🍻 cheers

README.md

@@ -2,6 +2,7 @@
 license: apache-2.0
 base_model: microsoft/swin-tiny-patch4-window7-224
 tags:
+- image-classification
 - generated_from_trainer
 metrics:
 - accuracy
@@ -15,9 +16,9 @@ should probably proofread and complete it, then remove this comment. -->
 
 # swin-tiny-patch4-window7-224-finetuned_ASL_Isolated_Swin_dataset2
 
-This model is a fine-tuned version of [microsoft/swin-tiny-patch4-window7-224](https://huggingface.co/microsoft/swin-tiny-patch4-window7-224) on an unknown dataset.
+This model is a fine-tuned version of [microsoft/swin-tiny-patch4-window7-224](https://huggingface.co/microsoft/swin-tiny-patch4-window7-224) on the ASL_Isolated_Swin_dataset dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.0834
+- Loss: 0.0784
 - Accuracy: 0.9846
 
 ## Model description

all_results.json

@@ -1,13 +1,13 @@
 {
-    "epoch": 10.0,
-    "eval_accuracy": 0.9769230769230769,
-    "eval_loss": 0.13189926743507385,
-    "eval_runtime": 1.9108,
-    "eval_samples_per_second": 136.069,
-    "eval_steps_per_second": 17.27,
-    "total_flos": 3.651306173035315e+17,
-    "train_loss": 0.7250858170830685,
-    "train_runtime": 208.4502,
-    "train_samples_per_second": 70.424,
-    "train_steps_per_second": 4.414
+    "epoch": 20.0,
+    "eval_accuracy": 0.9846153846153847,
+    "eval_loss": 0.07838542014360428,
+    "eval_runtime": 1.8983,
+    "eval_samples_per_second": 136.961,
+    "eval_steps_per_second": 17.384,
+    "total_flos": 7.30261234607063e+17,
+    "train_loss": 0.5343924296938855,
+    "train_runtime": 403.7255,
+    "train_samples_per_second": 72.723,
+    "train_steps_per_second": 4.558
 }

eval_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 10.0,
-    "eval_accuracy": 0.9769230769230769,
-    "eval_loss": 0.13189926743507385,
-    "eval_runtime": 1.9108,
-    "eval_samples_per_second": 136.069,
-    "eval_steps_per_second": 17.27
+    "epoch": 20.0,
+    "eval_accuracy": 0.9846153846153847,
+    "eval_loss": 0.07838542014360428,
+    "eval_runtime": 1.8983,
+    "eval_samples_per_second": 136.961,
+    "eval_steps_per_second": 17.384
 }

train_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 10.0,
-    "total_flos": 3.651306173035315e+17,
-    "train_loss": 0.7250858170830685,
-    "train_runtime": 208.4502,
-    "train_samples_per_second": 70.424,
-    "train_steps_per_second": 4.414
+    "epoch": 20.0,
+    "total_flos": 7.30261234607063e+17,
+    "train_loss": 0.5343924296938855,
+    "train_runtime": 403.7255,
+    "train_samples_per_second": 72.723,
+    "train_steps_per_second": 4.558
 }

trainer_state.json

@@ -1,661 +1,1294 @@
 {
-  "best_metric": 0.9769230769230769,
-  "best_model_checkpoint": "swin-tiny-patch4-window7-224-finetuned_ASL_Isolated_Swin_dataset2/checkpoint-700",
-  "epoch": 10.0,
+  "best_metric": 0.9846153846153847,
+  "best_model_checkpoint": "swin-tiny-patch4-window7-224-finetuned_ASL_Isolated_Swin_dataset2/checkpoint-1600",
+  "epoch": 20.0,
   "eval_steps": 100,
-  "global_step": 920,
+  "global_step": 1840,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
       "epoch": 0.11,
-      "learning_rate": 0.00019804347826086956,
-      "loss": 3.3183,
+      "learning_rate": 0.00019902173913043477,
+      "loss": 3.3338,
       "step": 10
     },
     {
       "epoch": 0.22,
-      "learning_rate": 0.0001958695652173913,
-      "loss": 3.1344,
+      "learning_rate": 0.00019804347826086956,
+      "loss": 3.1809,
       "step": 20
     },
     {
       "epoch": 0.33,
-      "learning_rate": 0.00019369565217391307,
-      "loss": 3.1166,
+      "learning_rate": 0.00019695652173913044,
+      "loss": 3.0792,
       "step": 30
     },
     {
       "epoch": 0.43,
-      "learning_rate": 0.00019152173913043478,
-      "loss": 2.8413,
+      "learning_rate": 0.0001958695652173913,
+      "loss": 2.7788,
       "step": 40
     },
     {
       "epoch": 0.54,
-      "learning_rate": 0.00018934782608695655,
-      "loss": 2.4092,
+      "learning_rate": 0.00019478260869565218,
+      "loss": 2.4649,
       "step": 50
     },
     {
       "epoch": 0.65,
-      "learning_rate": 0.00018717391304347826,
-      "loss": 2.173,
+      "learning_rate": 0.00019369565217391307,
+      "loss": 2.2866,
       "step": 60
     },
     {
       "epoch": 0.76,
-      "learning_rate": 0.00018500000000000002,
-      "loss": 1.9006,
+      "learning_rate": 0.00019260869565217392,
+      "loss": 2.1811,
       "step": 70
     },
     {
       "epoch": 0.87,
-      "learning_rate": 0.00018282608695652174,
-      "loss": 1.8329,
+      "learning_rate": 0.00019152173913043478,
+      "loss": 1.8357,
       "step": 80
     },
     {
       "epoch": 0.98,
-      "learning_rate": 0.00018086956521739132,
-      "loss": 1.7864,
+      "learning_rate": 0.00019043478260869566,
+      "loss": 1.8457,
       "step": 90
     },
     {
       "epoch": 1.09,
-      "learning_rate": 0.00017869565217391305,
-      "loss": 1.6751,
+      "learning_rate": 0.00018945652173913045,
+      "loss": 1.7296,
       "step": 100
     },
     {
       "epoch": 1.09,
-      "eval_accuracy": 0.6307692307692307,
-      "eval_loss": 1.1246747970581055,
-      "eval_runtime": 1.9197,
-      "eval_samples_per_second": 135.438,
-      "eval_steps_per_second": 17.19,
+      "eval_accuracy": 0.6115384615384616,
+      "eval_loss": 1.2288414239883423,
+      "eval_runtime": 2.0363,
+      "eval_samples_per_second": 127.683,
+      "eval_steps_per_second": 16.206,
       "step": 100
     },
     {
       "epoch": 1.2,
-      "learning_rate": 0.0001765217391304348,
-      "loss": 1.3096,
+      "learning_rate": 0.0001883695652173913,
+      "loss": 1.4324,
       "step": 110
     },
     {
       "epoch": 1.3,
-      "learning_rate": 0.00017434782608695653,
-      "loss": 1.3416,
+      "learning_rate": 0.00018728260869565216,
+      "loss": 1.2437,
       "step": 120
     },
     {
       "epoch": 1.41,
-      "learning_rate": 0.00017217391304347827,
-      "loss": 1.4005,
+      "learning_rate": 0.00018619565217391305,
+      "loss": 1.4662,
       "step": 130
     },
     {
       "epoch": 1.52,
-      "learning_rate": 0.00017,
-      "loss": 1.1935,
+      "learning_rate": 0.00018510869565217393,
+      "loss": 1.3662,
       "step": 140
     },
     {
       "epoch": 1.63,
-      "learning_rate": 0.00016782608695652175,
-      "loss": 1.0311,
+      "learning_rate": 0.0001840217391304348,
+      "loss": 1.2175,
       "step": 150
     },
     {
       "epoch": 1.74,
-      "learning_rate": 0.0001656521739130435,
-      "loss": 0.9773,
+      "learning_rate": 0.00018293478260869564,
+      "loss": 0.9581,
       "step": 160
     },
     {
       "epoch": 1.85,
-      "learning_rate": 0.00016347826086956523,
-      "loss": 1.0554,
+      "learning_rate": 0.00018184782608695655,
+      "loss": 0.9719,
       "step": 170
     },
     {
       "epoch": 1.96,
-      "learning_rate": 0.00016130434782608697,
-      "loss": 0.9596,
+      "learning_rate": 0.0001807608695652174,
+      "loss": 1.1808,
       "step": 180
     },
     {
       "epoch": 2.07,
-      "learning_rate": 0.00015913043478260868,
-      "loss": 1.1052,
+      "learning_rate": 0.00017967391304347827,
+      "loss": 1.1631,
       "step": 190
     },
     {
       "epoch": 2.17,
-      "learning_rate": 0.00015695652173913045,
-      "loss": 0.928,
+      "learning_rate": 0.00017858695652173912,
+      "loss": 1.0961,
       "step": 200
     },
     {
       "epoch": 2.17,
-      "eval_accuracy": 0.8461538461538461,
-      "eval_loss": 0.5504888892173767,
-      "eval_runtime": 1.8676,
-      "eval_samples_per_second": 139.218,
-      "eval_steps_per_second": 17.67,
+      "eval_accuracy": 0.8115384615384615,
+      "eval_loss": 0.6632584929466248,
+      "eval_runtime": 1.8061,
+      "eval_samples_per_second": 143.955,
+      "eval_steps_per_second": 18.271,
       "step": 200
     },
     {
       "epoch": 2.28,
-      "learning_rate": 0.0001547826086956522,
-      "loss": 0.7055,
+      "learning_rate": 0.0001775,
+      "loss": 0.8802,
       "step": 210
     },
     {
       "epoch": 2.39,
-      "learning_rate": 0.00015260869565217393,
-      "loss": 1.0022,
+      "learning_rate": 0.0001764130434782609,
+      "loss": 0.9565,
       "step": 220
     },
     {
       "epoch": 2.5,
-      "learning_rate": 0.00015043478260869567,
-      "loss": 0.8177,
+      "learning_rate": 0.00017532608695652174,
+      "loss": 0.8544,
       "step": 230
     },
     {
       "epoch": 2.61,
-      "learning_rate": 0.0001482608695652174,
-      "loss": 0.8889,
+      "learning_rate": 0.00017423913043478263,
+      "loss": 0.9685,
       "step": 240
     },
     {
       "epoch": 2.72,
-      "learning_rate": 0.00014608695652173914,
-      "loss": 0.7134,
+      "learning_rate": 0.00017315217391304348,
+      "loss": 0.952,
       "step": 250
     },
     {
       "epoch": 2.83,
-      "learning_rate": 0.00014391304347826086,
-      "loss": 0.7063,
+      "learning_rate": 0.00017206521739130434,
+      "loss": 0.908,
       "step": 260
     },
     {
       "epoch": 2.93,
-      "learning_rate": 0.00014173913043478262,
-      "loss": 0.558,
+      "learning_rate": 0.00017097826086956522,
+      "loss": 0.664,
       "step": 270
     },
     {
       "epoch": 3.04,
-      "learning_rate": 0.00013956521739130434,
-      "loss": 0.6297,
+      "learning_rate": 0.0001698913043478261,
+      "loss": 0.8025,
       "step": 280
     },
     {
       "epoch": 3.15,
-      "learning_rate": 0.0001373913043478261,
-      "loss": 0.6457,
+      "learning_rate": 0.00016880434782608696,
+      "loss": 0.8284,
       "step": 290
     },
     {
       "epoch": 3.26,
-      "learning_rate": 0.00013521739130434784,
-      "loss": 0.4477,
+      "learning_rate": 0.00016771739130434782,
+      "loss": 0.5971,
       "step": 300
     },
     {
       "epoch": 3.26,
-      "eval_accuracy": 0.9038461538461539,
-      "eval_loss": 0.31669339537620544,
-      "eval_runtime": 1.8909,
-      "eval_samples_per_second": 137.504,
-      "eval_steps_per_second": 17.452,
+      "eval_accuracy": 0.8846153846153846,
+      "eval_loss": 0.2988673150539398,
+      "eval_runtime": 1.8449,
+      "eval_samples_per_second": 140.926,
+      "eval_steps_per_second": 17.887,
       "step": 300
     },
     {
       "epoch": 3.37,
-      "learning_rate": 0.00013304347826086958,
-      "loss": 0.5847,
+      "learning_rate": 0.0001666304347826087,
+      "loss": 0.7568,
       "step": 310
     },
     {
       "epoch": 3.48,
-      "learning_rate": 0.00013086956521739132,
-      "loss": 0.8343,
+      "learning_rate": 0.00016554347826086958,
+      "loss": 0.8648,
       "step": 320
     },
     {
       "epoch": 3.59,
-      "learning_rate": 0.00012869565217391303,
-      "loss": 0.7171,
+      "learning_rate": 0.00016445652173913044,
+      "loss": 0.8262,
       "step": 330
     },
     {
       "epoch": 3.7,
-      "learning_rate": 0.0001265217391304348,
-      "loss": 0.5016,
+      "learning_rate": 0.0001633695652173913,
+      "loss": 0.6327,
       "step": 340
     },
     {
       "epoch": 3.8,
-      "learning_rate": 0.0001243478260869565,
-      "loss": 0.683,
+      "learning_rate": 0.00016228260869565218,
+      "loss": 0.6862,
       "step": 350
     },
     {
       "epoch": 3.91,
-      "learning_rate": 0.00012217391304347828,
-      "loss": 0.653,
+      "learning_rate": 0.00016119565217391306,
+      "loss": 0.7705,
       "step": 360
     },
     {
       "epoch": 4.02,
-      "learning_rate": 0.00012,
-      "loss": 0.6446,
+      "learning_rate": 0.00016010869565217392,
+      "loss": 0.6674,
       "step": 370
     },
     {
       "epoch": 4.13,
-      "learning_rate": 0.00011782608695652176,
-      "loss": 0.449,
+      "learning_rate": 0.00015902173913043478,
+      "loss": 0.7293,
       "step": 380
     },
     {
       "epoch": 4.24,
-      "learning_rate": 0.00011565217391304348,
-      "loss": 0.5571,
+      "learning_rate": 0.00015793478260869566,
+      "loss": 0.608,
       "step": 390
     },
     {
       "epoch": 4.35,
-      "learning_rate": 0.00011347826086956523,
-      "loss": 0.5154,
+      "learning_rate": 0.00015684782608695654,
+      "loss": 0.5663,
       "step": 400
     },
     {
       "epoch": 4.35,
-      "eval_accuracy": 0.9423076923076923,
-      "eval_loss": 0.21009394526481628,
-      "eval_runtime": 1.8763,
-      "eval_samples_per_second": 138.572,
-      "eval_steps_per_second": 17.588,
+      "eval_accuracy": 0.9538461538461539,
+      "eval_loss": 0.2658146917819977,
+      "eval_runtime": 1.8142,
+      "eval_samples_per_second": 143.313,
+      "eval_steps_per_second": 18.19,
       "step": 400
     },
     {
       "epoch": 4.46,
-      "learning_rate": 0.00011130434782608696,
-      "loss": 0.4713,
+      "learning_rate": 0.0001557608695652174,
+      "loss": 0.5467,
       "step": 410
     },
     {
       "epoch": 4.57,
-      "learning_rate": 0.00010913043478260869,
-      "loss": 0.451,
+      "learning_rate": 0.00015467391304347828,
+      "loss": 0.515,
       "step": 420
     },
     {
       "epoch": 4.67,
-      "learning_rate": 0.00010695652173913044,
-      "loss": 0.5028,
+      "learning_rate": 0.00015358695652173914,
+      "loss": 0.5092,
       "step": 430
     },
     {
       "epoch": 4.78,
-      "learning_rate": 0.00010478260869565218,
-      "loss": 0.4907,
+      "learning_rate": 0.0001525,
+      "loss": 0.6435,
       "step": 440
     },
     {
       "epoch": 4.89,
-      "learning_rate": 0.00010260869565217393,
-      "loss": 0.4996,
+      "learning_rate": 0.00015141304347826088,
+      "loss": 0.7066,
       "step": 450
     },
     {
       "epoch": 5.0,
-      "learning_rate": 0.00010043478260869566,
-      "loss": 0.5479,
+      "learning_rate": 0.00015032608695652176,
+      "loss": 0.6563,
       "step": 460
     },
     {
       "epoch": 5.11,
-      "learning_rate": 9.82608695652174e-05,
-      "loss": 0.4643,
+      "learning_rate": 0.00014923913043478262,
+      "loss": 0.5409,
       "step": 470
     },
     {
       "epoch": 5.22,
-      "learning_rate": 9.608695652173914e-05,
-      "loss": 0.5042,
+      "learning_rate": 0.00014815217391304347,
+      "loss": 0.5893,
       "step": 480
     },
     {
       "epoch": 5.33,
-      "learning_rate": 9.391304347826087e-05,
-      "loss": 0.4327,
+      "learning_rate": 0.00014706521739130436,
+      "loss": 0.5211,
       "step": 490
     },
     {
       "epoch": 5.43,
-      "learning_rate": 9.173913043478261e-05,
-      "loss": 0.4627,
+      "learning_rate": 0.00014597826086956524,
+      "loss": 0.6431,
       "step": 500
     },
     {
       "epoch": 5.43,
-      "eval_accuracy": 0.9346153846153846,
-      "eval_loss": 0.2289218157529831,
-      "eval_runtime": 2.011,
-      "eval_samples_per_second": 129.291,
-      "eval_steps_per_second": 16.41,
+      "eval_accuracy": 0.8846153846153846,
+      "eval_loss": 0.3099747896194458,
+      "eval_runtime": 1.8526,
+      "eval_samples_per_second": 140.344,
+      "eval_steps_per_second": 17.813,
       "step": 500
     },
     {
       "epoch": 5.54,
-      "learning_rate": 8.956521739130435e-05,
-      "loss": 0.391,
+      "learning_rate": 0.0001448913043478261,
+      "loss": 0.5512,
       "step": 510
     },
     {
       "epoch": 5.65,
-      "learning_rate": 8.739130434782609e-05,
-      "loss": 0.4032,
+      "learning_rate": 0.00014380434782608695,
+      "loss": 0.6073,
       "step": 520
     },
     {
       "epoch": 5.76,
-      "learning_rate": 8.521739130434783e-05,
-      "loss": 0.3878,
+      "learning_rate": 0.00014271739130434783,
+      "loss": 0.5843,
       "step": 530
     },
     {
       "epoch": 5.87,
-      "learning_rate": 8.304347826086957e-05,
-      "loss": 0.5423,
+      "learning_rate": 0.00014163043478260872,
+      "loss": 0.637,
       "step": 540
     },
     {
       "epoch": 5.98,
-      "learning_rate": 8.086956521739131e-05,
-      "loss": 0.4408,
+      "learning_rate": 0.00014054347826086957,
+      "loss": 0.4641,
       "step": 550
     },
     {
       "epoch": 6.09,
-      "learning_rate": 7.869565217391305e-05,
-      "loss": 0.5748,
+      "learning_rate": 0.00013945652173913043,
+      "loss": 0.6379,
       "step": 560
     },
     {
       "epoch": 6.2,
-      "learning_rate": 7.652173913043479e-05,
-      "loss": 0.4865,
+      "learning_rate": 0.0001383695652173913,
+      "loss": 0.5705,
       "step": 570
     },
     {
       "epoch": 6.3,
-      "learning_rate": 7.434782608695653e-05,
-      "loss": 0.2866,
+      "learning_rate": 0.00013728260869565217,
+      "loss": 0.3212,
       "step": 580
     },
     {
       "epoch": 6.41,
-      "learning_rate": 7.217391304347827e-05,
-      "loss": 0.3527,
+      "learning_rate": 0.00013619565217391305,
+      "loss": 0.5207,
       "step": 590
     },
     {
       "epoch": 6.52,
-      "learning_rate": 7e-05,
-      "loss": 0.3103,
+      "learning_rate": 0.00013510869565217394,
+      "loss": 0.49,
       "step": 600
     },
     {
       "epoch": 6.52,
-      "eval_accuracy": 0.9615384615384616,
-      "eval_loss": 0.15010055899620056,
-      "eval_runtime": 1.8802,
-      "eval_samples_per_second": 138.282,
-      "eval_steps_per_second": 17.551,
+      "eval_accuracy": 0.9192307692307692,
+      "eval_loss": 0.24573320150375366,
+      "eval_runtime": 2.0074,
+      "eval_samples_per_second": 129.523,
+      "eval_steps_per_second": 16.439,
       "step": 600
     },
     {
       "epoch": 6.63,
-      "learning_rate": 6.782608695652173e-05,
-      "loss": 0.4045,
+      "learning_rate": 0.0001340217391304348,
+      "loss": 0.452,
       "step": 610
     },
     {
       "epoch": 6.74,
-      "learning_rate": 6.565217391304349e-05,
-      "loss": 0.2758,
+      "learning_rate": 0.00013293478260869565,
+      "loss": 0.3719,
       "step": 620
     },
     {
       "epoch": 6.85,
-      "learning_rate": 6.347826086956523e-05,
-      "loss": 0.3185,
+      "learning_rate": 0.00013184782608695653,
+      "loss": 0.4104,
       "step": 630
     },
     {
       "epoch": 6.96,
-      "learning_rate": 6.130434782608696e-05,
-      "loss": 0.3133,
+      "learning_rate": 0.00013076086956521741,
+      "loss": 0.5211,
       "step": 640
     },
     {
       "epoch": 7.07,
-      "learning_rate": 5.9130434782608704e-05,
-      "loss": 0.2535,
+      "learning_rate": 0.00012967391304347827,
+      "loss": 0.4204,
       "step": 650
     },
     {
       "epoch": 7.17,
-      "learning_rate": 5.695652173913044e-05,
-      "loss": 0.2885,
+      "learning_rate": 0.00012858695652173913,
+      "loss": 0.312,
       "step": 660
     },
     {
       "epoch": 7.28,
-      "learning_rate": 5.478260869565217e-05,
-      "loss": 0.3363,
+      "learning_rate": 0.0001275,
+      "loss": 0.4948,
       "step": 670
     },
     {
       "epoch": 7.39,
-      "learning_rate": 5.260869565217391e-05,
-      "loss": 0.3127,
+      "learning_rate": 0.0001264130434782609,
+      "loss": 0.4796,
       "step": 680
     },
     {
       "epoch": 7.5,
-      "learning_rate": 5.0434782608695655e-05,
-      "loss": 0.3392,
+      "learning_rate": 0.00012532608695652175,
+      "loss": 0.5536,
       "step": 690
     },
     {
       "epoch": 7.61,
-      "learning_rate": 4.8260869565217394e-05,
-      "loss": 0.3637,
+      "learning_rate": 0.0001242391304347826,
+      "loss": 0.4639,
       "step": 700
     },
     {
       "epoch": 7.61,
-      "eval_accuracy": 0.9769230769230769,
-      "eval_loss": 0.13189926743507385,
-      "eval_runtime": 2.2676,
-      "eval_samples_per_second": 114.659,
-      "eval_steps_per_second": 14.553,
+      "eval_accuracy": 0.9153846153846154,
+      "eval_loss": 0.295836478471756,
+      "eval_runtime": 1.8268,
+      "eval_samples_per_second": 142.325,
+      "eval_steps_per_second": 18.064,
       "step": 700
     },
     {
       "epoch": 7.72,
-      "learning_rate": 4.608695652173913e-05,
-      "loss": 0.3521,
+      "learning_rate": 0.0001231521739130435,
+      "loss": 0.4924,
       "step": 710
     },
     {
       "epoch": 7.83,
-      "learning_rate": 4.391304347826087e-05,
-      "loss": 0.2299,
+      "learning_rate": 0.00012206521739130434,
+      "loss": 0.3013,
       "step": 720
     },
     {
       "epoch": 7.93,
-      "learning_rate": 4.1739130434782605e-05,
-      "loss": 0.3716,
+      "learning_rate": 0.00012097826086956523,
+      "loss": 0.4905,
       "step": 730
     },
     {
       "epoch": 8.04,
-      "learning_rate": 3.956521739130435e-05,
-      "loss": 0.3325,
+      "learning_rate": 0.0001198913043478261,
+      "loss": 0.3835,
       "step": 740
     },
     {
       "epoch": 8.15,
-      "learning_rate": 3.739130434782609e-05,
-      "loss": 0.2256,
+      "learning_rate": 0.00011880434782608695,
+      "loss": 0.3558,
       "step": 750
     },
     {
       "epoch": 8.26,
-      "learning_rate": 3.521739130434783e-05,
-      "loss": 0.4014,
+      "learning_rate": 0.00011771739130434782,
+      "loss": 0.5588,
       "step": 760
     },
     {
       "epoch": 8.37,
-      "learning_rate": 3.304347826086956e-05,
-      "loss": 0.2468,
+      "learning_rate": 0.0001166304347826087,
+      "loss": 0.3189,
       "step": 770
     },
     {
       "epoch": 8.48,
-      "learning_rate": 3.086956521739131e-05,
-      "loss": 0.3243,
+      "learning_rate": 0.00011554347826086958,
+      "loss": 0.3049,
       "step": 780
     },
     {
       "epoch": 8.59,
-      "learning_rate": 2.8695652173913044e-05,
-      "loss": 0.298,
+      "learning_rate": 0.00011445652173913045,
+      "loss": 0.4394,
       "step": 790
     },
     {
       "epoch": 8.7,
-      "learning_rate": 2.6521739130434787e-05,
-      "loss": 0.1941,
+      "learning_rate": 0.0001133695652173913,
+      "loss": 0.2436,
       "step": 800
     },
     {
       "epoch": 8.7,
-      "eval_accuracy": 0.9769230769230769,
-      "eval_loss": 0.08613809198141098,
-      "eval_runtime": 1.8262,
-      "eval_samples_per_second": 142.375,
-      "eval_steps_per_second": 18.071,
+      "eval_accuracy": 0.9192307692307692,
+      "eval_loss": 0.2146582454442978,
+      "eval_runtime": 1.8308,
+      "eval_samples_per_second": 142.012,
+      "eval_steps_per_second": 18.025,
       "step": 800
     },
     {
       "epoch": 8.8,
-      "learning_rate": 2.4347826086956523e-05,
-      "loss": 0.1929,
+      "learning_rate": 0.00011228260869565217,
+      "loss": 0.3024,
       "step": 810
     },
     {
       "epoch": 8.91,
-      "learning_rate": 2.2173913043478262e-05,
-      "loss": 0.1907,
+      "learning_rate": 0.00011119565217391305,
+      "loss": 0.3009,
       "step": 820
     },
     {
       "epoch": 9.02,
-      "learning_rate": 2e-05,
-      "loss": 0.2992,
+      "learning_rate": 0.00011010869565217392,
+      "loss": 0.3867,
       "step": 830
     },
     {
       "epoch": 9.13,
-      "learning_rate": 1.782608695652174e-05,
-      "loss": 0.2516,
+      "learning_rate": 0.00010902173913043478,
+      "loss": 0.3032,
       "step": 840
     },
     {
       "epoch": 9.24,
-      "learning_rate": 1.565217391304348e-05,
-      "loss": 0.2113,
+      "learning_rate": 0.00010793478260869565,
+      "loss": 0.3227,
       "step": 850
     },
     {
       "epoch": 9.35,
-      "learning_rate": 1.3478260869565218e-05,
-      "loss": 0.1503,
+      "learning_rate": 0.00010684782608695653,
+      "loss": 0.2126,
       "step": 860
     },
     {
       "epoch": 9.46,
-      "learning_rate": 1.1304347826086957e-05,
-      "loss": 0.1664,
+      "learning_rate": 0.0001057608695652174,
+      "loss": 0.2606,
       "step": 870
     },
     {
       "epoch": 9.57,
-      "learning_rate": 9.130434782608697e-06,
-      "loss": 0.398,
+      "learning_rate": 0.00010467391304347827,
+      "loss": 0.4299,
       "step": 880
     },
     {
       "epoch": 9.67,
-      "learning_rate": 6.956521739130435e-06,
-      "loss": 0.3558,
+      "learning_rate": 0.00010358695652173913,
+      "loss": 0.5039,
       "step": 890
     },
     {
       "epoch": 9.78,
-      "learning_rate": 4.782608695652174e-06,
-      "loss": 0.1663,
+      "learning_rate": 0.0001025,
+      "loss": 0.2652,
       "step": 900
     },
     {
       "epoch": 9.78,
-      "eval_accuracy": 0.9769230769230769,
-      "eval_loss": 0.08665793389081955,
-      "eval_runtime": 2.6289,
-      "eval_samples_per_second": 98.9,
-      "eval_steps_per_second": 12.553,
+      "eval_accuracy": 0.9615384615384616,
+      "eval_loss": 0.18220607936382294,
+      "eval_runtime": 1.8325,
+      "eval_samples_per_second": 141.881,
+      "eval_steps_per_second": 18.008,
       "step": 900
     },
     {
       "epoch": 9.89,
-      "learning_rate": 2.608695652173913e-06,
-      "loss": 0.1478,
+      "learning_rate": 0.00010141304347826088,
+      "loss": 0.2374,
       "step": 910
     },
     {
       "epoch": 10.0,
-      "learning_rate": 4.347826086956522e-07,
-      "loss": 0.2479,
+      "learning_rate": 0.00010032608695652175,
+      "loss": 0.3522,
       "step": 920
     },
     {
-      "epoch": 10.0,
-      "step": 920,
-      "total_flos": 3.651306173035315e+17,
-      "train_loss": 0.7250858170830685,
-      "train_runtime": 208.4502,
-      "train_samples_per_second": 70.424,
-      "train_steps_per_second": 4.414
+      "epoch": 10.11,
+      "learning_rate": 9.923913043478261e-05,
+      "loss": 0.2455,
+      "step": 930
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 9.815217391304349e-05,
+      "loss": 0.2028,
+      "step": 940
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 9.706521739130435e-05,
+      "loss": 0.4767,
+      "step": 950
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 9.597826086956522e-05,
+      "loss": 0.2277,
+      "step": 960
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 9.48913043478261e-05,
+      "loss": 0.3051,
+      "step": 970
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 9.380434782608696e-05,
+      "loss": 0.1665,
+      "step": 980
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 9.271739130434784e-05,
+      "loss": 0.3633,
+      "step": 990
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 9.16304347826087e-05,
+      "loss": 0.2346,
+      "step": 1000
+    },
+    {
+      "epoch": 10.87,
+      "eval_accuracy": 0.9384615384615385,
+      "eval_loss": 0.27161160111427307,
+      "eval_runtime": 1.9489,
+      "eval_samples_per_second": 133.405,
+      "eval_steps_per_second": 16.932,
+      "step": 1000
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 9.054347826086958e-05,
+      "loss": 0.3591,
+      "step": 1010
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 8.945652173913043e-05,
+      "loss": 0.1646,
+      "step": 1020
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 8.83695652173913e-05,
+      "loss": 0.2254,
+      "step": 1030
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 8.728260869565217e-05,
+      "loss": 0.2712,
+      "step": 1040
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 8.619565217391304e-05,
+      "loss": 0.3244,
+      "step": 1050
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 8.510869565217393e-05,
+      "loss": 0.5431,
+      "step": 1060
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 8.402173913043478e-05,
+      "loss": 0.3306,
+      "step": 1070
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 8.293478260869567e-05,
+      "loss": 0.2765,
+      "step": 1080
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 8.184782608695652e-05,
+      "loss": 0.1797,
+      "step": 1090
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 8.076086956521739e-05,
+      "loss": 0.3326,
+      "step": 1100
+    },
+    {
+      "epoch": 11.96,
+      "eval_accuracy": 0.9692307692307692,
+      "eval_loss": 0.09300269931554794,
+      "eval_runtime": 1.8336,
+      "eval_samples_per_second": 141.797,
+      "eval_steps_per_second": 17.997,
+      "step": 1100
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 7.967391304347826e-05,
+      "loss": 0.2801,
+      "step": 1110
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 7.858695652173913e-05,
+      "loss": 0.1963,
+      "step": 1120
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 7.75e-05,
+      "loss": 0.3328,
+      "step": 1130
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 7.641304347826087e-05,
+      "loss": 0.2452,
+      "step": 1140
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 7.532608695652175e-05,
+      "loss": 0.3876,
+      "step": 1150
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 7.423913043478261e-05,
+      "loss": 0.3523,
+      "step": 1160
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 7.315217391304349e-05,
+      "loss": 0.2854,
+      "step": 1170
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 7.206521739130435e-05,
+      "loss": 0.3044,
+      "step": 1180
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 7.097826086956522e-05,
+      "loss": 0.2829,
+      "step": 1190
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 6.989130434782609e-05,
+      "loss": 0.4255,
+      "step": 1200
+    },
+    {
+      "epoch": 13.04,
+      "eval_accuracy": 0.95,
+      "eval_loss": 0.15650111436843872,
+      "eval_runtime": 2.2463,
+      "eval_samples_per_second": 115.745,
+      "eval_steps_per_second": 14.691,
+      "step": 1200
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 6.880434782608696e-05,
+      "loss": 0.4168,
+      "step": 1210
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 6.771739130434783e-05,
+      "loss": 0.1493,
+      "step": 1220
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 6.66304347826087e-05,
+      "loss": 0.3662,
+      "step": 1230
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 6.554347826086957e-05,
+      "loss": 0.2641,
+      "step": 1240
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 6.445652173913044e-05,
+      "loss": 0.1724,
+      "step": 1250
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 6.33695652173913e-05,
+      "loss": 0.2753,
+      "step": 1260
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 6.228260869565218e-05,
+      "loss": 0.3445,
+      "step": 1270
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 6.119565217391304e-05,
+      "loss": 0.1927,
+      "step": 1280
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 6.010869565217392e-05,
+      "loss": 0.2233,
+      "step": 1290
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 5.9021739130434784e-05,
+      "loss": 0.3486,
+      "step": 1300
+    },
+    {
+      "epoch": 14.13,
+      "eval_accuracy": 0.9807692307692307,
+      "eval_loss": 0.08807563781738281,
+      "eval_runtime": 1.846,
+      "eval_samples_per_second": 140.846,
+      "eval_steps_per_second": 17.877,
+      "step": 1300
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 5.793478260869566e-05,
+      "loss": 0.2479,
+      "step": 1310
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 5.6847826086956524e-05,
+      "loss": 0.3112,
+      "step": 1320
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 5.5760869565217386e-05,
+      "loss": 0.2721,
+      "step": 1330
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 5.467391304347826e-05,
+      "loss": 0.145,
+      "step": 1340
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 5.3586956521739126e-05,
+      "loss": 0.323,
+      "step": 1350
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 5.25e-05,
+      "loss": 0.2676,
+      "step": 1360
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 5.141304347826087e-05,
+      "loss": 0.2826,
+      "step": 1370
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 5.032608695652175e-05,
+      "loss": 0.219,
+      "step": 1380
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 4.923913043478261e-05,
+      "loss": 0.176,
+      "step": 1390
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 4.815217391304348e-05,
+      "loss": 0.2801,
+      "step": 1400
+    },
+    {
+      "epoch": 15.22,
+      "eval_accuracy": 0.9576923076923077,
+      "eval_loss": 0.10833106189966202,
+      "eval_runtime": 2.5226,
+      "eval_samples_per_second": 103.068,
+      "eval_steps_per_second": 13.082,
+      "step": 1400
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 4.706521739130435e-05,
+      "loss": 0.1336,
+      "step": 1410
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 4.597826086956522e-05,
+      "loss": 0.1522,
+      "step": 1420
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 4.489130434782609e-05,
+      "loss": 0.2929,
+      "step": 1430
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 4.380434782608696e-05,
+      "loss": 0.2111,
+      "step": 1440
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 4.271739130434783e-05,
+      "loss": 0.3614,
+      "step": 1450
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 4.16304347826087e-05,
+      "loss": 0.3736,
+      "step": 1460
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 4.054347826086957e-05,
+      "loss": 0.2388,
+      "step": 1470
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 3.945652173913044e-05,
+      "loss": 0.212,
+      "step": 1480
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 3.836956521739131e-05,
+      "loss": 0.1982,
+      "step": 1490
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 3.728260869565218e-05,
+      "loss": 0.164,
+      "step": 1500
+    },
+    {
+      "epoch": 16.3,
+      "eval_accuracy": 0.9730769230769231,
+      "eval_loss": 0.09941530972719193,
+      "eval_runtime": 1.8161,
+      "eval_samples_per_second": 143.163,
+      "eval_steps_per_second": 18.171,
+      "step": 1500
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 3.619565217391305e-05,
+      "loss": 0.1593,
+      "step": 1510
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 3.510869565217392e-05,
+      "loss": 0.127,
+      "step": 1520
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 3.402173913043478e-05,
+      "loss": 0.2097,
+      "step": 1530
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 3.293478260869565e-05,
+      "loss": 0.2947,
+      "step": 1540
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 3.1847826086956526e-05,
+      "loss": 0.2027,
+      "step": 1550
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 3.0760869565217395e-05,
+      "loss": 0.29,
+      "step": 1560
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 2.967391304347826e-05,
+      "loss": 0.1293,
+      "step": 1570
+    },
+    {
+      "epoch": 17.17,
+      "learning_rate": 2.8586956521739135e-05,
+      "loss": 0.2181,
+      "step": 1580
+    },
+    {
+      "epoch": 17.28,
+      "learning_rate": 2.7500000000000004e-05,
+      "loss": 0.2735,
+      "step": 1590
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 2.6413043478260867e-05,
+      "loss": 0.2892,
+      "step": 1600
+    },
+    {
+      "epoch": 17.39,
+      "eval_accuracy": 0.9846153846153847,
+      "eval_loss": 0.07838542014360428,
+      "eval_runtime": 2.5953,
+      "eval_samples_per_second": 100.181,
+      "eval_steps_per_second": 12.715,
+      "step": 1600
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 2.5326086956521737e-05,
+      "loss": 0.1602,
+      "step": 1610
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 2.423913043478261e-05,
+      "loss": 0.1958,
+      "step": 1620
+    },
+    {
+      "epoch": 17.72,
+      "learning_rate": 2.315217391304348e-05,
+      "loss": 0.323,
+      "step": 1630
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 2.206521739130435e-05,
+      "loss": 0.1871,
+      "step": 1640
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 2.097826086956522e-05,
+      "loss": 0.2017,
+      "step": 1650
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 1.9891304347826085e-05,
+      "loss": 0.1395,
+      "step": 1660
+    },
+    {
+      "epoch": 18.15,
+      "learning_rate": 1.8804347826086958e-05,
+      "loss": 0.217,
+      "step": 1670
+    },
+    {
+      "epoch": 18.26,
+      "learning_rate": 1.7717391304347828e-05,
+      "loss": 0.1622,
+      "step": 1680
+    },
+    {
+      "epoch": 18.37,
+      "learning_rate": 1.6630434782608698e-05,
+      "loss": 0.2306,
+      "step": 1690
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 1.5543478260869564e-05,
+      "loss": 0.1033,
+      "step": 1700
+    },
+    {
+      "epoch": 18.48,
+      "eval_accuracy": 0.9807692307692307,
+      "eval_loss": 0.09326102584600449,
+      "eval_runtime": 1.8301,
+      "eval_samples_per_second": 142.071,
+      "eval_steps_per_second": 18.032,
+      "step": 1700
+    },
+    {
+      "epoch": 18.59,
+      "learning_rate": 1.4456521739130435e-05,
+      "loss": 0.2286,
+      "step": 1710
+    },
+    {
+      "epoch": 18.7,
+      "learning_rate": 1.3369565217391305e-05,
+      "loss": 0.1722,
+      "step": 1720
+    },
+    {
+      "epoch": 18.8,
+      "learning_rate": 1.2282608695652175e-05,
+      "loss": 0.1801,
+      "step": 1730
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 1.1195652173913044e-05,
+      "loss": 0.1624,
+      "step": 1740
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 1.0108695652173914e-05,
+      "loss": 0.1473,
+      "step": 1750
+    },
+    {
+      "epoch": 19.13,
+      "learning_rate": 9.021739130434784e-06,
+      "loss": 0.2875,
+      "step": 1760
+    },
+    {
+      "epoch": 19.24,
+      "learning_rate": 7.934782608695653e-06,
+      "loss": 0.193,
+      "step": 1770
+    },
+    {
+      "epoch": 19.35,
+      "learning_rate": 6.847826086956521e-06,
+      "loss": 0.1731,
+      "step": 1780
+    },
+    {
+      "epoch": 19.46,
+      "learning_rate": 5.760869565217392e-06,
+      "loss": 0.1206,
+      "step": 1790
+    },
+    {
+      "epoch": 19.57,
+      "learning_rate": 4.673913043478261e-06,
+      "loss": 0.1612,
+      "step": 1800
+    },
+    {
+      "epoch": 19.57,
+      "eval_accuracy": 0.9846153846153847,
+      "eval_loss": 0.0834396481513977,
+      "eval_runtime": 2.6459,
+      "eval_samples_per_second": 98.267,
+      "eval_steps_per_second": 12.472,
+      "step": 1800
+    },
+    {
+      "epoch": 19.67,
+      "learning_rate": 3.5869565217391305e-06,
+      "loss": 0.2104,
+      "step": 1810
+    },
+    {
+      "epoch": 19.78,
+      "learning_rate": 2.5e-06,
+      "loss": 0.1541,
+      "step": 1820
+    },
+    {
+      "epoch": 19.89,
+      "learning_rate": 1.4130434782608697e-06,
+      "loss": 0.1168,
+      "step": 1830
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 3.260869565217391e-07,
+      "loss": 0.1346,
+      "step": 1840
+    },
+    {
+      "epoch": 20.0,
+      "step": 1840,
+      "total_flos": 7.30261234607063e+17,
+      "train_loss": 0.5343924296938855,
+      "train_runtime": 403.7255,
+      "train_samples_per_second": 72.723,
+      "train_steps_per_second": 4.558
     }
   ],
   "logging_steps": 10,
-  "max_steps": 920,
-  "num_train_epochs": 10,
+  "max_steps": 1840,
+  "num_train_epochs": 20,
   "save_steps": 100,
-  "total_flos": 3.651306173035315e+17,
+  "total_flos": 7.30261234607063e+17,
   "trial_name": null,
   "trial_params": null
 }