End of training

README.md

@@ -1,11 +1,23 @@
 ---
 tags:
 - generated_from_trainer
+datasets:
+- Hartunka/processed_wikitext-103-raw-v1-km-20_v2
 metrics:
 - accuracy
 model-index:
 - name: tiny_bert_km_20_v2
-  results: []
+  results:
+  - task:
+      name: Masked Language Modeling
+      type: fill-mask
+    dataset:
+      name: Hartunka/processed_wikitext-103-raw-v1-km-20_v2
+      type: Hartunka/processed_wikitext-103-raw-v1-km-20_v2
+    metrics:
+    - name: Accuracy
+      type: accuracy
+      value: 0.15406566084647116
 ---
 
 <!-- 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_20_v2
 
-This model is a fine-tuned version of [](https://huggingface.co/) on an unknown dataset.
+This model is a fine-tuned version of [](https://huggingface.co/) on the Hartunka/processed_wikitext-103-raw-v1-km-20_v2 dataset.
 It achieves the following results on the evaluation set:
-- Loss: 6.7075
-- Accuracy: 0.1511
+- Loss: 6.5066
+- Accuracy: 0.1541
 
 ## Model description
 

all_results.json

@@ -1,16 +1,16 @@
 {
     "epoch": 25.0,
-    "eval_accuracy": 0.15400915409391422,
-    "eval_loss": 6.759331226348877,
-    "eval_runtime": 0.6778,
+    "eval_accuracy": 0.15406566084647116,
+    "eval_loss": 6.5065765380859375,
+    "eval_runtime": 0.6822,
     "eval_samples": 479,
-    "eval_samples_per_second": 706.67,
-    "eval_steps_per_second": 7.377,
-    "perplexity": 862.0654762872451,
+    "eval_samples_per_second": 702.098,
+    "eval_steps_per_second": 7.329,
+    "perplexity": 669.5303778823405,
     "total_flos": 3.051148107142656e+17,
-    "train_loss": 6.524886467434797,
-    "train_runtime": 11610.4539,
+    "train_loss": 6.453962587345357,
+    "train_runtime": 11281.7551,
     "train_samples": 228639,
-    "train_samples_per_second": 492.313,
-    "train_steps_per_second": 5.129
+    "train_samples_per_second": 506.657,
+    "train_steps_per_second": 5.278
 }

eval_results.json

@@ -1,10 +1,10 @@
 {
     "epoch": 25.0,
-    "eval_accuracy": 0.15400915409391422,
-    "eval_loss": 6.759331226348877,
-    "eval_runtime": 0.6778,
+    "eval_accuracy": 0.15406566084647116,
+    "eval_loss": 6.5065765380859375,
+    "eval_runtime": 0.6822,
     "eval_samples": 479,
-    "eval_samples_per_second": 706.67,
-    "eval_steps_per_second": 7.377,
-    "perplexity": 862.0654762872451
+    "eval_samples_per_second": 702.098,
+    "eval_steps_per_second": 7.329,
+    "perplexity": 669.5303778823405
 }

train_results.json

@@ -1,9 +1,9 @@
 {
     "epoch": 25.0,
     "total_flos": 3.051148107142656e+17,
-    "train_loss": 6.524886467434797,
-    "train_runtime": 11610.4539,
+    "train_loss": 6.453962587345357,
+    "train_runtime": 11281.7551,
     "train_samples": 228639,
-    "train_samples_per_second": 492.313,
-    "train_steps_per_second": 5.129
+    "train_samples_per_second": 506.657,
+    "train_steps_per_second": 5.278
 }

trainer_state.json

@@ -1,5 +1,5 @@
 {
-  "best_metric": 6.741695880889893,
+  "best_metric": 6.54965877532959,
   "best_model_checkpoint": "tiny_bert_km_20_v2/checkpoint-20000",
   "epoch": 25.0,
   "eval_steps": 10000,
@@ -10,890 +10,890 @@
   "log_history": [
     {
       "epoch": 0.20990764063811923,
-      "grad_norm": 2.626953125,
+      "grad_norm": 2.71126389503479,
       "learning_rate": 5e-06,
-      "loss": 12.7573,
+      "loss": 12.7506,
       "step": 500
     },
     {
       "epoch": 0.41981528127623846,
-      "grad_norm": 1.9100334644317627,
+      "grad_norm": 1.9715861082077026,
       "learning_rate": 1e-05,
-      "loss": 11.1676,
+      "loss": 11.1644,
       "step": 1000
     },
     {
       "epoch": 0.6297229219143576,
-      "grad_norm": 2.928952932357788,
+      "grad_norm": 5.0529069900512695,
       "learning_rate": 1.5e-05,
-      "loss": 9.7025,
+      "loss": 9.7797,
       "step": 1500
     },
     {
       "epoch": 0.8396305625524769,
-      "grad_norm": 5.528791904449463,
+      "grad_norm": 4.67584228515625,
       "learning_rate": 2e-05,
-      "loss": 8.6156,
+      "loss": 8.7486,
       "step": 2000
     },
     {
       "epoch": 1.0495382031905962,
-      "grad_norm": 8.937955856323242,
+      "grad_norm": 4.135430335998535,
       "learning_rate": 2.5e-05,
-      "loss": 7.9967,
+      "loss": 8.1275,
       "step": 2500
     },
     {
       "epoch": 1.2594458438287153,
-      "grad_norm": 3.489656925201416,
+      "grad_norm": 5.515730857849121,
       "learning_rate": 3e-05,
-      "loss": 7.6977,
+      "loss": 7.8004,
       "step": 3000
     },
     {
       "epoch": 1.4693534844668346,
-      "grad_norm": 7.35763692855835,
+      "grad_norm": 3.708866834640503,
       "learning_rate": 3.5e-05,
-      "loss": 7.5306,
+      "loss": 7.6218,
       "step": 3500
     },
     {
       "epoch": 1.6792611251049538,
-      "grad_norm": 8.683979988098145,
+      "grad_norm": 4.445878505706787,
       "learning_rate": 4e-05,
-      "loss": 7.4398,
+      "loss": 7.4752,
       "step": 4000
     },
     {
       "epoch": 1.8891687657430731,
-      "grad_norm": 6.048260688781738,
+      "grad_norm": 3.3583083152770996,
       "learning_rate": 4.5e-05,
-      "loss": 7.3427,
+      "loss": 7.3382,
       "step": 4500
     },
     {
       "epoch": 2.0990764063811924,
-      "grad_norm": 6.637815475463867,
+      "grad_norm": 3.607733964920044,
       "learning_rate": 5e-05,
-      "loss": 7.2302,
+      "loss": 7.2035,
       "step": 5000
     },
     {
       "epoch": 2.3089840470193117,
-      "grad_norm": 3.465994358062744,
+      "grad_norm": 4.7366204261779785,
       "learning_rate": 5.500000000000001e-05,
-      "loss": 7.1513,
+      "loss": 7.0941,
       "step": 5500
     },
     {
       "epoch": 2.5188916876574305,
-      "grad_norm": 5.877224922180176,
+      "grad_norm": 4.448856830596924,
       "learning_rate": 6e-05,
-      "loss": 7.1207,
+      "loss": 7.044,
       "step": 6000
     },
     {
       "epoch": 2.72879932829555,
-      "grad_norm": 6.2582573890686035,
+      "grad_norm": 3.4905285835266113,
       "learning_rate": 6.500000000000001e-05,
-      "loss": 7.0765,
+      "loss": 6.9928,
       "step": 6500
     },
     {
       "epoch": 2.938706968933669,
-      "grad_norm": 9.96641731262207,
+      "grad_norm": 2.962218761444092,
       "learning_rate": 7e-05,
-      "loss": 7.0505,
+      "loss": 6.9435,
       "step": 7000
     },
     {
       "epoch": 3.1486146095717884,
-      "grad_norm": 5.077490329742432,
+      "grad_norm": 2.484970808029175,
       "learning_rate": 7.500000000000001e-05,
-      "loss": 6.9615,
+      "loss": 6.8375,
       "step": 7500
     },
     {
       "epoch": 3.3585222502099077,
-      "grad_norm": 5.034363269805908,
+      "grad_norm": 5.155412673950195,
       "learning_rate": 8e-05,
-      "loss": 6.9372,
+      "loss": 6.7932,
       "step": 8000
     },
     {
       "epoch": 3.568429890848027,
-      "grad_norm": 4.603774547576904,
+      "grad_norm": 2.8160481452941895,
       "learning_rate": 8.5e-05,
-      "loss": 6.9235,
+      "loss": 6.7949,
       "step": 8500
     },
     {
       "epoch": 3.7783375314861463,
-      "grad_norm": 6.528964996337891,
+      "grad_norm": 2.7087278366088867,
       "learning_rate": 9e-05,
-      "loss": 6.9028,
+      "loss": 6.7745,
       "step": 9000
     },
     {
       "epoch": 3.988245172124265,
-      "grad_norm": 4.090594291687012,
+      "grad_norm": 2.5564920902252197,
       "learning_rate": 9.5e-05,
-      "loss": 6.8855,
+      "loss": 6.759,
       "step": 9500
     },
     {
       "epoch": 4.198152812762385,
-      "grad_norm": 3.793991804122925,
+      "grad_norm": 2.553255796432495,
       "learning_rate": 0.0001,
-      "loss": 6.7927,
+      "loss": 6.6514,
       "step": 10000
     },
     {
       "epoch": 4.198152812762385,
-      "eval_accuracy": 0.14772252984223266,
-      "eval_loss": 6.8603620529174805,
-      "eval_runtime": 0.6918,
-      "eval_samples_per_second": 692.389,
-      "eval_steps_per_second": 7.227,
+      "eval_accuracy": 0.14850162775814574,
+      "eval_loss": 6.7570953369140625,
+      "eval_runtime": 0.7121,
+      "eval_samples_per_second": 672.648,
+      "eval_steps_per_second": 7.021,
       "step": 10000
     },
     {
       "epoch": 4.408060453400504,
-      "grad_norm": 4.0062031745910645,
+      "grad_norm": 4.206502437591553,
       "learning_rate": 9.899091826437942e-05,
-      "loss": 6.7996,
+      "loss": 6.6586,
       "step": 10500
     },
     {
       "epoch": 4.617968094038623,
-      "grad_norm": 2.6802749633789062,
+      "grad_norm": 3.871795415878296,
       "learning_rate": 9.798183652875883e-05,
-      "loss": 6.792,
+      "loss": 6.6555,
       "step": 11000
     },
     {
       "epoch": 4.827875734676742,
-      "grad_norm": 2.7341952323913574,
+      "grad_norm": 2.5797126293182373,
       "learning_rate": 9.697275479313825e-05,
-      "loss": 6.778,
+      "loss": 6.6546,
       "step": 11500
     },
     {
       "epoch": 5.037783375314861,
-      "grad_norm": 3.6597518920898438,
+      "grad_norm": 2.1714375019073486,
       "learning_rate": 9.596367305751767e-05,
-      "loss": 6.7364,
+      "loss": 6.6229,
       "step": 12000
     },
     {
       "epoch": 5.247691015952981,
-      "grad_norm": 2.878286361694336,
+      "grad_norm": 2.7962558269500732,
       "learning_rate": 9.495459132189707e-05,
-      "loss": 6.6724,
+      "loss": 6.5348,
       "step": 12500
     },
     {
       "epoch": 5.4575986565911,
-      "grad_norm": 3.2767865657806396,
+      "grad_norm": 3.548794984817505,
       "learning_rate": 9.394550958627649e-05,
-      "loss": 6.6691,
+      "loss": 6.5446,
       "step": 13000
     },
     {
       "epoch": 5.667506297229219,
-      "grad_norm": 2.9648525714874268,
+      "grad_norm": 2.6331114768981934,
       "learning_rate": 9.293642785065591e-05,
-      "loss": 6.6692,
+      "loss": 6.5454,
       "step": 13500
     },
     {
       "epoch": 5.877413937867338,
-      "grad_norm": 2.79672908782959,
+      "grad_norm": 4.004150390625,
       "learning_rate": 9.192734611503532e-05,
-      "loss": 6.6608,
+      "loss": 6.54,
       "step": 14000
     },
     {
       "epoch": 6.087321578505458,
-      "grad_norm": 4.109856128692627,
+      "grad_norm": 2.731539249420166,
       "learning_rate": 9.091826437941474e-05,
-      "loss": 6.6172,
+      "loss": 6.499,
       "step": 14500
     },
     {
       "epoch": 6.297229219143577,
-      "grad_norm": 2.0831563472747803,
+      "grad_norm": 2.7149033546447754,
       "learning_rate": 8.990918264379415e-05,
-      "loss": 6.5674,
+      "loss": 6.4526,
       "step": 15000
     },
     {
       "epoch": 6.507136859781696,
-      "grad_norm": 2.1591970920562744,
+      "grad_norm": 2.085320472717285,
       "learning_rate": 8.890010090817356e-05,
-      "loss": 6.5737,
+      "loss": 6.4567,
       "step": 15500
     },
     {
       "epoch": 6.717044500419815,
-      "grad_norm": 2.8322367668151855,
+      "grad_norm": 3.539597272872925,
       "learning_rate": 8.789101917255298e-05,
-      "loss": 6.5691,
+      "loss": 6.4563,
       "step": 16000
     },
     {
       "epoch": 6.926952141057934,
-      "grad_norm": 2.9014763832092285,
+      "grad_norm": 2.0627574920654297,
       "learning_rate": 8.68819374369324e-05,
-      "loss": 6.5687,
+      "loss": 6.4603,
       "step": 16500
     },
     {
       "epoch": 7.136859781696054,
-      "grad_norm": 2.0010359287261963,
+      "grad_norm": 2.7053449153900146,
       "learning_rate": 8.58728557013118e-05,
-      "loss": 6.5101,
+      "loss": 6.3975,
       "step": 17000
     },
     {
       "epoch": 7.346767422334173,
-      "grad_norm": 4.102756500244141,
+      "grad_norm": 2.3060944080352783,
       "learning_rate": 8.486377396569122e-05,
-      "loss": 6.489,
+      "loss": 6.3735,
       "step": 17500
     },
     {
       "epoch": 7.5566750629722925,
-      "grad_norm": 3.5026400089263916,
+      "grad_norm": 2.0664925575256348,
       "learning_rate": 8.385469223007064e-05,
-      "loss": 6.4962,
+      "loss": 6.3892,
       "step": 18000
     },
     {
       "epoch": 7.766582703610411,
-      "grad_norm": 3.1486175060272217,
+      "grad_norm": 3.5700786113739014,
       "learning_rate": 8.284561049445006e-05,
-      "loss": 6.4984,
+      "loss": 6.3867,
       "step": 18500
     },
     {
       "epoch": 7.97649034424853,
-      "grad_norm": 2.6257808208465576,
+      "grad_norm": 2.6246838569641113,
       "learning_rate": 8.183652875882947e-05,
-      "loss": 6.4964,
+      "loss": 6.3911,
       "step": 19000
     },
     {
       "epoch": 8.18639798488665,
-      "grad_norm": 2.178621530532837,
+      "grad_norm": 2.0920815467834473,
       "learning_rate": 8.082744702320889e-05,
-      "loss": 6.4233,
+      "loss": 6.3265,
       "step": 19500
     },
     {
       "epoch": 8.39630562552477,
-      "grad_norm": 2.481250286102295,
+      "grad_norm": 2.441361665725708,
       "learning_rate": 7.98183652875883e-05,
-      "loss": 6.4283,
+      "loss": 6.326,
       "step": 20000
     },
     {
       "epoch": 8.39630562552477,
-      "eval_accuracy": 0.15295041948427285,
-      "eval_loss": 6.741695880889893,
-      "eval_runtime": 0.6713,
-      "eval_samples_per_second": 713.519,
-      "eval_steps_per_second": 7.448,
+      "eval_accuracy": 0.15264176884873312,
+      "eval_loss": 6.54965877532959,
+      "eval_runtime": 0.6605,
+      "eval_samples_per_second": 725.235,
+      "eval_steps_per_second": 7.57,
       "step": 20000
     },
     {
       "epoch": 8.606213266162888,
-      "grad_norm": 3.4793574810028076,
+      "grad_norm": 2.759524345397949,
       "learning_rate": 7.880928355196771e-05,
-      "loss": 6.4295,
+      "loss": 6.3265,
       "step": 20500
     },
     {
       "epoch": 8.816120906801007,
-      "grad_norm": 3.2521069049835205,
+      "grad_norm": 3.454348564147949,
       "learning_rate": 7.780020181634713e-05,
-      "loss": 6.4308,
+      "loss": 6.3344,
       "step": 21000
     },
     {
       "epoch": 9.026028547439127,
-      "grad_norm": 2.734280824661255,
+      "grad_norm": 2.499746322631836,
       "learning_rate": 7.679112008072655e-05,
-      "loss": 6.4231,
+      "loss": 6.3274,
       "step": 21500
     },
     {
       "epoch": 9.235936188077247,
-      "grad_norm": 2.5010018348693848,
+      "grad_norm": 2.2949016094207764,
       "learning_rate": 7.578203834510595e-05,
-      "loss": 6.369,
+      "loss": 6.2767,
       "step": 22000
     },
     {
       "epoch": 9.445843828715365,
-      "grad_norm": 2.327057123184204,
+      "grad_norm": 1.830916166305542,
       "learning_rate": 7.477295660948537e-05,
-      "loss": 6.374,
+      "loss": 6.2786,
       "step": 22500
     },
     {
       "epoch": 9.655751469353484,
-      "grad_norm": 5.40192985534668,
+      "grad_norm": 3.42680025100708,
       "learning_rate": 7.376387487386479e-05,
-      "loss": 6.3733,
+      "loss": 6.2788,
       "step": 23000
     },
     {
       "epoch": 9.865659109991604,
-      "grad_norm": 3.349362850189209,
+      "grad_norm": 1.889884114265442,
       "learning_rate": 7.27547931382442e-05,
-      "loss": 6.3776,
+      "loss": 6.2782,
       "step": 23500
     },
     {
       "epoch": 10.075566750629722,
-      "grad_norm": 3.010235071182251,
+      "grad_norm": 1.8403366804122925,
       "learning_rate": 7.174571140262362e-05,
-      "loss": 6.3518,
+      "loss": 6.2553,
       "step": 24000
     },
     {
       "epoch": 10.285474391267842,
-      "grad_norm": 3.9260144233703613,
+      "grad_norm": 2.413959503173828,
       "learning_rate": 7.073662966700304e-05,
-      "loss": 6.323,
+      "loss": 6.2381,
       "step": 24500
     },
     {
       "epoch": 10.495382031905962,
-      "grad_norm": 2.4476406574249268,
+      "grad_norm": 2.354292631149292,
       "learning_rate": 6.972754793138244e-05,
-      "loss": 6.326,
+      "loss": 6.2367,
       "step": 25000
     },
     {
       "epoch": 10.705289672544081,
-      "grad_norm": 3.331364631652832,
+      "grad_norm": 2.1689646244049072,
       "learning_rate": 6.871846619576186e-05,
-      "loss": 6.3317,
+      "loss": 6.2348,
       "step": 25500
     },
     {
       "epoch": 10.9151973131822,
-      "grad_norm": 3.0888099670410156,
+      "grad_norm": 2.645822286605835,
       "learning_rate": 6.770938446014128e-05,
-      "loss": 6.3366,
+      "loss": 6.2401,
       "step": 26000
     },
     {
       "epoch": 11.125104953820319,
-      "grad_norm": 2.096848964691162,
+      "grad_norm": 2.8303282260894775,
       "learning_rate": 6.670030272452068e-05,
-      "loss": 6.2984,
+      "loss": 6.212,
       "step": 26500
     },
     {
       "epoch": 11.335012594458439,
-      "grad_norm": 3.3840293884277344,
+      "grad_norm": 3.3768234252929688,
       "learning_rate": 6.56912209889001e-05,
-      "loss": 6.2809,
+      "loss": 6.2001,
       "step": 27000
     },
     {
       "epoch": 11.544920235096557,
-      "grad_norm": 3.304489850997925,
+      "grad_norm": 3.458333730697632,
       "learning_rate": 6.468213925327952e-05,
-      "loss": 6.2798,
+      "loss": 6.2017,
       "step": 27500
     },
     {
       "epoch": 11.754827875734676,
-      "grad_norm": 3.8819022178649902,
+      "grad_norm": 2.546410322189331,
       "learning_rate": 6.367305751765893e-05,
-      "loss": 6.2888,
+      "loss": 6.2012,
       "step": 28000
     },
     {
       "epoch": 11.964735516372796,
-      "grad_norm": 1.6290197372436523,
+      "grad_norm": 2.704577922821045,
       "learning_rate": 6.266397578203835e-05,
-      "loss": 6.2852,
+      "loss": 6.2013,
       "step": 28500
     },
     {
       "epoch": 12.174643157010916,
-      "grad_norm": 3.2812535762786865,
+      "grad_norm": 1.857717752456665,
       "learning_rate": 6.165489404641777e-05,
-      "loss": 6.2519,
+      "loss": 6.1723,
       "step": 29000
     },
     {
       "epoch": 12.384550797649034,
-      "grad_norm": 1.8753509521484375,
+      "grad_norm": 2.8042995929718018,
       "learning_rate": 6.064581231079718e-05,
-      "loss": 6.241,
+      "loss": 6.1654,
       "step": 29500
     },
     {
       "epoch": 12.594458438287154,
-      "grad_norm": 2.2194597721099854,
+      "grad_norm": 2.007753849029541,
       "learning_rate": 5.963673057517659e-05,
-      "loss": 6.2529,
+      "loss": 6.1671,
       "step": 30000
     },
     {
       "epoch": 12.594458438287154,
-      "eval_accuracy": 0.15337045638973612,
-      "eval_loss": 6.791427135467529,
-      "eval_runtime": 0.6839,
-      "eval_samples_per_second": 700.414,
-      "eval_steps_per_second": 7.311,
+      "eval_accuracy": 0.15440579791252762,
+      "eval_loss": 6.653005599975586,
+      "eval_runtime": 0.6695,
+      "eval_samples_per_second": 715.463,
+      "eval_steps_per_second": 7.468,
       "step": 30000
     },
     {
       "epoch": 12.804366078925273,
-      "grad_norm": 2.6068763732910156,
+      "grad_norm": 2.3782219886779785,
       "learning_rate": 5.8627648839556e-05,
-      "loss": 6.248,
+      "loss": 6.1671,
       "step": 30500
     },
     {
       "epoch": 13.014273719563391,
-      "grad_norm": 2.780801773071289,
+      "grad_norm": 2.7273099422454834,
       "learning_rate": 5.7618567103935416e-05,
-      "loss": 6.2496,
+      "loss": 6.1705,
       "step": 31000
     },
     {
       "epoch": 13.224181360201511,
-      "grad_norm": 2.0558760166168213,
+      "grad_norm": 1.9647295475006104,
       "learning_rate": 5.660948536831484e-05,
-      "loss": 6.2119,
+      "loss": 6.1404,
       "step": 31500
     },
     {
       "epoch": 13.43408900083963,
-      "grad_norm": 2.1015372276306152,
+      "grad_norm": 1.891290307044983,
       "learning_rate": 5.5600403632694254e-05,
-      "loss": 6.2098,
+      "loss": 6.1373,
       "step": 32000
     },
     {
       "epoch": 13.64399664147775,
-      "grad_norm": 2.079190969467163,
+      "grad_norm": 1.9195438623428345,
       "learning_rate": 5.4591321897073666e-05,
-      "loss": 6.2202,
+      "loss": 6.1417,
       "step": 32500
     },
     {
       "epoch": 13.853904282115868,
-      "grad_norm": 2.539520502090454,
+      "grad_norm": 2.3004908561706543,
       "learning_rate": 5.358224016145308e-05,
-      "loss": 6.2203,
+      "loss": 6.1431,
       "step": 33000
     },
     {
       "epoch": 14.063811922753988,
-      "grad_norm": 3.4186887741088867,
+      "grad_norm": 2.5193397998809814,
       "learning_rate": 5.257315842583249e-05,
-      "loss": 6.2066,
+      "loss": 6.131,
       "step": 33500
     },
     {
       "epoch": 14.273719563392108,
-      "grad_norm": 1.6018637418746948,
+      "grad_norm": 2.3650710582733154,
       "learning_rate": 5.15640766902119e-05,
-      "loss": 6.1873,
+      "loss": 6.117,
       "step": 34000
     },
     {
       "epoch": 14.483627204030226,
-      "grad_norm": 1.6987675428390503,
+      "grad_norm": 2.837047576904297,
       "learning_rate": 5.055499495459133e-05,
-      "loss": 6.1895,
+      "loss": 6.116,
       "step": 34500
     },
     {
       "epoch": 14.693534844668346,
-      "grad_norm": 2.854663848876953,
+      "grad_norm": 1.7063215970993042,
       "learning_rate": 4.954591321897074e-05,
-      "loss": 6.1907,
+      "loss": 6.116,
       "step": 35000
     },
     {
       "epoch": 14.903442485306465,
-      "grad_norm": 3.046041965484619,
+      "grad_norm": 1.8279320001602173,
       "learning_rate": 4.853683148335015e-05,
-      "loss": 6.1922,
+      "loss": 6.1207,
       "step": 35500
     },
     {
       "epoch": 15.113350125944585,
-      "grad_norm": 2.113647699356079,
+      "grad_norm": 1.15080988407135,
       "learning_rate": 4.7527749747729566e-05,
-      "loss": 6.1726,
+      "loss": 6.1035,
       "step": 36000
     },
     {
       "epoch": 15.323257766582703,
-      "grad_norm": 1.7852251529693604,
+      "grad_norm": 2.4123611450195312,
       "learning_rate": 4.6518668012108985e-05,
-      "loss": 6.1672,
+      "loss": 6.1014,
       "step": 36500
     },
     {
       "epoch": 15.533165407220823,
-      "grad_norm": 3.86224627494812,
+      "grad_norm": 3.021183490753174,
       "learning_rate": 4.55095862764884e-05,
-      "loss": 6.1663,
+      "loss": 6.0981,
       "step": 37000
     },
     {
       "epoch": 15.743073047858942,
-      "grad_norm": 2.544255495071411,
+      "grad_norm": 3.244154930114746,
       "learning_rate": 4.450050454086781e-05,
-      "loss": 6.1688,
+      "loss": 6.0933,
       "step": 37500
     },
     {
       "epoch": 15.95298068849706,
-      "grad_norm": 3.937136173248291,
+      "grad_norm": 2.689694881439209,
       "learning_rate": 4.349142280524723e-05,
-      "loss": 6.1671,
+      "loss": 6.0943,
       "step": 38000
     },
     {
       "epoch": 16.162888329135182,
-      "grad_norm": 2.6473629474639893,
+      "grad_norm": 1.9141645431518555,
       "learning_rate": 4.248234106962664e-05,
-      "loss": 6.1481,
+      "loss": 6.0836,
       "step": 38500
     },
     {
       "epoch": 16.3727959697733,
-      "grad_norm": 1.8369827270507812,
+      "grad_norm": 2.8854215145111084,
       "learning_rate": 4.147325933400605e-05,
-      "loss": 6.139,
+      "loss": 6.0775,
       "step": 39000
     },
     {
       "epoch": 16.582703610411418,
-      "grad_norm": 2.922276496887207,
+      "grad_norm": 4.890077590942383,
       "learning_rate": 4.046417759838547e-05,
-      "loss": 6.1462,
+      "loss": 6.0798,
       "step": 39500
     },
     {
       "epoch": 16.79261125104954,
-      "grad_norm": 2.60176157951355,
+      "grad_norm": 2.3635575771331787,
       "learning_rate": 3.9455095862764884e-05,
-      "loss": 6.1349,
+      "loss": 6.0706,
       "step": 40000
     },
     {
       "epoch": 16.79261125104954,
-      "eval_accuracy": 0.15209711786475272,
-      "eval_loss": 7.011668682098389,
-      "eval_runtime": 0.6729,
-      "eval_samples_per_second": 711.819,
-      "eval_steps_per_second": 7.43,
+      "eval_accuracy": 0.1513962094874958,
+      "eval_loss": 6.6304779052734375,
+      "eval_runtime": 0.6596,
+      "eval_samples_per_second": 726.182,
+      "eval_steps_per_second": 7.58,
       "step": 40000
     },
     {
       "epoch": 17.002518891687657,
-      "grad_norm": 2.5567150115966797,
+      "grad_norm": 1.3440437316894531,
       "learning_rate": 3.84460141271443e-05,
-      "loss": 6.1428,
+      "loss": 6.076,
       "step": 40500
     },
     {
       "epoch": 17.212426532325775,
-      "grad_norm": 1.887994408607483,
+      "grad_norm": 1.7824230194091797,
       "learning_rate": 3.7436932391523715e-05,
-      "loss": 6.1187,
+      "loss": 6.0612,
       "step": 41000
     },
     {
       "epoch": 17.422334172963897,
-      "grad_norm": 3.7771365642547607,
+      "grad_norm": 3.6087536811828613,
       "learning_rate": 3.642785065590313e-05,
-      "loss": 6.1212,
+      "loss": 6.0588,
       "step": 41500
     },
     {
       "epoch": 17.632241813602015,
-      "grad_norm": 2.04042911529541,
+      "grad_norm": 2.307131767272949,
       "learning_rate": 3.541876892028255e-05,
-      "loss": 6.1231,
+      "loss": 6.061,
       "step": 42000
     },
     {
       "epoch": 17.842149454240133,
-      "grad_norm": 1.855548620223999,
+      "grad_norm": 1.3711094856262207,
       "learning_rate": 3.440968718466196e-05,
-      "loss": 6.1239,
+      "loss": 6.0569,
       "step": 42500
     },
     {
       "epoch": 18.052057094878254,
-      "grad_norm": 3.5681562423706055,
+      "grad_norm": 1.2603286504745483,
       "learning_rate": 3.340060544904137e-05,
-      "loss": 6.115,
+      "loss": 6.0503,
       "step": 43000
     },
     {
       "epoch": 18.261964735516372,
-      "grad_norm": 3.338836431503296,
+      "grad_norm": 1.4356471300125122,
       "learning_rate": 3.239152371342079e-05,
-      "loss": 6.1032,
+      "loss": 6.0412,
       "step": 43500
     },
     {
       "epoch": 18.471872376154494,
-      "grad_norm": 3.386713981628418,
+      "grad_norm": 3.093308925628662,
       "learning_rate": 3.13824419778002e-05,
-      "loss": 6.1031,
+      "loss": 6.0432,
       "step": 44000
     },
     {
       "epoch": 18.68178001679261,
-      "grad_norm": 2.1127493381500244,
+      "grad_norm": 2.986450433731079,
       "learning_rate": 3.0373360242179615e-05,
-      "loss": 6.1048,
+      "loss": 6.0436,
       "step": 44500
     },
     {
       "epoch": 18.89168765743073,
-      "grad_norm": 3.2893552780151367,
+      "grad_norm": 3.6752383708953857,
       "learning_rate": 2.9364278506559034e-05,
-      "loss": 6.1075,
+      "loss": 6.0456,
       "step": 45000
     },
     {
       "epoch": 19.10159529806885,
-      "grad_norm": 2.278205394744873,
+      "grad_norm": 2.0228183269500732,
       "learning_rate": 2.8355196770938446e-05,
-      "loss": 6.0984,
+      "loss": 6.0376,
       "step": 45500
     },
     {
       "epoch": 19.31150293870697,
-      "grad_norm": 1.848292589187622,
+      "grad_norm": 2.9569408893585205,
       "learning_rate": 2.734611503531786e-05,
-      "loss": 6.0839,
+      "loss": 6.0257,
       "step": 46000
     },
     {
       "epoch": 19.521410579345087,
-      "grad_norm": 1.2474650144577026,
+      "grad_norm": 1.9467380046844482,
       "learning_rate": 2.6337033299697278e-05,
-      "loss": 6.0884,
+      "loss": 6.0264,
       "step": 46500
     },
     {
       "epoch": 19.73131821998321,
-      "grad_norm": 2.065427303314209,
+      "grad_norm": 2.2577433586120605,
       "learning_rate": 2.532795156407669e-05,
-      "loss": 6.0908,
+      "loss": 6.0273,
       "step": 47000
     },
     {
       "epoch": 19.941225860621326,
-      "grad_norm": 2.256314754486084,
+      "grad_norm": 1.0765509605407715,
       "learning_rate": 2.4318869828456106e-05,
-      "loss": 6.0856,
+      "loss": 6.0226,
       "step": 47500
     },
     {
       "epoch": 20.151133501259444,
-      "grad_norm": 0.9675717949867249,
+      "grad_norm": 0.9119879603385925,
       "learning_rate": 2.330978809283552e-05,
-      "loss": 6.077,
+      "loss": 6.0192,
       "step": 48000
     },
     {
       "epoch": 20.361041141897566,
-      "grad_norm": 3.822497606277466,
+      "grad_norm": 3.756011724472046,
       "learning_rate": 2.2300706357214934e-05,
-      "loss": 6.0758,
+      "loss": 6.0165,
       "step": 48500
     },
     {
       "epoch": 20.570948782535684,
-      "grad_norm": 3.19783878326416,
+      "grad_norm": 2.2077417373657227,
       "learning_rate": 2.129162462159435e-05,
-      "loss": 6.0683,
+      "loss": 6.0134,
       "step": 49000
     },
     {
       "epoch": 20.7808564231738,
-      "grad_norm": 3.127821683883667,
+      "grad_norm": 0.7892888784408569,
       "learning_rate": 2.0282542885973765e-05,
-      "loss": 6.0698,
+      "loss": 6.0117,
       "step": 49500
     },
     {
       "epoch": 20.990764063811923,
-      "grad_norm": 2.9081919193267822,
+      "grad_norm": 2.5961363315582275,
       "learning_rate": 1.927346115035318e-05,
-      "loss": 6.0723,
+      "loss": 6.0123,
       "step": 50000
     },
     {
       "epoch": 20.990764063811923,
-      "eval_accuracy": 0.1515126026859562,
-      "eval_loss": 7.073210716247559,
-      "eval_runtime": 0.66,
-      "eval_samples_per_second": 725.714,
-      "eval_steps_per_second": 7.575,
+      "eval_accuracy": 0.1510920458687302,
+      "eval_loss": 6.707541465759277,
+      "eval_runtime": 0.6579,
+      "eval_samples_per_second": 728.118,
+      "eval_steps_per_second": 7.6,
       "step": 50000
     },
     {
       "epoch": 21.20067170445004,
-      "grad_norm": 2.8496124744415283,
+      "grad_norm": 3.521597385406494,
       "learning_rate": 1.8264379414732593e-05,
-      "loss": 6.0612,
+      "loss": 6.001,
       "step": 50500
     },
     {
       "epoch": 21.410579345088163,
-      "grad_norm": 3.7069056034088135,
+      "grad_norm": 1.4102064371109009,
       "learning_rate": 1.725529767911201e-05,
-      "loss": 6.0624,
+      "loss": 6.0054,
       "step": 51000
     },
     {
       "epoch": 21.62048698572628,
-      "grad_norm": 1.678023338317871,
+      "grad_norm": 1.0189828872680664,
       "learning_rate": 1.6246215943491424e-05,
-      "loss": 6.0617,
+      "loss": 6.0023,
       "step": 51500
     },
     {
       "epoch": 21.8303946263644,
-      "grad_norm": 2.310828924179077,
+      "grad_norm": 3.020364999771118,
       "learning_rate": 1.5237134207870837e-05,
-      "loss": 6.0612,
+      "loss": 6.0019,
       "step": 52000
     },
     {
       "epoch": 22.04030226700252,
-      "grad_norm": 1.4076191186904907,
+      "grad_norm": 3.071946859359741,
       "learning_rate": 1.4228052472250252e-05,
-      "loss": 6.0608,
+      "loss": 6.0005,
       "step": 52500
     },
     {
       "epoch": 22.250209907640638,
-      "grad_norm": 0.9361013770103455,
+      "grad_norm": 2.5311036109924316,
       "learning_rate": 1.3218970736629668e-05,
-      "loss": 6.0452,
+      "loss": 5.9933,
       "step": 53000
     },
     {
       "epoch": 22.460117548278756,
-      "grad_norm": 1.3987303972244263,
+      "grad_norm": 1.2899010181427002,
       "learning_rate": 1.2209889001009082e-05,
-      "loss": 6.0492,
+      "loss": 5.9943,
       "step": 53500
     },
     {
       "epoch": 22.670025188916878,
-      "grad_norm": 0.8329377770423889,
+      "grad_norm": 2.720703363418579,
       "learning_rate": 1.1200807265388496e-05,
-      "loss": 6.0528,
+      "loss": 5.9969,
       "step": 54000
     },
     {
       "epoch": 22.879932829554996,
-      "grad_norm": 2.0292911529541016,
+      "grad_norm": 1.5236635208129883,
       "learning_rate": 1.0191725529767912e-05,
-      "loss": 6.0474,
+      "loss": 5.9938,
       "step": 54500
     },
     {
       "epoch": 23.089840470193113,
-      "grad_norm": 2.003349781036377,
+      "grad_norm": 0.9731554985046387,
       "learning_rate": 9.182643794147326e-06,
-      "loss": 6.044,
+      "loss": 5.9893,
       "step": 55000
     },
     {
       "epoch": 23.299748110831235,
-      "grad_norm": 0.8627718091011047,
+      "grad_norm": 0.8480926752090454,
       "learning_rate": 8.173562058526741e-06,
-      "loss": 6.0394,
+      "loss": 5.9875,
       "step": 55500
     },
     {
       "epoch": 23.509655751469353,
-      "grad_norm": 2.9067883491516113,
+      "grad_norm": 2.845705032348633,
       "learning_rate": 7.164480322906156e-06,
-      "loss": 6.0447,
+      "loss": 5.9885,
       "step": 56000
     },
     {
       "epoch": 23.719563392107474,
-      "grad_norm": 3.6389567852020264,
+      "grad_norm": 0.7594144940376282,
       "learning_rate": 6.155398587285571e-06,
-      "loss": 6.0377,
+      "loss": 5.9807,
       "step": 56500
     },
     {
       "epoch": 23.929471032745592,
-      "grad_norm": 1.3027358055114746,
+      "grad_norm": 0.811726987361908,
       "learning_rate": 5.146316851664986e-06,
-      "loss": 6.0396,
+      "loss": 5.9865,
       "step": 57000
     },
     {
       "epoch": 24.13937867338371,
-      "grad_norm": 1.0353599786758423,
+      "grad_norm": 1.1903959512710571,
       "learning_rate": 4.1372351160444e-06,
-      "loss": 6.035,
+      "loss": 5.9809,
       "step": 57500
     },
     {
       "epoch": 24.349286314021832,
-      "grad_norm": 1.6786377429962158,
+      "grad_norm": 2.064216375350952,
       "learning_rate": 3.1281533804238145e-06,
-      "loss": 6.0382,
+      "loss": 5.9826,
       "step": 58000
     },
     {
       "epoch": 24.55919395465995,
-      "grad_norm": 2.3209586143493652,
+      "grad_norm": 0.8064579367637634,
       "learning_rate": 2.1190716448032293e-06,
-      "loss": 6.0339,
+      "loss": 5.9788,
       "step": 58500
     },
     {
       "epoch": 24.769101595298068,
-      "grad_norm": 2.0108327865600586,
+      "grad_norm": 0.7879181504249573,
       "learning_rate": 1.109989909182644e-06,
-      "loss": 6.0322,
+      "loss": 5.9791,
       "step": 59000
     },
     {
       "epoch": 24.97900923593619,
-      "grad_norm": 1.021445393562317,
+      "grad_norm": 1.4866392612457275,
       "learning_rate": 1.0090817356205853e-07,
-      "loss": 6.0311,
+      "loss": 5.9796,
       "step": 59500
     },
     {
       "epoch": 25.0,
       "step": 59550,
       "total_flos": 3.051148107142656e+17,
-      "train_loss": 6.524886467434797,
-      "train_runtime": 11610.4539,
-      "train_samples_per_second": 492.313,
-      "train_steps_per_second": 5.129
+      "train_loss": 6.453962587345357,
+      "train_runtime": 11281.7551,
+      "train_samples_per_second": 506.657,
+      "train_steps_per_second": 5.278
     }
   ],
   "logging_steps": 500,