End of training

README.md

@@ -1,11 +1,23 @@
 ---
 tags:
 - generated_from_trainer
+datasets:
+- Hartunka/processed_wikitext-103-raw-v1-km-100_v2
 metrics:
 - accuracy
 model-index:
 - name: tiny_bert_km_100_v2
-  results: []
+  results:
+  - task:
+      name: Masked Language Modeling
+      type: fill-mask
+    dataset:
+      name: Hartunka/processed_wikitext-103-raw-v1-km-100_v2
+      type: Hartunka/processed_wikitext-103-raw-v1-km-100_v2
+    metrics:
+    - name: Accuracy
+      type: accuracy
+      value: 0.15307679267672486
 ---
 
 <!-- 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_100_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-100_v2 dataset.
 It achieves the following results on the evaluation set:
-- Loss: 7.1608
-- Accuracy: 0.1509
+- Loss: 6.9988
+- Accuracy: 0.1531
 
 ## Model description
 

all_results.json

@@ -1,16 +1,16 @@
 {
     "epoch": 25.0,
-    "eval_accuracy": 0.15333107306323104,
-    "eval_loss": 7.090662479400635,
-    "eval_runtime": 0.7408,
+    "eval_accuracy": 0.15307679267672486,
+    "eval_loss": 6.998800277709961,
+    "eval_runtime": 0.7225,
     "eval_samples": 479,
-    "eval_samples_per_second": 646.57,
-    "eval_steps_per_second": 6.749,
-    "perplexity": 1200.7029781767033,
+    "eval_samples_per_second": 662.945,
+    "eval_steps_per_second": 6.92,
+    "perplexity": 1095.3182920792408,
     "total_flos": 3.058354515764736e+17,
-    "train_loss": 6.935516868866761,
-    "train_runtime": 11960.537,
+    "train_loss": 6.8441891810035225,
+    "train_runtime": 11476.7871,
     "train_samples": 228639,
-    "train_samples_per_second": 477.903,
-    "train_steps_per_second": 4.979
+    "train_samples_per_second": 498.047,
+    "train_steps_per_second": 5.189
 }

eval_results.json

@@ -1,10 +1,10 @@
 {
     "epoch": 25.0,
-    "eval_accuracy": 0.15333107306323104,
-    "eval_loss": 7.090662479400635,
-    "eval_runtime": 0.7408,
+    "eval_accuracy": 0.15307679267672486,
+    "eval_loss": 6.998800277709961,
+    "eval_runtime": 0.7225,
     "eval_samples": 479,
-    "eval_samples_per_second": 646.57,
-    "eval_steps_per_second": 6.749,
-    "perplexity": 1200.7029781767033
+    "eval_samples_per_second": 662.945,
+    "eval_steps_per_second": 6.92,
+    "perplexity": 1095.3182920792408
 }

train_results.json

@@ -1,9 +1,9 @@
 {
     "epoch": 25.0,
     "total_flos": 3.058354515764736e+17,
-    "train_loss": 6.935516868866761,
-    "train_runtime": 11960.537,
+    "train_loss": 6.8441891810035225,
+    "train_runtime": 11476.7871,
     "train_samples": 228639,
-    "train_samples_per_second": 477.903,
-    "train_steps_per_second": 4.979
+    "train_samples_per_second": 498.047,
+    "train_steps_per_second": 5.189
 }

trainer_state.json

@@ -1,5 +1,5 @@
 {
-  "best_metric": 7.090819835662842,
+  "best_metric": 6.99865198135376,
   "best_model_checkpoint": "tiny_bert_km_100_v2/checkpoint-20000",
   "epoch": 25.0,
   "eval_steps": 10000,
@@ -10,890 +10,890 @@
   "log_history": [
     {
       "epoch": 0.20990764063811923,
-      "grad_norm": 2.571293830871582,
+      "grad_norm": 2.613888740539551,
       "learning_rate": 5e-06,
-      "loss": 14.3662,
+      "loss": 14.3723,
       "step": 500
     },
     {
       "epoch": 0.41981528127623846,
-      "grad_norm": 1.9099379777908325,
+      "grad_norm": 1.9190033674240112,
       "learning_rate": 1e-05,
-      "loss": 12.7832,
+      "loss": 12.7977,
       "step": 1000
     },
     {
       "epoch": 0.6297229219143576,
-      "grad_norm": 2.101893186569214,
+      "grad_norm": 2.1179254055023193,
       "learning_rate": 1.5e-05,
-      "loss": 11.7445,
+      "loss": 11.6993,
       "step": 1500
     },
     {
       "epoch": 0.8396305625524769,
-      "grad_norm": 2.5784850120544434,
+      "grad_norm": 3.613637685775757,
       "learning_rate": 2e-05,
-      "loss": 11.0967,
+      "loss": 10.987,
       "step": 2000
     },
     {
       "epoch": 1.0495382031905962,
-      "grad_norm": 3.0886240005493164,
+      "grad_norm": 4.624087810516357,
       "learning_rate": 2.5e-05,
-      "loss": 10.0396,
+      "loss": 10.1839,
       "step": 2500
     },
     {
       "epoch": 1.2594458438287153,
-      "grad_norm": 3.1394476890563965,
+      "grad_norm": 5.024355888366699,
       "learning_rate": 3e-05,
-      "loss": 9.1878,
+      "loss": 9.5,
       "step": 3000
     },
     {
       "epoch": 1.4693534844668346,
-      "grad_norm": 6.415482997894287,
+      "grad_norm": 4.341619968414307,
       "learning_rate": 3.5e-05,
-      "loss": 8.6737,
+      "loss": 9.0842,
       "step": 3500
     },
     {
       "epoch": 1.6792611251049538,
-      "grad_norm": 5.410119533538818,
+      "grad_norm": 6.015387058258057,
       "learning_rate": 4e-05,
-      "loss": 8.3257,
+      "loss": 8.7792,
       "step": 4000
     },
     {
       "epoch": 1.8891687657430731,
-      "grad_norm": 6.013356685638428,
+      "grad_norm": 3.739591360092163,
       "learning_rate": 4.5e-05,
-      "loss": 8.0861,
+      "loss": 8.5295,
       "step": 4500
     },
     {
       "epoch": 2.0990764063811924,
-      "grad_norm": 4.655304431915283,
+      "grad_norm": 5.641452312469482,
       "learning_rate": 5e-05,
-      "loss": 7.8725,
+      "loss": 8.2694,
       "step": 5000
     },
     {
       "epoch": 2.3089840470193117,
-      "grad_norm": 6.407634258270264,
+      "grad_norm": 3.95416259765625,
       "learning_rate": 5.500000000000001e-05,
-      "loss": 7.741,
+      "loss": 8.054,
       "step": 5500
     },
     {
       "epoch": 2.5188916876574305,
-      "grad_norm": 8.330334663391113,
+      "grad_norm": 4.106296062469482,
       "learning_rate": 6e-05,
-      "loss": 7.6701,
+      "loss": 7.8926,
       "step": 6000
     },
     {
       "epoch": 2.72879932829555,
-      "grad_norm": 6.971424102783203,
+      "grad_norm": 4.829506874084473,
       "learning_rate": 6.500000000000001e-05,
-      "loss": 7.6108,
+      "loss": 7.7508,
       "step": 6500
     },
     {
       "epoch": 2.938706968933669,
-      "grad_norm": 4.134661674499512,
+      "grad_norm": 5.512585163116455,
       "learning_rate": 7e-05,
-      "loss": 7.5738,
+      "loss": 7.63,
       "step": 7000
     },
     {
       "epoch": 3.1486146095717884,
-      "grad_norm": 5.728518009185791,
+      "grad_norm": 4.711154937744141,
       "learning_rate": 7.500000000000001e-05,
-      "loss": 7.4708,
+      "loss": 7.4475,
       "step": 7500
     },
     {
       "epoch": 3.3585222502099077,
-      "grad_norm": 4.252316951751709,
+      "grad_norm": 4.342106819152832,
       "learning_rate": 8e-05,
-      "loss": 7.4294,
+      "loss": 7.3893,
       "step": 8000
     },
     {
       "epoch": 3.568429890848027,
-      "grad_norm": 6.315696716308594,
+      "grad_norm": 4.720828056335449,
       "learning_rate": 8.5e-05,
-      "loss": 7.4369,
+      "loss": 7.362,
       "step": 8500
     },
     {
       "epoch": 3.7783375314861463,
-      "grad_norm": 7.669189453125,
+      "grad_norm": 4.400397777557373,
       "learning_rate": 9e-05,
-      "loss": 7.4119,
+      "loss": 7.3261,
       "step": 9000
     },
     {
       "epoch": 3.988245172124265,
-      "grad_norm": 7.449468612670898,
+      "grad_norm": 3.921447515487671,
       "learning_rate": 9.5e-05,
-      "loss": 7.3898,
+      "loss": 7.3036,
       "step": 9500
     },
     {
       "epoch": 4.198152812762385,
-      "grad_norm": 3.829033374786377,
+      "grad_norm": 4.298178195953369,
       "learning_rate": 0.0001,
-      "loss": 7.2843,
+      "loss": 7.1346,
       "step": 10000
     },
     {
       "epoch": 4.198152812762385,
-      "eval_accuracy": 0.14702690670302457,
-      "eval_loss": 7.433688163757324,
-      "eval_runtime": 0.7141,
-      "eval_samples_per_second": 670.762,
-      "eval_steps_per_second": 7.002,
+      "eval_accuracy": 0.14769470491666434,
+      "eval_loss": 7.205446243286133,
+      "eval_runtime": 0.6835,
+      "eval_samples_per_second": 700.776,
+      "eval_steps_per_second": 7.315,
       "step": 10000
     },
     {
       "epoch": 4.408060453400504,
-      "grad_norm": 5.837904930114746,
+      "grad_norm": 4.446288585662842,
       "learning_rate": 9.899091826437942e-05,
-      "loss": 7.2872,
+      "loss": 7.1464,
       "step": 10500
     },
     {
       "epoch": 4.617968094038623,
-      "grad_norm": 8.196144104003906,
+      "grad_norm": 4.7515645027160645,
       "learning_rate": 9.798183652875883e-05,
-      "loss": 7.2746,
+      "loss": 7.1424,
       "step": 11000
     },
     {
       "epoch": 4.827875734676742,
-      "grad_norm": 5.862619400024414,
+      "grad_norm": 5.114405632019043,
       "learning_rate": 9.697275479313825e-05,
-      "loss": 7.2717,
+      "loss": 7.1318,
       "step": 11500
     },
     {
       "epoch": 5.037783375314861,
-      "grad_norm": 6.175967216491699,
+      "grad_norm": 4.081265449523926,
       "learning_rate": 9.596367305751767e-05,
-      "loss": 7.2288,
+      "loss": 7.0895,
       "step": 12000
     },
     {
       "epoch": 5.247691015952981,
-      "grad_norm": 4.868844032287598,
+      "grad_norm": 3.7170498371124268,
       "learning_rate": 9.495459132189707e-05,
-      "loss": 7.1307,
+      "loss": 6.9629,
       "step": 12500
     },
     {
       "epoch": 5.4575986565911,
-      "grad_norm": 5.5049519538879395,
+      "grad_norm": 4.041747570037842,
       "learning_rate": 9.394550958627649e-05,
-      "loss": 7.1425,
+      "loss": 6.9708,
       "step": 13000
     },
     {
       "epoch": 5.667506297229219,
-      "grad_norm": 4.215301036834717,
+      "grad_norm": 4.427477836608887,
       "learning_rate": 9.293642785065591e-05,
-      "loss": 7.1466,
+      "loss": 6.9789,
       "step": 13500
     },
     {
       "epoch": 5.877413937867338,
-      "grad_norm": 4.993971347808838,
+      "grad_norm": 3.7849512100219727,
       "learning_rate": 9.192734611503532e-05,
-      "loss": 7.1235,
+      "loss": 6.9938,
       "step": 14000
     },
     {
       "epoch": 6.087321578505458,
-      "grad_norm": 5.772039890289307,
+      "grad_norm": 3.943079948425293,
       "learning_rate": 9.091826437941474e-05,
-      "loss": 7.0675,
+      "loss": 6.9069,
       "step": 14500
     },
     {
       "epoch": 6.297229219143577,
-      "grad_norm": 7.119624614715576,
+      "grad_norm": 3.590202808380127,
       "learning_rate": 8.990918264379415e-05,
-      "loss": 7.02,
+      "loss": 6.8403,
       "step": 15000
     },
     {
       "epoch": 6.507136859781696,
-      "grad_norm": 4.125405311584473,
+      "grad_norm": 3.936716318130493,
       "learning_rate": 8.890010090817356e-05,
-      "loss": 7.0209,
+      "loss": 6.849,
       "step": 15500
     },
     {
       "epoch": 6.717044500419815,
-      "grad_norm": 6.1558942794799805,
+      "grad_norm": 3.7201178073883057,
       "learning_rate": 8.789101917255298e-05,
-      "loss": 7.0228,
+      "loss": 6.8739,
       "step": 16000
     },
     {
       "epoch": 6.926952141057934,
-      "grad_norm": 4.960447788238525,
+      "grad_norm": 4.077293872833252,
       "learning_rate": 8.68819374369324e-05,
-      "loss": 7.0323,
+      "loss": 6.8711,
       "step": 16500
     },
     {
       "epoch": 7.136859781696054,
-      "grad_norm": 5.657976150512695,
+      "grad_norm": 5.359983444213867,
       "learning_rate": 8.58728557013118e-05,
-      "loss": 6.9427,
+      "loss": 6.7684,
       "step": 17000
     },
     {
       "epoch": 7.346767422334173,
-      "grad_norm": 3.434267520904541,
+      "grad_norm": 3.2259602546691895,
       "learning_rate": 8.486377396569122e-05,
-      "loss": 6.9043,
+      "loss": 6.7366,
       "step": 17500
     },
     {
       "epoch": 7.5566750629722925,
-      "grad_norm": 5.602636337280273,
+      "grad_norm": 3.0643773078918457,
       "learning_rate": 8.385469223007064e-05,
-      "loss": 6.9223,
+      "loss": 6.749,
       "step": 18000
     },
     {
       "epoch": 7.766582703610411,
-      "grad_norm": 6.14320182800293,
+      "grad_norm": 3.9621706008911133,
       "learning_rate": 8.284561049445006e-05,
-      "loss": 6.9332,
+      "loss": 6.7635,
       "step": 18500
     },
     {
       "epoch": 7.97649034424853,
-      "grad_norm": 4.824306011199951,
+      "grad_norm": 6.353549003601074,
       "learning_rate": 8.183652875882947e-05,
-      "loss": 6.9384,
+      "loss": 6.7757,
       "step": 19000
     },
     {
       "epoch": 8.18639798488665,
-      "grad_norm": 4.595215320587158,
+      "grad_norm": 5.328207015991211,
       "learning_rate": 8.082744702320889e-05,
-      "loss": 6.8172,
+      "loss": 6.6499,
       "step": 19500
     },
     {
       "epoch": 8.39630562552477,
-      "grad_norm": 4.526421546936035,
+      "grad_norm": 3.7361538410186768,
       "learning_rate": 7.98183652875883e-05,
-      "loss": 6.823,
+      "loss": 6.6576,
       "step": 20000
     },
     {
       "epoch": 8.39630562552477,
-      "eval_accuracy": 0.15208058587502454,
-      "eval_loss": 7.090819835662842,
-      "eval_runtime": 0.7267,
-      "eval_samples_per_second": 659.184,
-      "eval_steps_per_second": 6.881,
+      "eval_accuracy": 0.15213670417239541,
+      "eval_loss": 6.99865198135376,
+      "eval_runtime": 0.6716,
+      "eval_samples_per_second": 713.241,
+      "eval_steps_per_second": 7.445,
       "step": 20000
     },
     {
       "epoch": 8.606213266162888,
-      "grad_norm": 5.234260082244873,
+      "grad_norm": 3.931507110595703,
       "learning_rate": 7.880928355196771e-05,
-      "loss": 6.8398,
+      "loss": 6.6643,
       "step": 20500
     },
     {
       "epoch": 8.816120906801007,
-      "grad_norm": 5.007996082305908,
+      "grad_norm": 3.609084129333496,
       "learning_rate": 7.780020181634713e-05,
-      "loss": 6.8443,
+      "loss": 6.681,
       "step": 21000
     },
     {
       "epoch": 9.026028547439127,
-      "grad_norm": 3.5988104343414307,
+      "grad_norm": 3.898296594619751,
       "learning_rate": 7.679112008072655e-05,
-      "loss": 6.841,
+      "loss": 6.6663,
       "step": 21500
     },
     {
       "epoch": 9.235936188077247,
-      "grad_norm": 4.956523418426514,
+      "grad_norm": 3.7589988708496094,
       "learning_rate": 7.578203834510595e-05,
-      "loss": 6.7343,
+      "loss": 6.5639,
       "step": 22000
     },
     {
       "epoch": 9.445843828715365,
-      "grad_norm": 4.012487411499023,
+      "grad_norm": 3.570358991622925,
       "learning_rate": 7.477295660948537e-05,
-      "loss": 6.7515,
+      "loss": 6.577,
       "step": 22500
     },
     {
       "epoch": 9.655751469353484,
-      "grad_norm": 5.125402450561523,
+      "grad_norm": 4.482255458831787,
       "learning_rate": 7.376387487386479e-05,
-      "loss": 6.7666,
+      "loss": 6.5885,
       "step": 23000
     },
     {
       "epoch": 9.865659109991604,
-      "grad_norm": 3.2222142219543457,
+      "grad_norm": 4.959719657897949,
       "learning_rate": 7.27547931382442e-05,
-      "loss": 6.7686,
+      "loss": 6.5963,
       "step": 23500
     },
     {
       "epoch": 10.075566750629722,
-      "grad_norm": 5.873488903045654,
+      "grad_norm": 3.6764087677001953,
       "learning_rate": 7.174571140262362e-05,
-      "loss": 6.7328,
+      "loss": 6.5642,
       "step": 24000
     },
     {
       "epoch": 10.285474391267842,
-      "grad_norm": 3.7630276679992676,
+      "grad_norm": 3.9904074668884277,
       "learning_rate": 7.073662966700304e-05,
-      "loss": 6.6707,
+      "loss": 6.5085,
       "step": 24500
     },
     {
       "epoch": 10.495382031905962,
-      "grad_norm": 4.140413761138916,
+      "grad_norm": 5.6959333419799805,
       "learning_rate": 6.972754793138244e-05,
-      "loss": 6.6835,
+      "loss": 6.5113,
       "step": 25000
     },
     {
       "epoch": 10.705289672544081,
-      "grad_norm": 3.8728461265563965,
+      "grad_norm": 4.355558395385742,
       "learning_rate": 6.871846619576186e-05,
-      "loss": 6.6993,
+      "loss": 6.5253,
       "step": 25500
     },
     {
       "epoch": 10.9151973131822,
-      "grad_norm": 4.102002143859863,
+      "grad_norm": 3.352579116821289,
       "learning_rate": 6.770938446014128e-05,
-      "loss": 6.7049,
+      "loss": 6.5373,
       "step": 26000
     },
     {
       "epoch": 11.125104953820319,
-      "grad_norm": 3.693803548812866,
+      "grad_norm": 3.4871373176574707,
       "learning_rate": 6.670030272452068e-05,
-      "loss": 6.6372,
+      "loss": 6.4823,
       "step": 26500
     },
     {
       "epoch": 11.335012594458439,
-      "grad_norm": 4.381335258483887,
+      "grad_norm": 4.604550361633301,
       "learning_rate": 6.56912209889001e-05,
-      "loss": 6.6139,
+      "loss": 6.4544,
       "step": 27000
     },
     {
       "epoch": 11.544920235096557,
-      "grad_norm": 5.517989635467529,
+      "grad_norm": 3.4807066917419434,
       "learning_rate": 6.468213925327952e-05,
-      "loss": 6.6212,
+      "loss": 6.4589,
       "step": 27500
     },
     {
       "epoch": 11.754827875734676,
-      "grad_norm": 5.365880012512207,
+      "grad_norm": 5.020534515380859,
       "learning_rate": 6.367305751765893e-05,
-      "loss": 6.6352,
+      "loss": 6.4649,
       "step": 28000
     },
     {
       "epoch": 11.964735516372796,
-      "grad_norm": 4.104738235473633,
+      "grad_norm": 4.235457897186279,
       "learning_rate": 6.266397578203835e-05,
-      "loss": 6.6441,
+      "loss": 6.4815,
       "step": 28500
     },
     {
       "epoch": 12.174643157010916,
-      "grad_norm": 6.24871301651001,
+      "grad_norm": 4.625157356262207,
       "learning_rate": 6.165489404641777e-05,
-      "loss": 6.5664,
+      "loss": 6.4089,
       "step": 29000
     },
     {
       "epoch": 12.384550797649034,
-      "grad_norm": 4.707885265350342,
+      "grad_norm": 3.269484043121338,
       "learning_rate": 6.064581231079718e-05,
-      "loss": 6.5564,
+      "loss": 6.4035,
       "step": 29500
     },
     {
       "epoch": 12.594458438287154,
-      "grad_norm": 6.131391525268555,
+      "grad_norm": 2.9809415340423584,
       "learning_rate": 5.963673057517659e-05,
-      "loss": 6.5727,
+      "loss": 6.4107,
       "step": 30000
     },
     {
       "epoch": 12.594458438287154,
-      "eval_accuracy": 0.15283879452668103,
-      "eval_loss": 7.092159748077393,
-      "eval_runtime": 0.7224,
-      "eval_samples_per_second": 663.103,
-      "eval_steps_per_second": 6.922,
+      "eval_accuracy": 0.1536782606262417,
+      "eval_loss": 7.106044769287109,
+      "eval_runtime": 0.6791,
+      "eval_samples_per_second": 705.326,
+      "eval_steps_per_second": 7.362,
       "step": 30000
     },
     {
       "epoch": 12.804366078925273,
-      "grad_norm": 4.942874431610107,
+      "grad_norm": 4.064988613128662,
       "learning_rate": 5.8627648839556e-05,
-      "loss": 6.582,
+      "loss": 6.409,
       "step": 30500
     },
     {
       "epoch": 13.014273719563391,
-      "grad_norm": 4.9076361656188965,
+      "grad_norm": 4.205665588378906,
       "learning_rate": 5.7618567103935416e-05,
-      "loss": 6.5792,
+      "loss": 6.4191,
       "step": 31000
     },
     {
       "epoch": 13.224181360201511,
-      "grad_norm": 5.406895637512207,
+      "grad_norm": 3.8401434421539307,
       "learning_rate": 5.660948536831484e-05,
-      "loss": 6.5005,
+      "loss": 6.3531,
       "step": 31500
     },
     {
       "epoch": 13.43408900083963,
-      "grad_norm": 3.737612247467041,
+      "grad_norm": 3.983386754989624,
       "learning_rate": 5.5600403632694254e-05,
-      "loss": 6.5139,
+      "loss": 6.3577,
       "step": 32000
     },
     {
       "epoch": 13.64399664147775,
-      "grad_norm": 4.210270404815674,
+      "grad_norm": 4.048731327056885,
       "learning_rate": 5.4591321897073666e-05,
-      "loss": 6.5233,
+      "loss": 6.3713,
       "step": 32500
     },
     {
       "epoch": 13.853904282115868,
-      "grad_norm": 4.193741321563721,
+      "grad_norm": 3.627760410308838,
       "learning_rate": 5.358224016145308e-05,
-      "loss": 6.5293,
+      "loss": 6.377,
       "step": 33000
     },
     {
       "epoch": 14.063811922753988,
-      "grad_norm": 4.392582893371582,
+      "grad_norm": 4.206369400024414,
       "learning_rate": 5.257315842583249e-05,
-      "loss": 6.4962,
+      "loss": 6.3519,
       "step": 33500
     },
     {
       "epoch": 14.273719563392108,
-      "grad_norm": 4.345175743103027,
+      "grad_norm": 5.4676737785339355,
       "learning_rate": 5.15640766902119e-05,
-      "loss": 6.4511,
+      "loss": 6.315,
       "step": 34000
     },
     {
       "epoch": 14.483627204030226,
-      "grad_norm": 4.675127983093262,
+      "grad_norm": 3.866525650024414,
       "learning_rate": 5.055499495459133e-05,
-      "loss": 6.4591,
+      "loss": 6.3184,
       "step": 34500
     },
     {
       "epoch": 14.693534844668346,
-      "grad_norm": 5.318904399871826,
+      "grad_norm": 4.393762111663818,
       "learning_rate": 4.954591321897074e-05,
-      "loss": 6.4735,
+      "loss": 6.3267,
       "step": 35000
     },
     {
       "epoch": 14.903442485306465,
-      "grad_norm": 4.547414779663086,
+      "grad_norm": 2.93205189704895,
       "learning_rate": 4.853683148335015e-05,
-      "loss": 6.4836,
+      "loss": 6.3311,
       "step": 35500
     },
     {
       "epoch": 15.113350125944585,
-      "grad_norm": 3.7999465465545654,
+      "grad_norm": 3.1061670780181885,
       "learning_rate": 4.7527749747729566e-05,
-      "loss": 6.4444,
+      "loss": 6.3003,
       "step": 36000
     },
     {
       "epoch": 15.323257766582703,
-      "grad_norm": 3.7221741676330566,
+      "grad_norm": 3.3738293647766113,
       "learning_rate": 4.6518668012108985e-05,
-      "loss": 6.4183,
+      "loss": 6.2854,
       "step": 36500
     },
     {
       "epoch": 15.533165407220823,
-      "grad_norm": 5.092400550842285,
+      "grad_norm": 3.3892741203308105,
       "learning_rate": 4.55095862764884e-05,
-      "loss": 6.4232,
+      "loss": 6.2834,
       "step": 37000
     },
     {
       "epoch": 15.743073047858942,
-      "grad_norm": 4.360398769378662,
+      "grad_norm": 4.4049882888793945,
       "learning_rate": 4.450050454086781e-05,
-      "loss": 6.4322,
+      "loss": 6.2922,
       "step": 37500
     },
     {
       "epoch": 15.95298068849706,
-      "grad_norm": 6.7608771324157715,
+      "grad_norm": 3.8126187324523926,
       "learning_rate": 4.349142280524723e-05,
-      "loss": 6.4268,
+      "loss": 6.289,
       "step": 38000
     },
     {
       "epoch": 16.162888329135182,
-      "grad_norm": 3.7004191875457764,
+      "grad_norm": 3.5885815620422363,
       "learning_rate": 4.248234106962664e-05,
-      "loss": 6.3873,
+      "loss": 6.2624,
       "step": 38500
     },
     {
       "epoch": 16.3727959697733,
-      "grad_norm": 5.158031940460205,
+      "grad_norm": 3.0007107257843018,
       "learning_rate": 4.147325933400605e-05,
-      "loss": 6.3745,
+      "loss": 6.2549,
       "step": 39000
     },
     {
       "epoch": 16.582703610411418,
-      "grad_norm": 5.433754920959473,
+      "grad_norm": 4.066105842590332,
       "learning_rate": 4.046417759838547e-05,
-      "loss": 6.3898,
+      "loss": 6.2595,
       "step": 39500
     },
     {
       "epoch": 16.79261125104954,
-      "grad_norm": 4.476456642150879,
+      "grad_norm": 4.098573207855225,
       "learning_rate": 3.9455095862764884e-05,
-      "loss": 6.3832,
+      "loss": 6.2498,
       "step": 40000
     },
     {
       "epoch": 16.79261125104954,
-      "eval_accuracy": 0.15167657283839855,
-      "eval_loss": 7.213654041290283,
-      "eval_runtime": 0.7254,
-      "eval_samples_per_second": 660.348,
-      "eval_steps_per_second": 6.893,
+      "eval_accuracy": 0.15184479084894023,
+      "eval_loss": 7.238094806671143,
+      "eval_runtime": 0.6696,
+      "eval_samples_per_second": 715.358,
+      "eval_steps_per_second": 7.467,
       "step": 40000
     },
     {
       "epoch": 17.002518891687657,
-      "grad_norm": 5.207530498504639,
+      "grad_norm": 3.418558120727539,
       "learning_rate": 3.84460141271443e-05,
-      "loss": 6.3929,
+      "loss": 6.2602,
       "step": 40500
     },
     {
       "epoch": 17.212426532325775,
-      "grad_norm": 6.294618606567383,
+      "grad_norm": 2.369943141937256,
       "learning_rate": 3.7436932391523715e-05,
-      "loss": 6.3353,
+      "loss": 6.2228,
       "step": 41000
     },
     {
       "epoch": 17.422334172963897,
-      "grad_norm": 3.660827398300171,
+      "grad_norm": 3.2603626251220703,
       "learning_rate": 3.642785065590313e-05,
-      "loss": 6.3375,
+      "loss": 6.2204,
       "step": 41500
     },
     {
       "epoch": 17.632241813602015,
-      "grad_norm": 5.005442142486572,
+      "grad_norm": 3.710078716278076,
       "learning_rate": 3.541876892028255e-05,
-      "loss": 6.3469,
+      "loss": 6.228,
       "step": 42000
     },
     {
       "epoch": 17.842149454240133,
-      "grad_norm": 4.2572021484375,
+      "grad_norm": 3.4049365520477295,
       "learning_rate": 3.440968718466196e-05,
-      "loss": 6.354,
+      "loss": 6.2248,
       "step": 42500
     },
     {
       "epoch": 18.052057094878254,
-      "grad_norm": 4.661770343780518,
+      "grad_norm": 2.79130220413208,
       "learning_rate": 3.340060544904137e-05,
-      "loss": 6.3376,
+      "loss": 6.2231,
       "step": 43000
     },
     {
       "epoch": 18.261964735516372,
-      "grad_norm": 3.612496852874756,
+      "grad_norm": 3.3371455669403076,
       "learning_rate": 3.239152371342079e-05,
-      "loss": 6.3026,
+      "loss": 6.1973,
       "step": 43500
     },
     {
       "epoch": 18.471872376154494,
-      "grad_norm": 5.20078706741333,
+      "grad_norm": 2.428114652633667,
       "learning_rate": 3.13824419778002e-05,
-      "loss": 6.3132,
+      "loss": 6.1998,
       "step": 44000
     },
     {
       "epoch": 18.68178001679261,
-      "grad_norm": 4.2581377029418945,
+      "grad_norm": 2.6758224964141846,
       "learning_rate": 3.0373360242179615e-05,
-      "loss": 6.3149,
+      "loss": 6.2021,
       "step": 44500
     },
     {
       "epoch": 18.89168765743073,
-      "grad_norm": 3.2295942306518555,
+      "grad_norm": 3.9563393592834473,
       "learning_rate": 2.9364278506559034e-05,
-      "loss": 6.3142,
+      "loss": 6.2033,
       "step": 45000
     },
     {
       "epoch": 19.10159529806885,
-      "grad_norm": 3.3266310691833496,
+      "grad_norm": 3.574024200439453,
       "learning_rate": 2.8355196770938446e-05,
-      "loss": 6.2997,
+      "loss": 6.1908,
       "step": 45500
     },
     {
       "epoch": 19.31150293870697,
-      "grad_norm": 4.008757591247559,
+      "grad_norm": 4.8180365562438965,
       "learning_rate": 2.734611503531786e-05,
-      "loss": 6.2756,
+      "loss": 6.1749,
       "step": 46000
     },
     {
       "epoch": 19.521410579345087,
-      "grad_norm": 5.275732040405273,
+      "grad_norm": 3.9917449951171875,
       "learning_rate": 2.6337033299697278e-05,
-      "loss": 6.2822,
+      "loss": 6.1795,
       "step": 46500
     },
     {
       "epoch": 19.73131821998321,
-      "grad_norm": 3.3725156784057617,
+      "grad_norm": 2.8837454319000244,
       "learning_rate": 2.532795156407669e-05,
-      "loss": 6.2861,
+      "loss": 6.1818,
       "step": 47000
     },
     {
       "epoch": 19.941225860621326,
-      "grad_norm": 5.234038829803467,
+      "grad_norm": 2.209712505340576,
       "learning_rate": 2.4318869828456106e-05,
-      "loss": 6.2813,
+      "loss": 6.1768,
       "step": 47500
     },
     {
       "epoch": 20.151133501259444,
-      "grad_norm": 3.5982561111450195,
+      "grad_norm": 2.953699827194214,
       "learning_rate": 2.330978809283552e-05,
-      "loss": 6.2652,
+      "loss": 6.1659,
       "step": 48000
     },
     {
       "epoch": 20.361041141897566,
-      "grad_norm": 4.841806888580322,
+      "grad_norm": 1.9596891403198242,
       "learning_rate": 2.2300706357214934e-05,
-      "loss": 6.2542,
+      "loss": 6.1623,
       "step": 48500
     },
     {
       "epoch": 20.570948782535684,
-      "grad_norm": 3.018764019012451,
+      "grad_norm": 2.1720826625823975,
       "learning_rate": 2.129162462159435e-05,
-      "loss": 6.2579,
+      "loss": 6.1551,
       "step": 49000
     },
     {
       "epoch": 20.7808564231738,
-      "grad_norm": 3.967012643814087,
+      "grad_norm": 2.9273760318756104,
       "learning_rate": 2.0282542885973765e-05,
-      "loss": 6.2571,
+      "loss": 6.1578,
       "step": 49500
     },
     {
       "epoch": 20.990764063811923,
-      "grad_norm": 4.151165008544922,
+      "grad_norm": 3.1823031902313232,
       "learning_rate": 1.927346115035318e-05,
-      "loss": 6.2578,
+      "loss": 6.1573,
       "step": 50000
     },
     {
       "epoch": 20.990764063811923,
-      "eval_accuracy": 0.1502789693554266,
-      "eval_loss": 7.287151336669922,
-      "eval_runtime": 0.7133,
-      "eval_samples_per_second": 671.56,
-      "eval_steps_per_second": 7.01,
+      "eval_accuracy": 0.1508957860206914,
+      "eval_loss": 7.1608357429504395,
+      "eval_runtime": 0.6746,
+      "eval_samples_per_second": 710.021,
+      "eval_steps_per_second": 7.411,
       "step": 50000
     },
     {
       "epoch": 21.20067170445004,
-      "grad_norm": 3.9882612228393555,
+      "grad_norm": 3.508307456970215,
       "learning_rate": 1.8264379414732593e-05,
-      "loss": 6.2318,
+      "loss": 6.1446,
       "step": 50500
     },
     {
       "epoch": 21.410579345088163,
-      "grad_norm": 5.8506622314453125,
+      "grad_norm": 4.2598114013671875,
       "learning_rate": 1.725529767911201e-05,
-      "loss": 6.2353,
+      "loss": 6.1444,
       "step": 51000
     },
     {
       "epoch": 21.62048698572628,
-      "grad_norm": 3.6497795581817627,
+      "grad_norm": 3.6656954288482666,
       "learning_rate": 1.6246215943491424e-05,
-      "loss": 6.2369,
+      "loss": 6.141,
       "step": 51500
     },
     {
       "epoch": 21.8303946263644,
-      "grad_norm": 5.096824645996094,
+      "grad_norm": 1.9702938795089722,
       "learning_rate": 1.5237134207870837e-05,
-      "loss": 6.2346,
+      "loss": 6.1406,
       "step": 52000
     },
     {
       "epoch": 22.04030226700252,
-      "grad_norm": 3.092406988143921,
+      "grad_norm": 2.336904287338257,
       "learning_rate": 1.4228052472250252e-05,
-      "loss": 6.2302,
+      "loss": 6.145,
       "step": 52500
     },
     {
       "epoch": 22.250209907640638,
-      "grad_norm": 3.3946568965911865,
+      "grad_norm": 1.8393386602401733,
       "learning_rate": 1.3218970736629668e-05,
-      "loss": 6.2096,
+      "loss": 6.124,
       "step": 53000
     },
     {
       "epoch": 22.460117548278756,
-      "grad_norm": 4.148237705230713,
+      "grad_norm": 3.7537992000579834,
       "learning_rate": 1.2209889001009082e-05,
-      "loss": 6.2113,
+      "loss": 6.1278,
       "step": 53500
     },
     {
       "epoch": 22.670025188916878,
-      "grad_norm": 4.098395824432373,
+      "grad_norm": 3.7978570461273193,
       "learning_rate": 1.1200807265388496e-05,
-      "loss": 6.2178,
+      "loss": 6.1345,
       "step": 54000
     },
     {
       "epoch": 22.879932829554996,
-      "grad_norm": 3.1911346912384033,
+      "grad_norm": 1.7555248737335205,
       "learning_rate": 1.0191725529767912e-05,
-      "loss": 6.2094,
+      "loss": 6.1229,
       "step": 54500
     },
     {
       "epoch": 23.089840470193113,
-      "grad_norm": 3.468717098236084,
+      "grad_norm": 2.5084242820739746,
       "learning_rate": 9.182643794147326e-06,
-      "loss": 6.2047,
+      "loss": 6.1222,
       "step": 55000
     },
     {
       "epoch": 23.299748110831235,
-      "grad_norm": 3.5712525844573975,
+      "grad_norm": 3.398545980453491,
       "learning_rate": 8.173562058526741e-06,
-      "loss": 6.1994,
+      "loss": 6.1161,
       "step": 55500
     },
     {
       "epoch": 23.509655751469353,
-      "grad_norm": 3.9495656490325928,
+      "grad_norm": 3.7442221641540527,
       "learning_rate": 7.164480322906156e-06,
-      "loss": 6.1992,
+      "loss": 6.1174,
       "step": 56000
     },
     {
       "epoch": 23.719563392107474,
-      "grad_norm": 3.7151174545288086,
+      "grad_norm": 1.4284420013427734,
       "learning_rate": 6.155398587285571e-06,
-      "loss": 6.194,
+      "loss": 6.1128,
       "step": 56500
     },
     {
       "epoch": 23.929471032745592,
-      "grad_norm": 4.365117073059082,
+      "grad_norm": 3.252730369567871,
       "learning_rate": 5.146316851664986e-06,
-      "loss": 6.1972,
+      "loss": 6.1144,
       "step": 57000
     },
     {
       "epoch": 24.13937867338371,
-      "grad_norm": 2.835233211517334,
+      "grad_norm": 1.8585278987884521,
       "learning_rate": 4.1372351160444e-06,
-      "loss": 6.185,
+      "loss": 6.1093,
       "step": 57500
     },
     {
       "epoch": 24.349286314021832,
-      "grad_norm": 3.5119664669036865,
+      "grad_norm": 1.2302392721176147,
       "learning_rate": 3.1281533804238145e-06,
-      "loss": 6.1889,
+      "loss": 6.1091,
       "step": 58000
     },
     {
       "epoch": 24.55919395465995,
-      "grad_norm": 4.836373329162598,
+      "grad_norm": 2.466769218444824,
       "learning_rate": 2.1190716448032293e-06,
-      "loss": 6.1841,
+      "loss": 6.1083,
       "step": 58500
     },
     {
       "epoch": 24.769101595298068,
-      "grad_norm": 2.775986909866333,
+      "grad_norm": 2.1164252758026123,
       "learning_rate": 1.109989909182644e-06,
-      "loss": 6.1826,
+      "loss": 6.1027,
       "step": 59000
     },
     {
       "epoch": 24.97900923593619,
-      "grad_norm": 3.45804762840271,
+      "grad_norm": 1.3311982154846191,
       "learning_rate": 1.0090817356205853e-07,
-      "loss": 6.1831,
+      "loss": 6.1052,
       "step": 59500
     },
     {
       "epoch": 25.0,
       "step": 59550,
       "total_flos": 3.058354515764736e+17,
-      "train_loss": 6.935516868866761,
-      "train_runtime": 11960.537,
-      "train_samples_per_second": 477.903,
-      "train_steps_per_second": 4.979
+      "train_loss": 6.8441891810035225,
+      "train_runtime": 11476.7871,
+      "train_samples_per_second": 498.047,
+      "train_steps_per_second": 5.189
     }
   ],
   "logging_steps": 500,