End of training

README.md

@@ -1,11 +1,23 @@
 ---
 tags:
 - generated_from_trainer
+datasets:
+- Hartunka/processed_wikitext-103-raw-v1-km-50_v2
 metrics:
 - accuracy
 model-index:
 - name: tiny_bert_km_50_v2
-  results: []
+  results:
+  - task:
+      name: Masked Language Modeling
+      type: fill-mask
+    dataset:
+      name: Hartunka/processed_wikitext-103-raw-v1-km-50_v2
+      type: Hartunka/processed_wikitext-103-raw-v1-km-50_v2
+    metrics:
+    - name: Accuracy
+      type: accuracy
+      value: 0.15262473865626944
 ---
 
 <!-- This model card has been generated automatically according to the information the Trainer had access to. You
@@ -13,10 +25,10 @@ should probably proofread and complete it, then remove this comment. -->
 
 # tiny_bert_km_50_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-50_v2 dataset.
 It achieves the following results on the evaluation set:
-- Loss: 7.0162
-- Accuracy: 0.1513
+- Loss: 6.8757
+- Accuracy: 0.1526
 
 ## Model description
 

all_results.json

@@ -1,16 +1,16 @@
 {
     "epoch": 25.0,
-    "eval_accuracy": 0.15287901904277562,
-    "eval_loss": 6.919051647186279,
-    "eval_runtime": 3.1728,
+    "eval_accuracy": 0.15262473865626944,
+    "eval_loss": 6.875680923461914,
+    "eval_runtime": 0.6832,
     "eval_samples": 479,
-    "eval_samples_per_second": 150.969,
-    "eval_steps_per_second": 1.576,
-    "perplexity": 1011.3604131025066,
+    "eval_samples_per_second": 701.123,
+    "eval_steps_per_second": 7.319,
+    "perplexity": 968.4345709456935,
     "total_flos": 3.053850510375936e+17,
-    "train_loss": 6.746545254877892,
-    "train_runtime": 11365.8868,
+    "train_loss": 6.665182815410828,
+    "train_runtime": 11393.6254,
     "train_samples": 228639,
-    "train_samples_per_second": 502.906,
-    "train_steps_per_second": 5.239
+    "train_samples_per_second": 501.682,
+    "train_steps_per_second": 5.227
 }

eval_results.json

@@ -1,10 +1,10 @@
 {
     "epoch": 25.0,
-    "eval_accuracy": 0.15287901904277562,
-    "eval_loss": 6.919051647186279,
-    "eval_runtime": 3.1728,
+    "eval_accuracy": 0.15262473865626944,
+    "eval_loss": 6.875680923461914,
+    "eval_runtime": 0.6832,
     "eval_samples": 479,
-    "eval_samples_per_second": 150.969,
-    "eval_steps_per_second": 1.576,
-    "perplexity": 1011.3604131025066
+    "eval_samples_per_second": 701.123,
+    "eval_steps_per_second": 7.319,
+    "perplexity": 968.4345709456935
 }

train_results.json

@@ -1,9 +1,9 @@
 {
     "epoch": 25.0,
     "total_flos": 3.053850510375936e+17,
-    "train_loss": 6.746545254877892,
-    "train_runtime": 11365.8868,
+    "train_loss": 6.665182815410828,
+    "train_runtime": 11393.6254,
     "train_samples": 228639,
-    "train_samples_per_second": 502.906,
-    "train_steps_per_second": 5.239
+    "train_samples_per_second": 501.682,
+    "train_steps_per_second": 5.227
 }

trainer_state.json

@@ -1,5 +1,5 @@
 {
-  "best_metric": 6.908801555633545,
+  "best_metric": 6.862132549285889,
   "best_model_checkpoint": "tiny_bert_km_50_v2/checkpoint-20000",
   "epoch": 25.0,
   "eval_steps": 10000,
@@ -10,890 +10,890 @@
   "log_history": [
     {
       "epoch": 0.20990764063811923,
-      "grad_norm": 2.5278005599975586,
+      "grad_norm": 2.5457608699798584,
       "learning_rate": 5e-06,
-      "loss": 13.6879,
+      "loss": 13.6904,
       "step": 500
     },
     {
       "epoch": 0.41981528127623846,
-      "grad_norm": 1.8328272104263306,
+      "grad_norm": 1.9976685047149658,
       "learning_rate": 1e-05,
-      "loss": 12.0905,
+      "loss": 12.0952,
       "step": 1000
     },
     {
       "epoch": 0.6297229219143576,
-      "grad_norm": 2.447697401046753,
+      "grad_norm": 2.022425651550293,
       "learning_rate": 1.5e-05,
-      "loss": 10.9219,
+      "loss": 10.9426,
       "step": 1500
     },
     {
       "epoch": 0.8396305625524769,
-      "grad_norm": 2.882899045944214,
+      "grad_norm": 4.413027763366699,
       "learning_rate": 2e-05,
-      "loss": 9.9398,
+      "loss": 10.0169,
       "step": 2000
     },
     {
       "epoch": 1.0495382031905962,
-      "grad_norm": 4.983320713043213,
+      "grad_norm": 4.057147026062012,
       "learning_rate": 2.5e-05,
-      "loss": 8.9743,
+      "loss": 9.1861,
       "step": 2500
     },
     {
       "epoch": 1.2594458438287153,
-      "grad_norm": 4.867501258850098,
+      "grad_norm": 4.614315509796143,
       "learning_rate": 3e-05,
-      "loss": 8.3759,
+      "loss": 8.6599,
       "step": 3000
     },
     {
       "epoch": 1.4693534844668346,
-      "grad_norm": 4.105288505554199,
+      "grad_norm": 4.219094276428223,
       "learning_rate": 3.5e-05,
-      "loss": 8.0367,
+      "loss": 8.3461,
       "step": 3500
     },
     {
       "epoch": 1.6792611251049538,
-      "grad_norm": 5.218428134918213,
+      "grad_norm": 4.526509761810303,
       "learning_rate": 4e-05,
-      "loss": 7.8283,
+      "loss": 8.1246,
       "step": 4000
     },
     {
       "epoch": 1.8891687657430731,
-      "grad_norm": 8.200092315673828,
+      "grad_norm": 5.163930416107178,
       "learning_rate": 4.5e-05,
-      "loss": 7.6834,
+      "loss": 7.9188,
       "step": 4500
     },
     {
       "epoch": 2.0990764063811924,
-      "grad_norm": 4.4852681159973145,
+      "grad_norm": 6.749960422515869,
       "learning_rate": 5e-05,
-      "loss": 7.5336,
+      "loss": 7.6939,
       "step": 5000
     },
     {
       "epoch": 2.3089840470193117,
-      "grad_norm": 7.6733503341674805,
+      "grad_norm": 4.7997212409973145,
       "learning_rate": 5.500000000000001e-05,
-      "loss": 7.4372,
+      "loss": 7.5121,
       "step": 5500
     },
     {
       "epoch": 2.5188916876574305,
-      "grad_norm": 9.709551811218262,
+      "grad_norm": 4.419384002685547,
       "learning_rate": 6e-05,
-      "loss": 7.4102,
+      "loss": 7.4201,
       "step": 6000
     },
     {
       "epoch": 2.72879932829555,
-      "grad_norm": 6.545389652252197,
+      "grad_norm": 5.388558387756348,
       "learning_rate": 6.500000000000001e-05,
-      "loss": 7.3711,
+      "loss": 7.3441,
       "step": 6500
     },
     {
       "epoch": 2.938706968933669,
-      "grad_norm": 4.4055681228637695,
+      "grad_norm": 4.232640743255615,
       "learning_rate": 7e-05,
-      "loss": 7.3186,
+      "loss": 7.2806,
       "step": 7000
     },
     {
       "epoch": 3.1486146095717884,
-      "grad_norm": 7.446049690246582,
+      "grad_norm": 4.257312297821045,
       "learning_rate": 7.500000000000001e-05,
-      "loss": 7.2252,
+      "loss": 7.1467,
       "step": 7500
     },
     {
       "epoch": 3.3585222502099077,
-      "grad_norm": 4.695464134216309,
+      "grad_norm": 5.981511116027832,
       "learning_rate": 8e-05,
-      "loss": 7.1983,
+      "loss": 7.0902,
       "step": 8000
     },
     {
       "epoch": 3.568429890848027,
-      "grad_norm": 5.45769739151001,
+      "grad_norm": 5.908140659332275,
       "learning_rate": 8.5e-05,
-      "loss": 7.2056,
+      "loss": 7.0863,
       "step": 8500
     },
     {
       "epoch": 3.7783375314861463,
-      "grad_norm": 6.298837661743164,
+      "grad_norm": 5.025755882263184,
       "learning_rate": 9e-05,
-      "loss": 7.1736,
+      "loss": 7.0683,
       "step": 9000
     },
     {
       "epoch": 3.988245172124265,
-      "grad_norm": 3.8956034183502197,
+      "grad_norm": 3.24112868309021,
       "learning_rate": 9.5e-05,
-      "loss": 7.169,
+      "loss": 7.0487,
       "step": 9500
     },
     {
       "epoch": 4.198152812762385,
-      "grad_norm": 4.570950508117676,
+      "grad_norm": 3.6224069595336914,
       "learning_rate": 0.0001,
-      "loss": 7.0543,
+      "loss": 6.9045,
       "step": 10000
     },
     {
       "epoch": 4.198152812762385,
-      "eval_accuracy": 0.14674865744734134,
-      "eval_loss": 7.051877975463867,
-      "eval_runtime": 0.6661,
-      "eval_samples_per_second": 719.089,
-      "eval_steps_per_second": 7.506,
+      "eval_accuracy": 0.1481120788001892,
+      "eval_loss": 6.953248500823975,
+      "eval_runtime": 0.6751,
+      "eval_samples_per_second": 709.51,
+      "eval_steps_per_second": 7.406,
       "step": 10000
     },
     {
       "epoch": 4.408060453400504,
-      "grad_norm": 3.881148338317871,
+      "grad_norm": 3.222562313079834,
       "learning_rate": 9.899091826437942e-05,
-      "loss": 7.065,
+      "loss": 6.9207,
       "step": 10500
     },
     {
       "epoch": 4.617968094038623,
-      "grad_norm": 4.209325790405273,
+      "grad_norm": 5.635316371917725,
       "learning_rate": 9.798183652875883e-05,
-      "loss": 7.0634,
+      "loss": 6.9195,
       "step": 11000
     },
     {
       "epoch": 4.827875734676742,
-      "grad_norm": 3.949389696121216,
+      "grad_norm": 3.868974447250366,
       "learning_rate": 9.697275479313825e-05,
-      "loss": 7.0477,
+      "loss": 6.9001,
       "step": 11500
     },
     {
       "epoch": 5.037783375314861,
-      "grad_norm": 3.636336326599121,
+      "grad_norm": 3.7188217639923096,
       "learning_rate": 9.596367305751767e-05,
-      "loss": 7.013,
+      "loss": 6.8697,
       "step": 12000
     },
     {
       "epoch": 5.247691015952981,
-      "grad_norm": 4.009646892547607,
+      "grad_norm": 4.994904041290283,
       "learning_rate": 9.495459132189707e-05,
-      "loss": 6.9222,
+      "loss": 6.7593,
       "step": 12500
     },
     {
       "epoch": 5.4575986565911,
-      "grad_norm": 4.858743190765381,
+      "grad_norm": 3.7432363033294678,
       "learning_rate": 9.394550958627649e-05,
-      "loss": 6.9282,
+      "loss": 6.7644,
       "step": 13000
     },
     {
       "epoch": 5.667506297229219,
-      "grad_norm": 4.67729377746582,
+      "grad_norm": 3.920727014541626,
       "learning_rate": 9.293642785065591e-05,
-      "loss": 6.9348,
+      "loss": 6.7767,
       "step": 13500
     },
     {
       "epoch": 5.877413937867338,
-      "grad_norm": 3.6789445877075195,
+      "grad_norm": 4.117716312408447,
       "learning_rate": 9.192734611503532e-05,
-      "loss": 6.9283,
+      "loss": 6.7797,
       "step": 14000
     },
     {
       "epoch": 6.087321578505458,
-      "grad_norm": 4.320778846740723,
+      "grad_norm": 2.823979377746582,
       "learning_rate": 9.091826437941474e-05,
-      "loss": 6.8775,
+      "loss": 6.717,
       "step": 14500
     },
     {
       "epoch": 6.297229219143577,
-      "grad_norm": 5.599698066711426,
+      "grad_norm": 3.6684699058532715,
       "learning_rate": 8.990918264379415e-05,
-      "loss": 6.82,
+      "loss": 6.6533,
       "step": 15000
     },
     {
       "epoch": 6.507136859781696,
-      "grad_norm": 4.778390407562256,
+      "grad_norm": 2.289646625518799,
       "learning_rate": 8.890010090817356e-05,
-      "loss": 6.8253,
+      "loss": 6.6614,
       "step": 15500
     },
     {
       "epoch": 6.717044500419815,
-      "grad_norm": 6.671377658843994,
+      "grad_norm": 2.6362833976745605,
       "learning_rate": 8.789101917255298e-05,
-      "loss": 6.8281,
+      "loss": 6.6781,
       "step": 16000
     },
     {
       "epoch": 6.926952141057934,
-      "grad_norm": 5.022947788238525,
+      "grad_norm": 3.094486713409424,
       "learning_rate": 8.68819374369324e-05,
-      "loss": 6.8246,
+      "loss": 6.6794,
       "step": 16500
     },
     {
       "epoch": 7.136859781696054,
-      "grad_norm": 4.0362868309021,
+      "grad_norm": 3.2316012382507324,
       "learning_rate": 8.58728557013118e-05,
-      "loss": 6.7473,
+      "loss": 6.5975,
       "step": 17000
     },
     {
       "epoch": 7.346767422334173,
-      "grad_norm": 5.0283966064453125,
+      "grad_norm": 4.46132755279541,
       "learning_rate": 8.486377396569122e-05,
-      "loss": 6.7194,
+      "loss": 6.5673,
       "step": 17500
     },
     {
       "epoch": 7.5566750629722925,
-      "grad_norm": 3.8201303482055664,
+      "grad_norm": 3.2164196968078613,
       "learning_rate": 8.385469223007064e-05,
-      "loss": 6.7321,
+      "loss": 6.5826,
       "step": 18000
     },
     {
       "epoch": 7.766582703610411,
-      "grad_norm": 4.430297374725342,
+      "grad_norm": 2.46340274810791,
       "learning_rate": 8.284561049445006e-05,
-      "loss": 6.7489,
+      "loss": 6.5899,
       "step": 18500
     },
     {
       "epoch": 7.97649034424853,
-      "grad_norm": 3.80474591255188,
+      "grad_norm": 4.145327568054199,
       "learning_rate": 8.183652875882947e-05,
-      "loss": 6.7513,
+      "loss": 6.5954,
       "step": 19000
     },
     {
       "epoch": 8.18639798488665,
-      "grad_norm": 3.42175555229187,
+      "grad_norm": 3.2221481800079346,
       "learning_rate": 8.082744702320889e-05,
-      "loss": 6.6528,
+      "loss": 6.4945,
       "step": 19500
     },
     {
       "epoch": 8.39630562552477,
-      "grad_norm": 3.7943239212036133,
+      "grad_norm": 2.3382699489593506,
       "learning_rate": 7.98183652875883e-05,
-      "loss": 6.6574,
+      "loss": 6.4983,
       "step": 20000
     },
     {
       "epoch": 8.39630562552477,
-      "eval_accuracy": 0.15168775779342855,
-      "eval_loss": 6.908801555633545,
-      "eval_runtime": 0.6558,
-      "eval_samples_per_second": 730.4,
-      "eval_steps_per_second": 7.624,
+      "eval_accuracy": 0.15236117736187885,
+      "eval_loss": 6.862132549285889,
+      "eval_runtime": 0.66,
+      "eval_samples_per_second": 725.741,
+      "eval_steps_per_second": 7.576,
       "step": 20000
     },
     {
       "epoch": 8.606213266162888,
-      "grad_norm": 6.637362003326416,
+      "grad_norm": 4.08376407623291,
       "learning_rate": 7.880928355196771e-05,
-      "loss": 6.664,
+      "loss": 6.5109,
       "step": 20500
     },
     {
       "epoch": 8.816120906801007,
-      "grad_norm": 3.60398006439209,
+      "grad_norm": 4.051946640014648,
       "learning_rate": 7.780020181634713e-05,
-      "loss": 6.6703,
+      "loss": 6.5219,
       "step": 21000
     },
     {
       "epoch": 9.026028547439127,
-      "grad_norm": 2.7140209674835205,
+      "grad_norm": 4.16067361831665,
       "learning_rate": 7.679112008072655e-05,
-      "loss": 6.6515,
+      "loss": 6.5118,
       "step": 21500
     },
     {
       "epoch": 9.235936188077247,
-      "grad_norm": 4.183799743652344,
+      "grad_norm": 3.599525213241577,
       "learning_rate": 7.578203834510595e-05,
-      "loss": 6.5675,
+      "loss": 6.4274,
       "step": 22000
     },
     {
       "epoch": 9.445843828715365,
-      "grad_norm": 4.17147159576416,
+      "grad_norm": 3.1556954383850098,
       "learning_rate": 7.477295660948537e-05,
-      "loss": 6.5816,
+      "loss": 6.4345,
       "step": 22500
     },
     {
       "epoch": 9.655751469353484,
-      "grad_norm": 3.254685878753662,
+      "grad_norm": 3.792712688446045,
       "learning_rate": 7.376387487386479e-05,
-      "loss": 6.5925,
+      "loss": 6.4482,
       "step": 23000
     },
     {
       "epoch": 9.865659109991604,
-      "grad_norm": 3.2318437099456787,
+      "grad_norm": 3.364772319793701,
       "learning_rate": 7.27547931382442e-05,
-      "loss": 6.5859,
+      "loss": 6.4528,
       "step": 23500
     },
     {
       "epoch": 10.075566750629722,
-      "grad_norm": 3.645916700363159,
+      "grad_norm": 3.4910032749176025,
       "learning_rate": 7.174571140262362e-05,
-      "loss": 6.5574,
+      "loss": 6.4249,
       "step": 24000
     },
     {
       "epoch": 10.285474391267842,
-      "grad_norm": 3.421893835067749,
+      "grad_norm": 3.6031293869018555,
       "learning_rate": 7.073662966700304e-05,
-      "loss": 6.514,
+      "loss": 6.3817,
       "step": 24500
     },
     {
       "epoch": 10.495382031905962,
-      "grad_norm": 4.209316253662109,
+      "grad_norm": 4.115103721618652,
       "learning_rate": 6.972754793138244e-05,
-      "loss": 6.5228,
+      "loss": 6.3884,
       "step": 25000
     },
     {
       "epoch": 10.705289672544081,
-      "grad_norm": 4.944457530975342,
+      "grad_norm": 3.949532985687256,
       "learning_rate": 6.871846619576186e-05,
-      "loss": 6.5328,
+      "loss": 6.3931,
       "step": 25500
     },
     {
       "epoch": 10.9151973131822,
-      "grad_norm": 3.14280366897583,
+      "grad_norm": 3.3997602462768555,
       "learning_rate": 6.770938446014128e-05,
-      "loss": 6.5329,
+      "loss": 6.4059,
       "step": 26000
     },
     {
       "epoch": 11.125104953820319,
-      "grad_norm": 3.3328511714935303,
+      "grad_norm": 3.8906116485595703,
       "learning_rate": 6.670030272452068e-05,
-      "loss": 6.4866,
+      "loss": 6.3588,
       "step": 26500
     },
     {
       "epoch": 11.335012594458439,
-      "grad_norm": 4.425465106964111,
+      "grad_norm": 4.786259174346924,
       "learning_rate": 6.56912209889001e-05,
-      "loss": 6.4601,
+      "loss": 6.3402,
       "step": 27000
     },
     {
       "epoch": 11.544920235096557,
-      "grad_norm": 3.1440634727478027,
+      "grad_norm": 4.122541904449463,
       "learning_rate": 6.468213925327952e-05,
-      "loss": 6.4694,
+      "loss": 6.3444,
       "step": 27500
     },
     {
       "epoch": 11.754827875734676,
-      "grad_norm": 3.96699595451355,
+      "grad_norm": 3.0653414726257324,
       "learning_rate": 6.367305751765893e-05,
-      "loss": 6.4803,
+      "loss": 6.3425,
       "step": 28000
     },
     {
       "epoch": 11.964735516372796,
-      "grad_norm": 3.266310214996338,
+      "grad_norm": 3.5233154296875,
       "learning_rate": 6.266397578203835e-05,
-      "loss": 6.4802,
+      "loss": 6.3581,
       "step": 28500
     },
     {
       "epoch": 12.174643157010916,
-      "grad_norm": 2.9504129886627197,
+      "grad_norm": 3.1181325912475586,
       "learning_rate": 6.165489404641777e-05,
-      "loss": 6.4162,
+      "loss": 6.3099,
       "step": 29000
     },
     {
       "epoch": 12.384550797649034,
-      "grad_norm": 3.244772434234619,
+      "grad_norm": 3.411161422729492,
       "learning_rate": 6.064581231079718e-05,
-      "loss": 6.4129,
+      "loss": 6.2904,
       "step": 29500
     },
     {
       "epoch": 12.594458438287154,
-      "grad_norm": 3.7076940536499023,
+      "grad_norm": 2.3188114166259766,
       "learning_rate": 5.963673057517659e-05,
-      "loss": 6.4302,
+      "loss": 6.3069,
       "step": 30000
     },
     {
       "epoch": 12.594458438287154,
-      "eval_accuracy": 0.1537062428295604,
-      "eval_loss": 6.943292617797852,
-      "eval_runtime": 0.6615,
-      "eval_samples_per_second": 724.101,
-      "eval_steps_per_second": 7.558,
+      "eval_accuracy": 0.15325852757646138,
+      "eval_loss": 6.876902103424072,
+      "eval_runtime": 0.6675,
+      "eval_samples_per_second": 717.648,
+      "eval_steps_per_second": 7.491,
       "step": 30000
     },
     {
       "epoch": 12.804366078925273,
-      "grad_norm": 5.103267192840576,
+      "grad_norm": 2.651759147644043,
       "learning_rate": 5.8627648839556e-05,
-      "loss": 6.4281,
+      "loss": 6.3056,
       "step": 30500
     },
     {
       "epoch": 13.014273719563391,
-      "grad_norm": 4.839158058166504,
+      "grad_norm": 2.748534679412842,
       "learning_rate": 5.7618567103935416e-05,
-      "loss": 6.4262,
+      "loss": 6.3087,
       "step": 31000
     },
     {
       "epoch": 13.224181360201511,
-      "grad_norm": 3.3252675533294678,
+      "grad_norm": 3.192431688308716,
       "learning_rate": 5.660948536831484e-05,
-      "loss": 6.3687,
+      "loss": 6.2597,
       "step": 31500
     },
     {
       "epoch": 13.43408900083963,
-      "grad_norm": 3.2508044242858887,
+      "grad_norm": 3.152599334716797,
       "learning_rate": 5.5600403632694254e-05,
-      "loss": 6.3736,
+      "loss": 6.264,
       "step": 32000
     },
     {
       "epoch": 13.64399664147775,
-      "grad_norm": 5.2709784507751465,
+      "grad_norm": 2.5589845180511475,
       "learning_rate": 5.4591321897073666e-05,
-      "loss": 6.3843,
+      "loss": 6.27,
       "step": 32500
     },
     {
       "epoch": 13.853904282115868,
-      "grad_norm": 4.30772590637207,
+      "grad_norm": 3.1426684856414795,
       "learning_rate": 5.358224016145308e-05,
-      "loss": 6.3874,
+      "loss": 6.2703,
       "step": 33000
     },
     {
       "epoch": 14.063811922753988,
-      "grad_norm": 2.8513612747192383,
+      "grad_norm": 3.0625624656677246,
       "learning_rate": 5.257315842583249e-05,
-      "loss": 6.3712,
+      "loss": 6.2625,
       "step": 33500
     },
     {
       "epoch": 14.273719563392108,
-      "grad_norm": 3.778689384460449,
+      "grad_norm": 2.392235517501831,
       "learning_rate": 5.15640766902119e-05,
-      "loss": 6.3288,
+      "loss": 6.2273,
       "step": 34000
     },
     {
       "epoch": 14.483627204030226,
-      "grad_norm": 4.107575416564941,
+      "grad_norm": 2.8677453994750977,
       "learning_rate": 5.055499495459133e-05,
-      "loss": 6.337,
+      "loss": 6.2331,
       "step": 34500
     },
     {
       "epoch": 14.693534844668346,
-      "grad_norm": 4.152520179748535,
+      "grad_norm": 2.934575080871582,
       "learning_rate": 4.954591321897074e-05,
-      "loss": 6.3444,
+      "loss": 6.2335,
       "step": 35000
     },
     {
       "epoch": 14.903442485306465,
-      "grad_norm": 4.295997619628906,
+      "grad_norm": 2.294266700744629,
       "learning_rate": 4.853683148335015e-05,
-      "loss": 6.35,
+      "loss": 6.2387,
       "step": 35500
     },
     {
       "epoch": 15.113350125944585,
-      "grad_norm": 4.052658557891846,
+      "grad_norm": 4.032185077667236,
       "learning_rate": 4.7527749747729566e-05,
-      "loss": 6.3181,
+      "loss": 6.2165,
       "step": 36000
     },
     {
       "epoch": 15.323257766582703,
-      "grad_norm": 3.2357518672943115,
+      "grad_norm": 3.949481248855591,
       "learning_rate": 4.6518668012108985e-05,
-      "loss": 6.3042,
+      "loss": 6.2054,
       "step": 36500
     },
     {
       "epoch": 15.533165407220823,
-      "grad_norm": 4.015797138214111,
+      "grad_norm": 4.527189254760742,
       "learning_rate": 4.55095862764884e-05,
-      "loss": 6.3018,
+      "loss": 6.2032,
       "step": 37000
     },
     {
       "epoch": 15.743073047858942,
-      "grad_norm": 4.377602577209473,
+      "grad_norm": 3.2532660961151123,
       "learning_rate": 4.450050454086781e-05,
-      "loss": 6.312,
+      "loss": 6.2092,
       "step": 37500
     },
     {
       "epoch": 15.95298068849706,
-      "grad_norm": 4.442850589752197,
+      "grad_norm": 3.8207716941833496,
       "learning_rate": 4.349142280524723e-05,
-      "loss": 6.3126,
+      "loss": 6.208,
       "step": 38000
     },
     {
       "epoch": 16.162888329135182,
-      "grad_norm": 4.529189586639404,
+      "grad_norm": 3.3611695766448975,
       "learning_rate": 4.248234106962664e-05,
-      "loss": 6.2777,
+      "loss": 6.185,
       "step": 38500
     },
     {
       "epoch": 16.3727959697733,
-      "grad_norm": 4.3285698890686035,
+      "grad_norm": 3.6895010471343994,
       "learning_rate": 4.147325933400605e-05,
-      "loss": 6.267,
+      "loss": 6.1816,
       "step": 39000
     },
     {
       "epoch": 16.582703610411418,
-      "grad_norm": 2.7106502056121826,
+      "grad_norm": 4.3816118240356445,
       "learning_rate": 4.046417759838547e-05,
-      "loss": 6.281,
+      "loss": 6.1858,
       "step": 39500
     },
     {
       "epoch": 16.79261125104954,
-      "grad_norm": 3.3317830562591553,
+      "grad_norm": 2.5966224670410156,
       "learning_rate": 3.9455095862764884e-05,
-      "loss": 6.2677,
+      "loss": 6.1769,
       "step": 40000
     },
     {
       "epoch": 16.79261125104954,
-      "eval_accuracy": 0.15170460917348885,
-      "eval_loss": 7.032588481903076,
-      "eval_runtime": 0.6538,
-      "eval_samples_per_second": 732.588,
-      "eval_steps_per_second": 7.647,
+      "eval_accuracy": 0.1522653358752944,
+      "eval_loss": 6.953736782073975,
+      "eval_runtime": 0.6715,
+      "eval_samples_per_second": 713.277,
+      "eval_steps_per_second": 7.445,
       "step": 40000
     },
     {
       "epoch": 17.002518891687657,
-      "grad_norm": 4.439164638519287,
+      "grad_norm": 4.307312965393066,
       "learning_rate": 3.84460141271443e-05,
-      "loss": 6.2766,
+      "loss": 6.1829,
       "step": 40500
     },
     {
       "epoch": 17.212426532325775,
-      "grad_norm": 2.927025556564331,
+      "grad_norm": 1.7894271612167358,
       "learning_rate": 3.7436932391523715e-05,
-      "loss": 6.2411,
+      "loss": 6.1541,
       "step": 41000
     },
     {
       "epoch": 17.422334172963897,
-      "grad_norm": 3.9676928520202637,
+      "grad_norm": 5.252026081085205,
       "learning_rate": 3.642785065590313e-05,
-      "loss": 6.238,
+      "loss": 6.1546,
       "step": 41500
     },
     {
       "epoch": 17.632241813602015,
-      "grad_norm": 2.0777578353881836,
+      "grad_norm": 2.3743743896484375,
       "learning_rate": 3.541876892028255e-05,
-      "loss": 6.2385,
+      "loss": 6.1615,
       "step": 42000
     },
     {
       "epoch": 17.842149454240133,
-      "grad_norm": 2.9264719486236572,
+      "grad_norm": 3.007002353668213,
       "learning_rate": 3.440968718466196e-05,
-      "loss": 6.2479,
+      "loss": 6.1567,
       "step": 42500
     },
     {
       "epoch": 18.052057094878254,
-      "grad_norm": 2.951683282852173,
+      "grad_norm": 3.2413620948791504,
       "learning_rate": 3.340060544904137e-05,
-      "loss": 6.2383,
+      "loss": 6.1539,
       "step": 43000
     },
     {
       "epoch": 18.261964735516372,
-      "grad_norm": 3.7961654663085938,
+      "grad_norm": 3.3119919300079346,
       "learning_rate": 3.239152371342079e-05,
-      "loss": 6.2117,
+      "loss": 6.1324,
       "step": 43500
     },
     {
       "epoch": 18.471872376154494,
-      "grad_norm": 4.096261501312256,
+      "grad_norm": 2.2704648971557617,
       "learning_rate": 3.13824419778002e-05,
-      "loss": 6.222,
+      "loss": 6.1371,
       "step": 44000
     },
     {
       "epoch": 18.68178001679261,
-      "grad_norm": 3.113459587097168,
+      "grad_norm": 2.4617161750793457,
       "learning_rate": 3.0373360242179615e-05,
-      "loss": 6.2172,
+      "loss": 6.1387,
       "step": 44500
     },
     {
       "epoch": 18.89168765743073,
-      "grad_norm": 4.582091808319092,
+      "grad_norm": 2.70706844329834,
       "learning_rate": 2.9364278506559034e-05,
-      "loss": 6.2237,
+      "loss": 6.1423,
       "step": 45000
     },
     {
       "epoch": 19.10159529806885,
-      "grad_norm": 4.693905830383301,
+      "grad_norm": 3.604302406311035,
       "learning_rate": 2.8355196770938446e-05,
-      "loss": 6.2129,
+      "loss": 6.1338,
       "step": 45500
     },
     {
       "epoch": 19.31150293870697,
-      "grad_norm": 5.093323707580566,
+      "grad_norm": 1.6419793367385864,
       "learning_rate": 2.734611503531786e-05,
-      "loss": 6.1936,
+      "loss": 6.1181,
       "step": 46000
     },
     {
       "epoch": 19.521410579345087,
-      "grad_norm": 3.8527095317840576,
+      "grad_norm": 3.340660810470581,
       "learning_rate": 2.6337033299697278e-05,
-      "loss": 6.197,
+      "loss": 6.1174,
       "step": 46500
     },
     {
       "epoch": 19.73131821998321,
-      "grad_norm": 3.2054660320281982,
+      "grad_norm": 3.506154775619507,
       "learning_rate": 2.532795156407669e-05,
-      "loss": 6.1978,
+      "loss": 6.1221,
       "step": 47000
     },
     {
       "epoch": 19.941225860621326,
-      "grad_norm": 2.120891571044922,
+      "grad_norm": 3.1330461502075195,
       "learning_rate": 2.4318869828456106e-05,
-      "loss": 6.196,
+      "loss": 6.1199,
       "step": 47500
     },
     {
       "epoch": 20.151133501259444,
-      "grad_norm": 3.4394898414611816,
+      "grad_norm": 2.368105888366699,
       "learning_rate": 2.330978809283552e-05,
-      "loss": 6.1801,
+      "loss": 6.1094,
       "step": 48000
     },
     {
       "epoch": 20.361041141897566,
-      "grad_norm": 4.150263786315918,
+      "grad_norm": 4.75828218460083,
       "learning_rate": 2.2300706357214934e-05,
-      "loss": 6.175,
+      "loss": 6.1054,
       "step": 48500
     },
     {
       "epoch": 20.570948782535684,
-      "grad_norm": 4.106333255767822,
+      "grad_norm": 2.1988234519958496,
       "learning_rate": 2.129162462159435e-05,
-      "loss": 6.1741,
+      "loss": 6.0968,
       "step": 49000
     },
     {
       "epoch": 20.7808564231738,
-      "grad_norm": 3.705505132675171,
+      "grad_norm": 2.619251251220703,
       "learning_rate": 2.0282542885973765e-05,
-      "loss": 6.1741,
+      "loss": 6.1015,
       "step": 49500
     },
     {
       "epoch": 20.990764063811923,
-      "grad_norm": 3.1185195446014404,
+      "grad_norm": 2.415771961212158,
       "learning_rate": 1.927346115035318e-05,
-      "loss": 6.1767,
+      "loss": 6.0989,
       "step": 50000
     },
     {
       "epoch": 20.990764063811923,
-      "eval_accuracy": 0.15019485799198137,
-      "eval_loss": 7.045273780822754,
-      "eval_runtime": 0.6572,
-      "eval_samples_per_second": 728.885,
-      "eval_steps_per_second": 7.608,
+      "eval_accuracy": 0.1513163428379174,
+      "eval_loss": 7.016210079193115,
+      "eval_runtime": 0.6608,
+      "eval_samples_per_second": 724.908,
+      "eval_steps_per_second": 7.567,
       "step": 50000
     },
     {
       "epoch": 21.20067170445004,
-      "grad_norm": 3.2303192615509033,
+      "grad_norm": 4.247645378112793,
       "learning_rate": 1.8264379414732593e-05,
-      "loss": 6.1572,
+      "loss": 6.0886,
       "step": 50500
     },
     {
       "epoch": 21.410579345088163,
-      "grad_norm": 5.81098747253418,
+      "grad_norm": 2.1635053157806396,
       "learning_rate": 1.725529767911201e-05,
-      "loss": 6.1592,
+      "loss": 6.0907,
       "step": 51000
     },
     {
       "epoch": 21.62048698572628,
-      "grad_norm": 3.357577323913574,
+      "grad_norm": 1.7626135349273682,
       "learning_rate": 1.6246215943491424e-05,
-      "loss": 6.1551,
+      "loss": 6.0892,
       "step": 51500
     },
     {
       "epoch": 21.8303946263644,
-      "grad_norm": 2.948676109313965,
+      "grad_norm": 3.2080373764038086,
       "learning_rate": 1.5237134207870837e-05,
-      "loss": 6.1586,
+      "loss": 6.088,
       "step": 52000
     },
     {
       "epoch": 22.04030226700252,
-      "grad_norm": 3.3979671001434326,
+      "grad_norm": 1.8654497861862183,
       "learning_rate": 1.4228052472250252e-05,
-      "loss": 6.1513,
+      "loss": 6.0884,
       "step": 52500
     },
     {
       "epoch": 22.250209907640638,
-      "grad_norm": 4.48105001449585,
+      "grad_norm": 1.773589015007019,
       "learning_rate": 1.3218970736629668e-05,
-      "loss": 6.1389,
+      "loss": 6.0777,
       "step": 53000
     },
     {
       "epoch": 22.460117548278756,
-      "grad_norm": 2.395787000656128,
+      "grad_norm": 3.663158655166626,
       "learning_rate": 1.2209889001009082e-05,
-      "loss": 6.1418,
+      "loss": 6.0786,
       "step": 53500
     },
     {
       "epoch": 22.670025188916878,
-      "grad_norm": 2.878861904144287,
+      "grad_norm": 1.273491621017456,
       "learning_rate": 1.1200807265388496e-05,
-      "loss": 6.1421,
+      "loss": 6.0808,
       "step": 54000
     },
     {
       "epoch": 22.879932829554996,
-      "grad_norm": 1.2426501512527466,
+      "grad_norm": 3.485517978668213,
       "learning_rate": 1.0191725529767912e-05,
-      "loss": 6.1405,
+      "loss": 6.0746,
       "step": 54500
     },
     {
       "epoch": 23.089840470193113,
-      "grad_norm": 3.5291991233825684,
+      "grad_norm": 2.721423387527466,
       "learning_rate": 9.182643794147326e-06,
-      "loss": 6.1368,
+      "loss": 6.0722,
       "step": 55000
     },
     {
       "epoch": 23.299748110831235,
-      "grad_norm": 1.9686766862869263,
+      "grad_norm": 4.095690727233887,
       "learning_rate": 8.173562058526741e-06,
-      "loss": 6.1298,
+      "loss": 6.0675,
       "step": 55500
     },
     {
       "epoch": 23.509655751469353,
-      "grad_norm": 3.947535276412964,
+      "grad_norm": 3.137434482574463,
       "learning_rate": 7.164480322906156e-06,
-      "loss": 6.1326,
+      "loss": 6.0715,
       "step": 56000
     },
     {
       "epoch": 23.719563392107474,
-      "grad_norm": 3.063832998275757,
+      "grad_norm": 3.0779168605804443,
       "learning_rate": 6.155398587285571e-06,
-      "loss": 6.1256,
+      "loss": 6.0643,
       "step": 56500
     },
     {
       "epoch": 23.929471032745592,
-      "grad_norm": 4.72531270980835,
+      "grad_norm": 1.6233700513839722,
       "learning_rate": 5.146316851664986e-06,
-      "loss": 6.1318,
+      "loss": 6.0682,
       "step": 57000
     },
     {
       "epoch": 24.13937867338371,
-      "grad_norm": 4.100183486938477,
+      "grad_norm": 1.3678042888641357,
       "learning_rate": 4.1372351160444e-06,
-      "loss": 6.1242,
+      "loss": 6.0611,
       "step": 57500
     },
     {
       "epoch": 24.349286314021832,
-      "grad_norm": 3.9492807388305664,
+      "grad_norm": 3.5903029441833496,
       "learning_rate": 3.1281533804238145e-06,
-      "loss": 6.1216,
+      "loss": 6.0639,
       "step": 58000
     },
     {
       "epoch": 24.55919395465995,
-      "grad_norm": 4.388216972351074,
+      "grad_norm": 1.1615397930145264,
       "learning_rate": 2.1190716448032293e-06,
-      "loss": 6.1179,
+      "loss": 6.0601,
       "step": 58500
     },
     {
       "epoch": 24.769101595298068,
-      "grad_norm": 2.9845166206359863,
+      "grad_norm": 0.8376772999763489,
       "learning_rate": 1.109989909182644e-06,
-      "loss": 6.1155,
+      "loss": 6.0579,
       "step": 59000
     },
     {
       "epoch": 24.97900923593619,
-      "grad_norm": 2.361599922180176,
+      "grad_norm": 2.6765823364257812,
       "learning_rate": 1.0090817356205853e-07,
-      "loss": 6.1164,
+      "loss": 6.0586,
       "step": 59500
     },
     {
       "epoch": 25.0,
       "step": 59550,
       "total_flos": 3.053850510375936e+17,
-      "train_loss": 6.746545254877892,
-      "train_runtime": 11365.8868,
-      "train_samples_per_second": 502.906,
-      "train_steps_per_second": 5.239
+      "train_loss": 6.665182815410828,
+      "train_runtime": 11393.6254,
+      "train_samples_per_second": 501.682,
+      "train_steps_per_second": 5.227
     }
   ],
   "logging_steps": 500,