Training in progress, epoch 0

all_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 49.02,
-    "eval_accuracy": 0.8850574712643678,
-    "eval_loss": 0.38414719700813293,
-    "eval_runtime": 6.1302,
-    "eval_samples_per_second": 14.192,
-    "eval_steps_per_second": 3.589
+    "eval_accuracy": 0.9078947368421053,
+    "eval_loss": 0.41711878776550293,
+    "eval_runtime": 5.3059,
+    "eval_samples_per_second": 14.324,
+    "eval_steps_per_second": 3.581
 }

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b7766ef368aacab55697adbd46e013c9f7c6750ab26d73d882c5ed619162c21f
-size 345004552
+oid sha256:b806e98fc7300981b861a34391efa5a6fb36c3df4d06b4bc2682ff6735535ef0
+size 345004517

test_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 49.02,
-    "eval_accuracy": 0.8850574712643678,
-    "eval_loss": 0.38414719700813293,
-    "eval_runtime": 6.1302,
-    "eval_samples_per_second": 14.192,
-    "eval_steps_per_second": 3.589
+    "eval_accuracy": 0.9078947368421053,
+    "eval_loss": 0.41711878776550293,
+    "eval_runtime": 5.3059,
+    "eval_samples_per_second": 14.324,
+    "eval_steps_per_second": 3.581
 }

trainer_state.json

@@ -1,6 +1,6 @@
 {
-  "best_metric": 0.972972972972973,
-  "best_model_checkpoint": "videomae-base-finetuned-ucf101-subset/checkpoint-375",
+  "best_metric": 1.0,
+  "best_model_checkpoint": "videomae-base-finetuned-ucf101-subset/checkpoint-1950",
   "epoch": 49.02,
   "global_step": 3750,
   "is_hyper_param_search": false,
@@ -10,559 +10,559 @@
     {
       "epoch": 0.0,
       "learning_rate": 1.3333333333333334e-06,
-      "loss": 2.4049,
+      "loss": 2.3504,
       "step": 10
     },
     {
       "epoch": 0.01,
       "learning_rate": 2.666666666666667e-06,
-      "loss": 2.3782,
+      "loss": 2.409,
       "step": 20
     },
     {
       "epoch": 0.01,
       "learning_rate": 4.000000000000001e-06,
-      "loss": 2.3345,
+      "loss": 2.3102,
       "step": 30
     },
     {
       "epoch": 0.01,
       "learning_rate": 5.333333333333334e-06,
-      "loss": 2.3762,
+      "loss": 2.3537,
       "step": 40
     },
     {
       "epoch": 0.01,
       "learning_rate": 6.666666666666667e-06,
-      "loss": 2.3277,
+      "loss": 2.2566,
       "step": 50
     },
     {
       "epoch": 0.02,
       "learning_rate": 8.000000000000001e-06,
-      "loss": 2.2711,
+      "loss": 2.2599,
       "step": 60
     },
     {
       "epoch": 0.02,
       "learning_rate": 9.333333333333334e-06,
-      "loss": 2.369,
+      "loss": 2.3212,
       "step": 70
     },
     {
       "epoch": 0.02,
-      "eval_accuracy": 0.2972972972972973,
-      "eval_loss": 2.2216224670410156,
-      "eval_runtime": 2.6845,
-      "eval_samples_per_second": 13.783,
-      "eval_steps_per_second": 3.725,
+      "eval_accuracy": 0.25806451612903225,
+      "eval_loss": 2.2295100688934326,
+      "eval_runtime": 2.0436,
+      "eval_samples_per_second": 15.169,
+      "eval_steps_per_second": 3.915,
       "step": 75
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.0666666666666667e-05,
-      "loss": 2.2595,
+      "loss": 2.235,
       "step": 80
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.2e-05,
-      "loss": 2.2602,
+      "loss": 2.1721,
       "step": 90
     },
     {
       "epoch": 1.01,
       "learning_rate": 1.3333333333333333e-05,
-      "loss": 2.1993,
+      "loss": 2.1171,
       "step": 100
     },
     {
       "epoch": 1.01,
       "learning_rate": 1.4666666666666668e-05,
-      "loss": 2.2198,
+      "loss": 2.1702,
       "step": 110
     },
     {
       "epoch": 1.01,
       "learning_rate": 1.6000000000000003e-05,
-      "loss": 2.1327,
+      "loss": 2.0646,
       "step": 120
     },
     {
       "epoch": 1.01,
       "learning_rate": 1.7333333333333336e-05,
-      "loss": 2.0154,
+      "loss": 1.9602,
       "step": 130
     },
     {
       "epoch": 1.02,
       "learning_rate": 1.866666666666667e-05,
-      "loss": 2.0248,
+      "loss": 1.9113,
       "step": 140
     },
     {
       "epoch": 1.02,
       "learning_rate": 2e-05,
-      "loss": 1.8283,
+      "loss": 1.7775,
       "step": 150
     },
     {
       "epoch": 1.02,
-      "eval_accuracy": 0.4864864864864865,
-      "eval_loss": 1.7583507299423218,
-      "eval_runtime": 2.7247,
-      "eval_samples_per_second": 13.58,
-      "eval_steps_per_second": 3.67,
+      "eval_accuracy": 0.3870967741935484,
+      "eval_loss": 1.780012845993042,
+      "eval_runtime": 2.1066,
+      "eval_samples_per_second": 14.716,
+      "eval_steps_per_second": 3.798,
       "step": 150
     },
     {
       "epoch": 2.0,
       "learning_rate": 2.1333333333333335e-05,
-      "loss": 1.5075,
+      "loss": 1.5096,
       "step": 160
     },
     {
       "epoch": 2.01,
       "learning_rate": 2.2666666666666668e-05,
-      "loss": 1.4104,
+      "loss": 1.4483,
       "step": 170
     },
     {
       "epoch": 2.01,
       "learning_rate": 2.4e-05,
-      "loss": 1.1892,
+      "loss": 1.4243,
       "step": 180
     },
     {
       "epoch": 2.01,
       "learning_rate": 2.5333333333333337e-05,
-      "loss": 0.9929,
+      "loss": 1.1883,
       "step": 190
     },
     {
       "epoch": 2.01,
       "learning_rate": 2.6666666666666667e-05,
-      "loss": 0.8859,
+      "loss": 0.9991,
       "step": 200
     },
     {
       "epoch": 2.02,
       "learning_rate": 2.8000000000000003e-05,
-      "loss": 0.7763,
+      "loss": 1.0043,
       "step": 210
     },
     {
       "epoch": 2.02,
       "learning_rate": 2.9333333333333336e-05,
-      "loss": 0.8729,
+      "loss": 1.0633,
       "step": 220
     },
     {
       "epoch": 2.02,
-      "eval_accuracy": 0.7027027027027027,
-      "eval_loss": 1.0192049741744995,
-      "eval_runtime": 2.6893,
-      "eval_samples_per_second": 13.758,
-      "eval_steps_per_second": 3.718,
+      "eval_accuracy": 0.5483870967741935,
+      "eval_loss": 0.9655246734619141,
+      "eval_runtime": 2.0203,
+      "eval_samples_per_second": 15.344,
+      "eval_steps_per_second": 3.96,
       "step": 225
     },
     {
       "epoch": 3.0,
       "learning_rate": 3.066666666666667e-05,
-      "loss": 0.7605,
+      "loss": 0.7053,
       "step": 230
     },
     {
       "epoch": 3.0,
       "learning_rate": 3.2000000000000005e-05,
-      "loss": 0.709,
+      "loss": 0.7653,
       "step": 240
     },
     {
       "epoch": 3.01,
       "learning_rate": 3.3333333333333335e-05,
-      "loss": 0.4213,
+      "loss": 0.5797,
       "step": 250
     },
     {
       "epoch": 3.01,
       "learning_rate": 3.466666666666667e-05,
-      "loss": 0.6716,
+      "loss": 0.5341,
       "step": 260
     },
     {
       "epoch": 3.01,
       "learning_rate": 3.6e-05,
-      "loss": 0.7687,
+      "loss": 0.7323,
       "step": 270
     },
     {
       "epoch": 3.01,
       "learning_rate": 3.733333333333334e-05,
-      "loss": 0.4992,
+      "loss": 0.4958,
       "step": 280
     },
     {
       "epoch": 3.02,
       "learning_rate": 3.866666666666667e-05,
-      "loss": 0.5896,
+      "loss": 0.6096,
       "step": 290
     },
     {
       "epoch": 3.02,
       "learning_rate": 4e-05,
-      "loss": 0.4077,
+      "loss": 0.3783,
       "step": 300
     },
     {
       "epoch": 3.02,
-      "eval_accuracy": 0.8378378378378378,
-      "eval_loss": 0.48492956161499023,
-      "eval_runtime": 2.6873,
-      "eval_samples_per_second": 13.768,
-      "eval_steps_per_second": 3.721,
+      "eval_accuracy": 0.7419354838709677,
+      "eval_loss": 0.5900937914848328,
+      "eval_runtime": 2.0722,
+      "eval_samples_per_second": 14.96,
+      "eval_steps_per_second": 3.861,
       "step": 300
     },
     {
       "epoch": 4.0,
       "learning_rate": 4.133333333333333e-05,
-      "loss": 0.2703,
+      "loss": 0.2128,
       "step": 310
     },
     {
       "epoch": 4.01,
       "learning_rate": 4.266666666666667e-05,
-      "loss": 0.3431,
+      "loss": 0.5407,
       "step": 320
     },
     {
       "epoch": 4.01,
       "learning_rate": 4.4000000000000006e-05,
-      "loss": 0.4284,
+      "loss": 0.3279,
       "step": 330
     },
     {
       "epoch": 4.01,
       "learning_rate": 4.5333333333333335e-05,
-      "loss": 0.5786,
+      "loss": 0.4606,
       "step": 340
     },
     {
       "epoch": 4.01,
       "learning_rate": 4.666666666666667e-05,
-      "loss": 0.5347,
+      "loss": 0.4715,
       "step": 350
     },
     {
       "epoch": 4.02,
       "learning_rate": 4.8e-05,
-      "loss": 0.2638,
+      "loss": 0.3578,
       "step": 360
     },
     {
       "epoch": 4.02,
       "learning_rate": 4.933333333333334e-05,
-      "loss": 0.3742,
+      "loss": 0.756,
       "step": 370
     },
     {
       "epoch": 4.02,
-      "eval_accuracy": 0.972972972972973,
-      "eval_loss": 0.1343977451324463,
-      "eval_runtime": 2.7118,
-      "eval_samples_per_second": 13.644,
-      "eval_steps_per_second": 3.688,
+      "eval_accuracy": 0.6774193548387096,
+      "eval_loss": 0.9142343401908875,
+      "eval_runtime": 2.1263,
+      "eval_samples_per_second": 14.579,
+      "eval_steps_per_second": 3.762,
       "step": 375
     },
     {
       "epoch": 5.0,
       "learning_rate": 4.9925925925925926e-05,
-      "loss": 0.2036,
+      "loss": 0.2887,
       "step": 380
     },
     {
       "epoch": 5.0,
       "learning_rate": 4.977777777777778e-05,
-      "loss": 0.1899,
+      "loss": 1.0616,
       "step": 390
     },
     {
       "epoch": 5.01,
       "learning_rate": 4.962962962962963e-05,
-      "loss": 0.3761,
+      "loss": 0.2231,
       "step": 400
     },
     {
       "epoch": 5.01,
       "learning_rate": 4.9481481481481485e-05,
-      "loss": 0.3704,
+      "loss": 0.4401,
       "step": 410
     },
     {
       "epoch": 5.01,
       "learning_rate": 4.933333333333334e-05,
-      "loss": 0.2038,
+      "loss": 0.1261,
       "step": 420
     },
     {
       "epoch": 5.01,
       "learning_rate": 4.918518518518519e-05,
-      "loss": 0.4751,
+      "loss": 0.655,
       "step": 430
     },
     {
       "epoch": 5.02,
       "learning_rate": 4.903703703703704e-05,
-      "loss": 0.1951,
+      "loss": 0.4523,
       "step": 440
     },
     {
       "epoch": 5.02,
       "learning_rate": 4.888888888888889e-05,
-      "loss": 0.094,
+      "loss": 0.5186,
       "step": 450
     },
     {
       "epoch": 5.02,
-      "eval_accuracy": 0.8918918918918919,
-      "eval_loss": 0.24485082924365997,
-      "eval_runtime": 2.7174,
-      "eval_samples_per_second": 13.616,
-      "eval_steps_per_second": 3.68,
+      "eval_accuracy": 0.7741935483870968,
+      "eval_loss": 0.7383989095687866,
+      "eval_runtime": 2.1403,
+      "eval_samples_per_second": 14.484,
+      "eval_steps_per_second": 3.738,
       "step": 450
     },
     {
       "epoch": 6.0,
       "learning_rate": 4.874074074074074e-05,
-      "loss": 0.1336,
+      "loss": 0.1326,
       "step": 460
     },
     {
       "epoch": 6.01,
       "learning_rate": 4.8592592592592596e-05,
-      "loss": 0.0813,
+      "loss": 0.0233,
       "step": 470
     },
     {
       "epoch": 6.01,
       "learning_rate": 4.844444444444445e-05,
-      "loss": 0.1722,
+      "loss": 0.3254,
       "step": 480
     },
     {
       "epoch": 6.01,
       "learning_rate": 4.82962962962963e-05,
-      "loss": 0.5579,
+      "loss": 0.2502,
       "step": 490
     },
     {
       "epoch": 6.01,
       "learning_rate": 4.814814814814815e-05,
-      "loss": 0.3787,
+      "loss": 0.1076,
       "step": 500
     },
     {
       "epoch": 6.02,
       "learning_rate": 4.8e-05,
-      "loss": 0.5287,
+      "loss": 0.5042,
       "step": 510
     },
     {
       "epoch": 6.02,
       "learning_rate": 4.7851851851851854e-05,
-      "loss": 0.1005,
+      "loss": 0.3714,
       "step": 520
     },
     {
       "epoch": 6.02,
-      "eval_accuracy": 0.7837837837837838,
-      "eval_loss": 1.0794074535369873,
-      "eval_runtime": 2.7101,
-      "eval_samples_per_second": 13.653,
-      "eval_steps_per_second": 3.69,
+      "eval_accuracy": 0.7741935483870968,
+      "eval_loss": 1.1661574840545654,
+      "eval_runtime": 2.1688,
+      "eval_samples_per_second": 14.293,
+      "eval_steps_per_second": 3.689,
       "step": 525
     },
     {
       "epoch": 7.0,
       "learning_rate": 4.770370370370371e-05,
-      "loss": 0.5154,
+      "loss": 0.8849,
       "step": 530
     },
     {
       "epoch": 7.0,
       "learning_rate": 4.755555555555556e-05,
-      "loss": 0.1985,
+      "loss": 0.3047,
       "step": 540
     },
     {
       "epoch": 7.01,
       "learning_rate": 4.740740740740741e-05,
-      "loss": 0.2665,
+      "loss": 0.4302,
       "step": 550
     },
     {
       "epoch": 7.01,
       "learning_rate": 4.7259259259259266e-05,
-      "loss": 0.0321,
+      "loss": 0.454,
       "step": 560
     },
     {
       "epoch": 7.01,
       "learning_rate": 4.711111111111111e-05,
-      "loss": 0.159,
+      "loss": 0.1688,
       "step": 570
     },
     {
       "epoch": 7.01,
       "learning_rate": 4.6962962962962966e-05,
-      "loss": 0.3711,
+      "loss": 0.1852,
       "step": 580
     },
     {
       "epoch": 7.02,
       "learning_rate": 4.681481481481482e-05,
-      "loss": 0.0089,
+      "loss": 0.0478,
       "step": 590
     },
     {
       "epoch": 7.02,
       "learning_rate": 4.666666666666667e-05,
-      "loss": 0.0053,
+      "loss": 0.0263,
       "step": 600
     },
     {
       "epoch": 7.02,
-      "eval_accuracy": 0.9459459459459459,
-      "eval_loss": 0.23636393249034882,
-      "eval_runtime": 2.6811,
-      "eval_samples_per_second": 13.8,
-      "eval_steps_per_second": 3.73,
+      "eval_accuracy": 0.8064516129032258,
+      "eval_loss": 0.910193681716919,
+      "eval_runtime": 2.227,
+      "eval_samples_per_second": 13.92,
+      "eval_steps_per_second": 3.592,
       "step": 600
     },
     {
       "epoch": 8.0,
       "learning_rate": 4.6518518518518525e-05,
-      "loss": 0.0056,
+      "loss": 0.0189,
       "step": 610
     },
     {
       "epoch": 8.01,
       "learning_rate": 4.637037037037038e-05,
-      "loss": 0.3093,
+      "loss": 0.0965,
       "step": 620
     },
     {
       "epoch": 8.01,
       "learning_rate": 4.6222222222222224e-05,
-      "loss": 0.3216,
+      "loss": 0.0689,
       "step": 630
     },
     {
       "epoch": 8.01,
       "learning_rate": 4.607407407407408e-05,
-      "loss": 0.1304,
+      "loss": 0.0171,
       "step": 640
     },
     {
       "epoch": 8.01,
       "learning_rate": 4.592592592592593e-05,
-      "loss": 0.148,
+      "loss": 0.281,
       "step": 650
     },
     {
       "epoch": 8.02,
       "learning_rate": 4.577777777777778e-05,
-      "loss": 0.132,
+      "loss": 0.2834,
       "step": 660
     },
     {
       "epoch": 8.02,
       "learning_rate": 4.5629629629629636e-05,
-      "loss": 0.0807,
+      "loss": 0.0848,
       "step": 670
     },
     {
       "epoch": 8.02,
-      "eval_accuracy": 0.8378378378378378,
-      "eval_loss": 0.6658951640129089,
-      "eval_runtime": 2.6902,
-      "eval_samples_per_second": 13.754,
-      "eval_steps_per_second": 3.717,
+      "eval_accuracy": 0.9354838709677419,
+      "eval_loss": 0.12099169194698334,
+      "eval_runtime": 2.1891,
+      "eval_samples_per_second": 14.161,
+      "eval_steps_per_second": 3.654,
       "step": 675
     },
     {
       "epoch": 9.0,
       "learning_rate": 4.548148148148149e-05,
-      "loss": 0.0775,
+      "loss": 0.009,
       "step": 680
     },
     {
       "epoch": 9.0,
       "learning_rate": 4.5333333333333335e-05,
-      "loss": 0.1351,
+      "loss": 0.1009,
       "step": 690
     },
     {
       "epoch": 9.01,
       "learning_rate": 4.518518518518519e-05,
-      "loss": 0.1521,
+      "loss": 0.1219,
       "step": 700
     },
     {
       "epoch": 9.01,
       "learning_rate": 4.503703703703704e-05,
-      "loss": 0.4525,
+      "loss": 0.0463,
       "step": 710
     },
     {
       "epoch": 9.01,
       "learning_rate": 4.4888888888888894e-05,
-      "loss": 0.4013,
+      "loss": 0.0083,
       "step": 720
     },
     {
       "epoch": 9.01,
       "learning_rate": 4.474074074074075e-05,
-      "loss": 0.0261,
+      "loss": 0.0036,
       "step": 730
     },
     {
       "epoch": 9.02,
       "learning_rate": 4.4592592592592594e-05,
-      "loss": 0.0091,
+      "loss": 0.0295,
       "step": 740
     },
     {
       "epoch": 9.02,
       "learning_rate": 4.4444444444444447e-05,
-      "loss": 0.0031,
+      "loss": 0.0028,
       "step": 750
     },
     {
       "epoch": 9.02,
-      "eval_accuracy": 0.918918918918919,
-      "eval_loss": 0.4496133029460907,
-      "eval_runtime": 2.7338,
-      "eval_samples_per_second": 13.534,
-      "eval_steps_per_second": 3.658,
+      "eval_accuracy": 0.9354838709677419,
+      "eval_loss": 0.2373807281255722,
+      "eval_runtime": 2.1663,
+      "eval_samples_per_second": 14.31,
+      "eval_steps_per_second": 3.693,
       "step": 750
     },
     {
       "epoch": 10.0,
       "learning_rate": 4.42962962962963e-05,
-      "loss": 0.0035,
+      "loss": 0.005,
       "step": 760
     },
     {
       "epoch": 10.01,
       "learning_rate": 4.414814814814815e-05,
-      "loss": 0.0199,
+      "loss": 0.0043,
       "step": 770
     },
     {
       "epoch": 10.01,
       "learning_rate": 4.4000000000000006e-05,
-      "loss": 0.0028,
+      "loss": 0.0094,
       "step": 780
     },
     {
@@ -574,40 +574,40 @@
     {
       "epoch": 10.01,
       "learning_rate": 4.3703703703703705e-05,
-      "loss": 0.0035,
+      "loss": 0.0051,
       "step": 800
     },
     {
       "epoch": 10.02,
       "learning_rate": 4.355555555555556e-05,
-      "loss": 0.0025,
+      "loss": 0.0024,
       "step": 810
     },
     {
       "epoch": 10.02,
       "learning_rate": 4.340740740740741e-05,
-      "loss": 0.0203,
+      "loss": 0.0863,
       "step": 820
     },
     {
       "epoch": 10.02,
-      "eval_accuracy": 0.918918918918919,
-      "eval_loss": 0.3398858308792114,
-      "eval_runtime": 2.8361,
-      "eval_samples_per_second": 13.046,
-      "eval_steps_per_second": 3.526,
+      "eval_accuracy": 0.967741935483871,
+      "eval_loss": 0.17276477813720703,
+      "eval_runtime": 2.132,
+      "eval_samples_per_second": 14.54,
+      "eval_steps_per_second": 3.752,
       "step": 825
     },
     {
       "epoch": 11.0,
       "learning_rate": 4.325925925925926e-05,
-      "loss": 0.0042,
+      "loss": 0.0026,
       "step": 830
     },
     {
       "epoch": 11.0,
       "learning_rate": 4.311111111111111e-05,
-      "loss": 0.0024,
+      "loss": 0.0022,
       "step": 840
     },
     {
@@ -619,103 +619,103 @@
     {
       "epoch": 11.01,
       "learning_rate": 4.2814814814814816e-05,
-      "loss": 0.0266,
+      "loss": 0.002,
       "step": 860
     },
     {
       "epoch": 11.01,
       "learning_rate": 4.266666666666667e-05,
-      "loss": 0.018,
+      "loss": 0.002,
       "step": 870
     },
     {
       "epoch": 11.01,
       "learning_rate": 4.2518518518518515e-05,
-      "loss": 0.0028,
+      "loss": 0.0033,
       "step": 880
     },
     {
       "epoch": 11.02,
       "learning_rate": 4.237037037037037e-05,
-      "loss": 0.2435,
+      "loss": 0.1644,
       "step": 890
     },
     {
       "epoch": 11.02,
       "learning_rate": 4.222222222222222e-05,
-      "loss": 0.0093,
+      "loss": 0.0018,
       "step": 900
     },
     {
       "epoch": 11.02,
-      "eval_accuracy": 0.9459459459459459,
-      "eval_loss": 0.3724738359451294,
-      "eval_runtime": 2.729,
-      "eval_samples_per_second": 13.558,
-      "eval_steps_per_second": 3.664,
+      "eval_accuracy": 0.9354838709677419,
+      "eval_loss": 0.3042505085468292,
+      "eval_runtime": 2.1005,
+      "eval_samples_per_second": 14.759,
+      "eval_steps_per_second": 3.809,
       "step": 900
     },
     {
       "epoch": 12.0,
       "learning_rate": 4.2074074074074075e-05,
-      "loss": 0.1554,
+      "loss": 0.002,
       "step": 910
     },
     {
       "epoch": 12.01,
       "learning_rate": 4.192592592592593e-05,
-      "loss": 0.0049,
+      "loss": 0.0018,
       "step": 920
     },
     {
       "epoch": 12.01,
       "learning_rate": 4.177777777777778e-05,
-      "loss": 0.1019,
+      "loss": 0.0043,
       "step": 930
     },
     {
       "epoch": 12.01,
       "learning_rate": 4.162962962962963e-05,
-      "loss": 0.2167,
+      "loss": 0.0028,
       "step": 940
     },
     {
       "epoch": 12.01,
       "learning_rate": 4.148148148148148e-05,
-      "loss": 0.1541,
+      "loss": 0.002,
       "step": 950
     },
     {
       "epoch": 12.02,
       "learning_rate": 4.133333333333333e-05,
-      "loss": 0.0032,
+      "loss": 0.0027,
       "step": 960
     },
     {
       "epoch": 12.02,
       "learning_rate": 4.1185185185185186e-05,
-      "loss": 0.0022,
+      "loss": 0.1662,
       "step": 970
     },
     {
       "epoch": 12.02,
-      "eval_accuracy": 0.918918918918919,
-      "eval_loss": 0.5498412847518921,
-      "eval_runtime": 2.7424,
-      "eval_samples_per_second": 13.492,
-      "eval_steps_per_second": 3.647,
+      "eval_accuracy": 0.9032258064516129,
+      "eval_loss": 0.42439064383506775,
+      "eval_runtime": 2.1089,
+      "eval_samples_per_second": 14.699,
+      "eval_steps_per_second": 3.793,
       "step": 975
     },
     {
       "epoch": 13.0,
       "learning_rate": 4.103703703703704e-05,
-      "loss": 0.1514,
+      "loss": 0.0018,
       "step": 980
     },
     {
       "epoch": 13.0,
       "learning_rate": 4.088888888888889e-05,
-      "loss": 0.0019,
+      "loss": 0.041,
       "step": 990
     },
     {
@@ -727,40 +727,40 @@
     {
       "epoch": 13.01,
       "learning_rate": 4.059259259259259e-05,
-      "loss": 0.0017,
+      "loss": 0.0107,
       "step": 1010
     },
     {
       "epoch": 13.01,
       "learning_rate": 4.0444444444444444e-05,
-      "loss": 0.0017,
+      "loss": 0.0019,
       "step": 1020
     },
     {
       "epoch": 13.01,
       "learning_rate": 4.02962962962963e-05,
-      "loss": 0.0015,
+      "loss": 0.0016,
       "step": 1030
     },
     {
       "epoch": 13.02,
       "learning_rate": 4.014814814814815e-05,
-      "loss": 0.0015,
+      "loss": 0.0016,
       "step": 1040
     },
     {
       "epoch": 13.02,
       "learning_rate": 4e-05,
-      "loss": 0.0017,
+      "loss": 0.0132,
       "step": 1050
     },
     {
       "epoch": 13.02,
-      "eval_accuracy": 0.972972972972973,
-      "eval_loss": 0.16978278756141663,
-      "eval_runtime": 2.7539,
-      "eval_samples_per_second": 13.436,
-      "eval_steps_per_second": 3.631,
+      "eval_accuracy": 0.8709677419354839,
+      "eval_loss": 0.6965717673301697,
+      "eval_runtime": 2.0555,
+      "eval_samples_per_second": 15.082,
+      "eval_steps_per_second": 3.892,
       "step": 1050
     },
     {
@@ -772,25 +772,25 @@
     {
       "epoch": 14.01,
       "learning_rate": 3.97037037037037e-05,
-      "loss": 0.0014,
+      "loss": 0.0444,
       "step": 1070
     },
     {
       "epoch": 14.01,
       "learning_rate": 3.9555555555555556e-05,
-      "loss": 0.0014,
+      "loss": 0.0015,
       "step": 1080
     },
     {
       "epoch": 14.01,
       "learning_rate": 3.940740740740741e-05,
-      "loss": 0.0013,
+      "loss": 0.0015,
       "step": 1090
     },
     {
       "epoch": 14.01,
       "learning_rate": 3.925925925925926e-05,
-      "loss": 0.0012,
+      "loss": 0.0014,
       "step": 1100
     },
     {
@@ -802,73 +802,73 @@
     {
       "epoch": 14.02,
       "learning_rate": 3.896296296296296e-05,
-      "loss": 0.0014,
+      "loss": 0.0019,
       "step": 1120
     },
     {
       "epoch": 14.02,
-      "eval_accuracy": 0.9459459459459459,
-      "eval_loss": 0.19233068823814392,
-      "eval_runtime": 2.8803,
-      "eval_samples_per_second": 12.846,
-      "eval_steps_per_second": 3.472,
+      "eval_accuracy": 0.8709677419354839,
+      "eval_loss": 0.5602208971977234,
+      "eval_runtime": 2.0688,
+      "eval_samples_per_second": 14.985,
+      "eval_steps_per_second": 3.867,
       "step": 1125
     },
     {
       "epoch": 15.0,
       "learning_rate": 3.8814814814814814e-05,
-      "loss": 0.0012,
+      "loss": 0.0013,
       "step": 1130
     },
     {
       "epoch": 15.0,
       "learning_rate": 3.866666666666667e-05,
-      "loss": 0.0013,
+      "loss": 0.2104,
       "step": 1140
     },
     {
       "epoch": 15.01,
       "learning_rate": 3.851851851851852e-05,
-      "loss": 0.0206,
+      "loss": 0.0014,
       "step": 1150
     },
     {
       "epoch": 15.01,
       "learning_rate": 3.837037037037037e-05,
-      "loss": 0.0014,
+      "loss": 0.0015,
       "step": 1160
     },
     {
       "epoch": 15.01,
       "learning_rate": 3.8222222222222226e-05,
-      "loss": 0.2069,
+      "loss": 0.0011,
       "step": 1170
     },
     {
       "epoch": 15.01,
       "learning_rate": 3.807407407407408e-05,
-      "loss": 0.0014,
+      "loss": 0.0013,
       "step": 1180
     },
     {
       "epoch": 15.02,
       "learning_rate": 3.7925925925925925e-05,
-      "loss": 0.0012,
+      "loss": 0.0011,
       "step": 1190
     },
     {
       "epoch": 15.02,
       "learning_rate": 3.777777777777778e-05,
-      "loss": 0.0014,
+      "loss": 0.0012,
       "step": 1200
     },
     {
       "epoch": 15.02,
-      "eval_accuracy": 0.972972972972973,
-      "eval_loss": 0.15707702934741974,
-      "eval_runtime": 2.8394,
-      "eval_samples_per_second": 13.031,
-      "eval_steps_per_second": 3.522,
+      "eval_accuracy": 0.8709677419354839,
+      "eval_loss": 0.26489391922950745,
+      "eval_runtime": 2.1064,
+      "eval_samples_per_second": 14.717,
+      "eval_steps_per_second": 3.798,
       "step": 1200
     },
     {
@@ -880,139 +880,139 @@
     {
       "epoch": 16.01,
       "learning_rate": 3.7481481481481484e-05,
-      "loss": 0.0018,
+      "loss": 0.0014,
       "step": 1220
     },
     {
       "epoch": 16.01,
       "learning_rate": 3.733333333333334e-05,
-      "loss": 0.1871,
+      "loss": 0.0012,
       "step": 1230
     },
     {
       "epoch": 16.01,
       "learning_rate": 3.718518518518519e-05,
-      "loss": 0.0012,
+      "loss": 0.0011,
       "step": 1240
     },
     {
       "epoch": 16.01,
       "learning_rate": 3.7037037037037037e-05,
-      "loss": 0.0014,
+      "loss": 0.0013,
       "step": 1250
     },
     {
       "epoch": 16.02,
       "learning_rate": 3.688888888888889e-05,
-      "loss": 0.132,
+      "loss": 0.003,
       "step": 1260
     },
     {
       "epoch": 16.02,
       "learning_rate": 3.674074074074074e-05,
-      "loss": 0.0474,
+      "loss": 0.0061,
       "step": 1270
     },
     {
       "epoch": 16.02,
-      "eval_accuracy": 0.8918918918918919,
-      "eval_loss": 0.5192855596542358,
-      "eval_runtime": 2.8857,
-      "eval_samples_per_second": 12.822,
-      "eval_steps_per_second": 3.465,
+      "eval_accuracy": 0.8709677419354839,
+      "eval_loss": 0.7361043095588684,
+      "eval_runtime": 2.0949,
+      "eval_samples_per_second": 14.798,
+      "eval_steps_per_second": 3.819,
       "step": 1275
     },
     {
       "epoch": 17.0,
       "learning_rate": 3.6592592592592596e-05,
-      "loss": 0.0019,
+      "loss": 0.1429,
       "step": 1280
     },
     {
       "epoch": 17.0,
       "learning_rate": 3.644444444444445e-05,
-      "loss": 0.0013,
+      "loss": 0.0012,
       "step": 1290
     },
     {
       "epoch": 17.01,
       "learning_rate": 3.62962962962963e-05,
-      "loss": 0.1087,
+      "loss": 0.0011,
       "step": 1300
     },
     {
       "epoch": 17.01,
       "learning_rate": 3.614814814814815e-05,
-      "loss": 0.081,
+      "loss": 0.2073,
       "step": 1310
     },
     {
       "epoch": 17.01,
       "learning_rate": 3.6e-05,
-      "loss": 0.0019,
+      "loss": 0.0425,
       "step": 1320
     },
     {
       "epoch": 17.01,
       "learning_rate": 3.5851851851851854e-05,
-      "loss": 0.0012,
+      "loss": 0.0014,
       "step": 1330
     },
     {
       "epoch": 17.02,
       "learning_rate": 3.570370370370371e-05,
-      "loss": 0.0012,
+      "loss": 0.0016,
       "step": 1340
     },
     {
       "epoch": 17.02,
       "learning_rate": 3.555555555555556e-05,
-      "loss": 0.0011,
+      "loss": 0.0012,
       "step": 1350
     },
     {
       "epoch": 17.02,
-      "eval_accuracy": 0.972972972972973,
-      "eval_loss": 0.14079852402210236,
-      "eval_runtime": 3.0095,
-      "eval_samples_per_second": 12.294,
-      "eval_steps_per_second": 3.323,
+      "eval_accuracy": 0.9354838709677419,
+      "eval_loss": 0.2821272313594818,
+      "eval_runtime": 2.1139,
+      "eval_samples_per_second": 14.665,
+      "eval_steps_per_second": 3.784,
       "step": 1350
     },
     {
       "epoch": 18.0,
       "learning_rate": 3.540740740740741e-05,
-      "loss": 0.0009,
+      "loss": 0.0011,
       "step": 1360
     },
     {
       "epoch": 18.01,
       "learning_rate": 3.525925925925926e-05,
-      "loss": 0.0011,
+      "loss": 0.0951,
       "step": 1370
     },
     {
       "epoch": 18.01,
       "learning_rate": 3.511111111111111e-05,
-      "loss": 0.0277,
+      "loss": 0.0011,
       "step": 1380
     },
     {
       "epoch": 18.01,
       "learning_rate": 3.4962962962962965e-05,
-      "loss": 0.0011,
+      "loss": 0.0013,
       "step": 1390
     },
     {
       "epoch": 18.01,
       "learning_rate": 3.481481481481482e-05,
-      "loss": 0.0011,
+      "loss": 0.001,
       "step": 1400
     },
     {
       "epoch": 18.02,
       "learning_rate": 3.466666666666667e-05,
-      "loss": 0.1377,
+      "loss": 0.0094,
       "step": 1410
     },
     {
@@ -1023,137 +1023,137 @@
     },
     {
       "epoch": 18.02,
-      "eval_accuracy": 0.9459459459459459,
-      "eval_loss": 0.34057652950286865,
-      "eval_runtime": 2.8674,
-      "eval_samples_per_second": 12.904,
-      "eval_steps_per_second": 3.488,
+      "eval_accuracy": 0.9354838709677419,
+      "eval_loss": 0.3117120563983917,
+      "eval_runtime": 2.0573,
+      "eval_samples_per_second": 15.068,
+      "eval_steps_per_second": 3.889,
       "step": 1425
     },
     {
       "epoch": 19.0,
       "learning_rate": 3.437037037037037e-05,
-      "loss": 0.002,
+      "loss": 0.001,
       "step": 1430
     },
     {
       "epoch": 19.0,
       "learning_rate": 3.4222222222222224e-05,
-      "loss": 0.0018,
+      "loss": 0.001,
       "step": 1440
     },
     {
       "epoch": 19.01,
       "learning_rate": 3.4074074074074077e-05,
-      "loss": 0.2287,
+      "loss": 0.0147,
       "step": 1450
     },
     {
       "epoch": 19.01,
       "learning_rate": 3.392592592592593e-05,
-      "loss": 0.102,
+      "loss": 0.0013,
       "step": 1460
     },
     {
       "epoch": 19.01,
       "learning_rate": 3.377777777777778e-05,
-      "loss": 0.0785,
+      "loss": 0.063,
       "step": 1470
     },
     {
       "epoch": 19.01,
       "learning_rate": 3.3629629629629636e-05,
-      "loss": 0.002,
+      "loss": 0.1805,
       "step": 1480
     },
     {
       "epoch": 19.02,
       "learning_rate": 3.348148148148148e-05,
-      "loss": 0.0357,
+      "loss": 0.1541,
       "step": 1490
     },
     {
       "epoch": 19.02,
       "learning_rate": 3.3333333333333335e-05,
-      "loss": 0.0034,
+      "loss": 0.001,
       "step": 1500
     },
     {
       "epoch": 19.02,
-      "eval_accuracy": 0.9459459459459459,
-      "eval_loss": 0.25160789489746094,
-      "eval_runtime": 2.8519,
-      "eval_samples_per_second": 12.974,
-      "eval_steps_per_second": 3.506,
+      "eval_accuracy": 0.967741935483871,
+      "eval_loss": 0.11934669315814972,
+      "eval_runtime": 2.125,
+      "eval_samples_per_second": 14.588,
+      "eval_steps_per_second": 3.765,
       "step": 1500
     },
     {
       "epoch": 20.0,
       "learning_rate": 3.318518518518519e-05,
-      "loss": 0.0609,
+      "loss": 0.0022,
       "step": 1510
     },
     {
       "epoch": 20.01,
       "learning_rate": 3.303703703703704e-05,
-      "loss": 0.0013,
+      "loss": 0.0009,
       "step": 1520
     },
     {
       "epoch": 20.01,
       "learning_rate": 3.2888888888888894e-05,
-      "loss": 0.0012,
+      "loss": 0.0016,
       "step": 1530
     },
     {
       "epoch": 20.01,
       "learning_rate": 3.274074074074075e-05,
-      "loss": 0.0014,
+      "loss": 0.1235,
       "step": 1540
     },
     {
       "epoch": 20.01,
       "learning_rate": 3.25925925925926e-05,
-      "loss": 0.1104,
+      "loss": 0.0055,
       "step": 1550
     },
     {
       "epoch": 20.02,
       "learning_rate": 3.2444444444444446e-05,
-      "loss": 0.0011,
+      "loss": 0.001,
       "step": 1560
     },
     {
       "epoch": 20.02,
       "learning_rate": 3.22962962962963e-05,
-      "loss": 0.0029,
+      "loss": 0.0021,
       "step": 1570
     },
     {
       "epoch": 20.02,
-      "eval_accuracy": 0.918918918918919,
-      "eval_loss": 0.29616308212280273,
-      "eval_runtime": 2.9222,
-      "eval_samples_per_second": 12.662,
-      "eval_steps_per_second": 3.422,
+      "eval_accuracy": 0.9354838709677419,
+      "eval_loss": 0.24128369987010956,
+      "eval_runtime": 2.0867,
+      "eval_samples_per_second": 14.856,
+      "eval_steps_per_second": 3.834,
       "step": 1575
     },
     {
       "epoch": 21.0,
       "learning_rate": 3.214814814814815e-05,
-      "loss": 0.0014,
+      "loss": 0.001,
       "step": 1580
     },
     {
       "epoch": 21.0,
       "learning_rate": 3.2000000000000005e-05,
-      "loss": 0.0769,
+      "loss": 0.0009,
       "step": 1590
     },
     {
       "epoch": 21.01,
       "learning_rate": 3.185185185185185e-05,
-      "loss": 0.001,
+      "loss": 0.0012,
       "step": 1600
     },
     {
@@ -1165,64 +1165,64 @@
     {
       "epoch": 21.01,
       "learning_rate": 3.155555555555556e-05,
-      "loss": 0.1219,
+      "loss": 0.0009,
       "step": 1620
     },
     {
       "epoch": 21.01,
       "learning_rate": 3.140740740740741e-05,
-      "loss": 0.001,
+      "loss": 0.0008,
       "step": 1630
     },
     {
       "epoch": 21.02,
       "learning_rate": 3.1259259259259264e-05,
-      "loss": 0.0013,
+      "loss": 0.0009,
       "step": 1640
     },
     {
       "epoch": 21.02,
       "learning_rate": 3.111111111111111e-05,
-      "loss": 0.0008,
+      "loss": 0.0009,
       "step": 1650
     },
     {
       "epoch": 21.02,
-      "eval_accuracy": 0.918918918918919,
-      "eval_loss": 0.4023502469062805,
-      "eval_runtime": 2.9031,
-      "eval_samples_per_second": 12.745,
-      "eval_steps_per_second": 3.445,
+      "eval_accuracy": 0.967741935483871,
+      "eval_loss": 0.1641165167093277,
+      "eval_runtime": 2.0769,
+      "eval_samples_per_second": 14.926,
+      "eval_steps_per_second": 3.852,
       "step": 1650
     },
     {
       "epoch": 22.0,
       "learning_rate": 3.096296296296296e-05,
-      "loss": 0.001,
+      "loss": 0.0009,
       "step": 1660
     },
     {
       "epoch": 22.01,
       "learning_rate": 3.0814814814814816e-05,
-      "loss": 0.0009,
+      "loss": 0.0289,
       "step": 1670
     },
     {
       "epoch": 22.01,
       "learning_rate": 3.066666666666667e-05,
-      "loss": 0.0022,
+      "loss": 0.0008,
       "step": 1680
     },
     {
       "epoch": 22.01,
       "learning_rate": 3.0518518518518515e-05,
-      "loss": 0.0009,
+      "loss": 0.0008,
       "step": 1690
     },
     {
       "epoch": 22.01,
       "learning_rate": 3.037037037037037e-05,
-      "loss": 0.0009,
+      "loss": 0.0008,
       "step": 1700
     },
     {
@@ -1234,22 +1234,22 @@
     {
       "epoch": 22.02,
       "learning_rate": 3.0074074074074078e-05,
-      "loss": 0.0008,
+      "loss": 0.0016,
       "step": 1720
     },
     {
       "epoch": 22.02,
-      "eval_accuracy": 0.918918918918919,
-      "eval_loss": 0.4643724262714386,
-      "eval_runtime": 2.9019,
-      "eval_samples_per_second": 12.75,
-      "eval_steps_per_second": 3.446,
+      "eval_accuracy": 0.967741935483871,
+      "eval_loss": 0.1332567036151886,
+      "eval_runtime": 2.0589,
+      "eval_samples_per_second": 15.057,
+      "eval_steps_per_second": 3.886,
       "step": 1725
     },
     {
       "epoch": 23.0,
       "learning_rate": 2.992592592592593e-05,
-      "loss": 0.0009,
+      "loss": 0.0008,
       "step": 1730
     },
     {
@@ -1261,70 +1261,70 @@
     {
       "epoch": 23.01,
       "learning_rate": 2.962962962962963e-05,
-      "loss": 0.0336,
+      "loss": 0.0009,
       "step": 1750
     },
     {
       "epoch": 23.01,
       "learning_rate": 2.9481481481481483e-05,
-      "loss": 0.0044,
+      "loss": 0.0008,
       "step": 1760
     },
     {
       "epoch": 23.01,
       "learning_rate": 2.9333333333333336e-05,
-      "loss": 0.0008,
+      "loss": 0.0009,
       "step": 1770
     },
     {
       "epoch": 23.01,
       "learning_rate": 2.918518518518519e-05,
-      "loss": 0.001,
+      "loss": 0.0007,
       "step": 1780
     },
     {
       "epoch": 23.02,
       "learning_rate": 2.9037037037037042e-05,
-      "loss": 0.0008,
+      "loss": 0.0007,
       "step": 1790
     },
     {
       "epoch": 23.02,
       "learning_rate": 2.8888888888888888e-05,
-      "loss": 0.1521,
+      "loss": 0.0007,
       "step": 1800
     },
     {
       "epoch": 23.02,
-      "eval_accuracy": 0.918918918918919,
-      "eval_loss": 0.48252156376838684,
-      "eval_runtime": 2.9421,
-      "eval_samples_per_second": 12.576,
-      "eval_steps_per_second": 3.399,
+      "eval_accuracy": 0.967741935483871,
+      "eval_loss": 0.10598693042993546,
+      "eval_runtime": 2.0414,
+      "eval_samples_per_second": 15.186,
+      "eval_steps_per_second": 3.919,
       "step": 1800
     },
     {
       "epoch": 24.0,
       "learning_rate": 2.874074074074074e-05,
-      "loss": 0.0023,
+      "loss": 0.0007,
       "step": 1810
     },
     {
       "epoch": 24.01,
       "learning_rate": 2.8592592592592594e-05,
-      "loss": 0.0008,
+      "loss": 0.0007,
       "step": 1820
     },
     {
       "epoch": 24.01,
       "learning_rate": 2.8444444444444447e-05,
-      "loss": 0.0074,
+      "loss": 0.0007,
       "step": 1830
     },
     {
       "epoch": 24.01,
       "learning_rate": 2.8296296296296297e-05,
-      "loss": 0.0009,
+      "loss": 0.0007,
       "step": 1840
     },
     {
@@ -1336,22 +1336,22 @@
     {
       "epoch": 24.02,
       "learning_rate": 2.8000000000000003e-05,
-      "loss": 0.0007,
+      "loss": 0.0006,
       "step": 1860
     },
     {
       "epoch": 24.02,
       "learning_rate": 2.7851851851851853e-05,
-      "loss": 0.001,
+      "loss": 0.0008,
       "step": 1870
     },
     {
       "epoch": 24.02,
-      "eval_accuracy": 0.918918918918919,
-      "eval_loss": 0.6339796185493469,
-      "eval_runtime": 2.8461,
-      "eval_samples_per_second": 13.0,
-      "eval_steps_per_second": 3.514,
+      "eval_accuracy": 0.967741935483871,
+      "eval_loss": 0.1111554503440857,
+      "eval_runtime": 2.0546,
+      "eval_samples_per_second": 15.088,
+      "eval_steps_per_second": 3.894,
       "step": 1875
     },
     {
@@ -1369,97 +1369,97 @@
     {
       "epoch": 25.01,
       "learning_rate": 2.7407407407407408e-05,
-      "loss": 0.0008,
+      "loss": 0.0007,
       "step": 1900
     },
     {
       "epoch": 25.01,
       "learning_rate": 2.725925925925926e-05,
-      "loss": 0.001,
+      "loss": 0.0007,
       "step": 1910
     },
     {
       "epoch": 25.01,
       "learning_rate": 2.7111111111111114e-05,
-      "loss": 0.0007,
+      "loss": 0.0006,
       "step": 1920
     },
     {
       "epoch": 25.01,
       "learning_rate": 2.696296296296296e-05,
-      "loss": 0.0007,
+      "loss": 0.0006,
       "step": 1930
     },
     {
       "epoch": 25.02,
       "learning_rate": 2.6814814814814814e-05,
-      "loss": 0.0007,
+      "loss": 0.0006,
       "step": 1940
     },
     {
       "epoch": 25.02,
       "learning_rate": 2.6666666666666667e-05,
-      "loss": 0.0245,
+      "loss": 0.0006,
       "step": 1950
     },
     {
       "epoch": 25.02,
-      "eval_accuracy": 0.9459459459459459,
-      "eval_loss": 0.3778836727142334,
-      "eval_runtime": 2.8022,
-      "eval_samples_per_second": 13.204,
-      "eval_steps_per_second": 3.569,
+      "eval_accuracy": 1.0,
+      "eval_loss": 0.01604604721069336,
+      "eval_runtime": 2.0655,
+      "eval_samples_per_second": 15.008,
+      "eval_steps_per_second": 3.873,
       "step": 1950
     },
     {
       "epoch": 26.0,
       "learning_rate": 2.651851851851852e-05,
-      "loss": 0.0007,
+      "loss": 0.0006,
       "step": 1960
     },
     {
       "epoch": 26.01,
       "learning_rate": 2.6370370370370373e-05,
-      "loss": 0.0007,
+      "loss": 0.0006,
       "step": 1970
     },
     {
       "epoch": 26.01,
       "learning_rate": 2.6222222222222226e-05,
-      "loss": 0.0007,
+      "loss": 0.0006,
       "step": 1980
     },
     {
       "epoch": 26.01,
       "learning_rate": 2.6074074074074072e-05,
-      "loss": 0.0007,
+      "loss": 0.0006,
       "step": 1990
     },
     {
       "epoch": 26.01,
       "learning_rate": 2.5925925925925925e-05,
-      "loss": 0.0947,
+      "loss": 0.0006,
       "step": 2000
     },
     {
       "epoch": 26.02,
       "learning_rate": 2.5777777777777778e-05,
-      "loss": 0.001,
+      "loss": 0.0006,
       "step": 2010
     },
     {
       "epoch": 26.02,
       "learning_rate": 2.562962962962963e-05,
-      "loss": 0.0007,
+      "loss": 0.0006,
       "step": 2020
     },
     {
       "epoch": 26.02,
-      "eval_accuracy": 0.9459459459459459,
-      "eval_loss": 0.3375699818134308,
-      "eval_runtime": 2.7928,
-      "eval_samples_per_second": 13.248,
-      "eval_steps_per_second": 3.581,
+      "eval_accuracy": 0.967741935483871,
+      "eval_loss": 0.029128754511475563,
+      "eval_runtime": 2.1074,
+      "eval_samples_per_second": 14.71,
+      "eval_steps_per_second": 3.796,
       "step": 2025
     },
     {
@@ -1471,13 +1471,13 @@
     {
       "epoch": 27.0,
       "learning_rate": 2.5333333333333337e-05,
-      "loss": 0.0012,
+      "loss": 0.0006,
       "step": 2040
     },
     {
       "epoch": 27.01,
       "learning_rate": 2.5185185185185183e-05,
-      "loss": 0.0007,
+      "loss": 0.0006,
       "step": 2050
     },
     {
@@ -1489,64 +1489,64 @@
     {
       "epoch": 27.01,
       "learning_rate": 2.488888888888889e-05,
-      "loss": 0.0927,
+      "loss": 0.0006,
       "step": 2070
     },
     {
       "epoch": 27.01,
       "learning_rate": 2.4740740740740742e-05,
-      "loss": 0.0817,
+      "loss": 0.0006,
       "step": 2080
     },
     {
       "epoch": 27.02,
       "learning_rate": 2.4592592592592595e-05,
-      "loss": 0.2909,
+      "loss": 0.0006,
       "step": 2090
     },
     {
       "epoch": 27.02,
       "learning_rate": 2.4444444444444445e-05,
-      "loss": 0.0011,
+      "loss": 0.0006,
       "step": 2100
     },
     {
       "epoch": 27.02,
-      "eval_accuracy": 0.9459459459459459,
-      "eval_loss": 0.28334298729896545,
-      "eval_runtime": 2.7397,
-      "eval_samples_per_second": 13.505,
-      "eval_steps_per_second": 3.65,
+      "eval_accuracy": 0.967741935483871,
+      "eval_loss": 0.08601882308721542,
+      "eval_runtime": 2.1685,
+      "eval_samples_per_second": 14.296,
+      "eval_steps_per_second": 3.689,
       "step": 2100
     },
     {
       "epoch": 28.0,
       "learning_rate": 2.4296296296296298e-05,
-      "loss": 0.0007,
+      "loss": 0.0006,
       "step": 2110
     },
     {
       "epoch": 28.01,
       "learning_rate": 2.414814814814815e-05,
-      "loss": 0.0007,
+      "loss": 0.0006,
       "step": 2120
     },
     {
       "epoch": 28.01,
       "learning_rate": 2.4e-05,
-      "loss": 0.0007,
+      "loss": 0.0006,
       "step": 2130
     },
     {
       "epoch": 28.01,
       "learning_rate": 2.3851851851851854e-05,
-      "loss": 0.0007,
+      "loss": 0.0006,
       "step": 2140
     },
     {
       "epoch": 28.01,
       "learning_rate": 2.3703703703703707e-05,
-      "loss": 0.0007,
+      "loss": 0.0006,
       "step": 2150
     },
     {
@@ -1558,16 +1558,16 @@
     {
       "epoch": 28.02,
       "learning_rate": 2.340740740740741e-05,
-      "loss": 0.0008,
+      "loss": 0.0005,
       "step": 2170
     },
     {
       "epoch": 28.02,
-      "eval_accuracy": 0.972972972972973,
-      "eval_loss": 0.15925714373588562,
-      "eval_runtime": 2.6329,
-      "eval_samples_per_second": 14.053,
-      "eval_steps_per_second": 3.798,
+      "eval_accuracy": 0.967741935483871,
+      "eval_loss": 0.10801166296005249,
+      "eval_runtime": 2.1529,
+      "eval_samples_per_second": 14.399,
+      "eval_steps_per_second": 3.716,
       "step": 2175
     },
     {
@@ -1579,25 +1579,25 @@
     {
       "epoch": 29.0,
       "learning_rate": 2.3111111111111112e-05,
-      "loss": 0.0015,
+      "loss": 0.0005,
       "step": 2190
     },
     {
       "epoch": 29.01,
       "learning_rate": 2.2962962962962965e-05,
-      "loss": 0.0008,
+      "loss": 0.0006,
       "step": 2200
     },
     {
       "epoch": 29.01,
       "learning_rate": 2.2814814814814818e-05,
-      "loss": 0.0007,
+      "loss": 0.0006,
       "step": 2210
     },
     {
       "epoch": 29.01,
       "learning_rate": 2.2666666666666668e-05,
-      "loss": 0.0006,
+      "loss": 0.0005,
       "step": 2220
     },
     {
@@ -1609,28 +1609,28 @@
     {
       "epoch": 29.02,
       "learning_rate": 2.2370370370370374e-05,
-      "loss": 0.0953,
+      "loss": 0.0005,
       "step": 2240
     },
     {
       "epoch": 29.02,
       "learning_rate": 2.2222222222222223e-05,
-      "loss": 0.0008,
+      "loss": 0.0006,
       "step": 2250
     },
     {
       "epoch": 29.02,
-      "eval_accuracy": 0.972972972972973,
-      "eval_loss": 0.08563826233148575,
-      "eval_runtime": 2.7114,
-      "eval_samples_per_second": 13.646,
-      "eval_steps_per_second": 3.688,
+      "eval_accuracy": 0.967741935483871,
+      "eval_loss": 0.11195220053195953,
+      "eval_runtime": 2.1574,
+      "eval_samples_per_second": 14.369,
+      "eval_steps_per_second": 3.708,
       "step": 2250
     },
     {
       "epoch": 30.0,
       "learning_rate": 2.2074074074074076e-05,
-      "loss": 0.0006,
+      "loss": 0.0005,
       "step": 2260
     },
     {
@@ -1642,19 +1642,19 @@
     {
       "epoch": 30.01,
       "learning_rate": 2.177777777777778e-05,
-      "loss": 0.0006,
+      "loss": 0.1392,
       "step": 2280
     },
     {
       "epoch": 30.01,
       "learning_rate": 2.162962962962963e-05,
-      "loss": 0.0006,
+      "loss": 0.0005,
       "step": 2290
     },
     {
       "epoch": 30.01,
       "learning_rate": 2.148148148148148e-05,
-      "loss": 0.0007,
+      "loss": 0.0564,
       "step": 2300
     },
     {
@@ -1666,28 +1666,28 @@
     {
       "epoch": 30.02,
       "learning_rate": 2.1185185185185184e-05,
-      "loss": 0.0005,
+      "loss": 0.0006,
       "step": 2320
     },
     {
       "epoch": 30.02,
-      "eval_accuracy": 0.972972972972973,
-      "eval_loss": 0.1049196645617485,
-      "eval_runtime": 2.6802,
-      "eval_samples_per_second": 13.805,
-      "eval_steps_per_second": 3.731,
+      "eval_accuracy": 0.967741935483871,
+      "eval_loss": 0.05927049741148949,
+      "eval_runtime": 2.1955,
+      "eval_samples_per_second": 14.12,
+      "eval_steps_per_second": 3.644,
       "step": 2325
     },
     {
       "epoch": 31.0,
       "learning_rate": 2.1037037037037037e-05,
-      "loss": 0.0006,
+      "loss": 0.0007,
       "step": 2330
     },
     {
       "epoch": 31.0,
       "learning_rate": 2.088888888888889e-05,
-      "loss": 0.0005,
+      "loss": 0.0379,
       "step": 2340
     },
     {
@@ -1728,11 +1728,11 @@
     },
     {
       "epoch": 31.02,
-      "eval_accuracy": 0.972972972972973,
-      "eval_loss": 0.11323297023773193,
-      "eval_runtime": 3.2205,
-      "eval_samples_per_second": 11.489,
-      "eval_steps_per_second": 3.105,
+      "eval_accuracy": 0.967741935483871,
+      "eval_loss": 0.16602446138858795,
+      "eval_runtime": 2.1541,
+      "eval_samples_per_second": 14.391,
+      "eval_steps_per_second": 3.714,
       "step": 2400
     },
     {
@@ -1750,7 +1750,7 @@
     {
       "epoch": 32.01,
       "learning_rate": 1.9555555555555557e-05,
-      "loss": 0.0005,
+      "loss": 0.0007,
       "step": 2430
     },
     {
@@ -1779,11 +1779,11 @@
     },
     {
       "epoch": 32.02,
-      "eval_accuracy": 0.972972972972973,
-      "eval_loss": 0.11638977378606796,
-      "eval_runtime": 2.5685,
-      "eval_samples_per_second": 14.405,
-      "eval_steps_per_second": 3.893,
+      "eval_accuracy": 0.967741935483871,
+      "eval_loss": 0.04543553665280342,
+      "eval_runtime": 2.5566,
+      "eval_samples_per_second": 12.125,
+      "eval_steps_per_second": 3.129,
       "step": 2475
     },
     {
@@ -1795,7 +1795,7 @@
     {
       "epoch": 33.0,
       "learning_rate": 1.866666666666667e-05,
-      "loss": 0.0005,
+      "loss": 0.0006,
       "step": 2490
     },
     {
@@ -1813,7 +1813,7 @@
     {
       "epoch": 33.01,
       "learning_rate": 1.8222222222222224e-05,
-      "loss": 0.0005,
+      "loss": 0.0009,
       "step": 2520
     },
     {
@@ -1831,16 +1831,16 @@
     {
       "epoch": 33.02,
       "learning_rate": 1.777777777777778e-05,
-      "loss": 0.0005,
+      "loss": 0.0007,
       "step": 2550
     },
     {
       "epoch": 33.02,
-      "eval_accuracy": 0.972972972972973,
-      "eval_loss": 0.12427014112472534,
-      "eval_runtime": 2.5792,
-      "eval_samples_per_second": 14.345,
-      "eval_steps_per_second": 3.877,
+      "eval_accuracy": 0.967741935483871,
+      "eval_loss": 0.1014844998717308,
+      "eval_runtime": 2.1217,
+      "eval_samples_per_second": 14.611,
+      "eval_steps_per_second": 3.771,
       "step": 2550
     },
     {
@@ -1887,11 +1887,11 @@
     },
     {
       "epoch": 34.02,
-      "eval_accuracy": 0.972972972972973,
-      "eval_loss": 0.1306069791316986,
-      "eval_runtime": 2.6024,
-      "eval_samples_per_second": 14.217,
-      "eval_steps_per_second": 3.843,
+      "eval_accuracy": 0.967741935483871,
+      "eval_loss": 0.17122069001197815,
+      "eval_runtime": 2.1289,
+      "eval_samples_per_second": 14.561,
+      "eval_steps_per_second": 3.758,
       "step": 2625
     },
     {
@@ -1915,7 +1915,7 @@
     {
       "epoch": 35.01,
       "learning_rate": 1.614814814814815e-05,
-      "loss": 0.0007,
+      "loss": 0.0005,
       "step": 2660
     },
     {
@@ -1933,7 +1933,7 @@
     {
       "epoch": 35.02,
       "learning_rate": 1.5703703703703705e-05,
-      "loss": 0.0005,
+      "loss": 0.0004,
       "step": 2690
     },
     {
@@ -1944,17 +1944,17 @@
     },
     {
       "epoch": 35.02,
-      "eval_accuracy": 0.9459459459459459,
-      "eval_loss": 0.39192795753479004,
-      "eval_runtime": 2.5502,
-      "eval_samples_per_second": 14.509,
-      "eval_steps_per_second": 3.921,
+      "eval_accuracy": 0.967741935483871,
+      "eval_loss": 0.1599966138601303,
+      "eval_runtime": 2.0848,
+      "eval_samples_per_second": 14.87,
+      "eval_steps_per_second": 3.837,
       "step": 2700
     },
     {
       "epoch": 36.0,
       "learning_rate": 1.5407407407407408e-05,
-      "loss": 0.0005,
+      "loss": 0.0004,
       "step": 2710
     },
     {
@@ -1984,7 +1984,7 @@
     {
       "epoch": 36.02,
       "learning_rate": 1.4666666666666668e-05,
-      "loss": 0.0005,
+      "loss": 0.0004,
       "step": 2760
     },
     {
@@ -1995,17 +1995,17 @@
     },
     {
       "epoch": 36.02,
-      "eval_accuracy": 0.9459459459459459,
-      "eval_loss": 0.36302804946899414,
-      "eval_runtime": 2.4982,
-      "eval_samples_per_second": 14.811,
-      "eval_steps_per_second": 4.003,
+      "eval_accuracy": 0.967741935483871,
+      "eval_loss": 0.16178081929683685,
+      "eval_runtime": 2.0795,
+      "eval_samples_per_second": 14.908,
+      "eval_steps_per_second": 3.847,
       "step": 2775
     },
     {
       "epoch": 37.0,
       "learning_rate": 1.437037037037037e-05,
-      "loss": 0.0005,
+      "loss": 0.0004,
       "step": 2780
     },
     {
@@ -2023,7 +2023,7 @@
     {
       "epoch": 37.01,
       "learning_rate": 1.3925925925925926e-05,
-      "loss": 0.0005,
+      "loss": 0.0004,
       "step": 2810
     },
     {
@@ -2035,13 +2035,13 @@
     {
       "epoch": 37.01,
       "learning_rate": 1.362962962962963e-05,
-      "loss": 0.0005,
+      "loss": 0.0006,
       "step": 2830
     },
     {
       "epoch": 37.02,
       "learning_rate": 1.348148148148148e-05,
-      "loss": 0.0005,
+      "loss": 0.0004,
       "step": 2840
     },
     {
@@ -2052,11 +2052,11 @@
     },
     {
       "epoch": 37.02,
-      "eval_accuracy": 0.9459459459459459,
-      "eval_loss": 0.2762458324432373,
-      "eval_runtime": 2.5203,
-      "eval_samples_per_second": 14.681,
-      "eval_steps_per_second": 3.968,
+      "eval_accuracy": 0.967741935483871,
+      "eval_loss": 0.14675143361091614,
+      "eval_runtime": 2.0875,
+      "eval_samples_per_second": 14.851,
+      "eval_steps_per_second": 3.832,
       "step": 2850
     },
     {
@@ -2086,7 +2086,7 @@
     {
       "epoch": 38.01,
       "learning_rate": 1.2592592592592592e-05,
-      "loss": 0.0004,
+      "loss": 0.0005,
       "step": 2900
     },
     {
@@ -2098,16 +2098,16 @@
     {
       "epoch": 38.02,
       "learning_rate": 1.2296296296296298e-05,
-      "loss": 0.0005,
+      "loss": 0.0004,
       "step": 2920
     },
     {
       "epoch": 38.02,
-      "eval_accuracy": 0.9459459459459459,
-      "eval_loss": 0.23680266737937927,
-      "eval_runtime": 2.5362,
-      "eval_samples_per_second": 14.589,
-      "eval_steps_per_second": 3.943,
+      "eval_accuracy": 0.967741935483871,
+      "eval_loss": 0.11672930419445038,
+      "eval_runtime": 2.0961,
+      "eval_samples_per_second": 14.79,
+      "eval_steps_per_second": 3.817,
       "step": 2925
     },
     {
@@ -2125,13 +2125,13 @@
     {
       "epoch": 39.01,
       "learning_rate": 1.1851851851851853e-05,
-      "loss": 0.0005,
+      "loss": 0.0004,
       "step": 2950
     },
     {
       "epoch": 39.01,
       "learning_rate": 1.1703703703703705e-05,
-      "loss": 0.0005,
+      "loss": 0.0004,
       "step": 2960
     },
     {
@@ -2160,11 +2160,11 @@
     },
     {
       "epoch": 39.02,
-      "eval_accuracy": 0.972972972972973,
-      "eval_loss": 0.1935373693704605,
-      "eval_runtime": 2.5511,
-      "eval_samples_per_second": 14.504,
-      "eval_steps_per_second": 3.92,
+      "eval_accuracy": 0.967741935483871,
+      "eval_loss": 0.12776704132556915,
+      "eval_runtime": 2.1378,
+      "eval_samples_per_second": 14.501,
+      "eval_steps_per_second": 3.742,
       "step": 3000
     },
     {
@@ -2211,11 +2211,11 @@
     },
     {
       "epoch": 40.02,
-      "eval_accuracy": 0.972972972972973,
-      "eval_loss": 0.19308657944202423,
-      "eval_runtime": 2.5348,
-      "eval_samples_per_second": 14.597,
-      "eval_steps_per_second": 3.945,
+      "eval_accuracy": 0.967741935483871,
+      "eval_loss": 0.12002281844615936,
+      "eval_runtime": 2.1121,
+      "eval_samples_per_second": 14.677,
+      "eval_steps_per_second": 3.788,
       "step": 3075
     },
     {
@@ -2227,7 +2227,7 @@
     {
       "epoch": 41.0,
       "learning_rate": 9.777777777777779e-06,
-      "loss": 0.0085,
+      "loss": 0.0004,
       "step": 3090
     },
     {
@@ -2257,7 +2257,7 @@
     {
       "epoch": 41.02,
       "learning_rate": 9.037037037037037e-06,
-      "loss": 0.0005,
+      "loss": 0.0004,
       "step": 3140
     },
     {
@@ -2268,11 +2268,11 @@
     },
     {
       "epoch": 41.02,
-      "eval_accuracy": 0.9459459459459459,
-      "eval_loss": 0.21387839317321777,
-      "eval_runtime": 2.5103,
-      "eval_samples_per_second": 14.739,
-      "eval_steps_per_second": 3.984,
+      "eval_accuracy": 0.967741935483871,
+      "eval_loss": 0.12003983557224274,
+      "eval_runtime": 2.0588,
+      "eval_samples_per_second": 15.057,
+      "eval_steps_per_second": 3.886,
       "step": 3150
     },
     {
@@ -2284,7 +2284,7 @@
     {
       "epoch": 42.01,
       "learning_rate": 8.592592592592593e-06,
-      "loss": 0.0005,
+      "loss": 0.0004,
       "step": 3170
     },
     {
@@ -2319,17 +2319,17 @@
     },
     {
       "epoch": 42.02,
-      "eval_accuracy": 0.972972972972973,
-      "eval_loss": 0.1899683177471161,
-      "eval_runtime": 2.6252,
-      "eval_samples_per_second": 14.094,
-      "eval_steps_per_second": 3.809,
+      "eval_accuracy": 0.967741935483871,
+      "eval_loss": 0.12297246605157852,
+      "eval_runtime": 2.0736,
+      "eval_samples_per_second": 14.95,
+      "eval_steps_per_second": 3.858,
       "step": 3225
     },
     {
       "epoch": 43.0,
       "learning_rate": 7.703703703703704e-06,
-      "loss": 0.0015,
+      "loss": 0.0004,
       "step": 3230
     },
     {
@@ -2371,16 +2371,16 @@
     {
       "epoch": 43.02,
       "learning_rate": 6.666666666666667e-06,
-      "loss": 0.0006,
+      "loss": 0.0004,
       "step": 3300
     },
     {
       "epoch": 43.02,
-      "eval_accuracy": 0.972972972972973,
-      "eval_loss": 0.1750660091638565,
-      "eval_runtime": 2.5683,
-      "eval_samples_per_second": 14.407,
-      "eval_steps_per_second": 3.894,
+      "eval_accuracy": 0.967741935483871,
+      "eval_loss": 0.13225598633289337,
+      "eval_runtime": 2.0961,
+      "eval_samples_per_second": 14.789,
+      "eval_steps_per_second": 3.817,
       "step": 3300
     },
     {
@@ -2427,11 +2427,11 @@
     },
     {
       "epoch": 44.02,
-      "eval_accuracy": 0.9459459459459459,
-      "eval_loss": 0.29775792360305786,
-      "eval_runtime": 2.6767,
-      "eval_samples_per_second": 13.823,
-      "eval_steps_per_second": 3.736,
+      "eval_accuracy": 0.967741935483871,
+      "eval_loss": 0.1282513439655304,
+      "eval_runtime": 2.1105,
+      "eval_samples_per_second": 14.688,
+      "eval_steps_per_second": 3.791,
       "step": 3375
     },
     {
@@ -2484,11 +2484,11 @@
     },
     {
       "epoch": 45.02,
-      "eval_accuracy": 0.9459459459459459,
-      "eval_loss": 0.2776608467102051,
-      "eval_runtime": 2.5843,
-      "eval_samples_per_second": 14.317,
-      "eval_steps_per_second": 3.87,
+      "eval_accuracy": 0.967741935483871,
+      "eval_loss": 0.1329645812511444,
+      "eval_runtime": 2.079,
+      "eval_samples_per_second": 14.911,
+      "eval_steps_per_second": 3.848,
       "step": 3450
     },
     {
@@ -2535,11 +2535,11 @@
     },
     {
       "epoch": 46.02,
-      "eval_accuracy": 0.9459459459459459,
-      "eval_loss": 0.2706151008605957,
-      "eval_runtime": 2.6031,
-      "eval_samples_per_second": 14.214,
-      "eval_steps_per_second": 3.842,
+      "eval_accuracy": 0.967741935483871,
+      "eval_loss": 0.13405713438987732,
+      "eval_runtime": 2.0851,
+      "eval_samples_per_second": 14.867,
+      "eval_steps_per_second": 3.837,
       "step": 3525
     },
     {
@@ -2592,11 +2592,11 @@
     },
     {
       "epoch": 47.02,
-      "eval_accuracy": 0.9459459459459459,
-      "eval_loss": 0.26377302408218384,
-      "eval_runtime": 2.5538,
-      "eval_samples_per_second": 14.488,
-      "eval_steps_per_second": 3.916,
+      "eval_accuracy": 0.967741935483871,
+      "eval_loss": 0.13054759800434113,
+      "eval_runtime": 2.2343,
+      "eval_samples_per_second": 13.875,
+      "eval_steps_per_second": 3.581,
       "step": 3600
     },
     {
@@ -2608,7 +2608,7 @@
     {
       "epoch": 48.01,
       "learning_rate": 1.925925925925926e-06,
-      "loss": 0.0003,
+      "loss": 0.0004,
       "step": 3620
     },
     {
@@ -2620,7 +2620,7 @@
     {
       "epoch": 48.01,
       "learning_rate": 1.6296296296296295e-06,
-      "loss": 0.0005,
+      "loss": 0.0003,
       "step": 3640
     },
     {
@@ -2643,11 +2643,11 @@
     },
     {
       "epoch": 48.02,
-      "eval_accuracy": 0.9459459459459459,
-      "eval_loss": 0.21225357055664062,
-      "eval_runtime": 2.6586,
-      "eval_samples_per_second": 13.917,
-      "eval_steps_per_second": 3.761,
+      "eval_accuracy": 0.967741935483871,
+      "eval_loss": 0.1309322565793991,
+      "eval_runtime": 2.2018,
+      "eval_samples_per_second": 14.079,
+      "eval_steps_per_second": 3.633,
       "step": 3675
     },
     {
@@ -2700,38 +2700,38 @@
     },
     {
       "epoch": 49.02,
-      "eval_accuracy": 0.9459459459459459,
-      "eval_loss": 0.21062754094600677,
-      "eval_runtime": 2.7099,
-      "eval_samples_per_second": 13.653,
-      "eval_steps_per_second": 3.69,
+      "eval_accuracy": 0.967741935483871,
+      "eval_loss": 0.1303798407316208,
+      "eval_runtime": 2.1527,
+      "eval_samples_per_second": 14.4,
+      "eval_steps_per_second": 3.716,
       "step": 3750
     },
     {
       "epoch": 49.02,
       "step": 3750,
       "total_flos": 1.86923023515648e+19,
-      "train_loss": 0.16172246270999313,
-      "train_runtime": 3621.2802,
-      "train_samples_per_second": 4.142,
-      "train_steps_per_second": 1.036
+      "train_loss": 0.16177092336689433,
+      "train_runtime": 3695.0497,
+      "train_samples_per_second": 4.059,
+      "train_steps_per_second": 1.015
     },
     {
       "epoch": 49.02,
-      "eval_accuracy": 0.8850574712643678,
-      "eval_loss": 0.38414710760116577,
-      "eval_runtime": 8.6129,
-      "eval_samples_per_second": 10.101,
-      "eval_steps_per_second": 2.554,
+      "eval_accuracy": 0.9078947368421053,
+      "eval_loss": 0.4171188473701477,
+      "eval_runtime": 7.7151,
+      "eval_samples_per_second": 9.851,
+      "eval_steps_per_second": 2.463,
       "step": 3750
     },
     {
       "epoch": 49.02,
-      "eval_accuracy": 0.8850574712643678,
-      "eval_loss": 0.38414719700813293,
-      "eval_runtime": 6.1302,
-      "eval_samples_per_second": 14.192,
-      "eval_steps_per_second": 3.589,
+      "eval_accuracy": 0.9078947368421053,
+      "eval_loss": 0.41711878776550293,
+      "eval_runtime": 5.3059,
+      "eval_samples_per_second": 14.324,
+      "eval_steps_per_second": 3.581,
       "step": 3750
     }
   ],

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:54be8404f3a06187402196bf7251b8491aa663c5d9c26aa1c6f1e5be6bcb9d33
+oid sha256:03b467334c18ac42b9abe41982148898c6d99bf9fffb9c66a198c9df721fe5c8
 size 3439