End of training

all_results.json

@@ -1,13 +1,13 @@
 {
     "epoch": 44.44,
     "eval_accuracy": 1.0,
-    "eval_loss": 0.03582725301384926,
-    "eval_runtime": 0.3142,
-    "eval_samples_per_second": 171.879,
-    "eval_steps_per_second": 3.183,
+    "eval_loss": 0.03731463849544525,
+    "eval_runtime": 0.3143,
+    "eval_samples_per_second": 171.8,
+    "eval_steps_per_second": 3.181,
     "total_flos": 1.6586385457107272e+18,
-    "train_loss": 0.2113973332196474,
-    "train_runtime": 676.842,
-    "train_samples_per_second": 35.533,
-    "train_steps_per_second": 0.148
+    "train_loss": 0.20484884686768054,
+    "train_runtime": 700.3363,
+    "train_samples_per_second": 34.341,
+    "train_steps_per_second": 0.143
 }

eval_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 44.44,
     "eval_accuracy": 1.0,
-    "eval_loss": 0.03582725301384926,
-    "eval_runtime": 0.3142,
-    "eval_samples_per_second": 171.879,
-    "eval_steps_per_second": 3.183
+    "eval_loss": 0.03731463849544525,
+    "eval_runtime": 0.3143,
+    "eval_samples_per_second": 171.8,
+    "eval_steps_per_second": 3.181
 }

train_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 44.44,
     "total_flos": 1.6586385457107272e+18,
-    "train_loss": 0.2113973332196474,
-    "train_runtime": 676.842,
-    "train_samples_per_second": 35.533,
-    "train_steps_per_second": 0.148
+    "train_loss": 0.20484884686768054,
+    "train_runtime": 700.3363,
+    "train_samples_per_second": 34.341,
+    "train_steps_per_second": 0.143
 }

trainer_state.json

@@ -1,6 +1,6 @@
 {
   "best_metric": 1.0,
-  "best_model_checkpoint": "vit-base-patch16-224-Trial007-YEL_STEM/checkpoint-36",
+  "best_model_checkpoint": "vit-base-patch16-224-Trial007-YEL_STEM/checkpoint-29",
   "epoch": 44.44444444444444,
   "global_step": 100,
   "is_hyper_param_search": false,
@@ -10,1016 +10,1016 @@
     {
       "epoch": 0.44,
       "learning_rate": 5e-06,
-      "loss": 0.71,
+      "loss": 0.7059,
       "step": 1
     },
     {
       "epoch": 0.89,
       "learning_rate": 1e-05,
-      "loss": 0.7523,
+      "loss": 0.7081,
       "step": 2
     },
     {
       "epoch": 0.89,
-      "eval_accuracy": 0.6111111111111112,
-      "eval_loss": 0.6941519379615784,
-      "eval_runtime": 0.4141,
-      "eval_samples_per_second": 130.403,
-      "eval_steps_per_second": 2.415,
+      "eval_accuracy": 0.5555555555555556,
+      "eval_loss": 0.6817994713783264,
+      "eval_runtime": 0.2599,
+      "eval_samples_per_second": 207.804,
+      "eval_steps_per_second": 3.848,
       "step": 2
     },
     {
       "epoch": 1.33,
       "learning_rate": 1.5e-05,
-      "loss": 0.6296,
+      "loss": 0.6401,
       "step": 3
     },
     {
       "epoch": 1.78,
       "learning_rate": 2e-05,
-      "loss": 0.6785,
+      "loss": 0.6584,
       "step": 4
     },
     {
       "epoch": 1.78,
-      "eval_accuracy": 0.5925925925925926,
-      "eval_loss": 0.6312435269355774,
-      "eval_runtime": 0.2601,
-      "eval_samples_per_second": 207.61,
-      "eval_steps_per_second": 3.845,
+      "eval_accuracy": 0.7037037037037037,
+      "eval_loss": 0.5915446877479553,
+      "eval_runtime": 0.2575,
+      "eval_samples_per_second": 209.714,
+      "eval_steps_per_second": 3.884,
       "step": 4
     },
     {
       "epoch": 2.22,
       "learning_rate": 2.5e-05,
-      "loss": 0.6053,
+      "loss": 0.5794,
       "step": 5
     },
     {
       "epoch": 2.67,
       "learning_rate": 3e-05,
-      "loss": 0.5733,
+      "loss": 0.5552,
       "step": 6
     },
     {
       "epoch": 2.67,
-      "eval_accuracy": 0.8518518518518519,
-      "eval_loss": 0.4705863296985626,
-      "eval_runtime": 0.2543,
-      "eval_samples_per_second": 212.329,
-      "eval_steps_per_second": 3.932,
+      "eval_accuracy": 0.7407407407407407,
+      "eval_loss": 0.5365740656852722,
+      "eval_runtime": 0.2565,
+      "eval_samples_per_second": 210.539,
+      "eval_steps_per_second": 3.899,
       "step": 6
     },
     {
       "epoch": 3.11,
       "learning_rate": 3.5e-05,
-      "loss": 0.4955,
+      "loss": 0.4553,
       "step": 7
     },
     {
       "epoch": 3.56,
       "learning_rate": 4e-05,
-      "loss": 0.6037,
+      "loss": 0.5446,
       "step": 8
     },
     {
       "epoch": 4.0,
       "learning_rate": 4.5e-05,
-      "loss": 0.4473,
+      "loss": 0.3763,
       "step": 9
     },
     {
       "epoch": 4.0,
-      "eval_accuracy": 0.9074074074074074,
-      "eval_loss": 0.3479275405406952,
-      "eval_runtime": 0.2771,
-      "eval_samples_per_second": 194.879,
-      "eval_steps_per_second": 3.609,
+      "eval_accuracy": 0.8518518518518519,
+      "eval_loss": 0.35601484775543213,
+      "eval_runtime": 0.2638,
+      "eval_samples_per_second": 204.731,
+      "eval_steps_per_second": 3.791,
       "step": 9
     },
     {
       "epoch": 4.44,
       "learning_rate": 5e-05,
-      "loss": 0.463,
+      "loss": 0.4622,
       "step": 10
     },
     {
       "epoch": 4.89,
       "learning_rate": 4.9444444444444446e-05,
-      "loss": 0.4312,
+      "loss": 0.397,
       "step": 11
     },
     {
       "epoch": 4.89,
-      "eval_accuracy": 0.9444444444444444,
-      "eval_loss": 0.27067238092422485,
-      "eval_runtime": 0.2603,
-      "eval_samples_per_second": 207.479,
-      "eval_steps_per_second": 3.842,
+      "eval_accuracy": 0.8518518518518519,
+      "eval_loss": 0.2999265789985657,
+      "eval_runtime": 0.2681,
+      "eval_samples_per_second": 201.389,
+      "eval_steps_per_second": 3.729,
       "step": 11
     },
     {
       "epoch": 5.33,
       "learning_rate": 4.888888888888889e-05,
-      "loss": 0.3057,
+      "loss": 0.2849,
       "step": 12
     },
     {
       "epoch": 5.78,
       "learning_rate": 4.8333333333333334e-05,
-      "loss": 0.3499,
+      "loss": 0.3313,
       "step": 13
     },
     {
       "epoch": 5.78,
       "eval_accuracy": 0.9074074074074074,
-      "eval_loss": 0.24450431764125824,
-      "eval_runtime": 0.3251,
-      "eval_samples_per_second": 166.098,
-      "eval_steps_per_second": 3.076,
+      "eval_loss": 0.23073385655879974,
+      "eval_runtime": 0.2595,
+      "eval_samples_per_second": 208.115,
+      "eval_steps_per_second": 3.854,
       "step": 13
     },
     {
       "epoch": 6.22,
       "learning_rate": 4.7777777777777784e-05,
-      "loss": 0.2501,
+      "loss": 0.2448,
       "step": 14
     },
     {
       "epoch": 6.67,
       "learning_rate": 4.722222222222222e-05,
-      "loss": 0.3096,
+      "loss": 0.2957,
       "step": 15
     },
     {
       "epoch": 6.67,
-      "eval_accuracy": 0.9629629629629629,
-      "eval_loss": 0.17727157473564148,
-      "eval_runtime": 0.2547,
-      "eval_samples_per_second": 212.003,
-      "eval_steps_per_second": 3.926,
+      "eval_accuracy": 0.9259259259259259,
+      "eval_loss": 0.1745777279138565,
+      "eval_runtime": 0.2586,
+      "eval_samples_per_second": 208.833,
+      "eval_steps_per_second": 3.867,
       "step": 15
     },
     {
       "epoch": 7.11,
       "learning_rate": 4.666666666666667e-05,
-      "loss": 0.2454,
+      "loss": 0.2555,
       "step": 16
     },
     {
       "epoch": 7.56,
       "learning_rate": 4.6111111111111115e-05,
-      "loss": 0.272,
+      "loss": 0.3153,
       "step": 17
     },
     {
       "epoch": 8.0,
       "learning_rate": 4.555555555555556e-05,
-      "loss": 0.1992,
+      "loss": 0.2383,
       "step": 18
     },
     {
       "epoch": 8.0,
       "eval_accuracy": 0.9444444444444444,
-      "eval_loss": 0.14089126884937286,
-      "eval_runtime": 0.2563,
-      "eval_samples_per_second": 210.661,
-      "eval_steps_per_second": 3.901,
+      "eval_loss": 0.14317429065704346,
+      "eval_runtime": 0.2592,
+      "eval_samples_per_second": 208.343,
+      "eval_steps_per_second": 3.858,
       "step": 18
     },
     {
       "epoch": 8.44,
       "learning_rate": 4.5e-05,
-      "loss": 0.2011,
+      "loss": 0.1994,
       "step": 19
     },
     {
       "epoch": 8.89,
       "learning_rate": 4.4444444444444447e-05,
-      "loss": 0.2195,
+      "loss": 0.2664,
       "step": 20
     },
     {
       "epoch": 8.89,
-      "eval_accuracy": 0.9814814814814815,
-      "eval_loss": 0.11288923025131226,
-      "eval_runtime": 0.2576,
-      "eval_samples_per_second": 209.62,
-      "eval_steps_per_second": 3.882,
+      "eval_accuracy": 0.9074074074074074,
+      "eval_loss": 0.33199751377105713,
+      "eval_runtime": 0.2627,
+      "eval_samples_per_second": 205.531,
+      "eval_steps_per_second": 3.806,
       "step": 20
     },
     {
       "epoch": 9.33,
       "learning_rate": 4.388888888888889e-05,
-      "loss": 0.5572,
+      "loss": 0.506,
       "step": 21
     },
     {
       "epoch": 9.78,
       "learning_rate": 4.3333333333333334e-05,
-      "loss": 0.1992,
+      "loss": 0.2242,
       "step": 22
     },
     {
       "epoch": 9.78,
       "eval_accuracy": 0.9629629629629629,
-      "eval_loss": 0.10700082033872604,
-      "eval_runtime": 0.4132,
-      "eval_samples_per_second": 130.679,
-      "eval_steps_per_second": 2.42,
+      "eval_loss": 0.11195674538612366,
+      "eval_runtime": 0.2715,
+      "eval_samples_per_second": 198.931,
+      "eval_steps_per_second": 3.684,
       "step": 22
     },
     {
       "epoch": 10.22,
       "learning_rate": 4.277777777777778e-05,
-      "loss": 0.167,
+      "loss": 0.1685,
       "step": 23
     },
     {
       "epoch": 10.67,
       "learning_rate": 4.222222222222222e-05,
-      "loss": 0.2173,
+      "loss": 0.2072,
       "step": 24
     },
     {
       "epoch": 10.67,
-      "eval_accuracy": 0.9259259259259259,
-      "eval_loss": 0.2469346821308136,
-      "eval_runtime": 0.2584,
-      "eval_samples_per_second": 208.943,
-      "eval_steps_per_second": 3.869,
+      "eval_accuracy": 0.9629629629629629,
+      "eval_loss": 0.07178916037082672,
+      "eval_runtime": 0.2631,
+      "eval_samples_per_second": 205.262,
+      "eval_steps_per_second": 3.801,
       "step": 24
     },
     {
       "epoch": 11.11,
       "learning_rate": 4.166666666666667e-05,
-      "loss": 0.1728,
+      "loss": 0.1653,
       "step": 25
     },
     {
       "epoch": 11.56,
       "learning_rate": 4.111111111111111e-05,
-      "loss": 0.171,
+      "loss": 0.1573,
       "step": 26
     },
     {
       "epoch": 12.0,
       "learning_rate": 4.055555555555556e-05,
-      "loss": 0.1489,
+      "loss": 0.1399,
       "step": 27
     },
     {
       "epoch": 12.0,
       "eval_accuracy": 0.9814814814814815,
-      "eval_loss": 0.08100217580795288,
-      "eval_runtime": 0.2569,
-      "eval_samples_per_second": 210.192,
-      "eval_steps_per_second": 3.892,
+      "eval_loss": 0.049447279423475266,
+      "eval_runtime": 0.2574,
+      "eval_samples_per_second": 209.772,
+      "eval_steps_per_second": 3.885,
       "step": 27
     },
     {
       "epoch": 12.44,
       "learning_rate": 4e-05,
-      "loss": 0.2299,
+      "loss": 0.1696,
       "step": 28
     },
     {
       "epoch": 12.89,
       "learning_rate": 3.944444444444445e-05,
-      "loss": 0.1975,
+      "loss": 0.1846,
       "step": 29
     },
     {
       "epoch": 12.89,
-      "eval_accuracy": 0.9814814814814815,
-      "eval_loss": 0.08110610395669937,
+      "eval_accuracy": 1.0,
+      "eval_loss": 0.03731463849544525,
       "eval_runtime": 0.2604,
-      "eval_samples_per_second": 207.381,
+      "eval_samples_per_second": 207.336,
       "eval_steps_per_second": 3.84,
       "step": 29
     },
     {
       "epoch": 13.33,
       "learning_rate": 3.888888888888889e-05,
-      "loss": 0.1396,
+      "loss": 0.1438,
       "step": 30
     },
     {
       "epoch": 13.78,
       "learning_rate": 3.8333333333333334e-05,
-      "loss": 0.1944,
+      "loss": 0.1816,
       "step": 31
     },
     {
       "epoch": 13.78,
-      "eval_accuracy": 0.9814814814814815,
-      "eval_loss": 0.0979403406381607,
-      "eval_runtime": 0.2574,
-      "eval_samples_per_second": 209.778,
-      "eval_steps_per_second": 3.885,
+      "eval_accuracy": 1.0,
+      "eval_loss": 0.03541439026594162,
+      "eval_runtime": 0.2612,
+      "eval_samples_per_second": 206.767,
+      "eval_steps_per_second": 3.829,
       "step": 31
     },
     {
       "epoch": 14.22,
       "learning_rate": 3.777777777777778e-05,
-      "loss": 0.1272,
+      "loss": 0.2004,
       "step": 32
     },
     {
       "epoch": 14.67,
       "learning_rate": 3.722222222222222e-05,
-      "loss": 0.1556,
+      "loss": 0.1453,
       "step": 33
     },
     {
       "epoch": 14.67,
       "eval_accuracy": 0.9814814814814815,
-      "eval_loss": 0.059882353991270065,
-      "eval_runtime": 0.2583,
-      "eval_samples_per_second": 209.053,
-      "eval_steps_per_second": 3.871,
+      "eval_loss": 0.04606747254729271,
+      "eval_runtime": 0.2706,
+      "eval_samples_per_second": 199.556,
+      "eval_steps_per_second": 3.695,
       "step": 33
     },
     {
       "epoch": 15.11,
       "learning_rate": 3.6666666666666666e-05,
-      "loss": 0.1173,
+      "loss": 0.1129,
       "step": 34
     },
     {
       "epoch": 15.56,
       "learning_rate": 3.611111111111111e-05,
-      "loss": 0.1832,
+      "loss": 0.1819,
       "step": 35
     },
     {
       "epoch": 16.0,
       "learning_rate": 3.555555555555556e-05,
-      "loss": 0.1631,
+      "loss": 0.1406,
       "step": 36
     },
     {
       "epoch": 16.0,
       "eval_accuracy": 1.0,
-      "eval_loss": 0.03582725301384926,
-      "eval_runtime": 0.2565,
-      "eval_samples_per_second": 210.555,
-      "eval_steps_per_second": 3.899,
+      "eval_loss": 0.03328908979892731,
+      "eval_runtime": 0.2597,
+      "eval_samples_per_second": 207.893,
+      "eval_steps_per_second": 3.85,
       "step": 36
     },
     {
       "epoch": 16.44,
       "learning_rate": 3.5e-05,
-      "loss": 0.1637,
+      "loss": 0.1389,
       "step": 37
     },
     {
       "epoch": 16.89,
       "learning_rate": 3.444444444444445e-05,
-      "loss": 0.1853,
+      "loss": 0.1749,
       "step": 38
     },
     {
       "epoch": 16.89,
-      "eval_accuracy": 0.9814814814814815,
-      "eval_loss": 0.08257294446229935,
-      "eval_runtime": 0.2586,
-      "eval_samples_per_second": 208.83,
-      "eval_steps_per_second": 3.867,
+      "eval_accuracy": 1.0,
+      "eval_loss": 0.02746324986219406,
+      "eval_runtime": 0.2621,
+      "eval_samples_per_second": 206.026,
+      "eval_steps_per_second": 3.815,
       "step": 38
     },
     {
       "epoch": 17.33,
       "learning_rate": 3.388888888888889e-05,
-      "loss": 0.1153,
+      "loss": 0.1079,
       "step": 39
     },
     {
       "epoch": 17.78,
       "learning_rate": 3.3333333333333335e-05,
-      "loss": 0.1452,
+      "loss": 0.1383,
       "step": 40
     },
     {
       "epoch": 17.78,
-      "eval_accuracy": 0.9814814814814815,
-      "eval_loss": 0.08874496817588806,
-      "eval_runtime": 0.2563,
-      "eval_samples_per_second": 210.693,
-      "eval_steps_per_second": 3.902,
+      "eval_accuracy": 1.0,
+      "eval_loss": 0.02032529003918171,
+      "eval_runtime": 0.258,
+      "eval_samples_per_second": 209.286,
+      "eval_steps_per_second": 3.876,
       "step": 40
     },
     {
       "epoch": 18.22,
       "learning_rate": 3.277777777777778e-05,
-      "loss": 0.1271,
+      "loss": 0.1454,
       "step": 41
     },
     {
       "epoch": 18.67,
       "learning_rate": 3.222222222222223e-05,
-      "loss": 0.1682,
+      "loss": 0.1659,
       "step": 42
     },
     {
       "epoch": 18.67,
-      "eval_accuracy": 0.9814814814814815,
-      "eval_loss": 0.06881634145975113,
-      "eval_runtime": 0.2602,
-      "eval_samples_per_second": 207.551,
-      "eval_steps_per_second": 3.844,
+      "eval_accuracy": 1.0,
+      "eval_loss": 0.018641581758856773,
+      "eval_runtime": 0.2631,
+      "eval_samples_per_second": 205.27,
+      "eval_steps_per_second": 3.801,
       "step": 42
     },
     {
       "epoch": 19.11,
       "learning_rate": 3.1666666666666666e-05,
-      "loss": 0.521,
+      "loss": 0.5264,
       "step": 43
     },
     {
       "epoch": 19.56,
       "learning_rate": 3.111111111111111e-05,
-      "loss": 0.1546,
+      "loss": 0.1586,
       "step": 44
     },
     {
       "epoch": 20.0,
       "learning_rate": 3.055555555555556e-05,
-      "loss": 0.1674,
+      "loss": 0.153,
       "step": 45
     },
     {
       "epoch": 20.0,
-      "eval_accuracy": 0.9814814814814815,
-      "eval_loss": 0.10646378248929977,
-      "eval_runtime": 0.2574,
-      "eval_samples_per_second": 209.82,
-      "eval_steps_per_second": 3.886,
+      "eval_accuracy": 1.0,
+      "eval_loss": 0.018401814624667168,
+      "eval_runtime": 0.2579,
+      "eval_samples_per_second": 209.351,
+      "eval_steps_per_second": 3.877,
       "step": 45
     },
     {
       "epoch": 20.44,
       "learning_rate": 3e-05,
-      "loss": 0.1386,
+      "loss": 0.1323,
       "step": 46
     },
     {
       "epoch": 20.89,
       "learning_rate": 2.9444444444444448e-05,
-      "loss": 0.1729,
+      "loss": 0.1497,
       "step": 47
     },
     {
       "epoch": 20.89,
-      "eval_accuracy": 0.9814814814814815,
-      "eval_loss": 0.11758946627378464,
-      "eval_runtime": 0.2574,
-      "eval_samples_per_second": 209.801,
-      "eval_steps_per_second": 3.885,
+      "eval_accuracy": 1.0,
+      "eval_loss": 0.021507540717720985,
+      "eval_runtime": 0.2732,
+      "eval_samples_per_second": 197.68,
+      "eval_steps_per_second": 3.661,
       "step": 47
     },
     {
       "epoch": 21.33,
       "learning_rate": 2.8888888888888888e-05,
-      "loss": 0.249,
+      "loss": 0.2375,
       "step": 48
     },
     {
       "epoch": 21.78,
       "learning_rate": 2.8333333333333335e-05,
-      "loss": 0.1234,
+      "loss": 0.1088,
       "step": 49
     },
     {
       "epoch": 21.78,
       "eval_accuracy": 0.9814814814814815,
-      "eval_loss": 0.05895670875906944,
-      "eval_runtime": 0.2583,
-      "eval_samples_per_second": 209.081,
-      "eval_steps_per_second": 3.872,
+      "eval_loss": 0.040791917592287064,
+      "eval_runtime": 0.326,
+      "eval_samples_per_second": 165.619,
+      "eval_steps_per_second": 3.067,
       "step": 49
     },
     {
       "epoch": 22.22,
       "learning_rate": 2.777777777777778e-05,
-      "loss": 0.0955,
+      "loss": 0.0887,
       "step": 50
     },
     {
       "epoch": 22.67,
       "learning_rate": 2.7222222222222223e-05,
-      "loss": 0.1696,
+      "loss": 0.1796,
       "step": 51
     },
     {
       "epoch": 22.67,
-      "eval_accuracy": 1.0,
-      "eval_loss": 0.04187057167291641,
-      "eval_runtime": 0.4982,
-      "eval_samples_per_second": 108.388,
-      "eval_steps_per_second": 2.007,
+      "eval_accuracy": 0.9814814814814815,
+      "eval_loss": 0.03766679763793945,
+      "eval_runtime": 0.2593,
+      "eval_samples_per_second": 208.223,
+      "eval_steps_per_second": 3.856,
       "step": 51
     },
     {
       "epoch": 23.11,
       "learning_rate": 2.6666666666666667e-05,
-      "loss": 0.1788,
+      "loss": 0.1459,
       "step": 52
     },
     {
       "epoch": 23.56,
       "learning_rate": 2.6111111111111114e-05,
-      "loss": 0.2168,
+      "loss": 0.156,
       "step": 53
     },
     {
       "epoch": 24.0,
       "learning_rate": 2.5555555555555554e-05,
-      "loss": 0.12,
+      "loss": 0.1041,
       "step": 54
     },
     {
       "epoch": 24.0,
       "eval_accuracy": 0.9814814814814815,
-      "eval_loss": 0.05020841583609581,
-      "eval_runtime": 0.2609,
-      "eval_samples_per_second": 206.982,
-      "eval_steps_per_second": 3.833,
+      "eval_loss": 0.06314324587583542,
+      "eval_runtime": 0.2592,
+      "eval_samples_per_second": 208.351,
+      "eval_steps_per_second": 3.858,
       "step": 54
     },
     {
       "epoch": 24.44,
       "learning_rate": 2.5e-05,
-      "loss": 0.1713,
+      "loss": 0.1796,
       "step": 55
     },
     {
       "epoch": 24.89,
       "learning_rate": 2.4444444444444445e-05,
-      "loss": 0.1301,
+      "loss": 0.1193,
       "step": 56
     },
     {
       "epoch": 24.89,
       "eval_accuracy": 0.9814814814814815,
-      "eval_loss": 0.12487592548131943,
-      "eval_runtime": 0.2789,
-      "eval_samples_per_second": 193.603,
-      "eval_steps_per_second": 3.585,
+      "eval_loss": 0.06367243826389313,
+      "eval_runtime": 0.2815,
+      "eval_samples_per_second": 191.858,
+      "eval_steps_per_second": 3.553,
       "step": 56
     },
     {
       "epoch": 25.33,
       "learning_rate": 2.3888888888888892e-05,
-      "loss": 0.1284,
+      "loss": 0.0848,
       "step": 57
     },
     {
       "epoch": 25.78,
       "learning_rate": 2.3333333333333336e-05,
-      "loss": 0.1324,
+      "loss": 0.1653,
       "step": 58
     },
     {
       "epoch": 25.78,
       "eval_accuracy": 0.9814814814814815,
-      "eval_loss": 0.1310151070356369,
-      "eval_runtime": 0.2578,
-      "eval_samples_per_second": 209.456,
-      "eval_steps_per_second": 3.879,
+      "eval_loss": 0.07295241206884384,
+      "eval_runtime": 0.2602,
+      "eval_samples_per_second": 207.497,
+      "eval_steps_per_second": 3.843,
       "step": 58
     },
     {
       "epoch": 26.22,
       "learning_rate": 2.277777777777778e-05,
-      "loss": 0.0992,
+      "loss": 0.0973,
       "step": 59
     },
     {
       "epoch": 26.67,
       "learning_rate": 2.2222222222222223e-05,
-      "loss": 0.1128,
+      "loss": 0.1296,
       "step": 60
     },
     {
       "epoch": 26.67,
       "eval_accuracy": 0.9814814814814815,
-      "eval_loss": 0.07883566617965698,
-      "eval_runtime": 0.2581,
-      "eval_samples_per_second": 209.202,
-      "eval_steps_per_second": 3.874,
+      "eval_loss": 0.07793273031711578,
+      "eval_runtime": 0.2603,
+      "eval_samples_per_second": 207.468,
+      "eval_steps_per_second": 3.842,
       "step": 60
     },
     {
       "epoch": 27.11,
       "learning_rate": 2.1666666666666667e-05,
-      "loss": 0.1235,
+      "loss": 0.1243,
       "step": 61
     },
     {
       "epoch": 27.56,
       "learning_rate": 2.111111111111111e-05,
-      "loss": 0.1672,
+      "loss": 0.1566,
       "step": 62
     },
     {
       "epoch": 28.0,
       "learning_rate": 2.0555555555555555e-05,
-      "loss": 0.1083,
+      "loss": 0.1036,
       "step": 63
     },
     {
       "epoch": 28.0,
-      "eval_accuracy": 1.0,
-      "eval_loss": 0.024069679901003838,
-      "eval_runtime": 0.2592,
-      "eval_samples_per_second": 208.37,
-      "eval_steps_per_second": 3.859,
+      "eval_accuracy": 0.9814814814814815,
+      "eval_loss": 0.031223006546497345,
+      "eval_runtime": 0.2604,
+      "eval_samples_per_second": 207.369,
+      "eval_steps_per_second": 3.84,
       "step": 63
     },
     {
       "epoch": 28.44,
       "learning_rate": 2e-05,
-      "loss": 0.1343,
+      "loss": 0.1376,
       "step": 64
     },
     {
       "epoch": 28.89,
       "learning_rate": 1.9444444444444445e-05,
-      "loss": 0.1551,
+      "loss": 0.1287,
       "step": 65
     },
     {
       "epoch": 28.89,
-      "eval_accuracy": 0.9814814814814815,
-      "eval_loss": 0.032383933663368225,
-      "eval_runtime": 0.2601,
-      "eval_samples_per_second": 207.588,
-      "eval_steps_per_second": 3.844,
+      "eval_accuracy": 1.0,
+      "eval_loss": 0.011618535034358501,
+      "eval_runtime": 0.2729,
+      "eval_samples_per_second": 197.869,
+      "eval_steps_per_second": 3.664,
       "step": 65
     },
     {
       "epoch": 29.33,
       "learning_rate": 1.888888888888889e-05,
-      "loss": 0.0953,
+      "loss": 0.0961,
       "step": 66
     },
     {
       "epoch": 29.78,
       "learning_rate": 1.8333333333333333e-05,
-      "loss": 0.1311,
+      "loss": 0.1307,
       "step": 67
     },
     {
       "epoch": 29.78,
-      "eval_accuracy": 0.9814814814814815,
-      "eval_loss": 0.06356043368577957,
-      "eval_runtime": 0.2566,
-      "eval_samples_per_second": 210.412,
-      "eval_steps_per_second": 3.897,
+      "eval_accuracy": 1.0,
+      "eval_loss": 0.012949406169354916,
+      "eval_runtime": 0.2593,
+      "eval_samples_per_second": 208.256,
+      "eval_steps_per_second": 3.857,
       "step": 67
     },
     {
       "epoch": 30.22,
       "learning_rate": 1.777777777777778e-05,
-      "loss": 0.0912,
+      "loss": 0.0873,
       "step": 68
     },
     {
       "epoch": 30.67,
       "learning_rate": 1.7222222222222224e-05,
-      "loss": 0.1367,
+      "loss": 0.1337,
       "step": 69
     },
     {
       "epoch": 30.67,
-      "eval_accuracy": 0.9814814814814815,
-      "eval_loss": 0.06610292941331863,
-      "eval_runtime": 0.2596,
-      "eval_samples_per_second": 207.984,
-      "eval_steps_per_second": 3.852,
+      "eval_accuracy": 1.0,
+      "eval_loss": 0.01411823183298111,
+      "eval_runtime": 0.2669,
+      "eval_samples_per_second": 202.286,
+      "eval_steps_per_second": 3.746,
       "step": 69
     },
     {
       "epoch": 31.11,
       "learning_rate": 1.6666666666666667e-05,
-      "loss": 0.1008,
+      "loss": 0.0961,
       "step": 70
     },
     {
       "epoch": 31.56,
       "learning_rate": 1.6111111111111115e-05,
-      "loss": 0.1489,
+      "loss": 0.1316,
       "step": 71
     },
     {
       "epoch": 32.0,
       "learning_rate": 1.5555555555555555e-05,
-      "loss": 0.1315,
+      "loss": 0.1274,
       "step": 72
     },
     {
       "epoch": 32.0,
       "eval_accuracy": 1.0,
-      "eval_loss": 0.015112031251192093,
-      "eval_runtime": 0.265,
-      "eval_samples_per_second": 203.796,
-      "eval_steps_per_second": 3.774,
+      "eval_loss": 0.016076432541012764,
+      "eval_runtime": 0.2582,
+      "eval_samples_per_second": 209.143,
+      "eval_steps_per_second": 3.873,
       "step": 72
     },
     {
       "epoch": 32.44,
       "learning_rate": 1.5e-05,
-      "loss": 0.1153,
+      "loss": 0.1192,
       "step": 73
     },
     {
       "epoch": 32.89,
       "learning_rate": 1.4444444444444444e-05,
-      "loss": 0.1717,
+      "loss": 0.1612,
       "step": 74
     },
     {
       "epoch": 32.89,
       "eval_accuracy": 1.0,
-      "eval_loss": 0.018243148922920227,
-      "eval_runtime": 0.258,
-      "eval_samples_per_second": 209.273,
-      "eval_steps_per_second": 3.875,
+      "eval_loss": 0.017708102241158485,
+      "eval_runtime": 0.2623,
+      "eval_samples_per_second": 205.838,
+      "eval_steps_per_second": 3.812,
       "step": 74
     },
     {
       "epoch": 33.33,
       "learning_rate": 1.388888888888889e-05,
-      "loss": 0.1959,
+      "loss": 0.2107,
       "step": 75
     },
     {
       "epoch": 33.78,
       "learning_rate": 1.3333333333333333e-05,
-      "loss": 0.1795,
+      "loss": 0.1504,
       "step": 76
     },
     {
       "epoch": 33.78,
       "eval_accuracy": 1.0,
-      "eval_loss": 0.019817128777503967,
-      "eval_runtime": 0.2624,
-      "eval_samples_per_second": 205.8,
-      "eval_steps_per_second": 3.811,
+      "eval_loss": 0.018124129623174667,
+      "eval_runtime": 0.2587,
+      "eval_samples_per_second": 208.707,
+      "eval_steps_per_second": 3.865,
       "step": 76
     },
     {
       "epoch": 34.22,
       "learning_rate": 1.2777777777777777e-05,
-      "loss": 0.1076,
+      "loss": 0.1108,
       "step": 77
     },
     {
       "epoch": 34.67,
       "learning_rate": 1.2222222222222222e-05,
-      "loss": 0.1348,
+      "loss": 0.1307,
       "step": 78
     },
     {
       "epoch": 34.67,
       "eval_accuracy": 1.0,
-      "eval_loss": 0.017830517143011093,
-      "eval_runtime": 0.2576,
-      "eval_samples_per_second": 209.607,
-      "eval_steps_per_second": 3.882,
+      "eval_loss": 0.017455147579312325,
+      "eval_runtime": 0.2609,
+      "eval_samples_per_second": 206.946,
+      "eval_steps_per_second": 3.832,
       "step": 78
     },
     {
       "epoch": 35.11,
       "learning_rate": 1.1666666666666668e-05,
-      "loss": 0.3572,
+      "loss": 0.348,
       "step": 79
     },
     {
       "epoch": 35.56,
       "learning_rate": 1.1111111111111112e-05,
-      "loss": 0.138,
+      "loss": 0.1275,
       "step": 80
     },
     {
       "epoch": 36.0,
       "learning_rate": 1.0555555555555555e-05,
-      "loss": 0.136,
+      "loss": 0.125,
       "step": 81
     },
     {
       "epoch": 36.0,
       "eval_accuracy": 1.0,
-      "eval_loss": 0.01816863939166069,
-      "eval_runtime": 0.2585,
-      "eval_samples_per_second": 208.9,
-      "eval_steps_per_second": 3.869,
+      "eval_loss": 0.01702389493584633,
+      "eval_runtime": 0.2603,
+      "eval_samples_per_second": 207.432,
+      "eval_steps_per_second": 3.841,
       "step": 81
     },
     {
       "epoch": 36.44,
       "learning_rate": 1e-05,
-      "loss": 0.1155,
+      "loss": 0.1085,
       "step": 82
     },
     {
       "epoch": 36.89,
       "learning_rate": 9.444444444444445e-06,
-      "loss": 0.133,
+      "loss": 0.1357,
       "step": 83
     },
     {
       "epoch": 36.89,
       "eval_accuracy": 1.0,
-      "eval_loss": 0.019667092710733414,
-      "eval_runtime": 0.2563,
-      "eval_samples_per_second": 210.712,
-      "eval_steps_per_second": 3.902,
+      "eval_loss": 0.016507649794220924,
+      "eval_runtime": 0.2662,
+      "eval_samples_per_second": 202.839,
+      "eval_steps_per_second": 3.756,
       "step": 83
     },
     {
       "epoch": 37.33,
       "learning_rate": 8.88888888888889e-06,
-      "loss": 0.0956,
+      "loss": 0.1008,
       "step": 84
     },
     {
       "epoch": 37.78,
       "learning_rate": 8.333333333333334e-06,
-      "loss": 0.1006,
+      "loss": 0.1033,
       "step": 85
     },
     {
       "epoch": 37.78,
       "eval_accuracy": 1.0,
-      "eval_loss": 0.02239961363375187,
-      "eval_runtime": 0.2608,
-      "eval_samples_per_second": 207.09,
-      "eval_steps_per_second": 3.835,
+      "eval_loss": 0.016153164207935333,
+      "eval_runtime": 0.4053,
+      "eval_samples_per_second": 133.244,
+      "eval_steps_per_second": 2.467,
       "step": 85
     },
     {
       "epoch": 38.22,
       "learning_rate": 7.777777777777777e-06,
-      "loss": 0.1384,
+      "loss": 0.1346,
       "step": 86
     },
     {
       "epoch": 38.67,
       "learning_rate": 7.222222222222222e-06,
-      "loss": 0.159,
+      "loss": 0.1749,
       "step": 87
     },
     {
       "epoch": 38.67,
       "eval_accuracy": 1.0,
-      "eval_loss": 0.025877976790070534,
-      "eval_runtime": 0.2569,
-      "eval_samples_per_second": 210.173,
-      "eval_steps_per_second": 3.892,
+      "eval_loss": 0.016394561156630516,
+      "eval_runtime": 0.2605,
+      "eval_samples_per_second": 207.331,
+      "eval_steps_per_second": 3.839,
       "step": 87
     },
     {
       "epoch": 39.11,
       "learning_rate": 6.666666666666667e-06,
-      "loss": 0.0773,
+      "loss": 0.0824,
       "step": 88
     },
     {
       "epoch": 39.56,
       "learning_rate": 6.111111111111111e-06,
-      "loss": 0.1196,
+      "loss": 0.1232,
       "step": 89
     },
     {
       "epoch": 40.0,
       "learning_rate": 5.555555555555556e-06,
-      "loss": 0.0899,
+      "loss": 0.0906,
       "step": 90
     },
     {
       "epoch": 40.0,
-      "eval_accuracy": 0.9814814814814815,
-      "eval_loss": 0.02953430823981762,
-      "eval_runtime": 0.2609,
-      "eval_samples_per_second": 207.015,
-      "eval_steps_per_second": 3.834,
+      "eval_accuracy": 1.0,
+      "eval_loss": 0.015265186317265034,
+      "eval_runtime": 0.2605,
+      "eval_samples_per_second": 207.293,
+      "eval_steps_per_second": 3.839,
       "step": 90
     },
     {
       "epoch": 40.44,
       "learning_rate": 5e-06,
-      "loss": 0.1458,
+      "loss": 0.1454,
       "step": 91
     },
     {
       "epoch": 40.89,
       "learning_rate": 4.444444444444445e-06,
-      "loss": 0.1464,
+      "loss": 0.1349,
       "step": 92
     },
     {
       "epoch": 40.89,
-      "eval_accuracy": 0.9814814814814815,
-      "eval_loss": 0.032179489731788635,
-      "eval_runtime": 0.2579,
-      "eval_samples_per_second": 209.385,
-      "eval_steps_per_second": 3.878,
+      "eval_accuracy": 1.0,
+      "eval_loss": 0.01515868864953518,
+      "eval_runtime": 0.26,
+      "eval_samples_per_second": 207.728,
+      "eval_steps_per_second": 3.847,
       "step": 92
     },
     {
       "epoch": 41.33,
       "learning_rate": 3.888888888888889e-06,
-      "loss": 0.107,
+      "loss": 0.1047,
       "step": 93
     },
     {
       "epoch": 41.78,
       "learning_rate": 3.3333333333333333e-06,
-      "loss": 0.1098,
+      "loss": 0.1056,
       "step": 94
     },
     {
       "epoch": 41.78,
-      "eval_accuracy": 0.9814814814814815,
-      "eval_loss": 0.0338004007935524,
-      "eval_runtime": 0.2591,
-      "eval_samples_per_second": 208.453,
-      "eval_steps_per_second": 3.86,
+      "eval_accuracy": 1.0,
+      "eval_loss": 0.015033537521958351,
+      "eval_runtime": 0.2588,
+      "eval_samples_per_second": 208.667,
+      "eval_steps_per_second": 3.864,
       "step": 94
     },
     {
       "epoch": 42.22,
       "learning_rate": 2.777777777777778e-06,
-      "loss": 0.1228,
+      "loss": 0.1196,
       "step": 95
     },
     {
       "epoch": 42.67,
       "learning_rate": 2.2222222222222225e-06,
-      "loss": 0.128,
+      "loss": 0.1328,
       "step": 96
     },
     {
       "epoch": 42.67,
-      "eval_accuracy": 0.9814814814814815,
-      "eval_loss": 0.033080052584409714,
-      "eval_runtime": 0.2582,
-      "eval_samples_per_second": 209.121,
-      "eval_steps_per_second": 3.873,
+      "eval_accuracy": 1.0,
+      "eval_loss": 0.014817849732935429,
+      "eval_runtime": 0.2611,
+      "eval_samples_per_second": 206.804,
+      "eval_steps_per_second": 3.83,
       "step": 96
     },
     {
       "epoch": 43.11,
       "learning_rate": 1.6666666666666667e-06,
-      "loss": 0.1231,
+      "loss": 0.1358,
       "step": 97
     },
     {
       "epoch": 43.56,
       "learning_rate": 1.1111111111111112e-06,
-      "loss": 0.1276,
+      "loss": 0.1221,
       "step": 98
     },
     {
       "epoch": 44.0,
       "learning_rate": 5.555555555555556e-07,
-      "loss": 0.0841,
+      "loss": 0.0742,
       "step": 99
     },
     {
       "epoch": 44.0,
-      "eval_accuracy": 0.9814814814814815,
-      "eval_loss": 0.030649887397885323,
-      "eval_runtime": 0.2594,
-      "eval_samples_per_second": 208.196,
-      "eval_steps_per_second": 3.855,
+      "eval_accuracy": 1.0,
+      "eval_loss": 0.014790916815400124,
+      "eval_runtime": 0.2592,
+      "eval_samples_per_second": 208.324,
+      "eval_steps_per_second": 3.858,
       "step": 99
     },
     {
       "epoch": 44.44,
       "learning_rate": 0.0,
-      "loss": 0.0864,
+      "loss": 0.0875,
       "step": 100
     },
     {
       "epoch": 44.44,
-      "eval_accuracy": 0.9814814814814815,
-      "eval_loss": 0.030413048341870308,
-      "eval_runtime": 0.2709,
-      "eval_samples_per_second": 199.324,
-      "eval_steps_per_second": 3.691,
+      "eval_accuracy": 1.0,
+      "eval_loss": 0.014792277477681637,
+      "eval_runtime": 0.2608,
+      "eval_samples_per_second": 207.072,
+      "eval_steps_per_second": 3.835,
       "step": 100
     },
     {
       "epoch": 44.44,
       "step": 100,
       "total_flos": 1.6586385457107272e+18,
-      "train_loss": 0.2113973332196474,
-      "train_runtime": 676.842,
-      "train_samples_per_second": 35.533,
-      "train_steps_per_second": 0.148
+      "train_loss": 0.20484884686768054,
+      "train_runtime": 700.3363,
+      "train_samples_per_second": 34.341,
+      "train_steps_per_second": 0.143
     }
   ],
   "max_steps": 100,