End of training

README.md

@@ -22,7 +22,7 @@ model-index:
     metrics:
     - name: Accuracy
       type: accuracy
-      value: 0.5
+      value: 0.5909090909090909
 ---
 
 <!-- This model card has been generated automatically according to the information the Trainer had access to. You
@@ -32,8 +32,8 @@ should probably proofread and complete it, then remove this comment. -->
 
 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 imagefolder dataset.
 It achieves the following results on the evaluation set:
-- Loss: 2.7604
-- Accuracy: 0.5
+- Loss: 2.3724
+- Accuracy: 0.5909
 
 ## Model description
 

all_results.json

@@ -1,13 +1,13 @@
 {
     "epoch": 48.97959183673469,
-    "eval_accuracy": 0.45454545454545453,
-    "eval_loss": 1.5689846277236938,
-    "eval_runtime": 3.4914,
-    "eval_samples_per_second": 6.301,
-    "eval_steps_per_second": 3.151,
+    "eval_accuracy": 0.5909090909090909,
+    "eval_loss": 2.3724472522735596,
+    "eval_runtime": 4.1304,
+    "eval_samples_per_second": 5.326,
+    "eval_steps_per_second": 2.663,
     "total_flos": 2.374708462608384e+17,
-    "train_loss": 0.41171714147552846,
-    "train_runtime": 1608.8411,
-    "train_samples_per_second": 6.06,
-    "train_steps_per_second": 0.746
+    "train_loss": 0.4113796023776134,
+    "train_runtime": 1728.9004,
+    "train_samples_per_second": 5.639,
+    "train_steps_per_second": 0.694
 }

eval_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 9.795918367346939,
-    "eval_accuracy": 0.45454545454545453,
-    "eval_loss": 1.5689846277236938,
-    "eval_runtime": 3.4914,
-    "eval_samples_per_second": 6.301,
-    "eval_steps_per_second": 3.151
+    "epoch": 48.97959183673469,
+    "eval_accuracy": 0.5909090909090909,
+    "eval_loss": 2.3724472522735596,
+    "eval_runtime": 4.1304,
+    "eval_samples_per_second": 5.326,
+    "eval_steps_per_second": 2.663
 }

runs/May30_23-10-22_ip-10-110-161-167.ap-southeast-2.compute.internal/events.out.tfevents.1717112373.ip-10-110-161-167.ap-southeast-2.compute.internal.5685.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4871b8e522c2f4cede2301a32201bd242b18dc6170a0f437fa67f823b540e5dc
+size 411

train_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 48.97959183673469,
     "total_flos": 2.374708462608384e+17,
-    "train_loss": 0.41171714147552846,
-    "train_runtime": 1608.8411,
-    "train_samples_per_second": 6.06,
-    "train_steps_per_second": 0.746
+    "train_loss": 0.4113796023776134,
+    "train_runtime": 1728.9004,
+    "train_samples_per_second": 5.639,
+    "train_steps_per_second": 0.694
 }

trainer_state.json

@@ -1,6 +1,6 @@
 {
-  "best_metric": 0.5454545454545454,
-  "best_model_checkpoint": "swin-tiny-patch4-window7-224-finetuned-isic217/checkpoint-245",
+  "best_metric": 0.5909090909090909,
+  "best_model_checkpoint": "swin-tiny-patch4-window7-224-finetuned-isic217/checkpoint-563",
   "epoch": 48.97959183673469,
   "eval_steps": 500,
   "global_step": 1200,
@@ -10,1293 +10,1293 @@
   "log_history": [
     {
       "epoch": 0.40816326530612246,
-      "grad_norm": 20.10276222229004,
+      "grad_norm": 19.762880325317383,
       "learning_rate": 4.166666666666667e-06,
-      "loss": 2.1681,
+      "loss": 2.2252,
       "step": 10
     },
     {
       "epoch": 0.8163265306122449,
-      "grad_norm": 18.530778884887695,
+      "grad_norm": 15.404119491577148,
       "learning_rate": 8.333333333333334e-06,
-      "loss": 2.2404,
+      "loss": 2.2679,
       "step": 20
     },
     {
       "epoch": 0.9795918367346939,
-      "eval_accuracy": 0.13636363636363635,
-      "eval_loss": 2.1526713371276855,
-      "eval_runtime": 3.7709,
-      "eval_samples_per_second": 5.834,
-      "eval_steps_per_second": 2.917,
+      "eval_accuracy": 0.09090909090909091,
+      "eval_loss": 2.1550188064575195,
+      "eval_runtime": 6.6011,
+      "eval_samples_per_second": 3.333,
+      "eval_steps_per_second": 1.666,
       "step": 24
     },
     {
       "epoch": 1.2244897959183674,
-      "grad_norm": 19.15616798400879,
+      "grad_norm": 21.01092529296875,
       "learning_rate": 1.25e-05,
-      "loss": 2.1402,
+      "loss": 2.0909,
       "step": 30
     },
     {
       "epoch": 1.6326530612244898,
-      "grad_norm": 18.088478088378906,
+      "grad_norm": 19.893291473388672,
       "learning_rate": 1.6666666666666667e-05,
-      "loss": 2.0749,
+      "loss": 2.0504,
       "step": 40
     },
     {
       "epoch": 2.0,
-      "eval_accuracy": 0.13636363636363635,
-      "eval_loss": 2.115939140319824,
-      "eval_runtime": 4.2644,
-      "eval_samples_per_second": 5.159,
-      "eval_steps_per_second": 2.58,
+      "eval_accuracy": 0.2727272727272727,
+      "eval_loss": 2.0559206008911133,
+      "eval_runtime": 3.7884,
+      "eval_samples_per_second": 5.807,
+      "eval_steps_per_second": 2.904,
       "step": 49
     },
     {
       "epoch": 2.0408163265306123,
-      "grad_norm": 28.553932189941406,
+      "grad_norm": 18.42198944091797,
       "learning_rate": 2.0833333333333336e-05,
-      "loss": 2.0201,
+      "loss": 1.987,
       "step": 50
     },
     {
       "epoch": 2.4489795918367347,
-      "grad_norm": 21.90468406677246,
+      "grad_norm": 29.596494674682617,
       "learning_rate": 2.5e-05,
-      "loss": 1.8697,
+      "loss": 1.8355,
       "step": 60
     },
     {
       "epoch": 2.857142857142857,
-      "grad_norm": 29.221223831176758,
+      "grad_norm": 23.39991569519043,
       "learning_rate": 2.916666666666667e-05,
-      "loss": 1.947,
+      "loss": 1.8943,
       "step": 70
     },
     {
       "epoch": 2.979591836734694,
-      "eval_accuracy": 0.13636363636363635,
-      "eval_loss": 2.172260284423828,
-      "eval_runtime": 3.7285,
-      "eval_samples_per_second": 5.9,
-      "eval_steps_per_second": 2.95,
+      "eval_accuracy": 0.22727272727272727,
+      "eval_loss": 2.0185506343841553,
+      "eval_runtime": 3.7191,
+      "eval_samples_per_second": 5.915,
+      "eval_steps_per_second": 2.958,
       "step": 73
     },
     {
       "epoch": 3.2653061224489797,
-      "grad_norm": 26.88852882385254,
+      "grad_norm": 34.72861099243164,
       "learning_rate": 3.3333333333333335e-05,
-      "loss": 1.8989,
+      "loss": 1.6665,
       "step": 80
     },
     {
       "epoch": 3.673469387755102,
-      "grad_norm": 29.26166343688965,
+      "grad_norm": 23.47280502319336,
       "learning_rate": 3.7500000000000003e-05,
-      "loss": 1.732,
+      "loss": 1.5671,
       "step": 90
     },
     {
       "epoch": 4.0,
-      "eval_accuracy": 0.09090909090909091,
-      "eval_loss": 2.154468536376953,
-      "eval_runtime": 4.2209,
-      "eval_samples_per_second": 5.212,
-      "eval_steps_per_second": 2.606,
+      "eval_accuracy": 0.22727272727272727,
+      "eval_loss": 1.815375804901123,
+      "eval_runtime": 3.7909,
+      "eval_samples_per_second": 5.803,
+      "eval_steps_per_second": 2.902,
       "step": 98
     },
     {
       "epoch": 4.081632653061225,
-      "grad_norm": 39.108612060546875,
+      "grad_norm": 28.841514587402344,
       "learning_rate": 4.166666666666667e-05,
-      "loss": 1.6138,
+      "loss": 1.664,
       "step": 100
     },
     {
       "epoch": 4.489795918367347,
-      "grad_norm": 23.49874496459961,
+      "grad_norm": 30.53756332397461,
       "learning_rate": 4.5833333333333334e-05,
-      "loss": 1.4534,
+      "loss": 1.553,
       "step": 110
     },
     {
       "epoch": 4.8979591836734695,
-      "grad_norm": 26.28005599975586,
+      "grad_norm": 47.8935432434082,
       "learning_rate": 5e-05,
-      "loss": 1.446,
+      "loss": 1.3425,
       "step": 120
     },
     {
       "epoch": 4.979591836734694,
-      "eval_accuracy": 0.18181818181818182,
-      "eval_loss": 2.2917678356170654,
-      "eval_runtime": 3.8303,
-      "eval_samples_per_second": 5.744,
-      "eval_steps_per_second": 2.872,
+      "eval_accuracy": 0.22727272727272727,
+      "eval_loss": 2.047502279281616,
+      "eval_runtime": 3.683,
+      "eval_samples_per_second": 5.973,
+      "eval_steps_per_second": 2.987,
       "step": 122
     },
     {
       "epoch": 5.3061224489795915,
-      "grad_norm": 38.8511962890625,
+      "grad_norm": 21.777742385864258,
       "learning_rate": 4.9537037037037035e-05,
-      "loss": 1.2513,
+      "loss": 1.2543,
       "step": 130
     },
     {
       "epoch": 5.714285714285714,
-      "grad_norm": 29.820043563842773,
+      "grad_norm": 32.264713287353516,
       "learning_rate": 4.9074074074074075e-05,
-      "loss": 1.1175,
+      "loss": 1.2758,
       "step": 140
     },
     {
       "epoch": 6.0,
-      "eval_accuracy": 0.3181818181818182,
-      "eval_loss": 1.9290556907653809,
-      "eval_runtime": 4.301,
-      "eval_samples_per_second": 5.115,
-      "eval_steps_per_second": 2.558,
+      "eval_accuracy": 0.22727272727272727,
+      "eval_loss": 2.1913788318634033,
+      "eval_runtime": 3.8174,
+      "eval_samples_per_second": 5.763,
+      "eval_steps_per_second": 2.882,
       "step": 147
     },
     {
       "epoch": 6.122448979591836,
-      "grad_norm": 36.40614700317383,
+      "grad_norm": 30.771316528320312,
       "learning_rate": 4.8611111111111115e-05,
-      "loss": 1.3955,
+      "loss": 1.0049,
       "step": 150
     },
     {
       "epoch": 6.530612244897959,
-      "grad_norm": 31.91864776611328,
+      "grad_norm": 37.03094482421875,
       "learning_rate": 4.814814814814815e-05,
-      "loss": 1.0427,
+      "loss": 0.9949,
       "step": 160
     },
     {
       "epoch": 6.938775510204081,
-      "grad_norm": 35.82953643798828,
+      "grad_norm": 38.22206497192383,
       "learning_rate": 4.768518518518519e-05,
-      "loss": 1.1069,
+      "loss": 0.9808,
       "step": 170
     },
     {
       "epoch": 6.979591836734694,
       "eval_accuracy": 0.36363636363636365,
-      "eval_loss": 1.9551143646240234,
-      "eval_runtime": 4.268,
-      "eval_samples_per_second": 5.155,
-      "eval_steps_per_second": 2.577,
+      "eval_loss": 2.047807455062866,
+      "eval_runtime": 3.6897,
+      "eval_samples_per_second": 5.963,
+      "eval_steps_per_second": 2.981,
       "step": 171
     },
     {
       "epoch": 7.346938775510204,
-      "grad_norm": 30.75468635559082,
+      "grad_norm": 30.003612518310547,
       "learning_rate": 4.722222222222222e-05,
-      "loss": 0.9718,
+      "loss": 0.9797,
       "step": 180
     },
     {
       "epoch": 7.755102040816326,
-      "grad_norm": 33.49174880981445,
+      "grad_norm": 48.88200378417969,
       "learning_rate": 4.675925925925926e-05,
-      "loss": 0.7932,
+      "loss": 0.7246,
       "step": 190
     },
     {
       "epoch": 8.0,
       "eval_accuracy": 0.4090909090909091,
-      "eval_loss": 2.0534465312957764,
-      "eval_runtime": 4.3072,
-      "eval_samples_per_second": 5.108,
-      "eval_steps_per_second": 2.554,
+      "eval_loss": 1.884019374847412,
+      "eval_runtime": 4.5579,
+      "eval_samples_per_second": 4.827,
+      "eval_steps_per_second": 2.413,
       "step": 196
     },
     {
       "epoch": 8.16326530612245,
-      "grad_norm": 45.86684799194336,
+      "grad_norm": 25.771787643432617,
       "learning_rate": 4.62962962962963e-05,
-      "loss": 1.0077,
+      "loss": 0.8633,
       "step": 200
     },
     {
       "epoch": 8.571428571428571,
-      "grad_norm": 34.891082763671875,
+      "grad_norm": 22.62033462524414,
       "learning_rate": 4.5833333333333334e-05,
-      "loss": 0.7286,
+      "loss": 0.5814,
       "step": 210
     },
     {
       "epoch": 8.979591836734693,
-      "grad_norm": 30.408119201660156,
+      "grad_norm": 37.1585578918457,
       "learning_rate": 4.5370370370370374e-05,
-      "loss": 0.5994,
+      "loss": 0.7323,
       "step": 220
     },
     {
       "epoch": 8.979591836734693,
-      "eval_accuracy": 0.36363636363636365,
-      "eval_loss": 1.8135310411453247,
-      "eval_runtime": 3.8949,
-      "eval_samples_per_second": 5.648,
-      "eval_steps_per_second": 2.824,
+      "eval_accuracy": 0.4090909090909091,
+      "eval_loss": 2.1831090450286865,
+      "eval_runtime": 3.7295,
+      "eval_samples_per_second": 5.899,
+      "eval_steps_per_second": 2.949,
       "step": 220
     },
     {
       "epoch": 9.387755102040817,
-      "grad_norm": 20.37044906616211,
+      "grad_norm": 31.566368103027344,
       "learning_rate": 4.490740740740741e-05,
-      "loss": 0.7078,
+      "loss": 0.5923,
       "step": 230
     },
     {
       "epoch": 9.795918367346939,
-      "grad_norm": 14.344355583190918,
+      "grad_norm": 61.39928436279297,
       "learning_rate": 4.4444444444444447e-05,
-      "loss": 0.4671,
+      "loss": 0.4881,
       "step": 240
     },
     {
       "epoch": 10.0,
-      "eval_accuracy": 0.5454545454545454,
-      "eval_loss": 1.7822338342666626,
-      "eval_runtime": 4.1682,
-      "eval_samples_per_second": 5.278,
-      "eval_steps_per_second": 2.639,
+      "eval_accuracy": 0.36363636363636365,
+      "eval_loss": 2.286790132522583,
+      "eval_runtime": 3.7469,
+      "eval_samples_per_second": 5.872,
+      "eval_steps_per_second": 2.936,
       "step": 245
     },
     {
       "epoch": 10.204081632653061,
-      "grad_norm": 39.557579040527344,
+      "grad_norm": 57.27079772949219,
       "learning_rate": 4.3981481481481486e-05,
-      "loss": 0.5962,
+      "loss": 0.4739,
       "step": 250
     },
     {
       "epoch": 10.612244897959183,
-      "grad_norm": 7.858915328979492,
+      "grad_norm": 12.431879043579102,
       "learning_rate": 4.351851851851852e-05,
-      "loss": 0.4612,
+      "loss": 0.4346,
       "step": 260
     },
     {
       "epoch": 10.979591836734693,
-      "eval_accuracy": 0.5,
-      "eval_loss": 2.2529866695404053,
-      "eval_runtime": 3.7854,
-      "eval_samples_per_second": 5.812,
-      "eval_steps_per_second": 2.906,
+      "eval_accuracy": 0.45454545454545453,
+      "eval_loss": 2.231236457824707,
+      "eval_runtime": 3.6895,
+      "eval_samples_per_second": 5.963,
+      "eval_steps_per_second": 2.981,
       "step": 269
     },
     {
       "epoch": 11.020408163265307,
-      "grad_norm": 28.87532615661621,
+      "grad_norm": 13.14640998840332,
       "learning_rate": 4.305555555555556e-05,
-      "loss": 0.4548,
+      "loss": 0.3967,
       "step": 270
     },
     {
       "epoch": 11.428571428571429,
-      "grad_norm": 40.782554626464844,
+      "grad_norm": 37.04918670654297,
       "learning_rate": 4.259259259259259e-05,
-      "loss": 0.5538,
+      "loss": 0.4297,
       "step": 280
     },
     {
       "epoch": 11.83673469387755,
-      "grad_norm": 17.471019744873047,
+      "grad_norm": 35.07551574707031,
       "learning_rate": 4.212962962962963e-05,
-      "loss": 0.4016,
+      "loss": 0.5647,
       "step": 290
     },
     {
       "epoch": 12.0,
       "eval_accuracy": 0.4090909090909091,
-      "eval_loss": 1.8437938690185547,
-      "eval_runtime": 4.198,
-      "eval_samples_per_second": 5.241,
-      "eval_steps_per_second": 2.62,
+      "eval_loss": 1.9897090196609497,
+      "eval_runtime": 3.8455,
+      "eval_samples_per_second": 5.721,
+      "eval_steps_per_second": 2.86,
       "step": 294
     },
     {
       "epoch": 12.244897959183673,
-      "grad_norm": 38.50642395019531,
+      "grad_norm": 5.31343936920166,
       "learning_rate": 4.166666666666667e-05,
-      "loss": 0.4786,
+      "loss": 0.376,
       "step": 300
     },
     {
       "epoch": 12.653061224489797,
-      "grad_norm": 30.62510871887207,
+      "grad_norm": 2.7732479572296143,
       "learning_rate": 4.1203703703703705e-05,
-      "loss": 0.3947,
+      "loss": 0.1464,
       "step": 310
     },
     {
       "epoch": 12.979591836734693,
-      "eval_accuracy": 0.5,
-      "eval_loss": 1.8609530925750732,
-      "eval_runtime": 3.8332,
-      "eval_samples_per_second": 5.739,
-      "eval_steps_per_second": 2.87,
+      "eval_accuracy": 0.45454545454545453,
+      "eval_loss": 2.057887315750122,
+      "eval_runtime": 3.7271,
+      "eval_samples_per_second": 5.903,
+      "eval_steps_per_second": 2.951,
       "step": 318
     },
     {
       "epoch": 13.061224489795919,
-      "grad_norm": 1.0034515857696533,
+      "grad_norm": 27.16363525390625,
       "learning_rate": 4.074074074074074e-05,
-      "loss": 0.3527,
+      "loss": 0.4971,
       "step": 320
     },
     {
       "epoch": 13.46938775510204,
-      "grad_norm": 11.846328735351562,
+      "grad_norm": 21.138622283935547,
       "learning_rate": 4.027777777777778e-05,
-      "loss": 0.3415,
+      "loss": 0.5071,
       "step": 330
     },
     {
       "epoch": 13.877551020408163,
-      "grad_norm": 9.749055862426758,
+      "grad_norm": 79.29064178466797,
       "learning_rate": 3.981481481481482e-05,
-      "loss": 0.5033,
+      "loss": 0.5575,
       "step": 340
     },
     {
       "epoch": 14.0,
       "eval_accuracy": 0.45454545454545453,
-      "eval_loss": 1.8614706993103027,
-      "eval_runtime": 4.1423,
-      "eval_samples_per_second": 5.311,
-      "eval_steps_per_second": 2.656,
+      "eval_loss": 2.1859312057495117,
+      "eval_runtime": 3.7744,
+      "eval_samples_per_second": 5.829,
+      "eval_steps_per_second": 2.914,
       "step": 343
     },
     {
       "epoch": 14.285714285714286,
-      "grad_norm": 74.4493637084961,
+      "grad_norm": 1.0631258487701416,
       "learning_rate": 3.935185185185186e-05,
-      "loss": 0.2709,
+      "loss": 0.213,
       "step": 350
     },
     {
       "epoch": 14.693877551020408,
-      "grad_norm": 34.6863899230957,
+      "grad_norm": 59.045860290527344,
       "learning_rate": 3.888888888888889e-05,
-      "loss": 0.2846,
+      "loss": 0.3894,
       "step": 360
     },
     {
       "epoch": 14.979591836734693,
-      "eval_accuracy": 0.5,
-      "eval_loss": 1.5478615760803223,
-      "eval_runtime": 3.7136,
-      "eval_samples_per_second": 5.924,
-      "eval_steps_per_second": 2.962,
+      "eval_accuracy": 0.36363636363636365,
+      "eval_loss": 2.7353034019470215,
+      "eval_runtime": 4.1695,
+      "eval_samples_per_second": 5.276,
+      "eval_steps_per_second": 2.638,
       "step": 367
     },
     {
       "epoch": 15.10204081632653,
-      "grad_norm": 21.048105239868164,
+      "grad_norm": 7.641679763793945,
       "learning_rate": 3.8425925925925924e-05,
-      "loss": 0.361,
+      "loss": 0.2927,
       "step": 370
     },
     {
       "epoch": 15.510204081632653,
-      "grad_norm": 27.45090103149414,
+      "grad_norm": 30.950389862060547,
       "learning_rate": 3.7962962962962964e-05,
-      "loss": 0.1265,
+      "loss": 0.4006,
       "step": 380
     },
     {
       "epoch": 15.918367346938776,
-      "grad_norm": 2.5599782466888428,
+      "grad_norm": 6.0692524909973145,
       "learning_rate": 3.7500000000000003e-05,
-      "loss": 0.2828,
+      "loss": 0.4326,
       "step": 390
     },
     {
       "epoch": 16.0,
-      "eval_accuracy": 0.5454545454545454,
-      "eval_loss": 1.6410338878631592,
-      "eval_runtime": 4.2352,
-      "eval_samples_per_second": 5.195,
-      "eval_steps_per_second": 2.597,
+      "eval_accuracy": 0.36363636363636365,
+      "eval_loss": 2.4454774856567383,
+      "eval_runtime": 3.8093,
+      "eval_samples_per_second": 5.775,
+      "eval_steps_per_second": 2.888,
       "step": 392
     },
     {
       "epoch": 16.3265306122449,
-      "grad_norm": 44.01292037963867,
+      "grad_norm": 37.877662658691406,
       "learning_rate": 3.7037037037037037e-05,
-      "loss": 0.4138,
+      "loss": 0.381,
       "step": 400
     },
     {
       "epoch": 16.73469387755102,
-      "grad_norm": 49.22306823730469,
+      "grad_norm": 6.306503772735596,
       "learning_rate": 3.6574074074074076e-05,
-      "loss": 0.3426,
+      "loss": 0.3715,
       "step": 410
     },
     {
       "epoch": 16.979591836734695,
-      "eval_accuracy": 0.36363636363636365,
-      "eval_loss": 1.9146943092346191,
-      "eval_runtime": 3.73,
-      "eval_samples_per_second": 5.898,
-      "eval_steps_per_second": 2.949,
+      "eval_accuracy": 0.5454545454545454,
+      "eval_loss": 2.310415744781494,
+      "eval_runtime": 3.7058,
+      "eval_samples_per_second": 5.937,
+      "eval_steps_per_second": 2.968,
       "step": 416
     },
     {
       "epoch": 17.142857142857142,
-      "grad_norm": 20.12368392944336,
+      "grad_norm": 11.820294380187988,
       "learning_rate": 3.611111111111111e-05,
-      "loss": 0.3658,
+      "loss": 0.3212,
       "step": 420
     },
     {
       "epoch": 17.551020408163264,
-      "grad_norm": 5.72607946395874,
+      "grad_norm": 49.839988708496094,
       "learning_rate": 3.564814814814815e-05,
-      "loss": 0.0983,
+      "loss": 0.2577,
       "step": 430
     },
     {
       "epoch": 17.959183673469386,
-      "grad_norm": 127.7773208618164,
+      "grad_norm": 18.69984245300293,
       "learning_rate": 3.518518518518519e-05,
-      "loss": 0.3108,
+      "loss": 0.3966,
       "step": 440
     },
     {
       "epoch": 18.0,
-      "eval_accuracy": 0.5454545454545454,
-      "eval_loss": 1.4793992042541504,
-      "eval_runtime": 4.4219,
-      "eval_samples_per_second": 4.975,
-      "eval_steps_per_second": 2.488,
+      "eval_accuracy": 0.45454545454545453,
+      "eval_loss": 2.459726333618164,
+      "eval_runtime": 3.7623,
+      "eval_samples_per_second": 5.848,
+      "eval_steps_per_second": 2.924,
       "step": 441
     },
     {
       "epoch": 18.367346938775512,
-      "grad_norm": 2.2349815368652344,
+      "grad_norm": 53.232845306396484,
       "learning_rate": 3.472222222222222e-05,
-      "loss": 0.08,
+      "loss": 0.2414,
       "step": 450
     },
     {
       "epoch": 18.775510204081634,
-      "grad_norm": 84.97185516357422,
+      "grad_norm": 17.193056106567383,
       "learning_rate": 3.425925925925926e-05,
-      "loss": 0.2129,
+      "loss": 0.1855,
       "step": 460
     },
     {
       "epoch": 18.979591836734695,
-      "eval_accuracy": 0.4090909090909091,
-      "eval_loss": 1.776501178741455,
-      "eval_runtime": 3.7003,
-      "eval_samples_per_second": 5.946,
-      "eval_steps_per_second": 2.973,
+      "eval_accuracy": 0.36363636363636365,
+      "eval_loss": 2.3334853649139404,
+      "eval_runtime": 3.7675,
+      "eval_samples_per_second": 5.839,
+      "eval_steps_per_second": 2.92,
       "step": 465
     },
     {
       "epoch": 19.183673469387756,
-      "grad_norm": 6.593123912811279,
+      "grad_norm": 83.63492584228516,
       "learning_rate": 3.3796296296296295e-05,
-      "loss": 0.1714,
+      "loss": 0.147,
       "step": 470
     },
     {
       "epoch": 19.591836734693878,
-      "grad_norm": 2.1411988735198975,
+      "grad_norm": 18.476999282836914,
       "learning_rate": 3.3333333333333335e-05,
-      "loss": 0.1612,
+      "loss": 0.1192,
       "step": 480
     },
     {
       "epoch": 20.0,
-      "grad_norm": 7.131197929382324,
+      "grad_norm": 4.202638626098633,
       "learning_rate": 3.2870370370370375e-05,
-      "loss": 0.1946,
+      "loss": 0.1528,
       "step": 490
     },
     {
       "epoch": 20.0,
-      "eval_accuracy": 0.45454545454545453,
-      "eval_loss": 2.16646671295166,
-      "eval_runtime": 4.4203,
-      "eval_samples_per_second": 4.977,
-      "eval_steps_per_second": 2.489,
+      "eval_accuracy": 0.4090909090909091,
+      "eval_loss": 2.3629674911499023,
+      "eval_runtime": 3.7957,
+      "eval_samples_per_second": 5.796,
+      "eval_steps_per_second": 2.898,
       "step": 490
     },
     {
       "epoch": 20.408163265306122,
-      "grad_norm": 4.898344039916992,
+      "grad_norm": 69.58875274658203,
       "learning_rate": 3.240740740740741e-05,
-      "loss": 0.2809,
+      "loss": 0.1692,
       "step": 500
     },
     {
       "epoch": 20.816326530612244,
-      "grad_norm": 0.8991552591323853,
+      "grad_norm": 27.37713050842285,
       "learning_rate": 3.194444444444444e-05,
-      "loss": 0.1255,
+      "loss": 0.2036,
       "step": 510
     },
     {
       "epoch": 20.979591836734695,
-      "eval_accuracy": 0.5454545454545454,
-      "eval_loss": 1.8521960973739624,
-      "eval_runtime": 4.0094,
-      "eval_samples_per_second": 5.487,
-      "eval_steps_per_second": 2.744,
+      "eval_accuracy": 0.45454545454545453,
+      "eval_loss": 2.3520021438598633,
+      "eval_runtime": 3.6999,
+      "eval_samples_per_second": 5.946,
+      "eval_steps_per_second": 2.973,
       "step": 514
     },
     {
       "epoch": 21.224489795918366,
-      "grad_norm": 0.9338099956512451,
+      "grad_norm": 47.89410400390625,
       "learning_rate": 3.148148148148148e-05,
-      "loss": 0.0503,
+      "loss": 0.2093,
       "step": 520
     },
     {
       "epoch": 21.632653061224488,
-      "grad_norm": 1.708648920059204,
+      "grad_norm": 42.28773880004883,
       "learning_rate": 3.101851851851852e-05,
-      "loss": 0.2301,
+      "loss": 0.2026,
       "step": 530
     },
     {
       "epoch": 22.0,
-      "eval_accuracy": 0.45454545454545453,
-      "eval_loss": 2.028917074203491,
-      "eval_runtime": 4.3302,
-      "eval_samples_per_second": 5.081,
-      "eval_steps_per_second": 2.54,
+      "eval_accuracy": 0.4090909090909091,
+      "eval_loss": 2.7011702060699463,
+      "eval_runtime": 4.0581,
+      "eval_samples_per_second": 5.421,
+      "eval_steps_per_second": 2.711,
       "step": 539
     },
     {
       "epoch": 22.040816326530614,
-      "grad_norm": 68.97888946533203,
+      "grad_norm": 9.840641975402832,
       "learning_rate": 3.055555555555556e-05,
-      "loss": 0.1664,
+      "loss": 0.2354,
       "step": 540
     },
     {
       "epoch": 22.448979591836736,
-      "grad_norm": 8.795487403869629,
+      "grad_norm": 292.9500732421875,
       "learning_rate": 3.0092592592592593e-05,
-      "loss": 0.0495,
+      "loss": 0.1092,
       "step": 550
     },
     {
       "epoch": 22.857142857142858,
-      "grad_norm": 1.032072901725769,
+      "grad_norm": 3.9684982299804688,
       "learning_rate": 2.962962962962963e-05,
-      "loss": 0.0909,
+      "loss": 0.2127,
       "step": 560
     },
     {
       "epoch": 22.979591836734695,
-      "eval_accuracy": 0.4090909090909091,
-      "eval_loss": 2.051244020462036,
-      "eval_runtime": 4.0425,
-      "eval_samples_per_second": 5.442,
-      "eval_steps_per_second": 2.721,
+      "eval_accuracy": 0.5909090909090909,
+      "eval_loss": 2.3724472522735596,
+      "eval_runtime": 3.7338,
+      "eval_samples_per_second": 5.892,
+      "eval_steps_per_second": 2.946,
       "step": 563
     },
     {
       "epoch": 23.26530612244898,
-      "grad_norm": 28.04718589782715,
+      "grad_norm": 33.97622299194336,
       "learning_rate": 2.916666666666667e-05,
-      "loss": 0.1032,
+      "loss": 0.287,
       "step": 570
     },
     {
       "epoch": 23.6734693877551,
-      "grad_norm": 24.228317260742188,
+      "grad_norm": 64.21463012695312,
       "learning_rate": 2.8703703703703706e-05,
-      "loss": 0.1724,
+      "loss": 0.2719,
       "step": 580
     },
     {
       "epoch": 24.0,
-      "eval_accuracy": 0.4090909090909091,
-      "eval_loss": 2.3410568237304688,
-      "eval_runtime": 4.2445,
-      "eval_samples_per_second": 5.183,
-      "eval_steps_per_second": 2.592,
+      "eval_accuracy": 0.3181818181818182,
+      "eval_loss": 3.0376410484313965,
+      "eval_runtime": 3.8252,
+      "eval_samples_per_second": 5.751,
+      "eval_steps_per_second": 2.876,
       "step": 588
     },
     {
       "epoch": 24.081632653061224,
-      "grad_norm": 51.32829284667969,
+      "grad_norm": 18.13422203063965,
       "learning_rate": 2.824074074074074e-05,
-      "loss": 0.1376,
+      "loss": 0.1286,
       "step": 590
     },
     {
       "epoch": 24.489795918367346,
-      "grad_norm": 8.50900936126709,
+      "grad_norm": 1.6358630657196045,
       "learning_rate": 2.777777777777778e-05,
-      "loss": 0.0941,
+      "loss": 0.0524,
       "step": 600
     },
     {
       "epoch": 24.897959183673468,
-      "grad_norm": 53.41023254394531,
+      "grad_norm": 2.2687888145446777,
       "learning_rate": 2.7314814814814816e-05,
-      "loss": 0.2256,
+      "loss": 0.1292,
       "step": 610
     },
     {
       "epoch": 24.979591836734695,
       "eval_accuracy": 0.5,
-      "eval_loss": 2.162550687789917,
-      "eval_runtime": 3.7513,
-      "eval_samples_per_second": 5.865,
-      "eval_steps_per_second": 2.932,
+      "eval_loss": 2.568352699279785,
+      "eval_runtime": 3.7655,
+      "eval_samples_per_second": 5.842,
+      "eval_steps_per_second": 2.921,
       "step": 612
     },
     {
       "epoch": 25.306122448979593,
-      "grad_norm": 20.35555076599121,
+      "grad_norm": 0.1713099479675293,
       "learning_rate": 2.6851851851851855e-05,
-      "loss": 0.2228,
+      "loss": 0.2063,
       "step": 620
     },
     {
       "epoch": 25.714285714285715,
-      "grad_norm": 5.79302978515625,
+      "grad_norm": 0.19898605346679688,
       "learning_rate": 2.6388888888888892e-05,
-      "loss": 0.2471,
+      "loss": 0.2533,
       "step": 630
     },
     {
       "epoch": 26.0,
       "eval_accuracy": 0.4090909090909091,
-      "eval_loss": 2.0551793575286865,
-      "eval_runtime": 4.2279,
-      "eval_samples_per_second": 5.204,
-      "eval_steps_per_second": 2.602,
+      "eval_loss": 2.6974411010742188,
+      "eval_runtime": 4.1166,
+      "eval_samples_per_second": 5.344,
+      "eval_steps_per_second": 2.672,
       "step": 637
     },
     {
       "epoch": 26.122448979591837,
-      "grad_norm": 57.79982376098633,
+      "grad_norm": 10.351354598999023,
       "learning_rate": 2.5925925925925925e-05,
-      "loss": 0.1514,
+      "loss": 0.0784,
       "step": 640
     },
     {
       "epoch": 26.53061224489796,
-      "grad_norm": 9.047138214111328,
+      "grad_norm": 53.10356903076172,
       "learning_rate": 2.5462962962962965e-05,
-      "loss": 0.056,
+      "loss": 0.102,
       "step": 650
     },
     {
       "epoch": 26.93877551020408,
-      "grad_norm": 8.557161331176758,
+      "grad_norm": 34.5041389465332,
       "learning_rate": 2.5e-05,
-      "loss": 0.0671,
+      "loss": 0.1947,
       "step": 660
     },
     {
       "epoch": 26.979591836734695,
-      "eval_accuracy": 0.5454545454545454,
-      "eval_loss": 1.9339114427566528,
-      "eval_runtime": 3.8162,
-      "eval_samples_per_second": 5.765,
-      "eval_steps_per_second": 2.882,
+      "eval_accuracy": 0.4090909090909091,
+      "eval_loss": 2.695702075958252,
+      "eval_runtime": 3.6793,
+      "eval_samples_per_second": 5.979,
+      "eval_steps_per_second": 2.99,
       "step": 661
     },
     {
       "epoch": 27.346938775510203,
-      "grad_norm": 0.14792662858963013,
+      "grad_norm": 0.4286749064922333,
       "learning_rate": 2.4537037037037038e-05,
-      "loss": 0.1409,
+      "loss": 0.1586,
       "step": 670
     },
     {
       "epoch": 27.755102040816325,
-      "grad_norm": 112.82161712646484,
+      "grad_norm": 1.2722495794296265,
       "learning_rate": 2.4074074074074074e-05,
-      "loss": 0.2563,
+      "loss": 0.1805,
       "step": 680
     },
     {
       "epoch": 28.0,
-      "eval_accuracy": 0.45454545454545453,
-      "eval_loss": 2.2506721019744873,
-      "eval_runtime": 4.1968,
-      "eval_samples_per_second": 5.242,
-      "eval_steps_per_second": 2.621,
+      "eval_accuracy": 0.4090909090909091,
+      "eval_loss": 2.8953325748443604,
+      "eval_runtime": 3.8043,
+      "eval_samples_per_second": 5.783,
+      "eval_steps_per_second": 2.891,
       "step": 686
     },
     {
       "epoch": 28.163265306122447,
-      "grad_norm": 0.717983067035675,
+      "grad_norm": 9.904804229736328,
       "learning_rate": 2.361111111111111e-05,
-      "loss": 0.0499,
+      "loss": 0.3122,
       "step": 690
     },
     {
       "epoch": 28.571428571428573,
-      "grad_norm": 47.17780303955078,
+      "grad_norm": 34.21662521362305,
       "learning_rate": 2.314814814814815e-05,
-      "loss": 0.1607,
+      "loss": 0.0842,
       "step": 700
     },
     {
       "epoch": 28.979591836734695,
-      "grad_norm": 102.77259063720703,
+      "grad_norm": 0.2031356394290924,
       "learning_rate": 2.2685185185185187e-05,
-      "loss": 0.1865,
+      "loss": 0.1123,
       "step": 710
     },
     {
       "epoch": 28.979591836734695,
-      "eval_accuracy": 0.5,
-      "eval_loss": 2.0703585147857666,
-      "eval_runtime": 3.7043,
-      "eval_samples_per_second": 5.939,
-      "eval_steps_per_second": 2.969,
+      "eval_accuracy": 0.4090909090909091,
+      "eval_loss": 2.824019432067871,
+      "eval_runtime": 4.2297,
+      "eval_samples_per_second": 5.201,
+      "eval_steps_per_second": 2.601,
       "step": 710
     },
     {
       "epoch": 29.387755102040817,
-      "grad_norm": 0.42554885149002075,
+      "grad_norm": 0.74458909034729,
       "learning_rate": 2.2222222222222223e-05,
-      "loss": 0.1905,
+      "loss": 0.1034,
       "step": 720
     },
     {
       "epoch": 29.79591836734694,
-      "grad_norm": 92.93157196044922,
+      "grad_norm": 52.627017974853516,
       "learning_rate": 2.175925925925926e-05,
-      "loss": 0.0477,
+      "loss": 0.2143,
       "step": 730
     },
     {
       "epoch": 30.0,
-      "eval_accuracy": 0.3181818181818182,
-      "eval_loss": 2.839505195617676,
-      "eval_runtime": 4.1643,
-      "eval_samples_per_second": 5.283,
-      "eval_steps_per_second": 2.642,
+      "eval_accuracy": 0.45454545454545453,
+      "eval_loss": 2.3879992961883545,
+      "eval_runtime": 3.7999,
+      "eval_samples_per_second": 5.79,
+      "eval_steps_per_second": 2.895,
       "step": 735
     },
     {
       "epoch": 30.20408163265306,
-      "grad_norm": 10.255499839782715,
+      "grad_norm": 18.753475189208984,
       "learning_rate": 2.1296296296296296e-05,
-      "loss": 0.2298,
+      "loss": 0.1084,
       "step": 740
     },
     {
       "epoch": 30.612244897959183,
-      "grad_norm": 32.9160041809082,
+      "grad_norm": 36.20948028564453,
       "learning_rate": 2.0833333333333336e-05,
-      "loss": 0.0931,
+      "loss": 0.1845,
       "step": 750
     },
     {
       "epoch": 30.979591836734695,
       "eval_accuracy": 0.36363636363636365,
-      "eval_loss": 2.9483964443206787,
-      "eval_runtime": 3.6962,
-      "eval_samples_per_second": 5.952,
-      "eval_steps_per_second": 2.976,
+      "eval_loss": 2.607203483581543,
+      "eval_runtime": 3.7239,
+      "eval_samples_per_second": 5.908,
+      "eval_steps_per_second": 2.954,
       "step": 759
     },
     {
       "epoch": 31.020408163265305,
-      "grad_norm": 3.2979772090911865,
+      "grad_norm": 125.52183532714844,
       "learning_rate": 2.037037037037037e-05,
-      "loss": 0.0643,
+      "loss": 0.1467,
       "step": 760
     },
     {
       "epoch": 31.428571428571427,
-      "grad_norm": 1.9711153507232666,
+      "grad_norm": 0.5813603401184082,
       "learning_rate": 1.990740740740741e-05,
-      "loss": 0.1815,
+      "loss": 0.073,
       "step": 770
     },
     {
       "epoch": 31.836734693877553,
-      "grad_norm": 39.575653076171875,
+      "grad_norm": 29.528520584106445,
       "learning_rate": 1.9444444444444445e-05,
-      "loss": 0.047,
+      "loss": 0.0921,
       "step": 780
     },
     {
       "epoch": 32.0,
       "eval_accuracy": 0.45454545454545453,
-      "eval_loss": 2.548579216003418,
-      "eval_runtime": 4.2179,
-      "eval_samples_per_second": 5.216,
-      "eval_steps_per_second": 2.608,
+      "eval_loss": 2.7255632877349854,
+      "eval_runtime": 3.7809,
+      "eval_samples_per_second": 5.819,
+      "eval_steps_per_second": 2.909,
       "step": 784
     },
     {
       "epoch": 32.244897959183675,
-      "grad_norm": 0.1791549175977707,
+      "grad_norm": 23.8216495513916,
       "learning_rate": 1.8981481481481482e-05,
-      "loss": 0.1034,
+      "loss": 0.0362,
       "step": 790
     },
     {
       "epoch": 32.6530612244898,
-      "grad_norm": 18.35671043395996,
+      "grad_norm": 8.946208000183105,
       "learning_rate": 1.8518518518518518e-05,
-      "loss": 0.165,
+      "loss": 0.0276,
       "step": 800
     },
     {
       "epoch": 32.97959183673469,
-      "eval_accuracy": 0.45454545454545453,
-      "eval_loss": 2.601107120513916,
-      "eval_runtime": 3.894,
-      "eval_samples_per_second": 5.65,
-      "eval_steps_per_second": 2.825,
+      "eval_accuracy": 0.4090909090909091,
+      "eval_loss": 2.4073522090911865,
+      "eval_runtime": 3.7423,
+      "eval_samples_per_second": 5.879,
+      "eval_steps_per_second": 2.939,
       "step": 808
     },
     {
       "epoch": 33.06122448979592,
-      "grad_norm": 88.95061492919922,
+      "grad_norm": 11.886552810668945,
       "learning_rate": 1.8055555555555555e-05,
-      "loss": 0.0639,
+      "loss": 0.0145,
       "step": 810
     },
     {
       "epoch": 33.46938775510204,
-      "grad_norm": 0.11740878969430923,
+      "grad_norm": 2.590672492980957,
       "learning_rate": 1.7592592592592595e-05,
-      "loss": 0.032,
+      "loss": 0.0735,
       "step": 820
     },
     {
       "epoch": 33.87755102040816,
-      "grad_norm": 8.803248405456543,
+      "grad_norm": 2.8415122032165527,
       "learning_rate": 1.712962962962963e-05,
-      "loss": 0.0203,
+      "loss": 0.0876,
       "step": 830
     },
     {
       "epoch": 34.0,
-      "eval_accuracy": 0.5,
-      "eval_loss": 2.3598248958587646,
-      "eval_runtime": 4.1023,
-      "eval_samples_per_second": 5.363,
-      "eval_steps_per_second": 2.681,
+      "eval_accuracy": 0.45454545454545453,
+      "eval_loss": 2.6042609214782715,
+      "eval_runtime": 3.8508,
+      "eval_samples_per_second": 5.713,
+      "eval_steps_per_second": 2.857,
       "step": 833
     },
     {
       "epoch": 34.285714285714285,
-      "grad_norm": 3.050877571105957,
+      "grad_norm": 0.48890063166618347,
       "learning_rate": 1.6666666666666667e-05,
-      "loss": 0.0173,
+      "loss": 0.0343,
       "step": 840
     },
     {
       "epoch": 34.69387755102041,
-      "grad_norm": 10.551043510437012,
+      "grad_norm": 5.784631252288818,
       "learning_rate": 1.6203703703703704e-05,
-      "loss": 0.0143,
+      "loss": 0.0253,
       "step": 850
     },
     {
       "epoch": 34.97959183673469,
-      "eval_accuracy": 0.4090909090909091,
-      "eval_loss": 2.589207649230957,
-      "eval_runtime": 4.2169,
-      "eval_samples_per_second": 5.217,
-      "eval_steps_per_second": 2.609,
+      "eval_accuracy": 0.45454545454545453,
+      "eval_loss": 2.76198673248291,
+      "eval_runtime": 3.6773,
+      "eval_samples_per_second": 5.983,
+      "eval_steps_per_second": 2.991,
       "step": 857
     },
     {
       "epoch": 35.10204081632653,
-      "grad_norm": 2.492408514022827,
+      "grad_norm": 3.974219560623169,
       "learning_rate": 1.574074074074074e-05,
-      "loss": 0.1669,
+      "loss": 0.1737,
       "step": 860
     },
     {
       "epoch": 35.51020408163265,
-      "grad_norm": 28.920665740966797,
+      "grad_norm": 0.8458910584449768,
       "learning_rate": 1.527777777777778e-05,
-      "loss": 0.0546,
+      "loss": 0.1227,
       "step": 870
     },
     {
       "epoch": 35.91836734693877,
-      "grad_norm": 0.1343473196029663,
+      "grad_norm": 49.0916748046875,
       "learning_rate": 1.4814814814814815e-05,
-      "loss": 0.0248,
+      "loss": 0.1904,
       "step": 880
     },
     {
       "epoch": 36.0,
       "eval_accuracy": 0.4090909090909091,
-      "eval_loss": 2.8362252712249756,
-      "eval_runtime": 4.252,
-      "eval_samples_per_second": 5.174,
-      "eval_steps_per_second": 2.587,
+      "eval_loss": 2.6911137104034424,
+      "eval_runtime": 4.2847,
+      "eval_samples_per_second": 5.135,
+      "eval_steps_per_second": 2.567,
       "step": 882
     },
     {
       "epoch": 36.326530612244895,
-      "grad_norm": 0.00863693282008171,
+      "grad_norm": 33.6256103515625,
       "learning_rate": 1.4351851851851853e-05,
-      "loss": 0.0432,
+      "loss": 0.2075,
       "step": 890
     },
     {
       "epoch": 36.734693877551024,
-      "grad_norm": 1.786009430885315,
+      "grad_norm": 0.20336776971817017,
       "learning_rate": 1.388888888888889e-05,
-      "loss": 0.0812,
+      "loss": 0.072,
       "step": 900
     },
     {
       "epoch": 36.97959183673469,
-      "eval_accuracy": 0.4090909090909091,
-      "eval_loss": 2.4658091068267822,
-      "eval_runtime": 4.0409,
-      "eval_samples_per_second": 5.444,
-      "eval_steps_per_second": 2.722,
+      "eval_accuracy": 0.45454545454545453,
+      "eval_loss": 2.6528260707855225,
+      "eval_runtime": 3.7052,
+      "eval_samples_per_second": 5.938,
+      "eval_steps_per_second": 2.969,
       "step": 906
     },
     {
       "epoch": 37.142857142857146,
-      "grad_norm": 5.971357822418213,
+      "grad_norm": 18.30384063720703,
       "learning_rate": 1.3425925925925928e-05,
-      "loss": 0.1312,
+      "loss": 0.1398,
       "step": 910
     },
     {
       "epoch": 37.55102040816327,
-      "grad_norm": 9.33790111541748,
+      "grad_norm": 0.07602863758802414,
       "learning_rate": 1.2962962962962962e-05,
-      "loss": 0.0453,
+      "loss": 0.069,
       "step": 920
     },
     {
       "epoch": 37.95918367346939,
-      "grad_norm": 13.613574028015137,
+      "grad_norm": 8.712791442871094,
       "learning_rate": 1.25e-05,
-      "loss": 0.0662,
+      "loss": 0.169,
       "step": 930
     },
     {
       "epoch": 38.0,
-      "eval_accuracy": 0.4090909090909091,
-      "eval_loss": 2.640266180038452,
-      "eval_runtime": 4.0647,
-      "eval_samples_per_second": 5.412,
-      "eval_steps_per_second": 2.706,
+      "eval_accuracy": 0.45454545454545453,
+      "eval_loss": 2.645385265350342,
+      "eval_runtime": 3.7757,
+      "eval_samples_per_second": 5.827,
+      "eval_steps_per_second": 2.913,
       "step": 931
     },
     {
       "epoch": 38.36734693877551,
-      "grad_norm": 17.910255432128906,
+      "grad_norm": 0.9796821475028992,
       "learning_rate": 1.2037037037037037e-05,
-      "loss": 0.1061,
+      "loss": 0.0137,
       "step": 940
     },
     {
       "epoch": 38.775510204081634,
-      "grad_norm": 5.941530227661133,
+      "grad_norm": 0.2808334529399872,
       "learning_rate": 1.1574074074074075e-05,
-      "loss": 0.1855,
+      "loss": 0.0978,
       "step": 950
     },
     {
       "epoch": 38.97959183673469,
-      "eval_accuracy": 0.45454545454545453,
-      "eval_loss": 2.604184150695801,
-      "eval_runtime": 3.8038,
-      "eval_samples_per_second": 5.784,
-      "eval_steps_per_second": 2.892,
+      "eval_accuracy": 0.5,
+      "eval_loss": 2.626924514770508,
+      "eval_runtime": 3.7403,
+      "eval_samples_per_second": 5.882,
+      "eval_steps_per_second": 2.941,
       "step": 955
     },
     {
       "epoch": 39.183673469387756,
-      "grad_norm": 0.057647667825222015,
+      "grad_norm": 0.008785932324826717,
       "learning_rate": 1.1111111111111112e-05,
-      "loss": 0.0598,
+      "loss": 0.0195,
       "step": 960
     },
     {
       "epoch": 39.59183673469388,
-      "grad_norm": 1.981195092201233,
+      "grad_norm": 0.38346049189567566,
       "learning_rate": 1.0648148148148148e-05,
-      "loss": 0.0107,
+      "loss": 0.0283,
       "step": 970
     },
     {
       "epoch": 40.0,
-      "grad_norm": 74.54513549804688,
+      "grad_norm": 91.21365356445312,
       "learning_rate": 1.0185185185185185e-05,
-      "loss": 0.03,
+      "loss": 0.069,
       "step": 980
     },
     {
       "epoch": 40.0,
-      "eval_accuracy": 0.5,
-      "eval_loss": 2.5595407485961914,
-      "eval_runtime": 4.2131,
-      "eval_samples_per_second": 5.222,
-      "eval_steps_per_second": 2.611,
+      "eval_accuracy": 0.45454545454545453,
+      "eval_loss": 2.4153661727905273,
+      "eval_runtime": 3.8277,
+      "eval_samples_per_second": 5.748,
+      "eval_steps_per_second": 2.874,
       "step": 980
     },
     {
       "epoch": 40.40816326530612,
-      "grad_norm": 3.7874374389648438,
+      "grad_norm": 50.51347351074219,
       "learning_rate": 9.722222222222223e-06,
-      "loss": 0.0882,
+      "loss": 0.1111,
       "step": 990
     },
     {
       "epoch": 40.816326530612244,
-      "grad_norm": 0.011240988969802856,
+      "grad_norm": 0.06127191707491875,
       "learning_rate": 9.259259259259259e-06,
-      "loss": 0.1117,
+      "loss": 0.0159,
       "step": 1000
     },
     {
       "epoch": 40.97959183673469,
-      "eval_accuracy": 0.5,
-      "eval_loss": 2.4715521335601807,
-      "eval_runtime": 3.7805,
-      "eval_samples_per_second": 5.819,
-      "eval_steps_per_second": 2.91,
+      "eval_accuracy": 0.45454545454545453,
+      "eval_loss": 2.7026443481445312,
+      "eval_runtime": 3.6976,
+      "eval_samples_per_second": 5.95,
+      "eval_steps_per_second": 2.975,
       "step": 1004
     },
     {
       "epoch": 41.224489795918366,
-      "grad_norm": 30.458051681518555,
+      "grad_norm": 0.4312690794467926,
       "learning_rate": 8.796296296296297e-06,
-      "loss": 0.0819,
+      "loss": 0.1659,
       "step": 1010
     },
     {
       "epoch": 41.63265306122449,
-      "grad_norm": 0.2596540153026581,
+      "grad_norm": 0.1739516258239746,
       "learning_rate": 8.333333333333334e-06,
-      "loss": 0.0466,
+      "loss": 0.2046,
       "step": 1020
     },
     {
       "epoch": 42.0,
-      "eval_accuracy": 0.5,
-      "eval_loss": 2.325749397277832,
-      "eval_runtime": 4.0793,
-      "eval_samples_per_second": 5.393,
-      "eval_steps_per_second": 2.697,
+      "eval_accuracy": 0.45454545454545453,
+      "eval_loss": 2.521285057067871,
+      "eval_runtime": 4.0417,
+      "eval_samples_per_second": 5.443,
+      "eval_steps_per_second": 2.722,
       "step": 1029
     },
     {
       "epoch": 42.04081632653061,
-      "grad_norm": 0.8128257393836975,
+      "grad_norm": 0.023499183356761932,
       "learning_rate": 7.87037037037037e-06,
-      "loss": 0.2931,
+      "loss": 0.0502,
       "step": 1030
     },
     {
       "epoch": 42.44897959183673,
-      "grad_norm": 1.8079004287719727,
+      "grad_norm": 8.13757038116455,
       "learning_rate": 7.4074074074074075e-06,
-      "loss": 0.0667,
+      "loss": 0.1637,
       "step": 1040
     },
     {
       "epoch": 42.857142857142854,
-      "grad_norm": 13.098880767822266,
+      "grad_norm": 0.8705015778541565,
       "learning_rate": 6.944444444444445e-06,
-      "loss": 0.1349,
+      "loss": 0.0329,
       "step": 1050
     },
     {
       "epoch": 42.97959183673469,
-      "eval_accuracy": 0.45454545454545453,
-      "eval_loss": 2.5544846057891846,
-      "eval_runtime": 3.7821,
-      "eval_samples_per_second": 5.817,
-      "eval_steps_per_second": 2.908,
+      "eval_accuracy": 0.5,
+      "eval_loss": 2.639939069747925,
+      "eval_runtime": 4.1783,
+      "eval_samples_per_second": 5.265,
+      "eval_steps_per_second": 2.633,
       "step": 1053
     },
     {
       "epoch": 43.265306122448976,
-      "grad_norm": 0.03145942464470863,
+      "grad_norm": 0.7091279029846191,
       "learning_rate": 6.481481481481481e-06,
-      "loss": 0.0184,
+      "loss": 0.008,
       "step": 1060
     },
     {
       "epoch": 43.673469387755105,
-      "grad_norm": 0.008914113976061344,
+      "grad_norm": 0.27460241317749023,
       "learning_rate": 6.0185185185185185e-06,
-      "loss": 0.0069,
+      "loss": 0.0166,
       "step": 1070
     },
     {
       "epoch": 44.0,
-      "eval_accuracy": 0.5,
-      "eval_loss": 2.5814802646636963,
-      "eval_runtime": 4.2458,
-      "eval_samples_per_second": 5.182,
-      "eval_steps_per_second": 2.591,
+      "eval_accuracy": 0.45454545454545453,
+      "eval_loss": 2.778676986694336,
+      "eval_runtime": 3.8444,
+      "eval_samples_per_second": 5.723,
+      "eval_steps_per_second": 2.861,
       "step": 1078
     },
     {
       "epoch": 44.08163265306123,
-      "grad_norm": 0.0782785564661026,
+      "grad_norm": 9.933192253112793,
       "learning_rate": 5.555555555555556e-06,
-      "loss": 0.1017,
+      "loss": 0.0275,
       "step": 1080
     },
     {
       "epoch": 44.48979591836735,
-      "grad_norm": 0.24598410725593567,
+      "grad_norm": 0.6145612001419067,
       "learning_rate": 5.092592592592592e-06,
-      "loss": 0.0344,
+      "loss": 0.0086,
       "step": 1090
     },
     {
       "epoch": 44.89795918367347,
-      "grad_norm": 0.017518645152449608,
+      "grad_norm": 0.030872033908963203,
       "learning_rate": 4.6296296296296296e-06,
-      "loss": 0.0468,
+      "loss": 0.0812,
       "step": 1100
     },
     {
       "epoch": 44.97959183673469,
-      "eval_accuracy": 0.5,
-      "eval_loss": 2.3666675090789795,
-      "eval_runtime": 3.9709,
-      "eval_samples_per_second": 5.54,
-      "eval_steps_per_second": 2.77,
+      "eval_accuracy": 0.45454545454545453,
+      "eval_loss": 2.8176286220550537,
+      "eval_runtime": 3.7182,
+      "eval_samples_per_second": 5.917,
+      "eval_steps_per_second": 2.958,
       "step": 1102
     },
     {
       "epoch": 45.30612244897959,
-      "grad_norm": 3.6144909858703613,
+      "grad_norm": 0.21678821742534637,
       "learning_rate": 4.166666666666667e-06,
-      "loss": 0.0372,
+      "loss": 0.0005,
       "step": 1110
     },
     {
       "epoch": 45.714285714285715,
-      "grad_norm": 0.018579425290226936,
+      "grad_norm": 1.7509175539016724,
       "learning_rate": 3.7037037037037037e-06,
-      "loss": 0.1807,
+      "loss": 0.0197,
       "step": 1120
     },
     {
       "epoch": 46.0,
-      "eval_accuracy": 0.5,
-      "eval_loss": 2.472862482070923,
-      "eval_runtime": 4.5348,
-      "eval_samples_per_second": 4.851,
-      "eval_steps_per_second": 2.426,
+      "eval_accuracy": 0.45454545454545453,
+      "eval_loss": 2.804949998855591,
+      "eval_runtime": 3.8321,
+      "eval_samples_per_second": 5.741,
+      "eval_steps_per_second": 2.87,
       "step": 1127
     },
     {
       "epoch": 46.12244897959184,
-      "grad_norm": 6.9941864013671875,
+      "grad_norm": 0.18609049916267395,
       "learning_rate": 3.2407407407407406e-06,
-      "loss": 0.0091,
+      "loss": 0.2891,
       "step": 1130
     },
     {
       "epoch": 46.53061224489796,
-      "grad_norm": 1.61014986038208,
+      "grad_norm": 0.7729312777519226,
       "learning_rate": 2.777777777777778e-06,
-      "loss": 0.0388,
+      "loss": 0.0178,
       "step": 1140
     },
     {
       "epoch": 46.93877551020408,
-      "grad_norm": 0.021990543231368065,
+      "grad_norm": 1.1792634725570679,
       "learning_rate": 2.3148148148148148e-06,
-      "loss": 0.0667,
+      "loss": 0.0989,
       "step": 1150
     },
     {
       "epoch": 46.97959183673469,
-      "eval_accuracy": 0.5,
-      "eval_loss": 2.496938467025757,
-      "eval_runtime": 3.9012,
-      "eval_samples_per_second": 5.639,
-      "eval_steps_per_second": 2.82,
+      "eval_accuracy": 0.45454545454545453,
+      "eval_loss": 2.7478928565979004,
+      "eval_runtime": 3.7127,
+      "eval_samples_per_second": 5.926,
+      "eval_steps_per_second": 2.963,
       "step": 1151
     },
     {
       "epoch": 47.3469387755102,
-      "grad_norm": 0.5512164235115051,
+      "grad_norm": 0.029640546068549156,
       "learning_rate": 1.8518518518518519e-06,
-      "loss": 0.0191,
+      "loss": 0.0158,
       "step": 1160
     },
     {
       "epoch": 47.755102040816325,
-      "grad_norm": 0.2976542115211487,
+      "grad_norm": 0.2686777412891388,
       "learning_rate": 1.388888888888889e-06,
-      "loss": 0.0199,
+      "loss": 0.054,
       "step": 1170
     },
     {
       "epoch": 48.0,
-      "eval_accuracy": 0.5,
-      "eval_loss": 2.5520730018615723,
-      "eval_runtime": 4.4614,
-      "eval_samples_per_second": 4.931,
-      "eval_steps_per_second": 2.466,
+      "eval_accuracy": 0.45454545454545453,
+      "eval_loss": 2.761411190032959,
+      "eval_runtime": 3.8702,
+      "eval_samples_per_second": 5.684,
+      "eval_steps_per_second": 2.842,
       "step": 1176
     },
     {
       "epoch": 48.16326530612245,
-      "grad_norm": 0.37567588686943054,
+      "grad_norm": 0.17129023373126984,
       "learning_rate": 9.259259259259259e-07,
-      "loss": 0.0806,
+      "loss": 0.0709,
       "step": 1180
     },
     {
       "epoch": 48.57142857142857,
-      "grad_norm": 2.0820977687835693,
+      "grad_norm": 0.13873820006847382,
       "learning_rate": 4.6296296296296297e-07,
-      "loss": 0.1011,
+      "loss": 0.0133,
       "step": 1190
     },
     {
       "epoch": 48.97959183673469,
-      "grad_norm": 0.018370352685451508,
+      "grad_norm": 0.48775115609169006,
       "learning_rate": 0.0,
-      "loss": 0.2716,
+      "loss": 0.1095,
       "step": 1200
     },
     {
       "epoch": 48.97959183673469,
       "eval_accuracy": 0.5,
-      "eval_loss": 2.5371408462524414,
-      "eval_runtime": 3.8373,
-      "eval_samples_per_second": 5.733,
-      "eval_steps_per_second": 2.867,
+      "eval_loss": 2.7604033946990967,
+      "eval_runtime": 3.7642,
+      "eval_samples_per_second": 5.845,
+      "eval_steps_per_second": 2.922,
       "step": 1200
     },
     {
       "epoch": 48.97959183673469,
       "step": 1200,
       "total_flos": 2.374708462608384e+17,
-      "train_loss": 0.41171714147552846,
-      "train_runtime": 1608.8411,
-      "train_samples_per_second": 6.06,
-      "train_steps_per_second": 0.746
+      "train_loss": 0.4113796023776134,
+      "train_runtime": 1728.9004,
+      "train_samples_per_second": 5.639,
+      "train_steps_per_second": 0.694
     }
   ],
   "logging_steps": 10,