End of training

all_results.json

@@ -1,13 +1,13 @@
 {
     "epoch": 44.44,
     "eval_accuracy": 1.0,
-    "eval_loss": 0.017473837360739708,
-    "eval_runtime": 0.3229,
-    "eval_samples_per_second": 167.215,
-    "eval_steps_per_second": 3.097,
+    "eval_loss": 0.026890547946095467,
+    "eval_runtime": 0.3131,
+    "eval_samples_per_second": 172.459,
+    "eval_steps_per_second": 3.194,
     "total_flos": 1.6586385457107272e+18,
-    "train_loss": 0.15135242883116007,
-    "train_runtime": 1078.2452,
-    "train_samples_per_second": 22.305,
-    "train_steps_per_second": 0.093
+    "train_loss": 0.1744049086794257,
+    "train_runtime": 686.3928,
+    "train_samples_per_second": 35.038,
+    "train_steps_per_second": 0.146
 }

eval_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 44.44,
     "eval_accuracy": 1.0,
-    "eval_loss": 0.017473837360739708,
-    "eval_runtime": 0.3229,
-    "eval_samples_per_second": 167.215,
-    "eval_steps_per_second": 3.097
+    "eval_loss": 0.026890547946095467,
+    "eval_runtime": 0.3131,
+    "eval_samples_per_second": 172.459,
+    "eval_steps_per_second": 3.194
 }

train_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 44.44,
     "total_flos": 1.6586385457107272e+18,
-    "train_loss": 0.15135242883116007,
-    "train_runtime": 1078.2452,
-    "train_samples_per_second": 22.305,
-    "train_steps_per_second": 0.093
+    "train_loss": 0.1744049086794257,
+    "train_runtime": 686.3928,
+    "train_samples_per_second": 35.038,
+    "train_steps_per_second": 0.146
 }

trainer_state.json

@@ -1,6 +1,6 @@
 {
   "best_metric": 1.0,
-  "best_model_checkpoint": "vit-base-patch16-224-Trial007-YEL_STEM1/checkpoint-45",
+  "best_model_checkpoint": "vit-base-patch16-224-Trial007-YEL_STEM1/checkpoint-40",
   "epoch": 44.44444444444444,
   "global_step": 100,
   "is_hyper_param_search": false,
@@ -10,1016 +10,1016 @@
     {
       "epoch": 0.44,
       "learning_rate": 5e-06,
-      "loss": 0.6921,
+      "loss": 0.8294,
       "step": 1
     },
     {
       "epoch": 0.89,
       "learning_rate": 1e-05,
-      "loss": 0.6861,
+      "loss": 0.8443,
       "step": 2
     },
     {
       "epoch": 0.89,
-      "eval_accuracy": 0.5370370370370371,
-      "eval_loss": 0.6742474436759949,
-      "eval_runtime": 0.2725,
-      "eval_samples_per_second": 198.192,
-      "eval_steps_per_second": 3.67,
+      "eval_accuracy": 0.3148148148148148,
+      "eval_loss": 0.7812769412994385,
+      "eval_runtime": 0.2626,
+      "eval_samples_per_second": 205.601,
+      "eval_steps_per_second": 3.807,
       "step": 2
     },
     {
       "epoch": 1.33,
       "learning_rate": 1.5e-05,
-      "loss": 0.659,
+      "loss": 0.8169,
       "step": 3
     },
     {
       "epoch": 1.78,
       "learning_rate": 2e-05,
-      "loss": 0.6446,
+      "loss": 0.7501,
       "step": 4
     },
     {
       "epoch": 1.78,
-      "eval_accuracy": 0.7037037037037037,
-      "eval_loss": 0.5790768265724182,
-      "eval_runtime": 0.2769,
-      "eval_samples_per_second": 195.034,
-      "eval_steps_per_second": 3.612,
+      "eval_accuracy": 0.5555555555555556,
+      "eval_loss": 0.7086877226829529,
+      "eval_runtime": 0.2654,
+      "eval_samples_per_second": 203.478,
+      "eval_steps_per_second": 3.768,
       "step": 4
     },
     {
       "epoch": 2.22,
       "learning_rate": 2.5e-05,
-      "loss": 0.5569,
+      "loss": 0.663,
       "step": 5
     },
     {
       "epoch": 2.67,
       "learning_rate": 3e-05,
-      "loss": 0.5406,
+      "loss": 0.6312,
       "step": 6
     },
     {
       "epoch": 2.67,
-      "eval_accuracy": 0.8148148148148148,
-      "eval_loss": 0.43681636452674866,
-      "eval_runtime": 0.2645,
-      "eval_samples_per_second": 204.182,
-      "eval_steps_per_second": 3.781,
+      "eval_accuracy": 0.9074074074074074,
+      "eval_loss": 0.5305858254432678,
+      "eval_runtime": 0.2682,
+      "eval_samples_per_second": 201.351,
+      "eval_steps_per_second": 3.729,
       "step": 6
     },
     {
       "epoch": 3.11,
       "learning_rate": 3.5e-05,
-      "loss": 0.3935,
+      "loss": 0.4814,
       "step": 7
     },
     {
       "epoch": 3.56,
       "learning_rate": 4e-05,
-      "loss": 0.497,
+      "loss": 0.6005,
       "step": 8
     },
     {
       "epoch": 4.0,
       "learning_rate": 4.5e-05,
-      "loss": 0.328,
+      "loss": 0.4329,
       "step": 9
     },
     {
       "epoch": 4.0,
-      "eval_accuracy": 0.9259259259259259,
-      "eval_loss": 0.29564177989959717,
-      "eval_runtime": 0.2626,
-      "eval_samples_per_second": 205.67,
-      "eval_steps_per_second": 3.809,
+      "eval_accuracy": 0.9074074074074074,
+      "eval_loss": 0.3618362247943878,
+      "eval_runtime": 0.2584,
+      "eval_samples_per_second": 208.948,
+      "eval_steps_per_second": 3.869,
       "step": 9
     },
     {
       "epoch": 4.44,
       "learning_rate": 5e-05,
-      "loss": 0.3644,
+      "loss": 0.4233,
       "step": 10
     },
     {
       "epoch": 4.89,
       "learning_rate": 4.9444444444444446e-05,
-      "loss": 0.3238,
+      "loss": 0.4438,
       "step": 11
     },
     {
       "epoch": 4.89,
-      "eval_accuracy": 0.9259259259259259,
-      "eval_loss": 0.26564526557922363,
-      "eval_runtime": 0.2618,
-      "eval_samples_per_second": 206.277,
-      "eval_steps_per_second": 3.82,
+      "eval_accuracy": 0.9444444444444444,
+      "eval_loss": 0.26987290382385254,
+      "eval_runtime": 0.2579,
+      "eval_samples_per_second": 209.394,
+      "eval_steps_per_second": 3.878,
       "step": 11
     },
     {
       "epoch": 5.33,
       "learning_rate": 4.888888888888889e-05,
-      "loss": 0.2646,
+      "loss": 0.3444,
       "step": 12
     },
     {
       "epoch": 5.78,
       "learning_rate": 4.8333333333333334e-05,
-      "loss": 0.2851,
+      "loss": 0.3858,
       "step": 13
     },
     {
       "epoch": 5.78,
-      "eval_accuracy": 0.9259259259259259,
-      "eval_loss": 0.20708920061588287,
-      "eval_runtime": 0.3296,
-      "eval_samples_per_second": 163.85,
-      "eval_steps_per_second": 3.034,
+      "eval_accuracy": 0.7962962962962963,
+      "eval_loss": 0.3650400936603546,
+      "eval_runtime": 0.2868,
+      "eval_samples_per_second": 188.272,
+      "eval_steps_per_second": 3.487,
       "step": 13
     },
     {
       "epoch": 6.22,
       "learning_rate": 4.7777777777777784e-05,
-      "loss": 0.1854,
+      "loss": 0.3324,
       "step": 14
     },
     {
       "epoch": 6.67,
       "learning_rate": 4.722222222222222e-05,
-      "loss": 0.2767,
+      "loss": 0.339,
       "step": 15
     },
     {
       "epoch": 6.67,
-      "eval_accuracy": 0.9814814814814815,
-      "eval_loss": 0.13367433845996857,
-      "eval_runtime": 0.2809,
-      "eval_samples_per_second": 192.273,
-      "eval_steps_per_second": 3.561,
+      "eval_accuracy": 0.9629629629629629,
+      "eval_loss": 0.19111506640911102,
+      "eval_runtime": 0.2682,
+      "eval_samples_per_second": 201.363,
+      "eval_steps_per_second": 3.729,
       "step": 15
     },
     {
       "epoch": 7.11,
       "learning_rate": 4.666666666666667e-05,
-      "loss": 0.2161,
+      "loss": 0.2704,
       "step": 16
     },
     {
       "epoch": 7.56,
       "learning_rate": 4.6111111111111115e-05,
-      "loss": 0.1968,
+      "loss": 0.3059,
       "step": 17
     },
     {
       "epoch": 8.0,
       "learning_rate": 4.555555555555556e-05,
-      "loss": 0.1417,
+      "loss": 0.2852,
       "step": 18
     },
     {
       "epoch": 8.0,
-      "eval_accuracy": 0.8888888888888888,
-      "eval_loss": 0.24701863527297974,
-      "eval_runtime": 0.2963,
-      "eval_samples_per_second": 182.256,
-      "eval_steps_per_second": 3.375,
+      "eval_accuracy": 0.9629629629629629,
+      "eval_loss": 0.1610794961452484,
+      "eval_runtime": 0.2652,
+      "eval_samples_per_second": 203.582,
+      "eval_steps_per_second": 3.77,
       "step": 18
     },
     {
       "epoch": 8.44,
       "learning_rate": 4.5e-05,
-      "loss": 0.2184,
+      "loss": 0.1981,
       "step": 19
     },
     {
       "epoch": 8.89,
       "learning_rate": 4.4444444444444447e-05,
-      "loss": 0.1644,
+      "loss": 0.1866,
       "step": 20
     },
     {
       "epoch": 8.89,
-      "eval_accuracy": 0.9814814814814815,
-      "eval_loss": 0.09887748211622238,
-      "eval_runtime": 0.2603,
-      "eval_samples_per_second": 207.421,
-      "eval_steps_per_second": 3.841,
+      "eval_accuracy": 0.9444444444444444,
+      "eval_loss": 0.15157748758792877,
+      "eval_runtime": 0.2627,
+      "eval_samples_per_second": 205.573,
+      "eval_steps_per_second": 3.807,
       "step": 20
     },
     {
       "epoch": 9.33,
       "learning_rate": 4.388888888888889e-05,
-      "loss": 0.237,
+      "loss": 0.2223,
       "step": 21
     },
     {
       "epoch": 9.78,
       "learning_rate": 4.3333333333333334e-05,
-      "loss": 0.145,
+      "loss": 0.1748,
       "step": 22
     },
     {
       "epoch": 9.78,
-      "eval_accuracy": 0.9814814814814815,
-      "eval_loss": 0.12777921557426453,
-      "eval_runtime": 0.2613,
-      "eval_samples_per_second": 206.669,
-      "eval_steps_per_second": 3.827,
+      "eval_accuracy": 0.9629629629629629,
+      "eval_loss": 0.13326749205589294,
+      "eval_runtime": 0.2617,
+      "eval_samples_per_second": 206.317,
+      "eval_steps_per_second": 3.821,
       "step": 22
     },
     {
       "epoch": 10.22,
       "learning_rate": 4.277777777777778e-05,
-      "loss": 0.1986,
+      "loss": 0.1833,
       "step": 23
     },
     {
       "epoch": 10.67,
       "learning_rate": 4.222222222222222e-05,
-      "loss": 0.1948,
+      "loss": 0.1996,
       "step": 24
     },
     {
       "epoch": 10.67,
-      "eval_accuracy": 0.9814814814814815,
-      "eval_loss": 0.09025944769382477,
-      "eval_runtime": 0.2617,
-      "eval_samples_per_second": 206.322,
-      "eval_steps_per_second": 3.821,
+      "eval_accuracy": 0.9629629629629629,
+      "eval_loss": 0.11881023645401001,
+      "eval_runtime": 0.2602,
+      "eval_samples_per_second": 207.5,
+      "eval_steps_per_second": 3.843,
       "step": 24
     },
     {
       "epoch": 11.11,
       "learning_rate": 4.166666666666667e-05,
-      "loss": 0.101,
+      "loss": 0.1168,
       "step": 25
     },
     {
       "epoch": 11.56,
       "learning_rate": 4.111111111111111e-05,
-      "loss": 0.1034,
+      "loss": 0.1335,
       "step": 26
     },
     {
       "epoch": 12.0,
       "learning_rate": 4.055555555555556e-05,
-      "loss": 0.1407,
+      "loss": 0.1604,
       "step": 27
     },
     {
       "epoch": 12.0,
       "eval_accuracy": 0.9444444444444444,
-      "eval_loss": 0.10264571756124496,
-      "eval_runtime": 0.2631,
-      "eval_samples_per_second": 205.247,
-      "eval_steps_per_second": 3.801,
+      "eval_loss": 0.11688121408224106,
+      "eval_runtime": 0.2613,
+      "eval_samples_per_second": 206.693,
+      "eval_steps_per_second": 3.828,
       "step": 27
     },
     {
       "epoch": 12.44,
       "learning_rate": 4e-05,
-      "loss": 0.0916,
+      "loss": 0.1171,
       "step": 28
     },
     {
       "epoch": 12.89,
       "learning_rate": 3.944444444444445e-05,
-      "loss": 0.1213,
+      "loss": 0.1319,
       "step": 29
     },
     {
       "epoch": 12.89,
       "eval_accuracy": 0.9814814814814815,
-      "eval_loss": 0.0812113955616951,
-      "eval_runtime": 0.2626,
-      "eval_samples_per_second": 205.617,
-      "eval_steps_per_second": 3.808,
+      "eval_loss": 0.08352555334568024,
+      "eval_runtime": 0.2583,
+      "eval_samples_per_second": 209.02,
+      "eval_steps_per_second": 3.871,
       "step": 29
     },
     {
       "epoch": 13.33,
       "learning_rate": 3.888888888888889e-05,
-      "loss": 0.0859,
+      "loss": 0.0769,
       "step": 30
     },
     {
       "epoch": 13.78,
       "learning_rate": 3.8333333333333334e-05,
-      "loss": 0.1258,
+      "loss": 0.141,
       "step": 31
     },
     {
       "epoch": 13.78,
       "eval_accuracy": 0.9814814814814815,
-      "eval_loss": 0.09858040511608124,
-      "eval_runtime": 0.2841,
-      "eval_samples_per_second": 190.055,
-      "eval_steps_per_second": 3.52,
+      "eval_loss": 0.07039038091897964,
+      "eval_runtime": 0.2614,
+      "eval_samples_per_second": 206.613,
+      "eval_steps_per_second": 3.826,
       "step": 31
     },
     {
       "epoch": 14.22,
       "learning_rate": 3.777777777777778e-05,
-      "loss": 0.0752,
+      "loss": 0.0732,
       "step": 32
     },
     {
       "epoch": 14.67,
       "learning_rate": 3.722222222222222e-05,
-      "loss": 0.1204,
+      "loss": 0.123,
       "step": 33
     },
     {
       "epoch": 14.67,
       "eval_accuracy": 0.9814814814814815,
-      "eval_loss": 0.08133846521377563,
-      "eval_runtime": 0.2617,
-      "eval_samples_per_second": 206.327,
-      "eval_steps_per_second": 3.821,
+      "eval_loss": 0.05736014246940613,
+      "eval_runtime": 0.2587,
+      "eval_samples_per_second": 208.767,
+      "eval_steps_per_second": 3.866,
       "step": 33
     },
     {
       "epoch": 15.11,
       "learning_rate": 3.6666666666666666e-05,
-      "loss": 0.1135,
+      "loss": 0.1153,
       "step": 34
     },
     {
       "epoch": 15.56,
       "learning_rate": 3.611111111111111e-05,
-      "loss": 0.1176,
+      "loss": 0.1164,
       "step": 35
     },
     {
       "epoch": 16.0,
       "learning_rate": 3.555555555555556e-05,
-      "loss": 0.0857,
+      "loss": 0.0678,
       "step": 36
     },
     {
       "epoch": 16.0,
       "eval_accuracy": 0.9814814814814815,
-      "eval_loss": 0.04036113619804382,
-      "eval_runtime": 0.262,
-      "eval_samples_per_second": 206.086,
-      "eval_steps_per_second": 3.816,
+      "eval_loss": 0.0604197233915329,
+      "eval_runtime": 0.263,
+      "eval_samples_per_second": 205.352,
+      "eval_steps_per_second": 3.803,
       "step": 36
     },
     {
       "epoch": 16.44,
       "learning_rate": 3.5e-05,
-      "loss": 0.0889,
+      "loss": 0.0914,
       "step": 37
     },
     {
       "epoch": 16.89,
       "learning_rate": 3.444444444444445e-05,
-      "loss": 0.12,
+      "loss": 0.1208,
       "step": 38
     },
     {
       "epoch": 16.89,
       "eval_accuracy": 0.9814814814814815,
-      "eval_loss": 0.05832526460289955,
-      "eval_runtime": 0.2597,
-      "eval_samples_per_second": 207.917,
-      "eval_steps_per_second": 3.85,
+      "eval_loss": 0.03850555792450905,
+      "eval_runtime": 0.2602,
+      "eval_samples_per_second": 207.493,
+      "eval_steps_per_second": 3.842,
       "step": 38
     },
     {
       "epoch": 17.33,
       "learning_rate": 3.388888888888889e-05,
-      "loss": 0.0841,
+      "loss": 0.0833,
       "step": 39
     },
     {
       "epoch": 17.78,
       "learning_rate": 3.3333333333333335e-05,
-      "loss": 0.0929,
+      "loss": 0.0942,
       "step": 40
     },
     {
       "epoch": 17.78,
-      "eval_accuracy": 0.9814814814814815,
-      "eval_loss": 0.07853430509567261,
-      "eval_runtime": 0.262,
-      "eval_samples_per_second": 206.126,
-      "eval_steps_per_second": 3.817,
+      "eval_accuracy": 1.0,
+      "eval_loss": 0.026890547946095467,
+      "eval_runtime": 0.2611,
+      "eval_samples_per_second": 206.793,
+      "eval_steps_per_second": 3.829,
       "step": 40
     },
     {
       "epoch": 18.22,
       "learning_rate": 3.277777777777778e-05,
-      "loss": 0.0988,
+      "loss": 0.0736,
       "step": 41
     },
     {
       "epoch": 18.67,
       "learning_rate": 3.222222222222223e-05,
-      "loss": 0.0997,
+      "loss": 0.0822,
       "step": 42
     },
     {
       "epoch": 18.67,
-      "eval_accuracy": 0.9814814814814815,
-      "eval_loss": 0.04468917474150658,
-      "eval_runtime": 0.2628,
-      "eval_samples_per_second": 205.474,
-      "eval_steps_per_second": 3.805,
+      "eval_accuracy": 1.0,
+      "eval_loss": 0.01689229905605316,
+      "eval_runtime": 0.2602,
+      "eval_samples_per_second": 207.555,
+      "eval_steps_per_second": 3.844,
       "step": 42
     },
     {
       "epoch": 19.11,
       "learning_rate": 3.1666666666666666e-05,
-      "loss": 0.0674,
+      "loss": 0.086,
       "step": 43
     },
     {
       "epoch": 19.56,
       "learning_rate": 3.111111111111111e-05,
-      "loss": 0.0985,
+      "loss": 0.1013,
       "step": 44
     },
     {
       "epoch": 20.0,
       "learning_rate": 3.055555555555556e-05,
-      "loss": 0.0649,
+      "loss": 0.0578,
       "step": 45
     },
     {
       "epoch": 20.0,
       "eval_accuracy": 1.0,
-      "eval_loss": 0.017473837360739708,
-      "eval_runtime": 0.2637,
-      "eval_samples_per_second": 204.747,
-      "eval_steps_per_second": 3.792,
+      "eval_loss": 0.017471354454755783,
+      "eval_runtime": 0.2785,
+      "eval_samples_per_second": 193.881,
+      "eval_steps_per_second": 3.59,
       "step": 45
     },
     {
       "epoch": 20.44,
       "learning_rate": 3e-05,
-      "loss": 0.0953,
+      "loss": 0.0943,
       "step": 46
     },
     {
       "epoch": 20.89,
       "learning_rate": 2.9444444444444448e-05,
-      "loss": 0.0709,
+      "loss": 0.0611,
       "step": 47
     },
     {
       "epoch": 20.89,
       "eval_accuracy": 1.0,
-      "eval_loss": 0.021981148049235344,
-      "eval_runtime": 0.2758,
-      "eval_samples_per_second": 195.775,
-      "eval_steps_per_second": 3.625,
+      "eval_loss": 0.022001970559358597,
+      "eval_runtime": 0.2601,
+      "eval_samples_per_second": 207.599,
+      "eval_steps_per_second": 3.844,
       "step": 47
     },
     {
       "epoch": 21.33,
       "learning_rate": 2.8888888888888888e-05,
-      "loss": 0.0808,
+      "loss": 0.083,
       "step": 48
     },
     {
       "epoch": 21.78,
       "learning_rate": 2.8333333333333335e-05,
-      "loss": 0.1041,
+      "loss": 0.1053,
       "step": 49
     },
     {
       "epoch": 21.78,
-      "eval_accuracy": 0.9814814814814815,
-      "eval_loss": 0.03725693002343178,
-      "eval_runtime": 0.2628,
-      "eval_samples_per_second": 205.518,
-      "eval_steps_per_second": 3.806,
+      "eval_accuracy": 1.0,
+      "eval_loss": 0.0097891166806221,
+      "eval_runtime": 0.2612,
+      "eval_samples_per_second": 206.762,
+      "eval_steps_per_second": 3.829,
       "step": 49
     },
     {
       "epoch": 22.22,
       "learning_rate": 2.777777777777778e-05,
-      "loss": 0.0706,
+      "loss": 0.0638,
       "step": 50
     },
     {
       "epoch": 22.67,
       "learning_rate": 2.7222222222222223e-05,
-      "loss": 0.1194,
+      "loss": 0.1713,
       "step": 51
     },
     {
       "epoch": 22.67,
-      "eval_accuracy": 0.9814814814814815,
-      "eval_loss": 0.03398296236991882,
-      "eval_runtime": 0.3775,
-      "eval_samples_per_second": 143.045,
-      "eval_steps_per_second": 2.649,
+      "eval_accuracy": 1.0,
+      "eval_loss": 0.015564966946840286,
+      "eval_runtime": 0.2597,
+      "eval_samples_per_second": 207.902,
+      "eval_steps_per_second": 3.85,
       "step": 51
     },
     {
       "epoch": 23.11,
       "learning_rate": 2.6666666666666667e-05,
-      "loss": 0.082,
+      "loss": 0.1014,
       "step": 52
     },
     {
       "epoch": 23.56,
       "learning_rate": 2.6111111111111114e-05,
-      "loss": 0.087,
+      "loss": 0.0871,
       "step": 53
     },
     {
       "epoch": 24.0,
       "learning_rate": 2.5555555555555554e-05,
-      "loss": 0.07,
+      "loss": 0.0515,
       "step": 54
     },
     {
       "epoch": 24.0,
       "eval_accuracy": 1.0,
-      "eval_loss": 0.023272547870874405,
-      "eval_runtime": 0.2622,
-      "eval_samples_per_second": 205.956,
-      "eval_steps_per_second": 3.814,
+      "eval_loss": 0.011080882512032986,
+      "eval_runtime": 0.2623,
+      "eval_samples_per_second": 205.857,
+      "eval_steps_per_second": 3.812,
       "step": 54
     },
     {
       "epoch": 24.44,
       "learning_rate": 2.5e-05,
-      "loss": 0.0837,
+      "loss": 0.1139,
       "step": 55
     },
     {
       "epoch": 24.89,
       "learning_rate": 2.4444444444444445e-05,
-      "loss": 0.0778,
+      "loss": 0.1227,
       "step": 56
     },
     {
       "epoch": 24.89,
       "eval_accuracy": 1.0,
-      "eval_loss": 0.017261186614632607,
-      "eval_runtime": 0.2621,
-      "eval_samples_per_second": 206.01,
-      "eval_steps_per_second": 3.815,
+      "eval_loss": 0.01661752350628376,
+      "eval_runtime": 0.2611,
+      "eval_samples_per_second": 206.808,
+      "eval_steps_per_second": 3.83,
       "step": 56
     },
     {
       "epoch": 25.33,
       "learning_rate": 2.3888888888888892e-05,
-      "loss": 0.0507,
+      "loss": 0.0627,
       "step": 57
     },
     {
       "epoch": 25.78,
       "learning_rate": 2.3333333333333336e-05,
-      "loss": 0.07,
+      "loss": 0.0891,
       "step": 58
     },
     {
       "epoch": 25.78,
       "eval_accuracy": 1.0,
-      "eval_loss": 0.015340018086135387,
-      "eval_runtime": 0.2636,
-      "eval_samples_per_second": 204.892,
-      "eval_steps_per_second": 3.794,
+      "eval_loss": 0.0093127666041255,
+      "eval_runtime": 0.2618,
+      "eval_samples_per_second": 206.246,
+      "eval_steps_per_second": 3.819,
       "step": 58
     },
     {
       "epoch": 26.22,
       "learning_rate": 2.277777777777778e-05,
-      "loss": 0.0984,
+      "loss": 0.1174,
       "step": 59
     },
     {
       "epoch": 26.67,
       "learning_rate": 2.2222222222222223e-05,
-      "loss": 0.0616,
+      "loss": 0.0768,
       "step": 60
     },
     {
       "epoch": 26.67,
       "eval_accuracy": 1.0,
-      "eval_loss": 0.01479976624250412,
-      "eval_runtime": 0.271,
-      "eval_samples_per_second": 199.266,
-      "eval_steps_per_second": 3.69,
+      "eval_loss": 0.008994304575026035,
+      "eval_runtime": 0.2612,
+      "eval_samples_per_second": 206.756,
+      "eval_steps_per_second": 3.829,
       "step": 60
     },
     {
       "epoch": 27.11,
       "learning_rate": 2.1666666666666667e-05,
-      "loss": 0.103,
+      "loss": 0.1328,
       "step": 61
     },
     {
       "epoch": 27.56,
       "learning_rate": 2.111111111111111e-05,
-      "loss": 0.0802,
+      "loss": 0.0935,
       "step": 62
     },
     {
       "epoch": 28.0,
       "learning_rate": 2.0555555555555555e-05,
-      "loss": 0.0675,
+      "loss": 0.0755,
       "step": 63
     },
     {
       "epoch": 28.0,
       "eval_accuracy": 1.0,
-      "eval_loss": 0.0183400996029377,
-      "eval_runtime": 0.3105,
-      "eval_samples_per_second": 173.918,
-      "eval_steps_per_second": 3.221,
+      "eval_loss": 0.010785534046590328,
+      "eval_runtime": 0.2627,
+      "eval_samples_per_second": 205.56,
+      "eval_steps_per_second": 3.807,
       "step": 63
     },
     {
       "epoch": 28.44,
       "learning_rate": 2e-05,
-      "loss": 0.1179,
+      "loss": 0.117,
       "step": 64
     },
     {
       "epoch": 28.89,
       "learning_rate": 1.9444444444444445e-05,
-      "loss": 0.0763,
+      "loss": 0.0798,
       "step": 65
     },
     {
       "epoch": 28.89,
       "eval_accuracy": 1.0,
-      "eval_loss": 0.015255843289196491,
-      "eval_runtime": 0.2619,
-      "eval_samples_per_second": 206.15,
-      "eval_steps_per_second": 3.818,
+      "eval_loss": 0.020129548385739326,
+      "eval_runtime": 0.264,
+      "eval_samples_per_second": 204.578,
+      "eval_steps_per_second": 3.788,
       "step": 65
     },
     {
       "epoch": 29.33,
       "learning_rate": 1.888888888888889e-05,
-      "loss": 0.0636,
+      "loss": 0.0671,
       "step": 66
     },
     {
       "epoch": 29.78,
       "learning_rate": 1.8333333333333333e-05,
-      "loss": 0.083,
+      "loss": 0.1005,
       "step": 67
     },
     {
       "epoch": 29.78,
       "eval_accuracy": 1.0,
-      "eval_loss": 0.010834791697561741,
-      "eval_runtime": 0.2626,
-      "eval_samples_per_second": 205.618,
-      "eval_steps_per_second": 3.808,
+      "eval_loss": 0.011831928044557571,
+      "eval_runtime": 0.2639,
+      "eval_samples_per_second": 204.649,
+      "eval_steps_per_second": 3.79,
       "step": 67
     },
     {
       "epoch": 30.22,
       "learning_rate": 1.777777777777778e-05,
-      "loss": 0.1013,
+      "loss": 0.0572,
       "step": 68
     },
     {
       "epoch": 30.67,
       "learning_rate": 1.7222222222222224e-05,
-      "loss": 0.1185,
+      "loss": 0.1113,
       "step": 69
     },
     {
       "epoch": 30.67,
       "eval_accuracy": 1.0,
-      "eval_loss": 0.014811315573751926,
-      "eval_runtime": 0.2712,
-      "eval_samples_per_second": 199.085,
-      "eval_steps_per_second": 3.687,
+      "eval_loss": 0.013057045638561249,
+      "eval_runtime": 0.2625,
+      "eval_samples_per_second": 205.749,
+      "eval_steps_per_second": 3.81,
       "step": 69
     },
     {
       "epoch": 31.11,
       "learning_rate": 1.6666666666666667e-05,
-      "loss": 0.0456,
+      "loss": 0.0474,
       "step": 70
     },
     {
       "epoch": 31.56,
       "learning_rate": 1.6111111111111115e-05,
-      "loss": 0.1482,
+      "loss": 0.1374,
       "step": 71
     },
     {
       "epoch": 32.0,
       "learning_rate": 1.5555555555555555e-05,
-      "loss": 0.0911,
+      "loss": 0.1034,
       "step": 72
     },
     {
       "epoch": 32.0,
       "eval_accuracy": 1.0,
-      "eval_loss": 0.014465795829892159,
-      "eval_runtime": 0.2594,
-      "eval_samples_per_second": 208.185,
-      "eval_steps_per_second": 3.855,
+      "eval_loss": 0.017138680443167686,
+      "eval_runtime": 0.2623,
+      "eval_samples_per_second": 205.904,
+      "eval_steps_per_second": 3.813,
       "step": 72
     },
     {
       "epoch": 32.44,
       "learning_rate": 1.5e-05,
-      "loss": 0.0615,
+      "loss": 0.0604,
       "step": 73
     },
     {
       "epoch": 32.89,
       "learning_rate": 1.4444444444444444e-05,
-      "loss": 0.0868,
+      "loss": 0.0857,
       "step": 74
     },
     {
       "epoch": 32.89,
       "eval_accuracy": 1.0,
-      "eval_loss": 0.015387580730021,
-      "eval_runtime": 0.2739,
-      "eval_samples_per_second": 197.149,
-      "eval_steps_per_second": 3.651,
+      "eval_loss": 0.015838121995329857,
+      "eval_runtime": 0.2627,
+      "eval_samples_per_second": 205.551,
+      "eval_steps_per_second": 3.807,
       "step": 74
     },
     {
       "epoch": 33.33,
       "learning_rate": 1.388888888888889e-05,
-      "loss": 0.1134,
+      "loss": 0.1107,
       "step": 75
     },
     {
       "epoch": 33.78,
       "learning_rate": 1.3333333333333333e-05,
-      "loss": 0.0941,
+      "loss": 0.0864,
       "step": 76
     },
     {
       "epoch": 33.78,
       "eval_accuracy": 1.0,
-      "eval_loss": 0.014932823367416859,
-      "eval_runtime": 0.2614,
-      "eval_samples_per_second": 206.594,
-      "eval_steps_per_second": 3.826,
+      "eval_loss": 0.014146910980343819,
+      "eval_runtime": 0.2643,
+      "eval_samples_per_second": 204.346,
+      "eval_steps_per_second": 3.784,
       "step": 76
     },
     {
       "epoch": 34.22,
       "learning_rate": 1.2777777777777777e-05,
-      "loss": 0.0976,
+      "loss": 0.096,
       "step": 77
     },
     {
       "epoch": 34.67,
       "learning_rate": 1.2222222222222222e-05,
-      "loss": 0.1182,
+      "loss": 0.1241,
       "step": 78
     },
     {
       "epoch": 34.67,
       "eval_accuracy": 1.0,
-      "eval_loss": 0.014534077607095242,
-      "eval_runtime": 0.2631,
-      "eval_samples_per_second": 205.268,
-      "eval_steps_per_second": 3.801,
+      "eval_loss": 0.012656173668801785,
+      "eval_runtime": 0.2613,
+      "eval_samples_per_second": 206.683,
+      "eval_steps_per_second": 3.827,
       "step": 78
     },
     {
       "epoch": 35.11,
       "learning_rate": 1.1666666666666668e-05,
-      "loss": 0.1337,
+      "loss": 0.2137,
       "step": 79
     },
     {
       "epoch": 35.56,
       "learning_rate": 1.1111111111111112e-05,
-      "loss": 0.0878,
+      "loss": 0.1045,
       "step": 80
     },
     {
       "epoch": 36.0,
       "learning_rate": 1.0555555555555555e-05,
-      "loss": 0.0859,
+      "loss": 0.0868,
       "step": 81
     },
     {
       "epoch": 36.0,
       "eval_accuracy": 1.0,
-      "eval_loss": 0.01714230887591839,
-      "eval_runtime": 0.2613,
-      "eval_samples_per_second": 206.651,
-      "eval_steps_per_second": 3.827,
+      "eval_loss": 0.011819291859865189,
+      "eval_runtime": 0.2714,
+      "eval_samples_per_second": 198.995,
+      "eval_steps_per_second": 3.685,
       "step": 81
     },
     {
       "epoch": 36.44,
       "learning_rate": 1e-05,
-      "loss": 0.0814,
+      "loss": 0.0863,
       "step": 82
     },
     {
       "epoch": 36.89,
       "learning_rate": 9.444444444444445e-06,
-      "loss": 0.071,
+      "loss": 0.0704,
       "step": 83
     },
     {
       "epoch": 36.89,
       "eval_accuracy": 1.0,
-      "eval_loss": 0.018014799803495407,
-      "eval_runtime": 0.2725,
-      "eval_samples_per_second": 198.155,
-      "eval_steps_per_second": 3.67,
+      "eval_loss": 0.011303897947072983,
+      "eval_runtime": 0.2629,
+      "eval_samples_per_second": 205.397,
+      "eval_steps_per_second": 3.804,
       "step": 83
     },
     {
       "epoch": 37.33,
       "learning_rate": 8.88888888888889e-06,
-      "loss": 0.0635,
+      "loss": 0.0523,
       "step": 84
     },
     {
       "epoch": 37.78,
       "learning_rate": 8.333333333333334e-06,
-      "loss": 0.1003,
+      "loss": 0.0938,
       "step": 85
     },
     {
       "epoch": 37.78,
       "eval_accuracy": 1.0,
-      "eval_loss": 0.015952540561556816,
-      "eval_runtime": 0.2716,
-      "eval_samples_per_second": 198.786,
-      "eval_steps_per_second": 3.681,
+      "eval_loss": 0.010850733146071434,
+      "eval_runtime": 0.2773,
+      "eval_samples_per_second": 194.757,
+      "eval_steps_per_second": 3.607,
       "step": 85
     },
     {
       "epoch": 38.22,
       "learning_rate": 7.777777777777777e-06,
-      "loss": 0.0761,
+      "loss": 0.0838,
       "step": 86
     },
     {
       "epoch": 38.67,
       "learning_rate": 7.222222222222222e-06,
-      "loss": 0.1173,
+      "loss": 0.1181,
       "step": 87
     },
     {
       "epoch": 38.67,
       "eval_accuracy": 1.0,
-      "eval_loss": 0.016635410487651825,
-      "eval_runtime": 0.2617,
-      "eval_samples_per_second": 206.357,
-      "eval_steps_per_second": 3.821,
+      "eval_loss": 0.011989281512796879,
+      "eval_runtime": 0.2593,
+      "eval_samples_per_second": 208.25,
+      "eval_steps_per_second": 3.856,
       "step": 87
     },
     {
       "epoch": 39.11,
       "learning_rate": 6.666666666666667e-06,
-      "loss": 0.0596,
+      "loss": 0.0638,
       "step": 88
     },
     {
       "epoch": 39.56,
       "learning_rate": 6.111111111111111e-06,
-      "loss": 0.0549,
+      "loss": 0.0471,
       "step": 89
     },
     {
       "epoch": 40.0,
       "learning_rate": 5.555555555555556e-06,
-      "loss": 0.0478,
+      "loss": 0.0509,
       "step": 90
     },
     {
       "epoch": 40.0,
       "eval_accuracy": 1.0,
-      "eval_loss": 0.020745275542140007,
-      "eval_runtime": 0.2774,
-      "eval_samples_per_second": 194.642,
-      "eval_steps_per_second": 3.604,
+      "eval_loss": 0.014911294914782047,
+      "eval_runtime": 0.2616,
+      "eval_samples_per_second": 206.386,
+      "eval_steps_per_second": 3.822,
       "step": 90
     },
     {
       "epoch": 40.44,
       "learning_rate": 5e-06,
-      "loss": 0.1045,
+      "loss": 0.1063,
       "step": 91
     },
     {
       "epoch": 40.89,
       "learning_rate": 4.444444444444445e-06,
-      "loss": 0.0674,
+      "loss": 0.0684,
       "step": 92
     },
     {
       "epoch": 40.89,
-      "eval_accuracy": 0.9814814814814815,
-      "eval_loss": 0.022440657019615173,
-      "eval_runtime": 0.2613,
-      "eval_samples_per_second": 206.672,
-      "eval_steps_per_second": 3.827,
+      "eval_accuracy": 1.0,
+      "eval_loss": 0.015459185466170311,
+      "eval_runtime": 0.2635,
+      "eval_samples_per_second": 204.956,
+      "eval_steps_per_second": 3.795,
       "step": 92
     },
     {
       "epoch": 41.33,
       "learning_rate": 3.888888888888889e-06,
-      "loss": 0.077,
+      "loss": 0.0775,
       "step": 93
     },
     {
       "epoch": 41.78,
       "learning_rate": 3.3333333333333333e-06,
-      "loss": 0.0623,
+      "loss": 0.0625,
       "step": 94
     },
     {
       "epoch": 41.78,
-      "eval_accuracy": 0.9814814814814815,
-      "eval_loss": 0.02340044267475605,
-      "eval_runtime": 0.2622,
-      "eval_samples_per_second": 205.961,
-      "eval_steps_per_second": 3.814,
+      "eval_accuracy": 1.0,
+      "eval_loss": 0.015089095570147038,
+      "eval_runtime": 0.2645,
+      "eval_samples_per_second": 204.187,
+      "eval_steps_per_second": 3.781,
       "step": 94
     },
     {
       "epoch": 42.22,
       "learning_rate": 2.777777777777778e-06,
-      "loss": 0.0503,
+      "loss": 0.0474,
       "step": 95
     },
     {
       "epoch": 42.67,
       "learning_rate": 2.2222222222222225e-06,
-      "loss": 0.0711,
+      "loss": 0.0746,
       "step": 96
     },
     {
       "epoch": 42.67,
-      "eval_accuracy": 0.9814814814814815,
-      "eval_loss": 0.023668842390179634,
-      "eval_runtime": 0.2602,
-      "eval_samples_per_second": 207.57,
-      "eval_steps_per_second": 3.844,
+      "eval_accuracy": 1.0,
+      "eval_loss": 0.014267035759985447,
+      "eval_runtime": 0.2776,
+      "eval_samples_per_second": 194.496,
+      "eval_steps_per_second": 3.602,
       "step": 96
     },
     {
       "epoch": 43.11,
       "learning_rate": 1.6666666666666667e-06,
-      "loss": 0.0887,
+      "loss": 0.0868,
       "step": 97
     },
     {
       "epoch": 43.56,
       "learning_rate": 1.1111111111111112e-06,
-      "loss": 0.0767,
+      "loss": 0.0849,
       "step": 98
     },
     {
       "epoch": 44.0,
       "learning_rate": 5.555555555555556e-07,
-      "loss": 0.1057,
+      "loss": 0.1062,
       "step": 99
     },
     {
       "epoch": 44.0,
-      "eval_accuracy": 0.9814814814814815,
-      "eval_loss": 0.023580273613333702,
-      "eval_runtime": 0.2623,
-      "eval_samples_per_second": 205.883,
-      "eval_steps_per_second": 3.813,
+      "eval_accuracy": 1.0,
+      "eval_loss": 0.013263082131743431,
+      "eval_runtime": 0.2615,
+      "eval_samples_per_second": 206.506,
+      "eval_steps_per_second": 3.824,
       "step": 99
     },
     {
       "epoch": 44.44,
       "learning_rate": 0.0,
-      "loss": 0.0541,
+      "loss": 0.0579,
       "step": 100
     },
     {
       "epoch": 44.44,
-      "eval_accuracy": 0.9814814814814815,
-      "eval_loss": 0.02352498471736908,
-      "eval_runtime": 0.2719,
-      "eval_samples_per_second": 198.613,
-      "eval_steps_per_second": 3.678,
+      "eval_accuracy": 1.0,
+      "eval_loss": 0.013171697966754436,
+      "eval_runtime": 0.263,
+      "eval_samples_per_second": 205.31,
+      "eval_steps_per_second": 3.802,
       "step": 100
     },
     {
       "epoch": 44.44,
       "step": 100,
       "total_flos": 1.6586385457107272e+18,
-      "train_loss": 0.15135242883116007,
-      "train_runtime": 1078.2452,
-      "train_samples_per_second": 22.305,
-      "train_steps_per_second": 0.093
+      "train_loss": 0.1744049086794257,
+      "train_runtime": 686.3928,
+      "train_samples_per_second": 35.038,
+      "train_steps_per_second": 0.146
     }
   ],
   "max_steps": 100,