End of training

all_results.json

@@ -1,8 +1,13 @@
 {
     "epoch": 5.0,
-    "total_flos": 7.483912529986142e+18,
-    "train_loss": 1.0602939221811207,
-    "train_runtime": 5306.8618,
-    "train_samples_per_second": 66.23,
-    "train_steps_per_second": 0.517
+    "eval_accuracy": 0.9641475073981334,
+    "eval_loss": 0.1382417380809784,
+    "eval_runtime": 116.8179,
+    "eval_samples_per_second": 150.422,
+    "eval_steps_per_second": 4.708,
+    "total_flos": 7.485957524358513e+18,
+    "train_loss": 0.31472766021561754,
+    "train_runtime": 5227.0581,
+    "train_samples_per_second": 67.241,
+    "train_steps_per_second": 0.525
 }

eval_results.json

@@ -0,0 +1,8 @@
+{
+    "epoch": 5.0,
+    "eval_accuracy": 0.9641475073981334,
+    "eval_loss": 0.1382417380809784,
+    "eval_runtime": 116.8179,
+    "eval_samples_per_second": 150.422,
+    "eval_steps_per_second": 4.708
+}

runs/May01_11-37-02_f493e3a4fb8c/events.out.tfevents.1682952065.f493e3a4fb8c.8992.5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1a7c12ac19a8f9062a190764e2b623963f718b629993aa7a1f71ecfc544af009
+size 411

train_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 5.0,
-    "total_flos": 7.483912529986142e+18,
-    "train_loss": 1.0602939221811207,
-    "train_runtime": 5306.8618,
-    "train_samples_per_second": 66.23,
-    "train_steps_per_second": 0.517
+    "total_flos": 7.485957524358513e+18,
+    "train_loss": 0.31472766021561754,
+    "train_runtime": 5227.0581,
+    "train_samples_per_second": 67.241,
+    "train_steps_per_second": 0.525
 }

trainer_state.json

@@ -1,6 +1,6 @@
 {
-  "best_metric": 0.9624402458456636,
-  "best_model_checkpoint": "resnet-50-finetuned-eurosat/checkpoint-2745",
+  "best_metric": 0.9641475073981334,
+  "best_model_checkpoint": "resnet-50-finetuned-eurosat/checkpoint-1098",
   "epoch": 4.997724169321803,
   "global_step": 2745,
   "is_hyper_param_search": false,
@@ -9,1706 +9,1706 @@
   "log_history": [
     {
       "epoch": 0.02,
-      "learning_rate": 1.818181818181818e-06,
-      "loss": 3.6408,
+      "learning_rate": 0.0,
+      "loss": 0.3223,
       "step": 10
     },
     {
       "epoch": 0.04,
-      "learning_rate": 3.636363636363636e-06,
-      "loss": 3.6412,
+      "learning_rate": 0.0,
+      "loss": 0.3467,
       "step": 20
     },
     {
       "epoch": 0.05,
-      "learning_rate": 5.4545454545454545e-06,
-      "loss": 3.641,
+      "learning_rate": 0.0,
+      "loss": 0.2952,
       "step": 30
     },
     {
       "epoch": 0.07,
-      "learning_rate": 7.272727272727272e-06,
-      "loss": 3.6404,
+      "learning_rate": 0.0,
+      "loss": 0.3184,
       "step": 40
     },
     {
       "epoch": 0.09,
-      "learning_rate": 9.090909090909091e-06,
-      "loss": 3.6347,
+      "learning_rate": 0.0,
+      "loss": 0.3607,
       "step": 50
     },
     {
       "epoch": 0.11,
-      "learning_rate": 1.0909090909090909e-05,
-      "loss": 3.6377,
+      "learning_rate": 0.0,
+      "loss": 0.2938,
       "step": 60
     },
     {
       "epoch": 0.13,
-      "learning_rate": 1.2727272727272727e-05,
-      "loss": 3.6316,
+      "learning_rate": 0.0,
+      "loss": 0.3201,
       "step": 70
     },
     {
       "epoch": 0.15,
-      "learning_rate": 1.4545454545454545e-05,
-      "loss": 3.6339,
+      "learning_rate": 0.0,
+      "loss": 0.307,
       "step": 80
     },
     {
       "epoch": 0.16,
-      "learning_rate": 1.6363636363636366e-05,
-      "loss": 3.6301,
+      "learning_rate": 0.0,
+      "loss": 0.3133,
       "step": 90
     },
     {
       "epoch": 0.18,
-      "learning_rate": 1.8181818181818182e-05,
-      "loss": 3.626,
+      "learning_rate": 0.0,
+      "loss": 0.295,
       "step": 100
     },
     {
       "epoch": 0.2,
-      "learning_rate": 2e-05,
-      "loss": 3.621,
+      "learning_rate": 0.0,
+      "loss": 0.3321,
       "step": 110
     },
     {
       "epoch": 0.22,
-      "learning_rate": 2.1818181818181818e-05,
-      "loss": 3.6206,
+      "learning_rate": 0.0,
+      "loss": 0.3218,
       "step": 120
     },
     {
       "epoch": 0.24,
-      "learning_rate": 2.3636363636363637e-05,
-      "loss": 3.6118,
+      "learning_rate": 0.0,
+      "loss": 0.3313,
       "step": 130
     },
     {
       "epoch": 0.25,
-      "learning_rate": 2.5454545454545454e-05,
-      "loss": 3.6068,
+      "learning_rate": 0.0,
+      "loss": 0.3518,
       "step": 140
     },
     {
       "epoch": 0.27,
-      "learning_rate": 2.7272727272727273e-05,
-      "loss": 3.598,
+      "learning_rate": 0.0,
+      "loss": 0.319,
       "step": 150
     },
     {
       "epoch": 0.29,
-      "learning_rate": 2.909090909090909e-05,
-      "loss": 3.5991,
+      "learning_rate": 0.0,
+      "loss": 0.318,
       "step": 160
     },
     {
       "epoch": 0.31,
-      "learning_rate": 3.090909090909091e-05,
-      "loss": 3.5924,
+      "learning_rate": 0.0,
+      "loss": 0.3051,
       "step": 170
     },
     {
       "epoch": 0.33,
-      "learning_rate": 3.272727272727273e-05,
-      "loss": 3.5819,
+      "learning_rate": 0.0,
+      "loss": 0.304,
       "step": 180
     },
     {
       "epoch": 0.35,
-      "learning_rate": 3.454545454545455e-05,
-      "loss": 3.5726,
+      "learning_rate": 0.0,
+      "loss": 0.3102,
       "step": 190
     },
     {
       "epoch": 0.36,
-      "learning_rate": 3.6363636363636364e-05,
-      "loss": 3.5638,
+      "learning_rate": 0.0,
+      "loss": 0.2951,
       "step": 200
     },
     {
       "epoch": 0.38,
-      "learning_rate": 3.818181818181819e-05,
-      "loss": 3.5568,
+      "learning_rate": 0.0,
+      "loss": 0.3074,
       "step": 210
     },
     {
       "epoch": 0.4,
-      "learning_rate": 4e-05,
-      "loss": 3.5451,
+      "learning_rate": 0.0,
+      "loss": 0.3245,
       "step": 220
     },
     {
       "epoch": 0.42,
-      "learning_rate": 4.181818181818182e-05,
-      "loss": 3.5333,
+      "learning_rate": 0.0,
+      "loss": 0.3138,
       "step": 230
     },
     {
       "epoch": 0.44,
-      "learning_rate": 4.3636363636363636e-05,
-      "loss": 3.5256,
+      "learning_rate": 0.0,
+      "loss": 0.3479,
       "step": 240
     },
     {
       "epoch": 0.46,
-      "learning_rate": 4.545454545454546e-05,
-      "loss": 3.5079,
+      "learning_rate": 0.0,
+      "loss": 0.344,
       "step": 250
     },
     {
       "epoch": 0.47,
-      "learning_rate": 4.7272727272727275e-05,
-      "loss": 3.4865,
+      "learning_rate": 0.0,
+      "loss": 0.3239,
       "step": 260
     },
     {
       "epoch": 0.49,
-      "learning_rate": 4.909090909090909e-05,
-      "loss": 3.4667,
+      "learning_rate": 0.0,
+      "loss": 0.3223,
       "step": 270
     },
     {
       "epoch": 0.51,
-      "learning_rate": 4.989878542510122e-05,
-      "loss": 3.4516,
+      "learning_rate": 0.0,
+      "loss": 0.3292,
       "step": 280
     },
     {
       "epoch": 0.53,
-      "learning_rate": 4.9696356275303645e-05,
-      "loss": 3.427,
+      "learning_rate": 0.0,
+      "loss": 0.2866,
       "step": 290
     },
     {
       "epoch": 0.55,
-      "learning_rate": 4.9493927125506076e-05,
-      "loss": 3.3999,
+      "learning_rate": 0.0,
+      "loss": 0.3019,
       "step": 300
     },
     {
       "epoch": 0.56,
-      "learning_rate": 4.9291497975708506e-05,
-      "loss": 3.3732,
+      "learning_rate": 0.0,
+      "loss": 0.3334,
       "step": 310
     },
     {
       "epoch": 0.58,
-      "learning_rate": 4.908906882591093e-05,
-      "loss": 3.3437,
+      "learning_rate": 0.0,
+      "loss": 0.2834,
       "step": 320
     },
     {
       "epoch": 0.6,
-      "learning_rate": 4.888663967611336e-05,
-      "loss": 3.2873,
+      "learning_rate": 0.0,
+      "loss": 0.3198,
       "step": 330
     },
     {
       "epoch": 0.62,
-      "learning_rate": 4.868421052631579e-05,
-      "loss": 3.2593,
+      "learning_rate": 0.0,
+      "loss": 0.3173,
       "step": 340
     },
     {
       "epoch": 0.64,
-      "learning_rate": 4.848178137651822e-05,
-      "loss": 3.2006,
+      "learning_rate": 0.0,
+      "loss": 0.3495,
       "step": 350
     },
     {
       "epoch": 0.66,
-      "learning_rate": 4.8279352226720646e-05,
-      "loss": 3.1593,
+      "learning_rate": 0.0,
+      "loss": 0.3267,
       "step": 360
     },
     {
       "epoch": 0.67,
-      "learning_rate": 4.8076923076923084e-05,
-      "loss": 3.1129,
+      "learning_rate": 0.0,
+      "loss": 0.3229,
       "step": 370
     },
     {
       "epoch": 0.69,
-      "learning_rate": 4.787449392712551e-05,
-      "loss": 3.0435,
+      "learning_rate": 0.0,
+      "loss": 0.3154,
       "step": 380
     },
     {
       "epoch": 0.71,
-      "learning_rate": 4.767206477732794e-05,
-      "loss": 2.9742,
+      "learning_rate": 0.0,
+      "loss": 0.2958,
       "step": 390
     },
     {
       "epoch": 0.73,
-      "learning_rate": 4.746963562753037e-05,
-      "loss": 2.9306,
+      "learning_rate": 0.0,
+      "loss": 0.3068,
       "step": 400
     },
     {
       "epoch": 0.75,
-      "learning_rate": 4.726720647773279e-05,
-      "loss": 2.8627,
+      "learning_rate": 0.0,
+      "loss": 0.3443,
       "step": 410
     },
     {
       "epoch": 0.76,
-      "learning_rate": 4.7064777327935223e-05,
-      "loss": 2.8301,
+      "learning_rate": 0.0,
+      "loss": 0.3371,
       "step": 420
     },
     {
       "epoch": 0.78,
-      "learning_rate": 4.6862348178137654e-05,
-      "loss": 2.7201,
+      "learning_rate": 0.0,
+      "loss": 0.3311,
       "step": 430
     },
     {
       "epoch": 0.8,
-      "learning_rate": 4.6659919028340085e-05,
-      "loss": 2.6538,
+      "learning_rate": 0.0,
+      "loss": 0.3345,
       "step": 440
     },
     {
       "epoch": 0.82,
-      "learning_rate": 4.6457489878542516e-05,
-      "loss": 2.5858,
+      "learning_rate": 0.0,
+      "loss": 0.2798,
       "step": 450
     },
     {
       "epoch": 0.84,
-      "learning_rate": 4.625506072874494e-05,
-      "loss": 2.4747,
+      "learning_rate": 0.0,
+      "loss": 0.2834,
       "step": 460
     },
     {
       "epoch": 0.86,
-      "learning_rate": 4.605263157894737e-05,
-      "loss": 2.41,
+      "learning_rate": 0.0,
+      "loss": 0.3295,
       "step": 470
     },
     {
       "epoch": 0.87,
-      "learning_rate": 4.58502024291498e-05,
-      "loss": 2.366,
+      "learning_rate": 0.0,
+      "loss": 0.3176,
       "step": 480
     },
     {
       "epoch": 0.89,
-      "learning_rate": 4.564777327935223e-05,
-      "loss": 2.3017,
+      "learning_rate": 0.0,
+      "loss": 0.281,
       "step": 490
     },
     {
       "epoch": 0.91,
-      "learning_rate": 4.5445344129554655e-05,
-      "loss": 2.2318,
+      "learning_rate": 0.0,
+      "loss": 0.3066,
       "step": 500
     },
     {
       "epoch": 0.93,
-      "learning_rate": 4.524291497975709e-05,
-      "loss": 2.1774,
+      "learning_rate": 0.0,
+      "loss": 0.2898,
       "step": 510
     },
     {
       "epoch": 0.95,
-      "learning_rate": 4.504048582995952e-05,
-      "loss": 2.089,
+      "learning_rate": 0.0,
+      "loss": 0.307,
       "step": 520
     },
     {
       "epoch": 0.96,
-      "learning_rate": 4.483805668016194e-05,
-      "loss": 2.0213,
+      "learning_rate": 0.0,
+      "loss": 0.3126,
       "step": 530
     },
     {
       "epoch": 0.98,
-      "learning_rate": 4.463562753036438e-05,
-      "loss": 1.9386,
+      "learning_rate": 0.0,
+      "loss": 0.2976,
       "step": 540
     },
     {
       "epoch": 1.0,
-      "eval_accuracy": 0.7653084452538129,
-      "eval_loss": 1.521803379058838,
-      "eval_runtime": 116.2062,
-      "eval_samples_per_second": 151.214,
-      "eval_steps_per_second": 4.733,
+      "eval_accuracy": 0.9635784202139768,
+      "eval_loss": 0.1450216919183731,
+      "eval_runtime": 111.4659,
+      "eval_samples_per_second": 157.645,
+      "eval_steps_per_second": 4.934,
       "step": 549
     },
     {
       "epoch": 1.0,
-      "learning_rate": 4.44331983805668e-05,
-      "loss": 1.9215,
+      "learning_rate": 0.0,
+      "loss": 0.3563,
       "step": 550
     },
     {
       "epoch": 1.02,
-      "learning_rate": 4.423076923076923e-05,
-      "loss": 1.8288,
+      "learning_rate": 0.0,
+      "loss": 0.3112,
       "step": 560
     },
     {
       "epoch": 1.04,
-      "learning_rate": 4.402834008097166e-05,
-      "loss": 1.761,
+      "learning_rate": 0.0,
+      "loss": 0.3078,
       "step": 570
     },
     {
       "epoch": 1.06,
-      "learning_rate": 4.3825910931174094e-05,
-      "loss": 1.7306,
+      "learning_rate": 0.0,
+      "loss": 0.2967,
       "step": 580
     },
     {
       "epoch": 1.07,
-      "learning_rate": 4.362348178137652e-05,
-      "loss": 1.6337,
+      "learning_rate": 0.0,
+      "loss": 0.2955,
       "step": 590
     },
     {
       "epoch": 1.09,
-      "learning_rate": 4.342105263157895e-05,
-      "loss": 1.5869,
+      "learning_rate": 0.0,
+      "loss": 0.3216,
       "step": 600
     },
     {
       "epoch": 1.11,
-      "learning_rate": 4.321862348178138e-05,
-      "loss": 1.5347,
+      "learning_rate": 0.0,
+      "loss": 0.3227,
       "step": 610
     },
     {
       "epoch": 1.13,
-      "learning_rate": 4.30161943319838e-05,
-      "loss": 1.4658,
+      "learning_rate": 0.0,
+      "loss": 0.3164,
       "step": 620
     },
     {
       "epoch": 1.15,
-      "learning_rate": 4.281376518218624e-05,
-      "loss": 1.444,
+      "learning_rate": 0.0,
+      "loss": 0.3181,
       "step": 630
     },
     {
       "epoch": 1.17,
-      "learning_rate": 4.2611336032388664e-05,
-      "loss": 1.3742,
+      "learning_rate": 0.0,
+      "loss": 0.2877,
       "step": 640
     },
     {
       "epoch": 1.18,
-      "learning_rate": 4.2408906882591095e-05,
-      "loss": 1.3474,
+      "learning_rate": 0.0,
+      "loss": 0.2977,
       "step": 650
     },
     {
       "epoch": 1.2,
-      "learning_rate": 4.2206477732793526e-05,
-      "loss": 1.3503,
+      "learning_rate": 0.0,
+      "loss": 0.3257,
       "step": 660
     },
     {
       "epoch": 1.22,
-      "learning_rate": 4.200404858299595e-05,
-      "loss": 1.2557,
+      "learning_rate": 0.0,
+      "loss": 0.3415,
       "step": 670
     },
     {
       "epoch": 1.24,
-      "learning_rate": 4.180161943319838e-05,
-      "loss": 1.2054,
+      "learning_rate": 0.0,
+      "loss": 0.3257,
       "step": 680
     },
     {
       "epoch": 1.26,
-      "learning_rate": 4.159919028340081e-05,
-      "loss": 1.253,
+      "learning_rate": 0.0,
+      "loss": 0.3145,
       "step": 690
     },
     {
       "epoch": 1.27,
-      "learning_rate": 4.139676113360324e-05,
-      "loss": 1.1814,
+      "learning_rate": 0.0,
+      "loss": 0.3626,
       "step": 700
     },
     {
       "epoch": 1.29,
-      "learning_rate": 4.1194331983805666e-05,
-      "loss": 1.1699,
+      "learning_rate": 0.0,
+      "loss": 0.3251,
       "step": 710
     },
     {
       "epoch": 1.31,
-      "learning_rate": 4.09919028340081e-05,
-      "loss": 1.1464,
+      "learning_rate": 0.0,
+      "loss": 0.3272,
       "step": 720
     },
     {
       "epoch": 1.33,
-      "learning_rate": 4.078947368421053e-05,
-      "loss": 1.1289,
+      "learning_rate": 0.0,
+      "loss": 0.315,
       "step": 730
     },
     {
       "epoch": 1.35,
-      "learning_rate": 4.058704453441296e-05,
-      "loss": 1.0347,
+      "learning_rate": 0.0,
+      "loss": 0.318,
       "step": 740
     },
     {
       "epoch": 1.37,
-      "learning_rate": 4.038461538461539e-05,
-      "loss": 1.0617,
+      "learning_rate": 0.0,
+      "loss": 0.3067,
       "step": 750
     },
     {
       "epoch": 1.38,
-      "learning_rate": 4.018218623481781e-05,
-      "loss": 1.0069,
+      "learning_rate": 0.0,
+      "loss": 0.2829,
       "step": 760
     },
     {
       "epoch": 1.4,
-      "learning_rate": 3.997975708502025e-05,
-      "loss": 0.9519,
+      "learning_rate": 0.0,
+      "loss": 0.2894,
       "step": 770
     },
     {
       "epoch": 1.42,
-      "learning_rate": 3.9777327935222674e-05,
-      "loss": 0.9229,
+      "learning_rate": 0.0,
+      "loss": 0.3045,
       "step": 780
     },
     {
       "epoch": 1.44,
-      "learning_rate": 3.9574898785425104e-05,
-      "loss": 0.9309,
+      "learning_rate": 0.0,
+      "loss": 0.3003,
       "step": 790
     },
     {
       "epoch": 1.46,
-      "learning_rate": 3.9372469635627535e-05,
-      "loss": 0.8647,
+      "learning_rate": 0.0,
+      "loss": 0.2724,
       "step": 800
     },
     {
       "epoch": 1.47,
-      "learning_rate": 3.917004048582996e-05,
-      "loss": 0.8625,
+      "learning_rate": 0.0,
+      "loss": 0.2818,
       "step": 810
     },
     {
       "epoch": 1.49,
-      "learning_rate": 3.896761133603239e-05,
-      "loss": 0.9097,
+      "learning_rate": 0.0,
+      "loss": 0.2925,
       "step": 820
     },
     {
       "epoch": 1.51,
-      "learning_rate": 3.876518218623482e-05,
-      "loss": 0.8751,
+      "learning_rate": 0.0,
+      "loss": 0.3095,
       "step": 830
     },
     {
       "epoch": 1.53,
-      "learning_rate": 3.856275303643725e-05,
-      "loss": 0.8613,
+      "learning_rate": 0.0,
+      "loss": 0.3079,
       "step": 840
     },
     {
       "epoch": 1.55,
-      "learning_rate": 3.8360323886639675e-05,
-      "loss": 0.8289,
+      "learning_rate": 0.0,
+      "loss": 0.3226,
       "step": 850
     },
     {
       "epoch": 1.57,
-      "learning_rate": 3.815789473684211e-05,
-      "loss": 0.8145,
+      "learning_rate": 0.0,
+      "loss": 0.3346,
       "step": 860
     },
     {
       "epoch": 1.58,
-      "learning_rate": 3.7955465587044536e-05,
-      "loss": 0.8031,
+      "learning_rate": 0.0,
+      "loss": 0.3343,
       "step": 870
     },
     {
       "epoch": 1.6,
-      "learning_rate": 3.775303643724697e-05,
-      "loss": 0.844,
+      "learning_rate": 0.0,
+      "loss": 0.2956,
       "step": 880
     },
     {
       "epoch": 1.62,
-      "learning_rate": 3.75506072874494e-05,
-      "loss": 0.8124,
+      "learning_rate": 0.0,
+      "loss": 0.3006,
       "step": 890
     },
     {
       "epoch": 1.64,
-      "learning_rate": 3.734817813765182e-05,
-      "loss": 0.7609,
+      "learning_rate": 0.0,
+      "loss": 0.3165,
       "step": 900
     },
     {
       "epoch": 1.66,
-      "learning_rate": 3.714574898785425e-05,
-      "loss": 0.7603,
+      "learning_rate": 0.0,
+      "loss": 0.3185,
       "step": 910
     },
     {
       "epoch": 1.68,
-      "learning_rate": 3.694331983805668e-05,
-      "loss": 0.7959,
+      "learning_rate": 0.0,
+      "loss": 0.2932,
       "step": 920
     },
     {
       "epoch": 1.69,
-      "learning_rate": 3.6740890688259113e-05,
-      "loss": 0.719,
+      "learning_rate": 0.0,
+      "loss": 0.3233,
       "step": 930
     },
     {
       "epoch": 1.71,
-      "learning_rate": 3.653846153846154e-05,
-      "loss": 0.7447,
+      "learning_rate": 0.0,
+      "loss": 0.2804,
       "step": 940
     },
     {
       "epoch": 1.73,
-      "learning_rate": 3.633603238866397e-05,
-      "loss": 0.7588,
+      "learning_rate": 0.0,
+      "loss": 0.3244,
       "step": 950
     },
     {
       "epoch": 1.75,
-      "learning_rate": 3.61336032388664e-05,
-      "loss": 0.6941,
+      "learning_rate": 0.0,
+      "loss": 0.3166,
       "step": 960
     },
     {
       "epoch": 1.77,
-      "learning_rate": 3.593117408906882e-05,
-      "loss": 0.709,
+      "learning_rate": 0.0,
+      "loss": 0.2889,
       "step": 970
     },
     {
       "epoch": 1.78,
-      "learning_rate": 3.572874493927126e-05,
-      "loss": 0.7446,
+      "learning_rate": 0.0,
+      "loss": 0.3084,
       "step": 980
     },
     {
       "epoch": 1.8,
-      "learning_rate": 3.5526315789473684e-05,
-      "loss": 0.6729,
+      "learning_rate": 0.0,
+      "loss": 0.3083,
       "step": 990
     },
     {
       "epoch": 1.82,
-      "learning_rate": 3.5323886639676115e-05,
-      "loss": 0.6262,
+      "learning_rate": 0.0,
+      "loss": 0.3597,
       "step": 1000
     },
     {
       "epoch": 1.84,
-      "learning_rate": 3.5121457489878545e-05,
-      "loss": 0.6725,
+      "learning_rate": 0.0,
+      "loss": 0.3192,
       "step": 1010
     },
     {
       "epoch": 1.86,
-      "learning_rate": 3.491902834008097e-05,
-      "loss": 0.6287,
+      "learning_rate": 0.0,
+      "loss": 0.3044,
       "step": 1020
     },
     {
       "epoch": 1.88,
-      "learning_rate": 3.47165991902834e-05,
-      "loss": 0.6443,
+      "learning_rate": 0.0,
+      "loss": 0.3149,
       "step": 1030
     },
     {
       "epoch": 1.89,
-      "learning_rate": 3.451417004048583e-05,
-      "loss": 0.6444,
+      "learning_rate": 0.0,
+      "loss": 0.3192,
       "step": 1040
     },
     {
       "epoch": 1.91,
-      "learning_rate": 3.431174089068826e-05,
-      "loss": 0.6219,
+      "learning_rate": 0.0,
+      "loss": 0.2853,
       "step": 1050
     },
     {
       "epoch": 1.93,
-      "learning_rate": 3.410931174089069e-05,
-      "loss": 0.6271,
+      "learning_rate": 0.0,
+      "loss": 0.3255,
       "step": 1060
     },
     {
       "epoch": 1.95,
-      "learning_rate": 3.390688259109312e-05,
-      "loss": 0.6186,
+      "learning_rate": 0.0,
+      "loss": 0.3079,
       "step": 1070
     },
     {
       "epoch": 1.97,
-      "learning_rate": 3.3704453441295546e-05,
-      "loss": 0.5921,
+      "learning_rate": 0.0,
+      "loss": 0.2867,
       "step": 1080
     },
     {
       "epoch": 1.98,
-      "learning_rate": 3.350202429149798e-05,
-      "loss": 0.5662,
+      "learning_rate": 0.0,
+      "loss": 0.3388,
       "step": 1090
     },
     {
       "epoch": 2.0,
-      "eval_accuracy": 0.9288071932620078,
-      "eval_loss": 0.33246490359306335,
-      "eval_runtime": 116.3356,
-      "eval_samples_per_second": 151.046,
-      "eval_steps_per_second": 4.728,
+      "eval_accuracy": 0.9641475073981334,
+      "eval_loss": 0.1382417380809784,
+      "eval_runtime": 112.0412,
+      "eval_samples_per_second": 156.835,
+      "eval_steps_per_second": 4.909,
       "step": 1098
     },
     {
       "epoch": 2.0,
-      "learning_rate": 3.329959514170041e-05,
-      "loss": 0.6063,
+      "learning_rate": 0.0,
+      "loss": 0.3561,
       "step": 1100
     },
     {
       "epoch": 2.02,
-      "learning_rate": 3.309716599190283e-05,
-      "loss": 0.5733,
+      "learning_rate": 0.0,
+      "loss": 0.296,
       "step": 1110
     },
     {
       "epoch": 2.04,
-      "learning_rate": 3.289473684210527e-05,
-      "loss": 0.5662,
+      "learning_rate": 0.0,
+      "loss": 0.2846,
       "step": 1120
     },
     {
       "epoch": 2.06,
-      "learning_rate": 3.269230769230769e-05,
-      "loss": 0.5893,
+      "learning_rate": 0.0,
+      "loss": 0.3364,
       "step": 1130
     },
     {
       "epoch": 2.08,
-      "learning_rate": 3.2489878542510124e-05,
-      "loss": 0.5498,
+      "learning_rate": 0.0,
+      "loss": 0.3638,
       "step": 1140
     },
     {
       "epoch": 2.09,
-      "learning_rate": 3.2287449392712554e-05,
-      "loss": 0.5416,
+      "learning_rate": 0.0,
+      "loss": 0.3259,
       "step": 1150
     },
     {
       "epoch": 2.11,
-      "learning_rate": 3.208502024291498e-05,
-      "loss": 0.5271,
+      "learning_rate": 0.0,
+      "loss": 0.316,
       "step": 1160
     },
     {
       "epoch": 2.13,
-      "learning_rate": 3.188259109311741e-05,
-      "loss": 0.5506,
+      "learning_rate": 0.0,
+      "loss": 0.3298,
       "step": 1170
     },
     {
       "epoch": 2.15,
-      "learning_rate": 3.168016194331984e-05,
-      "loss": 0.5227,
+      "learning_rate": 0.0,
+      "loss": 0.3135,
       "step": 1180
     },
     {
       "epoch": 2.17,
-      "learning_rate": 3.147773279352227e-05,
-      "loss": 0.5379,
+      "learning_rate": 0.0,
+      "loss": 0.3031,
       "step": 1190
     },
     {
       "epoch": 2.18,
-      "learning_rate": 3.1275303643724694e-05,
-      "loss": 0.4872,
+      "learning_rate": 0.0,
+      "loss": 0.2982,
       "step": 1200
     },
     {
       "epoch": 2.2,
-      "learning_rate": 3.107287449392713e-05,
-      "loss": 0.5166,
+      "learning_rate": 0.0,
+      "loss": 0.3074,
       "step": 1210
     },
     {
       "epoch": 2.22,
-      "learning_rate": 3.0870445344129556e-05,
-      "loss": 0.5416,
+      "learning_rate": 0.0,
+      "loss": 0.3465,
       "step": 1220
     },
     {
       "epoch": 2.24,
-      "learning_rate": 3.0668016194331986e-05,
-      "loss": 0.5171,
+      "learning_rate": 0.0,
+      "loss": 0.3148,
       "step": 1230
     },
     {
       "epoch": 2.26,
-      "learning_rate": 3.0465587044534417e-05,
-      "loss": 0.4827,
+      "learning_rate": 0.0,
+      "loss": 0.3406,
       "step": 1240
     },
     {
       "epoch": 2.28,
-      "learning_rate": 3.0263157894736844e-05,
-      "loss": 0.5535,
+      "learning_rate": 0.0,
+      "loss": 0.2878,
       "step": 1250
     },
     {
       "epoch": 2.29,
-      "learning_rate": 3.006072874493927e-05,
-      "loss": 0.5095,
+      "learning_rate": 0.0,
+      "loss": 0.3026,
       "step": 1260
     },
     {
       "epoch": 2.31,
-      "learning_rate": 2.9858299595141702e-05,
-      "loss": 0.4874,
+      "learning_rate": 0.0,
+      "loss": 0.3189,
       "step": 1270
     },
     {
       "epoch": 2.33,
-      "learning_rate": 2.965587044534413e-05,
-      "loss": 0.4798,
+      "learning_rate": 0.0,
+      "loss": 0.312,
       "step": 1280
     },
     {
       "epoch": 2.35,
-      "learning_rate": 2.9453441295546557e-05,
-      "loss": 0.487,
+      "learning_rate": 0.0,
+      "loss": 0.3393,
       "step": 1290
     },
     {
       "epoch": 2.37,
-      "learning_rate": 2.925101214574899e-05,
-      "loss": 0.4855,
+      "learning_rate": 0.0,
+      "loss": 0.3389,
       "step": 1300
     },
     {
       "epoch": 2.39,
-      "learning_rate": 2.9048582995951418e-05,
-      "loss": 0.4818,
+      "learning_rate": 0.0,
+      "loss": 0.3073,
       "step": 1310
     },
     {
       "epoch": 2.4,
-      "learning_rate": 2.8846153846153845e-05,
-      "loss": 0.4359,
+      "learning_rate": 0.0,
+      "loss": 0.3525,
       "step": 1320
     },
     {
       "epoch": 2.42,
-      "learning_rate": 2.8643724696356276e-05,
-      "loss": 0.4553,
+      "learning_rate": 0.0,
+      "loss": 0.2958,
       "step": 1330
     },
     {
       "epoch": 2.44,
-      "learning_rate": 2.8441295546558703e-05,
-      "loss": 0.4566,
+      "learning_rate": 0.0,
+      "loss": 0.2998,
       "step": 1340
     },
     {
       "epoch": 2.46,
-      "learning_rate": 2.823886639676113e-05,
-      "loss": 0.468,
+      "learning_rate": 0.0,
+      "loss": 0.2992,
       "step": 1350
     },
     {
       "epoch": 2.48,
-      "learning_rate": 2.8036437246963565e-05,
-      "loss": 0.4448,
+      "learning_rate": 0.0,
+      "loss": 0.314,
       "step": 1360
     },
     {
       "epoch": 2.49,
-      "learning_rate": 2.7834008097165992e-05,
-      "loss": 0.4629,
+      "learning_rate": 0.0,
+      "loss": 0.3128,
       "step": 1370
     },
     {
       "epoch": 2.51,
-      "learning_rate": 2.7631578947368426e-05,
-      "loss": 0.4876,
+      "learning_rate": 0.0,
+      "loss": 0.3074,
       "step": 1380
     },
     {
       "epoch": 2.53,
-      "learning_rate": 2.7429149797570853e-05,
-      "loss": 0.4499,
+      "learning_rate": 0.0,
+      "loss": 0.3189,
       "step": 1390
     },
     {
       "epoch": 2.55,
-      "learning_rate": 2.722672064777328e-05,
-      "loss": 0.4449,
+      "learning_rate": 0.0,
+      "loss": 0.3284,
       "step": 1400
     },
     {
       "epoch": 2.57,
-      "learning_rate": 2.702429149797571e-05,
-      "loss": 0.404,
+      "learning_rate": 0.0,
+      "loss": 0.3278,
       "step": 1410
     },
     {
       "epoch": 2.59,
-      "learning_rate": 2.682186234817814e-05,
-      "loss": 0.4521,
+      "learning_rate": 0.0,
+      "loss": 0.2897,
       "step": 1420
     },
     {
       "epoch": 2.6,
-      "learning_rate": 2.6619433198380566e-05,
-      "loss": 0.4075,
+      "learning_rate": 0.0,
+      "loss": 0.3472,
       "step": 1430
     },
     {
       "epoch": 2.62,
-      "learning_rate": 2.6417004048583e-05,
-      "loss": 0.4528,
+      "learning_rate": 0.0,
+      "loss": 0.3571,
       "step": 1440
     },
     {
       "epoch": 2.64,
-      "learning_rate": 2.6214574898785427e-05,
-      "loss": 0.4615,
+      "learning_rate": 0.0,
+      "loss": 0.3078,
       "step": 1450
     },
     {
       "epoch": 2.66,
-      "learning_rate": 2.6012145748987855e-05,
-      "loss": 0.4654,
+      "learning_rate": 0.0,
+      "loss": 0.3512,
       "step": 1460
     },
     {
       "epoch": 2.68,
-      "learning_rate": 2.5809716599190285e-05,
-      "loss": 0.425,
+      "learning_rate": 0.0,
+      "loss": 0.3017,
       "step": 1470
     },
     {
       "epoch": 2.69,
-      "learning_rate": 2.5607287449392713e-05,
-      "loss": 0.409,
+      "learning_rate": 0.0,
+      "loss": 0.302,
       "step": 1480
     },
     {
       "epoch": 2.71,
-      "learning_rate": 2.540485829959514e-05,
-      "loss": 0.4284,
+      "learning_rate": 0.0,
+      "loss": 0.3281,
       "step": 1490
     },
     {
       "epoch": 2.73,
-      "learning_rate": 2.5202429149797574e-05,
-      "loss": 0.4122,
+      "learning_rate": 0.0,
+      "loss": 0.3511,
       "step": 1500
     },
     {
       "epoch": 2.75,
-      "learning_rate": 2.5e-05,
-      "loss": 0.4273,
+      "learning_rate": 0.0,
+      "loss": 0.3088,
       "step": 1510
     },
     {
       "epoch": 2.77,
-      "learning_rate": 2.4797570850202432e-05,
-      "loss": 0.3865,
+      "learning_rate": 0.0,
+      "loss": 0.2802,
       "step": 1520
     },
     {
       "epoch": 2.79,
-      "learning_rate": 2.459514170040486e-05,
-      "loss": 0.378,
+      "learning_rate": 0.0,
+      "loss": 0.304,
       "step": 1530
     },
     {
       "epoch": 2.8,
-      "learning_rate": 2.439271255060729e-05,
-      "loss": 0.4051,
+      "learning_rate": 0.0,
+      "loss": 0.3119,
       "step": 1540
     },
     {
       "epoch": 2.82,
-      "learning_rate": 2.4190283400809717e-05,
-      "loss": 0.4127,
+      "learning_rate": 0.0,
+      "loss": 0.3194,
       "step": 1550
     },
     {
       "epoch": 2.84,
-      "learning_rate": 2.3987854251012144e-05,
-      "loss": 0.4141,
+      "learning_rate": 0.0,
+      "loss": 0.3028,
       "step": 1560
     },
     {
       "epoch": 2.86,
-      "learning_rate": 2.3785425101214575e-05,
-      "loss": 0.3842,
+      "learning_rate": 0.0,
+      "loss": 0.2849,
       "step": 1570
     },
     {
       "epoch": 2.88,
-      "learning_rate": 2.3582995951417006e-05,
-      "loss": 0.423,
+      "learning_rate": 0.0,
+      "loss": 0.2897,
       "step": 1580
     },
     {
       "epoch": 2.89,
-      "learning_rate": 2.3380566801619436e-05,
-      "loss": 0.3714,
+      "learning_rate": 0.0,
+      "loss": 0.309,
       "step": 1590
     },
     {
       "epoch": 2.91,
-      "learning_rate": 2.3178137651821864e-05,
-      "loss": 0.3949,
+      "learning_rate": 0.0,
+      "loss": 0.3223,
       "step": 1600
     },
     {
       "epoch": 2.93,
-      "learning_rate": 2.2975708502024294e-05,
-      "loss": 0.398,
+      "learning_rate": 0.0,
+      "loss": 0.3157,
       "step": 1610
     },
     {
       "epoch": 2.95,
-      "learning_rate": 2.2773279352226722e-05,
-      "loss": 0.3838,
+      "learning_rate": 0.0,
+      "loss": 0.3112,
       "step": 1620
     },
     {
       "epoch": 2.97,
-      "learning_rate": 2.257085020242915e-05,
-      "loss": 0.365,
+      "learning_rate": 0.0,
+      "loss": 0.2911,
       "step": 1630
     },
     {
       "epoch": 2.99,
-      "learning_rate": 2.236842105263158e-05,
-      "loss": 0.4104,
+      "learning_rate": 0.0,
+      "loss": 0.361,
       "step": 1640
     },
     {
       "epoch": 3.0,
-      "eval_accuracy": 0.9561233781015251,
-      "eval_loss": 0.18532179296016693,
-      "eval_runtime": 113.1925,
-      "eval_samples_per_second": 155.24,
-      "eval_steps_per_second": 4.859,
+      "eval_accuracy": 0.9631800591850671,
+      "eval_loss": 0.14322593808174133,
+      "eval_runtime": 112.3273,
+      "eval_samples_per_second": 156.436,
+      "eval_steps_per_second": 4.896,
       "step": 1647
     },
     {
       "epoch": 3.0,
-      "learning_rate": 2.216599190283401e-05,
-      "loss": 0.4087,
+      "learning_rate": 0.0,
+      "loss": 0.3124,
       "step": 1650
     },
     {
       "epoch": 3.02,
-      "learning_rate": 2.1963562753036438e-05,
-      "loss": 0.3953,
+      "learning_rate": 0.0,
+      "loss": 0.3007,
       "step": 1660
     },
     {
       "epoch": 3.04,
-      "learning_rate": 2.1761133603238868e-05,
-      "loss": 0.3972,
+      "learning_rate": 0.0,
+      "loss": 0.3119,
       "step": 1670
     },
     {
       "epoch": 3.06,
-      "learning_rate": 2.15587044534413e-05,
-      "loss": 0.4105,
+      "learning_rate": 0.0,
+      "loss": 0.2744,
       "step": 1680
     },
     {
       "epoch": 3.08,
-      "learning_rate": 2.1356275303643726e-05,
-      "loss": 0.3761,
+      "learning_rate": 0.0,
+      "loss": 0.3164,
       "step": 1690
     },
     {
       "epoch": 3.1,
-      "learning_rate": 2.1153846153846154e-05,
-      "loss": 0.3555,
+      "learning_rate": 0.0,
+      "loss": 0.336,
       "step": 1700
     },
     {
       "epoch": 3.11,
-      "learning_rate": 2.0951417004048584e-05,
-      "loss": 0.3801,
+      "learning_rate": 0.0,
+      "loss": 0.3423,
       "step": 1710
     },
     {
       "epoch": 3.13,
-      "learning_rate": 2.074898785425101e-05,
-      "loss": 0.402,
+      "learning_rate": 0.0,
+      "loss": 0.347,
       "step": 1720
     },
     {
       "epoch": 3.15,
-      "learning_rate": 2.0546558704453442e-05,
-      "loss": 0.3929,
+      "learning_rate": 0.0,
+      "loss": 0.3064,
       "step": 1730
     },
     {
       "epoch": 3.17,
-      "learning_rate": 2.0344129554655873e-05,
-      "loss": 0.3884,
+      "learning_rate": 0.0,
+      "loss": 0.3241,
       "step": 1740
     },
     {
       "epoch": 3.19,
-      "learning_rate": 2.0141700404858304e-05,
-      "loss": 0.3853,
+      "learning_rate": 0.0,
+      "loss": 0.3381,
       "step": 1750
     },
     {
       "epoch": 3.2,
-      "learning_rate": 1.9939271255060727e-05,
-      "loss": 0.3869,
+      "learning_rate": 0.0,
+      "loss": 0.2719,
       "step": 1760
     },
     {
       "epoch": 3.22,
-      "learning_rate": 1.9736842105263158e-05,
-      "loss": 0.3758,
+      "learning_rate": 0.0,
+      "loss": 0.3141,
       "step": 1770
     },
     {
       "epoch": 3.24,
-      "learning_rate": 1.953441295546559e-05,
-      "loss": 0.3961,
+      "learning_rate": 0.0,
+      "loss": 0.3369,
       "step": 1780
     },
     {
       "epoch": 3.26,
-      "learning_rate": 1.9331983805668016e-05,
-      "loss": 0.3389,
+      "learning_rate": 0.0,
+      "loss": 0.2892,
       "step": 1790
     },
     {
       "epoch": 3.28,
-      "learning_rate": 1.9129554655870447e-05,
-      "loss": 0.3624,
+      "learning_rate": 0.0,
+      "loss": 0.3228,
       "step": 1800
     },
     {
       "epoch": 3.3,
-      "learning_rate": 1.8927125506072877e-05,
-      "loss": 0.3905,
+      "learning_rate": 0.0,
+      "loss": 0.3339,
       "step": 1810
     },
     {
       "epoch": 3.31,
-      "learning_rate": 1.8724696356275305e-05,
-      "loss": 0.3685,
+      "learning_rate": 0.0,
+      "loss": 0.2774,
       "step": 1820
     },
     {
       "epoch": 3.33,
-      "learning_rate": 1.8522267206477732e-05,
-      "loss": 0.3582,
+      "learning_rate": 0.0,
+      "loss": 0.3103,
       "step": 1830
     },
     {
       "epoch": 3.35,
-      "learning_rate": 1.8319838056680163e-05,
-      "loss": 0.3785,
+      "learning_rate": 0.0,
+      "loss": 0.2864,
       "step": 1840
     },
     {
       "epoch": 3.37,
-      "learning_rate": 1.811740890688259e-05,
-      "loss": 0.3758,
+      "learning_rate": 0.0,
+      "loss": 0.3264,
       "step": 1850
     },
     {
       "epoch": 3.39,
-      "learning_rate": 1.791497975708502e-05,
-      "loss": 0.386,
+      "learning_rate": 0.0,
+      "loss": 0.3145,
       "step": 1860
     },
     {
       "epoch": 3.4,
-      "learning_rate": 1.771255060728745e-05,
-      "loss": 0.3565,
+      "learning_rate": 0.0,
+      "loss": 0.3559,
       "step": 1870
     },
     {
       "epoch": 3.42,
-      "learning_rate": 1.751012145748988e-05,
-      "loss": 0.3423,
+      "learning_rate": 0.0,
+      "loss": 0.3365,
       "step": 1880
     },
     {
       "epoch": 3.44,
-      "learning_rate": 1.730769230769231e-05,
-      "loss": 0.3859,
+      "learning_rate": 0.0,
+      "loss": 0.2965,
       "step": 1890
     },
     {
       "epoch": 3.46,
-      "learning_rate": 1.7105263157894737e-05,
-      "loss": 0.3606,
+      "learning_rate": 0.0,
+      "loss": 0.3412,
       "step": 1900
     },
     {
       "epoch": 3.48,
-      "learning_rate": 1.6902834008097167e-05,
-      "loss": 0.3829,
+      "learning_rate": 0.0,
+      "loss": 0.3237,
       "step": 1910
     },
     {
       "epoch": 3.5,
-      "learning_rate": 1.6700404858299595e-05,
-      "loss": 0.3409,
+      "learning_rate": 0.0,
+      "loss": 0.3052,
       "step": 1920
     },
     {
       "epoch": 3.51,
-      "learning_rate": 1.6497975708502025e-05,
-      "loss": 0.358,
+      "learning_rate": 0.0,
+      "loss": 0.296,
       "step": 1930
     },
     {
       "epoch": 3.53,
-      "learning_rate": 1.6295546558704456e-05,
-      "loss": 0.3182,
+      "learning_rate": 0.0,
+      "loss": 0.3165,
       "step": 1940
     },
     {
       "epoch": 3.55,
-      "learning_rate": 1.6093117408906883e-05,
-      "loss": 0.3595,
+      "learning_rate": 0.0,
+      "loss": 0.3097,
       "step": 1950
     },
     {
       "epoch": 3.57,
-      "learning_rate": 1.5890688259109314e-05,
-      "loss": 0.3536,
+      "learning_rate": 0.0,
+      "loss": 0.3141,
       "step": 1960
     },
     {
       "epoch": 3.59,
-      "learning_rate": 1.568825910931174e-05,
-      "loss": 0.3818,
+      "learning_rate": 0.0,
+      "loss": 0.3165,
       "step": 1970
     },
     {
       "epoch": 3.6,
-      "learning_rate": 1.548582995951417e-05,
-      "loss": 0.3401,
+      "learning_rate": 0.0,
+      "loss": 0.3504,
       "step": 1980
     },
     {
       "epoch": 3.62,
-      "learning_rate": 1.52834008097166e-05,
-      "loss": 0.3554,
+      "learning_rate": 0.0,
+      "loss": 0.3353,
       "step": 1990
     },
     {
       "epoch": 3.64,
-      "learning_rate": 1.508097165991903e-05,
-      "loss": 0.3465,
+      "learning_rate": 0.0,
+      "loss": 0.3096,
       "step": 2000
     },
     {
       "epoch": 3.66,
-      "learning_rate": 1.4878542510121457e-05,
-      "loss": 0.3482,
+      "learning_rate": 0.0,
+      "loss": 0.3167,
       "step": 2010
     },
     {
       "epoch": 3.68,
-      "learning_rate": 1.4676113360323888e-05,
-      "loss": 0.3492,
+      "learning_rate": 0.0,
+      "loss": 0.2839,
       "step": 2020
     },
     {
       "epoch": 3.7,
-      "learning_rate": 1.4473684210526317e-05,
-      "loss": 0.3366,
+      "learning_rate": 0.0,
+      "loss": 0.2972,
       "step": 2030
     },
     {
       "epoch": 3.71,
-      "learning_rate": 1.4271255060728744e-05,
-      "loss": 0.3392,
+      "learning_rate": 0.0,
+      "loss": 0.3193,
       "step": 2040
     },
     {
       "epoch": 3.73,
-      "learning_rate": 1.4068825910931175e-05,
-      "loss": 0.3593,
+      "learning_rate": 0.0,
+      "loss": 0.2793,
       "step": 2050
     },
     {
       "epoch": 3.75,
-      "learning_rate": 1.3866396761133604e-05,
-      "loss": 0.3961,
+      "learning_rate": 0.0,
+      "loss": 0.3599,
       "step": 2060
     },
     {
       "epoch": 3.77,
-      "learning_rate": 1.3663967611336034e-05,
-      "loss": 0.3996,
+      "learning_rate": 0.0,
+      "loss": 0.3134,
       "step": 2070
     },
     {
       "epoch": 3.79,
-      "learning_rate": 1.3461538461538462e-05,
-      "loss": 0.3178,
+      "learning_rate": 0.0,
+      "loss": 0.3017,
       "step": 2080
     },
     {
       "epoch": 3.81,
-      "learning_rate": 1.3259109311740892e-05,
-      "loss": 0.3641,
+      "learning_rate": 0.0,
+      "loss": 0.3107,
       "step": 2090
     },
     {
       "epoch": 3.82,
-      "learning_rate": 1.3056680161943321e-05,
-      "loss": 0.3176,
+      "learning_rate": 0.0,
+      "loss": 0.3402,
       "step": 2100
     },
     {
       "epoch": 3.84,
-      "learning_rate": 1.2854251012145749e-05,
-      "loss": 0.3189,
+      "learning_rate": 0.0,
+      "loss": 0.3046,
       "step": 2110
     },
     {
       "epoch": 3.86,
-      "learning_rate": 1.265182186234818e-05,
-      "loss": 0.3663,
+      "learning_rate": 0.0,
+      "loss": 0.3675,
       "step": 2120
     },
     {
       "epoch": 3.88,
-      "learning_rate": 1.2449392712550607e-05,
-      "loss": 0.3105,
+      "learning_rate": 0.0,
+      "loss": 0.3236,
       "step": 2130
     },
     {
       "epoch": 3.9,
-      "learning_rate": 1.2246963562753037e-05,
-      "loss": 0.3479,
+      "learning_rate": 0.0,
+      "loss": 0.2909,
       "step": 2140
     },
     {
       "epoch": 3.91,
-      "learning_rate": 1.2044534412955466e-05,
-      "loss": 0.3301,
+      "learning_rate": 0.0,
+      "loss": 0.3472,
       "step": 2150
     },
     {
       "epoch": 3.93,
-      "learning_rate": 1.1842105263157895e-05,
-      "loss": 0.3344,
+      "learning_rate": 0.0,
+      "loss": 0.2666,
       "step": 2160
     },
     {
       "epoch": 3.95,
-      "learning_rate": 1.1639676113360324e-05,
-      "loss": 0.3169,
+      "learning_rate": 0.0,
+      "loss": 0.3138,
       "step": 2170
     },
     {
       "epoch": 3.97,
-      "learning_rate": 1.1437246963562753e-05,
-      "loss": 0.3486,
+      "learning_rate": 0.0,
+      "loss": 0.3314,
       "step": 2180
     },
     {
       "epoch": 3.99,
-      "learning_rate": 1.1234817813765184e-05,
-      "loss": 0.3551,
+      "learning_rate": 0.0,
+      "loss": 0.3163,
       "step": 2190
     },
     {
       "epoch": 4.0,
-      "eval_accuracy": 0.9622695196904165,
-      "eval_loss": 0.14936527609825134,
-      "eval_runtime": 112.1506,
-      "eval_samples_per_second": 156.682,
-      "eval_steps_per_second": 4.904,
+      "eval_accuracy": 0.9639767812428864,
+      "eval_loss": 0.14118124544620514,
+      "eval_runtime": 112.337,
+      "eval_samples_per_second": 156.422,
+      "eval_steps_per_second": 4.896,
       "step": 2197
     },
     {
       "epoch": 4.01,
-      "learning_rate": 1.1032388663967611e-05,
-      "loss": 0.3307,
+      "learning_rate": 0.0,
+      "loss": 0.2996,
       "step": 2200
     },
     {
       "epoch": 4.02,
-      "learning_rate": 1.082995951417004e-05,
-      "loss": 0.3035,
+      "learning_rate": 0.0,
+      "loss": 0.3324,
       "step": 2210
     },
     {
       "epoch": 4.04,
-      "learning_rate": 1.062753036437247e-05,
-      "loss": 0.3192,
+      "learning_rate": 0.0,
+      "loss": 0.2756,
       "step": 2220
     },
     {
       "epoch": 4.06,
-      "learning_rate": 1.04251012145749e-05,
-      "loss": 0.3357,
+      "learning_rate": 0.0,
+      "loss": 0.2944,
       "step": 2230
     },
     {
       "epoch": 4.08,
-      "learning_rate": 1.0222672064777327e-05,
-      "loss": 0.3081,
+      "learning_rate": 0.0,
+      "loss": 0.3082,
       "step": 2240
     },
     {
       "epoch": 4.1,
-      "learning_rate": 1.0020242914979758e-05,
-      "loss": 0.3343,
+      "learning_rate": 0.0,
+      "loss": 0.3241,
       "step": 2250
     },
     {
       "epoch": 4.11,
-      "learning_rate": 9.817813765182187e-06,
-      "loss": 0.3438,
+      "learning_rate": 0.0,
+      "loss": 0.3301,
       "step": 2260
     },
     {
       "epoch": 4.13,
-      "learning_rate": 9.615384615384616e-06,
-      "loss": 0.2939,
+      "learning_rate": 0.0,
+      "loss": 0.283,
       "step": 2270
     },
     {
       "epoch": 4.15,
-      "learning_rate": 9.412955465587045e-06,
-      "loss": 0.2954,
+      "learning_rate": 0.0,
+      "loss": 0.3029,
       "step": 2280
     },
     {
       "epoch": 4.17,
-      "learning_rate": 9.210526315789474e-06,
-      "loss": 0.3206,
+      "learning_rate": 0.0,
+      "loss": 0.3299,
       "step": 2290
     },
     {
       "epoch": 4.19,
-      "learning_rate": 9.008097165991904e-06,
-      "loss": 0.3116,
+      "learning_rate": 0.0,
+      "loss": 0.3163,
       "step": 2300
     },
     {
       "epoch": 4.21,
-      "learning_rate": 8.805668016194332e-06,
-      "loss": 0.3201,
+      "learning_rate": 0.0,
+      "loss": 0.2886,
       "step": 2310
     },
     {
       "epoch": 4.22,
-      "learning_rate": 8.60323886639676e-06,
-      "loss": 0.3439,
+      "learning_rate": 0.0,
+      "loss": 0.2912,
       "step": 2320
     },
     {
       "epoch": 4.24,
-      "learning_rate": 8.400809716599191e-06,
-      "loss": 0.3503,
+      "learning_rate": 0.0,
+      "loss": 0.3443,
       "step": 2330
     },
     {
       "epoch": 4.26,
-      "learning_rate": 8.19838056680162e-06,
-      "loss": 0.317,
+      "learning_rate": 0.0,
+      "loss": 0.3066,
       "step": 2340
     },
     {
       "epoch": 4.28,
-      "learning_rate": 7.99595141700405e-06,
-      "loss": 0.3113,
+      "learning_rate": 0.0,
+      "loss": 0.3289,
       "step": 2350
     },
     {
       "epoch": 4.3,
-      "learning_rate": 7.793522267206478e-06,
-      "loss": 0.311,
+      "learning_rate": 0.0,
+      "loss": 0.3178,
       "step": 2360
     },
     {
       "epoch": 4.31,
-      "learning_rate": 7.591093117408906e-06,
-      "loss": 0.3083,
+      "learning_rate": 0.0,
+      "loss": 0.3248,
       "step": 2370
     },
     {
       "epoch": 4.33,
-      "learning_rate": 7.388663967611337e-06,
-      "loss": 0.3165,
+      "learning_rate": 0.0,
+      "loss": 0.3366,
       "step": 2380
     },
     {
       "epoch": 4.35,
-      "learning_rate": 7.186234817813765e-06,
-      "loss": 0.3547,
+      "learning_rate": 0.0,
+      "loss": 0.3215,
       "step": 2390
     },
     {
       "epoch": 4.37,
-      "learning_rate": 6.983805668016195e-06,
-      "loss": 0.3071,
+      "learning_rate": 0.0,
+      "loss": 0.2989,
       "step": 2400
     },
     {
       "epoch": 4.39,
-      "learning_rate": 6.781376518218624e-06,
-      "loss": 0.3028,
+      "learning_rate": 0.0,
+      "loss": 0.3011,
       "step": 2410
     },
     {
       "epoch": 4.41,
-      "learning_rate": 6.578947368421053e-06,
-      "loss": 0.3215,
+      "learning_rate": 0.0,
+      "loss": 0.3113,
       "step": 2420
     },
     {
       "epoch": 4.42,
-      "learning_rate": 6.376518218623483e-06,
-      "loss": 0.3259,
+      "learning_rate": 0.0,
+      "loss": 0.3419,
       "step": 2430
     },
     {
       "epoch": 4.44,
-      "learning_rate": 6.174089068825911e-06,
-      "loss": 0.3515,
+      "learning_rate": 0.0,
+      "loss": 0.3369,
       "step": 2440
     },
     {
       "epoch": 4.46,
-      "learning_rate": 5.971659919028341e-06,
-      "loss": 0.3158,
+      "learning_rate": 0.0,
+      "loss": 0.3148,
       "step": 2450
     },
     {
       "epoch": 4.48,
-      "learning_rate": 5.76923076923077e-06,
-      "loss": 0.3067,
+      "learning_rate": 0.0,
+      "loss": 0.3224,
       "step": 2460
     },
     {
       "epoch": 4.5,
-      "learning_rate": 5.566801619433199e-06,
-      "loss": 0.324,
+      "learning_rate": 0.0,
+      "loss": 0.2957,
       "step": 2470
     },
     {
       "epoch": 4.52,
-      "learning_rate": 5.364372469635628e-06,
-      "loss": 0.3271,
+      "learning_rate": 0.0,
+      "loss": 0.2852,
       "step": 2480
     },
     {
       "epoch": 4.53,
-      "learning_rate": 5.161943319838057e-06,
-      "loss": 0.3367,
+      "learning_rate": 0.0,
+      "loss": 0.3099,
       "step": 2490
     },
     {
       "epoch": 4.55,
-      "learning_rate": 4.9595141700404865e-06,
-      "loss": 0.3166,
+      "learning_rate": 0.0,
+      "loss": 0.3152,
       "step": 2500
     },
     {
       "epoch": 4.57,
-      "learning_rate": 4.757085020242915e-06,
-      "loss": 0.3264,
+      "learning_rate": 0.0,
+      "loss": 0.328,
       "step": 2510
     },
     {
       "epoch": 4.59,
-      "learning_rate": 4.5546558704453445e-06,
-      "loss": 0.3129,
+      "learning_rate": 0.0,
+      "loss": 0.3273,
       "step": 2520
     },
     {
       "epoch": 4.61,
-      "learning_rate": 4.3522267206477735e-06,
-      "loss": 0.3114,
+      "learning_rate": 0.0,
+      "loss": 0.3105,
       "step": 2530
     },
     {
       "epoch": 4.62,
-      "learning_rate": 4.149797570850203e-06,
-      "loss": 0.2751,
+      "learning_rate": 0.0,
+      "loss": 0.2704,
       "step": 2540
     },
     {
       "epoch": 4.64,
-      "learning_rate": 3.9473684210526315e-06,
-      "loss": 0.3191,
+      "learning_rate": 0.0,
+      "loss": 0.3113,
       "step": 2550
     },
     {
       "epoch": 4.66,
-      "learning_rate": 3.744939271255061e-06,
-      "loss": 0.3171,
+      "learning_rate": 0.0,
+      "loss": 0.291,
       "step": 2560
     },
     {
       "epoch": 4.68,
-      "learning_rate": 3.5425101214574903e-06,
-      "loss": 0.2981,
+      "learning_rate": 0.0,
+      "loss": 0.3214,
       "step": 2570
     },
     {
       "epoch": 4.7,
-      "learning_rate": 3.3400809716599197e-06,
-      "loss": 0.3169,
+      "learning_rate": 0.0,
+      "loss": 0.3307,
       "step": 2580
     },
     {
       "epoch": 4.72,
-      "learning_rate": 3.1376518218623482e-06,
-      "loss": 0.304,
+      "learning_rate": 0.0,
+      "loss": 0.2952,
       "step": 2590
     },
     {
       "epoch": 4.73,
-      "learning_rate": 2.9352226720647772e-06,
-      "loss": 0.3021,
+      "learning_rate": 0.0,
+      "loss": 0.3218,
       "step": 2600
     },
     {
       "epoch": 4.75,
-      "learning_rate": 2.7327935222672066e-06,
-      "loss": 0.3106,
+      "learning_rate": 0.0,
+      "loss": 0.2881,
       "step": 2610
     },
     {
       "epoch": 4.77,
-      "learning_rate": 2.5303643724696356e-06,
-      "loss": 0.3066,
+      "learning_rate": 0.0,
+      "loss": 0.2984,
       "step": 2620
     },
     {
       "epoch": 4.79,
-      "learning_rate": 2.327935222672065e-06,
-      "loss": 0.3117,
+      "learning_rate": 0.0,
+      "loss": 0.3318,
       "step": 2630
     },
     {
       "epoch": 4.81,
-      "learning_rate": 2.125506072874494e-06,
-      "loss": 0.3079,
+      "learning_rate": 0.0,
+      "loss": 0.3192,
       "step": 2640
     },
     {
       "epoch": 4.82,
-      "learning_rate": 1.9230769230769234e-06,
-      "loss": 0.3316,
+      "learning_rate": 0.0,
+      "loss": 0.2928,
       "step": 2650
     },
     {
       "epoch": 4.84,
-      "learning_rate": 1.7206477732793522e-06,
-      "loss": 0.3103,
+      "learning_rate": 0.0,
+      "loss": 0.3178,
       "step": 2660
     },
     {
       "epoch": 4.86,
-      "learning_rate": 1.5182186234817814e-06,
-      "loss": 0.3034,
+      "learning_rate": 0.0,
+      "loss": 0.3158,
       "step": 2670
     },
     {
       "epoch": 4.88,
-      "learning_rate": 1.3157894736842106e-06,
-      "loss": 0.294,
+      "learning_rate": 0.0,
+      "loss": 0.3225,
       "step": 2680
     },
     {
       "epoch": 4.9,
-      "learning_rate": 1.1133603238866398e-06,
-      "loss": 0.3151,
+      "learning_rate": 0.0,
+      "loss": 0.3287,
       "step": 2690
     },
     {
       "epoch": 4.92,
-      "learning_rate": 9.109311740890688e-07,
-      "loss": 0.3544,
+      "learning_rate": 0.0,
+      "loss": 0.2937,
       "step": 2700
     },
     {
       "epoch": 4.93,
-      "learning_rate": 7.08502024291498e-07,
-      "loss": 0.3365,
+      "learning_rate": 0.0,
+      "loss": 0.2998,
       "step": 2710
     },
     {
       "epoch": 4.95,
-      "learning_rate": 5.060728744939271e-07,
-      "loss": 0.3181,
+      "learning_rate": 0.0,
+      "loss": 0.3084,
       "step": 2720
     },
     {
       "epoch": 4.97,
-      "learning_rate": 3.036437246963563e-07,
-      "loss": 0.303,
+      "learning_rate": 0.0,
+      "loss": 0.3264,
       "step": 2730
     },
     {
       "epoch": 4.99,
-      "learning_rate": 1.0121457489878543e-07,
-      "loss": 0.3174,
+      "learning_rate": 0.0,
+      "loss": 0.3103,
       "step": 2740
     },
     {
       "epoch": 5.0,
-      "eval_accuracy": 0.9624402458456636,
-      "eval_loss": 0.1446385681629181,
-      "eval_runtime": 113.1571,
-      "eval_samples_per_second": 155.288,
-      "eval_steps_per_second": 4.86,
+      "eval_accuracy": 0.9639198725244708,
+      "eval_loss": 0.13910169899463654,
+      "eval_runtime": 112.3738,
+      "eval_samples_per_second": 156.371,
+      "eval_steps_per_second": 4.894,
       "step": 2745
     },
     {
       "epoch": 5.0,
       "step": 2745,
-      "total_flos": 7.483912529986142e+18,
-      "train_loss": 1.0602939221811207,
-      "train_runtime": 5306.8618,
-      "train_samples_per_second": 66.23,
-      "train_steps_per_second": 0.517
+      "total_flos": 7.485957524358513e+18,
+      "train_loss": 0.31472766021561754,
+      "train_runtime": 5227.0581,
+      "train_samples_per_second": 67.241,
+      "train_steps_per_second": 0.525
     }
   ],
   "max_steps": 2745,
   "num_train_epochs": 5,
-  "total_flos": 7.483912529986142e+18,
+  "total_flos": 7.485957524358513e+18,
   "trial_name": null,
   "trial_params": null
 }