End of training

README.md

@@ -2,11 +2,23 @@
 library_name: transformers
 tags:
 - generated_from_trainer
+datasets:
+- Hartunka/processed_wikitext-103-raw-v1-km-20_v2
 metrics:
 - accuracy
 model-index:
 - name: bert_base_km_20_v2
-  results: []
+  results:
+  - task:
+      name: Masked Language Modeling
+      type: fill-mask
+    dataset:
+      name: Hartunka/processed_wikitext-103-raw-v1-km-20_v2
+      type: Hartunka/processed_wikitext-103-raw-v1-km-20_v2
+    metrics:
+    - name: Accuracy
+      type: accuracy
+      value: 0.15383963383624344
 ---
 
 <!-- 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_km_20_v2
 
-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-km-20_v2 dataset.
 It achieves the following results on the evaluation set:
-- Loss: 6.6477
-- Accuracy: 0.1524
+- Loss: 6.4338
+- Accuracy: 0.1538
 
 ## Model description
 

all_results.json

@@ -1,16 +1,16 @@
 {
     "epoch": 25.0,
-    "eval_accuracy": 0.15400915409391422,
-    "eval_loss": 6.723025321960449,
-    "eval_runtime": 1.0283,
+    "eval_accuracy": 0.15383963383624344,
+    "eval_loss": 6.433831214904785,
+    "eval_runtime": 1.0615,
     "eval_samples": 479,
-    "eval_samples_per_second": 465.802,
-    "eval_steps_per_second": 4.862,
-    "perplexity": 831.3287480343491,
+    "eval_samples_per_second": 451.232,
+    "eval_steps_per_second": 4.71,
+    "perplexity": 622.5545250250817,
     "total_flos": 1.5151126801503744e+18,
-    "train_loss": 6.378149575490495,
-    "train_runtime": 24807.4865,
+    "train_loss": 6.324693534844668,
+    "train_runtime": 24831.9556,
     "train_samples": 228639,
-    "train_samples_per_second": 230.413,
-    "train_steps_per_second": 2.4
+    "train_samples_per_second": 230.186,
+    "train_steps_per_second": 2.398
 }

eval_results.json

@@ -1,10 +1,10 @@
 {
     "epoch": 25.0,
-    "eval_accuracy": 0.15400915409391422,
-    "eval_loss": 6.723025321960449,
-    "eval_runtime": 1.0283,
+    "eval_accuracy": 0.15383963383624344,
+    "eval_loss": 6.433831214904785,
+    "eval_runtime": 1.0615,
     "eval_samples": 479,
-    "eval_samples_per_second": 465.802,
-    "eval_steps_per_second": 4.862,
-    "perplexity": 831.3287480343491
+    "eval_samples_per_second": 451.232,
+    "eval_steps_per_second": 4.71,
+    "perplexity": 622.5545250250817
 }

train_results.json

@@ -1,9 +1,9 @@
 {
     "epoch": 25.0,
     "total_flos": 1.5151126801503744e+18,
-    "train_loss": 6.378149575490495,
-    "train_runtime": 24807.4865,
+    "train_loss": 6.324693534844668,
+    "train_runtime": 24831.9556,
     "train_samples": 228639,
-    "train_samples_per_second": 230.413,
-    "train_steps_per_second": 2.4
+    "train_samples_per_second": 230.186,
+    "train_steps_per_second": 2.398
 }

trainer_state.json

@@ -1,6 +1,6 @@
 {
   "best_global_step": 20000,
-  "best_metric": 6.701448440551758,
+  "best_metric": 6.480999946594238,
   "best_model_checkpoint": "bert_base_km_20_v2/checkpoint-20000",
   "epoch": 25.0,
   "eval_steps": 10000,
@@ -11,890 +11,890 @@
   "log_history": [
     {
       "epoch": 0.20990764063811923,
-      "grad_norm": 3.2586686611175537,
+      "grad_norm": 3.3506252765655518,
       "learning_rate": 5e-06,
-      "loss": 12.1717,
+      "loss": 12.1648,
       "step": 500
     },
     {
       "epoch": 0.41981528127623846,
-      "grad_norm": 4.922037601470947,
+      "grad_norm": 5.457421779632568,
       "learning_rate": 1e-05,
-      "loss": 10.0311,
+      "loss": 10.0986,
       "step": 1000
     },
     {
       "epoch": 0.6297229219143576,
-      "grad_norm": 28.822181701660156,
+      "grad_norm": 23.159326553344727,
       "learning_rate": 1.5e-05,
-      "loss": 8.5295,
+      "loss": 8.7021,
       "step": 1500
     },
     {
       "epoch": 0.8396305625524769,
-      "grad_norm": 13.167506217956543,
+      "grad_norm": 8.975157737731934,
       "learning_rate": 2e-05,
-      "loss": 8.0722,
+      "loss": 8.1464,
       "step": 2000
     },
     {
       "epoch": 1.0495382031905962,
-      "grad_norm": 16.0736141204834,
+      "grad_norm": 10.226245880126953,
       "learning_rate": 2.5e-05,
-      "loss": 7.8283,
+      "loss": 7.8208,
       "step": 2500
     },
     {
       "epoch": 1.2594458438287153,
-      "grad_norm": 14.928688049316406,
+      "grad_norm": 9.012547492980957,
       "learning_rate": 3e-05,
-      "loss": 7.6626,
+      "loss": 7.5779,
       "step": 3000
     },
     {
       "epoch": 1.4693534844668346,
-      "grad_norm": 6.039346218109131,
+      "grad_norm": 8.201004028320312,
       "learning_rate": 3.5e-05,
-      "loss": 7.5058,
+      "loss": 7.4349,
       "step": 3500
     },
     {
       "epoch": 1.6792611251049538,
-      "grad_norm": 12.346041679382324,
+      "grad_norm": 6.068079471588135,
       "learning_rate": 4e-05,
-      "loss": 7.3977,
+      "loss": 7.3029,
       "step": 4000
     },
     {
       "epoch": 1.8891687657430731,
-      "grad_norm": 6.854773044586182,
+      "grad_norm": 4.390971660614014,
       "learning_rate": 4.5e-05,
-      "loss": 7.3026,
+      "loss": 7.189,
       "step": 4500
     },
     {
       "epoch": 2.0990764063811924,
-      "grad_norm": 7.6127848625183105,
+      "grad_norm": 4.879016399383545,
       "learning_rate": 5e-05,
-      "loss": 7.1895,
+      "loss": 7.0702,
       "step": 5000
     },
     {
       "epoch": 2.3089840470193117,
-      "grad_norm": 7.826136589050293,
+      "grad_norm": 4.469964981079102,
       "learning_rate": 5.500000000000001e-05,
-      "loss": 7.1128,
+      "loss": 6.9738,
       "step": 5500
     },
     {
       "epoch": 2.5188916876574305,
-      "grad_norm": 8.711495399475098,
+      "grad_norm": 4.512015342712402,
       "learning_rate": 6e-05,
-      "loss": 7.0718,
+      "loss": 6.9463,
       "step": 6000
     },
     {
       "epoch": 2.72879932829555,
-      "grad_norm": 4.3706536293029785,
+      "grad_norm": 3.9740757942199707,
       "learning_rate": 6.500000000000001e-05,
-      "loss": 7.0242,
+      "loss": 6.8989,
       "step": 6500
     },
     {
       "epoch": 2.938706968933669,
-      "grad_norm": 7.588512420654297,
+      "grad_norm": 4.45874547958374,
       "learning_rate": 7e-05,
-      "loss": 6.9993,
+      "loss": 6.8596,
       "step": 7000
     },
     {
       "epoch": 3.1486146095717884,
-      "grad_norm": 6.372596740722656,
+      "grad_norm": 3.5402514934539795,
       "learning_rate": 7.500000000000001e-05,
-      "loss": 6.9168,
+      "loss": 6.7614,
       "step": 7500
     },
     {
       "epoch": 3.3585222502099077,
-      "grad_norm": 5.352084159851074,
+      "grad_norm": 3.2395708560943604,
       "learning_rate": 8e-05,
-      "loss": 6.8948,
+      "loss": 6.7273,
       "step": 8000
     },
     {
       "epoch": 3.568429890848027,
-      "grad_norm": 4.005120277404785,
+      "grad_norm": 2.3133621215820312,
       "learning_rate": 8.5e-05,
-      "loss": 6.8742,
+      "loss": 6.737,
       "step": 8500
     },
     {
       "epoch": 3.7783375314861463,
-      "grad_norm": 3.1480915546417236,
+      "grad_norm": 2.676318883895874,
       "learning_rate": 9e-05,
-      "loss": 6.8502,
+      "loss": 6.7239,
       "step": 9000
     },
     {
       "epoch": 3.988245172124265,
-      "grad_norm": 2.2671101093292236,
+      "grad_norm": 2.319993734359741,
       "learning_rate": 9.5e-05,
-      "loss": 6.8378,
+      "loss": 6.7111,
       "step": 9500
     },
     {
       "epoch": 4.198152812762385,
-      "grad_norm": 3.7820088863372803,
+      "grad_norm": 4.243929386138916,
       "learning_rate": 0.0001,
-      "loss": 6.741,
+      "loss": 6.6058,
       "step": 10000
     },
     {
       "epoch": 4.198152812762385,
-      "eval_accuracy": 0.15047719747349675,
-      "eval_loss": 6.86174201965332,
-      "eval_runtime": 1.0657,
-      "eval_samples_per_second": 449.453,
-      "eval_steps_per_second": 4.692,
+      "eval_accuracy": 0.1499485238876986,
+      "eval_loss": 6.702857971191406,
+      "eval_runtime": 1.0645,
+      "eval_samples_per_second": 449.956,
+      "eval_steps_per_second": 4.697,
       "step": 10000
     },
     {
       "epoch": 4.408060453400504,
-      "grad_norm": 2.66556978225708,
+      "grad_norm": 4.093377590179443,
       "learning_rate": 9.899091826437942e-05,
-      "loss": 6.741,
+      "loss": 6.6124,
       "step": 10500
     },
     {
       "epoch": 4.617968094038623,
-      "grad_norm": 3.0062286853790283,
+      "grad_norm": 3.2889018058776855,
       "learning_rate": 9.798183652875883e-05,
-      "loss": 6.7367,
+      "loss": 6.6077,
       "step": 11000
     },
     {
       "epoch": 4.827875734676742,
-      "grad_norm": 2.927842855453491,
+      "grad_norm": 2.30513334274292,
       "learning_rate": 9.697275479313825e-05,
-      "loss": 6.7229,
+      "loss": 6.6029,
       "step": 11500
     },
     {
       "epoch": 5.037783375314861,
-      "grad_norm": 3.1912167072296143,
+      "grad_norm": 2.541769027709961,
       "learning_rate": 9.596367305751767e-05,
-      "loss": 6.6724,
+      "loss": 6.5729,
       "step": 12000
     },
     {
       "epoch": 5.247691015952981,
-      "grad_norm": 3.3745853900909424,
+      "grad_norm": 2.6730828285217285,
       "learning_rate": 9.495459132189707e-05,
-      "loss": 6.6049,
+      "loss": 6.4829,
       "step": 12500
     },
     {
       "epoch": 5.4575986565911,
-      "grad_norm": 2.804530382156372,
+      "grad_norm": 2.574977159500122,
       "learning_rate": 9.394550958627649e-05,
-      "loss": 6.5968,
+      "loss": 6.4891,
       "step": 13000
     },
     {
       "epoch": 5.667506297229219,
-      "grad_norm": 3.8129560947418213,
+      "grad_norm": 2.663959264755249,
       "learning_rate": 9.293642785065591e-05,
-      "loss": 6.5992,
+      "loss": 6.4903,
       "step": 13500
     },
     {
       "epoch": 5.877413937867338,
-      "grad_norm": 2.3966245651245117,
+      "grad_norm": 3.6568427085876465,
       "learning_rate": 9.192734611503532e-05,
-      "loss": 6.5829,
+      "loss": 6.4806,
       "step": 14000
     },
     {
       "epoch": 6.087321578505458,
-      "grad_norm": 3.265836477279663,
+      "grad_norm": 2.5574238300323486,
       "learning_rate": 9.091826437941474e-05,
-      "loss": 6.5406,
+      "loss": 6.4354,
       "step": 14500
     },
     {
       "epoch": 6.297229219143577,
-      "grad_norm": 2.9976537227630615,
+      "grad_norm": 2.4129903316497803,
       "learning_rate": 8.990918264379415e-05,
-      "loss": 6.4934,
+      "loss": 6.3888,
       "step": 15000
     },
     {
       "epoch": 6.507136859781696,
-      "grad_norm": 1.8768919706344604,
+      "grad_norm": 2.638054132461548,
       "learning_rate": 8.890010090817356e-05,
-      "loss": 6.4939,
+      "loss": 6.3903,
       "step": 15500
     },
     {
       "epoch": 6.717044500419815,
-      "grad_norm": 3.4282023906707764,
+      "grad_norm": 2.9566128253936768,
       "learning_rate": 8.789101917255298e-05,
-      "loss": 6.4879,
+      "loss": 6.3851,
       "step": 16000
     },
     {
       "epoch": 6.926952141057934,
-      "grad_norm": 3.0579543113708496,
+      "grad_norm": 2.3400323390960693,
       "learning_rate": 8.68819374369324e-05,
-      "loss": 6.4843,
+      "loss": 6.3894,
       "step": 16500
     },
     {
       "epoch": 7.136859781696054,
-      "grad_norm": 2.4434196949005127,
+      "grad_norm": 2.8594205379486084,
       "learning_rate": 8.58728557013118e-05,
-      "loss": 6.4169,
+      "loss": 6.325,
       "step": 17000
     },
     {
       "epoch": 7.346767422334173,
-      "grad_norm": 2.620242118835449,
+      "grad_norm": 1.7151496410369873,
       "learning_rate": 8.486377396569122e-05,
-      "loss": 6.3983,
+      "loss": 6.3014,
       "step": 17500
     },
     {
       "epoch": 7.5566750629722925,
-      "grad_norm": 2.902372121810913,
+      "grad_norm": 2.114755630493164,
       "learning_rate": 8.385469223007064e-05,
-      "loss": 6.4067,
+      "loss": 6.3106,
       "step": 18000
     },
     {
       "epoch": 7.766582703610411,
-      "grad_norm": 2.9418728351593018,
+      "grad_norm": 3.3576650619506836,
       "learning_rate": 8.284561049445006e-05,
-      "loss": 6.4046,
+      "loss": 6.31,
       "step": 18500
     },
     {
       "epoch": 7.97649034424853,
-      "grad_norm": 3.1700100898742676,
+      "grad_norm": 2.4794225692749023,
       "learning_rate": 8.183652875882947e-05,
-      "loss": 6.4043,
+      "loss": 6.3136,
       "step": 19000
     },
     {
       "epoch": 8.18639798488665,
-      "grad_norm": 2.1963798999786377,
+      "grad_norm": 2.520782470703125,
       "learning_rate": 8.082744702320889e-05,
-      "loss": 6.3264,
+      "loss": 6.2433,
       "step": 19500
     },
     {
       "epoch": 8.39630562552477,
-      "grad_norm": 2.7740824222564697,
+      "grad_norm": 2.2647593021392822,
       "learning_rate": 7.98183652875883e-05,
-      "loss": 6.3291,
+      "loss": 6.2431,
       "step": 20000
     },
     {
       "epoch": 8.39630562552477,
-      "eval_accuracy": 0.15233311821319342,
-      "eval_loss": 6.701448440551758,
-      "eval_runtime": 1.0499,
-      "eval_samples_per_second": 456.231,
-      "eval_steps_per_second": 4.762,
+      "eval_accuracy": 0.15323101097112712,
+      "eval_loss": 6.480999946594238,
+      "eval_runtime": 1.0251,
+      "eval_samples_per_second": 467.277,
+      "eval_steps_per_second": 4.878,
       "step": 20000
     },
     {
       "epoch": 8.606213266162888,
-      "grad_norm": 2.7811079025268555,
+      "grad_norm": 2.691263198852539,
       "learning_rate": 7.880928355196771e-05,
-      "loss": 6.3322,
+      "loss": 6.2411,
       "step": 20500
     },
     {
       "epoch": 8.816120906801007,
-      "grad_norm": 2.257051706314087,
+      "grad_norm": 3.366475820541382,
       "learning_rate": 7.780020181634713e-05,
-      "loss": 6.3292,
+      "loss": 6.2467,
       "step": 21000
     },
     {
       "epoch": 9.026028547439127,
-      "grad_norm": 2.737684726715088,
+      "grad_norm": 2.6554746627807617,
       "learning_rate": 7.679112008072655e-05,
-      "loss": 6.3188,
+      "loss": 6.2374,
       "step": 21500
     },
     {
       "epoch": 9.235936188077247,
-      "grad_norm": 3.8119022846221924,
+      "grad_norm": 2.4414002895355225,
       "learning_rate": 7.578203834510595e-05,
-      "loss": 6.2548,
+      "loss": 6.185,
       "step": 22000
     },
     {
       "epoch": 9.445843828715365,
-      "grad_norm": 2.7535812854766846,
+      "grad_norm": 2.6609840393066406,
       "learning_rate": 7.477295660948537e-05,
-      "loss": 6.2632,
+      "loss": 6.1858,
       "step": 22500
     },
     {
       "epoch": 9.655751469353484,
-      "grad_norm": 5.819505214691162,
+      "grad_norm": 2.6620678901672363,
       "learning_rate": 7.376387487386479e-05,
-      "loss": 6.2604,
+      "loss": 6.1865,
       "step": 23000
     },
     {
       "epoch": 9.865659109991604,
-      "grad_norm": 3.6783864498138428,
+      "grad_norm": 1.7680195569992065,
       "learning_rate": 7.27547931382442e-05,
-      "loss": 6.2687,
+      "loss": 6.1879,
       "step": 23500
     },
     {
       "epoch": 10.075566750629722,
-      "grad_norm": 5.783772945404053,
+      "grad_norm": 2.300396680831909,
       "learning_rate": 7.174571140262362e-05,
-      "loss": 6.231,
+      "loss": 6.159,
       "step": 24000
     },
     {
       "epoch": 10.285474391267842,
-      "grad_norm": 2.723422050476074,
+      "grad_norm": 2.125763177871704,
       "learning_rate": 7.073662966700304e-05,
-      "loss": 6.2044,
+      "loss": 6.1433,
       "step": 24500
     },
     {
       "epoch": 10.495382031905962,
-      "grad_norm": 3.333406448364258,
+      "grad_norm": 3.4411227703094482,
       "learning_rate": 6.972754793138244e-05,
-      "loss": 6.205,
+      "loss": 6.1372,
       "step": 25000
     },
     {
       "epoch": 10.705289672544081,
-      "grad_norm": 3.7053959369659424,
+      "grad_norm": 2.8306920528411865,
       "learning_rate": 6.871846619576186e-05,
-      "loss": 6.2037,
+      "loss": 6.1403,
       "step": 25500
     },
     {
       "epoch": 10.9151973131822,
-      "grad_norm": 3.2251012325286865,
+      "grad_norm": 2.745607614517212,
       "learning_rate": 6.770938446014128e-05,
-      "loss": 6.2136,
+      "loss": 6.1449,
       "step": 26000
     },
     {
       "epoch": 11.125104953820319,
-      "grad_norm": 2.389143466949463,
+      "grad_norm": 3.7489781379699707,
       "learning_rate": 6.670030272452068e-05,
-      "loss": 6.1707,
+      "loss": 6.114,
       "step": 26500
     },
     {
       "epoch": 11.335012594458439,
-      "grad_norm": 3.476702928543091,
+      "grad_norm": 2.851313352584839,
       "learning_rate": 6.56912209889001e-05,
-      "loss": 6.1547,
+      "loss": 6.1028,
       "step": 27000
     },
     {
       "epoch": 11.544920235096557,
-      "grad_norm": 3.1944007873535156,
+      "grad_norm": 2.392357349395752,
       "learning_rate": 6.468213925327952e-05,
-      "loss": 6.1514,
+      "loss": 6.0983,
       "step": 27500
     },
     {
       "epoch": 11.754827875734676,
-      "grad_norm": 3.1503257751464844,
+      "grad_norm": 3.2635204792022705,
       "learning_rate": 6.367305751765893e-05,
-      "loss": 6.155,
+      "loss": 6.0985,
       "step": 28000
     },
     {
       "epoch": 11.964735516372796,
-      "grad_norm": 2.4318435192108154,
+      "grad_norm": 1.8090300559997559,
       "learning_rate": 6.266397578203835e-05,
-      "loss": 6.1554,
+      "loss": 6.0976,
       "step": 28500
     },
     {
       "epoch": 12.174643157010916,
-      "grad_norm": 2.668941020965576,
+      "grad_norm": 2.6949682235717773,
       "learning_rate": 6.165489404641777e-05,
-      "loss": 6.1185,
+      "loss": 6.0729,
       "step": 29000
     },
     {
       "epoch": 12.384550797649034,
-      "grad_norm": 2.188319206237793,
+      "grad_norm": 2.046905755996704,
       "learning_rate": 6.064581231079718e-05,
-      "loss": 6.1029,
+      "loss": 6.0621,
       "step": 29500
     },
     {
       "epoch": 12.594458438287154,
-      "grad_norm": 2.9347007274627686,
+      "grad_norm": 2.0488154888153076,
       "learning_rate": 5.963673057517659e-05,
-      "loss": 6.1153,
+      "loss": 6.0675,
       "step": 30000
     },
     {
       "epoch": 12.594458438287154,
-      "eval_accuracy": 0.1539580826594286,
-      "eval_loss": 6.706422805786133,
-      "eval_runtime": 1.033,
-      "eval_samples_per_second": 463.676,
-      "eval_steps_per_second": 4.84,
+      "eval_accuracy": 0.15462965553907715,
+      "eval_loss": 6.531682014465332,
+      "eval_runtime": 1.0347,
+      "eval_samples_per_second": 462.951,
+      "eval_steps_per_second": 4.832,
       "step": 30000
     },
     {
       "epoch": 12.804366078925273,
-      "grad_norm": 3.4366564750671387,
+      "grad_norm": 1.8668218851089478,
       "learning_rate": 5.8627648839556e-05,
-      "loss": 6.1044,
+      "loss": 6.0623,
       "step": 30500
     },
     {
       "epoch": 13.014273719563391,
-      "grad_norm": 3.167576313018799,
+      "grad_norm": 3.1596615314483643,
       "learning_rate": 5.7618567103935416e-05,
-      "loss": 6.1106,
+      "loss": 6.0697,
       "step": 31000
     },
     {
       "epoch": 13.224181360201511,
-      "grad_norm": 2.0107972621917725,
+      "grad_norm": 2.794318437576294,
       "learning_rate": 5.660948536831484e-05,
-      "loss": 6.0733,
+      "loss": 6.037,
       "step": 31500
     },
     {
       "epoch": 13.43408900083963,
-      "grad_norm": 4.030771255493164,
+      "grad_norm": 3.688591718673706,
       "learning_rate": 5.5600403632694254e-05,
-      "loss": 6.072,
+      "loss": 6.0389,
       "step": 32000
     },
     {
       "epoch": 13.64399664147775,
-      "grad_norm": 3.8046839237213135,
+      "grad_norm": 3.4934194087982178,
       "learning_rate": 5.4591321897073666e-05,
-      "loss": 6.0786,
+      "loss": 6.0363,
       "step": 32500
     },
     {
       "epoch": 13.853904282115868,
-      "grad_norm": 2.889934778213501,
+      "grad_norm": 4.7184977531433105,
       "learning_rate": 5.358224016145308e-05,
-      "loss": 6.0757,
+      "loss": 6.0373,
       "step": 33000
     },
     {
       "epoch": 14.063811922753988,
-      "grad_norm": 4.883760929107666,
+      "grad_norm": 2.1444015502929688,
       "learning_rate": 5.257315842583249e-05,
-      "loss": 6.0633,
+      "loss": 6.0246,
       "step": 33500
     },
     {
       "epoch": 14.273719563392108,
-      "grad_norm": 0.8915508389472961,
+      "grad_norm": 3.46429443359375,
       "learning_rate": 5.15640766902119e-05,
-      "loss": 6.0436,
+      "loss": 6.0098,
       "step": 34000
     },
     {
       "epoch": 14.483627204030226,
-      "grad_norm": 2.643712043762207,
+      "grad_norm": 1.9107475280761719,
       "learning_rate": 5.055499495459133e-05,
-      "loss": 6.0413,
+      "loss": 6.0094,
       "step": 34500
     },
     {
       "epoch": 14.693534844668346,
-      "grad_norm": 2.5184438228607178,
+      "grad_norm": 1.8858654499053955,
       "learning_rate": 4.954591321897074e-05,
-      "loss": 6.0432,
+      "loss": 6.008,
       "step": 35000
     },
     {
       "epoch": 14.903442485306465,
-      "grad_norm": 2.847947835922241,
+      "grad_norm": 1.9284015893936157,
       "learning_rate": 4.853683148335015e-05,
-      "loss": 6.0409,
+      "loss": 6.0067,
       "step": 35500
     },
     {
       "epoch": 15.113350125944585,
-      "grad_norm": 1.147056221961975,
+      "grad_norm": 1.400378942489624,
       "learning_rate": 4.7527749747729566e-05,
-      "loss": 6.0247,
+      "loss": 5.9918,
       "step": 36000
     },
     {
       "epoch": 15.323257766582703,
-      "grad_norm": 2.816908597946167,
+      "grad_norm": 1.1706284284591675,
       "learning_rate": 4.6518668012108985e-05,
-      "loss": 6.0205,
+      "loss": 5.9912,
       "step": 36500
     },
     {
       "epoch": 15.533165407220823,
-      "grad_norm": 3.443824052810669,
+      "grad_norm": 2.7408406734466553,
       "learning_rate": 4.55095862764884e-05,
-      "loss": 6.0149,
+      "loss": 5.9854,
       "step": 37000
     },
     {
       "epoch": 15.743073047858942,
-      "grad_norm": 2.6054351329803467,
+      "grad_norm": 2.697648048400879,
       "learning_rate": 4.450050454086781e-05,
-      "loss": 6.0135,
+      "loss": 5.9834,
       "step": 37500
     },
     {
       "epoch": 15.95298068849706,
-      "grad_norm": 3.3810739517211914,
+      "grad_norm": 3.865612268447876,
       "learning_rate": 4.349142280524723e-05,
-      "loss": 6.0131,
+      "loss": 5.9813,
       "step": 38000
     },
     {
       "epoch": 16.162888329135182,
-      "grad_norm": 2.275756597518921,
+      "grad_norm": 0.8426138758659363,
       "learning_rate": 4.248234106962664e-05,
-      "loss": 5.9965,
+      "loss": 5.9715,
       "step": 38500
     },
     {
       "epoch": 16.3727959697733,
-      "grad_norm": 2.3946337699890137,
+      "grad_norm": 3.171299934387207,
       "learning_rate": 4.147325933400605e-05,
-      "loss": 5.9914,
+      "loss": 5.9639,
       "step": 39000
     },
     {
       "epoch": 16.582703610411418,
-      "grad_norm": 4.954773426055908,
+      "grad_norm": 4.164156913757324,
       "learning_rate": 4.046417759838547e-05,
-      "loss": 5.9909,
+      "loss": 5.9665,
       "step": 39500
     },
     {
       "epoch": 16.79261125104954,
-      "grad_norm": 4.085302829742432,
+      "grad_norm": 2.041938066482544,
       "learning_rate": 3.9455095862764884e-05,
-      "loss": 5.9862,
+      "loss": 5.9569,
       "step": 40000
     },
     {
       "epoch": 16.79261125104954,
-      "eval_accuracy": 0.1519849725243916,
-      "eval_loss": 7.014549255371094,
-      "eval_runtime": 1.0149,
-      "eval_samples_per_second": 471.967,
-      "eval_steps_per_second": 4.927,
+      "eval_accuracy": 0.15302231692273185,
+      "eval_loss": 6.651042938232422,
+      "eval_runtime": 1.021,
+      "eval_samples_per_second": 469.166,
+      "eval_steps_per_second": 4.897,
       "step": 40000
     },
     {
       "epoch": 17.002518891687657,
-      "grad_norm": 3.3821780681610107,
+      "grad_norm": 3.150923013687134,
       "learning_rate": 3.84460141271443e-05,
-      "loss": 5.9872,
+      "loss": 5.9617,
       "step": 40500
     },
     {
       "epoch": 17.212426532325775,
-      "grad_norm": 1.8607794046401978,
+      "grad_norm": 0.7818689346313477,
       "learning_rate": 3.7436932391523715e-05,
-      "loss": 5.9711,
+      "loss": 5.9453,
       "step": 41000
     },
     {
       "epoch": 17.422334172963897,
-      "grad_norm": 2.3415207862854004,
+      "grad_norm": 4.910539150238037,
       "learning_rate": 3.642785065590313e-05,
-      "loss": 5.9652,
+      "loss": 5.9428,
       "step": 41500
     },
     {
       "epoch": 17.632241813602015,
-      "grad_norm": 2.3748843669891357,
+      "grad_norm": 2.48958683013916,
       "learning_rate": 3.541876892028255e-05,
-      "loss": 5.97,
+      "loss": 5.9432,
       "step": 42000
     },
     {
       "epoch": 17.842149454240133,
-      "grad_norm": 1.7207521200180054,
+      "grad_norm": 2.8172192573547363,
       "learning_rate": 3.440968718466196e-05,
-      "loss": 5.9688,
+      "loss": 5.9413,
       "step": 42500
     },
     {
       "epoch": 18.052057094878254,
-      "grad_norm": 3.5838589668273926,
+      "grad_norm": 1.81460440158844,
       "learning_rate": 3.340060544904137e-05,
-      "loss": 5.9576,
+      "loss": 5.9341,
       "step": 43000
     },
     {
       "epoch": 18.261964735516372,
-      "grad_norm": 1.1076291799545288,
+      "grad_norm": 1.0337363481521606,
       "learning_rate": 3.239152371342079e-05,
-      "loss": 5.9478,
+      "loss": 5.9247,
       "step": 43500
     },
     {
       "epoch": 18.471872376154494,
-      "grad_norm": 2.307013750076294,
+      "grad_norm": 1.9823477268218994,
       "learning_rate": 3.13824419778002e-05,
-      "loss": 5.9504,
+      "loss": 5.9255,
       "step": 44000
     },
     {
       "epoch": 18.68178001679261,
-      "grad_norm": 1.877256989479065,
+      "grad_norm": 0.8637967109680176,
       "learning_rate": 3.0373360242179615e-05,
-      "loss": 5.95,
+      "loss": 5.9248,
       "step": 44500
     },
     {
       "epoch": 18.89168765743073,
-      "grad_norm": 2.667221784591675,
+      "grad_norm": 0.872508704662323,
       "learning_rate": 2.9364278506559034e-05,
-      "loss": 5.9522,
+      "loss": 5.9261,
       "step": 45000
     },
     {
       "epoch": 19.10159529806885,
-      "grad_norm": 1.781567096710205,
+      "grad_norm": 1.7914263010025024,
       "learning_rate": 2.8355196770938446e-05,
-      "loss": 5.9428,
+      "loss": 5.921,
       "step": 45500
     },
     {
       "epoch": 19.31150293870697,
-      "grad_norm": 2.600219488143921,
+      "grad_norm": 1.7649086713790894,
       "learning_rate": 2.734611503531786e-05,
-      "loss": 5.9273,
+      "loss": 5.9058,
       "step": 46000
     },
     {
       "epoch": 19.521410579345087,
-      "grad_norm": 3.2653560638427734,
+      "grad_norm": 0.9645077586174011,
       "learning_rate": 2.6337033299697278e-05,
-      "loss": 5.9327,
+      "loss": 5.907,
       "step": 46500
     },
     {
       "epoch": 19.73131821998321,
-      "grad_norm": 0.8226433992385864,
+      "grad_norm": 1.0561953783035278,
       "learning_rate": 2.532795156407669e-05,
-      "loss": 5.9303,
+      "loss": 5.9097,
       "step": 47000
     },
     {
       "epoch": 19.941225860621326,
-      "grad_norm": 2.6203231811523438,
+      "grad_norm": 0.970642626285553,
       "learning_rate": 2.4318869828456106e-05,
-      "loss": 5.9259,
+      "loss": 5.9055,
       "step": 47500
     },
     {
       "epoch": 20.151133501259444,
-      "grad_norm": 3.7636687755584717,
+      "grad_norm": 0.8695760369300842,
       "learning_rate": 2.330978809283552e-05,
-      "loss": 5.9187,
+      "loss": 5.8992,
       "step": 48000
     },
     {
       "epoch": 20.361041141897566,
-      "grad_norm": 2.3655290603637695,
+      "grad_norm": 1.0435950756072998,
       "learning_rate": 2.2300706357214934e-05,
-      "loss": 5.9176,
+      "loss": 5.8977,
       "step": 48500
     },
     {
       "epoch": 20.570948782535684,
-      "grad_norm": 1.55186927318573,
+      "grad_norm": 0.8484736680984497,
       "learning_rate": 2.129162462159435e-05,
-      "loss": 5.9103,
+      "loss": 5.8909,
       "step": 49000
     },
     {
       "epoch": 20.7808564231738,
-      "grad_norm": 4.357689380645752,
+      "grad_norm": 2.0572588443756104,
       "learning_rate": 2.0282542885973765e-05,
-      "loss": 5.9109,
+      "loss": 5.8899,
       "step": 49500
     },
     {
       "epoch": 20.990764063811923,
-      "grad_norm": 3.323455333709717,
+      "grad_norm": 0.7758699655532837,
       "learning_rate": 1.927346115035318e-05,
-      "loss": 5.9088,
+      "loss": 5.89,
       "step": 50000
     },
     {
       "epoch": 20.990764063811923,
-      "eval_accuracy": 0.15313875571256344,
-      "eval_loss": 7.005898952484131,
-      "eval_runtime": 1.0145,
-      "eval_samples_per_second": 472.149,
-      "eval_steps_per_second": 4.928,
+      "eval_accuracy": 0.15235371632040823,
+      "eval_loss": 6.6476731300354,
+      "eval_runtime": 1.0305,
+      "eval_samples_per_second": 464.818,
+      "eval_steps_per_second": 4.852,
       "step": 50000
     },
     {
       "epoch": 21.20067170445004,
-      "grad_norm": 0.8441404104232788,
+      "grad_norm": 1.1873598098754883,
       "learning_rate": 1.8264379414732593e-05,
-      "loss": 5.9017,
+      "loss": 5.8807,
       "step": 50500
     },
     {
       "epoch": 21.410579345088163,
-      "grad_norm": 2.293954372406006,
+      "grad_norm": 0.7782743573188782,
       "learning_rate": 1.725529767911201e-05,
-      "loss": 5.9007,
+      "loss": 5.8834,
       "step": 51000
     },
     {
       "epoch": 21.62048698572628,
-      "grad_norm": 2.9091224670410156,
+      "grad_norm": 0.8238320350646973,
       "learning_rate": 1.6246215943491424e-05,
-      "loss": 5.9001,
+      "loss": 5.88,
       "step": 51500
     },
     {
       "epoch": 21.8303946263644,
-      "grad_norm": 3.189171075820923,
+      "grad_norm": 4.082571506500244,
       "learning_rate": 1.5237134207870837e-05,
-      "loss": 5.8969,
+      "loss": 5.8796,
       "step": 52000
     },
     {
       "epoch": 22.04030226700252,
-      "grad_norm": 0.8301982283592224,
+      "grad_norm": 0.774568498134613,
       "learning_rate": 1.4228052472250252e-05,
-      "loss": 5.8956,
+      "loss": 5.8765,
       "step": 52500
     },
     {
       "epoch": 22.250209907640638,
-      "grad_norm": 3.955568552017212,
+      "grad_norm": 3.585805892944336,
       "learning_rate": 1.3218970736629668e-05,
-      "loss": 5.8826,
+      "loss": 5.8671,
       "step": 53000
     },
     {
       "epoch": 22.460117548278756,
-      "grad_norm": 1.694311261177063,
+      "grad_norm": 0.7828291654586792,
       "learning_rate": 1.2209889001009082e-05,
-      "loss": 5.8871,
+      "loss": 5.8694,
       "step": 53500
     },
     {
       "epoch": 22.670025188916878,
-      "grad_norm": 1.507860541343689,
+      "grad_norm": 1.8001397848129272,
       "learning_rate": 1.1200807265388496e-05,
-      "loss": 5.89,
+      "loss": 5.8712,
       "step": 54000
     },
     {
       "epoch": 22.879932829554996,
-      "grad_norm": 0.8334353566169739,
+      "grad_norm": 0.8042154908180237,
       "learning_rate": 1.0191725529767912e-05,
-      "loss": 5.8817,
+      "loss": 5.8654,
       "step": 54500
     },
     {
       "epoch": 23.089840470193113,
-      "grad_norm": 0.8316900134086609,
+      "grad_norm": 3.4733877182006836,
       "learning_rate": 9.182643794147326e-06,
-      "loss": 5.8804,
+      "loss": 5.8632,
       "step": 55000
     },
     {
       "epoch": 23.299748110831235,
-      "grad_norm": 2.653756618499756,
+      "grad_norm": 0.7870926260948181,
       "learning_rate": 8.173562058526741e-06,
-      "loss": 5.8749,
+      "loss": 5.8613,
       "step": 55500
     },
     {
       "epoch": 23.509655751469353,
-      "grad_norm": 3.7355833053588867,
+      "grad_norm": 3.1393957138061523,
       "learning_rate": 7.164480322906156e-06,
-      "loss": 5.8769,
+      "loss": 5.8606,
       "step": 56000
     },
     {
       "epoch": 23.719563392107474,
-      "grad_norm": 0.8122338652610779,
+      "grad_norm": 0.7654667496681213,
       "learning_rate": 6.155398587285571e-06,
-      "loss": 5.8699,
+      "loss": 5.8556,
       "step": 56500
     },
     {
       "epoch": 23.929471032745592,
-      "grad_norm": 0.8836779594421387,
+      "grad_norm": 0.7754602432250977,
       "learning_rate": 5.146316851664986e-06,
-      "loss": 5.8733,
+      "loss": 5.8596,
       "step": 57000
     },
     {
       "epoch": 24.13937867338371,
-      "grad_norm": 3.144986152648926,
+      "grad_norm": 0.7847160696983337,
       "learning_rate": 4.1372351160444e-06,
-      "loss": 5.8685,
+      "loss": 5.8528,
       "step": 57500
     },
     {
       "epoch": 24.349286314021832,
-      "grad_norm": 3.1895604133605957,
+      "grad_norm": 0.8076699376106262,
       "learning_rate": 3.1281533804238145e-06,
-      "loss": 5.8702,
+      "loss": 5.854,
       "step": 58000
     },
     {
       "epoch": 24.55919395465995,
-      "grad_norm": 0.9001981019973755,
+      "grad_norm": 3.1537649631500244,
       "learning_rate": 2.1190716448032293e-06,
-      "loss": 5.8651,
+      "loss": 5.851,
       "step": 58500
     },
     {
       "epoch": 24.769101595298068,
-      "grad_norm": 1.6653518676757812,
+      "grad_norm": 0.7634235620498657,
       "learning_rate": 1.109989909182644e-06,
-      "loss": 5.8625,
+      "loss": 5.8503,
       "step": 59000
     },
     {
       "epoch": 24.97900923593619,
-      "grad_norm": 1.8964729309082031,
+      "grad_norm": 4.646351337432861,
       "learning_rate": 1.0090817356205853e-07,
-      "loss": 5.8636,
+      "loss": 5.8501,
       "step": 59500
     },
     {
       "epoch": 25.0,
       "step": 59550,
       "total_flos": 1.5151126801503744e+18,
-      "train_loss": 6.378149575490495,
-      "train_runtime": 24807.4865,
-      "train_samples_per_second": 230.413,
-      "train_steps_per_second": 2.4
+      "train_loss": 6.324693534844668,
+      "train_runtime": 24831.9556,
+      "train_samples_per_second": 230.186,
+      "train_steps_per_second": 2.398
     }
   ],
   "logging_steps": 500,