3.47,
+      "learning_rate": 1.6989466530750937e-05,
+      "loss": 0.3324,
       "step": 2270
     },
     {
-      "epoch": 2.86,
-      "learning_rate": 2.9101879786493384e-05,
-      "loss": 0.6219,
+      "epoch": 3.48,
+      "learning_rate": 1.6819571865443427e-05,
+      "loss": 0.3249,
       "step": 2280
     },
     {
-      "epoch": 2.87,
-      "learning_rate": 2.8985843583197956e-05,
-      "loss": 0.6688,
+      "epoch": 3.5,
+      "learning_rate": 1.6649677200135916e-05,
+      "loss": 0.3113,
       "step": 2290
     },
     {
-      "epoch": 2.88,
-      "learning_rate": 2.886980737990253e-05,
-      "loss": 0.6476,
+      "epoch": 3.52,
+      "learning_rate": 1.647978253482841e-05,
+      "loss": 0.2582,
       "step": 2300
     },
     {
-      "epoch": 2.89,
-      "learning_rate": 2.87537711766071e-05,
-      "loss": 0.6278,
+      "epoch": 3.53,
+      "learning_rate": 1.63098878695209e-05,
+      "loss": 0.3834,
       "step": 2310
     },
     {
-      "epoch": 2.91,
-      "learning_rate": 2.8637734973311677e-05,
-      "loss": 0.5953,
+      "epoch": 3.55,
+      "learning_rate": 1.6139993204213388e-05,
+      "loss": 0.29,
       "step": 2320
     },
     {
-      "epoch": 2.92,
-      "learning_rate": 2.852169877001625e-05,
-      "loss": 0.6612,
+      "epoch": 3.56,
+      "learning_rate": 1.5970098538905877e-05,
+      "loss": 0.3144,
       "step": 2330
     },
     {
-      "epoch": 2.93,
-      "learning_rate": 2.840566256672082e-05,
-      "loss": 0.6813,
+      "epoch": 3.58,
+      "learning_rate": 1.580020387359837e-05,
+      "loss": 0.3653,
       "step": 2340
     },
     {
-      "epoch": 2.94,
-      "learning_rate": 2.8289626363425394e-05,
-      "loss": 0.6529,
+      "epoch": 3.59,
+      "learning_rate": 1.563030920829086e-05,
+      "loss": 0.3767,
       "step": 2350
     },
     {
-      "epoch": 2.96,
-      "learning_rate": 2.8173590160129966e-05,
-      "loss": 0.5897,
+      "epoch": 3.61,
+      "learning_rate": 1.546041454298335e-05,
+      "loss": 0.3475,
       "step": 2360
     },
     {
-      "epoch": 2.97,
-      "learning_rate": 2.805755395683453e-05,
-      "loss": 0.647,
+      "epoch": 3.62,
+      "learning_rate": 1.5290519877675842e-05,
+      "loss": 0.4123,
       "step": 2370
     },
     {
-      "epoch": 2.98,
-      "learning_rate": 2.7941517753539104e-05,
-      "loss": 0.6369,
+      "epoch": 3.64,
+      "learning_rate": 1.5120625212368333e-05,
+      "loss": 0.3363,
       "step": 2380
     },
     {
-      "epoch": 2.99,
-      "learning_rate": 2.7825481550243676e-05,
-      "loss": 0.6842,
+      "epoch": 3.65,
+      "learning_rate": 1.4950730547060823e-05,
+      "loss": 0.25,
       "step": 2390
     },
     {
-      "epoch": 3.0,
-      "eval_accuracy": 0.7837171052631579,
-      "eval_loss": 0.6286582946777344,
-      "eval_runtime": 244.0222,
-      "eval_samples_per_second": 104.646,
-      "eval_steps_per_second": 3.27,
-      "step": 2394
-    },
-    {
-      "epoch": 3.01,
-      "learning_rate": 2.770944534694825e-05,
-      "loss": 0.5777,
+      "epoch": 3.67,
+      "learning_rate": 1.4780835881753314e-05,
+      "loss": 0.2966,
       "step": 2400
     },
     {
-      "epoch": 3.02,
-      "learning_rate": 2.759340914365282e-05,
-      "loss": 0.5252,
+      "epoch": 3.68,
+      "learning_rate": 1.4610941216445803e-05,
+      "loss": 0.3758,
       "step": 2410
     },
     {
-      "epoch": 3.03,
-      "learning_rate": 2.7477372940357393e-05,
-      "loss": 0.5019,
+      "epoch": 3.7,
+      "learning_rate": 1.4441046551138296e-05,
+      "loss": 0.3117,
       "step": 2420
     },
     {
-      "epoch": 3.05,
-      "learning_rate": 2.7361336737061965e-05,
-      "loss": 0.5629,
+      "epoch": 3.71,
+      "learning_rate": 1.4271151885830786e-05,
+      "loss": 0.305,
       "step": 2430
     },
     {
-      "epoch": 3.06,
-      "learning_rate": 2.7245300533766538e-05,
-      "loss": 0.4852,
+      "epoch": 3.73,
+      "learning_rate": 1.4101257220523275e-05,
+      "loss": 0.3168,
       "step": 2440
     },
     {
-      "epoch": 3.07,
-      "learning_rate": 2.712926433047111e-05,
-      "loss": 0.5356,
+      "epoch": 3.74,
+      "learning_rate": 1.3931362555215766e-05,
+      "loss": 0.3642,
       "step": 2450
     },
     {
-      "epoch": 3.08,
-      "learning_rate": 2.7013228127175682e-05,
-      "loss": 0.5717,
+      "epoch": 3.76,
+      "learning_rate": 1.3761467889908258e-05,
+      "loss": 0.3745,
       "step": 2460
     },
     {
-      "epoch": 3.1,
-      "learning_rate": 2.6897191923880248e-05,
-      "loss": 0.5351,
+      "epoch": 3.78,
+      "learning_rate": 1.3591573224600749e-05,
+      "loss": 0.3162,
       "step": 2470
     },
     {
-      "epoch": 3.11,
-      "learning_rate": 2.678115572058482e-05,
-      "loss": 0.4641,
+      "epoch": 3.79,
+      "learning_rate": 1.3421678559293238e-05,
+      "loss": 0.3817,
       "step": 2480
     },
     {
-      "epoch": 3.12,
-      "learning_rate": 2.6665119517289396e-05,
-      "loss": 0.5437,
+      "epoch": 3.81,
+      "learning_rate": 1.3251783893985731e-05,
+      "loss": 0.2568,
       "step": 2490
     },
     {
-      "epoch": 3.13,
-      "learning_rate": 2.6549083313993968e-05,
-      "loss": 0.5385,
+      "epoch": 3.82,
+      "learning_rate": 1.308188922867822e-05,
+      "loss": 0.3392,
       "step": 2500
     },
     {
-      "epoch": 3.15,
-      "learning_rate": 2.643304711069854e-05,
-      "loss": 0.5425,
+      "epoch": 3.84,
+      "learning_rate": 1.291199456337071e-05,
+      "loss": 0.386,
       "step": 2510
     },
     {
-      "epoch": 3.16,
-      "learning_rate": 2.6317010907403113e-05,
-      "loss": 0.5478,
+      "epoch": 3.85,
+      "learning_rate": 1.2742099898063201e-05,
+      "loss": 0.392,
       "step": 2520
     },
     {
-      "epoch": 3.17,
-      "learning_rate": 2.6200974704107685e-05,
-      "loss": 0.5799,
+      "epoch": 3.87,
+      "learning_rate": 1.2572205232755694e-05,
+      "loss": 0.3166,
       "step": 2530
     },
     {
-      "epoch": 3.18,
-      "learning_rate": 2.6084938500812257e-05,
-      "loss": 0.5189,
+      "epoch": 3.88,
+      "learning_rate": 1.2402310567448183e-05,
+      "loss": 0.3222,
       "step": 2540
     },
     {
-      "epoch": 3.2,
-      "learning_rate": 2.596890229751683e-05,
-      "loss": 0.5384,
+      "epoch": 3.9,
+      "learning_rate": 1.2232415902140673e-05,
+      "loss": 0.3122,
       "step": 2550
     },
     {
-      "epoch": 3.21,
-      "learning_rate": 2.5852866094221402e-05,
-      "loss": 0.5281,
+      "epoch": 3.91,
+      "learning_rate": 1.2062521236833164e-05,
+      "loss": 0.3275,
       "step": 2560
     },
     {
-      "epoch": 3.22,
-      "learning_rate": 2.5736829890925968e-05,
-      "loss": 0.4708,
+      "epoch": 3.93,
+      "learning_rate": 1.1892626571525654e-05,
+      "loss": 0.3268,
       "step": 2570
     },
     {
-      "epoch": 3.23,
-      "learning_rate": 2.562079368763054e-05,
-      "loss": 0.5123,
+      "epoch": 3.94,
+      "learning_rate": 1.1722731906218146e-05,
+      "loss": 0.2926,
       "step": 2580
     },
     {
-      "epoch": 3.25,
-      "learning_rate": 2.5504757484335112e-05,
-      "loss": 0.4915,
+      "epoch": 3.96,
+      "learning_rate": 1.1552837240910636e-05,
+      "loss": 0.3422,
       "step": 2590
     },
     {
-      "epoch": 3.26,
-      "learning_rate": 2.5388721281039684e-05,
-      "loss": 0.5612,
+      "epoch": 3.97,
+      "learning_rate": 1.1382942575603127e-05,
+      "loss": 0.3107,
       "step": 2600
     },
     {
-      "epoch": 3.27,
-      "learning_rate": 2.5272685077744257e-05,
-      "loss": 0.5616,
+      "epoch": 3.99,
+      "learning_rate": 1.1213047910295617e-05,
+      "loss": 0.2461,
       "step": 2610
     },
     {
-      "epoch": 3.28,
-      "learning_rate": 2.515664887444883e-05,
-      "loss": 0.5495,
+      "epoch": 4.0,
+      "eval_accuracy": 0.7392550143266475,
+      "eval_loss": 0.7843212485313416,
+      "eval_runtime": 48.1379,
+      "eval_samples_per_second": 108.75,
+      "eval_steps_per_second": 13.607,
+      "step": 2617
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.1043153244988108e-05,
+      "loss": 0.2342,
       "step": 2620
     },
     {
-      "epoch": 3.3,
-      "learning_rate": 2.50406126711534e-05,
-      "loss": 0.5311,
+      "epoch": 4.02,
+      "learning_rate": 1.0873258579680597e-05,
+      "loss": 0.2295,
       "step": 2630
     },
     {
-      "epoch": 3.31,
-      "learning_rate": 2.4924576467857974e-05,
-      "loss": 0.4424,
+      "epoch": 4.04,
+      "learning_rate": 1.070336391437309e-05,
+      "loss": 0.1964,
       "step": 2640
     },
     {
-      "epoch": 3.32,
-      "learning_rate": 2.4808540264562543e-05,
-      "loss": 0.5303,
+      "epoch": 4.05,
+      "learning_rate": 1.053346924906558e-05,
+      "loss": 0.2504,
       "step": 2650
     },
     {
-      "epoch": 3.33,
-      "learning_rate": 2.4692504061267115e-05,
-      "loss": 0.5391,
+      "epoch": 4.07,
+      "learning_rate": 1.036357458375807e-05,
+      "loss": 0.2676,
       "step": 2660
     },
     {
-      "epoch": 3.35,
-      "learning_rate": 2.4576467857971687e-05,
-      "loss": 0.4972,
+      "epoch": 4.08,
+      "learning_rate": 1.019367991845056e-05,
+      "loss": 0.2241,
       "step": 2670
     },
     {
-      "epoch": 3.36,
-      "learning_rate": 2.446043165467626e-05,
-      "loss": 0.5716,
+      "epoch": 4.1,
+      "learning_rate": 1.0023785253143051e-05,
+      "loss": 0.189,
       "step": 2680
     },
     {
-      "epoch": 3.37,
-      "learning_rate": 2.4344395451380832e-05,
-      "loss": 0.4912,
+      "epoch": 4.11,
+      "learning_rate": 9.853890587835543e-06,
+      "loss": 0.263,
       "step": 2690
     },
     {
-      "epoch": 3.38,
-      "learning_rate": 2.4228359248085404e-05,
-      "loss": 0.5043,
+      "epoch": 4.13,
+      "learning_rate": 9.683995922528034e-06,
+      "loss": 0.2561,
       "step": 2700
     },
     {
-      "epoch": 3.4,
-      "learning_rate": 2.4112323044789976e-05,
-      "loss": 0.5148,
+      "epoch": 4.14,
+      "learning_rate": 9.514101257220525e-06,
+      "loss": 0.2154,
       "step": 2710
     },
     {
-      "epoch": 3.41,
-      "learning_rate": 2.399628684149455e-05,
-      "loss": 0.5139,
+      "epoch": 4.16,
+      "learning_rate": 9.344206591913014e-06,
+      "loss": 0.2351,
       "step": 2720
     },
     {
-      "epoch": 3.42,
-      "learning_rate": 2.388025063819912e-05,
-      "loss": 0.514,
+      "epoch": 4.17,
+      "learning_rate": 9.174311926605506e-06,
+      "loss": 0.2281,
       "step": 2730
     },
     {
-      "epoch": 3.43,
-      "learning_rate": 2.3764214434903693e-05,
-      "loss": 0.5302,
+      "epoch": 4.19,
+      "learning_rate": 9.004417261297995e-06,
+      "loss": 0.1526,
       "step": 2740
     },
     {
-      "epoch": 3.45,
-      "learning_rate": 2.3648178231608262e-05,
-      "loss": 0.5554,
+      "epoch": 4.2,
+      "learning_rate": 8.834522595990486e-06,
+      "loss": 0.2612,
       "step": 2750
     },
     {
-      "epoch": 3.46,
-      "learning_rate": 2.3532142028312835e-05,
-      "loss": 0.5057,
+      "epoch": 4.22,
+      "learning_rate": 8.664627930682977e-06,
+      "loss": 0.2339,
       "step": 2760
     },
     {
-      "epoch": 3.47,
-      "learning_rate": 2.3416105825017407e-05,
-      "loss": 0.5366,
+      "epoch": 4.23,
+      "learning_rate": 8.494733265375469e-06,
+      "loss": 0.2484,
       "step": 2770
     },
     {
-      "epoch": 3.48,
-      "learning_rate": 2.330006962172198e-05,
-      "loss": 0.5234,
+      "epoch": 4.25,
+      "learning_rate": 8.324838600067958e-06,
+      "loss": 0.2338,
       "step": 2780
     },
     {
-      "epoch": 3.5,
-      "learning_rate": 2.318403341842655e-05,
-      "loss": 0.5847,
+      "epoch": 4.26,
+      "learning_rate": 8.15494393476045e-06,
+      "loss": 0.2062,
       "step": 2790
     },
     {
-      "epoch": 3.51,
-      "learning_rate": 2.306799721513112e-05,
-      "loss": 0.5515,
+      "epoch": 4.28,
+      "learning_rate": 7.985049269452939e-06,
+      "loss": 0.2153,
       "step": 2800
     },
     {
-      "epoch": 3.52,
-      "learning_rate": 2.2951961011835693e-05,
-      "loss": 0.526,
+      "epoch": 4.29,
+      "learning_rate": 7.81515460414543e-06,
+      "loss": 0.1661,
       "step": 2810
     },
     {
-      "epoch": 3.53,
-      "learning_rate": 2.2835924808540265e-05,
-      "loss": 0.5571,
+      "epoch": 4.31,
+      "learning_rate": 7.645259938837921e-06,
+      "loss": 0.2059,
       "step": 2820
     },
     {
-      "epoch": 3.55,
-      "learning_rate": 2.2719888605244837e-05,
-      "loss": 0.5698,
+      "epoch": 4.33,
+      "learning_rate": 7.475365273530411e-06,
+      "loss": 0.2611,
       "step": 2830
     },
     {
-      "epoch": 3.56,
-      "learning_rate": 2.260385240194941e-05,
-      "loss": 0.5039,
+      "epoch": 4.34,
+      "learning_rate": 7.305470608222902e-06,
+      "loss": 0.2026,
       "step": 2840
     },
     {
-      "epoch": 3.57,
-      "learning_rate": 2.248781619865398e-05,
-      "loss": 0.5164,
+      "epoch": 4.36,
+      "learning_rate": 7.135575942915393e-06,
+      "loss": 0.2194,
       "step": 2850
     },
     {
-      "epoch": 3.58,
-      "learning_rate": 2.237177999535855e-05,
-      "loss": 0.5741,
+      "epoch": 4.37,
+      "learning_rate": 6.965681277607883e-06,
+      "loss": 0.1759,
       "step": 2860
     },
     {
-      "epoch": 3.6,
-      "learning_rate": 2.2255743792063123e-05,
-      "loss": 0.5243,
+      "epoch": 4.39,
+      "learning_rate": 6.795786612300374e-06,
+      "loss": 0.2087,
       "step": 2870
     },
     {
-      "epoch": 3.61,
-      "learning_rate": 2.2139707588767696e-05,
-      "loss": 0.5214,
+      "epoch": 4.4,
+      "learning_rate": 6.6258919469928655e-06,
+      "loss": 0.2105,
       "step": 2880
     },
     {
-      "epoch": 3.62,
-      "learning_rate": 2.2023671385472268e-05,
-      "loss": 0.5029,
+      "epoch": 4.42,
+      "learning_rate": 6.455997281685355e-06,
+      "loss": 0.2052,
       "step": 2890
     },
     {
-      "epoch": 3.63,
-      "learning_rate": 2.190763518217684e-05,
-      "loss": 0.5528,
+      "epoch": 4.43,
+      "learning_rate": 6.286102616377847e-06,
+      "loss": 0.2395,
       "step": 2900
     },
     {
-      "epoch": 3.65,
-      "learning_rate": 2.1791598978881413e-05,
-      "loss": 0.5115,
+      "epoch": 4.45,
+      "learning_rate": 6.1162079510703365e-06,
+      "loss": 0.1953,
       "step": 2910
     },
     {
-      "epoch": 3.66,
-      "learning_rate": 2.1675562775585985e-05,
-      "loss": 0.5399,
+      "epoch": 4.46,
+      "learning_rate": 5.946313285762827e-06,
+      "loss": 0.1944,
       "step": 2920
     },
     {
-      "epoch": 3.67,
-      "learning_rate": 2.1559526572290557e-05,
-      "loss": 0.5467,
+      "epoch": 4.48,
+      "learning_rate": 5.776418620455318e-06,
+      "loss": 0.2784,
       "step": 2930
     },
     {
-      "epoch": 3.68,
-      "learning_rate": 2.144349036899513e-05,
-      "loss": 0.5583,
+      "epoch": 4.49,
+      "learning_rate": 5.606523955147808e-06,
+      "loss": 0.1885,
       "step": 2940
     },
     {
-      "epoch": 3.7,
-      "learning_rate": 2.1327454165699702e-05,
-      "loss": 0.5826,
+      "epoch": 4.51,
+      "learning_rate": 5.436629289840299e-06,
+      "loss": 0.2575,
       "step": 2950
     },
     {
-      "epoch": 3.71,
-      "learning_rate": 2.121141796240427e-05,
-      "loss": 0.5042,
+      "epoch": 4.52,
+      "learning_rate": 5.26673462453279e-06,
+      "loss": 0.1941,
       "step": 2960
     },
     {
-      "epoch": 3.72,
-      "learning_rate": 2.1095381759108843e-05,
-      "loss": 0.4947,
+      "epoch": 4.54,
+      "learning_rate": 5.09683995922528e-06,
+      "loss": 0.2795,
       "step": 2970
     },
     {
-      "epoch": 3.73,
-      "learning_rate": 2.0979345555813415e-05,
-      "loss": 0.5185,
+      "epoch": 4.55,
+      "learning_rate": 4.926945293917771e-06,
+      "loss": 0.1971,
       "step": 2980
     },
     {
-      "epoch": 3.75,
-      "learning_rate": 2.0863309352517988e-05,
-      "loss": 0.539,
+      "epoch": 4.57,
+      "learning_rate": 4.7570506286102625e-06,
+      "loss": 0.224,
       "step": 2990
     },
     {
-      "epoch": 3.76,
-      "learning_rate": 2.074727314922256e-05,
-      "loss": 0.5194,
+      "epoch": 4.59,
+      "learning_rate": 4.587155963302753e-06,
+      "loss": 0.2044,
       "step": 3000
     },
     {
-      "epoch": 3.77,
-      "learning_rate": 2.063123694592713e-05,
-      "loss": 0.5049,
+      "epoch": 4.6,
+      "learning_rate": 4.417261297995243e-06,
+      "loss": 0.2073,
       "step": 3010
     },
     {
-      "epoch": 3.78,
-      "learning_rate": 2.05152007426317e-05,
-      "loss": 0.4813,
+      "epoch": 4.62,
+      "learning_rate": 4.247366632687734e-06,
+      "loss": 0.2055,
       "step": 3020
     },
     {
-      "epoch": 3.8,
-      "learning_rate": 2.0399164539336273e-05,
-      "loss": 0.4671,
+      "epoch": 4.63,
+      "learning_rate": 4.077471967380225e-06,
+      "loss": 0.1936,
       "step": 3030
     },
     {
-      "epoch": 3.81,
-      "learning_rate": 2.0283128336040846e-05,
-      "loss": 0.5214,
+      "epoch": 4.65,
+      "learning_rate": 3.907577302072715e-06,
+      "loss": 0.219,
       "step": 3040
     },
     {
-      "epoch": 3.82,
-      "learning_rate": 2.0167092132745418e-05,
-      "loss": 0.5476,
+      "epoch": 4.66,
+      "learning_rate": 3.7376826367652057e-06,
+      "loss": 0.198,
       "step": 3050
     },
     {
-      "epoch": 3.83,
-      "learning_rate": 2.0051055929449987e-05,
-      "loss": 0.5231,
+      "epoch": 4.68,
+      "learning_rate": 3.5677879714576964e-06,
+      "loss": 0.242,
       "step": 3060
     },
     {
-      "epoch": 3.85,
-      "learning_rate": 1.993501972615456e-05,
-      "loss": 0.5671,
+      "epoch": 4.69,
+      "learning_rate": 3.397893306150187e-06,
+      "loss": 0.2242,
       "step": 3070
     },
     {
-      "epoch": 3.86,
-      "learning_rate": 1.981898352285913e-05,
-      "loss": 0.5647,
+      "epoch": 4.71,
+      "learning_rate": 3.2279986408426775e-06,
+      "loss": 0.2083,
       "step": 3080
     },
     {
-      "epoch": 3.87,
-      "learning_rate": 1.9702947319563704e-05,
-      "loss": 0.5219,
+      "epoch": 4.72,
+      "learning_rate": 3.0581039755351682e-06,
+      "loss": 0.2268,
       "step": 3090
     },
     {
-      "epoch": 3.88,
-      "learning_rate": 1.9586911116268276e-05,
-      "loss": 0.5142,
+      "epoch": 4.74,
+      "learning_rate": 2.888209310227659e-06,
+      "loss": 0.1819,
       "step": 3100
     },
     {
-      "epoch": 3.9,
-      "learning_rate": 1.947087491297285e-05,
-      "loss": 0.5069,
+      "epoch": 4.75,
+      "learning_rate": 2.7183146449201493e-06,
+      "loss": 0.2127,
       "step": 3110
     },
     {
-      "epoch": 3.91,
-      "learning_rate": 1.935483870967742e-05,
-      "loss": 0.5581,
+      "epoch": 4.77,
+      "learning_rate": 2.54841997961264e-06,
+      "loss": 0.1816,
       "step": 3120
     },
     {
-      "epoch": 3.92,
-      "learning_rate": 1.9238802506381993e-05,
-      "loss": 0.4728,
+      "epoch": 4.78,
+      "learning_rate": 2.3785253143051312e-06,
+      "loss": 0.187,
       "step": 3130
     },
     {
-      "epoch": 3.93,
-      "learning_rate": 1.9122766303086565e-05,
-      "loss": 0.5513,
+      "epoch": 4.8,
+      "learning_rate": 2.2086306489976216e-06,
+      "loss": 0.2482,
       "step": 3140
     },
     {
-      "epoch": 3.95,
-      "learning_rate": 1.9006730099791138e-05,
-      "loss": 0.5409,
+      "epoch": 4.81,
+      "learning_rate": 2.0387359836901123e-06,
+      "loss": 0.2161,
       "step": 3150
     },
     {
-      "epoch": 3.96,
-      "learning_rate": 1.8890693896495707e-05,
-      "loss": 0.5164,
+      "epoch": 4.83,
+      "learning_rate": 1.8688413183826028e-06,
+      "loss": 0.2413,
       "step": 3160
     },
     {
-      "epoch": 3.97,
-      "learning_rate": 1.877465769320028e-05,
-      "loss": 0.5415,
+      "epoch": 4.85,
+      "learning_rate": 1.6989466530750936e-06,
+      "loss": 0.1959,
       "step": 3170
     },
     {
-      "epoch": 3.98,
-      "learning_rate": 1.865862148990485e-05,
-      "loss": 0.52,
+      "epoch": 4.86,
+      "learning_rate": 1.5290519877675841e-06,
+      "loss": 0.2644,
       "step": 3180
     },
     {
-      "epoch": 4.0,
-      "learning_rate": 1.8542585286609424e-05,
-      "loss": 0.5242,
+      "epoch": 4.88,
+      "learning_rate": 1.3591573224600747e-06,
+      "loss": 0.1793,
       "step": 3190
     },
     {
-      "epoch": 4.0,
-      "eval_accuracy": 0.7839129072681704,
-      "eval_loss": 0.62819904088974,
-      "eval_runtime": 245.126,
-      "eval_samples_per_second": 104.175,
-      "eval_steps_per_second": 3.255,
-      "step": 3192
-    },
-    {
-      "epoch": 4.01,
-      "learning_rate": 1.8426549083313996e-05,
-      "loss": 0.45,
+      "epoch": 4.89,
+      "learning_rate": 1.1892626571525656e-06,
+      "loss": 0.1869,
       "step": 3200
     },
     {
-      "epoch": 4.02,
-      "learning_rate": 1.8310512880018565e-05,
-      "loss": 0.4147,
+      "epoch": 4.91,
+      "learning_rate": 1.0193679918450562e-06,
+      "loss": 0.195,
       "step": 3210
     },
     {
-      "epoch": 4.04,
-      "learning_rate": 1.8194476676723137e-05,
-      "loss": 0.415,
+      "epoch": 4.92,
+      "learning_rate": 8.494733265375468e-07,
+      "loss": 0.2474,
       "step": 3220
     },
     {
-      "epoch": 4.05,
-      "learning_rate": 1.807844047342771e-05,
-      "loss": 0.4206,
+      "epoch": 4.94,
+      "learning_rate": 6.795786612300373e-07,
+      "loss": 0.2748,
       "step": 3230
     },
     {
-      "epoch": 4.06,
-      "learning_rate": 1.7962404270132282e-05,
-      "loss": 0.4103,
+      "epoch": 4.95,
+      "learning_rate": 5.096839959225281e-07,
+      "loss": 0.1932,
       "step": 3240
     },
     {
-      "epoch": 4.07,
-      "learning_rate": 1.7846368066836854e-05,
-      "loss": 0.4269,
+      "epoch": 4.97,
+      "learning_rate": 3.3978933061501866e-07,
+      "loss": 0.2393,
       "step": 3250
     },
     {
-      "epoch": 4.09,
-      "learning_rate": 1.7730331863541423e-05,
-      "loss": 0.4498,
+      "epoch": 4.98,
+      "learning_rate": 1.6989466530750933e-07,
+      "loss": 0.2547,
       "step": 3260
     },
     {
-      "epoch": 4.1,
-      "learning_rate": 1.7614295660245995e-05,
-      "loss": 0.4054,
+      "epoch": 5.0,
+      "learning_rate": 0.0,
+      "loss": 0.1846,
       "step": 3270
     },
-    {
-      "epoch": 4.11,
-      "learning_rate": 1.7498259456950568e-05,
-      "loss": 0.4487,
-      "step": 3280
-    },
-    {
-      "epoch": 4.12,
-      "learning_rate": 1.738222325365514e-05,
-      "loss": 0.4052,
-      "step": 3290
-    },
-    {
-      "epoch": 4.14,
-      "learning_rate": 1.7266187050359716e-05,
-      "loss": 0.4106,
-      "step": 3300
-    },
-    {
-      "epoch": 4.15,
-      "learning_rate": 1.7150150847064285e-05,
-      "loss": 0.4004,
-      "step": 3310
-    },
-    {
-      "epoch": 4.16,
-      "learning_rate": 1.7034114643768857e-05,
-      "loss": 0.433,
-      "step": 3320
-    },
-    {
-      "epoch": 4.17,
-      "learning_rate": 1.691807844047343e-05,
-      "loss": 0.3979,
-      "step": 3330
-    },
-    {
-      "epoch": 4.19,
-      "learning_rate": 1.6802042237178e-05,
-      "loss": 0.4445,
-      "step": 3340
-    },
-    {
-      "epoch": 4.2,
-      "learning_rate": 1.6686006033882574e-05,
-      "loss": 0.4357,
-      "step": 3350
-    },
-    {
-      "epoch": 4.21,
-      "learning_rate": 1.6569969830587146e-05,
-      "loss": 0.4279,
-      "step": 3360
-    },
-    {
-      "epoch": 4.22,
-      "learning_rate": 1.6453933627291715e-05,
-      "loss": 0.4089,
-      "step": 3370
-    },
-    {
-      "epoch": 4.24,
-      "learning_rate": 1.6337897423996287e-05,
-      "loss": 0.4039,
-      "step": 3380
-    },
-    {
-      "epoch": 4.25,
-      "learning_rate": 1.622186122070086e-05,
-      "loss": 0.4216,
-      "step": 3390
-    },
-    {
-      "epoch": 4.26,
-      "learning_rate": 1.6105825017405432e-05,
-      "loss": 0.4535,
-      "step": 3400
-    },
-    {
-      "epoch": 4.27,
-      "learning_rate": 1.5989788814110004e-05,
-      "loss": 0.4646,
-      "step": 3410
-    },
-    {
-      "epoch": 4.29,
-      "learning_rate": 1.5873752610814573e-05,
-      "loss": 0.4324,
-      "step": 3420
-    },
-    {
-      "epoch": 4.3,
-      "learning_rate": 1.5757716407519145e-05,
-      "loss": 0.4402,
-      "step": 3430
-    },
-    {
-      "epoch": 4.31,
-      "learning_rate": 1.5641680204223718e-05,
-      "loss": 0.4123,
-      "step": 3440
-    },
-    {
-      "epoch": 4.32,
-      "learning_rate": 1.552564400092829e-05,
-      "loss": 0.4742,
-      "step": 3450
-    },
-    {
-      "epoch": 4.34,
-      "learning_rate": 1.5409607797632862e-05,
-      "loss": 0.4177,
-      "step": 3460
-    },
-    {
-      "epoch": 4.35,
-      "learning_rate": 1.5293571594337435e-05,
-      "loss": 0.4215,
-      "step": 3470
-    },
-    {
-      "epoch": 4.36,
-      "learning_rate": 1.5177535391042005e-05,
-      "loss": 0.3837,
-      "step": 3480
-    },
-    {
-      "epoch": 4.37,
-      "learning_rate": 1.5061499187746578e-05,
-      "loss": 0.4275,
-      "step": 3490
-    },
-    {
-      "epoch": 4.39,
-      "learning_rate": 1.494546298445115e-05,
-      "loss": 0.4526,
-      "step": 3500
-    },
-    {
-      "epoch": 4.4,
-      "learning_rate": 1.4829426781155722e-05,
-      "loss": 0.3635,
-      "step": 3510
-    },
-    {
-      "epoch": 4.41,
-      "learning_rate": 1.4713390577860293e-05,
-      "loss": 0.4344,
-      "step": 3520
-    },
-    {
-      "epoch": 4.42,
-      "learning_rate": 1.4597354374564865e-05,
-      "loss": 0.4282,
-      "step": 3530
-    },
-    {
-      "epoch": 4.44,
-      "learning_rate": 1.4481318171269437e-05,
-      "loss": 0.4838,
-      "step": 3540
-    },
-    {
-      "epoch": 4.45,
-      "learning_rate": 1.436528196797401e-05,
-      "loss": 0.3926,
-      "step": 3550
-    },
-    {
-      "epoch": 4.46,
-      "learning_rate": 1.4249245764678582e-05,
-      "loss": 0.4181,
-      "step": 3560
-    },
-    {
-      "epoch": 4.47,
-      "learning_rate": 1.4133209561383151e-05,
-      "loss": 0.4614,
-      "step": 3570
-    },
-    {
-      "epoch": 4.49,
-      "learning_rate": 1.4017173358087723e-05,
-      "loss": 0.4338,
-      "step": 3580
-    },
-    {
-      "epoch": 4.5,
-      "learning_rate": 1.3901137154792296e-05,
-      "loss": 0.4306,
-      "step": 3590
-    },
-    {
-      "epoch": 4.51,
-      "learning_rate": 1.3785100951496868e-05,
-      "loss": 0.4294,
-      "step": 3600
-    },
-    {
-      "epoch": 4.52,
-      "learning_rate": 1.366906474820144e-05,
-      "loss": 0.4147,
-      "step": 3610
-    },
-    {
-      "epoch": 4.54,
-      "learning_rate": 1.3553028544906011e-05,
-      "loss": 0.4619,
-      "step": 3620
-    },
-    {
-      "epoch": 4.55,
-      "learning_rate": 1.3436992341610583e-05,
-      "loss": 0.4146,
-      "step": 3630
-    },
-    {
-      "epoch": 4.56,
-      "learning_rate": 1.3320956138315155e-05,
-      "loss": 0.4718,
-      "step": 3640
-    },
-    {
-      "epoch": 4.57,
-      "learning_rate": 1.3204919935019728e-05,
-      "loss": 0.3824,
-      "step": 3650
-    },
-    {
-      "epoch": 4.59,
-      "learning_rate": 1.30888837317243e-05,
-      "loss": 0.4173,
-      "step": 3660
-    },
-    {
-      "epoch": 4.6,
-      "learning_rate": 1.2972847528428869e-05,
-      "loss": 0.4221,
-      "step": 3670
-    },
-    {
-      "epoch": 4.61,
-      "learning_rate": 1.2856811325133441e-05,
-      "loss": 0.4027,
-      "step": 3680
-    },
-    {
-      "epoch": 4.62,
-      "learning_rate": 1.2740775121838014e-05,
-      "loss": 0.4,
-      "step": 3690
-    },
-    {
-      "epoch": 4.64,
-      "learning_rate": 1.2624738918542586e-05,
-      "loss": 0.3792,
-      "step": 3700
-    },
-    {
-      "epoch": 4.65,
-      "learning_rate": 1.2508702715247158e-05,
-      "loss": 0.4146,
-      "step": 3710
-    },
-    {
-      "epoch": 4.66,
-      "learning_rate": 1.2392666511951729e-05,
-      "loss": 0.4145,
-      "step": 3720
-    },
-    {
-      "epoch": 4.67,
-      "learning_rate": 1.2276630308656301e-05,
-      "loss": 0.4651,
-      "step": 3730
-    },
-    {
-      "epoch": 4.69,
-      "learning_rate": 1.2160594105360874e-05,
-      "loss": 0.4459,
-      "step": 3740
-    },
-    {
-      "epoch": 4.7,
-      "learning_rate": 1.2044557902065446e-05,
-      "loss": 0.4244,
-      "step": 3750
-    },
-    {
-      "epoch": 4.71,
-      "learning_rate": 1.1928521698770016e-05,
-      "loss": 0.476,
-      "step": 3760
-    },
-    {
-      "epoch": 4.72,
-      "learning_rate": 1.1812485495474589e-05,
-      "loss": 0.4,
-      "step": 3770
-    },
-    {
-      "epoch": 4.74,
-      "learning_rate": 1.1696449292179161e-05,
-      "loss": 0.4452,
-      "step": 3780
-    },
-    {
-      "epoch": 4.75,
-      "learning_rate": 1.1580413088883732e-05,
-      "loss": 0.3968,
-      "step": 3790
-    },
-    {
-      "epoch": 4.76,
-      "learning_rate": 1.1464376885588304e-05,
-      "loss": 0.4265,
-      "step": 3800
-    },
-    {
-      "epoch": 4.77,
-      "learning_rate": 1.1348340682292875e-05,
-      "loss": 0.414,
-      "step": 3810
-    },
-    {
-      "epoch": 4.79,
-      "learning_rate": 1.1232304478997447e-05,
-      "loss": 0.4283,
-      "step": 3820
-    },
-    {
-      "epoch": 4.8,
-      "learning_rate": 1.1116268275702021e-05,
-      "loss": 0.4335,
-      "step": 3830
-    },
-    {
-      "epoch": 4.81,
-      "learning_rate": 1.1000232072406592e-05,
-      "loss": 0.4735,
-      "step": 3840
-    },
-    {
-      "epoch": 4.82,
-      "learning_rate": 1.0884195869111164e-05,
-      "loss": 0.4222,
-      "step": 3850
-    },
-    {
-      "epoch": 4.84,
-      "learning_rate": 1.0768159665815734e-05,
-      "loss": 0.4328,
-      "step": 3860
-    },
-    {
-      "epoch": 4.85,
-      "learning_rate": 1.0652123462520307e-05,
-      "loss": 0.4262,
-      "step": 3870
-    },
-    {
-      "epoch": 4.86,
-      "learning_rate": 1.0536087259224879e-05,
-      "loss": 0.4021,
-      "step": 3880
-    },
-    {
-      "epoch": 4.87,
-      "learning_rate": 1.042005105592945e-05,
-      "loss": 0.4266,
-      "step": 3890
-    },
-    {
-      "epoch": 4.89,
-      "learning_rate": 1.0304014852634022e-05,
-      "loss": 0.4273,
-      "step": 3900
-    },
-    {
-      "epoch": 4.9,
-      "learning_rate": 1.0187978649338594e-05,
-      "loss": 0.4609,
-      "step": 3910
-    },
-    {
-      "epoch": 4.91,
-      "learning_rate": 1.0071942446043167e-05,
-      "loss": 0.4008,
-      "step": 3920
-    },
-    {
-      "epoch": 4.92,
-      "learning_rate": 9.955906242747739e-06,
-      "loss": 0.3794,
-      "step": 3930
-    },
-    {
-      "epoch": 4.94,
-      "learning_rate": 9.83987003945231e-06,
-      "loss": 0.3763,
-      "step": 3940
-    },
-    {
-      "epoch": 4.95,
-      "learning_rate": 9.723833836156882e-06,
-      "loss": 0.4246,
-      "step": 3950
-    },
-    {
-      "epoch": 4.96,
-      "learning_rate": 9.607797632861454e-06,
-      "loss": 0.4331,
-      "step": 3960
-    },
-    {
-      "epoch": 4.97,
-      "learning_rate": 9.491761429566025e-06,
-      "loss": 0.4279,
-      "step": 3970
-    },
-    {
-      "epoch": 4.99,
-      "learning_rate": 9.375725226270597e-06,
-      "loss": 0.4245,
-      "step": 3980
-    },
     {
       "epoch": 5.0,
-      "learning_rate": 9.259689022975168e-06,
-      "loss": 0.4321,
-      "step": 3990
+      "eval_accuracy": 0.7390639923591213,
+      "eval_loss": 0.8219425082206726,
+      "eval_runtime": 48.049,
+      "eval_samples_per_second": 108.951,
+      "eval_steps_per_second": 13.632,
+      "step": 3270
     },
     {
       "epoch": 5.0,
-      "eval_accuracy": 0.7823073308270677,
-      "eval_loss": 0.6423132419586182,
-      "eval_runtime": 243.4032,
-      "eval_samples_per_second": 104.912,
-      "eval_steps_per_second": 3.279,
-      "step": 3990
-    },
-    {
-      "epoch": 5.01,
-      "learning_rate": 9.14365281967974e-06,
-      "loss": 0.3513,
-      "step": 4000
-    },
-    {
-      "epoch": 5.03,
-      "learning_rate": 9.027616616384312e-06,
-      "loss": 0.337,
-      "step": 4010
-    },
-    {
-      "epoch": 5.04,
-      "learning_rate": 8.911580413088885e-06,
-      "loss": 0.3227,
-      "step": 4020
-    },
-    {
-      "epoch": 5.05,
-      "learning_rate": 8.795544209793457e-06,
-      "loss": 0.3443,
-      "step": 4030
-    },
-    {
-      "epoch": 5.06,
-      "learning_rate": 8.679508006498028e-06,
-      "loss": 0.3443,
-      "step": 4040
-    },
-    {
-      "epoch": 5.08,
-      "learning_rate": 8.5634718032026e-06,
-      "loss": 0.3033,
-      "step": 4050
-    },
-    {
-      "epoch": 5.09,
-      "learning_rate": 8.447435599907172e-06,
-      "loss": 0.3805,
-      "step": 4060
-    },
-    {
-      "epoch": 5.1,
-      "learning_rate": 8.331399396611743e-06,
-      "loss": 0.2883,
-      "step": 4070
-    },
-    {
-      "epoch": 5.11,
-      "learning_rate": 8.215363193316315e-06,
-      "loss": 0.3702,
-      "step": 4080
-    },
-    {
-      "epoch": 5.13,
-      "learning_rate": 8.099326990020886e-06,
-      "loss": 0.3468,
-      "step": 4090
-    },
-    {
-      "epoch": 5.14,
-      "learning_rate": 7.983290786725458e-06,
-      "loss": 0.3937,
-      "step": 4100
-    },
-    {
-      "epoch": 5.15,
-      "learning_rate": 7.86725458343003e-06,
-      "loss": 0.4071,
-      "step": 4110
-    },
-    {
-      "epoch": 5.16,
-      "learning_rate": 7.751218380134603e-06,
-      "loss": 0.3627,
-      "step": 4120
-    },
-    {
-      "epoch": 5.18,
-      "learning_rate": 7.635182176839175e-06,
-      "loss": 0.397,
-      "step": 4130
-    },
-    {
-      "epoch": 5.19,
-      "learning_rate": 7.519145973543746e-06,
-      "loss": 0.338,
-      "step": 4140
-    },
-    {
-      "epoch": 5.2,
-      "learning_rate": 7.403109770248318e-06,
-      "loss": 0.3559,
-      "step": 4150
-    },
-    {
-      "epoch": 5.21,
-      "learning_rate": 7.28707356695289e-06,
-      "loss": 0.3805,
-      "step": 4160
-    },
-    {
-      "epoch": 5.23,
-      "learning_rate": 7.171037363657461e-06,
-      "loss": 0.3679,
-      "step": 4170
-    },
-    {
-      "epoch": 5.24,
-      "learning_rate": 7.055001160362033e-06,
-      "loss": 0.3594,
-      "step": 4180
-    },
-    {
-      "epoch": 5.25,
-      "learning_rate": 6.938964957066605e-06,
-      "loss": 0.3311,
-      "step": 4190
-    },
-    {
-      "epoch": 5.26,
-      "learning_rate": 6.822928753771177e-06,
-      "loss": 0.3348,
-      "step": 4200
-    },
-    {
-      "epoch": 5.28,
-      "learning_rate": 6.706892550475749e-06,
-      "loss": 0.3552,
-      "step": 4210
-    },
-    {
-      "epoch": 5.29,
-      "learning_rate": 6.59085634718032e-06,
-      "loss": 0.34,
-      "step": 4220
-    },
-    {
-      "epoch": 5.3,
-      "learning_rate": 6.474820143884892e-06,
-      "loss": 0.3524,
-      "step": 4230
-    },
-    {
-      "epoch": 5.31,
-      "learning_rate": 6.358783940589464e-06,
-      "loss": 0.3401,
-      "step": 4240
-    },
-    {
-      "epoch": 5.33,
-      "learning_rate": 6.242747737294036e-06,
-      "loss": 0.3467,
-      "step": 4250
-    },
-    {
-      "epoch": 5.34,
-      "learning_rate": 6.126711533998608e-06,
-      "loss": 0.3115,
-      "step": 4260
-    },
-    {
-      "epoch": 5.35,
-      "learning_rate": 6.01067533070318e-06,
-      "loss": 0.344,
-      "step": 4270
-    },
-    {
-      "epoch": 5.36,
-      "learning_rate": 5.894639127407751e-06,
-      "loss": 0.3388,
-      "step": 4280
-    },
-    {
-      "epoch": 5.38,
-      "learning_rate": 5.778602924112323e-06,
-      "loss": 0.3483,
-      "step": 4290
-    },
-    {
-      "epoch": 5.39,
-      "learning_rate": 5.662566720816896e-06,
-      "loss": 0.355,
-      "step": 4300
-    },
-    {
-      "epoch": 5.4,
-      "learning_rate": 5.546530517521467e-06,
-      "loss": 0.3636,
-      "step": 4310
-    },
-    {
-      "epoch": 5.41,
-      "learning_rate": 5.430494314226039e-06,
-      "loss": 0.3675,
-      "step": 4320
-    },
-    {
-      "epoch": 5.43,
-      "learning_rate": 5.31445811093061e-06,
-      "loss": 0.3346,
-      "step": 4330
-    },
-    {
-      "epoch": 5.44,
-      "learning_rate": 5.198421907635182e-06,
-      "loss": 0.3191,
-      "step": 4340
-    },
-    {
-      "epoch": 5.45,
-      "learning_rate": 5.082385704339755e-06,
-      "loss": 0.3095,
-      "step": 4350
-    },
-    {
-      "epoch": 5.46,
-      "learning_rate": 4.966349501044326e-06,
-      "loss": 0.357,
-      "step": 4360
-    },
-    {
-      "epoch": 5.48,
-      "learning_rate": 4.850313297748898e-06,
-      "loss": 0.3123,
-      "step": 4370
-    },
-    {
-      "epoch": 5.49,
-      "learning_rate": 4.734277094453469e-06,
-      "loss": 0.347,
-      "step": 4380
-    },
-    {
-      "epoch": 5.5,
-      "learning_rate": 4.618240891158042e-06,
-      "loss": 0.3338,
-      "step": 4390
-    },
-    {
-      "epoch": 5.51,
-      "learning_rate": 4.502204687862614e-06,
-      "loss": 0.3698,
-      "step": 4400
-    },
-    {
-      "epoch": 5.53,
-      "learning_rate": 4.386168484567185e-06,
-      "loss": 0.3654,
-      "step": 4410
-    },
-    {
-      "epoch": 5.54,
-      "learning_rate": 4.270132281271757e-06,
-      "loss": 0.3458,
-      "step": 4420
-    },
-    {
-      "epoch": 5.55,
-      "learning_rate": 4.154096077976329e-06,
-      "loss": 0.3096,
-      "step": 4430
-    },
-    {
-      "epoch": 5.56,
-      "learning_rate": 4.038059874680901e-06,
-      "loss": 0.3391,
-      "step": 4440
-    },
-    {
-      "epoch": 5.58,
-      "learning_rate": 3.922023671385473e-06,
-      "loss": 0.3683,
-      "step": 4450
-    },
-    {
-      "epoch": 5.59,
-      "learning_rate": 3.805987468090044e-06,
-      "loss": 0.3855,
-      "step": 4460
-    },
-    {
-      "epoch": 5.6,
-      "learning_rate": 3.689951264794616e-06,
-      "loss": 0.353,
-      "step": 4470
-    },
-    {
-      "epoch": 5.61,
-      "learning_rate": 3.5739150614991875e-06,
-      "loss": 0.3303,
-      "step": 4480
-    },
-    {
-      "epoch": 5.63,
-      "learning_rate": 3.45787885820376e-06,
-      "loss": 0.2987,
-      "step": 4490
-    },
-    {
-      "epoch": 5.64,
-      "learning_rate": 3.3418426549083317e-06,
-      "loss": 0.3412,
-      "step": 4500
-    },
-    {
-      "epoch": 5.65,
-      "learning_rate": 3.225806451612903e-06,
-      "loss": 0.3066,
-      "step": 4510
-    },
-    {
-      "epoch": 5.66,
-      "learning_rate": 3.1097702483174755e-06,
-      "loss": 0.3252,
-      "step": 4520
-    },
-    {
-      "epoch": 5.68,
-      "learning_rate": 2.993734045022047e-06,
-      "loss": 0.3243,
-      "step": 4530
-    },
-    {
-      "epoch": 5.69,
-      "learning_rate": 2.877697841726619e-06,
-      "loss": 0.3888,
-      "step": 4540
-    },
-    {
-      "epoch": 5.7,
-      "learning_rate": 2.7616616384311907e-06,
-      "loss": 0.3635,
-      "step": 4550
-    },
-    {
-      "epoch": 5.71,
-      "learning_rate": 2.645625435135762e-06,
-      "loss": 0.2875,
-      "step": 4560
-    },
-    {
-      "epoch": 5.73,
-      "learning_rate": 2.5295892318403345e-06,
-      "loss": 0.3443,
-      "step": 4570
-    },
-    {
-      "epoch": 5.74,
-      "learning_rate": 2.413553028544906e-06,
-      "loss": 0.3334,
-      "step": 4580
-    },
-    {
-      "epoch": 5.75,
-      "learning_rate": 2.297516825249478e-06,
-      "loss": 0.337,
-      "step": 4590
-    },
-    {
-      "epoch": 5.76,
-      "learning_rate": 2.1814806219540497e-06,
-      "loss": 0.3204,
-      "step": 4600
-    },
-    {
-      "epoch": 5.78,
-      "learning_rate": 2.0654444186586216e-06,
-      "loss": 0.318,
-      "step": 4610
-    },
-    {
-      "epoch": 5.79,
-      "learning_rate": 1.9494082153631935e-06,
-      "loss": 0.3113,
-      "step": 4620
-    },
-    {
-      "epoch": 5.8,
-      "learning_rate": 1.833372012067765e-06,
-      "loss": 0.3623,
-      "step": 4630
-    },
-    {
-      "epoch": 5.81,
-      "learning_rate": 1.717335808772337e-06,
-      "loss": 0.3489,
-      "step": 4640
-    },
-    {
-      "epoch": 5.83,
-      "learning_rate": 1.6012996054769087e-06,
-      "loss": 0.3477,
-      "step": 4650
-    },
-    {
-      "epoch": 5.84,
-      "learning_rate": 1.4852634021814808e-06,
-      "loss": 0.3124,
-      "step": 4660
-    },
-    {
-      "epoch": 5.85,
-      "learning_rate": 1.3692271988860525e-06,
-      "loss": 0.3658,
-      "step": 4670
-    },
-    {
-      "epoch": 5.86,
-      "learning_rate": 1.2531909955906242e-06,
-      "loss": 0.3652,
-      "step": 4680
-    },
-    {
-      "epoch": 5.88,
-      "learning_rate": 1.137154792295196e-06,
-      "loss": 0.3415,
-      "step": 4690
-    },
-    {
-      "epoch": 5.89,
-      "learning_rate": 1.021118588999768e-06,
-      "loss": 0.3479,
-      "step": 4700
-    },
-    {
-      "epoch": 5.9,
-      "learning_rate": 9.050823857043398e-07,
-      "loss": 0.3746,
-      "step": 4710
-    },
-    {
-      "epoch": 5.91,
-      "learning_rate": 7.890461824089116e-07,
-      "loss": 0.3311,
-      "step": 4720
-    },
-    {
-      "epoch": 5.93,
-      "learning_rate": 6.730099791134834e-07,
-      "loss": 0.3282,
-      "step": 4730
-    },
-    {
-      "epoch": 5.94,
-      "learning_rate": 5.569737758180553e-07,
-      "loss": 0.3269,
-      "step": 4740
-    },
-    {
-      "epoch": 5.95,
-      "learning_rate": 4.4093757252262707e-07,
-      "loss": 0.3035,
-      "step": 4750
-    },
-    {
-      "epoch": 5.96,
-      "learning_rate": 3.249013692271989e-07,
-      "loss": 0.3208,
-      "step": 4760
-    },
-    {
-      "epoch": 5.98,
-      "learning_rate": 2.0886516593177073e-07,
-      "loss": 0.3553,
-      "step": 4770
-    },
-    {
-      "epoch": 5.99,
-      "learning_rate": 9.282896263634255e-08,
-      "loss": 0.3129,
-      "step": 4780
-    },
-    {
-      "epoch": 6.0,
-      "eval_accuracy": 0.7837562656641605,
-      "eval_loss": 0.6533188819885254,
-      "eval_runtime": 249.7809,
-      "eval_samples_per_second": 102.234,
-      "eval_steps_per_second": 3.195,
-      "step": 4788
-    },
-    {
-      "epoch": 6.0,
-      "learning_rate": 2.2208298524088e-05,
-      "loss": 0.4739,
-      "step": 4790
-    },
-    {
-      "epoch": 6.02,
-      "learning_rate": 2.2138680033416877e-05,
-      "loss": 0.4398,
-      "step": 4800
-    },
-    {
-      "epoch": 6.03,
-      "learning_rate": 2.2069061542745755e-05,
-      "loss": 0.4024,
-      "step": 4810
-    },
-    {
-      "epoch": 6.04,
-      "learning_rate": 2.199944305207463e-05,
-      "loss": 0.4506,
-      "step": 4820
-    },
-    {
-      "epoch": 6.05,
-      "learning_rate": 2.1929824561403507e-05,
-      "loss": 0.5156,
-      "step": 4830
-    },
-    {
-      "epoch": 6.07,
-      "learning_rate": 2.1860206070732385e-05,
-      "loss": 0.5007,
-      "step": 4840
-    },
-    {
-      "epoch": 6.08,
-      "learning_rate": 2.1790587580061263e-05,
-      "loss": 0.4959,
-      "step": 4850
-    },
-    {
-      "epoch": 6.09,
-      "learning_rate": 2.172096908939014e-05,
-      "loss": 0.5087,
-      "step": 4860
-    },
-    {
-      "epoch": 6.1,
-      "learning_rate": 2.165135059871902e-05,
-      "loss": 0.4577,
-      "step": 4870
-    },
-    {
-      "epoch": 6.12,
-      "learning_rate": 2.1581732108047897e-05,
-      "loss": 0.4851,
-      "step": 4880
-    },
-    {
-      "epoch": 6.13,
-      "learning_rate": 2.1512113617376775e-05,
-      "loss": 0.4541,
-      "step": 4890
-    },
-    {
-      "epoch": 6.14,
-      "learning_rate": 2.1442495126705653e-05,
-      "loss": 0.5341,
-      "step": 4900
-    },
-    {
-      "epoch": 6.15,
-      "learning_rate": 2.137287663603453e-05,
-      "loss": 0.4609,
-      "step": 4910
-    },
-    {
-      "epoch": 6.17,
-      "learning_rate": 2.130325814536341e-05,
-      "loss": 0.5178,
-      "step": 4920
-    },
-    {
-      "epoch": 6.18,
-      "learning_rate": 2.1233639654692286e-05,
-      "loss": 0.4792,
-      "step": 4930
-    },
-    {
-      "epoch": 6.19,
-      "learning_rate": 2.1164021164021164e-05,
-      "loss": 0.448,
-      "step": 4940
-    },
-    {
-      "epoch": 6.2,
-      "learning_rate": 2.1094402673350042e-05,
-      "loss": 0.4971,
-      "step": 4950
-    },
-    {
-      "epoch": 6.22,
-      "learning_rate": 2.102478418267892e-05,
-      "loss": 0.4521,
-      "step": 4960
-    },
-    {
-      "epoch": 6.23,
-      "learning_rate": 2.0955165692007798e-05,
-      "loss": 0.4584,
-      "step": 4970
-    },
-    {
-      "epoch": 6.24,
-      "learning_rate": 2.0885547201336676e-05,
-      "loss": 0.4984,
-      "step": 4980
-    },
-    {
-      "epoch": 6.25,
-      "learning_rate": 2.0815928710665554e-05,
-      "loss": 0.4894,
-      "step": 4990
-    },
-    {
-      "epoch": 6.27,
-      "learning_rate": 2.0746310219994432e-05,
-      "loss": 0.4887,
-      "step": 5000
-    },
-    {
-      "epoch": 6.28,
-      "learning_rate": 2.067669172932331e-05,
-      "loss": 0.5063,
-      "step": 5010
-    },
-    {
-      "epoch": 6.29,
-      "learning_rate": 2.0607073238652184e-05,
-      "loss": 0.4963,
-      "step": 5020
-    },
-    {
-      "epoch": 6.3,
-      "learning_rate": 2.0537454747981062e-05,
-      "loss": 0.524,
-      "step": 5030
-    },
-    {
-      "epoch": 6.32,
-      "learning_rate": 2.046783625730994e-05,
-      "loss": 0.4502,
-      "step": 5040
-    },
-    {
-      "epoch": 6.33,
-      "learning_rate": 2.0398217766638818e-05,
-      "loss": 0.4505,
-      "step": 5050
-    },
-    {
-      "epoch": 6.34,
-      "learning_rate": 2.0328599275967696e-05,
-      "loss": 0.4962,
-      "step": 5060
-    },
-    {
-      "epoch": 6.35,
-      "learning_rate": 2.0258980785296574e-05,
-      "loss": 0.4566,
-      "step": 5070
-    },
-    {
-      "epoch": 6.37,
-      "learning_rate": 2.018936229462545e-05,
-      "loss": 0.4464,
-      "step": 5080
-    },
-    {
-      "epoch": 6.38,
-      "learning_rate": 2.011974380395433e-05,
-      "loss": 0.4872,
-      "step": 5090
-    },
-    {
-      "epoch": 6.39,
-      "learning_rate": 2.0050125313283208e-05,
-      "loss": 0.4396,
-      "step": 5100
-    },
-    {
-      "epoch": 6.4,
-      "learning_rate": 1.9980506822612085e-05,
-      "loss": 0.5302,
-      "step": 5110
-    },
-    {
-      "epoch": 6.42,
-      "learning_rate": 1.9910888331940963e-05,
-      "loss": 0.4927,
-      "step": 5120
-    },
-    {
-      "epoch": 6.43,
-      "learning_rate": 1.984126984126984e-05,
-      "loss": 0.5129,
-      "step": 5130
-    },
-    {
-      "epoch": 6.44,
-      "learning_rate": 1.977165135059872e-05,
-      "loss": 0.4747,
-      "step": 5140
-    },
-    {
-      "epoch": 6.45,
-      "learning_rate": 1.9702032859927597e-05,
-      "loss": 0.4878,
-      "step": 5150
-    },
-    {
-      "epoch": 6.47,
-      "learning_rate": 1.9632414369256475e-05,
-      "loss": 0.4885,
-      "step": 5160
-    },
-    {
-      "epoch": 6.48,
-      "learning_rate": 1.9562795878585353e-05,
-      "loss": 0.4634,
-      "step": 5170
-    },
-    {
-      "epoch": 6.49,
-      "learning_rate": 1.949317738791423e-05,
-      "loss": 0.4665,
-      "step": 5180
-    },
-    {
-      "epoch": 6.5,
-      "learning_rate": 1.942355889724311e-05,
-      "loss": 0.4392,
-      "step": 5190
-    },
-    {
-      "epoch": 6.52,
-      "learning_rate": 1.9353940406571987e-05,
-      "loss": 0.4754,
-      "step": 5200
-    },
-    {
-      "epoch": 6.53,
-      "learning_rate": 1.9284321915900865e-05,
-      "loss": 0.457,
-      "step": 5210
-    },
-    {
-      "epoch": 6.54,
-      "learning_rate": 1.9214703425229743e-05,
-      "loss": 0.4734,
-      "step": 5220
-    },
-    {
-      "epoch": 6.55,
-      "learning_rate": 1.9145084934558617e-05,
-      "loss": 0.4556,
-      "step": 5230
-    },
-    {
-      "epoch": 6.57,
-      "learning_rate": 1.9075466443887495e-05,
-      "loss": 0.4844,
-      "step": 5240
-    },
-    {
-      "epoch": 6.58,
-      "learning_rate": 1.9005847953216373e-05,
-      "loss": 0.4161,
-      "step": 5250
-    },
-    {
-      "epoch": 6.59,
-      "learning_rate": 1.893622946254525e-05,
-      "loss": 0.4722,
-      "step": 5260
-    },
-    {
-      "epoch": 6.6,
-      "learning_rate": 1.886661097187413e-05,
-      "loss": 0.517,
-      "step": 5270
-    },
-    {
-      "epoch": 6.62,
-      "learning_rate": 1.8796992481203007e-05,
-      "loss": 0.4612,
-      "step": 5280
-    },
-    {
-      "epoch": 6.63,
-      "learning_rate": 1.8727373990531885e-05,
-      "loss": 0.5017,
-      "step": 5290
-    },
-    {
-      "epoch": 6.64,
-      "learning_rate": 1.8657755499860762e-05,
-      "loss": 0.4882,
-      "step": 5300
-    },
-    {
-      "epoch": 6.65,
-      "learning_rate": 1.858813700918964e-05,
-      "loss": 0.4722,
-      "step": 5310
-    },
-    {
-      "epoch": 6.67,
-      "learning_rate": 1.8518518518518518e-05,
-      "loss": 0.4674,
-      "step": 5320
-    },
-    {
-      "epoch": 6.68,
-      "learning_rate": 1.8448900027847396e-05,
-      "loss": 0.4928,
-      "step": 5330
-    },
-    {
-      "epoch": 6.69,
-      "learning_rate": 1.8379281537176274e-05,
-      "loss": 0.4426,
-      "step": 5340
-    },
-    {
-      "epoch": 6.7,
-      "learning_rate": 1.8309663046505152e-05,
-      "loss": 0.4778,
-      "step": 5350
-    },
-    {
-      "epoch": 6.72,
-      "learning_rate": 1.824004455583403e-05,
-      "loss": 0.4555,
-      "step": 5360
-    },
-    {
-      "epoch": 6.73,
-      "learning_rate": 1.8170426065162908e-05,
-      "loss": 0.4707,
-      "step": 5370
-    },
-    {
-      "epoch": 6.74,
-      "learning_rate": 1.8100807574491786e-05,
-      "loss": 0.4729,
-      "step": 5380
-    },
-    {
-      "epoch": 6.75,
-      "learning_rate": 1.8031189083820664e-05,
-      "loss": 0.5236,
-      "step": 5390
-    },
-    {
-      "epoch": 6.77,
-      "learning_rate": 1.796157059314954e-05,
-      "loss": 0.4635,
-      "step": 5400
-    },
-    {
-      "epoch": 6.78,
-      "learning_rate": 1.789195210247842e-05,
-      "loss": 0.4731,
-      "step": 5410
-    },
-    {
-      "epoch": 6.79,
-      "learning_rate": 1.7822333611807297e-05,
-      "loss": 0.4784,
-      "step": 5420
-    },
-    {
-      "epoch": 6.8,
-      "learning_rate": 1.7752715121136172e-05,
-      "loss": 0.4913,
-      "step": 5430
-    },
-    {
-      "epoch": 6.82,
-      "learning_rate": 1.768309663046505e-05,
-      "loss": 0.4617,
-      "step": 5440
-    },
-    {
-      "epoch": 6.83,
-      "learning_rate": 1.7613478139793928e-05,
-      "loss": 0.4888,
-      "step": 5450
-    },
-    {
-      "epoch": 6.84,
-      "learning_rate": 1.7543859649122806e-05,
-      "loss": 0.4976,
-      "step": 5460
-    },
-    {
-      "epoch": 6.85,
-      "learning_rate": 1.7474241158451684e-05,
-      "loss": 0.4778,
-      "step": 5470
-    },
-    {
-      "epoch": 6.87,
-      "learning_rate": 1.740462266778056e-05,
-      "loss": 0.4818,
-      "step": 5480
-    },
-    {
-      "epoch": 6.88,
-      "learning_rate": 1.733500417710944e-05,
-      "loss": 0.5,
-      "step": 5490
-    },
-    {
-      "epoch": 6.89,
-      "learning_rate": 1.7265385686438317e-05,
-      "loss": 0.4529,
-      "step": 5500
-    },
-    {
-      "epoch": 6.9,
-      "learning_rate": 1.7195767195767195e-05,
-      "loss": 0.4674,
-      "step": 5510
-    },
-    {
-      "epoch": 6.92,
-      "learning_rate": 1.7126148705096073e-05,
-      "loss": 0.4733,
-      "step": 5520
-    },
-    {
-      "epoch": 6.93,
-      "learning_rate": 1.705653021442495e-05,
-      "loss": 0.4702,
-      "step": 5530
-    },
-    {
-      "epoch": 6.94,
-      "learning_rate": 1.698691172375383e-05,
-      "loss": 0.4809,
-      "step": 5540
-    },
-    {
-      "epoch": 6.95,
-      "learning_rate": 1.6917293233082707e-05,
-      "loss": 0.509,
-      "step": 5550
-    },
-    {
-      "epoch": 6.97,
-      "learning_rate": 1.6847674742411585e-05,
-      "loss": 0.4692,
-      "step": 5560
-    },
-    {
-      "epoch": 6.98,
-      "learning_rate": 1.6778056251740463e-05,
-      "loss": 0.5013,
-      "step": 5570
-    },
-    {
-      "epoch": 6.99,
-      "learning_rate": 1.670843776106934e-05,
-      "loss": 0.4245,
-      "step": 5580
-    },
-    {
-      "epoch": 7.0,
-      "eval_accuracy": 0.8541666666666666,
-      "eval_loss": 0.43816250562667847,
-      "eval_runtime": 233.7655,
-      "eval_samples_per_second": 109.238,
-      "eval_steps_per_second": 3.414,
-      "step": 5586
-    },
-    {
-      "epoch": 7.01,
-      "learning_rate": 1.663881927039822e-05,
-      "loss": 0.4232,
-      "step": 5590
-    },
-    {
-      "epoch": 7.02,
-      "learning_rate": 1.6569200779727097e-05,
-      "loss": 0.4133,
-      "step": 5600
-    },
-    {
-      "epoch": 7.03,
-      "learning_rate": 1.6499582289055974e-05,
-      "loss": 0.4066,
-      "step": 5610
-    },
-    {
-      "epoch": 7.04,
-      "learning_rate": 1.6429963798384852e-05,
-      "loss": 0.4381,
-      "step": 5620
-    },
-    {
-      "epoch": 7.06,
-      "learning_rate": 1.636034530771373e-05,
-      "loss": 0.4063,
-      "step": 5630
-    },
-    {
-      "epoch": 7.07,
-      "learning_rate": 1.6290726817042605e-05,
-      "loss": 0.3721,
-      "step": 5640
-    },
-    {
-      "epoch": 7.08,
-      "learning_rate": 1.6221108326371483e-05,
-      "loss": 0.3574,
-      "step": 5650
-    },
-    {
-      "epoch": 7.09,
-      "learning_rate": 1.615148983570036e-05,
-      "loss": 0.4156,
-      "step": 5660
-    },
-    {
-      "epoch": 7.11,
-      "learning_rate": 1.608187134502924e-05,
-      "loss": 0.4311,
-      "step": 5670
-    },
-    {
-      "epoch": 7.12,
-      "learning_rate": 1.6012252854358116e-05,
-      "loss": 0.3922,
-      "step": 5680
-    },
-    {
-      "epoch": 7.13,
-      "learning_rate": 1.5942634363686994e-05,
-      "loss": 0.3568,
-      "step": 5690
-    },
-    {
-      "epoch": 7.14,
-      "learning_rate": 1.5873015873015872e-05,
-      "loss": 0.3943,
-      "step": 5700
-    },
-    {
-      "epoch": 7.16,
-      "learning_rate": 1.580339738234475e-05,
-      "loss": 0.3995,
-      "step": 5710
-    },
-    {
-      "epoch": 7.17,
-      "learning_rate": 1.5733778891673628e-05,
-      "loss": 0.3948,
-      "step": 5720
-    },
-    {
-      "epoch": 7.18,
-      "learning_rate": 1.5664160401002506e-05,
-      "loss": 0.3912,
-      "step": 5730
-    },
-    {
-      "epoch": 7.19,
-      "learning_rate": 1.5594541910331384e-05,
-      "loss": 0.3791,
-      "step": 5740
-    },
-    {
-      "epoch": 7.21,
-      "learning_rate": 1.5524923419660262e-05,
-      "loss": 0.4085,
-      "step": 5750
-    },
-    {
-      "epoch": 7.22,
-      "learning_rate": 1.545530492898914e-05,
-      "loss": 0.4175,
-      "step": 5760
-    },
-    {
-      "epoch": 7.23,
-      "learning_rate": 1.5385686438318018e-05,
-      "loss": 0.4018,
-      "step": 5770
-    },
-    {
-      "epoch": 7.24,
-      "learning_rate": 1.5316067947646896e-05,
-      "loss": 0.4203,
-      "step": 5780
-    },
-    {
-      "epoch": 7.26,
-      "learning_rate": 1.5246449456975772e-05,
-      "loss": 0.3939,
-      "step": 5790
-    },
-    {
-      "epoch": 7.27,
-      "learning_rate": 1.517683096630465e-05,
-      "loss": 0.4417,
-      "step": 5800
-    },
-    {
-      "epoch": 7.28,
-      "learning_rate": 1.5107212475633528e-05,
-      "loss": 0.3599,
-      "step": 5810
-    },
-    {
-      "epoch": 7.29,
-      "learning_rate": 1.5037593984962406e-05,
-      "loss": 0.3877,
-      "step": 5820
-    },
-    {
-      "epoch": 7.31,
-      "learning_rate": 1.4967975494291284e-05,
-      "loss": 0.4054,
-      "step": 5830
-    },
-    {
-      "epoch": 7.32,
-      "learning_rate": 1.4898357003620161e-05,
-      "loss": 0.3935,
-      "step": 5840
-    },
-    {
-      "epoch": 7.33,
-      "learning_rate": 1.482873851294904e-05,
-      "loss": 0.3589,
-      "step": 5850
-    },
-    {
-      "epoch": 7.34,
-      "learning_rate": 1.4759120022277917e-05,
-      "loss": 0.3815,
-      "step": 5860
-    },
-    {
-      "epoch": 7.36,
-      "learning_rate": 1.4689501531606795e-05,
-      "loss": 0.4185,
-      "step": 5870
-    },
-    {
-      "epoch": 7.37,
-      "learning_rate": 1.4619883040935673e-05,
-      "loss": 0.4318,
-      "step": 5880
-    },
-    {
-      "epoch": 7.38,
-      "learning_rate": 1.455026455026455e-05,
-      "loss": 0.3831,
-      "step": 5890
-    },
-    {
-      "epoch": 7.39,
-      "learning_rate": 1.4480646059593427e-05,
-      "loss": 0.3999,
-      "step": 5900
-    },
-    {
-      "epoch": 7.41,
-      "learning_rate": 1.4411027568922305e-05,
-      "loss": 0.4112,
-      "step": 5910
-    },
-    {
-      "epoch": 7.42,
-      "learning_rate": 1.4341409078251183e-05,
-      "loss": 0.4429,
-      "step": 5920
-    },
-    {
-      "epoch": 7.43,
-      "learning_rate": 1.4271790587580061e-05,
-      "loss": 0.4023,
-      "step": 5930
-    },
-    {
-      "epoch": 7.44,
-      "learning_rate": 1.4202172096908939e-05,
-      "loss": 0.4087,
-      "step": 5940
-    },
-    {
-      "epoch": 7.46,
-      "learning_rate": 1.4132553606237817e-05,
-      "loss": 0.3838,
-      "step": 5950
-    },
-    {
-      "epoch": 7.47,
-      "learning_rate": 1.4062935115566695e-05,
-      "loss": 0.4084,
-      "step": 5960
-    },
-    {
-      "epoch": 7.48,
-      "learning_rate": 1.3993316624895573e-05,
-      "loss": 0.3851,
-      "step": 5970
-    },
-    {
-      "epoch": 7.49,
-      "learning_rate": 1.392369813422445e-05,
-      "loss": 0.4234,
-      "step": 5980
-    },
-    {
-      "epoch": 7.51,
-      "learning_rate": 1.3854079643553327e-05,
-      "loss": 0.4183,
-      "step": 5990
-    },
-    {
-      "epoch": 7.52,
-      "learning_rate": 1.3784461152882205e-05,
-      "loss": 0.4108,
-      "step": 6000
-    },
-    {
-      "epoch": 7.53,
-      "learning_rate": 1.3714842662211083e-05,
-      "loss": 0.4314,
-      "step": 6010
-    },
-    {
-      "epoch": 7.54,
-      "learning_rate": 1.364522417153996e-05,
-      "loss": 0.4252,
-      "step": 6020
-    },
-    {
-      "epoch": 7.56,
-      "learning_rate": 1.3575605680868838e-05,
-      "loss": 0.4272,
-      "step": 6030
-    },
-    {
-      "epoch": 7.57,
-      "learning_rate": 1.3505987190197716e-05,
-      "loss": 0.3799,
-      "step": 6040
-    },
-    {
-      "epoch": 7.58,
-      "learning_rate": 1.3436368699526594e-05,
-      "loss": 0.3705,
-      "step": 6050
-    },
-    {
-      "epoch": 7.59,
-      "learning_rate": 1.3366750208855472e-05,
-      "loss": 0.4124,
-      "step": 6060
-    },
-    {
-      "epoch": 7.61,
-      "learning_rate": 1.329713171818435e-05,
-      "loss": 0.3976,
-      "step": 6070
-    },
-    {
-      "epoch": 7.62,
-      "learning_rate": 1.3227513227513228e-05,
-      "loss": 0.4187,
-      "step": 6080
-    },
-    {
-      "epoch": 7.63,
-      "learning_rate": 1.3157894736842106e-05,
-      "loss": 0.386,
-      "step": 6090
-    },
-    {
-      "epoch": 7.64,
-      "learning_rate": 1.3088276246170982e-05,
-      "loss": 0.4153,
-      "step": 6100
-    },
-    {
-      "epoch": 7.66,
-      "learning_rate": 1.301865775549986e-05,
-      "loss": 0.4067,
-      "step": 6110
-    },
-    {
-      "epoch": 7.67,
-      "learning_rate": 1.2949039264828738e-05,
-      "loss": 0.3874,
-      "step": 6120
-    },
-    {
-      "epoch": 7.68,
-      "learning_rate": 1.2879420774157616e-05,
-      "loss": 0.3921,
-      "step": 6130
-    },
-    {
-      "epoch": 7.69,
-      "learning_rate": 1.2809802283486494e-05,
-      "loss": 0.369,
-      "step": 6140
-    },
-    {
-      "epoch": 7.71,
-      "learning_rate": 1.2740183792815372e-05,
-      "loss": 0.4082,
-      "step": 6150
-    },
-    {
-      "epoch": 7.72,
-      "learning_rate": 1.267056530214425e-05,
-      "loss": 0.4238,
-      "step": 6160
-    },
-    {
-      "epoch": 7.73,
-      "learning_rate": 1.2600946811473128e-05,
-      "loss": 0.3899,
-      "step": 6170
-    },
-    {
-      "epoch": 7.74,
-      "learning_rate": 1.2531328320802006e-05,
-      "loss": 0.3806,
-      "step": 6180
-    },
-    {
-      "epoch": 7.76,
-      "learning_rate": 1.2461709830130883e-05,
-      "loss": 0.3617,
-      "step": 6190
-    },
-    {
-      "epoch": 7.77,
-      "learning_rate": 1.239209133945976e-05,
-      "loss": 0.4171,
-      "step": 6200
-    },
-    {
-      "epoch": 7.78,
-      "learning_rate": 1.2322472848788638e-05,
-      "loss": 0.3941,
-      "step": 6210
-    },
-    {
-      "epoch": 7.79,
-      "learning_rate": 1.2252854358117516e-05,
-      "loss": 0.3891,
-      "step": 6220
-    },
-    {
-      "epoch": 7.81,
-      "learning_rate": 1.2183235867446393e-05,
-      "loss": 0.3987,
-      "step": 6230
-    },
-    {
-      "epoch": 7.82,
-      "learning_rate": 1.2113617376775271e-05,
-      "loss": 0.3707,
-      "step": 6240
-    },
-    {
-      "epoch": 7.83,
-      "learning_rate": 1.204399888610415e-05,
-      "loss": 0.3835,
-      "step": 6250
-    },
-    {
-      "epoch": 7.84,
-      "learning_rate": 1.1974380395433027e-05,
-      "loss": 0.3925,
-      "step": 6260
-    },
-    {
-      "epoch": 7.86,
-      "learning_rate": 1.1904761904761905e-05,
-      "loss": 0.3693,
-      "step": 6270
-    },
-    {
-      "epoch": 7.87,
-      "learning_rate": 1.1835143414090783e-05,
-      "loss": 0.3809,
-      "step": 6280
-    },
-    {
-      "epoch": 7.88,
-      "learning_rate": 1.1765524923419661e-05,
-      "loss": 0.3872,
-      "step": 6290
-    },
-    {
-      "epoch": 7.89,
-      "learning_rate": 1.1695906432748537e-05,
-      "loss": 0.3964,
-      "step": 6300
-    },
-    {
-      "epoch": 7.91,
-      "learning_rate": 1.1626287942077415e-05,
-      "loss": 0.3767,
-      "step": 6310
-    },
-    {
-      "epoch": 7.92,
-      "learning_rate": 1.1556669451406293e-05,
-      "loss": 0.3902,
-      "step": 6320
-    },
-    {
-      "epoch": 7.93,
-      "learning_rate": 1.1487050960735171e-05,
-      "loss": 0.3551,
-      "step": 6330
-    },
-    {
-      "epoch": 7.94,
-      "learning_rate": 1.1417432470064049e-05,
-      "loss": 0.4097,
-      "step": 6340
-    },
-    {
-      "epoch": 7.96,
-      "learning_rate": 1.1347813979392927e-05,
-      "loss": 0.4374,
-      "step": 6350
-    },
-    {
-      "epoch": 7.97,
-      "learning_rate": 1.1278195488721805e-05,
-      "loss": 0.3815,
-      "step": 6360
-    },
-    {
-      "epoch": 7.98,
-      "learning_rate": 1.1208576998050683e-05,
-      "loss": 0.3829,
-      "step": 6370
-    },
-    {
-      "epoch": 7.99,
-      "learning_rate": 1.113895850737956e-05,
-      "loss": 0.3806,
-      "step": 6380
-    },
-    {
-      "epoch": 8.0,
-      "eval_accuracy": 0.853109335839599,
-      "eval_loss": 0.4375392496585846,
-      "eval_runtime": 234.5395,
-      "eval_samples_per_second": 108.877,
-      "eval_steps_per_second": 3.402,
-      "step": 6384
-    },
-    {
-      "epoch": 8.01,
-      "learning_rate": 1.1069340016708438e-05,
-      "loss": 0.3573,
-      "step": 6390
-    },
-    {
-      "epoch": 8.02,
-      "learning_rate": 1.0999721526037315e-05,
-      "loss": 0.3303,
-      "step": 6400
-    },
-    {
-      "epoch": 8.03,
-      "learning_rate": 1.0930103035366193e-05,
-      "loss": 0.3206,
-      "step": 6410
-    },
-    {
-      "epoch": 8.05,
-      "learning_rate": 1.086048454469507e-05,
-      "loss": 0.3416,
-      "step": 6420
-    },
-    {
-      "epoch": 8.06,
-      "learning_rate": 1.0790866054023948e-05,
-      "loss": 0.3379,
-      "step": 6430
-    },
-    {
-      "epoch": 8.07,
-      "learning_rate": 1.0721247563352826e-05,
-      "loss": 0.3255,
-      "step": 6440
-    },
-    {
-      "epoch": 8.08,
-      "learning_rate": 1.0651629072681704e-05,
-      "loss": 0.3278,
-      "step": 6450
-    },
-    {
-      "epoch": 8.1,
-      "learning_rate": 1.0582010582010582e-05,
-      "loss": 0.3136,
-      "step": 6460
-    },
-    {
-      "epoch": 8.11,
-      "learning_rate": 1.051239209133946e-05,
-      "loss": 0.2932,
-      "step": 6470
-    },
-    {
-      "epoch": 8.12,
-      "learning_rate": 1.0442773600668338e-05,
-      "loss": 0.3199,
-      "step": 6480
-    },
-    {
-      "epoch": 8.13,
-      "learning_rate": 1.0373155109997216e-05,
-      "loss": 0.3212,
-      "step": 6490
-    },
-    {
-      "epoch": 8.15,
-      "learning_rate": 1.0303536619326092e-05,
-      "loss": 0.3487,
-      "step": 6500
-    },
-    {
-      "epoch": 8.16,
-      "learning_rate": 1.023391812865497e-05,
-      "loss": 0.3364,
-      "step": 6510
-    },
-    {
-      "epoch": 8.17,
-      "learning_rate": 1.0164299637983848e-05,
-      "loss": 0.3261,
-      "step": 6520
-    },
-    {
-      "epoch": 8.18,
-      "learning_rate": 1.0094681147312726e-05,
-      "loss": 0.3398,
-      "step": 6530
-    },
-    {
-      "epoch": 8.2,
-      "learning_rate": 1.0025062656641604e-05,
-      "loss": 0.3111,
-      "step": 6540
-    },
-    {
-      "epoch": 8.21,
-      "learning_rate": 9.955444165970482e-06,
-      "loss": 0.3523,
-      "step": 6550
-    },
-    {
-      "epoch": 8.22,
-      "learning_rate": 9.88582567529936e-06,
-      "loss": 0.3325,
-      "step": 6560
-    },
-    {
-      "epoch": 8.23,
-      "learning_rate": 9.816207184628238e-06,
-      "loss": 0.329,
-      "step": 6570
-    },
-    {
-      "epoch": 8.25,
-      "learning_rate": 9.746588693957115e-06,
-      "loss": 0.3558,
-      "step": 6580
-    },
-    {
-      "epoch": 8.26,
-      "learning_rate": 9.676970203285993e-06,
-      "loss": 0.3286,
-      "step": 6590
-    },
-    {
-      "epoch": 8.27,
-      "learning_rate": 9.607351712614871e-06,
-      "loss": 0.3429,
-      "step": 6600
-    },
-    {
-      "epoch": 8.28,
-      "learning_rate": 9.537733221943747e-06,
-      "loss": 0.3146,
-      "step": 6610
-    },
-    {
-      "epoch": 8.3,
-      "learning_rate": 9.468114731272625e-06,
-      "loss": 0.3075,
-      "step": 6620
-    },
-    {
-      "epoch": 8.31,
-      "learning_rate": 9.398496240601503e-06,
-      "loss": 0.3053,
-      "step": 6630
-    },
-    {
-      "epoch": 8.32,
-      "learning_rate": 9.328877749930381e-06,
-      "loss": 0.3262,
-      "step": 6640
-    },
-    {
-      "epoch": 8.33,
-      "learning_rate": 9.259259259259259e-06,
-      "loss": 0.3315,
-      "step": 6650
-    },
-    {
-      "epoch": 8.35,
-      "learning_rate": 9.189640768588137e-06,
-      "loss": 0.3446,
-      "step": 6660
-    },
-    {
-      "epoch": 8.36,
-      "learning_rate": 9.120022277917015e-06,
-      "loss": 0.3406,
-      "step": 6670
-    },
-    {
-      "epoch": 8.37,
-      "learning_rate": 9.050403787245893e-06,
-      "loss": 0.3119,
-      "step": 6680
-    },
-    {
-      "epoch": 8.38,
-      "learning_rate": 8.98078529657477e-06,
-      "loss": 0.3102,
-      "step": 6690
-    },
-    {
-      "epoch": 8.4,
-      "learning_rate": 8.911166805903649e-06,
-      "loss": 0.3051,
-      "step": 6700
-    },
-    {
-      "epoch": 8.41,
-      "learning_rate": 8.841548315232525e-06,
-      "loss": 0.3509,
-      "step": 6710
-    },
-    {
-      "epoch": 8.42,
-      "learning_rate": 8.771929824561403e-06,
-      "loss": 0.3265,
-      "step": 6720
-    },
-    {
-      "epoch": 8.43,
-      "learning_rate": 8.70231133389028e-06,
-      "loss": 0.3522,
-      "step": 6730
-    },
-    {
-      "epoch": 8.45,
-      "learning_rate": 8.632692843219159e-06,
-      "loss": 0.3029,
-      "step": 6740
-    },
-    {
-      "epoch": 8.46,
-      "learning_rate": 8.563074352548037e-06,
-      "loss": 0.317,
-      "step": 6750
-    },
-    {
-      "epoch": 8.47,
-      "learning_rate": 8.493455861876915e-06,
-      "loss": 0.3446,
-      "step": 6760
-    },
-    {
-      "epoch": 8.48,
-      "learning_rate": 8.423837371205792e-06,
-      "loss": 0.3627,
-      "step": 6770
-    },
-    {
-      "epoch": 8.5,
-      "learning_rate": 8.35421888053467e-06,
-      "loss": 0.3583,
-      "step": 6780
-    },
-    {
-      "epoch": 8.51,
-      "learning_rate": 8.284600389863548e-06,
-      "loss": 0.3106,
-      "step": 6790
-    },
-    {
-      "epoch": 8.52,
-      "learning_rate": 8.214981899192426e-06,
-      "loss": 0.3142,
-      "step": 6800
-    },
-    {
-      "epoch": 8.53,
-      "learning_rate": 8.145363408521302e-06,
-      "loss": 0.3031,
-      "step": 6810
-    },
-    {
-      "epoch": 8.55,
-      "learning_rate": 8.07574491785018e-06,
-      "loss": 0.3235,
-      "step": 6820
-    },
-    {
-      "epoch": 8.56,
-      "learning_rate": 8.006126427179058e-06,
-      "loss": 0.3186,
-      "step": 6830
-    },
-    {
-      "epoch": 8.57,
-      "learning_rate": 7.936507936507936e-06,
-      "loss": 0.2935,
-      "step": 6840
-    },
-    {
-      "epoch": 8.58,
-      "learning_rate": 7.866889445836814e-06,
-      "loss": 0.2895,
-      "step": 6850
-    },
-    {
-      "epoch": 8.6,
-      "learning_rate": 7.797270955165692e-06,
-      "loss": 0.3252,
-      "step": 6860
-    },
-    {
-      "epoch": 8.61,
-      "learning_rate": 7.72765246449457e-06,
-      "loss": 0.3128,
-      "step": 6870
-    },
-    {
-      "epoch": 8.62,
-      "learning_rate": 7.658033973823448e-06,
-      "loss": 0.3176,
-      "step": 6880
-    },
-    {
-      "epoch": 8.63,
-      "learning_rate": 7.588415483152325e-06,
-      "loss": 0.3343,
-      "step": 6890
-    },
-    {
-      "epoch": 8.65,
-      "learning_rate": 7.518796992481203e-06,
-      "loss": 0.3738,
-      "step": 6900
-    },
-    {
-      "epoch": 8.66,
-      "learning_rate": 7.449178501810081e-06,
-      "loss": 0.3144,
-      "step": 6910
-    },
-    {
-      "epoch": 8.67,
-      "learning_rate": 7.379560011138959e-06,
-      "loss": 0.3808,
-      "step": 6920
-    },
-    {
-      "epoch": 8.68,
-      "learning_rate": 7.3099415204678366e-06,
-      "loss": 0.3089,
-      "step": 6930
-    },
-    {
-      "epoch": 8.7,
-      "learning_rate": 7.240323029796714e-06,
-      "loss": 0.3327,
-      "step": 6940
-    },
-    {
-      "epoch": 8.71,
-      "learning_rate": 7.1707045391255915e-06,
-      "loss": 0.3289,
-      "step": 6950
-    },
-    {
-      "epoch": 8.72,
-      "learning_rate": 7.1010860484544695e-06,
-      "loss": 0.3699,
-      "step": 6960
-    },
-    {
-      "epoch": 8.73,
-      "learning_rate": 7.031467557783347e-06,
-      "loss": 0.3061,
-      "step": 6970
-    },
-    {
-      "epoch": 8.75,
-      "learning_rate": 6.961849067112225e-06,
-      "loss": 0.3026,
-      "step": 6980
-    },
-    {
-      "epoch": 8.76,
-      "learning_rate": 6.892230576441102e-06,
-      "loss": 0.3233,
-      "step": 6990
-    },
-    {
-      "epoch": 8.77,
-      "learning_rate": 6.82261208576998e-06,
-      "loss": 0.3106,
-      "step": 7000
-    },
-    {
-      "epoch": 8.78,
-      "learning_rate": 6.752993595098858e-06,
-      "loss": 0.3429,
-      "step": 7010
-    },
-    {
-      "epoch": 8.8,
-      "learning_rate": 6.683375104427736e-06,
-      "loss": 0.3397,
-      "step": 7020
-    },
-    {
-      "epoch": 8.81,
-      "learning_rate": 6.613756613756614e-06,
-      "loss": 0.352,
-      "step": 7030
-    },
-    {
-      "epoch": 8.82,
-      "learning_rate": 6.544138123085491e-06,
-      "loss": 0.2803,
-      "step": 7040
-    },
-    {
-      "epoch": 8.83,
-      "learning_rate": 6.474519632414369e-06,
-      "loss": 0.3335,
-      "step": 7050
-    },
-    {
-      "epoch": 8.85,
-      "learning_rate": 6.404901141743247e-06,
-      "loss": 0.3054,
-      "step": 7060
-    },
-    {
-      "epoch": 8.86,
-      "learning_rate": 6.335282651072125e-06,
-      "loss": 0.3383,
-      "step": 7070
-    },
-    {
-      "epoch": 8.87,
-      "learning_rate": 6.265664160401003e-06,
-      "loss": 0.3237,
-      "step": 7080
-    },
-    {
-      "epoch": 8.88,
-      "learning_rate": 6.19604566972988e-06,
-      "loss": 0.3227,
-      "step": 7090
-    },
-    {
-      "epoch": 8.9,
-      "learning_rate": 6.126427179058758e-06,
-      "loss": 0.3375,
-      "step": 7100
-    },
-    {
-      "epoch": 8.91,
-      "learning_rate": 6.056808688387636e-06,
-      "loss": 0.2785,
-      "step": 7110
-    },
-    {
-      "epoch": 8.92,
-      "learning_rate": 5.987190197716514e-06,
-      "loss": 0.31,
-      "step": 7120
-    },
-    {
-      "epoch": 8.93,
-      "learning_rate": 5.9175717070453915e-06,
-      "loss": 0.3472,
-      "step": 7130
-    },
-    {
-      "epoch": 8.95,
-      "learning_rate": 5.8479532163742686e-06,
-      "loss": 0.3577,
-      "step": 7140
-    },
-    {
-      "epoch": 8.96,
-      "learning_rate": 5.7783347257031465e-06,
-      "loss": 0.3299,
-      "step": 7150
-    },
-    {
-      "epoch": 8.97,
-      "learning_rate": 5.708716235032024e-06,
-      "loss": 0.2989,
-      "step": 7160
-    },
-    {
-      "epoch": 8.98,
-      "learning_rate": 5.639097744360902e-06,
-      "loss": 0.2973,
-      "step": 7170
-    },
-    {
-      "epoch": 9.0,
-      "learning_rate": 5.56947925368978e-06,
-      "loss": 0.3112,
-      "step": 7180
-    },
-    {
-      "epoch": 9.0,
-      "eval_accuracy": 0.8556547619047619,
-      "eval_loss": 0.4372400641441345,
-      "eval_runtime": 234.8854,
-      "eval_samples_per_second": 108.717,
-      "eval_steps_per_second": 3.397,
-      "step": 7182
-    },
-    {
-      "epoch": 9.01,
-      "learning_rate": 5.499860763018657e-06,
-      "loss": 0.2907,
-      "step": 7190
-    },
-    {
-      "epoch": 9.02,
-      "learning_rate": 5.430242272347535e-06,
-      "loss": 0.3028,
-      "step": 7200
-    },
-    {
-      "epoch": 9.04,
-      "learning_rate": 5.360623781676413e-06,
-      "loss": 0.2706,
-      "step": 7210
-    },
-    {
-      "epoch": 9.05,
-      "learning_rate": 5.291005291005291e-06,
-      "loss": 0.2695,
-      "step": 7220
-    },
-    {
-      "epoch": 9.06,
-      "learning_rate": 5.221386800334169e-06,
-      "loss": 0.3298,
-      "step": 7230
-    },
-    {
-      "epoch": 9.07,
-      "learning_rate": 5.151768309663046e-06,
-      "loss": 0.2717,
-      "step": 7240
-    },
-    {
-      "epoch": 9.09,
-      "learning_rate": 5.082149818991924e-06,
-      "loss": 0.2785,
-      "step": 7250
-    },
-    {
-      "epoch": 9.1,
-      "learning_rate": 5.012531328320802e-06,
-      "loss": 0.2635,
-      "step": 7260
-    },
-    {
-      "epoch": 9.11,
-      "learning_rate": 4.94291283764968e-06,
-      "loss": 0.305,
-      "step": 7270
-    },
-    {
-      "epoch": 9.12,
-      "learning_rate": 4.873294346978558e-06,
-      "loss": 0.2786,
-      "step": 7280
-    },
-    {
-      "epoch": 9.14,
-      "learning_rate": 4.803675856307436e-06,
-      "loss": 0.2568,
-      "step": 7290
-    },
-    {
-      "epoch": 9.15,
-      "learning_rate": 4.734057365636313e-06,
-      "loss": 0.261,
-      "step": 7300
-    },
-    {
-      "epoch": 9.16,
-      "learning_rate": 4.664438874965191e-06,
-      "loss": 0.2548,
-      "step": 7310
-    },
-    {
-      "epoch": 9.17,
-      "learning_rate": 4.5948203842940685e-06,
-      "loss": 0.2698,
-      "step": 7320
-    },
-    {
-      "epoch": 9.19,
-      "learning_rate": 4.5252018936229464e-06,
-      "loss": 0.2965,
-      "step": 7330
-    },
-    {
-      "epoch": 9.2,
-      "learning_rate": 4.455583402951824e-06,
-      "loss": 0.2601,
-      "step": 7340
-    },
-    {
-      "epoch": 9.21,
-      "learning_rate": 4.3859649122807014e-06,
-      "loss": 0.2512,
-      "step": 7350
-    },
-    {
-      "epoch": 9.22,
-      "learning_rate": 4.316346421609579e-06,
-      "loss": 0.3007,
-      "step": 7360
-    },
-    {
-      "epoch": 9.24,
-      "learning_rate": 4.246727930938457e-06,
-      "loss": 0.2669,
-      "step": 7370
-    },
-    {
-      "epoch": 9.25,
-      "learning_rate": 4.177109440267335e-06,
-      "loss": 0.2709,
-      "step": 7380
-    },
-    {
-      "epoch": 9.26,
-      "learning_rate": 4.107490949596213e-06,
-      "loss": 0.2455,
-      "step": 7390
-    },
-    {
-      "epoch": 9.27,
-      "learning_rate": 4.03787245892509e-06,
-      "loss": 0.2737,
-      "step": 7400
-    },
-    {
-      "epoch": 9.29,
-      "learning_rate": 3.968253968253968e-06,
-      "loss": 0.2845,
-      "step": 7410
-    },
-    {
-      "epoch": 9.3,
-      "learning_rate": 3.898635477582846e-06,
-      "loss": 0.2437,
-      "step": 7420
-    },
-    {
-      "epoch": 9.31,
-      "learning_rate": 3.829016986911724e-06,
-      "loss": 0.2716,
-      "step": 7430
-    },
-    {
-      "epoch": 9.32,
-      "learning_rate": 3.7593984962406014e-06,
-      "loss": 0.2764,
-      "step": 7440
-    },
-    {
-      "epoch": 9.34,
-      "learning_rate": 3.6897800055694793e-06,
-      "loss": 0.2819,
-      "step": 7450
-    },
-    {
-      "epoch": 9.35,
-      "learning_rate": 3.620161514898357e-06,
-      "loss": 0.2976,
-      "step": 7460
-    },
-    {
-      "epoch": 9.36,
-      "learning_rate": 3.5505430242272347e-06,
-      "loss": 0.2553,
-      "step": 7470
-    },
-    {
-      "epoch": 9.37,
-      "learning_rate": 3.4809245335561126e-06,
-      "loss": 0.2817,
-      "step": 7480
-    },
-    {
-      "epoch": 9.39,
-      "learning_rate": 3.41130604288499e-06,
-      "loss": 0.2809,
-      "step": 7490
-    },
-    {
-      "epoch": 9.4,
-      "learning_rate": 3.341687552213868e-06,
-      "loss": 0.254,
-      "step": 7500
-    },
-    {
-      "epoch": 9.41,
-      "learning_rate": 3.2720690615427456e-06,
-      "loss": 0.2814,
-      "step": 7510
-    },
-    {
-      "epoch": 9.42,
-      "learning_rate": 3.2024505708716235e-06,
-      "loss": 0.2923,
-      "step": 7520
-    },
-    {
-      "epoch": 9.44,
-      "learning_rate": 3.1328320802005014e-06,
-      "loss": 0.2792,
-      "step": 7530
-    },
-    {
-      "epoch": 9.45,
-      "learning_rate": 3.063213589529379e-06,
-      "loss": 0.2843,
-      "step": 7540
-    },
-    {
-      "epoch": 9.46,
-      "learning_rate": 2.993595098858257e-06,
-      "loss": 0.2724,
-      "step": 7550
-    },
-    {
-      "epoch": 9.47,
-      "learning_rate": 2.9239766081871343e-06,
-      "loss": 0.2918,
-      "step": 7560
-    },
-    {
-      "epoch": 9.49,
-      "learning_rate": 2.854358117516012e-06,
-      "loss": 0.2756,
-      "step": 7570
-    },
-    {
-      "epoch": 9.5,
-      "learning_rate": 2.78473962684489e-06,
-      "loss": 0.303,
-      "step": 7580
-    },
-    {
-      "epoch": 9.51,
-      "learning_rate": 2.7151211361737676e-06,
-      "loss": 0.2556,
-      "step": 7590
-    },
-    {
-      "epoch": 9.52,
-      "learning_rate": 2.6455026455026455e-06,
-      "loss": 0.2687,
-      "step": 7600
-    },
-    {
-      "epoch": 9.54,
-      "learning_rate": 2.575884154831523e-06,
-      "loss": 0.2527,
-      "step": 7610
-    },
-    {
-      "epoch": 9.55,
-      "learning_rate": 2.506265664160401e-06,
-      "loss": 0.2665,
-      "step": 7620
-    },
-    {
-      "epoch": 9.56,
-      "learning_rate": 2.436647173489279e-06,
-      "loss": 0.287,
-      "step": 7630
-    },
-    {
-      "epoch": 9.57,
-      "learning_rate": 2.3670286828181563e-06,
-      "loss": 0.3368,
-      "step": 7640
-    },
-    {
-      "epoch": 9.59,
-      "learning_rate": 2.2974101921470343e-06,
-      "loss": 0.2389,
-      "step": 7650
-    },
-    {
-      "epoch": 9.6,
-      "learning_rate": 2.227791701475912e-06,
-      "loss": 0.2775,
-      "step": 7660
-    },
-    {
-      "epoch": 9.61,
-      "learning_rate": 2.1581732108047897e-06,
-      "loss": 0.2854,
-      "step": 7670
-    },
-    {
-      "epoch": 9.62,
-      "learning_rate": 2.0885547201336676e-06,
-      "loss": 0.2705,
-      "step": 7680
-    },
-    {
-      "epoch": 9.64,
-      "learning_rate": 2.018936229462545e-06,
-      "loss": 0.2666,
-      "step": 7690
-    },
-    {
-      "epoch": 9.65,
-      "learning_rate": 1.949317738791423e-06,
-      "loss": 0.2618,
-      "step": 7700
-    },
-    {
-      "epoch": 9.66,
-      "learning_rate": 1.8796992481203007e-06,
-      "loss": 0.2613,
-      "step": 7710
-    },
-    {
-      "epoch": 9.67,
-      "learning_rate": 1.8100807574491784e-06,
-      "loss": 0.2666,
-      "step": 7720
-    },
-    {
-      "epoch": 9.69,
-      "learning_rate": 1.7404622667780563e-06,
-      "loss": 0.2302,
-      "step": 7730
-    },
-    {
-      "epoch": 9.7,
-      "learning_rate": 1.670843776106934e-06,
-      "loss": 0.2484,
-      "step": 7740
-    },
-    {
-      "epoch": 9.71,
-      "learning_rate": 1.6012252854358117e-06,
-      "loss": 0.2419,
-      "step": 7750
-    },
-    {
-      "epoch": 9.72,
-      "learning_rate": 1.5316067947646894e-06,
-      "loss": 0.2479,
-      "step": 7760
-    },
-    {
-      "epoch": 9.74,
-      "learning_rate": 1.4619883040935671e-06,
-      "loss": 0.2976,
-      "step": 7770
-    },
-    {
-      "epoch": 9.75,
-      "learning_rate": 1.392369813422445e-06,
-      "loss": 0.2675,
-      "step": 7780
-    },
-    {
-      "epoch": 9.76,
-      "learning_rate": 1.3227513227513228e-06,
-      "loss": 0.2993,
-      "step": 7790
-    },
-    {
-      "epoch": 9.77,
-      "learning_rate": 1.2531328320802005e-06,
-      "loss": 0.2943,
-      "step": 7800
-    },
-    {
-      "epoch": 9.79,
-      "learning_rate": 1.1835143414090782e-06,
-      "loss": 0.3028,
-      "step": 7810
-    },
-    {
-      "epoch": 9.8,
-      "learning_rate": 1.113895850737956e-06,
-      "loss": 0.2926,
-      "step": 7820
-    },
-    {
-      "epoch": 9.81,
-      "learning_rate": 1.0442773600668338e-06,
-      "loss": 0.2558,
-      "step": 7830
-    },
-    {
-      "epoch": 9.82,
-      "learning_rate": 9.746588693957115e-07,
-      "loss": 0.2556,
-      "step": 7840
-    },
-    {
-      "epoch": 9.84,
-      "learning_rate": 9.050403787245892e-07,
-      "loss": 0.2531,
-      "step": 7850
-    },
-    {
-      "epoch": 9.85,
-      "learning_rate": 8.35421888053467e-07,
-      "loss": 0.2996,
-      "step": 7860
-    },
-    {
-      "epoch": 9.86,
-      "learning_rate": 7.658033973823447e-07,
-      "loss": 0.2825,
-      "step": 7870
-    },
-    {
-      "epoch": 9.87,
-      "learning_rate": 6.961849067112225e-07,
-      "loss": 0.2828,
-      "step": 7880
-    },
-    {
-      "epoch": 9.89,
-      "learning_rate": 6.265664160401002e-07,
-      "loss": 0.2608,
-      "step": 7890
-    },
-    {
-      "epoch": 9.9,
-      "learning_rate": 5.56947925368978e-07,
-      "loss": 0.2855,
-      "step": 7900
-    },
-    {
-      "epoch": 9.91,
-      "learning_rate": 4.873294346978557e-07,
-      "loss": 0.255,
-      "step": 7910
-    },
-    {
-      "epoch": 9.92,
-      "learning_rate": 4.177109440267335e-07,
-      "loss": 0.2583,
-      "step": 7920
-    },
-    {
-      "epoch": 9.94,
-      "learning_rate": 3.4809245335561126e-07,
-      "loss": 0.2945,
-      "step": 7930
-    },
-    {
-      "epoch": 9.95,
-      "learning_rate": 2.78473962684489e-07,
-      "loss": 0.2671,
-      "step": 7940
-    },
-    {
-      "epoch": 9.96,
-      "learning_rate": 2.0885547201336675e-07,
-      "loss": 0.2847,
-      "step": 7950
-    },
-    {
-      "epoch": 9.97,
-      "learning_rate": 1.392369813422445e-07,
-      "loss": 0.2756,
-      "step": 7960
-    },
-    {
-      "epoch": 9.99,
-      "learning_rate": 6.961849067112226e-08,
-      "loss": 0.2689,
-      "step": 7970
-    },
-    {
-      "epoch": 10.0,
-      "learning_rate": 0.0,
-      "loss": 0.2692,
-      "step": 7980
-    },
-    {
-      "epoch": 10.0,
-      "eval_accuracy": 0.8571428571428571,
-      "eval_loss": 0.4352613687515259,
-      "eval_runtime": 236.1141,
-      "eval_samples_per_second": 108.151,
-      "eval_steps_per_second": 3.38,
-      "step": 7980
-    },
-    {
-      "epoch": 10.0,
-      "step": 7980,
-      "total_flos": 7.915696500716863e+19,
-      "train_loss": 0.14746467811720712,
-      "train_runtime": 9966.943,
-      "train_samples_per_second": 102.483,
-      "train_steps_per_second": 0.801
+      "step": 3270,
+      "total_flos": 8.109125174606561e+18,
+      "train_loss": 0.5082513862064489,
+      "train_runtime": 2793.8499,
+      "train_samples_per_second": 37.468,
+      "train_steps_per_second": 1.17
     }
   ],
   "logging_steps": 10,
-  "max_steps": 7980,
+  "max_steps": 3270,
   "num_input_tokens_seen": 0,
-  "num_train_epochs": 10,
+  "num_train_epochs": 5,
   "save_steps": 500,
-  "total_flos": 7.915696500716863e+19,
-  "train_batch_size": 32,
+  "total_flos": 8.109125174606561e+18,
+  "train_batch_size": 8,
   "trial_name": null,
   "trial_params": null
 }