End of training

README.md

@@ -22,7 +22,7 @@ model-index:
     metrics:
     - name: Accuracy
       type: accuracy
-      value: 0.9953125
+      value: 0.996875
 ---
 
 <!-- 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 [facebook/vit-msn-small](https://huggingface.co/facebook/vit-msn-small) on the imagefolder dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.0131
-- Accuracy: 0.9953
+- Loss: 0.0160
+- Accuracy: 0.9969
 
 ## Model description
 

all_results.json

@@ -1,8 +1,13 @@
 {
-    "epoch": 9.777777777777779,
-    "total_flos": 1.102133137023959e+18,
-    "train_loss": 0.670017595724626,
-    "train_runtime": 696.1068,
-    "train_samples_per_second": 82.746,
-    "train_steps_per_second": 0.316
+    "epoch": 48.888888888888886,
+    "eval_accuracy": 0.996875,
+    "eval_loss": 0.015964530408382416,
+    "eval_runtime": 3.6348,
+    "eval_samples_per_second": 176.076,
+    "eval_steps_per_second": 2.751,
+    "total_flos": 5.510665685119795e+18,
+    "train_loss": 0.17931611462072894,
+    "train_runtime": 3528.3547,
+    "train_samples_per_second": 81.624,
+    "train_steps_per_second": 0.312
 }

eval_results.json

@@ -0,0 +1,8 @@
+{
+    "epoch": 48.888888888888886,
+    "eval_accuracy": 0.996875,
+    "eval_loss": 0.015964530408382416,
+    "eval_runtime": 3.6348,
+    "eval_samples_per_second": 176.076,
+    "eval_steps_per_second": 2.751
+}

runs/Apr26_13-20-29_1c59b5307e97/events.out.tfevents.1714141191.1c59b5307e97.453.6

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:14bc85df55fb09d6cf9f5c72d186133246ae0a3d0b6f858ad7af7a1e16c68889
+size 411

train_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 9.777777777777779,
-    "total_flos": 1.102133137023959e+18,
-    "train_loss": 0.670017595724626,
-    "train_runtime": 696.1068,
-    "train_samples_per_second": 82.746,
-    "train_steps_per_second": 0.316
+    "epoch": 48.888888888888886,
+    "total_flos": 5.510665685119795e+18,
+    "train_loss": 0.17931611462072894,
+    "train_runtime": 3528.3547,
+    "train_samples_per_second": 81.624,
+    "train_steps_per_second": 0.312
 }

trainer_state.json

@@ -1,273 +1,1240 @@
 {
-  "best_metric": 0.8625,
-  "best_model_checkpoint": "vit-msn-small-finetuned-alzheimers/checkpoint-220",
-  "epoch": 9.777777777777779,
+  "best_metric": 0.996875,
+  "best_model_checkpoint": "vit-msn-small-finetuned-alzheimers/checkpoint-765",
+  "epoch": 48.888888888888886,
   "eval_steps": 500,
-  "global_step": 220,
+  "global_step": 1100,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
       "epoch": 0.4444444444444444,
-      "grad_norm": 4.161161422729492,
-      "learning_rate": 2.272727272727273e-05,
-      "loss": 1.0688,
+      "grad_norm": 8.449820518493652,
+      "learning_rate": 4.5454545454545455e-06,
+      "loss": 0.2587,
       "step": 10
     },
     {
       "epoch": 0.8888888888888888,
-      "grad_norm": 19.670530319213867,
-      "learning_rate": 4.545454545454546e-05,
-      "loss": 0.9297,
+      "grad_norm": 13.680850982666016,
+      "learning_rate": 9.090909090909091e-06,
+      "loss": 0.2996,
       "step": 20
     },
     {
       "epoch": 0.9777777777777777,
-      "eval_accuracy": 0.615625,
-      "eval_loss": 0.8769256472587585,
-      "eval_runtime": 3.9355,
-      "eval_samples_per_second": 162.623,
-      "eval_steps_per_second": 2.541,
+      "eval_accuracy": 0.84375,
+      "eval_loss": 0.38971763849258423,
+      "eval_runtime": 3.5179,
+      "eval_samples_per_second": 181.926,
+      "eval_steps_per_second": 2.843,
       "step": 22
     },
     {
       "epoch": 1.3333333333333333,
-      "grad_norm": 7.473578929901123,
-      "learning_rate": 4.797979797979798e-05,
-      "loss": 0.8854,
+      "grad_norm": 9.488574981689453,
+      "learning_rate": 1.3636363636363637e-05,
+      "loss": 0.4023,
       "step": 30
     },
     {
       "epoch": 1.7777777777777777,
-      "grad_norm": 7.05064058303833,
-      "learning_rate": 4.545454545454546e-05,
-      "loss": 0.8601,
+      "grad_norm": 18.977561950683594,
+      "learning_rate": 1.8181818181818182e-05,
+      "loss": 0.3703,
       "step": 40
     },
     {
       "epoch": 2.0,
-      "eval_accuracy": 0.634375,
-      "eval_loss": 0.7798857688903809,
-      "eval_runtime": 3.5552,
-      "eval_samples_per_second": 180.019,
-      "eval_steps_per_second": 2.813,
+      "eval_accuracy": 0.859375,
+      "eval_loss": 0.3594878911972046,
+      "eval_runtime": 3.9024,
+      "eval_samples_per_second": 164.001,
+      "eval_steps_per_second": 2.563,
       "step": 45
     },
     {
       "epoch": 2.2222222222222223,
-      "grad_norm": 12.152384757995605,
-      "learning_rate": 4.292929292929293e-05,
-      "loss": 0.8571,
+      "grad_norm": 11.33133602142334,
+      "learning_rate": 2.272727272727273e-05,
+      "loss": 0.3541,
       "step": 50
     },
     {
       "epoch": 2.6666666666666665,
-      "grad_norm": 11.653566360473633,
-      "learning_rate": 4.0404040404040405e-05,
-      "loss": 0.7954,
+      "grad_norm": 16.366662979125977,
+      "learning_rate": 2.7272727272727273e-05,
+      "loss": 0.3087,
       "step": 60
     },
     {
       "epoch": 2.977777777777778,
-      "eval_accuracy": 0.6828125,
-      "eval_loss": 0.7196512222290039,
-      "eval_runtime": 3.5735,
-      "eval_samples_per_second": 179.098,
-      "eval_steps_per_second": 2.798,
+      "eval_accuracy": 0.8625,
+      "eval_loss": 0.3777163326740265,
+      "eval_runtime": 3.8599,
+      "eval_samples_per_second": 165.808,
+      "eval_steps_per_second": 2.591,
       "step": 67
     },
     {
       "epoch": 3.111111111111111,
-      "grad_norm": 10.308629989624023,
-      "learning_rate": 3.787878787878788e-05,
-      "loss": 0.7808,
+      "grad_norm": 18.307331085205078,
+      "learning_rate": 3.181818181818182e-05,
+      "loss": 0.3195,
       "step": 70
     },
     {
       "epoch": 3.5555555555555554,
-      "grad_norm": 15.87501335144043,
-      "learning_rate": 3.535353535353535e-05,
-      "loss": 0.7552,
+      "grad_norm": 42.80950164794922,
+      "learning_rate": 3.6363636363636364e-05,
+      "loss": 0.3483,
       "step": 80
     },
     {
       "epoch": 4.0,
-      "grad_norm": 9.567431449890137,
-      "learning_rate": 3.282828282828283e-05,
-      "loss": 0.7468,
+      "grad_norm": 18.051124572753906,
+      "learning_rate": 4.0909090909090915e-05,
+      "loss": 0.486,
       "step": 90
     },
     {
       "epoch": 4.0,
-      "eval_accuracy": 0.6734375,
-      "eval_loss": 0.7003158330917358,
-      "eval_runtime": 3.5895,
-      "eval_samples_per_second": 178.3,
-      "eval_steps_per_second": 2.786,
+      "eval_accuracy": 0.81875,
+      "eval_loss": 0.4530211389064789,
+      "eval_runtime": 3.6057,
+      "eval_samples_per_second": 177.495,
+      "eval_steps_per_second": 2.773,
       "step": 90
     },
     {
       "epoch": 4.444444444444445,
-      "grad_norm": 6.920608043670654,
-      "learning_rate": 3.0303030303030306e-05,
-      "loss": 0.683,
+      "grad_norm": 15.844127655029297,
+      "learning_rate": 4.545454545454546e-05,
+      "loss": 0.3521,
       "step": 100
     },
     {
       "epoch": 4.888888888888889,
-      "grad_norm": 8.078107833862305,
-      "learning_rate": 2.777777777777778e-05,
-      "loss": 0.6935,
+      "grad_norm": 11.87112808227539,
+      "learning_rate": 5e-05,
+      "loss": 0.3307,
       "step": 110
     },
     {
       "epoch": 4.977777777777778,
-      "eval_accuracy": 0.7546875,
-      "eval_loss": 0.6063631772994995,
-      "eval_runtime": 3.8885,
-      "eval_samples_per_second": 164.589,
-      "eval_steps_per_second": 2.572,
+      "eval_accuracy": 0.8234375,
+      "eval_loss": 0.45600825548171997,
+      "eval_runtime": 3.572,
+      "eval_samples_per_second": 179.171,
+      "eval_steps_per_second": 2.8,
       "step": 112
     },
     {
       "epoch": 5.333333333333333,
-      "grad_norm": 11.370922088623047,
-      "learning_rate": 2.5252525252525256e-05,
-      "loss": 0.6469,
+      "grad_norm": 17.418668746948242,
+      "learning_rate": 4.94949494949495e-05,
+      "loss": 0.3404,
       "step": 120
     },
     {
       "epoch": 5.777777777777778,
-      "grad_norm": 12.74963665008545,
-      "learning_rate": 2.272727272727273e-05,
-      "loss": 0.6271,
+      "grad_norm": 20.148906707763672,
+      "learning_rate": 4.898989898989899e-05,
+      "loss": 0.306,
       "step": 130
     },
     {
       "epoch": 6.0,
-      "eval_accuracy": 0.76875,
-      "eval_loss": 0.5647965669631958,
-      "eval_runtime": 3.9273,
-      "eval_samples_per_second": 162.963,
-      "eval_steps_per_second": 2.546,
+      "eval_accuracy": 0.8671875,
+      "eval_loss": 0.3470742106437683,
+      "eval_runtime": 3.5697,
+      "eval_samples_per_second": 179.289,
+      "eval_steps_per_second": 2.801,
       "step": 135
     },
     {
       "epoch": 6.222222222222222,
-      "grad_norm": 11.86828327178955,
-      "learning_rate": 2.0202020202020203e-05,
-      "loss": 0.5635,
+      "grad_norm": 11.451733589172363,
+      "learning_rate": 4.848484848484849e-05,
+      "loss": 0.2873,
       "step": 140
     },
     {
       "epoch": 6.666666666666667,
-      "grad_norm": 13.892380714416504,
-      "learning_rate": 1.7676767676767676e-05,
-      "loss": 0.5622,
+      "grad_norm": 15.43708610534668,
+      "learning_rate": 4.797979797979798e-05,
+      "loss": 0.3005,
       "step": 150
     },
     {
       "epoch": 6.977777777777778,
-      "eval_accuracy": 0.809375,
-      "eval_loss": 0.48242831230163574,
-      "eval_runtime": 3.5816,
-      "eval_samples_per_second": 178.692,
-      "eval_steps_per_second": 2.792,
+      "eval_accuracy": 0.8859375,
+      "eval_loss": 0.3024638891220093,
+      "eval_runtime": 3.8788,
+      "eval_samples_per_second": 164.998,
+      "eval_steps_per_second": 2.578,
       "step": 157
     },
     {
       "epoch": 7.111111111111111,
-      "grad_norm": 8.864046096801758,
-      "learning_rate": 1.5151515151515153e-05,
-      "loss": 0.525,
+      "grad_norm": 20.696516036987305,
+      "learning_rate": 4.7474747474747476e-05,
+      "loss": 0.3206,
       "step": 160
     },
     {
       "epoch": 7.555555555555555,
-      "grad_norm": 8.477625846862793,
-      "learning_rate": 1.2626262626262628e-05,
-      "loss": 0.4967,
+      "grad_norm": 12.01241397857666,
+      "learning_rate": 4.696969696969697e-05,
+      "loss": 0.2851,
       "step": 170
     },
     {
       "epoch": 8.0,
-      "grad_norm": 12.289462089538574,
-      "learning_rate": 1.0101010101010101e-05,
-      "loss": 0.4815,
+      "grad_norm": 17.638214111328125,
+      "learning_rate": 4.6464646464646464e-05,
+      "loss": 0.319,
       "step": 180
     },
     {
       "epoch": 8.0,
-      "eval_accuracy": 0.8609375,
-      "eval_loss": 0.4012059271335602,
-      "eval_runtime": 3.613,
-      "eval_samples_per_second": 177.138,
-      "eval_steps_per_second": 2.768,
+      "eval_accuracy": 0.8984375,
+      "eval_loss": 0.24509796500205994,
+      "eval_runtime": 3.9737,
+      "eval_samples_per_second": 161.061,
+      "eval_steps_per_second": 2.517,
       "step": 180
     },
     {
       "epoch": 8.444444444444445,
-      "grad_norm": 8.446834564208984,
-      "learning_rate": 7.5757575757575764e-06,
-      "loss": 0.45,
+      "grad_norm": 21.502132415771484,
+      "learning_rate": 4.595959595959596e-05,
+      "loss": 0.2968,
       "step": 190
     },
     {
       "epoch": 8.88888888888889,
-      "grad_norm": 15.683026313781738,
-      "learning_rate": 5.050505050505051e-06,
-      "loss": 0.4771,
+      "grad_norm": 20.09746742248535,
+      "learning_rate": 4.545454545454546e-05,
+      "loss": 0.3489,
       "step": 200
     },
     {
       "epoch": 8.977777777777778,
-      "eval_accuracy": 0.85625,
-      "eval_loss": 0.3798871636390686,
-      "eval_runtime": 3.5689,
-      "eval_samples_per_second": 179.325,
-      "eval_steps_per_second": 2.802,
+      "eval_accuracy": 0.928125,
+      "eval_loss": 0.18142804503440857,
+      "eval_runtime": 3.7455,
+      "eval_samples_per_second": 170.872,
+      "eval_steps_per_second": 2.67,
       "step": 202
     },
     {
       "epoch": 9.333333333333334,
-      "grad_norm": 9.576162338256836,
-      "learning_rate": 2.5252525252525253e-06,
-      "loss": 0.4376,
+      "grad_norm": 16.205760955810547,
+      "learning_rate": 4.494949494949495e-05,
+      "loss": 0.2915,
       "step": 210
     },
     {
       "epoch": 9.777777777777779,
-      "grad_norm": 9.685094833374023,
-      "learning_rate": 0.0,
-      "loss": 0.4171,
+      "grad_norm": 27.511030197143555,
+      "learning_rate": 4.4444444444444447e-05,
+      "loss": 0.3251,
       "step": 220
     },
     {
-      "epoch": 9.777777777777779,
-      "eval_accuracy": 0.8625,
-      "eval_loss": 0.3611668050289154,
-      "eval_runtime": 3.9731,
-      "eval_samples_per_second": 161.083,
-      "eval_steps_per_second": 2.517,
-      "step": 220
+      "epoch": 10.0,
+      "eval_accuracy": 0.915625,
+      "eval_loss": 0.24511559307575226,
+      "eval_runtime": 3.6361,
+      "eval_samples_per_second": 176.011,
+      "eval_steps_per_second": 2.75,
+      "step": 225
     },
     {
-      "epoch": 9.777777777777779,
-      "step": 220,
-      "total_flos": 1.102133137023959e+18,
-      "train_loss": 0.670017595724626,
-      "train_runtime": 696.1068,
-      "train_samples_per_second": 82.746,
-      "train_steps_per_second": 0.316
+      "epoch": 10.222222222222221,
+      "grad_norm": 11.171629905700684,
+      "learning_rate": 4.3939393939393944e-05,
+      "loss": 0.308,
+      "step": 230
+    },
+    {
+      "epoch": 10.666666666666666,
+      "grad_norm": 12.315302848815918,
+      "learning_rate": 4.343434343434344e-05,
+      "loss": 0.3034,
+      "step": 240
+    },
+    {
+      "epoch": 10.977777777777778,
+      "eval_accuracy": 0.940625,
+      "eval_loss": 0.15658709406852722,
+      "eval_runtime": 3.5876,
+      "eval_samples_per_second": 178.392,
+      "eval_steps_per_second": 2.787,
+      "step": 247
+    },
+    {
+      "epoch": 11.11111111111111,
+      "grad_norm": 14.539319038391113,
+      "learning_rate": 4.292929292929293e-05,
+      "loss": 0.2847,
+      "step": 250
+    },
+    {
+      "epoch": 11.555555555555555,
+      "grad_norm": 17.26177215576172,
+      "learning_rate": 4.242424242424243e-05,
+      "loss": 0.2754,
+      "step": 260
+    },
+    {
+      "epoch": 12.0,
+      "grad_norm": 12.163039207458496,
+      "learning_rate": 4.191919191919192e-05,
+      "loss": 0.2746,
+      "step": 270
+    },
+    {
+      "epoch": 12.0,
+      "eval_accuracy": 0.8921875,
+      "eval_loss": 0.24925951659679413,
+      "eval_runtime": 3.5808,
+      "eval_samples_per_second": 178.73,
+      "eval_steps_per_second": 2.793,
+      "step": 270
+    },
+    {
+      "epoch": 12.444444444444445,
+      "grad_norm": 11.92519474029541,
+      "learning_rate": 4.141414141414142e-05,
+      "loss": 0.2889,
+      "step": 280
+    },
+    {
+      "epoch": 12.88888888888889,
+      "grad_norm": 13.212408065795898,
+      "learning_rate": 4.0909090909090915e-05,
+      "loss": 0.2369,
+      "step": 290
+    },
+    {
+      "epoch": 12.977777777777778,
+      "eval_accuracy": 0.9375,
+      "eval_loss": 0.1622403860092163,
+      "eval_runtime": 3.8343,
+      "eval_samples_per_second": 166.914,
+      "eval_steps_per_second": 2.608,
+      "step": 292
+    },
+    {
+      "epoch": 13.333333333333334,
+      "grad_norm": 11.68896770477295,
+      "learning_rate": 4.0404040404040405e-05,
+      "loss": 0.2465,
+      "step": 300
+    },
+    {
+      "epoch": 13.777777777777779,
+      "grad_norm": 14.610076904296875,
+      "learning_rate": 3.98989898989899e-05,
+      "loss": 0.2231,
+      "step": 310
+    },
+    {
+      "epoch": 14.0,
+      "eval_accuracy": 0.9359375,
+      "eval_loss": 0.17805945873260498,
+      "eval_runtime": 3.9695,
+      "eval_samples_per_second": 161.231,
+      "eval_steps_per_second": 2.519,
+      "step": 315
+    },
+    {
+      "epoch": 14.222222222222221,
+      "grad_norm": 12.407272338867188,
+      "learning_rate": 3.939393939393939e-05,
+      "loss": 0.2177,
+      "step": 320
+    },
+    {
+      "epoch": 14.666666666666666,
+      "grad_norm": 7.3430256843566895,
+      "learning_rate": 3.888888888888889e-05,
+      "loss": 0.2281,
+      "step": 330
+    },
+    {
+      "epoch": 14.977777777777778,
+      "eval_accuracy": 0.953125,
+      "eval_loss": 0.12681424617767334,
+      "eval_runtime": 3.6554,
+      "eval_samples_per_second": 175.082,
+      "eval_steps_per_second": 2.736,
+      "step": 337
+    },
+    {
+      "epoch": 15.11111111111111,
+      "grad_norm": 10.262022018432617,
+      "learning_rate": 3.838383838383838e-05,
+      "loss": 0.209,
+      "step": 340
+    },
+    {
+      "epoch": 15.555555555555555,
+      "grad_norm": 9.078124046325684,
+      "learning_rate": 3.787878787878788e-05,
+      "loss": 0.2134,
+      "step": 350
+    },
+    {
+      "epoch": 16.0,
+      "grad_norm": 14.094355583190918,
+      "learning_rate": 3.7373737373737376e-05,
+      "loss": 0.2001,
+      "step": 360
+    },
+    {
+      "epoch": 16.0,
+      "eval_accuracy": 0.9140625,
+      "eval_loss": 0.24309130012989044,
+      "eval_runtime": 3.5892,
+      "eval_samples_per_second": 178.311,
+      "eval_steps_per_second": 2.786,
+      "step": 360
+    },
+    {
+      "epoch": 16.444444444444443,
+      "grad_norm": 12.868298530578613,
+      "learning_rate": 3.686868686868687e-05,
+      "loss": 0.2312,
+      "step": 370
+    },
+    {
+      "epoch": 16.88888888888889,
+      "grad_norm": 7.863047122955322,
+      "learning_rate": 3.6363636363636364e-05,
+      "loss": 0.183,
+      "step": 380
+    },
+    {
+      "epoch": 16.977777777777778,
+      "eval_accuracy": 0.9625,
+      "eval_loss": 0.10167054831981659,
+      "eval_runtime": 3.6007,
+      "eval_samples_per_second": 177.743,
+      "eval_steps_per_second": 2.777,
+      "step": 382
+    },
+    {
+      "epoch": 17.333333333333332,
+      "grad_norm": 8.919840812683105,
+      "learning_rate": 3.5858585858585855e-05,
+      "loss": 0.1997,
+      "step": 390
+    },
+    {
+      "epoch": 17.77777777777778,
+      "grad_norm": 11.91215705871582,
+      "learning_rate": 3.535353535353535e-05,
+      "loss": 0.1891,
+      "step": 400
+    },
+    {
+      "epoch": 18.0,
+      "eval_accuracy": 0.9390625,
+      "eval_loss": 0.180230051279068,
+      "eval_runtime": 3.5491,
+      "eval_samples_per_second": 180.325,
+      "eval_steps_per_second": 2.818,
+      "step": 405
+    },
+    {
+      "epoch": 18.22222222222222,
+      "grad_norm": 9.109786033630371,
+      "learning_rate": 3.484848484848485e-05,
+      "loss": 0.213,
+      "step": 410
+    },
+    {
+      "epoch": 18.666666666666668,
+      "grad_norm": 5.232081413269043,
+      "learning_rate": 3.434343434343435e-05,
+      "loss": 0.1862,
+      "step": 420
+    },
+    {
+      "epoch": 18.977777777777778,
+      "eval_accuracy": 0.9765625,
+      "eval_loss": 0.08689282089471817,
+      "eval_runtime": 3.9268,
+      "eval_samples_per_second": 162.982,
+      "eval_steps_per_second": 2.547,
+      "step": 427
+    },
+    {
+      "epoch": 19.11111111111111,
+      "grad_norm": 13.340733528137207,
+      "learning_rate": 3.3838383838383844e-05,
+      "loss": 0.1664,
+      "step": 430
+    },
+    {
+      "epoch": 19.555555555555557,
+      "grad_norm": 7.66475772857666,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 0.1727,
+      "step": 440
+    },
+    {
+      "epoch": 20.0,
+      "grad_norm": 13.612215995788574,
+      "learning_rate": 3.282828282828283e-05,
+      "loss": 0.1935,
+      "step": 450
+    },
+    {
+      "epoch": 20.0,
+      "eval_accuracy": 0.96875,
+      "eval_loss": 0.10791148245334625,
+      "eval_runtime": 3.9472,
+      "eval_samples_per_second": 162.141,
+      "eval_steps_per_second": 2.533,
+      "step": 450
+    },
+    {
+      "epoch": 20.444444444444443,
+      "grad_norm": 9.189305305480957,
+      "learning_rate": 3.232323232323233e-05,
+      "loss": 0.1945,
+      "step": 460
+    },
+    {
+      "epoch": 20.88888888888889,
+      "grad_norm": 9.650483131408691,
+      "learning_rate": 3.181818181818182e-05,
+      "loss": 0.1797,
+      "step": 470
+    },
+    {
+      "epoch": 20.977777777777778,
+      "eval_accuracy": 0.95625,
+      "eval_loss": 0.12502644956111908,
+      "eval_runtime": 3.6171,
+      "eval_samples_per_second": 176.939,
+      "eval_steps_per_second": 2.765,
+      "step": 472
+    },
+    {
+      "epoch": 21.333333333333332,
+      "grad_norm": 7.25011682510376,
+      "learning_rate": 3.131313131313132e-05,
+      "loss": 0.1767,
+      "step": 480
+    },
+    {
+      "epoch": 21.77777777777778,
+      "grad_norm": 9.036290168762207,
+      "learning_rate": 3.080808080808081e-05,
+      "loss": 0.1605,
+      "step": 490
+    },
+    {
+      "epoch": 22.0,
+      "eval_accuracy": 0.971875,
+      "eval_loss": 0.06545940786600113,
+      "eval_runtime": 3.5923,
+      "eval_samples_per_second": 178.16,
+      "eval_steps_per_second": 2.784,
+      "step": 495
+    },
+    {
+      "epoch": 22.22222222222222,
+      "grad_norm": 5.982744216918945,
+      "learning_rate": 3.0303030303030306e-05,
+      "loss": 0.1493,
+      "step": 500
+    },
+    {
+      "epoch": 22.666666666666668,
+      "grad_norm": 13.987672805786133,
+      "learning_rate": 2.9797979797979796e-05,
+      "loss": 0.1848,
+      "step": 510
+    },
+    {
+      "epoch": 22.977777777777778,
+      "eval_accuracy": 0.9765625,
+      "eval_loss": 0.08063917607069016,
+      "eval_runtime": 3.5801,
+      "eval_samples_per_second": 178.764,
+      "eval_steps_per_second": 2.793,
+      "step": 517
+    },
+    {
+      "epoch": 23.11111111111111,
+      "grad_norm": 15.488668441772461,
+      "learning_rate": 2.9292929292929294e-05,
+      "loss": 0.1426,
+      "step": 520
+    },
+    {
+      "epoch": 23.555555555555557,
+      "grad_norm": 11.646829605102539,
+      "learning_rate": 2.878787878787879e-05,
+      "loss": 0.1667,
+      "step": 530
+    },
+    {
+      "epoch": 24.0,
+      "grad_norm": 6.619264602661133,
+      "learning_rate": 2.8282828282828282e-05,
+      "loss": 0.1498,
+      "step": 540
+    },
+    {
+      "epoch": 24.0,
+      "eval_accuracy": 0.9578125,
+      "eval_loss": 0.11159060150384903,
+      "eval_runtime": 3.6016,
+      "eval_samples_per_second": 177.701,
+      "eval_steps_per_second": 2.777,
+      "step": 540
+    },
+    {
+      "epoch": 24.444444444444443,
+      "grad_norm": 7.8661627769470215,
+      "learning_rate": 2.777777777777778e-05,
+      "loss": 0.1287,
+      "step": 550
+    },
+    {
+      "epoch": 24.88888888888889,
+      "grad_norm": 7.934934616088867,
+      "learning_rate": 2.7272727272727273e-05,
+      "loss": 0.1394,
+      "step": 560
+    },
+    {
+      "epoch": 24.977777777777778,
+      "eval_accuracy": 0.9671875,
+      "eval_loss": 0.0806862860918045,
+      "eval_runtime": 3.8749,
+      "eval_samples_per_second": 165.167,
+      "eval_steps_per_second": 2.581,
+      "step": 562
+    },
+    {
+      "epoch": 25.333333333333332,
+      "grad_norm": 10.52723217010498,
+      "learning_rate": 2.676767676767677e-05,
+      "loss": 0.1524,
+      "step": 570
+    },
+    {
+      "epoch": 25.77777777777778,
+      "grad_norm": 7.490493297576904,
+      "learning_rate": 2.6262626262626268e-05,
+      "loss": 0.1584,
+      "step": 580
+    },
+    {
+      "epoch": 26.0,
+      "eval_accuracy": 0.9796875,
+      "eval_loss": 0.05252554267644882,
+      "eval_runtime": 3.9703,
+      "eval_samples_per_second": 161.197,
+      "eval_steps_per_second": 2.519,
+      "step": 585
+    },
+    {
+      "epoch": 26.22222222222222,
+      "grad_norm": 7.178821563720703,
+      "learning_rate": 2.575757575757576e-05,
+      "loss": 0.153,
+      "step": 590
+    },
+    {
+      "epoch": 26.666666666666668,
+      "grad_norm": 7.702730178833008,
+      "learning_rate": 2.5252525252525256e-05,
+      "loss": 0.1302,
+      "step": 600
+    },
+    {
+      "epoch": 26.977777777777778,
+      "eval_accuracy": 0.9828125,
+      "eval_loss": 0.05131406709551811,
+      "eval_runtime": 3.7014,
+      "eval_samples_per_second": 172.905,
+      "eval_steps_per_second": 2.702,
+      "step": 607
+    },
+    {
+      "epoch": 27.11111111111111,
+      "grad_norm": 9.482915878295898,
+      "learning_rate": 2.474747474747475e-05,
+      "loss": 0.1577,
+      "step": 610
+    },
+    {
+      "epoch": 27.555555555555557,
+      "grad_norm": 10.196369171142578,
+      "learning_rate": 2.4242424242424244e-05,
+      "loss": 0.1543,
+      "step": 620
+    },
+    {
+      "epoch": 28.0,
+      "grad_norm": 10.799006462097168,
+      "learning_rate": 2.3737373737373738e-05,
+      "loss": 0.1356,
+      "step": 630
+    },
+    {
+      "epoch": 28.0,
+      "eval_accuracy": 0.9875,
+      "eval_loss": 0.04204293340444565,
+      "eval_runtime": 3.9774,
+      "eval_samples_per_second": 160.91,
+      "eval_steps_per_second": 2.514,
+      "step": 630
+    },
+    {
+      "epoch": 28.444444444444443,
+      "grad_norm": 7.129752159118652,
+      "learning_rate": 2.3232323232323232e-05,
+      "loss": 0.1291,
+      "step": 640
+    },
+    {
+      "epoch": 28.88888888888889,
+      "grad_norm": 6.642085552215576,
+      "learning_rate": 2.272727272727273e-05,
+      "loss": 0.1101,
+      "step": 650
+    },
+    {
+      "epoch": 28.977777777777778,
+      "eval_accuracy": 0.9875,
+      "eval_loss": 0.03539272025227547,
+      "eval_runtime": 4.0166,
+      "eval_samples_per_second": 159.34,
+      "eval_steps_per_second": 2.49,
+      "step": 652
+    },
+    {
+      "epoch": 29.333333333333332,
+      "grad_norm": 8.5753755569458,
+      "learning_rate": 2.2222222222222223e-05,
+      "loss": 0.1445,
+      "step": 660
+    },
+    {
+      "epoch": 29.77777777777778,
+      "grad_norm": 13.183974266052246,
+      "learning_rate": 2.171717171717172e-05,
+      "loss": 0.1227,
+      "step": 670
+    },
+    {
+      "epoch": 30.0,
+      "eval_accuracy": 0.9765625,
+      "eval_loss": 0.0582769513130188,
+      "eval_runtime": 3.9913,
+      "eval_samples_per_second": 160.35,
+      "eval_steps_per_second": 2.505,
+      "step": 675
+    },
+    {
+      "epoch": 30.22222222222222,
+      "grad_norm": 9.026564598083496,
+      "learning_rate": 2.1212121212121215e-05,
+      "loss": 0.1209,
+      "step": 680
+    },
+    {
+      "epoch": 30.666666666666668,
+      "grad_norm": 12.504347801208496,
+      "learning_rate": 2.070707070707071e-05,
+      "loss": 0.1158,
+      "step": 690
+    },
+    {
+      "epoch": 30.977777777777778,
+      "eval_accuracy": 0.990625,
+      "eval_loss": 0.025325458496809006,
+      "eval_runtime": 3.9678,
+      "eval_samples_per_second": 161.299,
+      "eval_steps_per_second": 2.52,
+      "step": 697
+    },
+    {
+      "epoch": 31.11111111111111,
+      "grad_norm": 5.276214599609375,
+      "learning_rate": 2.0202020202020203e-05,
+      "loss": 0.119,
+      "step": 700
+    },
+    {
+      "epoch": 31.555555555555557,
+      "grad_norm": 8.732769012451172,
+      "learning_rate": 1.9696969696969697e-05,
+      "loss": 0.1156,
+      "step": 710
+    },
+    {
+      "epoch": 32.0,
+      "grad_norm": 5.604591369628906,
+      "learning_rate": 1.919191919191919e-05,
+      "loss": 0.117,
+      "step": 720
+    },
+    {
+      "epoch": 32.0,
+      "eval_accuracy": 0.990625,
+      "eval_loss": 0.023098567500710487,
+      "eval_runtime": 3.8505,
+      "eval_samples_per_second": 166.213,
+      "eval_steps_per_second": 2.597,
+      "step": 720
+    },
+    {
+      "epoch": 32.44444444444444,
+      "grad_norm": 7.332610607147217,
+      "learning_rate": 1.8686868686868688e-05,
+      "loss": 0.1213,
+      "step": 730
+    },
+    {
+      "epoch": 32.888888888888886,
+      "grad_norm": 12.890093803405762,
+      "learning_rate": 1.8181818181818182e-05,
+      "loss": 0.1022,
+      "step": 740
+    },
+    {
+      "epoch": 32.977777777777774,
+      "eval_accuracy": 0.9796875,
+      "eval_loss": 0.0725882276892662,
+      "eval_runtime": 3.8065,
+      "eval_samples_per_second": 168.135,
+      "eval_steps_per_second": 2.627,
+      "step": 742
+    },
+    {
+      "epoch": 33.333333333333336,
+      "grad_norm": 13.247682571411133,
+      "learning_rate": 1.7676767676767676e-05,
+      "loss": 0.1257,
+      "step": 750
+    },
+    {
+      "epoch": 33.77777777777778,
+      "grad_norm": 6.758236885070801,
+      "learning_rate": 1.7171717171717173e-05,
+      "loss": 0.1221,
+      "step": 760
+    },
+    {
+      "epoch": 34.0,
+      "eval_accuracy": 0.996875,
+      "eval_loss": 0.015964530408382416,
+      "eval_runtime": 3.7585,
+      "eval_samples_per_second": 170.283,
+      "eval_steps_per_second": 2.661,
+      "step": 765
+    },
+    {
+      "epoch": 34.22222222222222,
+      "grad_norm": 8.521262168884277,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 0.1014,
+      "step": 770
+    },
+    {
+      "epoch": 34.666666666666664,
+      "grad_norm": 5.949100971221924,
+      "learning_rate": 1.6161616161616165e-05,
+      "loss": 0.0956,
+      "step": 780
+    },
+    {
+      "epoch": 34.977777777777774,
+      "eval_accuracy": 0.984375,
+      "eval_loss": 0.048214979469776154,
+      "eval_runtime": 3.6909,
+      "eval_samples_per_second": 173.399,
+      "eval_steps_per_second": 2.709,
+      "step": 787
+    },
+    {
+      "epoch": 35.111111111111114,
+      "grad_norm": 10.151766777038574,
+      "learning_rate": 1.565656565656566e-05,
+      "loss": 0.1135,
+      "step": 790
+    },
+    {
+      "epoch": 35.55555555555556,
+      "grad_norm": 9.514137268066406,
+      "learning_rate": 1.5151515151515153e-05,
+      "loss": 0.1109,
+      "step": 800
+    },
+    {
+      "epoch": 36.0,
+      "grad_norm": 4.4278717041015625,
+      "learning_rate": 1.4646464646464647e-05,
+      "loss": 0.0856,
+      "step": 810
+    },
+    {
+      "epoch": 36.0,
+      "eval_accuracy": 0.9875,
+      "eval_loss": 0.025567293167114258,
+      "eval_runtime": 3.6537,
+      "eval_samples_per_second": 175.165,
+      "eval_steps_per_second": 2.737,
+      "step": 810
+    },
+    {
+      "epoch": 36.44444444444444,
+      "grad_norm": 8.582184791564941,
+      "learning_rate": 1.4141414141414141e-05,
+      "loss": 0.0994,
+      "step": 820
+    },
+    {
+      "epoch": 36.888888888888886,
+      "grad_norm": 9.628859519958496,
+      "learning_rate": 1.3636363636363637e-05,
+      "loss": 0.0996,
+      "step": 830
+    },
+    {
+      "epoch": 36.977777777777774,
+      "eval_accuracy": 0.990625,
+      "eval_loss": 0.021057253703475,
+      "eval_runtime": 3.6772,
+      "eval_samples_per_second": 174.046,
+      "eval_steps_per_second": 2.719,
+      "step": 832
+    },
+    {
+      "epoch": 37.333333333333336,
+      "grad_norm": 5.165952682495117,
+      "learning_rate": 1.3131313131313134e-05,
+      "loss": 0.0915,
+      "step": 840
+    },
+    {
+      "epoch": 37.77777777777778,
+      "grad_norm": 5.831385135650635,
+      "learning_rate": 1.2626262626262628e-05,
+      "loss": 0.0848,
+      "step": 850
+    },
+    {
+      "epoch": 38.0,
+      "eval_accuracy": 0.9796875,
+      "eval_loss": 0.04457371309399605,
+      "eval_runtime": 3.6584,
+      "eval_samples_per_second": 174.938,
+      "eval_steps_per_second": 2.733,
+      "step": 855
+    },
+    {
+      "epoch": 38.22222222222222,
+      "grad_norm": 9.629181861877441,
+      "learning_rate": 1.2121212121212122e-05,
+      "loss": 0.0972,
+      "step": 860
+    },
+    {
+      "epoch": 38.666666666666664,
+      "grad_norm": 6.214244365692139,
+      "learning_rate": 1.1616161616161616e-05,
+      "loss": 0.1001,
+      "step": 870
+    },
+    {
+      "epoch": 38.977777777777774,
+      "eval_accuracy": 0.9875,
+      "eval_loss": 0.02742326818406582,
+      "eval_runtime": 3.668,
+      "eval_samples_per_second": 174.481,
+      "eval_steps_per_second": 2.726,
+      "step": 877
+    },
+    {
+      "epoch": 39.111111111111114,
+      "grad_norm": 11.1734619140625,
+      "learning_rate": 1.1111111111111112e-05,
+      "loss": 0.0919,
+      "step": 880
+    },
+    {
+      "epoch": 39.55555555555556,
+      "grad_norm": 6.438005447387695,
+      "learning_rate": 1.0606060606060607e-05,
+      "loss": 0.0988,
+      "step": 890
+    },
+    {
+      "epoch": 40.0,
+      "grad_norm": 5.9803643226623535,
+      "learning_rate": 1.0101010101010101e-05,
+      "loss": 0.0976,
+      "step": 900
+    },
+    {
+      "epoch": 40.0,
+      "eval_accuracy": 0.9921875,
+      "eval_loss": 0.022529248148202896,
+      "eval_runtime": 3.7092,
+      "eval_samples_per_second": 172.543,
+      "eval_steps_per_second": 2.696,
+      "step": 900
+    },
+    {
+      "epoch": 40.44444444444444,
+      "grad_norm": 7.562661647796631,
+      "learning_rate": 9.595959595959595e-06,
+      "loss": 0.085,
+      "step": 910
+    },
+    {
+      "epoch": 40.888888888888886,
+      "grad_norm": 7.695030212402344,
+      "learning_rate": 9.090909090909091e-06,
+      "loss": 0.0864,
+      "step": 920
+    },
+    {
+      "epoch": 40.977777777777774,
+      "eval_accuracy": 0.9921875,
+      "eval_loss": 0.0207191314548254,
+      "eval_runtime": 4.0024,
+      "eval_samples_per_second": 159.904,
+      "eval_steps_per_second": 2.499,
+      "step": 922
+    },
+    {
+      "epoch": 41.333333333333336,
+      "grad_norm": 8.4052734375,
+      "learning_rate": 8.585858585858587e-06,
+      "loss": 0.088,
+      "step": 930
+    },
+    {
+      "epoch": 41.77777777777778,
+      "grad_norm": 8.705794334411621,
+      "learning_rate": 8.080808080808082e-06,
+      "loss": 0.0865,
+      "step": 940
+    },
+    {
+      "epoch": 42.0,
+      "eval_accuracy": 0.996875,
+      "eval_loss": 0.01933131366968155,
+      "eval_runtime": 3.9909,
+      "eval_samples_per_second": 160.365,
+      "eval_steps_per_second": 2.506,
+      "step": 945
+    },
+    {
+      "epoch": 42.22222222222222,
+      "grad_norm": 7.478874683380127,
+      "learning_rate": 7.5757575757575764e-06,
+      "loss": 0.0815,
+      "step": 950
+    },
+    {
+      "epoch": 42.666666666666664,
+      "grad_norm": 5.25657320022583,
+      "learning_rate": 7.0707070707070704e-06,
+      "loss": 0.0773,
+      "step": 960
+    },
+    {
+      "epoch": 42.977777777777774,
+      "eval_accuracy": 0.9921875,
+      "eval_loss": 0.020288193598389626,
+      "eval_runtime": 3.6594,
+      "eval_samples_per_second": 174.89,
+      "eval_steps_per_second": 2.733,
+      "step": 967
+    },
+    {
+      "epoch": 43.111111111111114,
+      "grad_norm": 4.1972246170043945,
+      "learning_rate": 6.565656565656567e-06,
+      "loss": 0.0799,
+      "step": 970
+    },
+    {
+      "epoch": 43.55555555555556,
+      "grad_norm": 6.9554972648620605,
+      "learning_rate": 6.060606060606061e-06,
+      "loss": 0.0772,
+      "step": 980
+    },
+    {
+      "epoch": 44.0,
+      "grad_norm": 6.343081951141357,
+      "learning_rate": 5.555555555555556e-06,
+      "loss": 0.075,
+      "step": 990
+    },
+    {
+      "epoch": 44.0,
+      "eval_accuracy": 0.996875,
+      "eval_loss": 0.013058523647487164,
+      "eval_runtime": 3.5774,
+      "eval_samples_per_second": 178.899,
+      "eval_steps_per_second": 2.795,
+      "step": 990
+    },
+    {
+      "epoch": 44.44444444444444,
+      "grad_norm": 4.900812149047852,
+      "learning_rate": 5.050505050505051e-06,
+      "loss": 0.0736,
+      "step": 1000
+    },
+    {
+      "epoch": 44.888888888888886,
+      "grad_norm": 5.955135345458984,
+      "learning_rate": 4.5454545454545455e-06,
+      "loss": 0.0761,
+      "step": 1010
+    },
+    {
+      "epoch": 44.977777777777774,
+      "eval_accuracy": 0.99375,
+      "eval_loss": 0.012860281392931938,
+      "eval_runtime": 3.5897,
+      "eval_samples_per_second": 178.288,
+      "eval_steps_per_second": 2.786,
+      "step": 1012
+    },
+    {
+      "epoch": 45.333333333333336,
+      "grad_norm": 4.250102996826172,
+      "learning_rate": 4.040404040404041e-06,
+      "loss": 0.0707,
+      "step": 1020
+    },
+    {
+      "epoch": 45.77777777777778,
+      "grad_norm": 6.8997931480407715,
+      "learning_rate": 3.5353535353535352e-06,
+      "loss": 0.0624,
+      "step": 1030
+    },
+    {
+      "epoch": 46.0,
+      "eval_accuracy": 0.996875,
+      "eval_loss": 0.011364495381712914,
+      "eval_runtime": 3.6128,
+      "eval_samples_per_second": 177.146,
+      "eval_steps_per_second": 2.768,
+      "step": 1035
+    },
+    {
+      "epoch": 46.22222222222222,
+      "grad_norm": 10.210082054138184,
+      "learning_rate": 3.0303030303030305e-06,
+      "loss": 0.0762,
+      "step": 1040
+    },
+    {
+      "epoch": 46.666666666666664,
+      "grad_norm": 4.9201788902282715,
+      "learning_rate": 2.5252525252525253e-06,
+      "loss": 0.0557,
+      "step": 1050
+    },
+    {
+      "epoch": 46.977777777777774,
+      "eval_accuracy": 0.9953125,
+      "eval_loss": 0.010208332911133766,
+      "eval_runtime": 3.8474,
+      "eval_samples_per_second": 166.347,
+      "eval_steps_per_second": 2.599,
+      "step": 1057
+    },
+    {
+      "epoch": 47.111111111111114,
+      "grad_norm": 3.725327491760254,
+      "learning_rate": 2.0202020202020206e-06,
+      "loss": 0.0613,
+      "step": 1060
+    },
+    {
+      "epoch": 47.55555555555556,
+      "grad_norm": 3.1549530029296875,
+      "learning_rate": 1.5151515151515152e-06,
+      "loss": 0.0601,
+      "step": 1070
+    },
+    {
+      "epoch": 48.0,
+      "grad_norm": 6.280518054962158,
+      "learning_rate": 1.0101010101010103e-06,
+      "loss": 0.0708,
+      "step": 1080
+    },
+    {
+      "epoch": 48.0,
+      "eval_accuracy": 0.9953125,
+      "eval_loss": 0.01160599384456873,
+      "eval_runtime": 3.9922,
+      "eval_samples_per_second": 160.314,
+      "eval_steps_per_second": 2.505,
+      "step": 1080
+    },
+    {
+      "epoch": 48.44444444444444,
+      "grad_norm": 6.1849260330200195,
+      "learning_rate": 5.050505050505052e-07,
+      "loss": 0.0699,
+      "step": 1090
+    },
+    {
+      "epoch": 48.888888888888886,
+      "grad_norm": 7.637501239776611,
+      "learning_rate": 0.0,
+      "loss": 0.0667,
+      "step": 1100
+    },
+    {
+      "epoch": 48.888888888888886,
+      "eval_accuracy": 0.9953125,
+      "eval_loss": 0.013088616542518139,
+      "eval_runtime": 3.601,
+      "eval_samples_per_second": 177.727,
+      "eval_steps_per_second": 2.777,
+      "step": 1100
+    },
+    {
+      "epoch": 48.888888888888886,
+      "step": 1100,
+      "total_flos": 5.510665685119795e+18,
+      "train_loss": 0.17931611462072894,
+      "train_runtime": 3528.3547,
+      "train_samples_per_second": 81.624,
+      "train_steps_per_second": 0.312
     }
   ],
   "logging_steps": 10,
-  "max_steps": 220,
+  "max_steps": 1100,
   "num_input_tokens_seen": 0,
-  "num_train_epochs": 10,
+  "num_train_epochs": 50,
   "save_steps": 500,
-  "total_flos": 1.102133137023959e+18,
+  "total_flos": 5.510665685119795e+18,
   "train_batch_size": 64,
   "trial_name": null,
   "trial_params": null