End of training

all_results.json

@@ -1,13 +1,13 @@
 {
     "epoch": 3.0,
-    "eval_accuracy": 0.7774538386783285,
-    "eval_loss": 0.39165472984313965,
-    "eval_runtime": 52.9368,
-    "eval_samples_per_second": 136.068,
-    "eval_steps_per_second": 4.269,
+    "eval_accuracy": 0.9372483687352492,
+    "eval_loss": 0.17838290333747864,
+    "eval_runtime": 51.4343,
+    "eval_samples_per_second": 140.043,
+    "eval_steps_per_second": 4.394,
     "total_flos": 4.829589697691566e+18,
-    "train_loss": 0.5911478077470078,
-    "train_runtime": 3253.5777,
-    "train_samples_per_second": 59.774,
-    "train_steps_per_second": 0.467
+    "train_loss": 0.4650797137551314,
+    "train_runtime": 3147.7418,
+    "train_samples_per_second": 61.783,
+    "train_steps_per_second": 0.482
 }

eval_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 3.0,
-    "eval_accuracy": 0.7774538386783285,
-    "eval_loss": 0.39165472984313965,
-    "eval_runtime": 52.9368,
-    "eval_samples_per_second": 136.068,
-    "eval_steps_per_second": 4.269
+    "eval_accuracy": 0.9372483687352492,
+    "eval_loss": 0.17838290333747864,
+    "eval_runtime": 51.4343,
+    "eval_samples_per_second": 140.043,
+    "eval_steps_per_second": 4.394
 }

train_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 3.0,
     "total_flos": 4.829589697691566e+18,
-    "train_loss": 0.5911478077470078,
-    "train_runtime": 3253.5777,
-    "train_samples_per_second": 59.774,
-    "train_steps_per_second": 0.467
+    "train_loss": 0.4650797137551314,
+    "train_runtime": 3147.7418,
+    "train_samples_per_second": 61.783,
+    "train_steps_per_second": 0.482
 }

trainer_state.json

@@ -1,5 +1,5 @@
 {
-  "best_metric": 0.7774538386783285,
+  "best_metric": 0.9372483687352492,
   "best_model_checkpoint": "swin-tiny-patch4-window7-224-img_orientation/checkpoint-1518",
   "epoch": 2.9970384995064165,
   "eval_steps": 500,
@@ -11,1856 +11,1856 @@
     {
       "epoch": 0.01,
       "learning_rate": 1.6447368421052632e-06,
-      "loss": 1.4583,
+      "loss": 1.4231,
       "step": 5
     },
     {
       "epoch": 0.02,
       "learning_rate": 3.2894736842105265e-06,
-      "loss": 1.4742,
+      "loss": 1.4652,
       "step": 10
     },
     {
       "epoch": 0.03,
       "learning_rate": 4.9342105263157895e-06,
-      "loss": 1.4362,
+      "loss": 1.4556,
       "step": 15
     },
     {
       "epoch": 0.04,
       "learning_rate": 6.578947368421053e-06,
-      "loss": 1.458,
+      "loss": 1.4316,
       "step": 20
     },
     {
       "epoch": 0.05,
       "learning_rate": 8.223684210526317e-06,
-      "loss": 1.3734,
+      "loss": 1.3826,
       "step": 25
     },
     {
       "epoch": 0.06,
       "learning_rate": 9.868421052631579e-06,
-      "loss": 1.3655,
+      "loss": 1.3646,
       "step": 30
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.1513157894736843e-05,
-      "loss": 1.3442,
+      "loss": 1.3281,
       "step": 35
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.3157894736842106e-05,
-      "loss": 1.3333,
+      "loss": 1.2863,
       "step": 40
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.4802631578947368e-05,
-      "loss": 1.2669,
+      "loss": 1.2726,
       "step": 45
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.6447368421052635e-05,
-      "loss": 1.2279,
+      "loss": 1.21,
       "step": 50
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.8092105263157896e-05,
-      "loss": 1.176,
+      "loss": 1.1925,
       "step": 55
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9736842105263158e-05,
-      "loss": 1.1198,
+      "loss": 1.1344,
       "step": 60
     },
     {
       "epoch": 0.13,
       "learning_rate": 2.1381578947368423e-05,
-      "loss": 1.0456,
+      "loss": 1.0479,
       "step": 65
     },
     {
       "epoch": 0.14,
       "learning_rate": 2.3026315789473685e-05,
-      "loss": 0.9941,
+      "loss": 1.0175,
       "step": 70
     },
     {
       "epoch": 0.15,
       "learning_rate": 2.4671052631578947e-05,
-      "loss": 0.991,
+      "loss": 0.9697,
       "step": 75
     },
     {
       "epoch": 0.16,
       "learning_rate": 2.6315789473684212e-05,
-      "loss": 0.9243,
+      "loss": 0.9277,
       "step": 80
     },
     {
       "epoch": 0.17,
       "learning_rate": 2.7960526315789477e-05,
-      "loss": 0.9057,
+      "loss": 0.9186,
       "step": 85
     },
     {
       "epoch": 0.18,
       "learning_rate": 2.9605263157894735e-05,
-      "loss": 0.8941,
+      "loss": 0.938,
       "step": 90
     },
     {
       "epoch": 0.19,
       "learning_rate": 3.125e-05,
-      "loss": 0.8725,
+      "loss": 0.8542,
       "step": 95
     },
     {
       "epoch": 0.2,
       "learning_rate": 3.289473684210527e-05,
-      "loss": 0.8545,
+      "loss": 0.815,
       "step": 100
     },
     {
       "epoch": 0.21,
       "learning_rate": 3.4539473684210524e-05,
-      "loss": 0.8274,
+      "loss": 0.8339,
       "step": 105
     },
     {
       "epoch": 0.22,
       "learning_rate": 3.618421052631579e-05,
-      "loss": 0.818,
+      "loss": 0.8058,
       "step": 110
     },
     {
       "epoch": 0.23,
       "learning_rate": 3.7828947368421054e-05,
-      "loss": 0.8329,
+      "loss": 0.7445,
       "step": 115
     },
     {
       "epoch": 0.24,
       "learning_rate": 3.9473684210526316e-05,
-      "loss": 0.8005,
+      "loss": 0.8001,
       "step": 120
     },
     {
       "epoch": 0.25,
       "learning_rate": 4.111842105263158e-05,
-      "loss": 0.7636,
+      "loss": 0.785,
       "step": 125
     },
     {
       "epoch": 0.26,
       "learning_rate": 4.2763157894736847e-05,
-      "loss": 0.8036,
+      "loss": 0.764,
       "step": 130
     },
     {
       "epoch": 0.27,
       "learning_rate": 4.440789473684211e-05,
-      "loss": 0.7663,
+      "loss": 0.7118,
       "step": 135
     },
     {
       "epoch": 0.28,
       "learning_rate": 4.605263157894737e-05,
-      "loss": 0.7535,
+      "loss": 0.7243,
       "step": 140
     },
     {
       "epoch": 0.29,
       "learning_rate": 4.769736842105263e-05,
-      "loss": 0.7418,
+      "loss": 0.6734,
       "step": 145
     },
     {
       "epoch": 0.3,
       "learning_rate": 4.9342105263157894e-05,
-      "loss": 0.7471,
+      "loss": 0.7305,
       "step": 150
     },
     {
       "epoch": 0.31,
       "learning_rate": 4.9890190336749635e-05,
-      "loss": 0.7603,
+      "loss": 0.7014,
       "step": 155
     },
     {
       "epoch": 0.32,
       "learning_rate": 4.970717423133236e-05,
-      "loss": 0.7101,
+      "loss": 0.7475,
       "step": 160
     },
     {
       "epoch": 0.33,
       "learning_rate": 4.952415812591508e-05,
-      "loss": 0.6832,
+      "loss": 0.6954,
       "step": 165
     },
     {
       "epoch": 0.34,
       "learning_rate": 4.934114202049781e-05,
-      "loss": 0.6959,
+      "loss": 0.6944,
       "step": 170
     },
     {
       "epoch": 0.35,
       "learning_rate": 4.9158125915080526e-05,
-      "loss": 0.7345,
+      "loss": 0.707,
       "step": 175
     },
     {
       "epoch": 0.36,
       "learning_rate": 4.897510980966325e-05,
-      "loss": 0.7131,
+      "loss": 0.7096,
       "step": 180
     },
     {
       "epoch": 0.37,
       "learning_rate": 4.879209370424598e-05,
-      "loss": 0.7713,
+      "loss": 0.6947,
       "step": 185
     },
     {
       "epoch": 0.38,
       "learning_rate": 4.86090775988287e-05,
-      "loss": 0.6945,
+      "loss": 0.674,
       "step": 190
     },
     {
       "epoch": 0.38,
       "learning_rate": 4.8426061493411424e-05,
-      "loss": 0.6644,
+      "loss": 0.6553,
       "step": 195
     },
     {
       "epoch": 0.39,
       "learning_rate": 4.824304538799414e-05,
-      "loss": 0.679,
+      "loss": 0.6839,
       "step": 200
     },
     {
       "epoch": 0.4,
       "learning_rate": 4.806002928257687e-05,
-      "loss": 0.6657,
+      "loss": 0.622,
       "step": 205
     },
     {
       "epoch": 0.41,
       "learning_rate": 4.787701317715959e-05,
-      "loss": 0.685,
+      "loss": 0.6905,
       "step": 210
     },
     {
       "epoch": 0.42,
       "learning_rate": 4.7693997071742315e-05,
-      "loss": 0.6931,
+      "loss": 0.6331,
       "step": 215
     },
     {
       "epoch": 0.43,
       "learning_rate": 4.751098096632504e-05,
-      "loss": 0.6803,
+      "loss": 0.6139,
       "step": 220
     },
     {
       "epoch": 0.44,
       "learning_rate": 4.732796486090777e-05,
-      "loss": 0.6415,
+      "loss": 0.6399,
       "step": 225
     },
     {
       "epoch": 0.45,
       "learning_rate": 4.7144948755490486e-05,
-      "loss": 0.6753,
+      "loss": 0.691,
       "step": 230
     },
     {
       "epoch": 0.46,
       "learning_rate": 4.696193265007321e-05,
-      "loss": 0.6676,
+      "loss": 0.602,
       "step": 235
     },
     {
       "epoch": 0.47,
       "learning_rate": 4.677891654465593e-05,
-      "loss": 0.6445,
+      "loss": 0.6097,
       "step": 240
     },
     {
       "epoch": 0.48,
       "learning_rate": 4.659590043923866e-05,
-      "loss": 0.7077,
+      "loss": 0.6069,
       "step": 245
     },
     {
       "epoch": 0.49,
       "learning_rate": 4.641288433382138e-05,
-      "loss": 0.6417,
+      "loss": 0.6243,
       "step": 250
     },
     {
       "epoch": 0.5,
       "learning_rate": 4.6229868228404096e-05,
-      "loss": 0.6383,
+      "loss": 0.6222,
       "step": 255
     },
     {
       "epoch": 0.51,
       "learning_rate": 4.604685212298683e-05,
-      "loss": 0.6662,
+      "loss": 0.603,
       "step": 260
     },
     {
       "epoch": 0.52,
       "learning_rate": 4.586383601756955e-05,
-      "loss": 0.6741,
+      "loss": 0.559,
       "step": 265
     },
     {
       "epoch": 0.53,
       "learning_rate": 4.5680819912152275e-05,
-      "loss": 0.6429,
+      "loss": 0.5601,
       "step": 270
     },
     {
       "epoch": 0.54,
       "learning_rate": 4.5497803806734994e-05,
-      "loss": 0.6361,
+      "loss": 0.5935,
       "step": 275
     },
     {
       "epoch": 0.55,
       "learning_rate": 4.531478770131772e-05,
-      "loss": 0.6221,
+      "loss": 0.5495,
       "step": 280
     },
     {
       "epoch": 0.56,
       "learning_rate": 4.513177159590044e-05,
-      "loss": 0.64,
+      "loss": 0.5694,
       "step": 285
     },
     {
       "epoch": 0.57,
       "learning_rate": 4.4948755490483165e-05,
-      "loss": 0.6432,
+      "loss": 0.5767,
       "step": 290
     },
     {
       "epoch": 0.58,
       "learning_rate": 4.4765739385065885e-05,
-      "loss": 0.6621,
+      "loss": 0.635,
       "step": 295
     },
     {
       "epoch": 0.59,
       "learning_rate": 4.458272327964861e-05,
-      "loss": 0.5987,
+      "loss": 0.593,
       "step": 300
     },
     {
       "epoch": 0.6,
       "learning_rate": 4.439970717423134e-05,
-      "loss": 0.6368,
+      "loss": 0.5241,
       "step": 305
     },
     {
       "epoch": 0.61,
       "learning_rate": 4.4216691068814056e-05,
-      "loss": 0.6285,
+      "loss": 0.5306,
       "step": 310
     },
     {
       "epoch": 0.62,
       "learning_rate": 4.403367496339678e-05,
-      "loss": 0.645,
+      "loss": 0.5309,
       "step": 315
     },
     {
       "epoch": 0.63,
       "learning_rate": 4.38506588579795e-05,
-      "loss": 0.6416,
+      "loss": 0.5205,
       "step": 320
     },
     {
       "epoch": 0.64,
       "learning_rate": 4.366764275256223e-05,
-      "loss": 0.6513,
+      "loss": 0.4878,
       "step": 325
     },
     {
       "epoch": 0.65,
       "learning_rate": 4.348462664714495e-05,
-      "loss": 0.6353,
+      "loss": 0.5199,
       "step": 330
     },
     {
       "epoch": 0.66,
       "learning_rate": 4.330161054172767e-05,
-      "loss": 0.6323,
+      "loss": 0.5524,
       "step": 335
     },
     {
       "epoch": 0.67,
       "learning_rate": 4.31185944363104e-05,
-      "loss": 0.6454,
+      "loss": 0.5278,
       "step": 340
     },
     {
       "epoch": 0.68,
       "learning_rate": 4.2935578330893125e-05,
-      "loss": 0.6282,
+      "loss": 0.4959,
       "step": 345
     },
     {
       "epoch": 0.69,
       "learning_rate": 4.2752562225475845e-05,
-      "loss": 0.6029,
+      "loss": 0.5055,
       "step": 350
     },
     {
       "epoch": 0.7,
       "learning_rate": 4.256954612005857e-05,
-      "loss": 0.6069,
+      "loss": 0.5002,
       "step": 355
     },
     {
       "epoch": 0.71,
       "learning_rate": 4.238653001464129e-05,
-      "loss": 0.5707,
+      "loss": 0.5296,
       "step": 360
     },
     {
       "epoch": 0.72,
       "learning_rate": 4.2203513909224016e-05,
-      "loss": 0.6161,
+      "loss": 0.5661,
       "step": 365
     },
     {
       "epoch": 0.73,
       "learning_rate": 4.2020497803806735e-05,
-      "loss": 0.6118,
+      "loss": 0.5451,
       "step": 370
     },
     {
       "epoch": 0.74,
       "learning_rate": 4.1837481698389455e-05,
-      "loss": 0.613,
+      "loss": 0.523,
       "step": 375
     },
     {
       "epoch": 0.75,
       "learning_rate": 4.165446559297219e-05,
-      "loss": 0.6044,
+      "loss": 0.5313,
       "step": 380
     },
     {
       "epoch": 0.76,
       "learning_rate": 4.147144948755491e-05,
-      "loss": 0.5668,
+      "loss": 0.5729,
       "step": 385
     },
     {
       "epoch": 0.77,
       "learning_rate": 4.128843338213763e-05,
-      "loss": 0.6372,
+      "loss": 0.5396,
       "step": 390
     },
     {
       "epoch": 0.78,
       "learning_rate": 4.110541727672035e-05,
-      "loss": 0.5783,
+      "loss": 0.5299,
       "step": 395
     },
     {
       "epoch": 0.79,
       "learning_rate": 4.092240117130308e-05,
-      "loss": 0.6007,
+      "loss": 0.5403,
       "step": 400
     },
     {
       "epoch": 0.8,
       "learning_rate": 4.07393850658858e-05,
-      "loss": 0.5667,
+      "loss": 0.5476,
       "step": 405
     },
     {
       "epoch": 0.81,
       "learning_rate": 4.0556368960468524e-05,
-      "loss": 0.5819,
+      "loss": 0.5198,
       "step": 410
     },
     {
       "epoch": 0.82,
       "learning_rate": 4.037335285505124e-05,
-      "loss": 0.5884,
+      "loss": 0.4541,
       "step": 415
     },
     {
       "epoch": 0.83,
       "learning_rate": 4.019033674963397e-05,
-      "loss": 0.6358,
+      "loss": 0.4666,
       "step": 420
     },
     {
       "epoch": 0.84,
       "learning_rate": 4.0007320644216695e-05,
-      "loss": 0.6365,
+      "loss": 0.4636,
       "step": 425
     },
     {
       "epoch": 0.85,
       "learning_rate": 3.9824304538799415e-05,
-      "loss": 0.569,
+      "loss": 0.5061,
       "step": 430
     },
     {
       "epoch": 0.86,
       "learning_rate": 3.964128843338214e-05,
-      "loss": 0.6281,
+      "loss": 0.54,
       "step": 435
     },
     {
       "epoch": 0.87,
       "learning_rate": 3.945827232796486e-05,
-      "loss": 0.5812,
+      "loss": 0.4544,
       "step": 440
     },
     {
       "epoch": 0.88,
       "learning_rate": 3.9275256222547586e-05,
-      "loss": 0.604,
+      "loss": 0.4486,
       "step": 445
     },
     {
       "epoch": 0.89,
       "learning_rate": 3.9092240117130305e-05,
-      "loss": 0.5833,
+      "loss": 0.4375,
       "step": 450
     },
     {
       "epoch": 0.9,
       "learning_rate": 3.890922401171303e-05,
-      "loss": 0.5247,
+      "loss": 0.5157,
       "step": 455
     },
     {
       "epoch": 0.91,
       "learning_rate": 3.872620790629576e-05,
-      "loss": 0.5513,
+      "loss": 0.5203,
       "step": 460
     },
     {
       "epoch": 0.92,
       "learning_rate": 3.8543191800878484e-05,
-      "loss": 0.5885,
+      "loss": 0.4583,
       "step": 465
     },
     {
       "epoch": 0.93,
       "learning_rate": 3.83601756954612e-05,
-      "loss": 0.5995,
+      "loss": 0.4744,
       "step": 470
     },
     {
       "epoch": 0.94,
       "learning_rate": 3.817715959004393e-05,
-      "loss": 0.591,
+      "loss": 0.4537,
       "step": 475
     },
     {
       "epoch": 0.95,
       "learning_rate": 3.799414348462665e-05,
-      "loss": 0.6269,
+      "loss": 0.4854,
       "step": 480
     },
     {
       "epoch": 0.96,
       "learning_rate": 3.7811127379209375e-05,
-      "loss": 0.5496,
+      "loss": 0.4668,
       "step": 485
     },
     {
       "epoch": 0.97,
       "learning_rate": 3.7628111273792094e-05,
-      "loss": 0.6007,
+      "loss": 0.4958,
       "step": 490
     },
     {
       "epoch": 0.98,
       "learning_rate": 3.744509516837481e-05,
-      "loss": 0.5894,
+      "loss": 0.4887,
       "step": 495
     },
     {
       "epoch": 0.99,
       "learning_rate": 3.7262079062957546e-05,
-      "loss": 0.6417,
+      "loss": 0.4975,
       "step": 500
     },
     {
       "epoch": 1.0,
       "learning_rate": 3.7079062957540265e-05,
-      "loss": 0.5654,
+      "loss": 0.4079,
       "step": 505
     },
     {
       "epoch": 1.0,
-      "eval_accuracy": 0.7356656948493683,
-      "eval_loss": 0.46419134736061096,
-      "eval_runtime": 54.4631,
-      "eval_samples_per_second": 132.255,
-      "eval_steps_per_second": 4.15,
+      "eval_accuracy": 0.8781063445786478,
+      "eval_loss": 0.30538803339004517,
+      "eval_runtime": 53.2404,
+      "eval_samples_per_second": 135.292,
+      "eval_steps_per_second": 4.245,
       "step": 506
     },
     {
       "epoch": 1.01,
       "learning_rate": 3.689604685212299e-05,
-      "loss": 0.5932,
+      "loss": 0.3657,
       "step": 510
     },
     {
       "epoch": 1.02,
       "learning_rate": 3.671303074670571e-05,
-      "loss": 0.5732,
+      "loss": 0.4015,
       "step": 515
     },
     {
       "epoch": 1.03,
       "learning_rate": 3.653001464128844e-05,
-      "loss": 0.5225,
+      "loss": 0.431,
       "step": 520
     },
     {
       "epoch": 1.04,
       "learning_rate": 3.6346998535871156e-05,
-      "loss": 0.57,
+      "loss": 0.4145,
       "step": 525
     },
     {
       "epoch": 1.05,
       "learning_rate": 3.616398243045388e-05,
-      "loss": 0.5341,
+      "loss": 0.4017,
       "step": 530
     },
     {
       "epoch": 1.06,
       "learning_rate": 3.59809663250366e-05,
-      "loss": 0.5654,
+      "loss": 0.411,
       "step": 535
     },
     {
       "epoch": 1.07,
       "learning_rate": 3.579795021961933e-05,
-      "loss": 0.5749,
+      "loss": 0.4549,
       "step": 540
     },
     {
       "epoch": 1.08,
       "learning_rate": 3.5614934114202054e-05,
-      "loss": 0.5546,
+      "loss": 0.3704,
       "step": 545
     },
     {
       "epoch": 1.09,
       "learning_rate": 3.543191800878477e-05,
-      "loss": 0.5608,
+      "loss": 0.439,
       "step": 550
     },
     {
       "epoch": 1.1,
       "learning_rate": 3.52489019033675e-05,
-      "loss": 0.5155,
+      "loss": 0.3795,
       "step": 555
     },
     {
       "epoch": 1.11,
       "learning_rate": 3.506588579795022e-05,
-      "loss": 0.541,
+      "loss": 0.4194,
       "step": 560
     },
     {
       "epoch": 1.12,
       "learning_rate": 3.4882869692532945e-05,
-      "loss": 0.5515,
+      "loss": 0.434,
       "step": 565
     },
     {
       "epoch": 1.13,
       "learning_rate": 3.4699853587115664e-05,
-      "loss": 0.5686,
+      "loss": 0.3572,
       "step": 570
     },
     {
       "epoch": 1.14,
       "learning_rate": 3.451683748169839e-05,
-      "loss": 0.566,
+      "loss": 0.4794,
       "step": 575
     },
     {
       "epoch": 1.15,
       "learning_rate": 3.4333821376281116e-05,
-      "loss": 0.5548,
+      "loss": 0.4555,
       "step": 580
     },
     {
       "epoch": 1.15,
       "learning_rate": 3.415080527086384e-05,
-      "loss": 0.5808,
+      "loss": 0.3989,
       "step": 585
     },
     {
       "epoch": 1.16,
       "learning_rate": 3.396778916544656e-05,
-      "loss": 0.5683,
+      "loss": 0.4357,
       "step": 590
     },
     {
       "epoch": 1.17,
       "learning_rate": 3.378477306002929e-05,
-      "loss": 0.5428,
+      "loss": 0.4427,
       "step": 595
     },
     {
       "epoch": 1.18,
       "learning_rate": 3.360175695461201e-05,
-      "loss": 0.541,
+      "loss": 0.3575,
       "step": 600
     },
     {
       "epoch": 1.19,
       "learning_rate": 3.3418740849194726e-05,
-      "loss": 0.5082,
+      "loss": 0.4116,
       "step": 605
     },
     {
       "epoch": 1.2,
       "learning_rate": 3.323572474377745e-05,
-      "loss": 0.5672,
+      "loss": 0.3938,
       "step": 610
     },
     {
       "epoch": 1.21,
       "learning_rate": 3.305270863836017e-05,
-      "loss": 0.5593,
+      "loss": 0.4135,
       "step": 615
     },
     {
       "epoch": 1.22,
       "learning_rate": 3.2869692532942905e-05,
-      "loss": 0.5492,
+      "loss": 0.3695,
       "step": 620
     },
     {
       "epoch": 1.23,
       "learning_rate": 3.2686676427525624e-05,
-      "loss": 0.553,
+      "loss": 0.4559,
       "step": 625
     },
     {
       "epoch": 1.24,
       "learning_rate": 3.250366032210835e-05,
-      "loss": 0.5812,
+      "loss": 0.4487,
       "step": 630
     },
     {
       "epoch": 1.25,
       "learning_rate": 3.232064421669107e-05,
-      "loss": 0.5467,
+      "loss": 0.3635,
       "step": 635
     },
     {
       "epoch": 1.26,
       "learning_rate": 3.2137628111273795e-05,
-      "loss": 0.5604,
+      "loss": 0.4071,
       "step": 640
     },
     {
       "epoch": 1.27,
       "learning_rate": 3.1954612005856515e-05,
-      "loss": 0.5613,
+      "loss": 0.4295,
       "step": 645
     },
     {
       "epoch": 1.28,
       "learning_rate": 3.177159590043924e-05,
-      "loss": 0.5247,
+      "loss": 0.4958,
       "step": 650
     },
     {
       "epoch": 1.29,
       "learning_rate": 3.158857979502196e-05,
-      "loss": 0.5596,
+      "loss": 0.4281,
       "step": 655
     },
     {
       "epoch": 1.3,
       "learning_rate": 3.1405563689604686e-05,
-      "loss": 0.5211,
+      "loss": 0.4023,
       "step": 660
     },
     {
       "epoch": 1.31,
       "learning_rate": 3.122254758418741e-05,
-      "loss": 0.5338,
+      "loss": 0.4148,
       "step": 665
     },
     {
       "epoch": 1.32,
       "learning_rate": 3.103953147877013e-05,
-      "loss": 0.5422,
+      "loss": 0.407,
       "step": 670
     },
     {
       "epoch": 1.33,
       "learning_rate": 3.085651537335286e-05,
-      "loss": 0.5502,
+      "loss": 0.3638,
       "step": 675
     },
     {
       "epoch": 1.34,
       "learning_rate": 3.067349926793558e-05,
-      "loss": 0.5305,
+      "loss": 0.366,
       "step": 680
     },
     {
       "epoch": 1.35,
       "learning_rate": 3.0490483162518303e-05,
-      "loss": 0.5415,
+      "loss": 0.3173,
       "step": 685
     },
     {
       "epoch": 1.36,
       "learning_rate": 3.0307467057101026e-05,
-      "loss": 0.5065,
+      "loss": 0.4269,
       "step": 690
     },
     {
       "epoch": 1.37,
       "learning_rate": 3.012445095168375e-05,
-      "loss": 0.551,
+      "loss": 0.3789,
       "step": 695
     },
     {
       "epoch": 1.38,
       "learning_rate": 2.9941434846266475e-05,
-      "loss": 0.5345,
+      "loss": 0.3952,
       "step": 700
     },
     {
       "epoch": 1.39,
       "learning_rate": 2.9758418740849197e-05,
-      "loss": 0.514,
+      "loss": 0.3611,
       "step": 705
     },
     {
       "epoch": 1.4,
       "learning_rate": 2.957540263543192e-05,
-      "loss": 0.5535,
+      "loss": 0.3679,
       "step": 710
     },
     {
       "epoch": 1.41,
       "learning_rate": 2.9392386530014643e-05,
-      "loss": 0.5537,
+      "loss": 0.3712,
       "step": 715
     },
     {
       "epoch": 1.42,
       "learning_rate": 2.9209370424597365e-05,
-      "loss": 0.5305,
+      "loss": 0.3672,
       "step": 720
     },
     {
       "epoch": 1.43,
       "learning_rate": 2.9026354319180088e-05,
-      "loss": 0.5551,
+      "loss": 0.4433,
       "step": 725
     },
     {
       "epoch": 1.44,
       "learning_rate": 2.884333821376281e-05,
-      "loss": 0.5723,
+      "loss": 0.3833,
       "step": 730
     },
     {
       "epoch": 1.45,
       "learning_rate": 2.8660322108345534e-05,
-      "loss": 0.5722,
+      "loss": 0.4158,
       "step": 735
     },
     {
       "epoch": 1.46,
       "learning_rate": 2.847730600292826e-05,
-      "loss": 0.5376,
+      "loss": 0.3979,
       "step": 740
     },
     {
       "epoch": 1.47,
       "learning_rate": 2.8294289897510982e-05,
-      "loss": 0.577,
+      "loss": 0.4005,
       "step": 745
     },
     {
       "epoch": 1.48,
       "learning_rate": 2.8111273792093705e-05,
-      "loss": 0.5322,
+      "loss": 0.3757,
       "step": 750
     },
     {
       "epoch": 1.49,
       "learning_rate": 2.7928257686676428e-05,
-      "loss": 0.5505,
+      "loss": 0.374,
       "step": 755
     },
     {
       "epoch": 1.5,
       "learning_rate": 2.774524158125915e-05,
-      "loss": 0.4995,
+      "loss": 0.4122,
       "step": 760
     },
     {
       "epoch": 1.51,
       "learning_rate": 2.7562225475841873e-05,
-      "loss": 0.5371,
+      "loss": 0.3892,
       "step": 765
     },
     {
       "epoch": 1.52,
       "learning_rate": 2.7379209370424596e-05,
-      "loss": 0.5526,
+      "loss": 0.4024,
       "step": 770
     },
     {
       "epoch": 1.53,
       "learning_rate": 2.719619326500732e-05,
-      "loss": 0.5275,
+      "loss": 0.3734,
       "step": 775
     },
     {
       "epoch": 1.54,
       "learning_rate": 2.7013177159590048e-05,
-      "loss": 0.5073,
+      "loss": 0.4271,
       "step": 780
     },
     {
       "epoch": 1.55,
       "learning_rate": 2.683016105417277e-05,
-      "loss": 0.5645,
+      "loss": 0.3551,
       "step": 785
     },
     {
       "epoch": 1.56,
       "learning_rate": 2.6647144948755493e-05,
-      "loss": 0.5253,
+      "loss": 0.4362,
       "step": 790
     },
     {
       "epoch": 1.57,
       "learning_rate": 2.6464128843338216e-05,
-      "loss": 0.5327,
+      "loss": 0.3928,
       "step": 795
     },
     {
       "epoch": 1.58,
       "learning_rate": 2.628111273792094e-05,
-      "loss": 0.5219,
+      "loss": 0.3794,
       "step": 800
     },
     {
       "epoch": 1.59,
       "learning_rate": 2.609809663250366e-05,
-      "loss": 0.5068,
+      "loss": 0.3925,
       "step": 805
     },
     {
       "epoch": 1.6,
       "learning_rate": 2.5915080527086384e-05,
-      "loss": 0.5378,
+      "loss": 0.3458,
       "step": 810
     },
     {
       "epoch": 1.61,
       "learning_rate": 2.5732064421669107e-05,
-      "loss": 0.5464,
+      "loss": 0.3678,
       "step": 815
     },
     {
       "epoch": 1.62,
       "learning_rate": 2.5549048316251833e-05,
-      "loss": 0.5195,
+      "loss": 0.3469,
       "step": 820
     },
     {
       "epoch": 1.63,
       "learning_rate": 2.5366032210834556e-05,
-      "loss": 0.5481,
+      "loss": 0.3549,
       "step": 825
     },
     {
       "epoch": 1.64,
       "learning_rate": 2.518301610541728e-05,
-      "loss": 0.512,
+      "loss": 0.3672,
       "step": 830
     },
     {
       "epoch": 1.65,
       "learning_rate": 2.5e-05,
-      "loss": 0.5342,
+      "loss": 0.3562,
       "step": 835
     },
     {
       "epoch": 1.66,
       "learning_rate": 2.4816983894582724e-05,
-      "loss": 0.5444,
+      "loss": 0.3552,
       "step": 840
     },
     {
       "epoch": 1.67,
       "learning_rate": 2.463396778916545e-05,
-      "loss": 0.533,
+      "loss": 0.4225,
       "step": 845
     },
     {
       "epoch": 1.68,
       "learning_rate": 2.4450951683748173e-05,
-      "loss": 0.5249,
+      "loss": 0.3528,
       "step": 850
     },
     {
       "epoch": 1.69,
       "learning_rate": 2.4267935578330895e-05,
-      "loss": 0.5867,
+      "loss": 0.3317,
       "step": 855
     },
     {
       "epoch": 1.7,
       "learning_rate": 2.4084919472913618e-05,
-      "loss": 0.4845,
+      "loss": 0.344,
       "step": 860
     },
     {
       "epoch": 1.71,
       "learning_rate": 2.390190336749634e-05,
-      "loss": 0.5419,
+      "loss": 0.3163,
       "step": 865
     },
     {
       "epoch": 1.72,
       "learning_rate": 2.3718887262079064e-05,
-      "loss": 0.5173,
+      "loss": 0.3705,
       "step": 870
     },
     {
       "epoch": 1.73,
       "learning_rate": 2.3535871156661786e-05,
-      "loss": 0.5138,
+      "loss": 0.3456,
       "step": 875
     },
     {
       "epoch": 1.74,
       "learning_rate": 2.335285505124451e-05,
-      "loss": 0.5291,
+      "loss": 0.326,
       "step": 880
     },
     {
       "epoch": 1.75,
       "learning_rate": 2.3169838945827235e-05,
-      "loss": 0.5299,
+      "loss": 0.3429,
       "step": 885
     },
     {
       "epoch": 1.76,
       "learning_rate": 2.2986822840409958e-05,
-      "loss": 0.5249,
+      "loss": 0.3485,
       "step": 890
     },
     {
       "epoch": 1.77,
       "learning_rate": 2.280380673499268e-05,
-      "loss": 0.4676,
+      "loss": 0.3599,
       "step": 895
     },
     {
       "epoch": 1.78,
       "learning_rate": 2.2620790629575403e-05,
-      "loss": 0.5323,
+      "loss": 0.3657,
       "step": 900
     },
     {
       "epoch": 1.79,
       "learning_rate": 2.243777452415813e-05,
-      "loss": 0.4845,
+      "loss": 0.4039,
       "step": 905
     },
     {
       "epoch": 1.8,
       "learning_rate": 2.2254758418740852e-05,
-      "loss": 0.5559,
+      "loss": 0.328,
       "step": 910
     },
     {
       "epoch": 1.81,
       "learning_rate": 2.2071742313323575e-05,
-      "loss": 0.5064,
+      "loss": 0.3446,
       "step": 915
     },
     {
       "epoch": 1.82,
       "learning_rate": 2.1888726207906297e-05,
-      "loss": 0.5281,
+      "loss": 0.3491,
       "step": 920
     },
     {
       "epoch": 1.83,
       "learning_rate": 2.170571010248902e-05,
-      "loss": 0.5471,
+      "loss": 0.3665,
       "step": 925
     },
     {
       "epoch": 1.84,
       "learning_rate": 2.1522693997071743e-05,
-      "loss": 0.4968,
+      "loss": 0.3392,
       "step": 930
     },
     {
       "epoch": 1.85,
       "learning_rate": 2.1339677891654465e-05,
-      "loss": 0.481,
+      "loss": 0.3244,
       "step": 935
     },
     {
       "epoch": 1.86,
       "learning_rate": 2.1156661786237188e-05,
-      "loss": 0.5199,
+      "loss": 0.377,
       "step": 940
     },
     {
       "epoch": 1.87,
       "learning_rate": 2.0973645680819914e-05,
-      "loss": 0.4974,
+      "loss": 0.3165,
       "step": 945
     },
     {
       "epoch": 1.88,
       "learning_rate": 2.0790629575402637e-05,
-      "loss": 0.5176,
+      "loss": 0.3318,
       "step": 950
     },
     {
       "epoch": 1.89,
       "learning_rate": 2.060761346998536e-05,
-      "loss": 0.5054,
+      "loss": 0.3049,
       "step": 955
     },
     {
       "epoch": 1.9,
       "learning_rate": 2.0424597364568082e-05,
-      "loss": 0.5211,
+      "loss": 0.3265,
       "step": 960
     },
     {
       "epoch": 1.91,
       "learning_rate": 2.024158125915081e-05,
-      "loss": 0.4668,
+      "loss": 0.3265,
       "step": 965
     },
     {
       "epoch": 1.92,
       "learning_rate": 2.005856515373353e-05,
-      "loss": 0.4834,
+      "loss": 0.3719,
       "step": 970
     },
     {
       "epoch": 1.92,
       "learning_rate": 1.9875549048316254e-05,
-      "loss": 0.5393,
+      "loss": 0.3507,
       "step": 975
     },
     {
       "epoch": 1.93,
       "learning_rate": 1.9692532942898977e-05,
-      "loss": 0.5269,
+      "loss": 0.3319,
       "step": 980
     },
     {
       "epoch": 1.94,
       "learning_rate": 1.95095168374817e-05,
-      "loss": 0.5445,
+      "loss": 0.382,
       "step": 985
     },
     {
       "epoch": 1.95,
       "learning_rate": 1.9326500732064422e-05,
-      "loss": 0.5473,
+      "loss": 0.3204,
       "step": 990
     },
     {
       "epoch": 1.96,
       "learning_rate": 1.9143484626647145e-05,
-      "loss": 0.5002,
+      "loss": 0.3793,
       "step": 995
     },
     {
       "epoch": 1.97,
       "learning_rate": 1.8960468521229867e-05,
-      "loss": 0.4839,
+      "loss": 0.3731,
       "step": 1000
     },
     {
       "epoch": 1.98,
       "learning_rate": 1.8777452415812594e-05,
-      "loss": 0.5268,
+      "loss": 0.3628,
       "step": 1005
     },
     {
       "epoch": 1.99,
       "learning_rate": 1.8594436310395316e-05,
-      "loss": 0.5194,
+      "loss": 0.3327,
       "step": 1010
     },
     {
       "epoch": 2.0,
-      "eval_accuracy": 0.7675968346522283,
-      "eval_loss": 0.4097585380077362,
-      "eval_runtime": 51.2327,
-      "eval_samples_per_second": 140.594,
-      "eval_steps_per_second": 4.411,
+      "eval_accuracy": 0.927391364709149,
+      "eval_loss": 0.19831140339374542,
+      "eval_runtime": 51.0671,
+      "eval_samples_per_second": 141.05,
+      "eval_steps_per_second": 4.426,
       "step": 1013
     },
     {
       "epoch": 2.0,
       "learning_rate": 1.841142020497804e-05,
-      "loss": 0.4812,
+      "loss": 0.2912,
       "step": 1015
     },
     {
       "epoch": 2.01,
       "learning_rate": 1.822840409956076e-05,
-      "loss": 0.5085,
+      "loss": 0.313,
       "step": 1020
     },
     {
       "epoch": 2.02,
       "learning_rate": 1.8045387994143488e-05,
-      "loss": 0.461,
+      "loss": 0.3181,
       "step": 1025
     },
     {
       "epoch": 2.03,
       "learning_rate": 1.786237188872621e-05,
-      "loss": 0.5006,
+      "loss": 0.3328,
       "step": 1030
     },
     {
       "epoch": 2.04,
       "learning_rate": 1.7679355783308933e-05,
-      "loss": 0.4821,
+      "loss": 0.3494,
       "step": 1035
     },
     {
       "epoch": 2.05,
       "learning_rate": 1.7496339677891656e-05,
-      "loss": 0.4414,
+      "loss": 0.3584,
       "step": 1040
     },
     {
       "epoch": 2.06,
       "learning_rate": 1.731332357247438e-05,
-      "loss": 0.4986,
+      "loss": 0.3275,
       "step": 1045
     },
     {
       "epoch": 2.07,
       "learning_rate": 1.71303074670571e-05,
-      "loss": 0.4747,
+      "loss": 0.3517,
       "step": 1050
     },
     {
       "epoch": 2.08,
       "learning_rate": 1.6947291361639824e-05,
-      "loss": 0.5198,
+      "loss": 0.3464,
       "step": 1055
     },
     {
       "epoch": 2.09,
       "learning_rate": 1.6764275256222547e-05,
-      "loss": 0.4425,
+      "loss": 0.2692,
       "step": 1060
     },
     {
       "epoch": 2.1,
       "learning_rate": 1.6581259150805273e-05,
-      "loss": 0.4593,
+      "loss": 0.3252,
       "step": 1065
     },
     {
       "epoch": 2.11,
       "learning_rate": 1.6398243045387995e-05,
-      "loss": 0.5233,
+      "loss": 0.3393,
       "step": 1070
     },
     {
       "epoch": 2.12,
       "learning_rate": 1.6215226939970718e-05,
-      "loss": 0.4698,
+      "loss": 0.2896,
       "step": 1075
     },
     {
       "epoch": 2.13,
       "learning_rate": 1.603221083455344e-05,
-      "loss": 0.5108,
+      "loss": 0.2643,
       "step": 1080
     },
     {
       "epoch": 2.14,
       "learning_rate": 1.5849194729136167e-05,
-      "loss": 0.5274,
+      "loss": 0.3854,
       "step": 1085
     },
     {
       "epoch": 2.15,
       "learning_rate": 1.566617862371889e-05,
-      "loss": 0.4443,
+      "loss": 0.3183,
       "step": 1090
     },
     {
       "epoch": 2.16,
       "learning_rate": 1.5483162518301612e-05,
-      "loss": 0.4591,
+      "loss": 0.3572,
       "step": 1095
     },
     {
       "epoch": 2.17,
       "learning_rate": 1.5300146412884335e-05,
-      "loss": 0.5101,
+      "loss": 0.2625,
       "step": 1100
     },
     {
       "epoch": 2.18,
       "learning_rate": 1.511713030746706e-05,
-      "loss": 0.4938,
+      "loss": 0.3089,
       "step": 1105
     },
     {
       "epoch": 2.19,
       "learning_rate": 1.4934114202049782e-05,
-      "loss": 0.5465,
+      "loss": 0.2938,
       "step": 1110
     },
     {
       "epoch": 2.2,
       "learning_rate": 1.4751098096632505e-05,
-      "loss": 0.4086,
+      "loss": 0.3711,
       "step": 1115
     },
     {
       "epoch": 2.21,
       "learning_rate": 1.4568081991215226e-05,
-      "loss": 0.4771,
+      "loss": 0.3456,
       "step": 1120
     },
     {
       "epoch": 2.22,
       "learning_rate": 1.4385065885797952e-05,
-      "loss": 0.5017,
+      "loss": 0.37,
       "step": 1125
     },
     {
       "epoch": 2.23,
       "learning_rate": 1.4202049780380675e-05,
-      "loss": 0.4975,
+      "loss": 0.3584,
       "step": 1130
     },
     {
       "epoch": 2.24,
       "learning_rate": 1.4019033674963397e-05,
-      "loss": 0.573,
+      "loss": 0.2956,
       "step": 1135
     },
     {
       "epoch": 2.25,
       "learning_rate": 1.383601756954612e-05,
-      "loss": 0.4946,
+      "loss": 0.3251,
       "step": 1140
     },
     {
       "epoch": 2.26,
       "learning_rate": 1.3653001464128845e-05,
-      "loss": 0.4861,
+      "loss": 0.2945,
       "step": 1145
     },
     {
       "epoch": 2.27,
       "learning_rate": 1.3469985358711567e-05,
-      "loss": 0.4579,
+      "loss": 0.3545,
       "step": 1150
     },
     {
       "epoch": 2.28,
       "learning_rate": 1.328696925329429e-05,
-      "loss": 0.4574,
+      "loss": 0.2872,
       "step": 1155
     },
     {
       "epoch": 2.29,
       "learning_rate": 1.3103953147877013e-05,
-      "loss": 0.5262,
+      "loss": 0.3433,
       "step": 1160
     },
     {
       "epoch": 2.3,
       "learning_rate": 1.2920937042459739e-05,
-      "loss": 0.4991,
+      "loss": 0.3048,
       "step": 1165
     },
     {
       "epoch": 2.31,
       "learning_rate": 1.2737920937042461e-05,
-      "loss": 0.5082,
+      "loss": 0.2649,
       "step": 1170
     },
     {
       "epoch": 2.32,
       "learning_rate": 1.2554904831625182e-05,
-      "loss": 0.4557,
+      "loss": 0.3022,
       "step": 1175
     },
     {
       "epoch": 2.33,
       "learning_rate": 1.2371888726207907e-05,
-      "loss": 0.5222,
+      "loss": 0.3398,
       "step": 1180
     },
     {
       "epoch": 2.34,
       "learning_rate": 1.218887262079063e-05,
-      "loss": 0.4846,
+      "loss": 0.272,
       "step": 1185
     },
     {
       "epoch": 2.35,
       "learning_rate": 1.2005856515373354e-05,
-      "loss": 0.4616,
+      "loss": 0.3443,
       "step": 1190
     },
     {
       "epoch": 2.36,
       "learning_rate": 1.1822840409956077e-05,
-      "loss": 0.4609,
+      "loss": 0.2978,
       "step": 1195
     },
     {
       "epoch": 2.37,
       "learning_rate": 1.1639824304538801e-05,
-      "loss": 0.501,
+      "loss": 0.3128,
       "step": 1200
     },
     {
       "epoch": 2.38,
       "learning_rate": 1.1456808199121522e-05,
-      "loss": 0.4591,
+      "loss": 0.3301,
       "step": 1205
     },
     {
       "epoch": 2.39,
       "learning_rate": 1.1273792093704246e-05,
-      "loss": 0.4996,
+      "loss": 0.2922,
       "step": 1210
     },
     {
       "epoch": 2.4,
       "learning_rate": 1.109077598828697e-05,
-      "loss": 0.4629,
+      "loss": 0.2871,
       "step": 1215
     },
     {
       "epoch": 2.41,
       "learning_rate": 1.0907759882869694e-05,
-      "loss": 0.4559,
+      "loss": 0.2858,
       "step": 1220
     },
     {
       "epoch": 2.42,
       "learning_rate": 1.0724743777452416e-05,
-      "loss": 0.4881,
+      "loss": 0.3008,
       "step": 1225
     },
     {
       "epoch": 2.43,
       "learning_rate": 1.054172767203514e-05,
-      "loss": 0.4593,
+      "loss": 0.345,
       "step": 1230
     },
     {
       "epoch": 2.44,
       "learning_rate": 1.0358711566617862e-05,
-      "loss": 0.4847,
+      "loss": 0.3253,
       "step": 1235
     },
     {
       "epoch": 2.45,
       "learning_rate": 1.0175695461200586e-05,
-      "loss": 0.4676,
+      "loss": 0.319,
       "step": 1240
     },
     {
       "epoch": 2.46,
       "learning_rate": 9.992679355783309e-06,
-      "loss": 0.4893,
+      "loss": 0.3208,
       "step": 1245
     },
     {
       "epoch": 2.47,
       "learning_rate": 9.809663250366033e-06,
-      "loss": 0.4845,
+      "loss": 0.3445,
       "step": 1250
     },
     {
       "epoch": 2.48,
       "learning_rate": 9.626647144948756e-06,
-      "loss": 0.4829,
+      "loss": 0.3328,
       "step": 1255
     },
     {
       "epoch": 2.49,
       "learning_rate": 9.44363103953148e-06,
-      "loss": 0.5043,
+      "loss": 0.3148,
       "step": 1260
     },
     {
       "epoch": 2.5,
       "learning_rate": 9.260614934114201e-06,
-      "loss": 0.497,
+      "loss": 0.2937,
       "step": 1265
     },
     {
       "epoch": 2.51,
       "learning_rate": 9.077598828696926e-06,
-      "loss": 0.5003,
+      "loss": 0.2848,
       "step": 1270
     },
     {
       "epoch": 2.52,
       "learning_rate": 8.894582723279648e-06,
-      "loss": 0.4462,
+      "loss": 0.2818,
       "step": 1275
     },
     {
       "epoch": 2.53,
       "learning_rate": 8.711566617862373e-06,
-      "loss": 0.5062,
+      "loss": 0.3272,
       "step": 1280
     },
     {
       "epoch": 2.54,
       "learning_rate": 8.528550512445096e-06,
-      "loss": 0.4795,
+      "loss": 0.3098,
       "step": 1285
     },
     {
       "epoch": 2.55,
       "learning_rate": 8.34553440702782e-06,
-      "loss": 0.4773,
+      "loss": 0.2721,
       "step": 1290
     },
     {
       "epoch": 2.56,
       "learning_rate": 8.162518301610541e-06,
-      "loss": 0.4667,
+      "loss": 0.3247,
       "step": 1295
     },
     {
       "epoch": 2.57,
       "learning_rate": 7.979502196193265e-06,
-      "loss": 0.5078,
+      "loss": 0.3047,
       "step": 1300
     },
     {
       "epoch": 2.58,
       "learning_rate": 7.796486090775988e-06,
-      "loss": 0.4586,
+      "loss": 0.3353,
       "step": 1305
     },
     {
       "epoch": 2.59,
       "learning_rate": 7.613469985358712e-06,
-      "loss": 0.4328,
+      "loss": 0.3642,
       "step": 1310
     },
     {
       "epoch": 2.6,
       "learning_rate": 7.430453879941435e-06,
-      "loss": 0.443,
+      "loss": 0.3076,
       "step": 1315
     },
     {
       "epoch": 2.61,
       "learning_rate": 7.247437774524159e-06,
-      "loss": 0.4597,
+      "loss": 0.3023,
       "step": 1320
     },
     {
       "epoch": 2.62,
       "learning_rate": 7.064421669106881e-06,
-      "loss": 0.5333,
+      "loss": 0.3034,
       "step": 1325
     },
     {
       "epoch": 2.63,
       "learning_rate": 6.881405563689605e-06,
-      "loss": 0.4654,
+      "loss": 0.3302,
       "step": 1330
     },
     {
       "epoch": 2.64,
       "learning_rate": 6.698389458272328e-06,
-      "loss": 0.4383,
+      "loss": 0.2926,
       "step": 1335
     },
     {
       "epoch": 2.65,
       "learning_rate": 6.515373352855052e-06,
-      "loss": 0.4828,
+      "loss": 0.3434,
       "step": 1340
     },
     {
       "epoch": 2.66,
       "learning_rate": 6.332357247437774e-06,
-      "loss": 0.4849,
+      "loss": 0.2747,
       "step": 1345
     },
     {
       "epoch": 2.67,
       "learning_rate": 6.149341142020498e-06,
-      "loss": 0.465,
+      "loss": 0.3549,
       "step": 1350
     },
     {
       "epoch": 2.68,
       "learning_rate": 5.966325036603222e-06,
-      "loss": 0.5227,
+      "loss": 0.2671,
       "step": 1355
     },
     {
       "epoch": 2.69,
       "learning_rate": 5.7833089311859446e-06,
-      "loss": 0.5229,
+      "loss": 0.3066,
       "step": 1360
     },
     {
       "epoch": 2.69,
       "learning_rate": 5.600292825768668e-06,
-      "loss": 0.4592,
+      "loss": 0.3247,
       "step": 1365
     },
     {
       "epoch": 2.7,
       "learning_rate": 5.417276720351392e-06,
-      "loss": 0.4769,
+      "loss": 0.3116,
       "step": 1370
     },
     {
       "epoch": 2.71,
       "learning_rate": 5.234260614934114e-06,
-      "loss": 0.4959,
+      "loss": 0.2785,
       "step": 1375
     },
     {
       "epoch": 2.72,
       "learning_rate": 5.051244509516838e-06,
-      "loss": 0.4731,
+      "loss": 0.2975,
       "step": 1380
     },
     {
       "epoch": 2.73,
       "learning_rate": 4.8682284040995615e-06,
-      "loss": 0.4718,
+      "loss": 0.3296,
       "step": 1385
     },
     {
       "epoch": 2.74,
       "learning_rate": 4.685212298682284e-06,
-      "loss": 0.5007,
+      "loss": 0.3022,
       "step": 1390
     },
     {
       "epoch": 2.75,
       "learning_rate": 4.502196193265008e-06,
-      "loss": 0.437,
+      "loss": 0.3067,
       "step": 1395
     },
     {
       "epoch": 2.76,
       "learning_rate": 4.319180087847731e-06,
-      "loss": 0.4732,
+      "loss": 0.3437,
       "step": 1400
     },
     {
       "epoch": 2.77,
       "learning_rate": 4.136163982430454e-06,
-      "loss": 0.4582,
+      "loss": 0.3191,
       "step": 1405
     },
     {
       "epoch": 2.78,
       "learning_rate": 3.9531478770131775e-06,
-      "loss": 0.4903,
+      "loss": 0.3234,
       "step": 1410
     },
     {
       "epoch": 2.79,
       "learning_rate": 3.7701317715959007e-06,
-      "loss": 0.5049,
+      "loss": 0.268,
       "step": 1415
     },
     {
       "epoch": 2.8,
       "learning_rate": 3.587115666178624e-06,
-      "loss": 0.457,
+      "loss": 0.3349,
       "step": 1420
     },
     {
       "epoch": 2.81,
       "learning_rate": 3.4040995607613473e-06,
-      "loss": 0.4516,
+      "loss": 0.3179,
       "step": 1425
     },
     {
       "epoch": 2.82,
       "learning_rate": 3.2210834553440705e-06,
-      "loss": 0.5026,
+      "loss": 0.3158,
       "step": 1430
     },
     {
       "epoch": 2.83,
       "learning_rate": 3.0380673499267936e-06,
-      "loss": 0.5247,
+      "loss": 0.3094,
       "step": 1435
     },
     {
       "epoch": 2.84,
       "learning_rate": 2.855051244509517e-06,
-      "loss": 0.4605,
+      "loss": 0.2834,
       "step": 1440
     },
     {
       "epoch": 2.85,
       "learning_rate": 2.6720351390922403e-06,
-      "loss": 0.4889,
+      "loss": 0.2633,
       "step": 1445
     },
     {
       "epoch": 2.86,
       "learning_rate": 2.4890190336749634e-06,
-      "loss": 0.4645,
+      "loss": 0.2923,
       "step": 1450
     },
     {
       "epoch": 2.87,
       "learning_rate": 2.306002928257687e-06,
-      "loss": 0.4493,
+      "loss": 0.3066,
       "step": 1455
     },
     {
       "epoch": 2.88,
       "learning_rate": 2.12298682284041e-06,
-      "loss": 0.492,
+      "loss": 0.3047,
       "step": 1460
     },
     {
       "epoch": 2.89,
       "learning_rate": 1.9399707174231332e-06,
-      "loss": 0.5066,
+      "loss": 0.3098,
       "step": 1465
     },
     {
       "epoch": 2.9,
       "learning_rate": 1.7569546120058566e-06,
-      "loss": 0.4364,
+      "loss": 0.3357,
       "step": 1470
     },
     {
       "epoch": 2.91,
       "learning_rate": 1.57393850658858e-06,
-      "loss": 0.4823,
+      "loss": 0.2852,
       "step": 1475
     },
     {
       "epoch": 2.92,
       "learning_rate": 1.3909224011713032e-06,
-      "loss": 0.5101,
+      "loss": 0.3455,
       "step": 1480
     },
     {
       "epoch": 2.93,
       "learning_rate": 1.2079062957540264e-06,
-      "loss": 0.5322,
+      "loss": 0.3222,
       "step": 1485
     },
     {
       "epoch": 2.94,
       "learning_rate": 1.0248901903367497e-06,
-      "loss": 0.4745,
+      "loss": 0.2713,
       "step": 1490
     },
     {
       "epoch": 2.95,
       "learning_rate": 8.41874084919473e-07,
-      "loss": 0.4728,
+      "loss": 0.2757,
       "step": 1495
     },
     {
       "epoch": 2.96,
       "learning_rate": 6.588579795021963e-07,
-      "loss": 0.5055,
+      "loss": 0.329,
       "step": 1500
     },
     {
       "epoch": 2.97,
       "learning_rate": 4.758418740849195e-07,
-      "loss": 0.5092,
+      "loss": 0.3608,
       "step": 1505
     },
     {
       "epoch": 2.98,
       "learning_rate": 2.9282576866764276e-07,
-      "loss": 0.4562,
+      "loss": 0.2705,
       "step": 1510
     },
     {
       "epoch": 2.99,
       "learning_rate": 1.0980966325036604e-07,
-      "loss": 0.4478,
+      "loss": 0.3041,
       "step": 1515
     },
     {
       "epoch": 3.0,
-      "eval_accuracy": 0.7774538386783285,
-      "eval_loss": 0.39165472984313965,
-      "eval_runtime": 56.6277,
-      "eval_samples_per_second": 127.199,
-      "eval_steps_per_second": 3.991,
+      "eval_accuracy": 0.9372483687352492,
+      "eval_loss": 0.17838290333747864,
+      "eval_runtime": 56.76,
+      "eval_samples_per_second": 126.903,
+      "eval_steps_per_second": 3.982,
       "step": 1518
     },
     {
       "epoch": 3.0,
       "step": 1518,
       "total_flos": 4.829589697691566e+18,
-      "train_loss": 0.5911478077470078,
-      "train_runtime": 3253.5777,
-      "train_samples_per_second": 59.774,
-      "train_steps_per_second": 0.467
+      "train_loss": 0.4650797137551314,
+      "train_runtime": 3147.7418,
+      "train_samples_per_second": 61.783,
+      "train_steps_per_second": 0.482
     }
   ],
   "logging_steps": 5,