End of training

README.md

@@ -18,8 +18,8 @@ should probably proofread and complete it, then remove this comment. -->
 
 This model is a fine-tuned version of [microsoft/swinv2-tiny-patch4-window8-256](https://huggingface.co/microsoft/swinv2-tiny-patch4-window8-256) on an unknown dataset.
 It achieves the following results on the evaluation set:
-- Loss: 1.2461
-- Accuracy: 0.6923
+- Loss: 0.8631
+- Accuracy: 0.7308
 
 ## Model description
 

all_results.json

@@ -1,13 +1,13 @@
 {
     "epoch": 41.93023255813954,
     "eval_accuracy": 0.7307692307692307,
-    "eval_loss": 0.9577280879020691,
-    "eval_runtime": 1.0772,
-    "eval_samples_per_second": 48.273,
-    "eval_steps_per_second": 1.857,
+    "eval_loss": 0.8631380796432495,
+    "eval_runtime": 1.3442,
+    "eval_samples_per_second": 38.683,
+    "eval_steps_per_second": 1.488,
     "total_flos": 1.8345690247389512e+18,
-    "train_loss": 1.2270459805216107,
-    "train_runtime": 1356.3174,
-    "train_samples_per_second": 42.176,
-    "train_steps_per_second": 0.31
+    "train_loss": 1.1112867690268018,
+    "train_runtime": 1368.6695,
+    "train_samples_per_second": 41.795,
+    "train_steps_per_second": 0.307
 }

eval_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 41.93023255813954,
     "eval_accuracy": 0.7307692307692307,
-    "eval_loss": 0.9577280879020691,
-    "eval_runtime": 1.0772,
-    "eval_samples_per_second": 48.273,
-    "eval_steps_per_second": 1.857
+    "eval_loss": 0.8631380796432495,
+    "eval_runtime": 1.3442,
+    "eval_samples_per_second": 38.683,
+    "eval_steps_per_second": 1.488
 }

runs/Jan21_16-36-23_31423d62c322/events.out.tfevents.1737480255.31423d62c322.1929.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:df65b5a9535834e72ff108061e1e39a0e3c9f13e5595cf23ff0e4c1ad3093d93
+size 411

train_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 41.93023255813954,
     "total_flos": 1.8345690247389512e+18,
-    "train_loss": 1.2270459805216107,
-    "train_runtime": 1356.3174,
-    "train_samples_per_second": 42.176,
-    "train_steps_per_second": 0.31
+    "train_loss": 1.1112867690268018,
+    "train_runtime": 1368.6695,
+    "train_samples_per_second": 41.795,
+    "train_steps_per_second": 0.307
 }

trainer_state.json

@@ -1,6 +1,6 @@
 {
   "best_metric": 0.7307692307692307,
-  "best_model_checkpoint": "swinv2-tiny-patch4-window8-256-dmae-humeda-DAV21/checkpoint-90",
+  "best_model_checkpoint": "swinv2-tiny-patch4-window8-256-dmae-humeda-DAV21/checkpoint-130",
   "epoch": 41.93023255813954,
   "eval_steps": 500,
   "global_step": 420,
@@ -10,635 +10,635 @@
   "log_history": [
     {
       "epoch": 0.9302325581395349,
-      "eval_accuracy": 0.23076923076923078,
-      "eval_loss": 1.600624680519104,
-      "eval_runtime": 0.9466,
-      "eval_samples_per_second": 54.935,
-      "eval_steps_per_second": 2.113,
+      "eval_accuracy": 0.34615384615384615,
+      "eval_loss": 1.550955891609192,
+      "eval_runtime": 0.9413,
+      "eval_samples_per_second": 55.245,
+      "eval_steps_per_second": 2.125,
       "step": 10
     },
     {
       "epoch": 1.1860465116279069,
-      "grad_norm": 11.869792938232422,
+      "grad_norm": 13.047734260559082,
       "learning_rate": 1.4285714285714285e-05,
-      "loss": 6.7273,
+      "loss": 6.5929,
       "step": 12
     },
     {
       "epoch": 1.9302325581395348,
-      "eval_accuracy": 0.3076923076923077,
-      "eval_loss": 1.4678974151611328,
-      "eval_runtime": 0.9407,
-      "eval_samples_per_second": 55.281,
-      "eval_steps_per_second": 2.126,
+      "eval_accuracy": 0.2692307692307692,
+      "eval_loss": 1.4801901578903198,
+      "eval_runtime": 0.914,
+      "eval_samples_per_second": 56.894,
+      "eval_steps_per_second": 2.188,
       "step": 20
     },
     {
       "epoch": 2.3720930232558137,
-      "grad_norm": 21.593393325805664,
+      "grad_norm": 18.362613677978516,
       "learning_rate": 2.857142857142857e-05,
-      "loss": 5.9474,
+      "loss": 5.6252,
       "step": 24
     },
     {
       "epoch": 2.9302325581395348,
-      "eval_accuracy": 0.5384615384615384,
-      "eval_loss": 1.1914443969726562,
-      "eval_runtime": 0.9141,
-      "eval_samples_per_second": 56.886,
-      "eval_steps_per_second": 2.188,
+      "eval_accuracy": 0.40384615384615385,
+      "eval_loss": 1.1115041971206665,
+      "eval_runtime": 3.1175,
+      "eval_samples_per_second": 16.68,
+      "eval_steps_per_second": 0.642,
       "step": 30
     },
     {
       "epoch": 3.558139534883721,
-      "grad_norm": 30.464523315429688,
+      "grad_norm": 30.29971694946289,
       "learning_rate": 4.2857142857142856e-05,
-      "loss": 4.3002,
+      "loss": 3.874,
       "step": 36
     },
     {
       "epoch": 3.9302325581395348,
-      "eval_accuracy": 0.46153846153846156,
-      "eval_loss": 1.1951944828033447,
-      "eval_runtime": 1.2786,
-      "eval_samples_per_second": 40.671,
-      "eval_steps_per_second": 1.564,
+      "eval_accuracy": 0.5576923076923077,
+      "eval_loss": 0.9996148347854614,
+      "eval_runtime": 0.9335,
+      "eval_samples_per_second": 55.705,
+      "eval_steps_per_second": 2.143,
       "step": 40
     },
     {
       "epoch": 4.7441860465116275,
-      "grad_norm": 25.736072540283203,
+      "grad_norm": 29.079408645629883,
       "learning_rate": 4.9206349206349204e-05,
-      "loss": 2.9717,
+      "loss": 2.7168,
       "step": 48
     },
     {
       "epoch": 4.930232558139535,
-      "eval_accuracy": 0.5769230769230769,
-      "eval_loss": 0.9782841205596924,
-      "eval_runtime": 0.9391,
-      "eval_samples_per_second": 55.375,
-      "eval_steps_per_second": 2.13,
+      "eval_accuracy": 0.6538461538461539,
+      "eval_loss": 0.8436340689659119,
+      "eval_runtime": 1.286,
+      "eval_samples_per_second": 40.437,
+      "eval_steps_per_second": 1.555,
       "step": 50
     },
     {
       "epoch": 5.930232558139535,
-      "grad_norm": 54.5507698059082,
+      "grad_norm": 38.7504768371582,
       "learning_rate": 4.761904761904762e-05,
-      "loss": 2.4828,
+      "loss": 2.2435,
       "step": 60
     },
     {
       "epoch": 5.930232558139535,
-      "eval_accuracy": 0.5769230769230769,
-      "eval_loss": 0.9832176566123962,
-      "eval_runtime": 0.9619,
-      "eval_samples_per_second": 54.06,
-      "eval_steps_per_second": 2.079,
+      "eval_accuracy": 0.6153846153846154,
+      "eval_loss": 0.9319868087768555,
+      "eval_runtime": 0.9627,
+      "eval_samples_per_second": 54.015,
+      "eval_steps_per_second": 2.078,
       "step": 60
     },
     {
       "epoch": 6.930232558139535,
-      "eval_accuracy": 0.5,
-      "eval_loss": 1.0621165037155151,
-      "eval_runtime": 1.3284,
-      "eval_samples_per_second": 39.145,
-      "eval_steps_per_second": 1.506,
+      "eval_accuracy": 0.6346153846153846,
+      "eval_loss": 0.8411852121353149,
+      "eval_runtime": 0.9446,
+      "eval_samples_per_second": 55.048,
+      "eval_steps_per_second": 2.117,
       "step": 70
     },
     {
       "epoch": 7.186046511627907,
-      "grad_norm": 26.069013595581055,
+      "grad_norm": 27.031049728393555,
       "learning_rate": 4.603174603174603e-05,
-      "loss": 2.0276,
+      "loss": 1.9334,
       "step": 72
     },
     {
       "epoch": 7.930232558139535,
-      "eval_accuracy": 0.6538461538461539,
-      "eval_loss": 0.9040729999542236,
-      "eval_runtime": 0.9561,
-      "eval_samples_per_second": 54.39,
-      "eval_steps_per_second": 2.092,
+      "eval_accuracy": 0.6730769230769231,
+      "eval_loss": 0.8622324466705322,
+      "eval_runtime": 1.1865,
+      "eval_samples_per_second": 43.825,
+      "eval_steps_per_second": 1.686,
       "step": 80
     },
     {
       "epoch": 8.372093023255815,
-      "grad_norm": 43.163116455078125,
+      "grad_norm": 29.655231475830078,
       "learning_rate": 4.4444444444444447e-05,
-      "loss": 1.7041,
+      "loss": 1.6303,
       "step": 84
     },
     {
       "epoch": 8.930232558139535,
-      "eval_accuracy": 0.7307692307692307,
-      "eval_loss": 0.9577280879020691,
-      "eval_runtime": 0.9834,
-      "eval_samples_per_second": 52.878,
-      "eval_steps_per_second": 2.034,
+      "eval_accuracy": 0.7115384615384616,
+      "eval_loss": 0.9151535034179688,
+      "eval_runtime": 0.9237,
+      "eval_samples_per_second": 56.296,
+      "eval_steps_per_second": 2.165,
       "step": 90
     },
     {
       "epoch": 9.55813953488372,
-      "grad_norm": 45.423728942871094,
+      "grad_norm": 22.004980087280273,
       "learning_rate": 4.2857142857142856e-05,
-      "loss": 1.3967,
+      "loss": 1.2748,
       "step": 96
     },
     {
       "epoch": 9.930232558139535,
       "eval_accuracy": 0.6730769230769231,
-      "eval_loss": 1.0290472507476807,
-      "eval_runtime": 0.9566,
-      "eval_samples_per_second": 54.362,
-      "eval_steps_per_second": 2.091,
+      "eval_loss": 0.9720916748046875,
+      "eval_runtime": 0.9318,
+      "eval_samples_per_second": 55.806,
+      "eval_steps_per_second": 2.146,
       "step": 100
     },
     {
       "epoch": 10.744186046511627,
-      "grad_norm": 29.1428279876709,
+      "grad_norm": 30.746623992919922,
       "learning_rate": 4.126984126984127e-05,
-      "loss": 1.3079,
+      "loss": 1.0945,
       "step": 108
     },
     {
       "epoch": 10.930232558139535,
-      "eval_accuracy": 0.7115384615384616,
-      "eval_loss": 0.9863845109939575,
-      "eval_runtime": 1.3391,
-      "eval_samples_per_second": 38.832,
-      "eval_steps_per_second": 1.494,
+      "eval_accuracy": 0.6538461538461539,
+      "eval_loss": 1.0826632976531982,
+      "eval_runtime": 1.2786,
+      "eval_samples_per_second": 40.671,
+      "eval_steps_per_second": 1.564,
       "step": 110
     },
     {
       "epoch": 11.930232558139535,
-      "grad_norm": 25.176292419433594,
+      "grad_norm": 22.60085105895996,
       "learning_rate": 3.968253968253968e-05,
-      "loss": 1.0486,
+      "loss": 0.8395,
       "step": 120
     },
     {
       "epoch": 11.930232558139535,
-      "eval_accuracy": 0.6730769230769231,
-      "eval_loss": 1.0005784034729004,
-      "eval_runtime": 0.9808,
-      "eval_samples_per_second": 53.016,
-      "eval_steps_per_second": 2.039,
+      "eval_accuracy": 0.7115384615384616,
+      "eval_loss": 0.9152665734291077,
+      "eval_runtime": 0.9881,
+      "eval_samples_per_second": 52.629,
+      "eval_steps_per_second": 2.024,
       "step": 120
     },
     {
       "epoch": 12.930232558139535,
-      "eval_accuracy": 0.6153846153846154,
-      "eval_loss": 0.9503486156463623,
-      "eval_runtime": 0.9746,
-      "eval_samples_per_second": 53.353,
-      "eval_steps_per_second": 2.052,
+      "eval_accuracy": 0.7307692307692307,
+      "eval_loss": 0.8631380796432495,
+      "eval_runtime": 0.9359,
+      "eval_samples_per_second": 55.559,
+      "eval_steps_per_second": 2.137,
       "step": 130
     },
     {
       "epoch": 13.186046511627907,
-      "grad_norm": 55.16592025756836,
+      "grad_norm": 28.28727912902832,
       "learning_rate": 3.809523809523809e-05,
-      "loss": 1.0704,
+      "loss": 0.8587,
       "step": 132
     },
     {
       "epoch": 13.930232558139535,
-      "eval_accuracy": 0.7307692307692307,
-      "eval_loss": 0.9434700012207031,
-      "eval_runtime": 1.3209,
-      "eval_samples_per_second": 39.368,
-      "eval_steps_per_second": 1.514,
+      "eval_accuracy": 0.6538461538461539,
+      "eval_loss": 1.1038739681243896,
+      "eval_runtime": 1.1762,
+      "eval_samples_per_second": 44.21,
+      "eval_steps_per_second": 1.7,
       "step": 140
     },
     {
       "epoch": 14.372093023255815,
-      "grad_norm": 31.888456344604492,
+      "grad_norm": 30.497623443603516,
       "learning_rate": 3.650793650793651e-05,
-      "loss": 0.9929,
+      "loss": 0.8574,
       "step": 144
     },
     {
       "epoch": 14.930232558139535,
-      "eval_accuracy": 0.6346153846153846,
-      "eval_loss": 1.0601013898849487,
-      "eval_runtime": 0.9793,
-      "eval_samples_per_second": 53.098,
-      "eval_steps_per_second": 2.042,
+      "eval_accuracy": 0.6923076923076923,
+      "eval_loss": 1.0462819337844849,
+      "eval_runtime": 0.9493,
+      "eval_samples_per_second": 54.776,
+      "eval_steps_per_second": 2.107,
       "step": 150
     },
     {
       "epoch": 15.55813953488372,
-      "grad_norm": 29.11701202392578,
+      "grad_norm": 19.123886108398438,
       "learning_rate": 3.492063492063492e-05,
-      "loss": 0.8911,
+      "loss": 0.7096,
       "step": 156
     },
     {
       "epoch": 15.930232558139535,
-      "eval_accuracy": 0.6730769230769231,
-      "eval_loss": 1.0281590223312378,
-      "eval_runtime": 0.9645,
-      "eval_samples_per_second": 53.912,
-      "eval_steps_per_second": 2.074,
+      "eval_accuracy": 0.7115384615384616,
+      "eval_loss": 0.9990596175193787,
+      "eval_runtime": 0.9614,
+      "eval_samples_per_second": 54.09,
+      "eval_steps_per_second": 2.08,
       "step": 160
     },
     {
       "epoch": 16.74418604651163,
-      "grad_norm": 32.15597152709961,
+      "grad_norm": 34.98741912841797,
       "learning_rate": 3.3333333333333335e-05,
-      "loss": 0.7702,
+      "loss": 0.6606,
       "step": 168
     },
     {
       "epoch": 16.930232558139537,
-      "eval_accuracy": 0.6923076923076923,
-      "eval_loss": 1.1857846975326538,
-      "eval_runtime": 0.9593,
-      "eval_samples_per_second": 54.207,
-      "eval_steps_per_second": 2.085,
+      "eval_accuracy": 0.6730769230769231,
+      "eval_loss": 1.051858901977539,
+      "eval_runtime": 1.0499,
+      "eval_samples_per_second": 49.531,
+      "eval_steps_per_second": 1.905,
       "step": 170
     },
     {
       "epoch": 17.930232558139537,
-      "grad_norm": 32.97303771972656,
+      "grad_norm": 29.33749771118164,
       "learning_rate": 3.1746031746031745e-05,
-      "loss": 0.718,
+      "loss": 0.5513,
       "step": 180
     },
     {
       "epoch": 17.930232558139537,
-      "eval_accuracy": 0.6538461538461539,
-      "eval_loss": 1.2644145488739014,
-      "eval_runtime": 1.3041,
-      "eval_samples_per_second": 39.875,
-      "eval_steps_per_second": 1.534,
+      "eval_accuracy": 0.7115384615384616,
+      "eval_loss": 1.0864715576171875,
+      "eval_runtime": 0.9297,
+      "eval_samples_per_second": 55.932,
+      "eval_steps_per_second": 2.151,
       "step": 180
     },
     {
       "epoch": 18.930232558139537,
-      "eval_accuracy": 0.6346153846153846,
-      "eval_loss": 1.1403751373291016,
-      "eval_runtime": 0.957,
-      "eval_samples_per_second": 54.339,
-      "eval_steps_per_second": 2.09,
+      "eval_accuracy": 0.6730769230769231,
+      "eval_loss": 1.1139813661575317,
+      "eval_runtime": 0.9356,
+      "eval_samples_per_second": 55.582,
+      "eval_steps_per_second": 2.138,
       "step": 190
     },
     {
       "epoch": 19.186046511627907,
-      "grad_norm": 18.562591552734375,
+      "grad_norm": 31.023462295532227,
       "learning_rate": 3.0158730158730158e-05,
-      "loss": 0.6526,
+      "loss": 0.61,
       "step": 192
     },
     {
       "epoch": 19.930232558139537,
-      "eval_accuracy": 0.6923076923076923,
-      "eval_loss": 1.2202544212341309,
-      "eval_runtime": 0.981,
-      "eval_samples_per_second": 53.008,
-      "eval_steps_per_second": 2.039,
+      "eval_accuracy": 0.6730769230769231,
+      "eval_loss": 1.0289796590805054,
+      "eval_runtime": 0.9508,
+      "eval_samples_per_second": 54.693,
+      "eval_steps_per_second": 2.104,
       "step": 200
     },
     {
       "epoch": 20.372093023255815,
-      "grad_norm": 41.22922897338867,
+      "grad_norm": 31.7921142578125,
       "learning_rate": 2.857142857142857e-05,
-      "loss": 0.6617,
+      "loss": 0.5278,
       "step": 204
     },
     {
       "epoch": 20.930232558139537,
-      "eval_accuracy": 0.6730769230769231,
-      "eval_loss": 1.3465415239334106,
-      "eval_runtime": 1.3212,
-      "eval_samples_per_second": 39.359,
-      "eval_steps_per_second": 1.514,
+      "eval_accuracy": 0.6923076923076923,
+      "eval_loss": 1.1002823114395142,
+      "eval_runtime": 0.9325,
+      "eval_samples_per_second": 55.763,
+      "eval_steps_per_second": 2.145,
       "step": 210
     },
     {
       "epoch": 21.558139534883722,
-      "grad_norm": 28.829620361328125,
+      "grad_norm": 33.69935989379883,
       "learning_rate": 2.6984126984126984e-05,
-      "loss": 0.5699,
+      "loss": 0.4639,
       "step": 216
     },
     {
       "epoch": 21.930232558139537,
       "eval_accuracy": 0.6538461538461539,
-      "eval_loss": 1.3862558603286743,
-      "eval_runtime": 0.9516,
-      "eval_samples_per_second": 54.644,
-      "eval_steps_per_second": 2.102,
+      "eval_loss": 1.2471978664398193,
+      "eval_runtime": 0.9316,
+      "eval_samples_per_second": 55.818,
+      "eval_steps_per_second": 2.147,
       "step": 220
     },
     {
       "epoch": 22.74418604651163,
-      "grad_norm": 23.03827476501465,
+      "grad_norm": 20.980804443359375,
       "learning_rate": 2.5396825396825397e-05,
-      "loss": 0.5228,
+      "loss": 0.4719,
       "step": 228
     },
     {
       "epoch": 22.930232558139537,
-      "eval_accuracy": 0.7115384615384616,
-      "eval_loss": 1.272707223892212,
-      "eval_runtime": 0.985,
-      "eval_samples_per_second": 52.794,
-      "eval_steps_per_second": 2.031,
+      "eval_accuracy": 0.6923076923076923,
+      "eval_loss": 1.1545660495758057,
+      "eval_runtime": 0.9225,
+      "eval_samples_per_second": 56.37,
+      "eval_steps_per_second": 2.168,
       "step": 230
     },
     {
       "epoch": 23.930232558139537,
-      "grad_norm": 27.27239990234375,
+      "grad_norm": 26.727859497070312,
       "learning_rate": 2.380952380952381e-05,
-      "loss": 0.4321,
+      "loss": 0.4212,
       "step": 240
     },
     {
       "epoch": 23.930232558139537,
-      "eval_accuracy": 0.7115384615384616,
-      "eval_loss": 1.194717288017273,
-      "eval_runtime": 0.9431,
-      "eval_samples_per_second": 55.135,
-      "eval_steps_per_second": 2.121,
+      "eval_accuracy": 0.7307692307692307,
+      "eval_loss": 1.1084016561508179,
+      "eval_runtime": 0.9361,
+      "eval_samples_per_second": 55.552,
+      "eval_steps_per_second": 2.137,
       "step": 240
     },
     {
       "epoch": 24.930232558139537,
-      "eval_accuracy": 0.6346153846153846,
-      "eval_loss": 1.397331714630127,
-      "eval_runtime": 1.2955,
-      "eval_samples_per_second": 40.139,
-      "eval_steps_per_second": 1.544,
+      "eval_accuracy": 0.6730769230769231,
+      "eval_loss": 1.2952685356140137,
+      "eval_runtime": 1.0838,
+      "eval_samples_per_second": 47.981,
+      "eval_steps_per_second": 1.845,
       "step": 250
     },
     {
       "epoch": 25.186046511627907,
-      "grad_norm": 24.13581085205078,
+      "grad_norm": 16.363554000854492,
       "learning_rate": 2.2222222222222223e-05,
-      "loss": 0.557,
+      "loss": 0.4109,
       "step": 252
     },
     {
       "epoch": 25.930232558139537,
-      "eval_accuracy": 0.6538461538461539,
-      "eval_loss": 1.4321506023406982,
-      "eval_runtime": 1.3216,
-      "eval_samples_per_second": 39.345,
-      "eval_steps_per_second": 1.513,
+      "eval_accuracy": 0.7307692307692307,
+      "eval_loss": 1.1867846250534058,
+      "eval_runtime": 0.9096,
+      "eval_samples_per_second": 57.167,
+      "eval_steps_per_second": 2.199,
       "step": 260
     },
     {
       "epoch": 26.372093023255815,
-      "grad_norm": 34.4373664855957,
+      "grad_norm": 42.207515716552734,
       "learning_rate": 2.0634920634920636e-05,
-      "loss": 0.4569,
+      "loss": 0.4236,
       "step": 264
     },
     {
       "epoch": 26.930232558139537,
-      "eval_accuracy": 0.6538461538461539,
-      "eval_loss": 1.3255730867385864,
-      "eval_runtime": 0.9379,
-      "eval_samples_per_second": 55.443,
-      "eval_steps_per_second": 2.132,
+      "eval_accuracy": 0.6346153846153846,
+      "eval_loss": 1.2559605836868286,
+      "eval_runtime": 0.9101,
+      "eval_samples_per_second": 57.135,
+      "eval_steps_per_second": 2.197,
       "step": 270
     },
     {
       "epoch": 27.558139534883722,
-      "grad_norm": 19.65439224243164,
+      "grad_norm": 15.23690414428711,
       "learning_rate": 1.9047619047619046e-05,
-      "loss": 0.4395,
+      "loss": 0.3638,
       "step": 276
     },
     {
       "epoch": 27.930232558139537,
-      "eval_accuracy": 0.6730769230769231,
-      "eval_loss": 1.3907530307769775,
-      "eval_runtime": 1.2894,
-      "eval_samples_per_second": 40.33,
-      "eval_steps_per_second": 1.551,
+      "eval_accuracy": 0.7115384615384616,
+      "eval_loss": 1.216145634651184,
+      "eval_runtime": 1.2163,
+      "eval_samples_per_second": 42.753,
+      "eval_steps_per_second": 1.644,
       "step": 280
     },
     {
       "epoch": 28.74418604651163,
-      "grad_norm": 31.272125244140625,
+      "grad_norm": 33.352203369140625,
       "learning_rate": 1.746031746031746e-05,
-      "loss": 0.5587,
+      "loss": 0.3944,
       "step": 288
     },
     {
       "epoch": 28.930232558139537,
-      "eval_accuracy": 0.6923076923076923,
-      "eval_loss": 1.5055606365203857,
-      "eval_runtime": 0.9347,
-      "eval_samples_per_second": 55.632,
-      "eval_steps_per_second": 2.14,
+      "eval_accuracy": 0.7307692307692307,
+      "eval_loss": 1.1581984758377075,
+      "eval_runtime": 0.9294,
+      "eval_samples_per_second": 55.948,
+      "eval_steps_per_second": 2.152,
       "step": 290
     },
     {
       "epoch": 29.930232558139537,
-      "grad_norm": 21.733734130859375,
+      "grad_norm": 19.420169830322266,
       "learning_rate": 1.5873015873015872e-05,
-      "loss": 0.439,
+      "loss": 0.3621,
       "step": 300
     },
     {
       "epoch": 29.930232558139537,
-      "eval_accuracy": 0.6538461538461539,
-      "eval_loss": 1.3810527324676514,
-      "eval_runtime": 0.9297,
-      "eval_samples_per_second": 55.934,
-      "eval_steps_per_second": 2.151,
+      "eval_accuracy": 0.6923076923076923,
+      "eval_loss": 1.2993353605270386,
+      "eval_runtime": 0.912,
+      "eval_samples_per_second": 57.019,
+      "eval_steps_per_second": 2.193,
       "step": 300
     },
     {
       "epoch": 30.930232558139537,
-      "eval_accuracy": 0.6153846153846154,
-      "eval_loss": 1.3693809509277344,
-      "eval_runtime": 1.2587,
-      "eval_samples_per_second": 41.311,
-      "eval_steps_per_second": 1.589,
+      "eval_accuracy": 0.7115384615384616,
+      "eval_loss": 1.1401245594024658,
+      "eval_runtime": 1.2431,
+      "eval_samples_per_second": 41.83,
+      "eval_steps_per_second": 1.609,
       "step": 310
     },
     {
       "epoch": 31.186046511627907,
-      "grad_norm": 24.466344833374023,
+      "grad_norm": 24.769433975219727,
       "learning_rate": 1.4285714285714285e-05,
-      "loss": 0.3989,
+      "loss": 0.3203,
       "step": 312
     },
     {
       "epoch": 31.930232558139537,
-      "eval_accuracy": 0.6538461538461539,
-      "eval_loss": 1.4759361743927002,
-      "eval_runtime": 0.9328,
-      "eval_samples_per_second": 55.746,
-      "eval_steps_per_second": 2.144,
+      "eval_accuracy": 0.7115384615384616,
+      "eval_loss": 1.3228098154067993,
+      "eval_runtime": 0.9358,
+      "eval_samples_per_second": 55.568,
+      "eval_steps_per_second": 2.137,
       "step": 320
     },
     {
       "epoch": 32.372093023255815,
-      "grad_norm": 35.87159729003906,
+      "grad_norm": 25.592016220092773,
       "learning_rate": 1.2698412698412699e-05,
-      "loss": 0.312,
+      "loss": 0.3014,
       "step": 324
     },
     {
       "epoch": 32.93023255813954,
-      "eval_accuracy": 0.6346153846153846,
-      "eval_loss": 1.5513783693313599,
-      "eval_runtime": 0.9364,
-      "eval_samples_per_second": 55.53,
-      "eval_steps_per_second": 2.136,
+      "eval_accuracy": 0.6923076923076923,
+      "eval_loss": 1.2812803983688354,
+      "eval_runtime": 1.1414,
+      "eval_samples_per_second": 45.557,
+      "eval_steps_per_second": 1.752,
       "step": 330
     },
     {
       "epoch": 33.55813953488372,
-      "grad_norm": 14.943767547607422,
+      "grad_norm": 17.85179328918457,
       "learning_rate": 1.1111111111111112e-05,
-      "loss": 0.3728,
+      "loss": 0.3464,
       "step": 336
     },
     {
       "epoch": 33.93023255813954,
       "eval_accuracy": 0.6538461538461539,
-      "eval_loss": 1.5128782987594604,
-      "eval_runtime": 1.2844,
-      "eval_samples_per_second": 40.485,
-      "eval_steps_per_second": 1.557,
+      "eval_loss": 1.4768296480178833,
+      "eval_runtime": 0.9065,
+      "eval_samples_per_second": 57.364,
+      "eval_steps_per_second": 2.206,
       "step": 340
     },
     {
       "epoch": 34.74418604651163,
-      "grad_norm": 15.3562593460083,
+      "grad_norm": 15.770562171936035,
       "learning_rate": 9.523809523809523e-06,
-      "loss": 0.4031,
+      "loss": 0.2891,
       "step": 348
     },
     {
       "epoch": 34.93023255813954,
-      "eval_accuracy": 0.6538461538461539,
-      "eval_loss": 1.4914475679397583,
-      "eval_runtime": 0.948,
-      "eval_samples_per_second": 54.852,
-      "eval_steps_per_second": 2.11,
+      "eval_accuracy": 0.7307692307692307,
+      "eval_loss": 1.2304089069366455,
+      "eval_runtime": 0.9275,
+      "eval_samples_per_second": 56.067,
+      "eval_steps_per_second": 2.156,
       "step": 350
     },
     {
       "epoch": 35.93023255813954,
-      "grad_norm": 23.48607635498047,
+      "grad_norm": 22.76992416381836,
       "learning_rate": 7.936507936507936e-06,
-      "loss": 0.3523,
+      "loss": 0.3153,
       "step": 360
     },
     {
       "epoch": 35.93023255813954,
-      "eval_accuracy": 0.6730769230769231,
-      "eval_loss": 1.5588839054107666,
-      "eval_runtime": 0.9436,
-      "eval_samples_per_second": 55.108,
-      "eval_steps_per_second": 2.12,
+      "eval_accuracy": 0.6923076923076923,
+      "eval_loss": 1.3096110820770264,
+      "eval_runtime": 1.2083,
+      "eval_samples_per_second": 43.037,
+      "eval_steps_per_second": 1.655,
       "step": 360
     },
     {
       "epoch": 36.93023255813954,
-      "eval_accuracy": 0.6730769230769231,
-      "eval_loss": 1.568899154663086,
-      "eval_runtime": 1.2885,
-      "eval_samples_per_second": 40.359,
-      "eval_steps_per_second": 1.552,
+      "eval_accuracy": 0.7115384615384616,
+      "eval_loss": 1.3564749956130981,
+      "eval_runtime": 0.9106,
+      "eval_samples_per_second": 57.103,
+      "eval_steps_per_second": 2.196,
       "step": 370
     },
     {
       "epoch": 37.18604651162791,
-      "grad_norm": 31.557363510131836,
+      "grad_norm": 32.75940704345703,
       "learning_rate": 6.349206349206349e-06,
-      "loss": 0.265,
+      "loss": 0.2762,
       "step": 372
     },
     {
       "epoch": 37.93023255813954,
-      "eval_accuracy": 0.6730769230769231,
-      "eval_loss": 1.6178526878356934,
-      "eval_runtime": 0.9634,
-      "eval_samples_per_second": 53.974,
-      "eval_steps_per_second": 2.076,
+      "eval_accuracy": 0.6923076923076923,
+      "eval_loss": 1.2931231260299683,
+      "eval_runtime": 0.9165,
+      "eval_samples_per_second": 56.736,
+      "eval_steps_per_second": 2.182,
       "step": 380
     },
     {
       "epoch": 38.372093023255815,
-      "grad_norm": 27.208581924438477,
+      "grad_norm": 28.327919006347656,
       "learning_rate": 4.7619047619047615e-06,
-      "loss": 0.3189,
+      "loss": 0.3191,
       "step": 384
     },
     {
       "epoch": 38.93023255813954,
-      "eval_accuracy": 0.6538461538461539,
-      "eval_loss": 1.5962971448898315,
-      "eval_runtime": 0.9485,
-      "eval_samples_per_second": 54.826,
-      "eval_steps_per_second": 2.109,
+      "eval_accuracy": 0.7307692307692307,
+      "eval_loss": 1.2440568208694458,
+      "eval_runtime": 1.2601,
+      "eval_samples_per_second": 41.266,
+      "eval_steps_per_second": 1.587,
       "step": 390
     },
     {
       "epoch": 39.55813953488372,
-      "grad_norm": 30.211496353149414,
+      "grad_norm": 17.981231689453125,
       "learning_rate": 3.1746031746031746e-06,
-      "loss": 0.2976,
+      "loss": 0.3009,
       "step": 396
     },
     {
       "epoch": 39.93023255813954,
-      "eval_accuracy": 0.6346153846153846,
-      "eval_loss": 1.5738095045089722,
-      "eval_runtime": 1.1624,
-      "eval_samples_per_second": 44.733,
-      "eval_steps_per_second": 1.721,
+      "eval_accuracy": 0.7307692307692307,
+      "eval_loss": 1.2109795808792114,
+      "eval_runtime": 0.9344,
+      "eval_samples_per_second": 55.649,
+      "eval_steps_per_second": 2.14,
       "step": 400
     },
     {
       "epoch": 40.74418604651163,
-      "grad_norm": 16.808286666870117,
+      "grad_norm": 14.792089462280273,
       "learning_rate": 1.5873015873015873e-06,
-      "loss": 0.3097,
+      "loss": 0.2645,
       "step": 408
     },
     {
       "epoch": 40.93023255813954,
-      "eval_accuracy": 0.6346153846153846,
-      "eval_loss": 1.578661561012268,
-      "eval_runtime": 0.9279,
-      "eval_samples_per_second": 56.041,
-      "eval_steps_per_second": 2.155,
+      "eval_accuracy": 0.7115384615384616,
+      "eval_loss": 1.2433098554611206,
+      "eval_runtime": 0.9244,
+      "eval_samples_per_second": 56.254,
+      "eval_steps_per_second": 2.164,
       "step": 410
     },
     {
       "epoch": 41.93023255813954,
-      "grad_norm": 11.15967082977295,
+      "grad_norm": 13.357348442077637,
       "learning_rate": 0.0,
-      "loss": 0.269,
+      "loss": 0.2497,
       "step": 420
     },
     {
       "epoch": 41.93023255813954,
-      "eval_accuracy": 0.6346153846153846,
-      "eval_loss": 1.5818698406219482,
-      "eval_runtime": 1.6499,
-      "eval_samples_per_second": 31.516,
-      "eval_steps_per_second": 1.212,
+      "eval_accuracy": 0.6923076923076923,
+      "eval_loss": 1.2460753917694092,
+      "eval_runtime": 1.8198,
+      "eval_samples_per_second": 28.575,
+      "eval_steps_per_second": 1.099,
       "step": 420
     },
     {
       "epoch": 41.93023255813954,
       "step": 420,
       "total_flos": 1.8345690247389512e+18,
-      "train_loss": 1.2270459805216107,
-      "train_runtime": 1356.3174,
-      "train_samples_per_second": 42.176,
-      "train_steps_per_second": 0.31
+      "train_loss": 1.1112867690268018,
+      "train_runtime": 1368.6695,
+      "train_samples_per_second": 41.795,
+      "train_steps_per_second": 0.307
     }
   ],
   "logging_steps": 12,