Training in progress, epoch 0

all_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 49.02,
-    "eval_accuracy": 0.9290322580645162,
-    "eval_loss": 0.18336611986160278,
-    "eval_runtime": 12.5538,
-    "eval_samples_per_second": 12.347,
-    "eval_steps_per_second": 3.107
+    "eval_accuracy": 0.8850574712643678,
+    "eval_loss": 0.38414719700813293,
+    "eval_runtime": 6.1302,
+    "eval_samples_per_second": 14.192,
+    "eval_steps_per_second": 3.589
 }

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9cb154e0d13e9e3d27632b626366e31b3a8cc4aaf395306ec53c67bb9cf8f30f
-size 345004516
+oid sha256:4883b0eb29ea209f1f563a296a757202cbed167c64820f562d4ec4df2c54d1a9
+size 345004552

test_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 49.02,
-    "eval_accuracy": 0.9290322580645162,
-    "eval_loss": 0.18336611986160278,
-    "eval_runtime": 12.5538,
-    "eval_samples_per_second": 12.347,
-    "eval_steps_per_second": 3.107
+    "eval_accuracy": 0.8850574712643678,
+    "eval_loss": 0.38414719700813293,
+    "eval_runtime": 6.1302,
+    "eval_samples_per_second": 14.192,
+    "eval_steps_per_second": 3.589
 }

trainer_state.json

@@ -1,6 +1,6 @@
 {
-  "best_metric": 0.9857142857142858,
-  "best_model_checkpoint": "videomae-base-finetuned-ucf101-subset/checkpoint-1425",
+  "best_metric": 0.972972972972973,
+  "best_model_checkpoint": "videomae-base-finetuned-ucf101-subset/checkpoint-375",
   "epoch": 49.02,
   "global_step": 3750,
   "is_hyper_param_search": false,
@@ -10,883 +10,883 @@
     {
       "epoch": 0.0,
       "learning_rate": 1.3333333333333334e-06,
-      "loss": 2.3169,
+      "loss": 2.4049,
       "step": 10
     },
     {
       "epoch": 0.01,
       "learning_rate": 2.666666666666667e-06,
-      "loss": 2.3252,
+      "loss": 2.3782,
       "step": 20
     },
     {
       "epoch": 0.01,
       "learning_rate": 4.000000000000001e-06,
-      "loss": 2.2807,
+      "loss": 2.3345,
       "step": 30
     },
     {
       "epoch": 0.01,
       "learning_rate": 5.333333333333334e-06,
-      "loss": 2.2992,
+      "loss": 2.3762,
       "step": 40
     },
     {
       "epoch": 0.01,
       "learning_rate": 6.666666666666667e-06,
-      "loss": 2.2765,
+      "loss": 2.3277,
       "step": 50
     },
     {
       "epoch": 0.02,
       "learning_rate": 8.000000000000001e-06,
-      "loss": 2.2692,
+      "loss": 2.2711,
       "step": 60
     },
     {
       "epoch": 0.02,
       "learning_rate": 9.333333333333334e-06,
-      "loss": 2.2357,
+      "loss": 2.369,
       "step": 70
     },
     {
       "epoch": 0.02,
-      "eval_accuracy": 0.34285714285714286,
-      "eval_loss": 2.195152759552002,
-      "eval_runtime": 5.4445,
-      "eval_samples_per_second": 12.857,
-      "eval_steps_per_second": 3.306,
+      "eval_accuracy": 0.2972972972972973,
+      "eval_loss": 2.2216224670410156,
+      "eval_runtime": 2.6845,
+      "eval_samples_per_second": 13.783,
+      "eval_steps_per_second": 3.725,
       "step": 75
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.0666666666666667e-05,
-      "loss": 2.2979,
+      "loss": 2.2595,
       "step": 80
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.2e-05,
-      "loss": 2.1564,
+      "loss": 2.2602,
       "step": 90
     },
     {
       "epoch": 1.01,
       "learning_rate": 1.3333333333333333e-05,
-      "loss": 2.1848,
+      "loss": 2.1993,
       "step": 100
     },
     {
       "epoch": 1.01,
       "learning_rate": 1.4666666666666668e-05,
-      "loss": 2.072,
+      "loss": 2.2198,
       "step": 110
     },
     {
       "epoch": 1.01,
       "learning_rate": 1.6000000000000003e-05,
-      "loss": 2.0733,
+      "loss": 2.1327,
       "step": 120
     },
     {
       "epoch": 1.01,
       "learning_rate": 1.7333333333333336e-05,
-      "loss": 2.0719,
+      "loss": 2.0154,
       "step": 130
     },
     {
       "epoch": 1.02,
       "learning_rate": 1.866666666666667e-05,
-      "loss": 2.0285,
+      "loss": 2.0248,
       "step": 140
     },
     {
       "epoch": 1.02,
       "learning_rate": 2e-05,
-      "loss": 1.8271,
+      "loss": 1.8283,
       "step": 150
     },
     {
       "epoch": 1.02,
-      "eval_accuracy": 0.35714285714285715,
-      "eval_loss": 1.8570266962051392,
-      "eval_runtime": 5.776,
-      "eval_samples_per_second": 12.119,
-      "eval_steps_per_second": 3.116,
+      "eval_accuracy": 0.4864864864864865,
+      "eval_loss": 1.7583507299423218,
+      "eval_runtime": 2.7247,
+      "eval_samples_per_second": 13.58,
+      "eval_steps_per_second": 3.67,
       "step": 150
     },
     {
       "epoch": 2.0,
       "learning_rate": 2.1333333333333335e-05,
-      "loss": 1.6139,
+      "loss": 1.5075,
       "step": 160
     },
     {
       "epoch": 2.01,
       "learning_rate": 2.2666666666666668e-05,
-      "loss": 1.424,
+      "loss": 1.4104,
       "step": 170
     },
     {
       "epoch": 2.01,
       "learning_rate": 2.4e-05,
-      "loss": 1.3944,
+      "loss": 1.1892,
       "step": 180
     },
     {
       "epoch": 2.01,
       "learning_rate": 2.5333333333333337e-05,
-      "loss": 1.1121,
+      "loss": 0.9929,
       "step": 190
     },
     {
       "epoch": 2.01,
       "learning_rate": 2.6666666666666667e-05,
-      "loss": 1.0835,
+      "loss": 0.8859,
       "step": 200
     },
     {
       "epoch": 2.02,
       "learning_rate": 2.8000000000000003e-05,
-      "loss": 0.9839,
+      "loss": 0.7763,
       "step": 210
     },
     {
       "epoch": 2.02,
       "learning_rate": 2.9333333333333336e-05,
-      "loss": 0.8947,
+      "loss": 0.8729,
       "step": 220
     },
     {
       "epoch": 2.02,
-      "eval_accuracy": 0.7285714285714285,
-      "eval_loss": 0.839695155620575,
-      "eval_runtime": 5.7359,
-      "eval_samples_per_second": 12.204,
-      "eval_steps_per_second": 3.138,
+      "eval_accuracy": 0.7027027027027027,
+      "eval_loss": 1.0192049741744995,
+      "eval_runtime": 2.6893,
+      "eval_samples_per_second": 13.758,
+      "eval_steps_per_second": 3.718,
       "step": 225
     },
     {
       "epoch": 3.0,
       "learning_rate": 3.066666666666667e-05,
-      "loss": 0.8324,
+      "loss": 0.7605,
       "step": 230
     },
     {
       "epoch": 3.0,
       "learning_rate": 3.2000000000000005e-05,
-      "loss": 0.6889,
+      "loss": 0.709,
       "step": 240
     },
     {
       "epoch": 3.01,
       "learning_rate": 3.3333333333333335e-05,
-      "loss": 0.6668,
+      "loss": 0.4213,
       "step": 250
     },
     {
       "epoch": 3.01,
       "learning_rate": 3.466666666666667e-05,
-      "loss": 0.6746,
+      "loss": 0.6716,
       "step": 260
     },
     {
       "epoch": 3.01,
       "learning_rate": 3.6e-05,
-      "loss": 0.7853,
+      "loss": 0.7687,
       "step": 270
     },
     {
       "epoch": 3.01,
       "learning_rate": 3.733333333333334e-05,
-      "loss": 0.3688,
+      "loss": 0.4992,
       "step": 280
     },
     {
       "epoch": 3.02,
       "learning_rate": 3.866666666666667e-05,
-      "loss": 0.4888,
+      "loss": 0.5896,
       "step": 290
     },
     {
       "epoch": 3.02,
       "learning_rate": 4e-05,
-      "loss": 0.5347,
+      "loss": 0.4077,
       "step": 300
     },
     {
       "epoch": 3.02,
-      "eval_accuracy": 0.8285714285714286,
-      "eval_loss": 0.5631577968597412,
-      "eval_runtime": 6.1294,
-      "eval_samples_per_second": 11.42,
-      "eval_steps_per_second": 2.937,
+      "eval_accuracy": 0.8378378378378378,
+      "eval_loss": 0.48492956161499023,
+      "eval_runtime": 2.6873,
+      "eval_samples_per_second": 13.768,
+      "eval_steps_per_second": 3.721,
       "step": 300
     },
     {
       "epoch": 4.0,
       "learning_rate": 4.133333333333333e-05,
-      "loss": 0.4236,
+      "loss": 0.2703,
       "step": 310
     },
     {
       "epoch": 4.01,
       "learning_rate": 4.266666666666667e-05,
-      "loss": 0.4206,
+      "loss": 0.3431,
       "step": 320
     },
     {
       "epoch": 4.01,
       "learning_rate": 4.4000000000000006e-05,
-      "loss": 0.4133,
+      "loss": 0.4284,
       "step": 330
     },
     {
       "epoch": 4.01,
       "learning_rate": 4.5333333333333335e-05,
-      "loss": 0.6152,
+      "loss": 0.5786,
       "step": 340
     },
     {
       "epoch": 4.01,
       "learning_rate": 4.666666666666667e-05,
-      "loss": 0.7022,
+      "loss": 0.5347,
       "step": 350
     },
     {
       "epoch": 4.02,
       "learning_rate": 4.8e-05,
-      "loss": 0.4414,
+      "loss": 0.2638,
       "step": 360
     },
     {
       "epoch": 4.02,
       "learning_rate": 4.933333333333334e-05,
-      "loss": 0.4101,
+      "loss": 0.3742,
       "step": 370
     },
     {
       "epoch": 4.02,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.690824031829834,
-      "eval_runtime": 5.752,
-      "eval_samples_per_second": 12.17,
-      "eval_steps_per_second": 3.129,
+      "eval_accuracy": 0.972972972972973,
+      "eval_loss": 0.1343977451324463,
+      "eval_runtime": 2.7118,
+      "eval_samples_per_second": 13.644,
+      "eval_steps_per_second": 3.688,
       "step": 375
     },
     {
       "epoch": 5.0,
       "learning_rate": 4.9925925925925926e-05,
-      "loss": 0.1599,
+      "loss": 0.2036,
       "step": 380
     },
     {
       "epoch": 5.0,
       "learning_rate": 4.977777777777778e-05,
-      "loss": 0.3312,
+      "loss": 0.1899,
       "step": 390
     },
     {
       "epoch": 5.01,
       "learning_rate": 4.962962962962963e-05,
-      "loss": 0.3967,
+      "loss": 0.3761,
       "step": 400
     },
     {
       "epoch": 5.01,
       "learning_rate": 4.9481481481481485e-05,
-      "loss": 0.3926,
+      "loss": 0.3704,
       "step": 410
     },
     {
       "epoch": 5.01,
       "learning_rate": 4.933333333333334e-05,
-      "loss": 0.8443,
+      "loss": 0.2038,
       "step": 420
     },
     {
       "epoch": 5.01,
       "learning_rate": 4.918518518518519e-05,
-      "loss": 0.4227,
+      "loss": 0.4751,
       "step": 430
     },
     {
       "epoch": 5.02,
       "learning_rate": 4.903703703703704e-05,
-      "loss": 0.2168,
+      "loss": 0.1951,
       "step": 440
     },
     {
       "epoch": 5.02,
       "learning_rate": 4.888888888888889e-05,
-      "loss": 0.0855,
+      "loss": 0.094,
       "step": 450
     },
     {
       "epoch": 5.02,
-      "eval_accuracy": 0.9571428571428572,
-      "eval_loss": 0.1541331708431244,
-      "eval_runtime": 5.9819,
-      "eval_samples_per_second": 11.702,
-      "eval_steps_per_second": 3.009,
+      "eval_accuracy": 0.8918918918918919,
+      "eval_loss": 0.24485082924365997,
+      "eval_runtime": 2.7174,
+      "eval_samples_per_second": 13.616,
+      "eval_steps_per_second": 3.68,
       "step": 450
     },
     {
       "epoch": 6.0,
       "learning_rate": 4.874074074074074e-05,
-      "loss": 0.0751,
+      "loss": 0.1336,
       "step": 460
     },
     {
       "epoch": 6.01,
       "learning_rate": 4.8592592592592596e-05,
-      "loss": 0.2454,
+      "loss": 0.0813,
       "step": 470
     },
     {
       "epoch": 6.01,
       "learning_rate": 4.844444444444445e-05,
-      "loss": 0.3209,
+      "loss": 0.1722,
       "step": 480
     },
     {
       "epoch": 6.01,
       "learning_rate": 4.82962962962963e-05,
-      "loss": 0.3679,
+      "loss": 0.5579,
       "step": 490
     },
     {
       "epoch": 6.01,
       "learning_rate": 4.814814814814815e-05,
-      "loss": 0.2122,
+      "loss": 0.3787,
       "step": 500
     },
     {
       "epoch": 6.02,
       "learning_rate": 4.8e-05,
-      "loss": 0.1339,
+      "loss": 0.5287,
       "step": 510
     },
     {
       "epoch": 6.02,
       "learning_rate": 4.7851851851851854e-05,
-      "loss": 0.2286,
+      "loss": 0.1005,
       "step": 520
     },
     {
       "epoch": 6.02,
-      "eval_accuracy": 0.9,
-      "eval_loss": 0.2762179374694824,
-      "eval_runtime": 5.6813,
-      "eval_samples_per_second": 12.321,
-      "eval_steps_per_second": 3.168,
+      "eval_accuracy": 0.7837837837837838,
+      "eval_loss": 1.0794074535369873,
+      "eval_runtime": 2.7101,
+      "eval_samples_per_second": 13.653,
+      "eval_steps_per_second": 3.69,
       "step": 525
     },
     {
       "epoch": 7.0,
       "learning_rate": 4.770370370370371e-05,
-      "loss": 0.3897,
+      "loss": 0.5154,
       "step": 530
     },
     {
       "epoch": 7.0,
       "learning_rate": 4.755555555555556e-05,
-      "loss": 0.0845,
+      "loss": 0.1985,
       "step": 540
     },
     {
       "epoch": 7.01,
       "learning_rate": 4.740740740740741e-05,
-      "loss": 0.3024,
+      "loss": 0.2665,
       "step": 550
     },
     {
       "epoch": 7.01,
       "learning_rate": 4.7259259259259266e-05,
-      "loss": 0.0144,
+      "loss": 0.0321,
       "step": 560
     },
     {
       "epoch": 7.01,
       "learning_rate": 4.711111111111111e-05,
-      "loss": 0.4922,
+      "loss": 0.159,
       "step": 570
     },
     {
       "epoch": 7.01,
       "learning_rate": 4.6962962962962966e-05,
-      "loss": 0.0182,
+      "loss": 0.3711,
       "step": 580
     },
     {
       "epoch": 7.02,
       "learning_rate": 4.681481481481482e-05,
-      "loss": 0.0056,
+      "loss": 0.0089,
       "step": 590
     },
     {
       "epoch": 7.02,
       "learning_rate": 4.666666666666667e-05,
-      "loss": 0.2905,
+      "loss": 0.0053,
       "step": 600
     },
     {
       "epoch": 7.02,
-      "eval_accuracy": 0.9,
-      "eval_loss": 0.23063896596431732,
-      "eval_runtime": 5.9851,
-      "eval_samples_per_second": 11.696,
-      "eval_steps_per_second": 3.007,
+      "eval_accuracy": 0.9459459459459459,
+      "eval_loss": 0.23636393249034882,
+      "eval_runtime": 2.6811,
+      "eval_samples_per_second": 13.8,
+      "eval_steps_per_second": 3.73,
       "step": 600
     },
     {
       "epoch": 8.0,
       "learning_rate": 4.6518518518518525e-05,
-      "loss": 0.0139,
+      "loss": 0.0056,
       "step": 610
     },
     {
       "epoch": 8.01,
       "learning_rate": 4.637037037037038e-05,
-      "loss": 0.3338,
+      "loss": 0.3093,
       "step": 620
     },
     {
       "epoch": 8.01,
       "learning_rate": 4.6222222222222224e-05,
-      "loss": 0.0218,
+      "loss": 0.3216,
       "step": 630
     },
     {
       "epoch": 8.01,
       "learning_rate": 4.607407407407408e-05,
-      "loss": 0.0053,
+      "loss": 0.1304,
       "step": 640
     },
     {
       "epoch": 8.01,
       "learning_rate": 4.592592592592593e-05,
-      "loss": 0.0042,
+      "loss": 0.148,
       "step": 650
     },
     {
       "epoch": 8.02,
       "learning_rate": 4.577777777777778e-05,
-      "loss": 0.0115,
+      "loss": 0.132,
       "step": 660
     },
     {
       "epoch": 8.02,
       "learning_rate": 4.5629629629629636e-05,
-      "loss": 0.0051,
+      "loss": 0.0807,
       "step": 670
     },
     {
       "epoch": 8.02,
-      "eval_accuracy": 0.9571428571428572,
-      "eval_loss": 0.2053740918636322,
-      "eval_runtime": 5.9611,
-      "eval_samples_per_second": 11.743,
-      "eval_steps_per_second": 3.02,
+      "eval_accuracy": 0.8378378378378378,
+      "eval_loss": 0.6658951640129089,
+      "eval_runtime": 2.6902,
+      "eval_samples_per_second": 13.754,
+      "eval_steps_per_second": 3.717,
       "step": 675
     },
     {
       "epoch": 9.0,
       "learning_rate": 4.548148148148149e-05,
-      "loss": 0.0034,
+      "loss": 0.0775,
       "step": 680
     },
     {
       "epoch": 9.0,
       "learning_rate": 4.5333333333333335e-05,
-      "loss": 0.006,
+      "loss": 0.1351,
       "step": 690
     },
     {
       "epoch": 9.01,
       "learning_rate": 4.518518518518519e-05,
-      "loss": 0.0663,
+      "loss": 0.1521,
       "step": 700
     },
     {
       "epoch": 9.01,
       "learning_rate": 4.503703703703704e-05,
-      "loss": 0.0074,
+      "loss": 0.4525,
       "step": 710
     },
     {
       "epoch": 9.01,
       "learning_rate": 4.4888888888888894e-05,
-      "loss": 0.1151,
+      "loss": 0.4013,
       "step": 720
     },
     {
       "epoch": 9.01,
       "learning_rate": 4.474074074074075e-05,
-      "loss": 0.0627,
+      "loss": 0.0261,
       "step": 730
     },
     {
       "epoch": 9.02,
       "learning_rate": 4.4592592592592594e-05,
-      "loss": 0.016,
+      "loss": 0.0091,
       "step": 740
     },
     {
       "epoch": 9.02,
       "learning_rate": 4.4444444444444447e-05,
-      "loss": 0.1142,
+      "loss": 0.0031,
       "step": 750
     },
     {
       "epoch": 9.02,
-      "eval_accuracy": 0.8714285714285714,
-      "eval_loss": 0.704865038394928,
-      "eval_runtime": 5.6807,
-      "eval_samples_per_second": 12.323,
-      "eval_steps_per_second": 3.169,
+      "eval_accuracy": 0.918918918918919,
+      "eval_loss": 0.4496133029460907,
+      "eval_runtime": 2.7338,
+      "eval_samples_per_second": 13.534,
+      "eval_steps_per_second": 3.658,
       "step": 750
     },
     {
       "epoch": 10.0,
       "learning_rate": 4.42962962962963e-05,
-      "loss": 0.0937,
+      "loss": 0.0035,
       "step": 760
     },
     {
       "epoch": 10.01,
       "learning_rate": 4.414814814814815e-05,
-      "loss": 0.1874,
+      "loss": 0.0199,
       "step": 770
     },
     {
       "epoch": 10.01,
       "learning_rate": 4.4000000000000006e-05,
-      "loss": 0.0798,
+      "loss": 0.0028,
       "step": 780
     },
     {
       "epoch": 10.01,
       "learning_rate": 4.385185185185185e-05,
-      "loss": 0.0028,
+      "loss": 0.0021,
       "step": 790
     },
     {
       "epoch": 10.01,
       "learning_rate": 4.3703703703703705e-05,
-      "loss": 0.1547,
+      "loss": 0.0035,
       "step": 800
     },
     {
       "epoch": 10.02,
       "learning_rate": 4.355555555555556e-05,
-      "loss": 0.1714,
+      "loss": 0.0025,
       "step": 810
     },
     {
       "epoch": 10.02,
       "learning_rate": 4.340740740740741e-05,
-      "loss": 0.0022,
+      "loss": 0.0203,
       "step": 820
     },
     {
       "epoch": 10.02,
-      "eval_accuracy": 0.9428571428571428,
-      "eval_loss": 0.1919040083885193,
-      "eval_runtime": 5.707,
-      "eval_samples_per_second": 12.266,
-      "eval_steps_per_second": 3.154,
+      "eval_accuracy": 0.918918918918919,
+      "eval_loss": 0.3398858308792114,
+      "eval_runtime": 2.8361,
+      "eval_samples_per_second": 13.046,
+      "eval_steps_per_second": 3.526,
       "step": 825
     },
     {
       "epoch": 11.0,
       "learning_rate": 4.325925925925926e-05,
-      "loss": 0.0026,
+      "loss": 0.0042,
       "step": 830
     },
     {
       "epoch": 11.0,
       "learning_rate": 4.311111111111111e-05,
-      "loss": 0.0029,
+      "loss": 0.0024,
       "step": 840
     },
     {
       "epoch": 11.01,
       "learning_rate": 4.296296296296296e-05,
-      "loss": 0.0794,
+      "loss": 0.0023,
       "step": 850
     },
     {
       "epoch": 11.01,
       "learning_rate": 4.2814814814814816e-05,
-      "loss": 0.0708,
+      "loss": 0.0266,
       "step": 860
     },
     {
       "epoch": 11.01,
       "learning_rate": 4.266666666666667e-05,
-      "loss": 0.0705,
+      "loss": 0.018,
       "step": 870
     },
     {
       "epoch": 11.01,
       "learning_rate": 4.2518518518518515e-05,
-      "loss": 0.0549,
+      "loss": 0.0028,
       "step": 880
     },
     {
       "epoch": 11.02,
       "learning_rate": 4.237037037037037e-05,
-      "loss": 0.0022,
+      "loss": 0.2435,
       "step": 890
     },
     {
       "epoch": 11.02,
       "learning_rate": 4.222222222222222e-05,
-      "loss": 0.0019,
+      "loss": 0.0093,
       "step": 900
     },
     {
       "epoch": 11.02,
-      "eval_accuracy": 0.8857142857142857,
-      "eval_loss": 0.5478063225746155,
-      "eval_runtime": 5.7039,
-      "eval_samples_per_second": 12.272,
-      "eval_steps_per_second": 3.156,
+      "eval_accuracy": 0.9459459459459459,
+      "eval_loss": 0.3724738359451294,
+      "eval_runtime": 2.729,
+      "eval_samples_per_second": 13.558,
+      "eval_steps_per_second": 3.664,
       "step": 900
     },
     {
       "epoch": 12.0,
       "learning_rate": 4.2074074074074075e-05,
-      "loss": 0.0032,
+      "loss": 0.1554,
       "step": 910
     },
     {
       "epoch": 12.01,
       "learning_rate": 4.192592592592593e-05,
-      "loss": 0.0159,
+      "loss": 0.0049,
       "step": 920
     },
     {
       "epoch": 12.01,
       "learning_rate": 4.177777777777778e-05,
-      "loss": 0.0861,
+      "loss": 0.1019,
       "step": 930
     },
     {
       "epoch": 12.01,
       "learning_rate": 4.162962962962963e-05,
-      "loss": 0.2058,
+      "loss": 0.2167,
       "step": 940
     },
     {
       "epoch": 12.01,
       "learning_rate": 4.148148148148148e-05,
-      "loss": 0.0042,
+      "loss": 0.1541,
       "step": 950
     },
     {
       "epoch": 12.02,
       "learning_rate": 4.133333333333333e-05,
-      "loss": 0.1814,
+      "loss": 0.0032,
       "step": 960
     },
     {
       "epoch": 12.02,
       "learning_rate": 4.1185185185185186e-05,
-      "loss": 0.0021,
+      "loss": 0.0022,
       "step": 970
     },
     {
       "epoch": 12.02,
-      "eval_accuracy": 0.9285714285714286,
-      "eval_loss": 0.4232262670993805,
-      "eval_runtime": 5.5509,
-      "eval_samples_per_second": 12.611,
-      "eval_steps_per_second": 3.243,
+      "eval_accuracy": 0.918918918918919,
+      "eval_loss": 0.5498412847518921,
+      "eval_runtime": 2.7424,
+      "eval_samples_per_second": 13.492,
+      "eval_steps_per_second": 3.647,
       "step": 975
     },
     {
       "epoch": 13.0,
       "learning_rate": 4.103703703703704e-05,
-      "loss": 0.0036,
+      "loss": 0.1514,
       "step": 980
     },
     {
       "epoch": 13.0,
       "learning_rate": 4.088888888888889e-05,
-      "loss": 0.0761,
+      "loss": 0.0019,
       "step": 990
     },
     {
       "epoch": 13.01,
       "learning_rate": 4.074074074074074e-05,
-      "loss": 0.0036,
+      "loss": 0.0018,
       "step": 1000
     },
     {
       "epoch": 13.01,
       "learning_rate": 4.059259259259259e-05,
-      "loss": 0.1052,
+      "loss": 0.0017,
       "step": 1010
     },
     {
       "epoch": 13.01,
       "learning_rate": 4.0444444444444444e-05,
-      "loss": 0.0165,
+      "loss": 0.0017,
       "step": 1020
     },
     {
       "epoch": 13.01,
       "learning_rate": 4.02962962962963e-05,
-      "loss": 0.0347,
+      "loss": 0.0015,
       "step": 1030
     },
     {
       "epoch": 13.02,
       "learning_rate": 4.014814814814815e-05,
-      "loss": 0.007,
+      "loss": 0.0015,
       "step": 1040
     },
     {
       "epoch": 13.02,
       "learning_rate": 4e-05,
-      "loss": 0.0019,
+      "loss": 0.0017,
       "step": 1050
     },
     {
       "epoch": 13.02,
-      "eval_accuracy": 0.8857142857142857,
-      "eval_loss": 0.5436500310897827,
-      "eval_runtime": 5.8387,
-      "eval_samples_per_second": 11.989,
-      "eval_steps_per_second": 3.083,
+      "eval_accuracy": 0.972972972972973,
+      "eval_loss": 0.16978278756141663,
+      "eval_runtime": 2.7539,
+      "eval_samples_per_second": 13.436,
+      "eval_steps_per_second": 3.631,
       "step": 1050
     },
     {
       "epoch": 14.0,
       "learning_rate": 3.985185185185185e-05,
-      "loss": 0.0304,
+      "loss": 0.0015,
       "step": 1060
     },
     {
       "epoch": 14.01,
       "learning_rate": 3.97037037037037e-05,
-      "loss": 0.0019,
+      "loss": 0.0014,
       "step": 1070
     },
     {
       "epoch": 14.01,
       "learning_rate": 3.9555555555555556e-05,
-      "loss": 0.1148,
+      "loss": 0.0014,
       "step": 1080
     },
     {
       "epoch": 14.01,
       "learning_rate": 3.940740740740741e-05,
-      "loss": 0.0148,
+      "loss": 0.0013,
       "step": 1090
     },
     {
       "epoch": 14.01,
       "learning_rate": 3.925925925925926e-05,
-      "loss": 0.0016,
+      "loss": 0.0012,
       "step": 1100
     },
     {
       "epoch": 14.02,
       "learning_rate": 3.9111111111111115e-05,
-      "loss": 0.0113,
+      "loss": 0.0013,
       "step": 1110
     },
     {
       "epoch": 14.02,
       "learning_rate": 3.896296296296296e-05,
-      "loss": 0.002,
+      "loss": 0.0014,
       "step": 1120
     },
     {
       "epoch": 14.02,
-      "eval_accuracy": 0.7857142857142857,
-      "eval_loss": 1.0354480743408203,
-      "eval_runtime": 5.6448,
-      "eval_samples_per_second": 12.401,
-      "eval_steps_per_second": 3.189,
+      "eval_accuracy": 0.9459459459459459,
+      "eval_loss": 0.19233068823814392,
+      "eval_runtime": 2.8803,
+      "eval_samples_per_second": 12.846,
+      "eval_steps_per_second": 3.472,
       "step": 1125
     },
     {
       "epoch": 15.0,
       "learning_rate": 3.8814814814814814e-05,
-      "loss": 0.0017,
+      "loss": 0.0012,
       "step": 1130
     },
     {
       "epoch": 15.0,
       "learning_rate": 3.866666666666667e-05,
-      "loss": 0.1468,
+      "loss": 0.0013,
       "step": 1140
     },
     {
       "epoch": 15.01,
       "learning_rate": 3.851851851851852e-05,
-      "loss": 0.0023,
+      "loss": 0.0206,
       "step": 1150
     },
     {
       "epoch": 15.01,
       "learning_rate": 3.837037037037037e-05,
-      "loss": 0.0082,
+      "loss": 0.0014,
       "step": 1160
     },
     {
       "epoch": 15.01,
       "learning_rate": 3.8222222222222226e-05,
-      "loss": 0.003,
+      "loss": 0.2069,
       "step": 1170
     },
     {
       "epoch": 15.01,
       "learning_rate": 3.807407407407408e-05,
-      "loss": 0.0822,
+      "loss": 0.0014,
       "step": 1180
     },
     {
       "epoch": 15.02,
       "learning_rate": 3.7925925925925925e-05,
-      "loss": 0.0015,
+      "loss": 0.0012,
       "step": 1190
     },
     {
       "epoch": 15.02,
       "learning_rate": 3.777777777777778e-05,
-      "loss": 0.0028,
+      "loss": 0.0014,
       "step": 1200
     },
     {
       "epoch": 15.02,
-      "eval_accuracy": 0.9714285714285714,
-      "eval_loss": 0.10393200069665909,
-      "eval_runtime": 5.6145,
-      "eval_samples_per_second": 12.468,
-      "eval_steps_per_second": 3.206,
+      "eval_accuracy": 0.972972972972973,
+      "eval_loss": 0.15707702934741974,
+      "eval_runtime": 2.8394,
+      "eval_samples_per_second": 13.031,
+      "eval_steps_per_second": 3.522,
       "step": 1200
     },
     {
       "epoch": 16.0,
       "learning_rate": 3.762962962962963e-05,
-      "loss": 0.0019,
+      "loss": 0.0012,
       "step": 1210
     },
     {
       "epoch": 16.01,
       "learning_rate": 3.7481481481481484e-05,
-      "loss": 0.0029,
+      "loss": 0.0018,
       "step": 1220
     },
     {
       "epoch": 16.01,
       "learning_rate": 3.733333333333334e-05,
-      "loss": 0.0042,
+      "loss": 0.1871,
       "step": 1230
     },
     {
@@ -904,301 +904,301 @@
     {
       "epoch": 16.02,
       "learning_rate": 3.688888888888889e-05,
-      "loss": 0.1956,
+      "loss": 0.132,
       "step": 1260
     },
     {
       "epoch": 16.02,
       "learning_rate": 3.674074074074074e-05,
-      "loss": 0.0013,
+      "loss": 0.0474,
       "step": 1270
     },
     {
       "epoch": 16.02,
-      "eval_accuracy": 0.9714285714285714,
-      "eval_loss": 0.1551610380411148,
-      "eval_runtime": 5.9927,
-      "eval_samples_per_second": 11.681,
-      "eval_steps_per_second": 3.004,
+      "eval_accuracy": 0.8918918918918919,
+      "eval_loss": 0.5192855596542358,
+      "eval_runtime": 2.8857,
+      "eval_samples_per_second": 12.822,
+      "eval_steps_per_second": 3.465,
       "step": 1275
     },
     {
       "epoch": 17.0,
       "learning_rate": 3.6592592592592596e-05,
-      "loss": 0.2917,
+      "loss": 0.0019,
       "step": 1280
     },
     {
       "epoch": 17.0,
       "learning_rate": 3.644444444444445e-05,
-      "loss": 0.0015,
+      "loss": 0.0013,
       "step": 1290
     },
     {
       "epoch": 17.01,
       "learning_rate": 3.62962962962963e-05,
-      "loss": 0.3129,
+      "loss": 0.1087,
       "step": 1300
     },
     {
       "epoch": 17.01,
       "learning_rate": 3.614814814814815e-05,
-      "loss": 0.0016,
+      "loss": 0.081,
       "step": 1310
     },
     {
       "epoch": 17.01,
       "learning_rate": 3.6e-05,
-      "loss": 0.1919,
+      "loss": 0.0019,
       "step": 1320
     },
     {
       "epoch": 17.01,
       "learning_rate": 3.5851851851851854e-05,
-      "loss": 0.0335,
+      "loss": 0.0012,
       "step": 1330
     },
     {
       "epoch": 17.02,
       "learning_rate": 3.570370370370371e-05,
-      "loss": 0.0014,
+      "loss": 0.0012,
       "step": 1340
     },
     {
       "epoch": 17.02,
       "learning_rate": 3.555555555555556e-05,
-      "loss": 0.2309,
+      "loss": 0.0011,
       "step": 1350
     },
     {
       "epoch": 17.02,
-      "eval_accuracy": 0.9285714285714286,
-      "eval_loss": 0.2719881236553192,
-      "eval_runtime": 5.759,
-      "eval_samples_per_second": 12.155,
-      "eval_steps_per_second": 3.126,
+      "eval_accuracy": 0.972972972972973,
+      "eval_loss": 0.14079852402210236,
+      "eval_runtime": 3.0095,
+      "eval_samples_per_second": 12.294,
+      "eval_steps_per_second": 3.323,
       "step": 1350
     },
     {
       "epoch": 18.0,
       "learning_rate": 3.540740740740741e-05,
-      "loss": 0.0013,
+      "loss": 0.0009,
       "step": 1360
     },
     {
       "epoch": 18.01,
       "learning_rate": 3.525925925925926e-05,
-      "loss": 0.0018,
+      "loss": 0.0011,
       "step": 1370
     },
     {
       "epoch": 18.01,
       "learning_rate": 3.511111111111111e-05,
-      "loss": 0.2065,
+      "loss": 0.0277,
       "step": 1380
     },
     {
       "epoch": 18.01,
       "learning_rate": 3.4962962962962965e-05,
-      "loss": 0.0106,
+      "loss": 0.0011,
       "step": 1390
     },
     {
       "epoch": 18.01,
       "learning_rate": 3.481481481481482e-05,
-      "loss": 0.0019,
+      "loss": 0.0011,
       "step": 1400
     },
     {
       "epoch": 18.02,
       "learning_rate": 3.466666666666667e-05,
-      "loss": 0.0179,
+      "loss": 0.1377,
       "step": 1410
     },
     {
       "epoch": 18.02,
       "learning_rate": 3.4518518518518524e-05,
-      "loss": 0.1662,
+      "loss": 0.001,
       "step": 1420
     },
     {
       "epoch": 18.02,
-      "eval_accuracy": 0.9857142857142858,
-      "eval_loss": 0.031152786687016487,
-      "eval_runtime": 5.8068,
-      "eval_samples_per_second": 12.055,
-      "eval_steps_per_second": 3.1,
+      "eval_accuracy": 0.9459459459459459,
+      "eval_loss": 0.34057652950286865,
+      "eval_runtime": 2.8674,
+      "eval_samples_per_second": 12.904,
+      "eval_steps_per_second": 3.488,
       "step": 1425
     },
     {
       "epoch": 19.0,
       "learning_rate": 3.437037037037037e-05,
-      "loss": 0.0945,
+      "loss": 0.002,
       "step": 1430
     },
     {
       "epoch": 19.0,
       "learning_rate": 3.4222222222222224e-05,
-      "loss": 0.1443,
+      "loss": 0.0018,
       "step": 1440
     },
     {
       "epoch": 19.01,
       "learning_rate": 3.4074074074074077e-05,
-      "loss": 0.0583,
+      "loss": 0.2287,
       "step": 1450
     },
     {
       "epoch": 19.01,
       "learning_rate": 3.392592592592593e-05,
-      "loss": 0.0015,
+      "loss": 0.102,
       "step": 1460
     },
     {
       "epoch": 19.01,
       "learning_rate": 3.377777777777778e-05,
-      "loss": 0.1515,
+      "loss": 0.0785,
       "step": 1470
     },
     {
       "epoch": 19.01,
       "learning_rate": 3.3629629629629636e-05,
-      "loss": 0.008,
+      "loss": 0.002,
       "step": 1480
     },
     {
       "epoch": 19.02,
       "learning_rate": 3.348148148148148e-05,
-      "loss": 0.2176,
+      "loss": 0.0357,
       "step": 1490
     },
     {
       "epoch": 19.02,
       "learning_rate": 3.3333333333333335e-05,
-      "loss": 0.0199,
+      "loss": 0.0034,
       "step": 1500
     },
     {
       "epoch": 19.02,
-      "eval_accuracy": 0.9571428571428572,
-      "eval_loss": 0.14778082072734833,
-      "eval_runtime": 5.8359,
-      "eval_samples_per_second": 11.995,
-      "eval_steps_per_second": 3.084,
+      "eval_accuracy": 0.9459459459459459,
+      "eval_loss": 0.25160789489746094,
+      "eval_runtime": 2.8519,
+      "eval_samples_per_second": 12.974,
+      "eval_steps_per_second": 3.506,
       "step": 1500
     },
     {
       "epoch": 20.0,
       "learning_rate": 3.318518518518519e-05,
-      "loss": 0.0017,
+      "loss": 0.0609,
       "step": 1510
     },
     {
       "epoch": 20.01,
       "learning_rate": 3.303703703703704e-05,
-      "loss": 0.001,
+      "loss": 0.0013,
       "step": 1520
     },
     {
       "epoch": 20.01,
       "learning_rate": 3.2888888888888894e-05,
-      "loss": 0.0442,
+      "loss": 0.0012,
       "step": 1530
     },
     {
       "epoch": 20.01,
       "learning_rate": 3.274074074074075e-05,
-      "loss": 0.0019,
+      "loss": 0.0014,
       "step": 1540
     },
     {
       "epoch": 20.01,
       "learning_rate": 3.25925925925926e-05,
-      "loss": 0.0015,
+      "loss": 0.1104,
       "step": 1550
     },
     {
       "epoch": 20.02,
       "learning_rate": 3.2444444444444446e-05,
-      "loss": 0.0012,
+      "loss": 0.0011,
       "step": 1560
     },
     {
       "epoch": 20.02,
       "learning_rate": 3.22962962962963e-05,
-      "loss": 0.001,
+      "loss": 0.0029,
       "step": 1570
     },
     {
       "epoch": 20.02,
-      "eval_accuracy": 0.9714285714285714,
-      "eval_loss": 0.21894989907741547,
-      "eval_runtime": 5.952,
-      "eval_samples_per_second": 11.761,
-      "eval_steps_per_second": 3.024,
+      "eval_accuracy": 0.918918918918919,
+      "eval_loss": 0.29616308212280273,
+      "eval_runtime": 2.9222,
+      "eval_samples_per_second": 12.662,
+      "eval_steps_per_second": 3.422,
       "step": 1575
     },
     {
       "epoch": 21.0,
       "learning_rate": 3.214814814814815e-05,
-      "loss": 0.0058,
+      "loss": 0.0014,
       "step": 1580
     },
     {
       "epoch": 21.0,
       "learning_rate": 3.2000000000000005e-05,
-      "loss": 0.001,
+      "loss": 0.0769,
       "step": 1590
     },
     {
       "epoch": 21.01,
       "learning_rate": 3.185185185185185e-05,
-      "loss": 0.0271,
+      "loss": 0.001,
       "step": 1600
     },
     {
       "epoch": 21.01,
       "learning_rate": 3.1703703703703705e-05,
-      "loss": 0.0009,
+      "loss": 0.001,
       "step": 1610
     },
     {
       "epoch": 21.01,
       "learning_rate": 3.155555555555556e-05,
-      "loss": 0.001,
+      "loss": 0.1219,
       "step": 1620
     },
     {
       "epoch": 21.01,
       "learning_rate": 3.140740740740741e-05,
-      "loss": 0.1786,
+      "loss": 0.001,
       "step": 1630
     },
     {
       "epoch": 21.02,
       "learning_rate": 3.1259259259259264e-05,
-      "loss": 0.0056,
+      "loss": 0.0013,
       "step": 1640
     },
     {
       "epoch": 21.02,
       "learning_rate": 3.111111111111111e-05,
-      "loss": 0.0009,
+      "loss": 0.0008,
       "step": 1650
     },
     {
       "epoch": 21.02,
-      "eval_accuracy": 0.9714285714285714,
-      "eval_loss": 0.15684819221496582,
-      "eval_runtime": 5.7784,
-      "eval_samples_per_second": 12.114,
-      "eval_steps_per_second": 3.115,
+      "eval_accuracy": 0.918918918918919,
+      "eval_loss": 0.4023502469062805,
+      "eval_runtime": 2.9031,
+      "eval_samples_per_second": 12.745,
+      "eval_steps_per_second": 3.445,
       "step": 1650
     },
     {
       "epoch": 22.0,
       "learning_rate": 3.096296296296296e-05,
-      "loss": 0.1822,
+      "loss": 0.001,
       "step": 1660
     },
     {
@@ -1210,13 +1210,13 @@
     {
       "epoch": 22.01,
       "learning_rate": 3.066666666666667e-05,
-      "loss": 0.0009,
+      "loss": 0.0022,
       "step": 1680
     },
     {
       "epoch": 22.01,
       "learning_rate": 3.0518518518518515e-05,
-      "loss": 0.0008,
+      "loss": 0.0009,
       "step": 1690
     },
     {
@@ -1228,7 +1228,7 @@
     {
       "epoch": 22.02,
       "learning_rate": 3.0222222222222225e-05,
-      "loss": 0.001,
+      "loss": 0.0008,
       "step": 1710
     },
     {
@@ -1239,17 +1239,17 @@
     },
     {
       "epoch": 22.02,
-      "eval_accuracy": 0.9428571428571428,
-      "eval_loss": 0.213576078414917,
-      "eval_runtime": 5.6561,
-      "eval_samples_per_second": 12.376,
-      "eval_steps_per_second": 3.182,
+      "eval_accuracy": 0.918918918918919,
+      "eval_loss": 0.4643724262714386,
+      "eval_runtime": 2.9019,
+      "eval_samples_per_second": 12.75,
+      "eval_steps_per_second": 3.446,
       "step": 1725
     },
     {
       "epoch": 23.0,
       "learning_rate": 2.992592592592593e-05,
-      "loss": 0.0008,
+      "loss": 0.0009,
       "step": 1730
     },
     {
@@ -1261,13 +1261,13 @@
     {
       "epoch": 23.01,
       "learning_rate": 2.962962962962963e-05,
-      "loss": 0.0008,
+      "loss": 0.0336,
       "step": 1750
     },
     {
       "epoch": 23.01,
       "learning_rate": 2.9481481481481483e-05,
-      "loss": 0.0238,
+      "loss": 0.0044,
       "step": 1760
     },
     {
@@ -1279,52 +1279,52 @@
     {
       "epoch": 23.01,
       "learning_rate": 2.918518518518519e-05,
-      "loss": 0.0007,
+      "loss": 0.001,
       "step": 1780
     },
     {
       "epoch": 23.02,
       "learning_rate": 2.9037037037037042e-05,
-      "loss": 0.0009,
+      "loss": 0.0008,
       "step": 1790
     },
     {
       "epoch": 23.02,
       "learning_rate": 2.8888888888888888e-05,
-      "loss": 0.0007,
+      "loss": 0.1521,
       "step": 1800
     },
     {
       "epoch": 23.02,
-      "eval_accuracy": 0.9714285714285714,
-      "eval_loss": 0.10322745889425278,
-      "eval_runtime": 5.6572,
-      "eval_samples_per_second": 12.374,
-      "eval_steps_per_second": 3.182,
+      "eval_accuracy": 0.918918918918919,
+      "eval_loss": 0.48252156376838684,
+      "eval_runtime": 2.9421,
+      "eval_samples_per_second": 12.576,
+      "eval_steps_per_second": 3.399,
       "step": 1800
     },
     {
       "epoch": 24.0,
       "learning_rate": 2.874074074074074e-05,
-      "loss": 0.0008,
+      "loss": 0.0023,
       "step": 1810
     },
     {
       "epoch": 24.01,
       "learning_rate": 2.8592592592592594e-05,
-      "loss": 0.0007,
+      "loss": 0.0008,
       "step": 1820
     },
     {
       "epoch": 24.01,
       "learning_rate": 2.8444444444444447e-05,
-      "loss": 0.0008,
+      "loss": 0.0074,
       "step": 1830
     },
     {
       "epoch": 24.01,
       "learning_rate": 2.8296296296296297e-05,
-      "loss": 0.0007,
+      "loss": 0.0009,
       "step": 1840
     },
     {
@@ -1336,22 +1336,22 @@
     {
       "epoch": 24.02,
       "learning_rate": 2.8000000000000003e-05,
-      "loss": 0.0006,
+      "loss": 0.0007,
       "step": 1860
     },
     {
       "epoch": 24.02,
       "learning_rate": 2.7851851851851853e-05,
-      "loss": 0.0007,
+      "loss": 0.001,
       "step": 1870
     },
     {
       "epoch": 24.02,
-      "eval_accuracy": 0.9714285714285714,
-      "eval_loss": 0.10255812108516693,
-      "eval_runtime": 5.7489,
-      "eval_samples_per_second": 12.176,
-      "eval_steps_per_second": 3.131,
+      "eval_accuracy": 0.918918918918919,
+      "eval_loss": 0.6339796185493469,
+      "eval_runtime": 2.8461,
+      "eval_samples_per_second": 13.0,
+      "eval_steps_per_second": 3.514,
       "step": 1875
     },
     {
@@ -1375,58 +1375,58 @@
     {
       "epoch": 25.01,
       "learning_rate": 2.725925925925926e-05,
-      "loss": 0.0007,
+      "loss": 0.001,
       "step": 1910
     },
     {
       "epoch": 25.01,
       "learning_rate": 2.7111111111111114e-05,
-      "loss": 0.0006,
+      "loss": 0.0007,
       "step": 1920
     },
     {
       "epoch": 25.01,
       "learning_rate": 2.696296296296296e-05,
-      "loss": 0.0006,
+      "loss": 0.0007,
       "step": 1930
     },
     {
       "epoch": 25.02,
       "learning_rate": 2.6814814814814814e-05,
-      "loss": 0.0006,
+      "loss": 0.0007,
       "step": 1940
     },
     {
       "epoch": 25.02,
       "learning_rate": 2.6666666666666667e-05,
-      "loss": 0.0006,
+      "loss": 0.0245,
       "step": 1950
     },
     {
       "epoch": 25.02,
-      "eval_accuracy": 0.9571428571428572,
-      "eval_loss": 0.11298384517431259,
-      "eval_runtime": 5.7507,
-      "eval_samples_per_second": 12.173,
-      "eval_steps_per_second": 3.13,
+      "eval_accuracy": 0.9459459459459459,
+      "eval_loss": 0.3778836727142334,
+      "eval_runtime": 2.8022,
+      "eval_samples_per_second": 13.204,
+      "eval_steps_per_second": 3.569,
       "step": 1950
     },
     {
       "epoch": 26.0,
       "learning_rate": 2.651851851851852e-05,
-      "loss": 0.0006,
+      "loss": 0.0007,
       "step": 1960
     },
     {
       "epoch": 26.01,
       "learning_rate": 2.6370370370370373e-05,
-      "loss": 0.0006,
+      "loss": 0.0007,
       "step": 1970
     },
     {
       "epoch": 26.01,
       "learning_rate": 2.6222222222222226e-05,
-      "loss": 0.0006,
+      "loss": 0.0007,
       "step": 1980
     },
     {
@@ -1438,115 +1438,115 @@
     {
       "epoch": 26.01,
       "learning_rate": 2.5925925925925925e-05,
-      "loss": 0.0006,
+      "loss": 0.0947,
       "step": 2000
     },
     {
       "epoch": 26.02,
       "learning_rate": 2.5777777777777778e-05,
-      "loss": 0.0006,
+      "loss": 0.001,
       "step": 2010
     },
     {
       "epoch": 26.02,
       "learning_rate": 2.562962962962963e-05,
-      "loss": 0.0006,
+      "loss": 0.0007,
       "step": 2020
     },
     {
       "epoch": 26.02,
-      "eval_accuracy": 0.9714285714285714,
-      "eval_loss": 0.11467116326093674,
-      "eval_runtime": 5.8139,
-      "eval_samples_per_second": 12.04,
-      "eval_steps_per_second": 3.096,
+      "eval_accuracy": 0.9459459459459459,
+      "eval_loss": 0.3375699818134308,
+      "eval_runtime": 2.7928,
+      "eval_samples_per_second": 13.248,
+      "eval_steps_per_second": 3.581,
       "step": 2025
     },
     {
       "epoch": 27.0,
       "learning_rate": 2.5481481481481484e-05,
-      "loss": 0.0007,
+      "loss": 0.0006,
       "step": 2030
     },
     {
       "epoch": 27.0,
       "learning_rate": 2.5333333333333337e-05,
-      "loss": 0.0007,
+      "loss": 0.0012,
       "step": 2040
     },
     {
       "epoch": 27.01,
       "learning_rate": 2.5185185185185183e-05,
-      "loss": 0.0006,
+      "loss": 0.0007,
       "step": 2050
     },
     {
       "epoch": 27.01,
       "learning_rate": 2.5037037037037036e-05,
-      "loss": 0.0006,
+      "loss": 0.0007,
       "step": 2060
     },
     {
       "epoch": 27.01,
       "learning_rate": 2.488888888888889e-05,
-      "loss": 0.0006,
+      "loss": 0.0927,
       "step": 2070
     },
     {
       "epoch": 27.01,
       "learning_rate": 2.4740740740740742e-05,
-      "loss": 0.0006,
+      "loss": 0.0817,
       "step": 2080
     },
     {
       "epoch": 27.02,
       "learning_rate": 2.4592592592592595e-05,
-      "loss": 0.0006,
+      "loss": 0.2909,
       "step": 2090
     },
     {
       "epoch": 27.02,
       "learning_rate": 2.4444444444444445e-05,
-      "loss": 0.0006,
+      "loss": 0.0011,
       "step": 2100
     },
     {
       "epoch": 27.02,
-      "eval_accuracy": 0.9857142857142858,
-      "eval_loss": 0.0858125165104866,
-      "eval_runtime": 5.7371,
-      "eval_samples_per_second": 12.201,
-      "eval_steps_per_second": 3.137,
+      "eval_accuracy": 0.9459459459459459,
+      "eval_loss": 0.28334298729896545,
+      "eval_runtime": 2.7397,
+      "eval_samples_per_second": 13.505,
+      "eval_steps_per_second": 3.65,
       "step": 2100
     },
     {
       "epoch": 28.0,
       "learning_rate": 2.4296296296296298e-05,
-      "loss": 0.0006,
+      "loss": 0.0007,
       "step": 2110
     },
     {
       "epoch": 28.01,
       "learning_rate": 2.414814814814815e-05,
-      "loss": 0.0006,
+      "loss": 0.0007,
       "step": 2120
     },
     {
       "epoch": 28.01,
       "learning_rate": 2.4e-05,
-      "loss": 0.0006,
+      "loss": 0.0007,
       "step": 2130
     },
     {
       "epoch": 28.01,
       "learning_rate": 2.3851851851851854e-05,
-      "loss": 0.0005,
+      "loss": 0.0007,
       "step": 2140
     },
     {
       "epoch": 28.01,
       "learning_rate": 2.3703703703703707e-05,
-      "loss": 0.0006,
+      "loss": 0.0007,
       "step": 2150
     },
     {
@@ -1558,16 +1558,16 @@
     {
       "epoch": 28.02,
       "learning_rate": 2.340740740740741e-05,
-      "loss": 0.0006,
+      "loss": 0.0008,
       "step": 2170
     },
     {
       "epoch": 28.02,
-      "eval_accuracy": 0.9857142857142858,
-      "eval_loss": 0.08683785051107407,
-      "eval_runtime": 5.8537,
-      "eval_samples_per_second": 11.958,
-      "eval_steps_per_second": 3.075,
+      "eval_accuracy": 0.972972972972973,
+      "eval_loss": 0.15925714373588562,
+      "eval_runtime": 2.6329,
+      "eval_samples_per_second": 14.053,
+      "eval_steps_per_second": 3.798,
       "step": 2175
     },
     {
@@ -1579,58 +1579,58 @@
     {
       "epoch": 29.0,
       "learning_rate": 2.3111111111111112e-05,
-      "loss": 0.0005,
+      "loss": 0.0015,
       "step": 2190
     },
     {
       "epoch": 29.01,
       "learning_rate": 2.2962962962962965e-05,
-      "loss": 0.0006,
+      "loss": 0.0008,
       "step": 2200
     },
     {
       "epoch": 29.01,
       "learning_rate": 2.2814814814814818e-05,
-      "loss": 0.0006,
+      "loss": 0.0007,
       "step": 2210
     },
     {
       "epoch": 29.01,
       "learning_rate": 2.2666666666666668e-05,
-      "loss": 0.0005,
+      "loss": 0.0006,
       "step": 2220
     },
     {
       "epoch": 29.01,
       "learning_rate": 2.251851851851852e-05,
-      "loss": 0.0005,
+      "loss": 0.0006,
       "step": 2230
     },
     {
       "epoch": 29.02,
       "learning_rate": 2.2370370370370374e-05,
-      "loss": 0.0005,
+      "loss": 0.0953,
       "step": 2240
     },
     {
       "epoch": 29.02,
       "learning_rate": 2.2222222222222223e-05,
-      "loss": 0.0006,
+      "loss": 0.0008,
       "step": 2250
     },
     {
       "epoch": 29.02,
-      "eval_accuracy": 0.9857142857142858,
-      "eval_loss": 0.08798635751008987,
-      "eval_runtime": 5.7007,
-      "eval_samples_per_second": 12.279,
-      "eval_steps_per_second": 3.158,
+      "eval_accuracy": 0.972972972972973,
+      "eval_loss": 0.08563826233148575,
+      "eval_runtime": 2.7114,
+      "eval_samples_per_second": 13.646,
+      "eval_steps_per_second": 3.688,
       "step": 2250
     },
     {
       "epoch": 30.0,
       "learning_rate": 2.2074074074074076e-05,
-      "loss": 0.0005,
+      "loss": 0.0006,
       "step": 2260
     },
     {
@@ -1642,25 +1642,25 @@
     {
       "epoch": 30.01,
       "learning_rate": 2.177777777777778e-05,
-      "loss": 0.0005,
+      "loss": 0.0006,
       "step": 2280
     },
     {
       "epoch": 30.01,
       "learning_rate": 2.162962962962963e-05,
-      "loss": 0.0005,
+      "loss": 0.0006,
       "step": 2290
     },
     {
       "epoch": 30.01,
       "learning_rate": 2.148148148148148e-05,
-      "loss": 0.0006,
+      "loss": 0.0007,
       "step": 2300
     },
     {
       "epoch": 30.02,
       "learning_rate": 2.1333333333333335e-05,
-      "loss": 0.0005,
+      "loss": 0.0006,
       "step": 2310
     },
     {
@@ -1671,17 +1671,17 @@
     },
     {
       "epoch": 30.02,
-      "eval_accuracy": 0.9857142857142858,
-      "eval_loss": 0.08717390149831772,
-      "eval_runtime": 5.8675,
-      "eval_samples_per_second": 11.93,
-      "eval_steps_per_second": 3.068,
+      "eval_accuracy": 0.972972972972973,
+      "eval_loss": 0.1049196645617485,
+      "eval_runtime": 2.6802,
+      "eval_samples_per_second": 13.805,
+      "eval_steps_per_second": 3.731,
       "step": 2325
     },
     {
       "epoch": 31.0,
       "learning_rate": 2.1037037037037037e-05,
-      "loss": 0.0005,
+      "loss": 0.0006,
       "step": 2330
     },
     {
@@ -1693,7 +1693,7 @@
     {
       "epoch": 31.01,
       "learning_rate": 2.074074074074074e-05,
-      "loss": 0.0005,
+      "loss": 0.0006,
       "step": 2350
     },
     {
@@ -1728,11 +1728,11 @@
     },
     {
       "epoch": 31.02,
-      "eval_accuracy": 0.9857142857142858,
-      "eval_loss": 0.09013553708791733,
-      "eval_runtime": 5.7699,
-      "eval_samples_per_second": 12.132,
-      "eval_steps_per_second": 3.12,
+      "eval_accuracy": 0.972972972972973,
+      "eval_loss": 0.11323297023773193,
+      "eval_runtime": 3.2205,
+      "eval_samples_per_second": 11.489,
+      "eval_steps_per_second": 3.105,
       "step": 2400
     },
     {
@@ -1779,11 +1779,11 @@
     },
     {
       "epoch": 32.02,
-      "eval_accuracy": 0.9857142857142858,
-      "eval_loss": 0.08940133452415466,
-      "eval_runtime": 5.7037,
-      "eval_samples_per_second": 12.273,
-      "eval_steps_per_second": 3.156,
+      "eval_accuracy": 0.972972972972973,
+      "eval_loss": 0.11638977378606796,
+      "eval_runtime": 2.5685,
+      "eval_samples_per_second": 14.405,
+      "eval_steps_per_second": 3.893,
       "step": 2475
     },
     {
@@ -1813,7 +1813,7 @@
     {
       "epoch": 33.01,
       "learning_rate": 1.8222222222222224e-05,
-      "loss": 0.0004,
+      "loss": 0.0005,
       "step": 2520
     },
     {
@@ -1836,17 +1836,17 @@
     },
     {
       "epoch": 33.02,
-      "eval_accuracy": 0.9857142857142858,
-      "eval_loss": 0.08584433048963547,
-      "eval_runtime": 5.7242,
-      "eval_samples_per_second": 12.229,
-      "eval_steps_per_second": 3.145,
+      "eval_accuracy": 0.972972972972973,
+      "eval_loss": 0.12427014112472534,
+      "eval_runtime": 2.5792,
+      "eval_samples_per_second": 14.345,
+      "eval_steps_per_second": 3.877,
       "step": 2550
     },
     {
       "epoch": 34.0,
       "learning_rate": 1.762962962962963e-05,
-      "loss": 0.0004,
+      "loss": 0.0005,
       "step": 2560
     },
     {
@@ -1864,7 +1864,7 @@
     {
       "epoch": 34.01,
       "learning_rate": 1.7185185185185185e-05,
-      "loss": 0.0004,
+      "loss": 0.0005,
       "step": 2590
     },
     {
@@ -1887,11 +1887,11 @@
     },
     {
       "epoch": 34.02,
-      "eval_accuracy": 0.9857142857142858,
-      "eval_loss": 0.0912405401468277,
-      "eval_runtime": 5.7854,
-      "eval_samples_per_second": 12.099,
-      "eval_steps_per_second": 3.111,
+      "eval_accuracy": 0.972972972972973,
+      "eval_loss": 0.1306069791316986,
+      "eval_runtime": 2.6024,
+      "eval_samples_per_second": 14.217,
+      "eval_steps_per_second": 3.843,
       "step": 2625
     },
     {
@@ -1903,7 +1903,7 @@
     {
       "epoch": 35.0,
       "learning_rate": 1.6444444444444447e-05,
-      "loss": 0.0004,
+      "loss": 0.0005,
       "step": 2640
     },
     {
@@ -1915,13 +1915,13 @@
     {
       "epoch": 35.01,
       "learning_rate": 1.614814814814815e-05,
-      "loss": 0.0005,
+      "loss": 0.0007,
       "step": 2660
     },
     {
       "epoch": 35.01,
       "learning_rate": 1.6000000000000003e-05,
-      "loss": 0.1029,
+      "loss": 0.0005,
       "step": 2670
     },
     {
@@ -1933,7 +1933,7 @@
     {
       "epoch": 35.02,
       "learning_rate": 1.5703703703703705e-05,
-      "loss": 0.0006,
+      "loss": 0.0005,
       "step": 2690
     },
     {
@@ -1944,11 +1944,11 @@
     },
     {
       "epoch": 35.02,
-      "eval_accuracy": 0.9428571428571428,
-      "eval_loss": 0.33030033111572266,
-      "eval_runtime": 5.6632,
-      "eval_samples_per_second": 12.361,
-      "eval_steps_per_second": 3.178,
+      "eval_accuracy": 0.9459459459459459,
+      "eval_loss": 0.39192795753479004,
+      "eval_runtime": 2.5502,
+      "eval_samples_per_second": 14.509,
+      "eval_steps_per_second": 3.921,
       "step": 2700
     },
     {
@@ -1960,13 +1960,13 @@
     {
       "epoch": 36.01,
       "learning_rate": 1.5259259259259258e-05,
-      "loss": 0.0004,
+      "loss": 0.0005,
       "step": 2720
     },
     {
       "epoch": 36.01,
       "learning_rate": 1.5111111111111112e-05,
-      "loss": 0.0004,
+      "loss": 0.0005,
       "step": 2730
     },
     {
@@ -1984,7 +1984,7 @@
     {
       "epoch": 36.02,
       "learning_rate": 1.4666666666666668e-05,
-      "loss": 0.0004,
+      "loss": 0.0005,
       "step": 2760
     },
     {
@@ -1995,17 +1995,17 @@
     },
     {
       "epoch": 36.02,
-      "eval_accuracy": 0.9428571428571428,
-      "eval_loss": 0.17365017533302307,
-      "eval_runtime": 5.6606,
-      "eval_samples_per_second": 12.366,
-      "eval_steps_per_second": 3.18,
+      "eval_accuracy": 0.9459459459459459,
+      "eval_loss": 0.36302804946899414,
+      "eval_runtime": 2.4982,
+      "eval_samples_per_second": 14.811,
+      "eval_steps_per_second": 4.003,
       "step": 2775
     },
     {
       "epoch": 37.0,
       "learning_rate": 1.437037037037037e-05,
-      "loss": 0.0004,
+      "loss": 0.0005,
       "step": 2780
     },
     {
@@ -2023,7 +2023,7 @@
     {
       "epoch": 37.01,
       "learning_rate": 1.3925925925925926e-05,
-      "loss": 0.0004,
+      "loss": 0.0005,
       "step": 2810
     },
     {
@@ -2035,13 +2035,13 @@
     {
       "epoch": 37.01,
       "learning_rate": 1.362962962962963e-05,
-      "loss": 0.0004,
+      "loss": 0.0005,
       "step": 2830
     },
     {
       "epoch": 37.02,
       "learning_rate": 1.348148148148148e-05,
-      "loss": 0.0004,
+      "loss": 0.0005,
       "step": 2840
     },
     {
@@ -2052,11 +2052,11 @@
     },
     {
       "epoch": 37.02,
-      "eval_accuracy": 0.9714285714285714,
-      "eval_loss": 0.15336854755878448,
-      "eval_runtime": 5.7371,
-      "eval_samples_per_second": 12.201,
-      "eval_steps_per_second": 3.137,
+      "eval_accuracy": 0.9459459459459459,
+      "eval_loss": 0.2762458324432373,
+      "eval_runtime": 2.5203,
+      "eval_samples_per_second": 14.681,
+      "eval_steps_per_second": 3.968,
       "step": 2850
     },
     {
@@ -2068,7 +2068,7 @@
     {
       "epoch": 38.01,
       "learning_rate": 1.3037037037037036e-05,
-      "loss": 0.0004,
+      "loss": 0.0005,
       "step": 2870
     },
     {
@@ -2098,16 +2098,16 @@
     {
       "epoch": 38.02,
       "learning_rate": 1.2296296296296298e-05,
-      "loss": 0.0004,
+      "loss": 0.0005,
       "step": 2920
     },
     {
       "epoch": 38.02,
-      "eval_accuracy": 0.9714285714285714,
-      "eval_loss": 0.13503512740135193,
-      "eval_runtime": 5.8822,
-      "eval_samples_per_second": 11.9,
-      "eval_steps_per_second": 3.06,
+      "eval_accuracy": 0.9459459459459459,
+      "eval_loss": 0.23680266737937927,
+      "eval_runtime": 2.5362,
+      "eval_samples_per_second": 14.589,
+      "eval_steps_per_second": 3.943,
       "step": 2925
     },
     {
@@ -2131,7 +2131,7 @@
     {
       "epoch": 39.01,
       "learning_rate": 1.1703703703703705e-05,
-      "loss": 0.0004,
+      "loss": 0.0005,
       "step": 2960
     },
     {
@@ -2160,11 +2160,11 @@
     },
     {
       "epoch": 39.02,
-      "eval_accuracy": 0.9714285714285714,
-      "eval_loss": 0.1270487755537033,
-      "eval_runtime": 5.5912,
-      "eval_samples_per_second": 12.52,
-      "eval_steps_per_second": 3.219,
+      "eval_accuracy": 0.972972972972973,
+      "eval_loss": 0.1935373693704605,
+      "eval_runtime": 2.5511,
+      "eval_samples_per_second": 14.504,
+      "eval_steps_per_second": 3.92,
       "step": 3000
     },
     {
@@ -2211,11 +2211,11 @@
     },
     {
       "epoch": 40.02,
-      "eval_accuracy": 0.9714285714285714,
-      "eval_loss": 0.1252710521221161,
-      "eval_runtime": 5.672,
-      "eval_samples_per_second": 12.341,
-      "eval_steps_per_second": 3.173,
+      "eval_accuracy": 0.972972972972973,
+      "eval_loss": 0.19308657944202423,
+      "eval_runtime": 2.5348,
+      "eval_samples_per_second": 14.597,
+      "eval_steps_per_second": 3.945,
       "step": 3075
     },
     {
@@ -2227,7 +2227,7 @@
     {
       "epoch": 41.0,
       "learning_rate": 9.777777777777779e-06,
-      "loss": 0.0004,
+      "loss": 0.0085,
       "step": 3090
     },
     {
@@ -2257,7 +2257,7 @@
     {
       "epoch": 41.02,
       "learning_rate": 9.037037037037037e-06,
-      "loss": 0.0004,
+      "loss": 0.0005,
       "step": 3140
     },
     {
@@ -2268,11 +2268,11 @@
     },
     {
       "epoch": 41.02,
-      "eval_accuracy": 0.9714285714285714,
-      "eval_loss": 0.12408071011304855,
-      "eval_runtime": 5.7302,
-      "eval_samples_per_second": 12.216,
-      "eval_steps_per_second": 3.141,
+      "eval_accuracy": 0.9459459459459459,
+      "eval_loss": 0.21387839317321777,
+      "eval_runtime": 2.5103,
+      "eval_samples_per_second": 14.739,
+      "eval_steps_per_second": 3.984,
       "step": 3150
     },
     {
@@ -2284,7 +2284,7 @@
     {
       "epoch": 42.01,
       "learning_rate": 8.592592592592593e-06,
-      "loss": 0.0004,
+      "loss": 0.0005,
       "step": 3170
     },
     {
@@ -2319,17 +2319,17 @@
     },
     {
       "epoch": 42.02,
-      "eval_accuracy": 0.9714285714285714,
-      "eval_loss": 0.12468262761831284,
-      "eval_runtime": 5.8752,
-      "eval_samples_per_second": 11.915,
-      "eval_steps_per_second": 3.064,
+      "eval_accuracy": 0.972972972972973,
+      "eval_loss": 0.1899683177471161,
+      "eval_runtime": 2.6252,
+      "eval_samples_per_second": 14.094,
+      "eval_steps_per_second": 3.809,
       "step": 3225
     },
     {
       "epoch": 43.0,
       "learning_rate": 7.703703703703704e-06,
-      "loss": 0.0004,
+      "loss": 0.0015,
       "step": 3230
     },
     {
@@ -2371,16 +2371,16 @@
     {
       "epoch": 43.02,
       "learning_rate": 6.666666666666667e-06,
-      "loss": 0.0004,
+      "loss": 0.0006,
       "step": 3300
     },
     {
       "epoch": 43.02,
-      "eval_accuracy": 0.9714285714285714,
-      "eval_loss": 0.12618018686771393,
-      "eval_runtime": 5.7926,
-      "eval_samples_per_second": 12.084,
-      "eval_steps_per_second": 3.107,
+      "eval_accuracy": 0.972972972972973,
+      "eval_loss": 0.1750660091638565,
+      "eval_runtime": 2.5683,
+      "eval_samples_per_second": 14.407,
+      "eval_steps_per_second": 3.894,
       "step": 3300
     },
     {
@@ -2427,11 +2427,11 @@
     },
     {
       "epoch": 44.02,
-      "eval_accuracy": 0.9571428571428572,
-      "eval_loss": 0.12661002576351166,
-      "eval_runtime": 5.8019,
-      "eval_samples_per_second": 12.065,
-      "eval_steps_per_second": 3.102,
+      "eval_accuracy": 0.9459459459459459,
+      "eval_loss": 0.29775792360305786,
+      "eval_runtime": 2.6767,
+      "eval_samples_per_second": 13.823,
+      "eval_steps_per_second": 3.736,
       "step": 3375
     },
     {
@@ -2461,7 +2461,7 @@
     {
       "epoch": 45.01,
       "learning_rate": 4.888888888888889e-06,
-      "loss": 0.0253,
+      "loss": 0.0004,
       "step": 3420
     },
     {
@@ -2484,11 +2484,11 @@
     },
     {
       "epoch": 45.02,
-      "eval_accuracy": 0.9714285714285714,
-      "eval_loss": 0.17405559122562408,
-      "eval_runtime": 5.6182,
-      "eval_samples_per_second": 12.459,
-      "eval_steps_per_second": 3.204,
+      "eval_accuracy": 0.9459459459459459,
+      "eval_loss": 0.2776608467102051,
+      "eval_runtime": 2.5843,
+      "eval_samples_per_second": 14.317,
+      "eval_steps_per_second": 3.87,
       "step": 3450
     },
     {
@@ -2535,11 +2535,11 @@
     },
     {
       "epoch": 46.02,
-      "eval_accuracy": 0.9714285714285714,
-      "eval_loss": 0.17525674402713776,
-      "eval_runtime": 5.8413,
-      "eval_samples_per_second": 11.984,
-      "eval_steps_per_second": 3.082,
+      "eval_accuracy": 0.9459459459459459,
+      "eval_loss": 0.2706151008605957,
+      "eval_runtime": 2.6031,
+      "eval_samples_per_second": 14.214,
+      "eval_steps_per_second": 3.842,
       "step": 3525
     },
     {
@@ -2581,7 +2581,7 @@
     {
       "epoch": 47.02,
       "learning_rate": 2.3703703703703703e-06,
-      "loss": 0.0006,
+      "loss": 0.0004,
       "step": 3590
     },
     {
@@ -2592,11 +2592,11 @@
     },
     {
       "epoch": 47.02,
-      "eval_accuracy": 0.9714285714285714,
-      "eval_loss": 0.1633566915988922,
-      "eval_runtime": 5.6957,
-      "eval_samples_per_second": 12.29,
-      "eval_steps_per_second": 3.16,
+      "eval_accuracy": 0.9459459459459459,
+      "eval_loss": 0.26377302408218384,
+      "eval_runtime": 2.5538,
+      "eval_samples_per_second": 14.488,
+      "eval_steps_per_second": 3.916,
       "step": 3600
     },
     {
@@ -2620,7 +2620,7 @@
     {
       "epoch": 48.01,
       "learning_rate": 1.6296296296296295e-06,
-      "loss": 0.0004,
+      "loss": 0.0005,
       "step": 3640
     },
     {
@@ -2643,11 +2643,11 @@
     },
     {
       "epoch": 48.02,
-      "eval_accuracy": 0.9714285714285714,
-      "eval_loss": 0.1603454053401947,
-      "eval_runtime": 5.9217,
-      "eval_samples_per_second": 11.821,
-      "eval_steps_per_second": 3.04,
+      "eval_accuracy": 0.9459459459459459,
+      "eval_loss": 0.21225357055664062,
+      "eval_runtime": 2.6586,
+      "eval_samples_per_second": 13.917,
+      "eval_steps_per_second": 3.761,
       "step": 3675
     },
     {
@@ -2700,38 +2700,38 @@
     },
     {
       "epoch": 49.02,
-      "eval_accuracy": 0.9714285714285714,
-      "eval_loss": 0.16023388504981995,
-      "eval_runtime": 5.6893,
-      "eval_samples_per_second": 12.304,
-      "eval_steps_per_second": 3.164,
+      "eval_accuracy": 0.9459459459459459,
+      "eval_loss": 0.21062754094600677,
+      "eval_runtime": 2.7099,
+      "eval_samples_per_second": 13.653,
+      "eval_steps_per_second": 3.69,
       "step": 3750
     },
     {
       "epoch": 49.02,
       "step": 3750,
       "total_flos": 1.86923023515648e+19,
-      "train_loss": 0.1646428666052719,
-      "train_runtime": 3006.4065,
-      "train_samples_per_second": 4.989,
-      "train_steps_per_second": 1.247
+      "train_loss": 0.16172246270999313,
+      "train_runtime": 3621.2802,
+      "train_samples_per_second": 4.142,
+      "train_steps_per_second": 1.036
     },
     {
       "epoch": 49.02,
-      "eval_accuracy": 0.9290322580645162,
-      "eval_loss": 0.18336611986160278,
-      "eval_runtime": 15.2988,
-      "eval_samples_per_second": 10.132,
-      "eval_steps_per_second": 2.549,
+      "eval_accuracy": 0.8850574712643678,
+      "eval_loss": 0.38414710760116577,
+      "eval_runtime": 8.6129,
+      "eval_samples_per_second": 10.101,
+      "eval_steps_per_second": 2.554,
       "step": 3750
     },
     {
       "epoch": 49.02,
-      "eval_accuracy": 0.9290322580645162,
-      "eval_loss": 0.18336611986160278,
-      "eval_runtime": 12.5538,
-      "eval_samples_per_second": 12.347,
-      "eval_steps_per_second": 3.107,
+      "eval_accuracy": 0.8850574712643678,
+      "eval_loss": 0.38414719700813293,
+      "eval_runtime": 6.1302,
+      "eval_samples_per_second": 14.192,
+      "eval_steps_per_second": 3.589,
       "step": 3750
     }
   ],

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d6e3454d677f838f72c44c0d25ed5379fc596af582a8d2edaaedf4baaf9e9edc
+oid sha256:54be8404f3a06187402196bf7251b8491aa663c5d9c26aa1c6f1e5be6bcb9d33
 size 3439