End of training

README.md

@@ -2,11 +2,23 @@
 library_name: transformers
 tags:
 - generated_from_trainer
+datasets:
+- Hartunka/processed_wikitext-103-raw-v1-rand-20_v2
 metrics:
 - accuracy
 model-index:
 - name: bert_base_rand_20_v1
-  results: []
+  results:
+  - task:
+      name: Masked Language Modeling
+      type: fill-mask
+    dataset:
+      name: Hartunka/processed_wikitext-103-raw-v1-rand-20_v2
+      type: Hartunka/processed_wikitext-103-raw-v1-rand-20_v2
+    metrics:
+    - name: Accuracy
+      type: accuracy
+      value: 0.15341583319206645
 ---
 
 <!-- This model card has been generated automatically according to the information the Trainer had access to. You
@@ -14,10 +26,10 @@ should probably proofread and complete it, then remove this comment. -->
 
 # bert_base_rand_20_v1
 
-This model is a fine-tuned version of [](https://huggingface.co/) on an unknown dataset.
+This model is a fine-tuned version of [](https://huggingface.co/) on the Hartunka/processed_wikitext-103-raw-v1-rand-20_v2 dataset.
 It achieves the following results on the evaluation set:
-- Loss: 12.3375
-- Accuracy: 0.1518
+- Loss: 9.0827
+- Accuracy: 0.1534
 
 ## Model description
 

all_results.json

@@ -1,16 +1,16 @@
 {
     "epoch": 25.0,
-    "eval_accuracy": 0.15290727241905408,
-    "eval_loss": 9.105982780456543,
-    "eval_runtime": 1.0411,
+    "eval_accuracy": 0.15341583319206645,
+    "eval_loss": 9.082741737365723,
+    "eval_runtime": 1.0384,
     "eval_samples": 479,
-    "eval_samples_per_second": 460.085,
-    "eval_steps_per_second": 4.803,
-    "perplexity": 9009.030845112977,
+    "eval_samples_per_second": 461.307,
+    "eval_steps_per_second": 4.815,
+    "perplexity": 8802.065927180123,
     "total_flos": 1.5151126801503744e+18,
-    "train_loss": 7.635663399364045,
-    "train_runtime": 24680.2989,
+    "train_loss": 7.655053733486172,
+    "train_runtime": 24646.3655,
     "train_samples": 228639,
-    "train_samples_per_second": 231.601,
-    "train_steps_per_second": 2.413
+    "train_samples_per_second": 231.92,
+    "train_steps_per_second": 2.416
 }

eval_results.json

@@ -1,10 +1,10 @@
 {
     "epoch": 25.0,
-    "eval_accuracy": 0.15290727241905408,
-    "eval_loss": 9.105982780456543,
-    "eval_runtime": 1.0411,
+    "eval_accuracy": 0.15341583319206645,
+    "eval_loss": 9.082741737365723,
+    "eval_runtime": 1.0384,
     "eval_samples": 479,
-    "eval_samples_per_second": 460.085,
-    "eval_steps_per_second": 4.803,
-    "perplexity": 9009.030845112977
+    "eval_samples_per_second": 461.307,
+    "eval_steps_per_second": 4.815,
+    "perplexity": 8802.065927180123
 }

train_results.json

@@ -1,9 +1,9 @@
 {
     "epoch": 25.0,
     "total_flos": 1.5151126801503744e+18,
-    "train_loss": 7.635663399364045,
-    "train_runtime": 24680.2989,
+    "train_loss": 7.655053733486172,
+    "train_runtime": 24646.3655,
     "train_samples": 228639,
-    "train_samples_per_second": 231.601,
-    "train_steps_per_second": 2.413
+    "train_samples_per_second": 231.92,
+    "train_steps_per_second": 2.416
 }

trainer_state.json

@@ -1,6 +1,6 @@
 {
   "best_global_step": 10000,
-  "best_metric": 9.088469505310059,
+  "best_metric": 9.063735008239746,
   "best_model_checkpoint": "bert_base_rand_20_v1/checkpoint-10000",
   "epoch": 25.0,
   "eval_steps": 10000,
@@ -11,890 +11,890 @@
   "log_history": [
     {
       "epoch": 0.20990764063811923,
-      "grad_norm": 2.788844347000122,
+      "grad_norm": 3.0303101539611816,
       "learning_rate": 5e-06,
-      "loss": 12.1969,
+      "loss": 12.1977,
       "step": 500
     },
     {
       "epoch": 0.41981528127623846,
-      "grad_norm": 2.116429328918457,
+      "grad_norm": 2.4331130981445312,
       "learning_rate": 1e-05,
-      "loss": 10.5463,
+      "loss": 10.5454,
       "step": 1000
     },
     {
       "epoch": 0.6297229219143576,
-      "grad_norm": 2.5773391723632812,
+      "grad_norm": 1.8983516693115234,
       "learning_rate": 1.5e-05,
-      "loss": 10.007,
+      "loss": 10.0079,
       "step": 1500
     },
     {
       "epoch": 0.8396305625524769,
-      "grad_norm": 2.456319808959961,
+      "grad_norm": 1.7948098182678223,
       "learning_rate": 2e-05,
-      "loss": 9.8521,
+      "loss": 9.8503,
       "step": 2000
     },
     {
       "epoch": 1.0495382031905962,
-      "grad_norm": 3.1209683418273926,
+      "grad_norm": 2.0405168533325195,
       "learning_rate": 2.5e-05,
-      "loss": 9.7313,
+      "loss": 9.7294,
       "step": 2500
     },
     {
       "epoch": 1.2594458438287153,
-      "grad_norm": 2.2184574604034424,
+      "grad_norm": 2.248918056488037,
       "learning_rate": 3e-05,
-      "loss": 9.6159,
+      "loss": 9.6169,
       "step": 3000
     },
     {
       "epoch": 1.4693534844668346,
-      "grad_norm": 3.01564359664917,
+      "grad_norm": 2.2189340591430664,
       "learning_rate": 3.5e-05,
-      "loss": 9.5253,
+      "loss": 9.5259,
       "step": 3500
     },
     {
       "epoch": 1.6792611251049538,
-      "grad_norm": 2.510653018951416,
+      "grad_norm": 1.8111761808395386,
       "learning_rate": 4e-05,
-      "loss": 9.4561,
+      "loss": 9.4574,
       "step": 4000
     },
     {
       "epoch": 1.8891687657430731,
-      "grad_norm": 2.3645122051239014,
+      "grad_norm": 2.3392271995544434,
       "learning_rate": 4.5e-05,
-      "loss": 9.3983,
+      "loss": 9.3994,
       "step": 4500
     },
     {
       "epoch": 2.0990764063811924,
-      "grad_norm": 2.2504115104675293,
+      "grad_norm": 2.3432321548461914,
       "learning_rate": 5e-05,
-      "loss": 9.3482,
+      "loss": 9.3492,
       "step": 5000
     },
     {
       "epoch": 2.3089840470193117,
-      "grad_norm": 2.744396448135376,
+      "grad_norm": 2.1577751636505127,
       "learning_rate": 5.500000000000001e-05,
-      "loss": 9.3084,
+      "loss": 9.3096,
       "step": 5500
     },
     {
       "epoch": 2.5188916876574305,
-      "grad_norm": 2.3779869079589844,
+      "grad_norm": 2.56933856010437,
       "learning_rate": 6e-05,
-      "loss": 9.2706,
+      "loss": 9.2719,
       "step": 6000
     },
     {
       "epoch": 2.72879932829555,
-      "grad_norm": 2.8609466552734375,
+      "grad_norm": 2.7785491943359375,
       "learning_rate": 6.500000000000001e-05,
-      "loss": 9.2375,
+      "loss": 9.2391,
       "step": 6500
     },
     {
       "epoch": 2.938706968933669,
-      "grad_norm": 3.2861833572387695,
+      "grad_norm": 2.7166359424591064,
       "learning_rate": 7e-05,
-      "loss": 9.2088,
+      "loss": 9.2089,
       "step": 7000
     },
     {
       "epoch": 3.1486146095717884,
-      "grad_norm": 2.6351678371429443,
+      "grad_norm": 2.3371636867523193,
       "learning_rate": 7.500000000000001e-05,
-      "loss": 9.1833,
+      "loss": 9.1855,
       "step": 7500
     },
     {
       "epoch": 3.3585222502099077,
-      "grad_norm": 2.62437105178833,
+      "grad_norm": 2.322885036468506,
       "learning_rate": 8e-05,
-      "loss": 9.1637,
+      "loss": 9.1648,
       "step": 8000
     },
     {
       "epoch": 3.568429890848027,
-      "grad_norm": 2.4397714138031006,
+      "grad_norm": 3.072427272796631,
       "learning_rate": 8.5e-05,
-      "loss": 9.1463,
+      "loss": 9.148,
       "step": 8500
     },
     {
       "epoch": 3.7783375314861463,
-      "grad_norm": 2.203540563583374,
+      "grad_norm": 3.082549810409546,
       "learning_rate": 9e-05,
-      "loss": 9.1205,
+      "loss": 9.1222,
       "step": 9000
     },
     {
       "epoch": 3.988245172124265,
-      "grad_norm": 2.892096996307373,
+      "grad_norm": 2.9923737049102783,
       "learning_rate": 9.5e-05,
-      "loss": 9.1048,
+      "loss": 9.1058,
       "step": 9500
     },
     {
       "epoch": 4.198152812762385,
-      "grad_norm": 2.932500123977661,
+      "grad_norm": 3.6475846767425537,
       "learning_rate": 0.0001,
-      "loss": 9.0619,
+      "loss": 9.059,
       "step": 10000
     },
     {
       "epoch": 4.198152812762385,
-      "eval_accuracy": 0.15022677314338184,
-      "eval_loss": 9.088469505310059,
-      "eval_runtime": 1.0303,
-      "eval_samples_per_second": 464.905,
-      "eval_steps_per_second": 4.853,
+      "eval_accuracy": 0.15075544672918,
+      "eval_loss": 9.063735008239746,
+      "eval_runtime": 1.0347,
+      "eval_samples_per_second": 462.944,
+      "eval_steps_per_second": 4.832,
       "step": 10000
     },
     {
       "epoch": 4.408060453400504,
-      "grad_norm": 3.4973864555358887,
+      "grad_norm": 2.6016314029693604,
       "learning_rate": 9.899091826437942e-05,
-      "loss": 9.0499,
+      "loss": 9.0436,
       "step": 10500
     },
     {
       "epoch": 4.617968094038623,
-      "grad_norm": 3.0019209384918213,
+      "grad_norm": 2.505039691925049,
       "learning_rate": 9.798183652875883e-05,
-      "loss": 9.0349,
+      "loss": 9.0327,
       "step": 11000
     },
     {
       "epoch": 4.827875734676742,
-      "grad_norm": 3.0911049842834473,
+      "grad_norm": 3.311245918273926,
       "learning_rate": 9.697275479313825e-05,
-      "loss": 9.0313,
+      "loss": 9.0299,
       "step": 11500
     },
     {
       "epoch": 5.037783375314861,
-      "grad_norm": 3.753380298614502,
+      "grad_norm": 3.3555872440338135,
       "learning_rate": 9.596367305751767e-05,
-      "loss": 8.9948,
+      "loss": 8.9945,
       "step": 12000
     },
     {
       "epoch": 5.247691015952981,
-      "grad_norm": 3.388258218765259,
+      "grad_norm": 3.520192861557007,
       "learning_rate": 9.495459132189707e-05,
-      "loss": 8.9036,
+      "loss": 8.8964,
       "step": 12500
     },
     {
       "epoch": 5.4575986565911,
-      "grad_norm": 3.387521266937256,
+      "grad_norm": 3.8760409355163574,
       "learning_rate": 9.394550958627649e-05,
-      "loss": 8.9114,
+      "loss": 8.9097,
       "step": 13000
     },
     {
       "epoch": 5.667506297229219,
-      "grad_norm": 2.734049081802368,
+      "grad_norm": 3.049903392791748,
       "learning_rate": 9.293642785065591e-05,
-      "loss": 8.9194,
+      "loss": 8.9157,
       "step": 13500
     },
     {
       "epoch": 5.877413937867338,
-      "grad_norm": 3.0717275142669678,
+      "grad_norm": 3.5967395305633545,
       "learning_rate": 9.192734611503532e-05,
-      "loss": 8.9175,
+      "loss": 8.9138,
       "step": 14000
     },
     {
       "epoch": 6.087321578505458,
-      "grad_norm": 2.9726364612579346,
+      "grad_norm": 3.550504684448242,
       "learning_rate": 9.091826437941474e-05,
-      "loss": 8.8464,
+      "loss": 8.8439,
       "step": 14500
     },
     {
       "epoch": 6.297229219143577,
-      "grad_norm": 3.359595537185669,
+      "grad_norm": 2.9916577339172363,
       "learning_rate": 8.990918264379415e-05,
-      "loss": 8.7767,
+      "loss": 8.7789,
       "step": 15000
     },
     {
       "epoch": 6.507136859781696,
-      "grad_norm": 3.2138357162475586,
+      "grad_norm": 2.8933894634246826,
       "learning_rate": 8.890010090817356e-05,
-      "loss": 8.7899,
+      "loss": 8.7911,
       "step": 15500
     },
     {
       "epoch": 6.717044500419815,
-      "grad_norm": 2.9690675735473633,
+      "grad_norm": 3.477431058883667,
       "learning_rate": 8.789101917255298e-05,
-      "loss": 8.8029,
+      "loss": 8.8049,
       "step": 16000
     },
     {
       "epoch": 6.926952141057934,
-      "grad_norm": 3.0250117778778076,
+      "grad_norm": 4.021463871002197,
       "learning_rate": 8.68819374369324e-05,
-      "loss": 8.8139,
+      "loss": 8.8122,
       "step": 16500
     },
     {
       "epoch": 7.136859781696054,
-      "grad_norm": 3.242534875869751,
+      "grad_norm": 3.246682643890381,
       "learning_rate": 8.58728557013118e-05,
-      "loss": 8.6658,
+      "loss": 8.6677,
       "step": 17000
     },
     {
       "epoch": 7.346767422334173,
-      "grad_norm": 4.330350399017334,
+      "grad_norm": 3.860541582107544,
       "learning_rate": 8.486377396569122e-05,
-      "loss": 8.6172,
+      "loss": 8.6197,
       "step": 17500
     },
     {
       "epoch": 7.5566750629722925,
-      "grad_norm": 3.748150110244751,
+      "grad_norm": 3.9295053482055664,
       "learning_rate": 8.385469223007064e-05,
-      "loss": 8.647,
+      "loss": 8.6517,
       "step": 18000
     },
     {
       "epoch": 7.766582703610411,
-      "grad_norm": 3.4176602363586426,
+      "grad_norm": 3.767359972000122,
       "learning_rate": 8.284561049445006e-05,
-      "loss": 8.6641,
+      "loss": 8.6665,
       "step": 18500
     },
     {
       "epoch": 7.97649034424853,
-      "grad_norm": 4.2462005615234375,
+      "grad_norm": 3.7014143466949463,
       "learning_rate": 8.183652875882947e-05,
-      "loss": 8.6723,
+      "loss": 8.6741,
       "step": 19000
     },
     {
       "epoch": 8.18639798488665,
-      "grad_norm": 4.500391483306885,
+      "grad_norm": 4.395498752593994,
       "learning_rate": 8.082744702320889e-05,
-      "loss": 8.3854,
+      "loss": 8.3858,
       "step": 19500
     },
     {
       "epoch": 8.39630562552477,
-      "grad_norm": 5.132614612579346,
+      "grad_norm": 4.721158981323242,
       "learning_rate": 7.98183652875883e-05,
-      "loss": 8.3754,
+      "loss": 8.3813,
       "step": 20000
     },
     {
       "epoch": 8.39630562552477,
-      "eval_accuracy": 0.15306265607901456,
-      "eval_loss": 9.548609733581543,
-      "eval_runtime": 1.012,
-      "eval_samples_per_second": 473.323,
-      "eval_steps_per_second": 4.941,
+      "eval_accuracy": 0.15216476332108084,
+      "eval_loss": 9.491806030273438,
+      "eval_runtime": 1.0152,
+      "eval_samples_per_second": 471.811,
+      "eval_steps_per_second": 4.925,
       "step": 20000
     },
     {
       "epoch": 8.606213266162888,
-      "grad_norm": 5.027251243591309,
+      "grad_norm": 5.142504692077637,
       "learning_rate": 7.880928355196771e-05,
-      "loss": 8.3966,
+      "loss": 8.4025,
       "step": 20500
     },
     {
       "epoch": 8.816120906801007,
-      "grad_norm": 4.8901591300964355,
+      "grad_norm": 4.701345443725586,
       "learning_rate": 7.780020181634713e-05,
-      "loss": 8.4262,
+      "loss": 8.4315,
       "step": 21000
     },
     {
       "epoch": 9.026028547439127,
-      "grad_norm": 6.076493740081787,
+      "grad_norm": 6.819747447967529,
       "learning_rate": 7.679112008072655e-05,
-      "loss": 8.3814,
+      "loss": 8.3835,
       "step": 21500
     },
     {
       "epoch": 9.235936188077247,
-      "grad_norm": 5.670348167419434,
+      "grad_norm": 6.080567359924316,
       "learning_rate": 7.578203834510595e-05,
-      "loss": 8.0239,
+      "loss": 8.0291,
       "step": 22000
     },
     {
       "epoch": 9.445843828715365,
-      "grad_norm": 5.736623287200928,
+      "grad_norm": 6.5088677406311035,
       "learning_rate": 7.477295660948537e-05,
-      "loss": 8.0512,
+      "loss": 8.055,
       "step": 22500
     },
     {
       "epoch": 9.655751469353484,
-      "grad_norm": 5.6201863288879395,
+      "grad_norm": 5.474933624267578,
       "learning_rate": 7.376387487386479e-05,
-      "loss": 8.0952,
+      "loss": 8.1011,
       "step": 23000
     },
     {
       "epoch": 9.865659109991604,
-      "grad_norm": 5.4278106689453125,
+      "grad_norm": 6.140969276428223,
       "learning_rate": 7.27547931382442e-05,
-      "loss": 8.105,
+      "loss": 8.1063,
       "step": 23500
     },
     {
       "epoch": 10.075566750629722,
-      "grad_norm": 5.7619829177856445,
+      "grad_norm": 5.9476165771484375,
       "learning_rate": 7.174571140262362e-05,
-      "loss": 7.9669,
+      "loss": 7.9752,
       "step": 24000
     },
     {
       "epoch": 10.285474391267842,
-      "grad_norm": 6.371272563934326,
+      "grad_norm": 5.563359260559082,
       "learning_rate": 7.073662966700304e-05,
-      "loss": 7.7393,
+      "loss": 7.7502,
       "step": 24500
     },
     {
       "epoch": 10.495382031905962,
-      "grad_norm": 5.298580646514893,
+      "grad_norm": 7.092005252838135,
       "learning_rate": 6.972754793138244e-05,
-      "loss": 7.7664,
+      "loss": 7.7686,
       "step": 25000
     },
     {
       "epoch": 10.705289672544081,
-      "grad_norm": 7.094499111175537,
+      "grad_norm": 6.459993362426758,
       "learning_rate": 6.871846619576186e-05,
-      "loss": 7.803,
+      "loss": 7.8079,
       "step": 25500
     },
     {
       "epoch": 10.9151973131822,
-      "grad_norm": 6.834150791168213,
+      "grad_norm": 7.596024990081787,
       "learning_rate": 6.770938446014128e-05,
-      "loss": 7.8322,
+      "loss": 7.8431,
       "step": 26000
     },
     {
       "epoch": 11.125104953820319,
-      "grad_norm": 5.807174205780029,
+      "grad_norm": 7.701043605804443,
       "learning_rate": 6.670030272452068e-05,
-      "loss": 7.6204,
+      "loss": 7.6226,
       "step": 26500
     },
     {
       "epoch": 11.335012594458439,
-      "grad_norm": 6.464725017547607,
+      "grad_norm": 7.442851543426514,
       "learning_rate": 6.56912209889001e-05,
-      "loss": 7.5096,
+      "loss": 7.5202,
       "step": 27000
     },
     {
       "epoch": 11.544920235096557,
-      "grad_norm": 7.25095796585083,
+      "grad_norm": 7.431483745574951,
       "learning_rate": 6.468213925327952e-05,
-      "loss": 7.5471,
+      "loss": 7.5567,
       "step": 27500
     },
     {
       "epoch": 11.754827875734676,
-      "grad_norm": 8.751776695251465,
+      "grad_norm": 8.853947639465332,
       "learning_rate": 6.367305751765893e-05,
-      "loss": 7.5654,
+      "loss": 7.5767,
       "step": 28000
     },
     {
       "epoch": 11.964735516372796,
-      "grad_norm": 9.918253898620605,
+      "grad_norm": 7.045031547546387,
       "learning_rate": 6.266397578203835e-05,
-      "loss": 7.5977,
+      "loss": 7.6044,
       "step": 28500
     },
     {
       "epoch": 12.174643157010916,
-      "grad_norm": 6.876755714416504,
+      "grad_norm": 7.04722785949707,
       "learning_rate": 6.165489404641777e-05,
-      "loss": 7.3312,
+      "loss": 7.3292,
       "step": 29000
     },
     {
       "epoch": 12.384550797649034,
-      "grad_norm": 7.298816680908203,
+      "grad_norm": 7.797435760498047,
       "learning_rate": 6.064581231079718e-05,
-      "loss": 7.2921,
+      "loss": 7.2958,
       "step": 29500
     },
     {
       "epoch": 12.594458438287154,
-      "grad_norm": 6.345142364501953,
+      "grad_norm": 7.942636489868164,
       "learning_rate": 5.963673057517659e-05,
-      "loss": 7.3395,
+      "loss": 7.3399,
       "step": 30000
     },
     {
       "epoch": 12.594458438287154,
-      "eval_accuracy": 0.15351036740632956,
-      "eval_loss": 10.677895545959473,
-      "eval_runtime": 1.0178,
-      "eval_samples_per_second": 470.607,
-      "eval_steps_per_second": 4.912,
+      "eval_accuracy": 0.15365027842292303,
+      "eval_loss": 10.872977256774902,
+      "eval_runtime": 1.0173,
+      "eval_samples_per_second": 470.868,
+      "eval_steps_per_second": 4.915,
       "step": 30000
     },
     {
       "epoch": 12.804366078925273,
-      "grad_norm": 9.623331069946289,
+      "grad_norm": 6.244444847106934,
       "learning_rate": 5.8627648839556e-05,
-      "loss": 7.3548,
+      "loss": 7.366,
       "step": 30500
     },
     {
       "epoch": 13.014273719563391,
-      "grad_norm": 7.847203731536865,
+      "grad_norm": 7.656332969665527,
       "learning_rate": 5.7618567103935416e-05,
-      "loss": 7.3521,
+      "loss": 7.3598,
       "step": 31000
     },
     {
       "epoch": 13.224181360201511,
-      "grad_norm": 8.091148376464844,
+      "grad_norm": 8.64603328704834,
       "learning_rate": 5.660948536831484e-05,
-      "loss": 7.0706,
+      "loss": 7.0718,
       "step": 31500
     },
     {
       "epoch": 13.43408900083963,
-      "grad_norm": 8.715632438659668,
+      "grad_norm": 9.710521697998047,
       "learning_rate": 5.5600403632694254e-05,
-      "loss": 7.1087,
+      "loss": 7.1131,
       "step": 32000
     },
     {
       "epoch": 13.64399664147775,
-      "grad_norm": 7.3595123291015625,
+      "grad_norm": 7.6166486740112305,
       "learning_rate": 5.4591321897073666e-05,
-      "loss": 7.1468,
+      "loss": 7.1517,
       "step": 32500
     },
     {
       "epoch": 13.853904282115868,
-      "grad_norm": 8.611331939697266,
+      "grad_norm": 6.435095310211182,
       "learning_rate": 5.358224016145308e-05,
-      "loss": 7.152,
+      "loss": 7.1658,
       "step": 33000
     },
     {
       "epoch": 14.063811922753988,
-      "grad_norm": 7.117791175842285,
+      "grad_norm": 6.226130485534668,
       "learning_rate": 5.257315842583249e-05,
-      "loss": 7.0842,
+      "loss": 7.0913,
       "step": 33500
     },
     {
       "epoch": 14.273719563392108,
-      "grad_norm": 7.401384353637695,
+      "grad_norm": 7.975910186767578,
       "learning_rate": 5.15640766902119e-05,
-      "loss": 6.9085,
+      "loss": 6.9147,
       "step": 34000
     },
     {
       "epoch": 14.483627204030226,
-      "grad_norm": 6.674845218658447,
+      "grad_norm": 9.481583595275879,
       "learning_rate": 5.055499495459133e-05,
-      "loss": 6.9509,
+      "loss": 6.9445,
       "step": 34500
     },
     {
       "epoch": 14.693534844668346,
-      "grad_norm": 9.794930458068848,
+      "grad_norm": 7.345225811004639,
       "learning_rate": 4.954591321897074e-05,
-      "loss": 6.9669,
+      "loss": 6.9676,
       "step": 35000
     },
     {
       "epoch": 14.903442485306465,
-      "grad_norm": 8.372832298278809,
+      "grad_norm": 7.864171981811523,
       "learning_rate": 4.853683148335015e-05,
-      "loss": 6.9779,
+      "loss": 6.9842,
       "step": 35500
     },
     {
       "epoch": 15.113350125944585,
-      "grad_norm": 7.102290630340576,
+      "grad_norm": 10.979301452636719,
       "learning_rate": 4.7527749747729566e-05,
-      "loss": 6.8617,
+      "loss": 6.8731,
       "step": 36000
     },
     {
       "epoch": 15.323257766582703,
-      "grad_norm": 7.259284496307373,
+      "grad_norm": 8.653640747070312,
       "learning_rate": 4.6518668012108985e-05,
-      "loss": 6.7725,
+      "loss": 6.7832,
       "step": 36500
     },
     {
       "epoch": 15.533165407220823,
-      "grad_norm": 10.708537101745605,
+      "grad_norm": 7.870504379272461,
       "learning_rate": 4.55095862764884e-05,
-      "loss": 6.8013,
+      "loss": 6.7981,
       "step": 37000
     },
     {
       "epoch": 15.743073047858942,
-      "grad_norm": 9.083553314208984,
+      "grad_norm": 8.514162063598633,
       "learning_rate": 4.450050454086781e-05,
-      "loss": 6.8136,
+      "loss": 6.8266,
       "step": 37500
     },
     {
       "epoch": 15.95298068849706,
-      "grad_norm": 6.711920738220215,
+      "grad_norm": 10.380403518676758,
       "learning_rate": 4.349142280524723e-05,
-      "loss": 6.8337,
+      "loss": 6.8385,
       "step": 38000
     },
     {
       "epoch": 16.162888329135182,
-      "grad_norm": 7.46658182144165,
+      "grad_norm": 9.118815422058105,
       "learning_rate": 4.248234106962664e-05,
-      "loss": 6.6739,
+      "loss": 6.6982,
       "step": 38500
     },
     {
       "epoch": 16.3727959697733,
-      "grad_norm": 7.668158054351807,
+      "grad_norm": 8.186168670654297,
       "learning_rate": 4.147325933400605e-05,
-      "loss": 6.6617,
+      "loss": 6.6818,
       "step": 39000
     },
     {
       "epoch": 16.582703610411418,
-      "grad_norm": 9.162720680236816,
+      "grad_norm": 9.249139785766602,
       "learning_rate": 4.046417759838547e-05,
-      "loss": 6.6621,
+      "loss": 6.6891,
       "step": 39500
     },
     {
       "epoch": 16.79261125104954,
-      "grad_norm": 8.438434600830078,
+      "grad_norm": 9.438446998596191,
       "learning_rate": 3.9455095862764884e-05,
-      "loss": 6.68,
+      "loss": 6.6933,
       "step": 40000
     },
     {
       "epoch": 16.79261125104954,
-      "eval_accuracy": 0.1523774812156555,
-      "eval_loss": 11.671748161315918,
-      "eval_runtime": 1.0169,
-      "eval_samples_per_second": 471.05,
-      "eval_steps_per_second": 4.917,
+      "eval_accuracy": 0.15117191880677358,
+      "eval_loss": 11.692638397216797,
+      "eval_runtime": 1.0214,
+      "eval_samples_per_second": 468.985,
+      "eval_steps_per_second": 4.895,
       "step": 40000
     },
     {
       "epoch": 17.002518891687657,
-      "grad_norm": 7.6917266845703125,
+      "grad_norm": 6.471468925476074,
       "learning_rate": 3.84460141271443e-05,
-      "loss": 6.6957,
+      "loss": 6.718,
       "step": 40500
     },
     {
       "epoch": 17.212426532325775,
-      "grad_norm": 5.1674299240112305,
+      "grad_norm": 7.857778549194336,
       "learning_rate": 3.7436932391523715e-05,
-      "loss": 6.532,
+      "loss": 6.5762,
       "step": 41000
     },
     {
       "epoch": 17.422334172963897,
-      "grad_norm": 7.299459934234619,
+      "grad_norm": 8.556367874145508,
       "learning_rate": 3.642785065590313e-05,
-      "loss": 6.5489,
+      "loss": 6.59,
       "step": 41500
     },
     {
       "epoch": 17.632241813602015,
-      "grad_norm": 8.432929039001465,
+      "grad_norm": 9.771214485168457,
       "learning_rate": 3.541876892028255e-05,
-      "loss": 6.5588,
+      "loss": 6.6022,
       "step": 42000
     },
     {
       "epoch": 17.842149454240133,
-      "grad_norm": 8.720917701721191,
+      "grad_norm": 7.696030616760254,
       "learning_rate": 3.440968718466196e-05,
-      "loss": 6.5819,
+      "loss": 6.6177,
       "step": 42500
     },
     {
       "epoch": 18.052057094878254,
-      "grad_norm": 6.441305160522461,
+      "grad_norm": 6.658695697784424,
       "learning_rate": 3.340060544904137e-05,
-      "loss": 6.536,
+      "loss": 6.5793,
       "step": 43000
     },
     {
       "epoch": 18.261964735516372,
-      "grad_norm": 6.2586212158203125,
+      "grad_norm": 5.75900411605835,
       "learning_rate": 3.239152371342079e-05,
-      "loss": 6.443,
+      "loss": 6.513,
       "step": 43500
     },
     {
       "epoch": 18.471872376154494,
-      "grad_norm": 7.705145359039307,
+      "grad_norm": 7.845323085784912,
       "learning_rate": 3.13824419778002e-05,
-      "loss": 6.4618,
+      "loss": 6.5133,
       "step": 44000
     },
     {
       "epoch": 18.68178001679261,
-      "grad_norm": 6.204861164093018,
+      "grad_norm": 7.73126745223999,
       "learning_rate": 3.0373360242179615e-05,
-      "loss": 6.4703,
+      "loss": 6.5342,
       "step": 44500
     },
     {
       "epoch": 18.89168765743073,
-      "grad_norm": 7.980417728424072,
+      "grad_norm": 5.368079662322998,
       "learning_rate": 2.9364278506559034e-05,
-      "loss": 6.4856,
+      "loss": 6.5397,
       "step": 45000
     },
     {
       "epoch": 19.10159529806885,
-      "grad_norm": 8.686908721923828,
+      "grad_norm": 5.72926139831543,
       "learning_rate": 2.8355196770938446e-05,
-      "loss": 6.4174,
+      "loss": 6.4836,
       "step": 45500
     },
     {
       "epoch": 19.31150293870697,
-      "grad_norm": 8.06748104095459,
+      "grad_norm": 6.304308891296387,
       "learning_rate": 2.734611503531786e-05,
-      "loss": 6.3644,
+      "loss": 6.4446,
       "step": 46000
     },
     {
       "epoch": 19.521410579345087,
-      "grad_norm": 6.80294942855835,
+      "grad_norm": 7.8987298011779785,
       "learning_rate": 2.6337033299697278e-05,
-      "loss": 6.3693,
+      "loss": 6.4487,
       "step": 46500
     },
     {
       "epoch": 19.73131821998321,
-      "grad_norm": 5.868018627166748,
+      "grad_norm": 6.950328350067139,
       "learning_rate": 2.532795156407669e-05,
-      "loss": 6.3747,
+      "loss": 6.45,
       "step": 47000
     },
     {
       "epoch": 19.941225860621326,
-      "grad_norm": 7.769217491149902,
+      "grad_norm": 7.982358932495117,
       "learning_rate": 2.4318869828456106e-05,
-      "loss": 6.3798,
+      "loss": 6.4512,
       "step": 47500
     },
     {
       "epoch": 20.151133501259444,
-      "grad_norm": 7.198081970214844,
+      "grad_norm": 6.615007400512695,
       "learning_rate": 2.330978809283552e-05,
-      "loss": 6.328,
+      "loss": 6.4027,
       "step": 48000
     },
     {
       "epoch": 20.361041141897566,
-      "grad_norm": 8.08232593536377,
+      "grad_norm": 9.836413383483887,
       "learning_rate": 2.2300706357214934e-05,
-      "loss": 6.2978,
+      "loss": 6.3731,
       "step": 48500
     },
     {
       "epoch": 20.570948782535684,
-      "grad_norm": 7.389130115509033,
+      "grad_norm": 3.978142738342285,
       "learning_rate": 2.129162462159435e-05,
-      "loss": 6.2922,
+      "loss": 6.3711,
       "step": 49000
     },
     {
       "epoch": 20.7808564231738,
-      "grad_norm": 7.787246227264404,
+      "grad_norm": 5.0921630859375,
       "learning_rate": 2.0282542885973765e-05,
-      "loss": 6.2921,
+      "loss": 6.3797,
       "step": 49500
     },
     {
       "epoch": 20.990764063811923,
-      "grad_norm": 6.166048526763916,
+      "grad_norm": 5.254364490509033,
       "learning_rate": 1.927346115035318e-05,
-      "loss": 6.3016,
+      "loss": 6.3773,
       "step": 50000
     },
     {
       "epoch": 20.990764063811923,
-      "eval_accuracy": 0.1517929738974402,
-      "eval_loss": 12.51880168914795,
-      "eval_runtime": 1.0243,
-      "eval_samples_per_second": 467.648,
+      "eval_accuracy": 0.15184904813973701,
+      "eval_loss": 12.33754825592041,
+      "eval_runtime": 1.0241,
+      "eval_samples_per_second": 467.713,
       "eval_steps_per_second": 4.882,
       "step": 50000
     },
     {
       "epoch": 21.20067170445004,
-      "grad_norm": 7.7119646072387695,
+      "grad_norm": 6.045699119567871,
       "learning_rate": 1.8264379414732593e-05,
-      "loss": 6.2415,
+      "loss": 6.303,
       "step": 50500
     },
     {
       "epoch": 21.410579345088163,
-      "grad_norm": 8.19467830657959,
+      "grad_norm": 8.00769329071045,
       "learning_rate": 1.725529767911201e-05,
-      "loss": 6.2444,
+      "loss": 6.3101,
       "step": 51000
     },
     {
       "epoch": 21.62048698572628,
-      "grad_norm": 7.477888107299805,
+      "grad_norm": 7.591960906982422,
       "learning_rate": 1.6246215943491424e-05,
-      "loss": 6.2466,
+      "loss": 6.307,
       "step": 51500
     },
     {
       "epoch": 21.8303946263644,
-      "grad_norm": 9.23111629486084,
+      "grad_norm": 8.466202735900879,
       "learning_rate": 1.5237134207870837e-05,
-      "loss": 6.2426,
+      "loss": 6.3103,
       "step": 52000
     },
     {
       "epoch": 22.04030226700252,
-      "grad_norm": 5.021499156951904,
+      "grad_norm": 6.745018482208252,
       "learning_rate": 1.4228052472250252e-05,
-      "loss": 6.2322,
+      "loss": 6.2889,
       "step": 52500
     },
     {
       "epoch": 22.250209907640638,
-      "grad_norm": 6.123244762420654,
+      "grad_norm": 4.916234493255615,
       "learning_rate": 1.3218970736629668e-05,
-      "loss": 6.1935,
+      "loss": 6.2402,
       "step": 53000
     },
     {
       "epoch": 22.460117548278756,
-      "grad_norm": 7.231016159057617,
+      "grad_norm": 9.260708808898926,
       "learning_rate": 1.2209889001009082e-05,
-      "loss": 6.1908,
+      "loss": 6.2379,
       "step": 53500
     },
     {
       "epoch": 22.670025188916878,
-      "grad_norm": 6.407002925872803,
+      "grad_norm": 8.743579864501953,
       "learning_rate": 1.1200807265388496e-05,
-      "loss": 6.2041,
+      "loss": 6.2426,
       "step": 54000
     },
     {
       "epoch": 22.879932829554996,
-      "grad_norm": 4.805995941162109,
+      "grad_norm": 5.662677764892578,
       "learning_rate": 1.0191725529767912e-05,
-      "loss": 6.1923,
+      "loss": 6.2395,
       "step": 54500
     },
     {
       "epoch": 23.089840470193113,
-      "grad_norm": 3.93101167678833,
+      "grad_norm": 7.829682350158691,
       "learning_rate": 9.182643794147326e-06,
-      "loss": 6.1766,
+      "loss": 6.2224,
       "step": 55000
     },
     {
       "epoch": 23.299748110831235,
-      "grad_norm": 7.635568618774414,
+      "grad_norm": 6.931691646575928,
       "learning_rate": 8.173562058526741e-06,
-      "loss": 6.1604,
+      "loss": 6.1901,
       "step": 55500
     },
     {
       "epoch": 23.509655751469353,
-      "grad_norm": 4.48831844329834,
+      "grad_norm": 6.559401988983154,
       "learning_rate": 7.164480322906156e-06,
-      "loss": 6.1716,
+      "loss": 6.1994,
       "step": 56000
     },
     {
       "epoch": 23.719563392107474,
-      "grad_norm": 7.754641532897949,
+      "grad_norm": 4.310472011566162,
       "learning_rate": 6.155398587285571e-06,
-      "loss": 6.1572,
+      "loss": 6.1863,
       "step": 56500
     },
     {
       "epoch": 23.929471032745592,
-      "grad_norm": 8.06408405303955,
+      "grad_norm": 5.716533184051514,
       "learning_rate": 5.146316851664986e-06,
-      "loss": 6.1613,
+      "loss": 6.1963,
       "step": 57000
     },
     {
       "epoch": 24.13937867338371,
-      "grad_norm": 5.528585433959961,
+      "grad_norm": 7.209354400634766,
       "learning_rate": 4.1372351160444e-06,
-      "loss": 6.1488,
+      "loss": 6.1719,
       "step": 57500
     },
     {
       "epoch": 24.349286314021832,
-      "grad_norm": 8.36520004272461,
+      "grad_norm": 7.885115623474121,
       "learning_rate": 3.1281533804238145e-06,
-      "loss": 6.1515,
+      "loss": 6.1706,
       "step": 58000
     },
     {
       "epoch": 24.55919395465995,
-      "grad_norm": 5.141292095184326,
+      "grad_norm": 5.144560813903809,
       "learning_rate": 2.1190716448032293e-06,
-      "loss": 6.1481,
+      "loss": 6.1735,
       "step": 58500
     },
     {
       "epoch": 24.769101595298068,
-      "grad_norm": 5.405873775482178,
+      "grad_norm": 6.012503623962402,
       "learning_rate": 1.109989909182644e-06,
-      "loss": 6.1381,
+      "loss": 6.1674,
       "step": 59000
     },
     {
       "epoch": 24.97900923593619,
-      "grad_norm": 5.042454242706299,
+      "grad_norm": 6.069438457489014,
       "learning_rate": 1.0090817356205853e-07,
-      "loss": 6.1344,
+      "loss": 6.1569,
       "step": 59500
     },
     {
       "epoch": 25.0,
       "step": 59550,
       "total_flos": 1.5151126801503744e+18,
-      "train_loss": 7.635663399364045,
-      "train_runtime": 24680.2989,
-      "train_samples_per_second": 231.601,
-      "train_steps_per_second": 2.413
+      "train_loss": 7.655053733486172,
+      "train_runtime": 24646.3655,
+      "train_samples_per_second": 231.92,
+      "train_steps_per_second": 2.416
     }
   ],
   "logging_steps": 500,