End of training

README.md

@@ -2,11 +2,23 @@
 library_name: transformers
 tags:
 - generated_from_trainer
+datasets:
+- Hartunka/processed_wikitext-103-raw-v1-rand-5_v2
 metrics:
 - accuracy
 model-index:
 - name: bert_base_rand_5_v2
-  results: []
+  results:
+  - task:
+      name: Masked Language Modeling
+      type: fill-mask
+    dataset:
+      name: Hartunka/processed_wikitext-103-raw-v1-rand-5_v2
+      type: Hartunka/processed_wikitext-103-raw-v1-rand-5_v2
+    metrics:
+    - name: Accuracy
+      type: accuracy
+      value: 0.1531332994292818
 ---
 
 <!-- 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_5_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-rand-5_v2 dataset.
 It achieves the following results on the evaluation set:
-- Loss: 9.6431
-- Accuracy: 0.1513
+- Loss: 7.6460
+- Accuracy: 0.1531
 
 ## Model description
 

all_results.json

@@ -1,16 +1,16 @@
 {
     "epoch": 25.0,
-    "eval_accuracy": 0.15375487370740804,
-    "eval_loss": 7.654566287994385,
-    "eval_runtime": 1.0263,
+    "eval_accuracy": 0.1531332994292818,
+    "eval_loss": 7.646039962768555,
+    "eval_runtime": 1.0253,
     "eval_samples": 479,
-    "eval_samples_per_second": 466.738,
-    "eval_steps_per_second": 4.872,
-    "perplexity": 2110.259675793937,
+    "eval_samples_per_second": 467.182,
+    "eval_steps_per_second": 4.877,
+    "perplexity": 2092.3434040071675,
     "total_flos": 1.5149101316039424e+18,
-    "train_loss": 6.953392696813012,
-    "train_runtime": 24679.3114,
+    "train_loss": 6.939091333602278,
+    "train_runtime": 24667.3069,
     "train_samples": 228639,
-    "train_samples_per_second": 231.61,
-    "train_steps_per_second": 2.413
+    "train_samples_per_second": 231.723,
+    "train_steps_per_second": 2.414
 }

eval_results.json

@@ -1,10 +1,10 @@
 {
     "epoch": 25.0,
-    "eval_accuracy": 0.15375487370740804,
-    "eval_loss": 7.654566287994385,
-    "eval_runtime": 1.0263,
+    "eval_accuracy": 0.1531332994292818,
+    "eval_loss": 7.646039962768555,
+    "eval_runtime": 1.0253,
     "eval_samples": 479,
-    "eval_samples_per_second": 466.738,
-    "eval_steps_per_second": 4.872,
-    "perplexity": 2110.259675793937
+    "eval_samples_per_second": 467.182,
+    "eval_steps_per_second": 4.877,
+    "perplexity": 2092.3434040071675
 }

train_results.json

@@ -1,9 +1,9 @@
 {
     "epoch": 25.0,
     "total_flos": 1.5149101316039424e+18,
-    "train_loss": 6.953392696813012,
-    "train_runtime": 24679.3114,
+    "train_loss": 6.939091333602278,
+    "train_runtime": 24667.3069,
     "train_samples": 228639,
-    "train_samples_per_second": 231.61,
-    "train_steps_per_second": 2.413
+    "train_samples_per_second": 231.723,
+    "train_steps_per_second": 2.414
 }

trainer_state.json

@@ -1,7 +1,7 @@
 {
   "best_global_step": 10000,
-  "best_metric": 7.634960174560547,
-  "best_model_checkpoint": "bert_base_rand_5_v12/checkpoint-10000",
+  "best_metric": 7.631558895111084,
+  "best_model_checkpoint": "bert_base_rand_5_v2/checkpoint-10000",
   "epoch": 25.0,
   "eval_steps": 10000,
   "global_step": 59550,
@@ -11,890 +11,890 @@
   "log_history": [
     {
       "epoch": 0.20990764063811923,
-      "grad_norm": 2.4992802143096924,
+      "grad_norm": 2.9877593517303467,
       "learning_rate": 5e-06,
-      "loss": 10.8092,
+      "loss": 10.8091,
       "step": 500
     },
     {
       "epoch": 0.41981528127623846,
-      "grad_norm": 2.681934118270874,
+      "grad_norm": 1.135245680809021,
       "learning_rate": 1e-05,
-      "loss": 9.1411,
+      "loss": 9.1407,
       "step": 1000
     },
     {
       "epoch": 0.6297229219143576,
-      "grad_norm": 1.7480603456497192,
+      "grad_norm": 2.194422960281372,
       "learning_rate": 1.5e-05,
-      "loss": 8.596,
+      "loss": 8.5953,
       "step": 1500
     },
     {
       "epoch": 0.8396305625524769,
-      "grad_norm": 1.8540552854537964,
+      "grad_norm": 3.5926201343536377,
       "learning_rate": 2e-05,
-      "loss": 8.4357,
+      "loss": 8.4387,
       "step": 2000
     },
     {
       "epoch": 1.0495382031905962,
-      "grad_norm": 1.5082471370697021,
+      "grad_norm": 3.6953790187835693,
       "learning_rate": 2.5e-05,
-      "loss": 8.3197,
+      "loss": 8.3186,
       "step": 2500
     },
     {
       "epoch": 1.2594458438287153,
-      "grad_norm": 1.7294883728027344,
+      "grad_norm": 2.253312110900879,
       "learning_rate": 3e-05,
-      "loss": 8.2113,
+      "loss": 8.2111,
       "step": 3000
     },
     {
       "epoch": 1.4693534844668346,
-      "grad_norm": 1.7149721384048462,
+      "grad_norm": 2.3300347328186035,
       "learning_rate": 3.5e-05,
-      "loss": 8.1225,
+      "loss": 8.1229,
       "step": 3500
     },
     {
       "epoch": 1.6792611251049538,
-      "grad_norm": 2.204758882522583,
+      "grad_norm": 1.5966769456863403,
       "learning_rate": 4e-05,
-      "loss": 8.0566,
+      "loss": 8.0564,
       "step": 4000
     },
     {
       "epoch": 1.8891687657430731,
-      "grad_norm": 1.0799752473831177,
+      "grad_norm": 2.017082691192627,
       "learning_rate": 4.5e-05,
-      "loss": 8.0003,
+      "loss": 8.0011,
       "step": 4500
     },
     {
       "epoch": 2.0990764063811924,
-      "grad_norm": 1.8463094234466553,
+      "grad_norm": 2.4728705883026123,
       "learning_rate": 5e-05,
-      "loss": 7.9515,
+      "loss": 7.9518,
       "step": 5000
     },
     {
       "epoch": 2.3089840470193117,
-      "grad_norm": 1.6064136028289795,
+      "grad_norm": 3.458080530166626,
       "learning_rate": 5.500000000000001e-05,
-      "loss": 7.9105,
+      "loss": 7.9114,
       "step": 5500
     },
     {
       "epoch": 2.5188916876574305,
-      "grad_norm": 1.6856950521469116,
+      "grad_norm": 2.0525100231170654,
       "learning_rate": 6e-05,
-      "loss": 7.8761,
+      "loss": 7.8754,
       "step": 6000
     },
     {
       "epoch": 2.72879932829555,
-      "grad_norm": 1.9870349168777466,
+      "grad_norm": 2.5022454261779785,
       "learning_rate": 6.500000000000001e-05,
-      "loss": 7.8423,
+      "loss": 7.8429,
       "step": 6500
     },
     {
       "epoch": 2.938706968933669,
-      "grad_norm": 2.8902649879455566,
+      "grad_norm": 2.1538009643554688,
       "learning_rate": 7e-05,
-      "loss": 7.813,
+      "loss": 7.8132,
       "step": 7000
     },
     {
       "epoch": 3.1486146095717884,
-      "grad_norm": 3.3757693767547607,
+      "grad_norm": 3.1521589756011963,
       "learning_rate": 7.500000000000001e-05,
-      "loss": 7.7895,
+      "loss": 7.789,
       "step": 7500
     },
     {
       "epoch": 3.3585222502099077,
-      "grad_norm": 2.6722750663757324,
+      "grad_norm": 1.6730432510375977,
       "learning_rate": 8e-05,
-      "loss": 7.7668,
+      "loss": 7.7688,
       "step": 8000
     },
     {
       "epoch": 3.568429890848027,
-      "grad_norm": 1.3434144258499146,
+      "grad_norm": 1.1812211275100708,
       "learning_rate": 8.5e-05,
       "loss": 7.7514,
       "step": 8500
     },
     {
       "epoch": 3.7783375314861463,
-      "grad_norm": 1.5891029834747314,
+      "grad_norm": 1.405652403831482,
       "learning_rate": 9e-05,
-      "loss": 7.7252,
+      "loss": 7.7258,
       "step": 9000
     },
     {
       "epoch": 3.988245172124265,
-      "grad_norm": 3.59718918800354,
+      "grad_norm": 3.2779672145843506,
       "learning_rate": 9.5e-05,
-      "loss": 7.7062,
+      "loss": 7.7091,
       "step": 9500
     },
     {
       "epoch": 4.198152812762385,
-      "grad_norm": 2.891531229019165,
+      "grad_norm": 2.5300495624542236,
       "learning_rate": 0.0001,
-      "loss": 7.6838,
+      "loss": 7.6861,
       "step": 10000
     },
     {
       "epoch": 4.198152812762385,
-      "eval_accuracy": 0.1512006455382732,
-      "eval_loss": 7.634960174560547,
-      "eval_runtime": 1.0308,
-      "eval_samples_per_second": 464.675,
-      "eval_steps_per_second": 4.85,
+      "eval_accuracy": 0.15056067225020173,
+      "eval_loss": 7.631558895111084,
+      "eval_runtime": 1.0449,
+      "eval_samples_per_second": 458.402,
+      "eval_steps_per_second": 4.785,
       "step": 10000
     },
     {
       "epoch": 4.408060453400504,
-      "grad_norm": 2.096656322479248,
+      "grad_norm": 2.1099071502685547,
       "learning_rate": 9.899091826437942e-05,
-      "loss": 7.6684,
+      "loss": 7.6706,
       "step": 10500
     },
     {
       "epoch": 4.617968094038623,
-      "grad_norm": 3.7218902111053467,
+      "grad_norm": 1.7518905401229858,
       "learning_rate": 9.798183652875883e-05,
-      "loss": 7.6509,
+      "loss": 7.6495,
       "step": 11000
     },
     {
       "epoch": 4.827875734676742,
-      "grad_norm": 2.0484023094177246,
+      "grad_norm": 2.30505633354187,
       "learning_rate": 9.697275479313825e-05,
-      "loss": 7.6441,
+      "loss": 7.643,
       "step": 11500
     },
     {
       "epoch": 5.037783375314861,
-      "grad_norm": 2.0777432918548584,
+      "grad_norm": 2.463966131210327,
       "learning_rate": 9.596367305751767e-05,
-      "loss": 7.6211,
+      "loss": 7.6224,
       "step": 12000
     },
     {
       "epoch": 5.247691015952981,
-      "grad_norm": 2.098064422607422,
+      "grad_norm": 2.924703359603882,
       "learning_rate": 9.495459132189707e-05,
-      "loss": 7.5826,
+      "loss": 7.58,
       "step": 12500
     },
     {
       "epoch": 5.4575986565911,
-      "grad_norm": 3.8706886768341064,
+      "grad_norm": 1.974431037902832,
       "learning_rate": 9.394550958627649e-05,
-      "loss": 7.5745,
+      "loss": 7.5758,
       "step": 13000
     },
     {
       "epoch": 5.667506297229219,
-      "grad_norm": 1.157472848892212,
+      "grad_norm": 1.5063343048095703,
       "learning_rate": 9.293642785065591e-05,
-      "loss": 7.575,
+      "loss": 7.5774,
       "step": 13500
     },
     {
       "epoch": 5.877413937867338,
-      "grad_norm": 1.768904685974121,
+      "grad_norm": 2.6250364780426025,
       "learning_rate": 9.192734611503532e-05,
-      "loss": 7.5669,
+      "loss": 7.5647,
       "step": 14000
     },
     {
       "epoch": 6.087321578505458,
-      "grad_norm": 2.8934755325317383,
+      "grad_norm": 3.129988193511963,
       "learning_rate": 9.091826437941474e-05,
-      "loss": 7.5344,
+      "loss": 7.5356,
       "step": 14500
     },
     {
       "epoch": 6.297229219143577,
-      "grad_norm": 2.821071147918701,
+      "grad_norm": 3.8247897624969482,
       "learning_rate": 8.990918264379415e-05,
-      "loss": 7.5118,
+      "loss": 7.5102,
       "step": 15000
     },
     {
       "epoch": 6.507136859781696,
-      "grad_norm": 2.228764533996582,
+      "grad_norm": 4.086785793304443,
       "learning_rate": 8.890010090817356e-05,
-      "loss": 7.5057,
+      "loss": 7.5054,
       "step": 15500
     },
     {
       "epoch": 6.717044500419815,
-      "grad_norm": 2.4731714725494385,
+      "grad_norm": 3.1727752685546875,
       "learning_rate": 8.789101917255298e-05,
-      "loss": 7.5106,
+      "loss": 7.5083,
       "step": 16000
     },
     {
       "epoch": 6.926952141057934,
-      "grad_norm": 2.037497043609619,
+      "grad_norm": 2.41412615776062,
       "learning_rate": 8.68819374369324e-05,
-      "loss": 7.5061,
+      "loss": 7.5063,
       "step": 16500
     },
     {
       "epoch": 7.136859781696054,
-      "grad_norm": 3.60455060005188,
+      "grad_norm": 3.926177501678467,
       "learning_rate": 8.58728557013118e-05,
-      "loss": 7.4598,
+      "loss": 7.4584,
       "step": 17000
     },
     {
       "epoch": 7.346767422334173,
-      "grad_norm": 3.124608278274536,
+      "grad_norm": 2.417818069458008,
       "learning_rate": 8.486377396569122e-05,
-      "loss": 7.4435,
+      "loss": 7.4392,
       "step": 17500
     },
     {
       "epoch": 7.5566750629722925,
-      "grad_norm": 2.735703468322754,
+      "grad_norm": 3.9381933212280273,
       "learning_rate": 8.385469223007064e-05,
-      "loss": 7.4499,
+      "loss": 7.4503,
       "step": 18000
     },
     {
       "epoch": 7.766582703610411,
-      "grad_norm": 2.900106430053711,
+      "grad_norm": 2.326669216156006,
       "learning_rate": 8.284561049445006e-05,
-      "loss": 7.4567,
+      "loss": 7.4544,
       "step": 18500
     },
     {
       "epoch": 7.97649034424853,
-      "grad_norm": 3.210784435272217,
+      "grad_norm": 2.3803982734680176,
       "learning_rate": 8.183652875882947e-05,
-      "loss": 7.456,
+      "loss": 7.4558,
       "step": 19000
     },
     {
       "epoch": 8.18639798488665,
-      "grad_norm": 2.3770394325256348,
+      "grad_norm": 3.4016687870025635,
       "learning_rate": 8.082744702320889e-05,
-      "loss": 7.375,
+      "loss": 7.3677,
       "step": 19500
     },
     {
       "epoch": 8.39630562552477,
-      "grad_norm": 3.6196975708007812,
+      "grad_norm": 2.824284791946411,
       "learning_rate": 7.98183652875883e-05,
-      "loss": 7.3816,
+      "loss": 7.3707,
       "step": 20000
     },
     {
       "epoch": 8.39630562552477,
-      "eval_accuracy": 0.152697887146104,
-      "eval_loss": 7.671019077301025,
-      "eval_runtime": 1.0196,
-      "eval_samples_per_second": 469.786,
-      "eval_steps_per_second": 4.904,
+      "eval_accuracy": 0.15272594629478942,
+      "eval_loss": 7.690751075744629,
+      "eval_runtime": 1.0289,
+      "eval_samples_per_second": 465.551,
+      "eval_steps_per_second": 4.86,
       "step": 20000
     },
     {
       "epoch": 8.606213266162888,
-      "grad_norm": 1.907096028327942,
+      "grad_norm": 3.187249183654785,
       "learning_rate": 7.880928355196771e-05,
-      "loss": 7.3829,
+      "loss": 7.3728,
       "step": 20500
     },
     {
       "epoch": 8.816120906801007,
-      "grad_norm": 2.3949108123779297,
+      "grad_norm": 2.782331943511963,
       "learning_rate": 7.780020181634713e-05,
-      "loss": 7.3945,
+      "loss": 7.3852,
       "step": 21000
     },
     {
       "epoch": 9.026028547439127,
-      "grad_norm": 4.06590461730957,
+      "grad_norm": 3.291903018951416,
       "learning_rate": 7.679112008072655e-05,
-      "loss": 7.3721,
+      "loss": 7.3591,
       "step": 21500
     },
     {
       "epoch": 9.235936188077247,
-      "grad_norm": 4.082590103149414,
+      "grad_norm": 4.126809597015381,
       "learning_rate": 7.578203834510595e-05,
-      "loss": 7.2535,
+      "loss": 7.242,
       "step": 22000
     },
     {
       "epoch": 9.445843828715365,
-      "grad_norm": 3.5404210090637207,
+      "grad_norm": 3.587277412414551,
       "learning_rate": 7.477295660948537e-05,
-      "loss": 7.2698,
+      "loss": 7.2571,
       "step": 22500
     },
     {
       "epoch": 9.655751469353484,
-      "grad_norm": 4.227196216583252,
+      "grad_norm": 3.321702003479004,
       "learning_rate": 7.376387487386479e-05,
-      "loss": 7.282,
+      "loss": 7.2719,
       "step": 23000
     },
     {
       "epoch": 9.865659109991604,
-      "grad_norm": 3.573904037475586,
+      "grad_norm": 3.5752756595611572,
       "learning_rate": 7.27547931382442e-05,
-      "loss": 7.2842,
+      "loss": 7.2716,
       "step": 23500
     },
     {
       "epoch": 10.075566750629722,
-      "grad_norm": 4.746927738189697,
+      "grad_norm": 3.9233832359313965,
       "learning_rate": 7.174571140262362e-05,
-      "loss": 7.2038,
+      "loss": 7.1901,
       "step": 24000
     },
     {
       "epoch": 10.285474391267842,
-      "grad_norm": 4.1370086669921875,
+      "grad_norm": 3.8932623863220215,
       "learning_rate": 7.073662966700304e-05,
-      "loss": 7.0839,
+      "loss": 7.0631,
       "step": 24500
     },
     {
       "epoch": 10.495382031905962,
-      "grad_norm": 4.36155891418457,
+      "grad_norm": 3.942333936691284,
       "learning_rate": 6.972754793138244e-05,
-      "loss": 7.1023,
+      "loss": 7.0795,
       "step": 25000
     },
     {
       "epoch": 10.705289672544081,
-      "grad_norm": 5.503416538238525,
+      "grad_norm": 4.349544525146484,
       "learning_rate": 6.871846619576186e-05,
-      "loss": 7.1142,
+      "loss": 7.0909,
       "step": 25500
     },
     {
       "epoch": 10.9151973131822,
-      "grad_norm": 4.3891377449035645,
+      "grad_norm": 4.525393009185791,
       "learning_rate": 6.770938446014128e-05,
-      "loss": 7.1249,
+      "loss": 7.1072,
       "step": 26000
     },
     {
       "epoch": 11.125104953820319,
-      "grad_norm": 5.554305076599121,
+      "grad_norm": 5.4399213790893555,
       "learning_rate": 6.670030272452068e-05,
-      "loss": 6.9825,
+      "loss": 6.9649,
       "step": 26500
     },
     {
       "epoch": 11.335012594458439,
-      "grad_norm": 5.704439640045166,
+      "grad_norm": 5.478682041168213,
       "learning_rate": 6.56912209889001e-05,
-      "loss": 6.9082,
+      "loss": 6.8882,
       "step": 27000
     },
     {
       "epoch": 11.544920235096557,
-      "grad_norm": 4.971321105957031,
+      "grad_norm": 5.21258544921875,
       "learning_rate": 6.468213925327952e-05,
-      "loss": 6.9273,
+      "loss": 6.9039,
       "step": 27500
     },
     {
       "epoch": 11.754827875734676,
-      "grad_norm": 5.947397232055664,
+      "grad_norm": 5.533293724060059,
       "learning_rate": 6.367305751765893e-05,
-      "loss": 6.9341,
+      "loss": 6.9185,
       "step": 28000
     },
     {
       "epoch": 11.964735516372796,
-      "grad_norm": 3.416337251663208,
+      "grad_norm": 6.176639080047607,
       "learning_rate": 6.266397578203835e-05,
-      "loss": 6.9555,
+      "loss": 6.9336,
       "step": 28500
     },
     {
       "epoch": 12.174643157010916,
-      "grad_norm": 5.537859916687012,
+      "grad_norm": 6.256444454193115,
       "learning_rate": 6.165489404641777e-05,
-      "loss": 6.789,
+      "loss": 6.7603,
       "step": 29000
     },
     {
       "epoch": 12.384550797649034,
-      "grad_norm": 4.691187381744385,
+      "grad_norm": 7.563621520996094,
       "learning_rate": 6.064581231079718e-05,
-      "loss": 6.7612,
+      "loss": 6.7354,
       "step": 29500
     },
     {
       "epoch": 12.594458438287154,
-      "grad_norm": 6.227050304412842,
+      "grad_norm": 5.112383842468262,
       "learning_rate": 5.963673057517659e-05,
-      "loss": 6.784,
+      "loss": 6.7638,
       "step": 30000
     },
     {
       "epoch": 12.594458438287154,
-      "eval_accuracy": 0.15443378011584633,
-      "eval_loss": 8.568798065185547,
-      "eval_runtime": 1.0111,
-      "eval_samples_per_second": 473.724,
-      "eval_steps_per_second": 4.945,
+      "eval_accuracy": 0.15401404706606597,
+      "eval_loss": 8.326098442077637,
+      "eval_runtime": 1.0042,
+      "eval_samples_per_second": 476.997,
+      "eval_steps_per_second": 4.979,
       "step": 30000
     },
     {
       "epoch": 12.804366078925273,
-      "grad_norm": 7.301233291625977,
+      "grad_norm": 5.6610188484191895,
       "learning_rate": 5.8627648839556e-05,
-      "loss": 6.8001,
+      "loss": 6.7782,
       "step": 30500
     },
     {
       "epoch": 13.014273719563391,
-      "grad_norm": 4.9339518547058105,
+      "grad_norm": 7.352931976318359,
       "learning_rate": 5.7618567103935416e-05,
-      "loss": 6.794,
+      "loss": 6.7723,
       "step": 31000
     },
     {
       "epoch": 13.224181360201511,
-      "grad_norm": 6.148602485656738,
+      "grad_norm": 7.229602336883545,
       "learning_rate": 5.660948536831484e-05,
-      "loss": 6.6352,
+      "loss": 6.6128,
       "step": 31500
     },
     {
       "epoch": 13.43408900083963,
-      "grad_norm": 5.351287841796875,
+      "grad_norm": 8.164566040039062,
       "learning_rate": 5.5600403632694254e-05,
-      "loss": 6.65,
+      "loss": 6.6314,
       "step": 32000
     },
     {
       "epoch": 13.64399664147775,
-      "grad_norm": 6.611981391906738,
+      "grad_norm": 7.342796802520752,
       "learning_rate": 5.4591321897073666e-05,
-      "loss": 6.6717,
+      "loss": 6.6541,
       "step": 32500
     },
     {
       "epoch": 13.853904282115868,
-      "grad_norm": 6.155622959136963,
+      "grad_norm": 6.204566478729248,
       "learning_rate": 5.358224016145308e-05,
-      "loss": 6.6795,
+      "loss": 6.6497,
       "step": 33000
     },
     {
       "epoch": 14.063811922753988,
-      "grad_norm": 6.759790897369385,
+      "grad_norm": 6.311522960662842,
       "learning_rate": 5.257315842583249e-05,
-      "loss": 6.6339,
+      "loss": 6.6163,
       "step": 33500
     },
     {
       "epoch": 14.273719563392108,
-      "grad_norm": 5.430413722991943,
+      "grad_norm": 6.8089728355407715,
       "learning_rate": 5.15640766902119e-05,
-      "loss": 6.5395,
+      "loss": 6.5209,
       "step": 34000
     },
     {
       "epoch": 14.483627204030226,
-      "grad_norm": 6.663574695587158,
+      "grad_norm": 6.853890419006348,
       "learning_rate": 5.055499495459133e-05,
-      "loss": 6.5573,
+      "loss": 6.5408,
       "step": 34500
     },
     {
       "epoch": 14.693534844668346,
-      "grad_norm": 5.055428981781006,
+      "grad_norm": 4.601109504699707,
       "learning_rate": 4.954591321897074e-05,
-      "loss": 6.5632,
+      "loss": 6.5459,
       "step": 35000
     },
     {
       "epoch": 14.903442485306465,
-      "grad_norm": 5.806168079376221,
+      "grad_norm": 5.2091240882873535,
       "learning_rate": 4.853683148335015e-05,
-      "loss": 6.5726,
+      "loss": 6.5543,
       "step": 35500
     },
     {
       "epoch": 15.113350125944585,
-      "grad_norm": 5.446578025817871,
+      "grad_norm": 4.657880783081055,
       "learning_rate": 4.7527749747729566e-05,
-      "loss": 6.5048,
+      "loss": 6.493,
       "step": 36000
     },
     {
       "epoch": 15.323257766582703,
-      "grad_norm": 4.39936637878418,
+      "grad_norm": 6.046952724456787,
       "learning_rate": 4.6518668012108985e-05,
-      "loss": 6.457,
+      "loss": 6.4523,
       "step": 36500
     },
     {
       "epoch": 15.533165407220823,
-      "grad_norm": 5.585454940795898,
+      "grad_norm": 5.572329521179199,
       "learning_rate": 4.55095862764884e-05,
-      "loss": 6.4674,
+      "loss": 6.4567,
       "step": 37000
     },
     {
       "epoch": 15.743073047858942,
-      "grad_norm": 4.23904275894165,
+      "grad_norm": 6.628154754638672,
       "learning_rate": 4.450050454086781e-05,
-      "loss": 6.477,
+      "loss": 6.4699,
       "step": 37500
     },
     {
       "epoch": 15.95298068849706,
-      "grad_norm": 8.903462409973145,
+      "grad_norm": 4.219488143920898,
       "learning_rate": 4.349142280524723e-05,
-      "loss": 6.4846,
+      "loss": 6.4746,
       "step": 38000
     },
     {
       "epoch": 16.162888329135182,
-      "grad_norm": 7.746410846710205,
+      "grad_norm": 7.043789863586426,
       "learning_rate": 4.248234106962664e-05,
-      "loss": 6.4045,
+      "loss": 6.3962,
       "step": 38500
     },
     {
       "epoch": 16.3727959697733,
-      "grad_norm": 6.386459827423096,
+      "grad_norm": 5.139469146728516,
       "learning_rate": 4.147325933400605e-05,
-      "loss": 6.3964,
+      "loss": 6.3856,
       "step": 39000
     },
     {
       "epoch": 16.582703610411418,
-      "grad_norm": 7.453471660614014,
+      "grad_norm": 6.3856682777404785,
       "learning_rate": 4.046417759838547e-05,
-      "loss": 6.401,
+      "loss": 6.3909,
       "step": 39500
     },
     {
       "epoch": 16.79261125104954,
-      "grad_norm": 8.481389999389648,
+      "grad_norm": 5.964265823364258,
       "learning_rate": 3.9455095862764884e-05,
-      "loss": 6.4046,
+      "loss": 6.3877,
       "step": 40000
     },
     {
       "epoch": 16.79261125104954,
-      "eval_accuracy": 0.15229337221038466,
-      "eval_loss": 9.075252532958984,
-      "eval_runtime": 1.0089,
-      "eval_samples_per_second": 474.758,
-      "eval_steps_per_second": 4.956,
+      "eval_accuracy": 0.15192889985421107,
+      "eval_loss": 9.037202835083008,
+      "eval_runtime": 1.0095,
+      "eval_samples_per_second": 474.495,
+      "eval_steps_per_second": 4.953,
       "step": 40000
     },
     {
       "epoch": 17.002518891687657,
-      "grad_norm": 5.657889366149902,
+      "grad_norm": 4.393352508544922,
       "learning_rate": 3.84460141271443e-05,
-      "loss": 6.4185,
+      "loss": 6.4003,
       "step": 40500
     },
     {
       "epoch": 17.212426532325775,
-      "grad_norm": 4.647271633148193,
+      "grad_norm": 7.377375602722168,
       "learning_rate": 3.7436932391523715e-05,
-      "loss": 6.3273,
+      "loss": 6.3061,
       "step": 41000
     },
     {
       "epoch": 17.422334172963897,
-      "grad_norm": 4.699731349945068,
+      "grad_norm": 7.525390625,
       "learning_rate": 3.642785065590313e-05,
-      "loss": 6.334,
+      "loss": 6.3168,
       "step": 41500
     },
     {
       "epoch": 17.632241813602015,
-      "grad_norm": 7.401587009429932,
+      "grad_norm": 7.932316780090332,
       "learning_rate": 3.541876892028255e-05,
-      "loss": 6.3376,
+      "loss": 6.3231,
       "step": 42000
     },
     {
       "epoch": 17.842149454240133,
-      "grad_norm": 4.6369829177856445,
+      "grad_norm": 5.1866254806518555,
       "learning_rate": 3.440968718466196e-05,
-      "loss": 6.3509,
+      "loss": 6.3289,
       "step": 42500
     },
     {
       "epoch": 18.052057094878254,
-      "grad_norm": 4.509687423706055,
+      "grad_norm": 5.159563064575195,
       "learning_rate": 3.340060544904137e-05,
-      "loss": 6.3251,
+      "loss": 6.3032,
       "step": 43000
     },
     {
       "epoch": 18.261964735516372,
-      "grad_norm": 5.544124126434326,
+      "grad_norm": 4.022120952606201,
       "learning_rate": 3.239152371342079e-05,
-      "loss": 6.2804,
+      "loss": 6.2456,
       "step": 43500
     },
     {
       "epoch": 18.471872376154494,
-      "grad_norm": 3.798172950744629,
+      "grad_norm": 8.648042678833008,
       "learning_rate": 3.13824419778002e-05,
-      "loss": 6.2926,
+      "loss": 6.257,
       "step": 44000
     },
     {
       "epoch": 18.68178001679261,
-      "grad_norm": 6.0228986740112305,
+      "grad_norm": 5.765574932098389,
       "learning_rate": 3.0373360242179615e-05,
-      "loss": 6.3,
+      "loss": 6.266,
       "step": 44500
     },
     {
       "epoch": 18.89168765743073,
-      "grad_norm": 5.522843837738037,
+      "grad_norm": 4.173483848571777,
       "learning_rate": 2.9364278506559034e-05,
-      "loss": 6.2983,
+      "loss": 6.266,
       "step": 45000
     },
     {
       "epoch": 19.10159529806885,
-      "grad_norm": 3.5186538696289062,
+      "grad_norm": 3.917922019958496,
       "learning_rate": 2.8355196770938446e-05,
-      "loss": 6.2643,
+      "loss": 6.2256,
       "step": 45500
     },
     {
       "epoch": 19.31150293870697,
-      "grad_norm": 5.115077972412109,
+      "grad_norm": 6.6297454833984375,
       "learning_rate": 2.734611503531786e-05,
-      "loss": 6.2344,
+      "loss": 6.1916,
       "step": 46000
     },
     {
       "epoch": 19.521410579345087,
-      "grad_norm": 5.235415458679199,
+      "grad_norm": 3.7506961822509766,
       "learning_rate": 2.6337033299697278e-05,
-      "loss": 6.238,
+      "loss": 6.1905,
       "step": 46500
     },
     {
       "epoch": 19.73131821998321,
-      "grad_norm": 5.123752117156982,
+      "grad_norm": 7.5525803565979,
       "learning_rate": 2.532795156407669e-05,
-      "loss": 6.2391,
+      "loss": 6.1934,
       "step": 47000
     },
     {
       "epoch": 19.941225860621326,
-      "grad_norm": 3.6134514808654785,
+      "grad_norm": 4.820414066314697,
       "learning_rate": 2.4318869828456106e-05,
-      "loss": 6.2322,
+      "loss": 6.1913,
       "step": 47500
     },
     {
       "epoch": 20.151133501259444,
-      "grad_norm": 4.235367298126221,
+      "grad_norm": 6.592065334320068,
       "learning_rate": 2.330978809283552e-05,
-      "loss": 6.2009,
+      "loss": 6.1547,
       "step": 48000
     },
     {
       "epoch": 20.361041141897566,
-      "grad_norm": 5.81327486038208,
+      "grad_norm": 4.041726589202881,
       "learning_rate": 2.2300706357214934e-05,
-      "loss": 6.1797,
+      "loss": 6.1346,
       "step": 48500
     },
     {
       "epoch": 20.570948782535684,
-      "grad_norm": 5.603066444396973,
+      "grad_norm": 5.67519998550415,
       "learning_rate": 2.129162462159435e-05,
-      "loss": 6.1779,
+      "loss": 6.135,
       "step": 49000
     },
     {
       "epoch": 20.7808564231738,
-      "grad_norm": 6.231359004974365,
+      "grad_norm": 4.125024318695068,
       "learning_rate": 2.0282542885973765e-05,
-      "loss": 6.1839,
+      "loss": 6.1374,
       "step": 49500
     },
     {
       "epoch": 20.990764063811923,
-      "grad_norm": 4.366462707519531,
+      "grad_norm": 5.8608717918396,
       "learning_rate": 1.927346115035318e-05,
-      "loss": 6.1809,
+      "loss": 6.1372,
       "step": 50000
     },
     {
       "epoch": 20.990764063811923,
-      "eval_accuracy": 0.15201727086662742,
-      "eval_loss": 9.93927001953125,
-      "eval_runtime": 1.0202,
-      "eval_samples_per_second": 469.507,
-      "eval_steps_per_second": 4.901,
+      "eval_accuracy": 0.1513443799590658,
+      "eval_loss": 9.643128395080566,
+      "eval_runtime": 1.0027,
+      "eval_samples_per_second": 477.713,
+      "eval_steps_per_second": 4.987,
       "step": 50000
     },
     {
       "epoch": 21.20067170445004,
-      "grad_norm": 4.904627799987793,
+      "grad_norm": 5.987393856048584,
       "learning_rate": 1.8264379414732593e-05,
-      "loss": 6.1187,
+      "loss": 6.0877,
       "step": 50500
     },
     {
       "epoch": 21.410579345088163,
-      "grad_norm": 5.952512264251709,
+      "grad_norm": 7.018879413604736,
       "learning_rate": 1.725529767911201e-05,
-      "loss": 6.1288,
+      "loss": 6.0939,
       "step": 51000
     },
     {
       "epoch": 21.62048698572628,
-      "grad_norm": 3.655320644378662,
+      "grad_norm": 7.489809036254883,
       "learning_rate": 1.6246215943491424e-05,
-      "loss": 6.1259,
+      "loss": 6.0883,
       "step": 51500
     },
     {
       "epoch": 21.8303946263644,
-      "grad_norm": 4.746984004974365,
+      "grad_norm": 6.501410007476807,
       "learning_rate": 1.5237134207870837e-05,
-      "loss": 6.1266,
+      "loss": 6.0909,
       "step": 52000
     },
     {
       "epoch": 22.04030226700252,
-      "grad_norm": 6.43018913269043,
+      "grad_norm": 4.224008083343506,
       "learning_rate": 1.4228052472250252e-05,
-      "loss": 6.1138,
+      "loss": 6.0784,
       "step": 52500
     },
     {
       "epoch": 22.250209907640638,
-      "grad_norm": 4.370454788208008,
+      "grad_norm": 3.761359691619873,
       "learning_rate": 1.3218970736629668e-05,
-      "loss": 6.0685,
+      "loss": 6.0484,
       "step": 53000
     },
     {
       "epoch": 22.460117548278756,
-      "grad_norm": 4.814517021179199,
+      "grad_norm": 8.934931755065918,
       "learning_rate": 1.2209889001009082e-05,
-      "loss": 6.074,
+      "loss": 6.0498,
       "step": 53500
     },
     {
       "epoch": 22.670025188916878,
-      "grad_norm": 4.246694564819336,
+      "grad_norm": 6.961613655090332,
       "learning_rate": 1.1200807265388496e-05,
-      "loss": 6.079,
+      "loss": 6.0567,
       "step": 54000
     },
     {
       "epoch": 22.879932829554996,
-      "grad_norm": 3.308920383453369,
+      "grad_norm": 4.204840660095215,
       "learning_rate": 1.0191725529767912e-05,
-      "loss": 6.0699,
+      "loss": 6.0513,
       "step": 54500
     },
     {
       "epoch": 23.089840470193113,
-      "grad_norm": 5.225849151611328,
+      "grad_norm": 4.377452373504639,
       "learning_rate": 9.182643794147326e-06,
-      "loss": 6.0549,
+      "loss": 6.0411,
       "step": 55000
     },
     {
       "epoch": 23.299748110831235,
-      "grad_norm": 5.21868896484375,
+      "grad_norm": 7.943485736846924,
       "learning_rate": 8.173562058526741e-06,
-      "loss": 6.0331,
+      "loss": 6.0242,
       "step": 55500
     },
     {
       "epoch": 23.509655751469353,
-      "grad_norm": 3.2828941345214844,
+      "grad_norm": 3.8972129821777344,
       "learning_rate": 7.164480322906156e-06,
-      "loss": 6.043,
+      "loss": 6.0325,
       "step": 56000
     },
     {
       "epoch": 23.719563392107474,
-      "grad_norm": 3.3772292137145996,
+      "grad_norm": 7.181693077087402,
       "learning_rate": 6.155398587285571e-06,
-      "loss": 6.0326,
+      "loss": 6.0228,
       "step": 56500
     },
     {
       "epoch": 23.929471032745592,
-      "grad_norm": 6.689046382904053,
+      "grad_norm": 3.5020320415496826,
       "learning_rate": 5.146316851664986e-06,
-      "loss": 6.0336,
+      "loss": 6.0215,
       "step": 57000
     },
     {
       "epoch": 24.13937867338371,
-      "grad_norm": 4.235395908355713,
+      "grad_norm": 3.9920434951782227,
       "learning_rate": 4.1372351160444e-06,
-      "loss": 6.0193,
+      "loss": 6.0147,
       "step": 57500
     },
     {
       "epoch": 24.349286314021832,
-      "grad_norm": 6.533717632293701,
+      "grad_norm": 6.99743127822876,
       "learning_rate": 3.1281533804238145e-06,
-      "loss": 6.0159,
+      "loss": 6.0165,
       "step": 58000
     },
     {
       "epoch": 24.55919395465995,
-      "grad_norm": 4.675073146820068,
+      "grad_norm": 4.573281288146973,
       "learning_rate": 2.1190716448032293e-06,
-      "loss": 6.0125,
+      "loss": 6.011,
       "step": 58500
     },
     {
       "epoch": 24.769101595298068,
-      "grad_norm": 4.745194911956787,
+      "grad_norm": 3.944985866546631,
       "learning_rate": 1.109989909182644e-06,
-      "loss": 6.0105,
+      "loss": 6.0039,
       "step": 59000
     },
     {
       "epoch": 24.97900923593619,
-      "grad_norm": 5.249101638793945,
+      "grad_norm": 3.4093751907348633,
       "learning_rate": 1.0090817356205853e-07,
-      "loss": 6.0042,
+      "loss": 6.0003,
       "step": 59500
     },
     {
       "epoch": 25.0,
       "step": 59550,
       "total_flos": 1.5149101316039424e+18,
-      "train_loss": 6.953392696813012,
-      "train_runtime": 24679.3114,
-      "train_samples_per_second": 231.61,
-      "train_steps_per_second": 2.413
+      "train_loss": 6.939091333602278,
+      "train_runtime": 24667.3069,
+      "train_samples_per_second": 231.723,
+      "train_steps_per_second": 2.414
     }
   ],
   "logging_steps": 500,