End of training

README.md

@@ -1,11 +1,23 @@
 ---
 tags:
 - generated_from_trainer
+datasets:
+- Hartunka/processed_wikitext-103-raw-v1-rand-20_v2
 metrics:
 - accuracy
 model-index:
 - name: tiny_bert_rand_20_v2
-  results: []
+  results:
+  - task:
+      name: Masked Language Modeling
+      type: fill-mask
+    dataset:
+      name: Hartunka/processed_wikitext-103-raw-v1-rand-20_v2
+      type: Hartunka/processed_wikitext-103-raw-v1-rand-20_v2
+    metrics:
+    - name: Accuracy
+      type: accuracy
+      value: 0.15220093801209245
 ---
 
 <!-- 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_rand_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-rand-20_v2 dataset.
 It achieves the following results on the evaluation set:
-- Loss: 11.3958
-- Accuracy: 0.1511
+- Loss: 9.2854
+- Accuracy: 0.1522
 
 ## Model description
 

all_results.json

@@ -1,16 +1,16 @@
 {
     "epoch": 25.0,
-    "eval_accuracy": 0.15242696502232017,
-    "eval_loss": 9.300938606262207,
-    "eval_runtime": 0.7233,
+    "eval_accuracy": 0.15220093801209245,
+    "eval_loss": 9.285418510437012,
+    "eval_runtime": 0.7076,
     "eval_samples": 479,
-    "eval_samples_per_second": 662.21,
-    "eval_steps_per_second": 6.912,
-    "perplexity": 10948.290521095358,
+    "eval_samples_per_second": 676.961,
+    "eval_steps_per_second": 7.066,
+    "perplexity": 10779.683783837943,
     "total_flos": 3.051148107142656e+17,
-    "train_loss": 8.18093173522693,
-    "train_runtime": 11970.8684,
+    "train_loss": 8.201283628782438,
+    "train_runtime": 12621.2502,
     "train_samples": 228639,
-    "train_samples_per_second": 477.49,
-    "train_steps_per_second": 4.975
+    "train_samples_per_second": 452.885,
+    "train_steps_per_second": 4.718
 }

eval_results.json

@@ -1,10 +1,10 @@
 {
     "epoch": 25.0,
-    "eval_accuracy": 0.15242696502232017,
-    "eval_loss": 9.300938606262207,
-    "eval_runtime": 0.7233,
+    "eval_accuracy": 0.15220093801209245,
+    "eval_loss": 9.285418510437012,
+    "eval_runtime": 0.7076,
     "eval_samples": 479,
-    "eval_samples_per_second": 662.21,
-    "eval_steps_per_second": 6.912,
-    "perplexity": 10948.290521095358
+    "eval_samples_per_second": 676.961,
+    "eval_steps_per_second": 7.066,
+    "perplexity": 10779.683783837943
 }

train_results.json

@@ -1,9 +1,9 @@
 {
     "epoch": 25.0,
     "total_flos": 3.051148107142656e+17,
-    "train_loss": 8.18093173522693,
-    "train_runtime": 11970.8684,
+    "train_loss": 8.201283628782438,
+    "train_runtime": 12621.2502,
     "train_samples": 228639,
-    "train_samples_per_second": 477.49,
-    "train_steps_per_second": 4.975
+    "train_samples_per_second": 452.885,
+    "train_steps_per_second": 4.718
 }

trainer_state.json

@@ -1,5 +1,5 @@
 {
-  "best_metric": 9.296181678771973,
+  "best_metric": 9.243675231933594,
   "best_model_checkpoint": "tiny_bert_rand_20_v2/checkpoint-10000",
   "epoch": 25.0,
   "eval_steps": 10000,
@@ -10,890 +10,890 @@
   "log_history": [
     {
       "epoch": 0.20990764063811923,
-      "grad_norm": 2.4613113403320312,
+      "grad_norm": 2.6130077838897705,
       "learning_rate": 5e-06,
-      "loss": 12.7747,
+      "loss": 12.7744,
       "step": 500
     },
     {
       "epoch": 0.41981528127623846,
-      "grad_norm": 1.6948860883712769,
+      "grad_norm": 1.9487887620925903,
       "learning_rate": 1e-05,
-      "loss": 11.2072,
+      "loss": 11.2064,
       "step": 1000
     },
     {
       "epoch": 0.6297229219143576,
-      "grad_norm": 1.6647202968597412,
+      "grad_norm": 1.1609364748001099,
       "learning_rate": 1.5e-05,
-      "loss": 10.212,
+      "loss": 10.2112,
       "step": 1500
     },
     {
       "epoch": 0.8396305625524769,
-      "grad_norm": 1.5998878479003906,
+      "grad_norm": 1.2349486351013184,
       "learning_rate": 2e-05,
-      "loss": 10.0127,
+      "loss": 10.0102,
       "step": 2000
     },
     {
       "epoch": 1.0495382031905962,
-      "grad_norm": 1.9302012920379639,
+      "grad_norm": 1.3904448747634888,
       "learning_rate": 2.5e-05,
-      "loss": 9.8957,
+      "loss": 9.8928,
       "step": 2500
     },
     {
       "epoch": 1.2594458438287153,
-      "grad_norm": 1.427160382270813,
+      "grad_norm": 1.5051811933517456,
       "learning_rate": 3e-05,
-      "loss": 9.7907,
+      "loss": 9.7899,
       "step": 3000
     },
     {
       "epoch": 1.4693534844668346,
-      "grad_norm": 1.95688796043396,
+      "grad_norm": 1.536996603012085,
       "learning_rate": 3.5e-05,
-      "loss": 9.7081,
+      "loss": 9.7069,
       "step": 3500
     },
     {
       "epoch": 1.6792611251049538,
-      "grad_norm": 1.8251227140426636,
+      "grad_norm": 1.406592607498169,
       "learning_rate": 4e-05,
-      "loss": 9.6313,
+      "loss": 9.6314,
       "step": 4000
     },
     {
       "epoch": 1.8891687657430731,
-      "grad_norm": 1.562404990196228,
+      "grad_norm": 1.5203056335449219,
       "learning_rate": 4.5e-05,
-      "loss": 9.5521,
+      "loss": 9.5531,
       "step": 4500
     },
     {
       "epoch": 2.0990764063811924,
-      "grad_norm": 1.5219001770019531,
+      "grad_norm": 1.6348485946655273,
       "learning_rate": 5e-05,
-      "loss": 9.4857,
+      "loss": 9.4856,
       "step": 5000
     },
     {
       "epoch": 2.3089840470193117,
-      "grad_norm": 1.657734990119934,
+      "grad_norm": 1.514229416847229,
       "learning_rate": 5.500000000000001e-05,
-      "loss": 9.4397,
+      "loss": 9.4401,
       "step": 5500
     },
     {
       "epoch": 2.5188916876574305,
-      "grad_norm": 1.5999040603637695,
+      "grad_norm": 1.7469381093978882,
       "learning_rate": 6e-05,
-      "loss": 9.4062,
+      "loss": 9.4057,
       "step": 6000
     },
     {
       "epoch": 2.72879932829555,
-      "grad_norm": 1.791552186012268,
+      "grad_norm": 1.8264902830123901,
       "learning_rate": 6.500000000000001e-05,
       "loss": 9.3766,
       "step": 6500
     },
     {
       "epoch": 2.938706968933669,
-      "grad_norm": 1.9645187854766846,
+      "grad_norm": 1.6915169954299927,
       "learning_rate": 7e-05,
-      "loss": 9.3495,
+      "loss": 9.3484,
       "step": 7000
     },
     {
       "epoch": 3.1486146095717884,
-      "grad_norm": 1.849722981452942,
+      "grad_norm": 1.769294261932373,
       "learning_rate": 7.500000000000001e-05,
-      "loss": 9.2999,
+      "loss": 9.3007,
       "step": 7500
     },
     {
       "epoch": 3.3585222502099077,
-      "grad_norm": 2.0951216220855713,
+      "grad_norm": 2.033007860183716,
       "learning_rate": 8e-05,
-      "loss": 9.2647,
+      "loss": 9.2661,
       "step": 8000
     },
     {
       "epoch": 3.568429890848027,
-      "grad_norm": 1.8428375720977783,
+      "grad_norm": 2.237866163253784,
       "learning_rate": 8.5e-05,
-      "loss": 9.2545,
+      "loss": 9.2551,
       "step": 8500
     },
     {
       "epoch": 3.7783375314861463,
-      "grad_norm": 1.943477988243103,
+      "grad_norm": 2.1374595165252686,
       "learning_rate": 9e-05,
-      "loss": 9.2298,
+      "loss": 9.2313,
       "step": 9000
     },
     {
       "epoch": 3.988245172124265,
-      "grad_norm": 2.1256041526794434,
+      "grad_norm": 2.173703908920288,
       "learning_rate": 9.5e-05,
-      "loss": 9.2199,
+      "loss": 9.2191,
       "step": 9500
     },
     {
       "epoch": 4.198152812762385,
-      "grad_norm": 2.4727325439453125,
+      "grad_norm": 2.779006242752075,
       "learning_rate": 0.0001,
-      "loss": 9.1042,
+      "loss": 9.1024,
       "step": 10000
     },
     {
       "epoch": 4.198152812762385,
-      "eval_accuracy": 0.15142324494281978,
-      "eval_loss": 9.296181678771973,
-      "eval_runtime": 0.7206,
-      "eval_samples_per_second": 664.709,
-      "eval_steps_per_second": 6.939,
+      "eval_accuracy": 0.15069979687804336,
+      "eval_loss": 9.243675231933594,
+      "eval_runtime": 0.9575,
+      "eval_samples_per_second": 500.245,
+      "eval_steps_per_second": 5.222,
       "step": 10000
     },
     {
       "epoch": 4.408060453400504,
-      "grad_norm": 2.5808792114257812,
+      "grad_norm": 2.316368818283081,
       "learning_rate": 9.899091826437942e-05,
-      "loss": 9.1028,
+      "loss": 9.1014,
       "step": 10500
     },
     {
       "epoch": 4.617968094038623,
-      "grad_norm": 2.69742488861084,
+      "grad_norm": 2.4037418365478516,
       "learning_rate": 9.798183652875883e-05,
-      "loss": 9.0977,
+      "loss": 9.0989,
       "step": 11000
     },
     {
       "epoch": 4.827875734676742,
-      "grad_norm": 2.6872336864471436,
+      "grad_norm": 2.6702349185943604,
       "learning_rate": 9.697275479313825e-05,
-      "loss": 9.1018,
+      "loss": 9.1023,
       "step": 11500
     },
     {
       "epoch": 5.037783375314861,
-      "grad_norm": 2.8443214893341064,
+      "grad_norm": 2.4351463317871094,
       "learning_rate": 9.596367305751767e-05,
-      "loss": 9.0638,
+      "loss": 9.0706,
       "step": 12000
     },
     {
       "epoch": 5.247691015952981,
-      "grad_norm": 2.7174601554870605,
+      "grad_norm": 2.828047513961792,
       "learning_rate": 9.495459132189707e-05,
-      "loss": 8.9614,
+      "loss": 8.9589,
       "step": 12500
     },
     {
       "epoch": 5.4575986565911,
-      "grad_norm": 2.59043288230896,
+      "grad_norm": 3.002562999725342,
       "learning_rate": 9.394550958627649e-05,
-      "loss": 8.97,
+      "loss": 8.9743,
       "step": 13000
     },
     {
       "epoch": 5.667506297229219,
-      "grad_norm": 2.4165260791778564,
+      "grad_norm": 2.3240981101989746,
       "learning_rate": 9.293642785065591e-05,
-      "loss": 8.9835,
+      "loss": 8.9848,
       "step": 13500
     },
     {
       "epoch": 5.877413937867338,
-      "grad_norm": 2.565976858139038,
+      "grad_norm": 2.7959797382354736,
       "learning_rate": 9.192734611503532e-05,
-      "loss": 8.9812,
+      "loss": 8.9831,
       "step": 14000
     },
     {
       "epoch": 6.087321578505458,
-      "grad_norm": 2.414792060852051,
+      "grad_norm": 2.686533212661743,
       "learning_rate": 9.091826437941474e-05,
-      "loss": 8.9258,
+      "loss": 8.9289,
       "step": 14500
     },
     {
       "epoch": 6.297229219143577,
-      "grad_norm": 2.7424213886260986,
+      "grad_norm": 2.45792555809021,
       "learning_rate": 8.990918264379415e-05,
-      "loss": 8.8699,
+      "loss": 8.875,
       "step": 15000
     },
     {
       "epoch": 6.507136859781696,
-      "grad_norm": 2.612027645111084,
+      "grad_norm": 2.2763113975524902,
       "learning_rate": 8.890010090817356e-05,
-      "loss": 8.8812,
+      "loss": 8.884,
       "step": 15500
     },
     {
       "epoch": 6.717044500419815,
-      "grad_norm": 2.407137632369995,
+      "grad_norm": 2.498940944671631,
       "learning_rate": 8.789101917255298e-05,
-      "loss": 8.8932,
+      "loss": 8.8944,
       "step": 16000
     },
     {
       "epoch": 6.926952141057934,
-      "grad_norm": 2.4915878772735596,
+      "grad_norm": 3.049781084060669,
       "learning_rate": 8.68819374369324e-05,
-      "loss": 8.9013,
+      "loss": 8.9012,
       "step": 16500
     },
     {
       "epoch": 7.136859781696054,
-      "grad_norm": 2.228576898574829,
+      "grad_norm": 2.3419675827026367,
       "learning_rate": 8.58728557013118e-05,
-      "loss": 8.8171,
+      "loss": 8.8163,
       "step": 17000
     },
     {
       "epoch": 7.346767422334173,
-      "grad_norm": 3.1212477684020996,
+      "grad_norm": 2.602226734161377,
       "learning_rate": 8.486377396569122e-05,
-      "loss": 8.7841,
+      "loss": 8.7904,
       "step": 17500
     },
     {
       "epoch": 7.5566750629722925,
-      "grad_norm": 2.709784984588623,
+      "grad_norm": 2.762700319290161,
       "learning_rate": 8.385469223007064e-05,
-      "loss": 8.8064,
+      "loss": 8.815,
       "step": 18000
     },
     {
       "epoch": 7.766582703610411,
-      "grad_norm": 2.305772542953491,
+      "grad_norm": 2.8332741260528564,
       "learning_rate": 8.284561049445006e-05,
-      "loss": 8.823,
+      "loss": 8.8283,
       "step": 18500
     },
     {
       "epoch": 7.97649034424853,
-      "grad_norm": 3.071176767349243,
+      "grad_norm": 2.8833227157592773,
       "learning_rate": 8.183652875882947e-05,
-      "loss": 8.8276,
+      "loss": 8.8307,
       "step": 19000
     },
     {
       "epoch": 8.18639798488665,
-      "grad_norm": 2.496626377105713,
+      "grad_norm": 2.686835527420044,
       "learning_rate": 8.082744702320889e-05,
-      "loss": 8.7147,
+      "loss": 8.7211,
       "step": 19500
     },
     {
       "epoch": 8.39630562552477,
-      "grad_norm": 2.8181748390197754,
+      "grad_norm": 2.750887155532837,
       "learning_rate": 7.98183652875883e-05,
-      "loss": 8.718,
+      "loss": 8.7308,
       "step": 20000
     },
     {
       "epoch": 8.39630562552477,
-      "eval_accuracy": 0.15250147310530598,
-      "eval_loss": 9.342524528503418,
-      "eval_runtime": 0.7037,
-      "eval_samples_per_second": 680.718,
-      "eval_steps_per_second": 7.106,
+      "eval_accuracy": 0.15230505906450797,
+      "eval_loss": 9.404132843017578,
+      "eval_runtime": 0.687,
+      "eval_samples_per_second": 697.285,
+      "eval_steps_per_second": 7.279,
       "step": 20000
     },
     {
       "epoch": 8.606213266162888,
-      "grad_norm": 3.054976463317871,
+      "grad_norm": 2.781379461288452,
       "learning_rate": 7.880928355196771e-05,
-      "loss": 8.7277,
+      "loss": 8.7374,
       "step": 20500
     },
     {
       "epoch": 8.816120906801007,
-      "grad_norm": 2.586456060409546,
+      "grad_norm": 3.0466349124908447,
       "learning_rate": 7.780020181634713e-05,
-      "loss": 8.7504,
+      "loss": 8.7558,
       "step": 21000
     },
     {
       "epoch": 9.026028547439127,
-      "grad_norm": 3.341202974319458,
+      "grad_norm": 3.205801248550415,
       "learning_rate": 7.679112008072655e-05,
-      "loss": 8.7284,
+      "loss": 8.7396,
       "step": 21500
     },
     {
       "epoch": 9.235936188077247,
-      "grad_norm": 3.1252424716949463,
+      "grad_norm": 3.279179334640503,
       "learning_rate": 7.578203834510595e-05,
-      "loss": 8.602,
+      "loss": 8.6237,
       "step": 22000
     },
     {
       "epoch": 9.445843828715365,
-      "grad_norm": 3.131685256958008,
+      "grad_norm": 2.839015007019043,
       "learning_rate": 7.477295660948537e-05,
-      "loss": 8.6209,
+      "loss": 8.635,
       "step": 22500
     },
     {
       "epoch": 9.655751469353484,
-      "grad_norm": 2.8262670040130615,
+      "grad_norm": 2.7180089950561523,
       "learning_rate": 7.376387487386479e-05,
-      "loss": 8.6446,
+      "loss": 8.6638,
       "step": 23000
     },
     {
       "epoch": 9.865659109991604,
-      "grad_norm": 2.8560707569122314,
+      "grad_norm": 3.3253207206726074,
       "learning_rate": 7.27547931382442e-05,
-      "loss": 8.6485,
+      "loss": 8.6629,
       "step": 23500
     },
     {
       "epoch": 10.075566750629722,
-      "grad_norm": 3.7638072967529297,
+      "grad_norm": 3.483107089996338,
       "learning_rate": 7.174571140262362e-05,
-      "loss": 8.5796,
+      "loss": 8.6009,
       "step": 24000
     },
     {
       "epoch": 10.285474391267842,
-      "grad_norm": 3.2438437938690186,
+      "grad_norm": 3.4180562496185303,
       "learning_rate": 7.073662966700304e-05,
-      "loss": 8.478,
+      "loss": 8.5001,
       "step": 24500
     },
     {
       "epoch": 10.495382031905962,
-      "grad_norm": 3.6268389225006104,
+      "grad_norm": 3.278217077255249,
       "learning_rate": 6.972754793138244e-05,
-      "loss": 8.4837,
+      "loss": 8.5159,
       "step": 25000
     },
     {
       "epoch": 10.705289672544081,
-      "grad_norm": 4.221205711364746,
+      "grad_norm": 3.70385479927063,
       "learning_rate": 6.871846619576186e-05,
-      "loss": 8.5092,
+      "loss": 8.5354,
       "step": 25500
     },
     {
       "epoch": 10.9151973131822,
-      "grad_norm": 3.950782299041748,
+      "grad_norm": 3.487703561782837,
       "learning_rate": 6.770938446014128e-05,
-      "loss": 8.5295,
+      "loss": 8.56,
       "step": 26000
     },
     {
       "epoch": 11.125104953820319,
-      "grad_norm": 3.7950942516326904,
+      "grad_norm": 4.552535057067871,
       "learning_rate": 6.670030272452068e-05,
-      "loss": 8.3933,
+      "loss": 8.4245,
       "step": 26500
     },
     {
       "epoch": 11.335012594458439,
-      "grad_norm": 4.182861328125,
+      "grad_norm": 4.1042022705078125,
       "learning_rate": 6.56912209889001e-05,
-      "loss": 8.3174,
+      "loss": 8.3508,
       "step": 27000
     },
     {
       "epoch": 11.544920235096557,
-      "grad_norm": 5.146470069885254,
+      "grad_norm": 4.3246941566467285,
       "learning_rate": 6.468213925327952e-05,
-      "loss": 8.3402,
+      "loss": 8.3758,
       "step": 27500
     },
     {
       "epoch": 11.754827875734676,
-      "grad_norm": 4.480419158935547,
+      "grad_norm": 3.9653408527374268,
       "learning_rate": 6.367305751765893e-05,
-      "loss": 8.3621,
+      "loss": 8.3897,
       "step": 28000
     },
     {
       "epoch": 11.964735516372796,
-      "grad_norm": 4.593987464904785,
+      "grad_norm": 4.222130298614502,
       "learning_rate": 6.266397578203835e-05,
-      "loss": 8.3828,
+      "loss": 8.4155,
       "step": 28500
     },
     {
       "epoch": 12.174643157010916,
-      "grad_norm": 4.7530317306518555,
+      "grad_norm": 4.950220108032227,
       "learning_rate": 6.165489404641777e-05,
-      "loss": 8.1637,
+      "loss": 8.2098,
       "step": 29000
     },
     {
       "epoch": 12.384550797649034,
-      "grad_norm": 4.161499977111816,
+      "grad_norm": 4.2607550621032715,
       "learning_rate": 6.064581231079718e-05,
-      "loss": 8.1471,
+      "loss": 8.175,
       "step": 29500
     },
     {
       "epoch": 12.594458438287154,
-      "grad_norm": 5.079950332641602,
+      "grad_norm": 5.873214244842529,
       "learning_rate": 5.963673057517659e-05,
-      "loss": 8.1812,
+      "loss": 8.2147,
       "step": 30000
     },
     {
       "epoch": 12.594458438287154,
-      "eval_accuracy": 0.15415395808265944,
-      "eval_loss": 9.84601879119873,
-      "eval_runtime": 0.7055,
-      "eval_samples_per_second": 678.919,
-      "eval_steps_per_second": 7.087,
+      "eval_accuracy": 0.1539860648627473,
+      "eval_loss": 9.970804214477539,
+      "eval_runtime": 0.7285,
+      "eval_samples_per_second": 657.471,
+      "eval_steps_per_second": 6.863,
       "step": 30000
     },
     {
       "epoch": 12.804366078925273,
-      "grad_norm": 4.768843650817871,
+      "grad_norm": 4.897546291351318,
       "learning_rate": 5.8627648839556e-05,
-      "loss": 8.1889,
+      "loss": 8.2332,
       "step": 30500
     },
     {
       "epoch": 13.014273719563391,
-      "grad_norm": 5.478571891784668,
+      "grad_norm": 4.54090690612793,
       "learning_rate": 5.7618567103935416e-05,
-      "loss": 8.1997,
+      "loss": 8.2306,
       "step": 31000
     },
     {
       "epoch": 13.224181360201511,
-      "grad_norm": 5.266770839691162,
+      "grad_norm": 5.108377456665039,
       "learning_rate": 5.660948536831484e-05,
-      "loss": 7.9457,
+      "loss": 7.9821,
       "step": 31500
     },
     {
       "epoch": 13.43408900083963,
-      "grad_norm": 5.525669574737549,
+      "grad_norm": 5.725313186645508,
       "learning_rate": 5.5600403632694254e-05,
-      "loss": 7.9753,
+      "loss": 8.0109,
       "step": 32000
     },
     {
       "epoch": 13.64399664147775,
-      "grad_norm": 5.723587512969971,
+      "grad_norm": 5.594811916351318,
       "learning_rate": 5.4591321897073666e-05,
-      "loss": 8.0119,
+      "loss": 8.0542,
       "step": 32500
     },
     {
       "epoch": 13.853904282115868,
-      "grad_norm": 5.669743061065674,
+      "grad_norm": 5.017056941986084,
       "learning_rate": 5.358224016145308e-05,
-      "loss": 8.0331,
+      "loss": 8.0684,
       "step": 33000
     },
     {
       "epoch": 14.063811922753988,
-      "grad_norm": 5.793339729309082,
+      "grad_norm": 5.904685020446777,
       "learning_rate": 5.257315842583249e-05,
-      "loss": 7.9612,
+      "loss": 7.9952,
       "step": 33500
     },
     {
       "epoch": 14.273719563392108,
-      "grad_norm": 6.351025104522705,
+      "grad_norm": 7.006627559661865,
       "learning_rate": 5.15640766902119e-05,
-      "loss": 7.7883,
+      "loss": 7.8221,
       "step": 34000
     },
     {
       "epoch": 14.483627204030226,
-      "grad_norm": 6.8214192390441895,
+      "grad_norm": 6.2029266357421875,
       "learning_rate": 5.055499495459133e-05,
-      "loss": 7.8215,
+      "loss": 7.858,
       "step": 34500
     },
     {
       "epoch": 14.693534844668346,
-      "grad_norm": 6.922965049743652,
+      "grad_norm": 5.595657825469971,
       "learning_rate": 4.954591321897074e-05,
-      "loss": 7.8464,
+      "loss": 7.8832,
       "step": 35000
     },
     {
       "epoch": 14.903442485306465,
-      "grad_norm": 6.26981258392334,
+      "grad_norm": 6.32289457321167,
       "learning_rate": 4.853683148335015e-05,
-      "loss": 7.8625,
+      "loss": 7.9024,
       "step": 35500
     },
     {
       "epoch": 15.113350125944585,
-      "grad_norm": 6.1497015953063965,
+      "grad_norm": 5.871936798095703,
       "learning_rate": 4.7527749747729566e-05,
-      "loss": 7.7257,
+      "loss": 7.7625,
       "step": 36000
     },
     {
       "epoch": 15.323257766582703,
-      "grad_norm": 7.00075101852417,
+      "grad_norm": 6.5099992752075195,
       "learning_rate": 4.6518668012108985e-05,
-      "loss": 7.6424,
+      "loss": 7.6751,
       "step": 36500
     },
     {
       "epoch": 15.533165407220823,
-      "grad_norm": 7.382388114929199,
+      "grad_norm": 6.116209506988525,
       "learning_rate": 4.55095862764884e-05,
-      "loss": 7.6672,
+      "loss": 7.7103,
       "step": 37000
     },
     {
       "epoch": 15.743073047858942,
-      "grad_norm": 6.985136032104492,
+      "grad_norm": 6.561167240142822,
       "learning_rate": 4.450050454086781e-05,
-      "loss": 7.6944,
+      "loss": 7.7243,
       "step": 37500
     },
     {
       "epoch": 15.95298068849706,
-      "grad_norm": 7.3510589599609375,
+      "grad_norm": 6.933249473571777,
       "learning_rate": 4.349142280524723e-05,
-      "loss": 7.7155,
+      "loss": 7.7405,
       "step": 38000
     },
     {
       "epoch": 16.162888329135182,
-      "grad_norm": 6.908013820648193,
+      "grad_norm": 7.447343826293945,
       "learning_rate": 4.248234106962664e-05,
-      "loss": 7.5252,
+      "loss": 7.5561,
       "step": 38500
     },
     {
       "epoch": 16.3727959697733,
-      "grad_norm": 7.940768241882324,
+      "grad_norm": 7.856605529785156,
       "learning_rate": 4.147325933400605e-05,
-      "loss": 7.4949,
+      "loss": 7.5321,
       "step": 39000
     },
     {
       "epoch": 16.582703610411418,
-      "grad_norm": 7.501976490020752,
+      "grad_norm": 6.8672871589660645,
       "learning_rate": 4.046417759838547e-05,
-      "loss": 7.5189,
+      "loss": 7.5549,
       "step": 39500
     },
     {
       "epoch": 16.79261125104954,
-      "grad_norm": 7.551584720611572,
+      "grad_norm": 7.581363201141357,
       "learning_rate": 3.9455095862764884e-05,
-      "loss": 7.5418,
+      "loss": 7.5753,
       "step": 40000
     },
     {
       "epoch": 16.79261125104954,
-      "eval_accuracy": 0.15159246383312774,
-      "eval_loss": 10.532922744750977,
-      "eval_runtime": 0.7293,
-      "eval_samples_per_second": 656.822,
-      "eval_steps_per_second": 6.856,
+      "eval_accuracy": 0.15097566446114163,
+      "eval_loss": 10.770827293395996,
+      "eval_runtime": 1.8438,
+      "eval_samples_per_second": 259.784,
+      "eval_steps_per_second": 2.712,
       "step": 40000
     },
     {
       "epoch": 17.002518891687657,
-      "grad_norm": 7.355690002441406,
+      "grad_norm": 7.250075817108154,
       "learning_rate": 3.84460141271443e-05,
-      "loss": 7.5643,
+      "loss": 7.6011,
       "step": 40500
     },
     {
       "epoch": 17.212426532325775,
-      "grad_norm": 7.690167427062988,
+      "grad_norm": 6.583193778991699,
       "learning_rate": 3.7436932391523715e-05,
-      "loss": 7.3301,
+      "loss": 7.3702,
       "step": 41000
     },
     {
       "epoch": 17.422334172963897,
-      "grad_norm": 7.068574905395508,
+      "grad_norm": 7.613584995269775,
       "learning_rate": 3.642785065590313e-05,
-      "loss": 7.3695,
+      "loss": 7.3946,
       "step": 41500
     },
     {
       "epoch": 17.632241813602015,
-      "grad_norm": 9.541287422180176,
+      "grad_norm": 7.3677802085876465,
       "learning_rate": 3.541876892028255e-05,
-      "loss": 7.3864,
+      "loss": 7.4185,
       "step": 42000
     },
     {
       "epoch": 17.842149454240133,
-      "grad_norm": 7.779609680175781,
+      "grad_norm": 7.598528861999512,
       "learning_rate": 3.440968718466196e-05,
-      "loss": 7.4135,
+      "loss": 7.4436,
       "step": 42500
     },
     {
       "epoch": 18.052057094878254,
-      "grad_norm": 7.6669158935546875,
+      "grad_norm": 7.344916820526123,
       "learning_rate": 3.340060544904137e-05,
-      "loss": 7.3612,
+      "loss": 7.4036,
       "step": 43000
     },
     {
       "epoch": 18.261964735516372,
-      "grad_norm": 6.732392311096191,
+      "grad_norm": 6.601145267486572,
       "learning_rate": 3.239152371342079e-05,
-      "loss": 7.2212,
+      "loss": 7.2479,
       "step": 43500
     },
     {
       "epoch": 18.471872376154494,
-      "grad_norm": 7.032383918762207,
+      "grad_norm": 8.270407676696777,
       "learning_rate": 3.13824419778002e-05,
-      "loss": 7.2441,
+      "loss": 7.2709,
       "step": 44000
     },
     {
       "epoch": 18.68178001679261,
-      "grad_norm": 7.678667068481445,
+      "grad_norm": 8.377107620239258,
       "learning_rate": 3.0373360242179615e-05,
-      "loss": 7.266,
+      "loss": 7.3051,
       "step": 44500
     },
     {
       "epoch": 18.89168765743073,
-      "grad_norm": 9.278508186340332,
+      "grad_norm": 8.57275104522705,
       "learning_rate": 2.9364278506559034e-05,
-      "loss": 7.2953,
+      "loss": 7.3261,
       "step": 45000
     },
     {
       "epoch": 19.10159529806885,
-      "grad_norm": 8.756003379821777,
+      "grad_norm": 8.127633094787598,
       "learning_rate": 2.8355196770938446e-05,
-      "loss": 7.1966,
+      "loss": 7.2216,
       "step": 45500
     },
     {
       "epoch": 19.31150293870697,
-      "grad_norm": 8.082226753234863,
+      "grad_norm": 8.819077491760254,
       "learning_rate": 2.734611503531786e-05,
-      "loss": 7.1147,
+      "loss": 7.1496,
       "step": 46000
     },
     {
       "epoch": 19.521410579345087,
-      "grad_norm": 8.553418159484863,
+      "grad_norm": 8.62332534790039,
       "learning_rate": 2.6337033299697278e-05,
-      "loss": 7.1318,
+      "loss": 7.1705,
       "step": 46500
     },
     {
       "epoch": 19.73131821998321,
-      "grad_norm": 7.318296909332275,
+      "grad_norm": 8.724237442016602,
       "learning_rate": 2.532795156407669e-05,
-      "loss": 7.145,
+      "loss": 7.1817,
       "step": 47000
     },
     {
       "epoch": 19.941225860621326,
-      "grad_norm": 9.50650405883789,
+      "grad_norm": 8.85011100769043,
       "learning_rate": 2.4318869828456106e-05,
-      "loss": 7.157,
+      "loss": 7.1885,
       "step": 47500
     },
     {
       "epoch": 20.151133501259444,
-      "grad_norm": 8.413248062133789,
+      "grad_norm": 8.974445343017578,
       "learning_rate": 2.330978809283552e-05,
-      "loss": 7.0529,
+      "loss": 7.0751,
       "step": 48000
     },
     {
       "epoch": 20.361041141897566,
-      "grad_norm": 8.851900100708008,
+      "grad_norm": 11.303741455078125,
       "learning_rate": 2.2300706357214934e-05,
-      "loss": 7.0145,
+      "loss": 7.0533,
       "step": 48500
     },
     {
       "epoch": 20.570948782535684,
-      "grad_norm": 7.856457710266113,
+      "grad_norm": 8.581236839294434,
       "learning_rate": 2.129162462159435e-05,
-      "loss": 7.031,
+      "loss": 7.0637,
       "step": 49000
     },
     {
       "epoch": 20.7808564231738,
-      "grad_norm": 8.467928886413574,
+      "grad_norm": 9.926290512084961,
       "learning_rate": 2.0282542885973765e-05,
-      "loss": 7.0423,
+      "loss": 7.0734,
       "step": 49500
     },
     {
       "epoch": 20.990764063811923,
-      "grad_norm": 7.82729434967041,
+      "grad_norm": 8.554856300354004,
       "learning_rate": 1.927346115035318e-05,
-      "loss": 7.0563,
+      "loss": 7.0936,
       "step": 50000
     },
     {
       "epoch": 20.990764063811923,
-      "eval_accuracy": 0.15187708526088542,
-      "eval_loss": 11.335107803344727,
-      "eval_runtime": 0.6908,
-      "eval_samples_per_second": 693.418,
-      "eval_steps_per_second": 7.238,
+      "eval_accuracy": 0.1511200829898786,
+      "eval_loss": 11.395831108093262,
+      "eval_runtime": 0.6862,
+      "eval_samples_per_second": 698.003,
+      "eval_steps_per_second": 7.286,
       "step": 50000
     },
     {
       "epoch": 21.20067170445004,
-      "grad_norm": 9.393904685974121,
+      "grad_norm": 8.856547355651855,
       "learning_rate": 1.8264379414732593e-05,
-      "loss": 6.9212,
+      "loss": 6.9504,
       "step": 50500
     },
     {
       "epoch": 21.410579345088163,
-      "grad_norm": 8.09192943572998,
+      "grad_norm": 8.438743591308594,
       "learning_rate": 1.725529767911201e-05,
-      "loss": 6.9342,
+      "loss": 6.9607,
       "step": 51000
     },
     {
       "epoch": 21.62048698572628,
-      "grad_norm": 8.69682788848877,
+      "grad_norm": 9.606910705566406,
       "learning_rate": 1.6246215943491424e-05,
-      "loss": 6.9565,
+      "loss": 6.9817,
       "step": 51500
     },
     {
       "epoch": 21.8303946263644,
-      "grad_norm": 10.353145599365234,
+      "grad_norm": 9.459112167358398,
       "learning_rate": 1.5237134207870837e-05,
-      "loss": 6.963,
+      "loss": 6.9872,
       "step": 52000
     },
     {
       "epoch": 22.04030226700252,
-      "grad_norm": 9.867997169494629,
+      "grad_norm": 8.481375694274902,
       "learning_rate": 1.4228052472250252e-05,
-      "loss": 6.9292,
+      "loss": 6.9726,
       "step": 52500
     },
     {
       "epoch": 22.250209907640638,
-      "grad_norm": 8.513895034790039,
+      "grad_norm": 8.331954002380371,
       "learning_rate": 1.3218970736629668e-05,
-      "loss": 6.8571,
+      "loss": 6.8827,
       "step": 53000
     },
     {
       "epoch": 22.460117548278756,
-      "grad_norm": 7.132133483886719,
+      "grad_norm": 8.738556861877441,
       "learning_rate": 1.2209889001009082e-05,
-      "loss": 6.8649,
+      "loss": 6.8867,
       "step": 53500
     },
     {
       "epoch": 22.670025188916878,
-      "grad_norm": 9.139779090881348,
+      "grad_norm": 8.790620803833008,
       "learning_rate": 1.1200807265388496e-05,
-      "loss": 6.8722,
+      "loss": 6.8981,
       "step": 54000
     },
     {
       "epoch": 22.879932829554996,
-      "grad_norm": 7.9002556800842285,
+      "grad_norm": 9.088656425476074,
       "learning_rate": 1.0191725529767912e-05,
-      "loss": 6.8763,
+      "loss": 6.8966,
       "step": 54500
     },
     {
       "epoch": 23.089840470193113,
-      "grad_norm": 8.529388427734375,
+      "grad_norm": 7.995791435241699,
       "learning_rate": 9.182643794147326e-06,
-      "loss": 6.8346,
+      "loss": 6.8673,
       "step": 55000
     },
     {
       "epoch": 23.299748110831235,
-      "grad_norm": 7.972701549530029,
+      "grad_norm": 9.11633014678955,
       "learning_rate": 8.173562058526741e-06,
-      "loss": 6.8021,
+      "loss": 6.8176,
       "step": 55500
     },
     {
       "epoch": 23.509655751469353,
-      "grad_norm": 10.245627403259277,
+      "grad_norm": 9.16334342956543,
       "learning_rate": 7.164480322906156e-06,
-      "loss": 6.8104,
+      "loss": 6.8408,
       "step": 56000
     },
     {
       "epoch": 23.719563392107474,
-      "grad_norm": 8.268402099609375,
+      "grad_norm": 9.88434886932373,
       "learning_rate": 6.155398587285571e-06,
-      "loss": 6.802,
+      "loss": 6.8215,
       "step": 56500
     },
     {
       "epoch": 23.929471032745592,
-      "grad_norm": 9.488409042358398,
+      "grad_norm": 9.812824249267578,
       "learning_rate": 5.146316851664986e-06,
-      "loss": 6.812,
+      "loss": 6.844,
       "step": 57000
     },
     {
       "epoch": 24.13937867338371,
-      "grad_norm": 8.589398384094238,
+      "grad_norm": 9.951552391052246,
       "learning_rate": 4.1372351160444e-06,
-      "loss": 6.7823,
+      "loss": 6.7985,
       "step": 57500
     },
     {
       "epoch": 24.349286314021832,
-      "grad_norm": 8.700592994689941,
+      "grad_norm": 8.857964515686035,
       "learning_rate": 3.1281533804238145e-06,
-      "loss": 6.7744,
+      "loss": 6.8,
       "step": 58000
     },
     {
       "epoch": 24.55919395465995,
-      "grad_norm": 11.097859382629395,
+      "grad_norm": 10.355735778808594,
       "learning_rate": 2.1190716448032293e-06,
-      "loss": 6.7689,
+      "loss": 6.8036,
       "step": 58500
     },
     {
       "epoch": 24.769101595298068,
-      "grad_norm": 8.395718574523926,
+      "grad_norm": 9.625215530395508,
       "learning_rate": 1.109989909182644e-06,
-      "loss": 6.7658,
+      "loss": 6.7907,
       "step": 59000
     },
     {
       "epoch": 24.97900923593619,
-      "grad_norm": 10.317686080932617,
+      "grad_norm": 9.651956558227539,
       "learning_rate": 1.0090817356205853e-07,
-      "loss": 6.7607,
+      "loss": 6.7857,
       "step": 59500
     },
     {
       "epoch": 25.0,
       "step": 59550,
       "total_flos": 3.051148107142656e+17,
-      "train_loss": 8.18093173522693,
-      "train_runtime": 11970.8684,
-      "train_samples_per_second": 477.49,
-      "train_steps_per_second": 4.975
+      "train_loss": 8.201283628782438,
+      "train_runtime": 12621.2502,
+      "train_samples_per_second": 452.885,
+      "train_steps_per_second": 4.718
     }
   ],
   "logging_steps": 500,