End of training

README.md

@@ -1,11 +1,23 @@
 ---
 tags:
 - generated_from_trainer
+datasets:
+- Hartunka/processed_wikitext-103-raw-v1-rand-5_v2
 metrics:
 - accuracy
 model-index:
 - name: tiny_bert_rand_5_v2
-  results: []
+  results:
+  - task:
+      name: Masked Language Modeling
+      type: fill-mask
+    dataset:
+      name: Hartunka/processed_wikitext-103-raw-v1-rand-5_v2
+      type: Hartunka/processed_wikitext-103-raw-v1-rand-5_v2
+    metrics:
+    - name: Accuracy
+      type: accuracy
+      value: 0.15245521839859863
 ---
 
 <!-- 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_5_v2
 
-This model is a fine-tuned version of [](https://huggingface.co/) on an unknown dataset.
+This model is a fine-tuned version of [](https://huggingface.co/) on the Hartunka/processed_wikitext-103-raw-v1-rand-5_v2 dataset.
 It achieves the following results on the evaluation set:
-- Loss: 8.9658
-- Accuracy: 0.1511
+- Loss: 7.7873
+- Accuracy: 0.1525
 
 ## Model description
 

all_results.json

@@ -1,16 +1,16 @@
 {
     "epoch": 25.0,
-    "eval_accuracy": 0.1545177148669266,
-    "eval_loss": 7.7626118659973145,
-    "eval_runtime": 0.7079,
+    "eval_accuracy": 0.15245521839859863,
+    "eval_loss": 7.787292003631592,
+    "eval_runtime": 0.7336,
     "eval_samples": 479,
-    "eval_samples_per_second": 676.604,
-    "eval_steps_per_second": 7.063,
-    "perplexity": 2351.037187136085,
+    "eval_samples_per_second": 652.912,
+    "eval_steps_per_second": 6.815,
+    "perplexity": 2409.783054689172,
     "total_flos": 3.049796905526016e+17,
-    "train_loss": 7.280221478184364,
-    "train_runtime": 11794.2322,
+    "train_loss": 7.265635145365141,
+    "train_runtime": 11576.6001,
     "train_samples": 228639,
-    "train_samples_per_second": 484.642,
-    "train_steps_per_second": 5.049
+    "train_samples_per_second": 493.752,
+    "train_steps_per_second": 5.144
 }

eval_results.json

@@ -1,10 +1,10 @@
 {
     "epoch": 25.0,
-    "eval_accuracy": 0.1545177148669266,
-    "eval_loss": 7.7626118659973145,
-    "eval_runtime": 0.7079,
+    "eval_accuracy": 0.15245521839859863,
+    "eval_loss": 7.787292003631592,
+    "eval_runtime": 0.7336,
     "eval_samples": 479,
-    "eval_samples_per_second": 676.604,
-    "eval_steps_per_second": 7.063,
-    "perplexity": 2351.037187136085
+    "eval_samples_per_second": 652.912,
+    "eval_steps_per_second": 6.815,
+    "perplexity": 2409.783054689172
 }

train_results.json

@@ -1,9 +1,9 @@
 {
     "epoch": 25.0,
     "total_flos": 3.049796905526016e+17,
-    "train_loss": 7.280221478184364,
-    "train_runtime": 11794.2322,
+    "train_loss": 7.265635145365141,
+    "train_runtime": 11576.6001,
     "train_samples": 228639,
-    "train_samples_per_second": 484.642,
-    "train_steps_per_second": 5.049
+    "train_samples_per_second": 493.752,
+    "train_steps_per_second": 5.144
 }

trainer_state.json

@@ -1,6 +1,6 @@
 {
-  "best_metric": 7.75156307220459,
-  "best_model_checkpoint": "tiny_bert_rand_5_v2/checkpoint-20000",
+  "best_metric": 7.779261589050293,
+  "best_model_checkpoint": "tiny_bert_rand_5_v2/checkpoint-10000",
   "epoch": 25.0,
   "eval_steps": 10000,
   "global_step": 59550,
@@ -10,890 +10,890 @@
   "log_history": [
     {
       "epoch": 0.20990764063811923,
-      "grad_norm": 2.2963080406188965,
+      "grad_norm": 2.5618419647216797,
       "learning_rate": 5e-06,
       "loss": 11.3941,
       "step": 500
     },
     {
       "epoch": 0.41981528127623846,
-      "grad_norm": 2.127878189086914,
+      "grad_norm": 1.3906711339950562,
       "learning_rate": 1e-05,
-      "loss": 9.8095,
+      "loss": 9.8068,
       "step": 1000
     },
     {
       "epoch": 0.6297229219143576,
-      "grad_norm": 1.2294187545776367,
+      "grad_norm": 1.7807637453079224,
       "learning_rate": 1.5e-05,
-      "loss": 8.8105,
+      "loss": 8.8075,
       "step": 1500
     },
     {
       "epoch": 0.8396305625524769,
-      "grad_norm": 1.2945270538330078,
+      "grad_norm": 2.1833412647247314,
       "learning_rate": 2e-05,
-      "loss": 8.6093,
+      "loss": 8.6113,
       "step": 2000
     },
     {
       "epoch": 1.0495382031905962,
-      "grad_norm": 0.8322573304176331,
+      "grad_norm": 2.479278326034546,
       "learning_rate": 2.5e-05,
-      "loss": 8.4914,
+      "loss": 8.4919,
       "step": 2500
     },
     {
       "epoch": 1.2594458438287153,
-      "grad_norm": 1.406333088874817,
+      "grad_norm": 1.3648505210876465,
       "learning_rate": 3e-05,
-      "loss": 8.3892,
+      "loss": 8.3896,
       "step": 3000
     },
     {
       "epoch": 1.4693534844668346,
-      "grad_norm": 1.4099137783050537,
+      "grad_norm": 1.6506822109222412,
       "learning_rate": 3.5e-05,
-      "loss": 8.3033,
+      "loss": 8.3041,
       "step": 3500
     },
     {
       "epoch": 1.6792611251049538,
-      "grad_norm": 1.617598056793213,
+      "grad_norm": 1.3045011758804321,
       "learning_rate": 4e-05,
-      "loss": 8.2222,
+      "loss": 8.2242,
       "step": 4000
     },
     {
       "epoch": 1.8891687657430731,
-      "grad_norm": 1.0762231349945068,
+      "grad_norm": 1.553267240524292,
       "learning_rate": 4.5e-05,
-      "loss": 8.1478,
+      "loss": 8.1502,
       "step": 4500
     },
     {
       "epoch": 2.0990764063811924,
-      "grad_norm": 1.354127287864685,
+      "grad_norm": 1.7191903591156006,
       "learning_rate": 5e-05,
-      "loss": 8.0885,
+      "loss": 8.0897,
       "step": 5000
     },
     {
       "epoch": 2.3089840470193117,
-      "grad_norm": 1.1586333513259888,
+      "grad_norm": 2.1493537425994873,
       "learning_rate": 5.500000000000001e-05,
-      "loss": 8.0462,
+      "loss": 8.0477,
       "step": 5500
     },
     {
       "epoch": 2.5188916876574305,
-      "grad_norm": 1.2256927490234375,
+      "grad_norm": 1.5696853399276733,
       "learning_rate": 6e-05,
-      "loss": 8.0097,
+      "loss": 8.0114,
       "step": 6000
     },
     {
       "epoch": 2.72879932829555,
-      "grad_norm": 1.3555954694747925,
+      "grad_norm": 1.5219547748565674,
       "learning_rate": 6.500000000000001e-05,
-      "loss": 7.9773,
+      "loss": 7.9791,
       "step": 6500
     },
     {
       "epoch": 2.938706968933669,
-      "grad_norm": 1.890107274055481,
+      "grad_norm": 1.3124215602874756,
       "learning_rate": 7e-05,
-      "loss": 7.9489,
+      "loss": 7.9502,
       "step": 7000
     },
     {
       "epoch": 3.1486146095717884,
-      "grad_norm": 2.2107865810394287,
+      "grad_norm": 1.9786266088485718,
       "learning_rate": 7.500000000000001e-05,
-      "loss": 7.9152,
+      "loss": 7.9157,
       "step": 7500
     },
     {
       "epoch": 3.3585222502099077,
-      "grad_norm": 1.768619179725647,
+      "grad_norm": 1.4063630104064941,
       "learning_rate": 8e-05,
-      "loss": 7.888,
+      "loss": 7.8895,
       "step": 8000
     },
     {
       "epoch": 3.568429890848027,
-      "grad_norm": 1.212744116783142,
+      "grad_norm": 1.1584075689315796,
       "learning_rate": 8.5e-05,
-      "loss": 7.873,
+      "loss": 7.8752,
       "step": 8500
     },
     {
       "epoch": 3.7783375314861463,
-      "grad_norm": 1.150292992591858,
+      "grad_norm": 1.1364918947219849,
       "learning_rate": 9e-05,
-      "loss": 7.847,
+      "loss": 7.8471,
       "step": 9000
     },
     {
       "epoch": 3.988245172124265,
-      "grad_norm": 2.340731143951416,
+      "grad_norm": 2.191779375076294,
       "learning_rate": 9.5e-05,
-      "loss": 7.8276,
+      "loss": 7.8304,
       "step": 9500
     },
     {
       "epoch": 4.198152812762385,
-      "grad_norm": 2.221745729446411,
+      "grad_norm": 1.9462758302688599,
       "learning_rate": 0.0001,
-      "loss": 7.7647,
+      "loss": 7.7678,
       "step": 10000
     },
     {
       "epoch": 4.198152812762385,
-      "eval_accuracy": 0.15072762180361168,
-      "eval_loss": 7.783298492431641,
-      "eval_runtime": 0.7227,
-      "eval_samples_per_second": 662.752,
-      "eval_steps_per_second": 6.918,
+      "eval_accuracy": 0.15106152091043157,
+      "eval_loss": 7.779261589050293,
+      "eval_runtime": 0.7554,
+      "eval_samples_per_second": 634.074,
+      "eval_steps_per_second": 6.619,
       "step": 10000
     },
     {
       "epoch": 4.408060453400504,
-      "grad_norm": 1.6444823741912842,
+      "grad_norm": 1.7463555335998535,
       "learning_rate": 9.899091826437942e-05,
-      "loss": 7.7616,
+      "loss": 7.7624,
       "step": 10500
     },
     {
       "epoch": 4.617968094038623,
-      "grad_norm": 2.4590201377868652,
+      "grad_norm": 1.4476852416992188,
       "learning_rate": 9.798183652875883e-05,
-      "loss": 7.7484,
+      "loss": 7.7465,
       "step": 11000
     },
     {
       "epoch": 4.827875734676742,
-      "grad_norm": 1.886684775352478,
+      "grad_norm": 1.9277458190917969,
       "learning_rate": 9.697275479313825e-05,
-      "loss": 7.7465,
+      "loss": 7.7467,
       "step": 11500
     },
     {
       "epoch": 5.037783375314861,
-      "grad_norm": 1.8157285451889038,
+      "grad_norm": 1.926012635231018,
       "learning_rate": 9.596367305751767e-05,
-      "loss": 7.7214,
+      "loss": 7.7233,
       "step": 12000
     },
     {
       "epoch": 5.247691015952981,
-      "grad_norm": 1.9011366367340088,
+      "grad_norm": 2.074347972869873,
       "learning_rate": 9.495459132189707e-05,
-      "loss": 7.6707,
+      "loss": 7.6669,
       "step": 12500
     },
     {
       "epoch": 5.4575986565911,
-      "grad_norm": 2.44659686088562,
+      "grad_norm": 1.6645373106002808,
       "learning_rate": 9.394550958627649e-05,
-      "loss": 7.6693,
+      "loss": 7.666,
       "step": 13000
     },
     {
       "epoch": 5.667506297229219,
-      "grad_norm": 1.3702588081359863,
+      "grad_norm": 1.4428155422210693,
       "learning_rate": 9.293642785065591e-05,
-      "loss": 7.6702,
+      "loss": 7.6709,
       "step": 13500
     },
     {
       "epoch": 5.877413937867338,
-      "grad_norm": 1.5496770143508911,
+      "grad_norm": 1.8915469646453857,
       "learning_rate": 9.192734611503532e-05,
-      "loss": 7.6663,
+      "loss": 7.6643,
       "step": 14000
     },
     {
       "epoch": 6.087321578505458,
-      "grad_norm": 2.4334986209869385,
+      "grad_norm": 2.146442413330078,
       "learning_rate": 9.091826437941474e-05,
-      "loss": 7.6372,
+      "loss": 7.636,
       "step": 14500
     },
     {
       "epoch": 6.297229219143577,
-      "grad_norm": 1.7889429330825806,
+      "grad_norm": 1.9808622598648071,
       "learning_rate": 8.990918264379415e-05,
-      "loss": 7.6113,
+      "loss": 7.6095,
       "step": 15000
     },
     {
       "epoch": 6.507136859781696,
-      "grad_norm": 1.9616079330444336,
+      "grad_norm": 2.204853057861328,
       "learning_rate": 8.890010090817356e-05,
-      "loss": 7.6099,
+      "loss": 7.6076,
       "step": 15500
     },
     {
       "epoch": 6.717044500419815,
-      "grad_norm": 1.6861013174057007,
+      "grad_norm": 2.3092031478881836,
       "learning_rate": 8.789101917255298e-05,
-      "loss": 7.611,
+      "loss": 7.6085,
       "step": 16000
     },
     {
       "epoch": 6.926952141057934,
-      "grad_norm": 2.1229841709136963,
+      "grad_norm": 2.117516279220581,
       "learning_rate": 8.68819374369324e-05,
-      "loss": 7.6095,
+      "loss": 7.6055,
       "step": 16500
     },
     {
       "epoch": 7.136859781696054,
-      "grad_norm": 2.1880974769592285,
+      "grad_norm": 2.7730908393859863,
       "learning_rate": 8.58728557013118e-05,
-      "loss": 7.5725,
+      "loss": 7.5702,
       "step": 17000
     },
     {
       "epoch": 7.346767422334173,
-      "grad_norm": 1.7154184579849243,
+      "grad_norm": 1.9820139408111572,
       "learning_rate": 8.486377396569122e-05,
-      "loss": 7.5551,
+      "loss": 7.5519,
       "step": 17500
     },
     {
       "epoch": 7.5566750629722925,
-      "grad_norm": 2.275742530822754,
+      "grad_norm": 2.9039790630340576,
       "learning_rate": 8.385469223007064e-05,
-      "loss": 7.5621,
+      "loss": 7.5594,
       "step": 18000
     },
     {
       "epoch": 7.766582703610411,
-      "grad_norm": 1.6617145538330078,
+      "grad_norm": 1.491622805595398,
       "learning_rate": 8.284561049445006e-05,
-      "loss": 7.565,
+      "loss": 7.5605,
       "step": 18500
     },
     {
       "epoch": 7.97649034424853,
-      "grad_norm": 2.4727964401245117,
+      "grad_norm": 2.0691475868225098,
       "learning_rate": 8.183652875882947e-05,
-      "loss": 7.5637,
+      "loss": 7.5599,
       "step": 19000
     },
     {
       "epoch": 8.18639798488665,
-      "grad_norm": 1.8340699672698975,
+      "grad_norm": 1.8290640115737915,
       "learning_rate": 8.082744702320889e-05,
-      "loss": 7.517,
+      "loss": 7.5146,
       "step": 19500
     },
     {
       "epoch": 8.39630562552477,
-      "grad_norm": 3.5926151275634766,
+      "grad_norm": 2.2066845893859863,
       "learning_rate": 7.98183652875883e-05,
-      "loss": 7.5213,
+      "loss": 7.5134,
       "step": 20000
     },
     {
       "epoch": 8.39630562552477,
-      "eval_accuracy": 0.15210864502371,
-      "eval_loss": 7.75156307220459,
-      "eval_runtime": 0.6982,
-      "eval_samples_per_second": 686.089,
-      "eval_steps_per_second": 7.162,
+      "eval_accuracy": 0.152697887146104,
+      "eval_loss": 7.784549713134766,
+      "eval_runtime": 0.6772,
+      "eval_samples_per_second": 707.317,
+      "eval_steps_per_second": 7.383,
       "step": 20000
     },
     {
       "epoch": 8.606213266162888,
-      "grad_norm": 1.9679731130599976,
+      "grad_norm": 1.7037521600723267,
       "learning_rate": 7.880928355196771e-05,
-      "loss": 7.5134,
+      "loss": 7.509,
       "step": 20500
     },
     {
       "epoch": 8.816120906801007,
-      "grad_norm": 1.7791990041732788,
+      "grad_norm": 1.9685962200164795,
       "learning_rate": 7.780020181634713e-05,
-      "loss": 7.5251,
+      "loss": 7.5195,
       "step": 21000
     },
     {
       "epoch": 9.026028547439127,
-      "grad_norm": 2.018057346343994,
+      "grad_norm": 1.6262836456298828,
       "learning_rate": 7.679112008072655e-05,
-      "loss": 7.5122,
+      "loss": 7.5054,
       "step": 21500
     },
     {
       "epoch": 9.235936188077247,
-      "grad_norm": 2.222806453704834,
+      "grad_norm": 2.75607967376709,
       "learning_rate": 7.578203834510595e-05,
-      "loss": 7.4671,
+      "loss": 7.4567,
       "step": 22000
     },
     {
       "epoch": 9.445843828715365,
-      "grad_norm": 2.0874717235565186,
+      "grad_norm": 1.7600328922271729,
       "learning_rate": 7.477295660948537e-05,
-      "loss": 7.4708,
+      "loss": 7.4624,
       "step": 22500
     },
     {
       "epoch": 9.655751469353484,
-      "grad_norm": 2.1058754920959473,
+      "grad_norm": 1.7434253692626953,
       "learning_rate": 7.376387487386479e-05,
-      "loss": 7.4767,
+      "loss": 7.47,
       "step": 23000
     },
     {
       "epoch": 9.865659109991604,
-      "grad_norm": 1.7424660921096802,
+      "grad_norm": 2.3567728996276855,
       "learning_rate": 7.27547931382442e-05,
-      "loss": 7.473,
+      "loss": 7.4656,
       "step": 23500
     },
     {
       "epoch": 10.075566750629722,
-      "grad_norm": 1.92734956741333,
+      "grad_norm": 2.2403724193573,
       "learning_rate": 7.174571140262362e-05,
-      "loss": 7.4439,
+      "loss": 7.4362,
       "step": 24000
     },
     {
       "epoch": 10.285474391267842,
-      "grad_norm": 1.9836821556091309,
+      "grad_norm": 2.0520102977752686,
       "learning_rate": 7.073662966700304e-05,
-      "loss": 7.4124,
+      "loss": 7.3998,
       "step": 24500
     },
     {
       "epoch": 10.495382031905962,
-      "grad_norm": 2.5103793144226074,
+      "grad_norm": 2.4070234298706055,
       "learning_rate": 6.972754793138244e-05,
-      "loss": 7.4117,
+      "loss": 7.4008,
       "step": 25000
     },
     {
       "epoch": 10.705289672544081,
-      "grad_norm": 2.2446186542510986,
+      "grad_norm": 2.0839731693267822,
       "learning_rate": 6.871846619576186e-05,
-      "loss": 7.4205,
+      "loss": 7.4078,
       "step": 25500
     },
     {
       "epoch": 10.9151973131822,
-      "grad_norm": 2.133932113647461,
+      "grad_norm": 2.588850736618042,
       "learning_rate": 6.770938446014128e-05,
-      "loss": 7.4262,
+      "loss": 7.4176,
       "step": 26000
     },
     {
       "epoch": 11.125104953820319,
-      "grad_norm": 2.5607736110687256,
+      "grad_norm": 2.3594658374786377,
       "learning_rate": 6.670030272452068e-05,
-      "loss": 7.3748,
+      "loss": 7.3569,
       "step": 26500
     },
     {
       "epoch": 11.335012594458439,
-      "grad_norm": 2.75422739982605,
+      "grad_norm": 2.502192497253418,
       "learning_rate": 6.56912209889001e-05,
-      "loss": 7.3439,
+      "loss": 7.3333,
       "step": 27000
     },
     {
       "epoch": 11.544920235096557,
-      "grad_norm": 2.7491469383239746,
+      "grad_norm": 2.7358298301696777,
       "learning_rate": 6.468213925327952e-05,
-      "loss": 7.3546,
+      "loss": 7.336,
       "step": 27500
     },
     {
       "epoch": 11.754827875734676,
-      "grad_norm": 2.96463680267334,
+      "grad_norm": 4.04531192779541,
       "learning_rate": 6.367305751765893e-05,
-      "loss": 7.3584,
+      "loss": 7.3417,
       "step": 28000
     },
     {
       "epoch": 11.964735516372796,
-      "grad_norm": 2.1535544395446777,
+      "grad_norm": 2.9066739082336426,
       "learning_rate": 6.266397578203835e-05,
-      "loss": 7.363,
+      "loss": 7.3503,
       "step": 28500
     },
     {
       "epoch": 12.174643157010916,
-      "grad_norm": 3.3787529468536377,
+      "grad_norm": 2.6150119304656982,
       "learning_rate": 6.165489404641777e-05,
-      "loss": 7.2749,
+      "loss": 7.2586,
       "step": 29000
     },
     {
       "epoch": 12.384550797649034,
-      "grad_norm": 1.9232056140899658,
+      "grad_norm": 2.6906931400299072,
       "learning_rate": 6.064581231079718e-05,
-      "loss": 7.2695,
+      "loss": 7.2467,
       "step": 29500
     },
     {
       "epoch": 12.594458438287154,
-      "grad_norm": 2.4514613151550293,
+      "grad_norm": 2.852776527404785,
       "learning_rate": 5.963673057517659e-05,
-      "loss": 7.2819,
+      "loss": 7.2592,
       "step": 30000
     },
     {
       "epoch": 12.594458438287154,
-      "eval_accuracy": 0.1536782606262417,
-      "eval_loss": 7.925209999084473,
-      "eval_runtime": 0.7071,
-      "eval_samples_per_second": 677.454,
-      "eval_steps_per_second": 7.072,
+      "eval_accuracy": 0.15412597587934074,
+      "eval_loss": 7.9266839027404785,
+      "eval_runtime": 0.682,
+      "eval_samples_per_second": 702.365,
+      "eval_steps_per_second": 7.332,
       "step": 30000
     },
     {
       "epoch": 12.804366078925273,
-      "grad_norm": 2.9329562187194824,
+      "grad_norm": 3.1905484199523926,
       "learning_rate": 5.8627648839556e-05,
-      "loss": 7.2829,
+      "loss": 7.2682,
       "step": 30500
     },
     {
       "epoch": 13.014273719563391,
-      "grad_norm": 3.388092279434204,
+      "grad_norm": 2.745398759841919,
       "learning_rate": 5.7618567103935416e-05,
-      "loss": 7.2814,
+      "loss": 7.268,
       "step": 31000
     },
     {
       "epoch": 13.224181360201511,
-      "grad_norm": 3.280879497528076,
+      "grad_norm": 3.094973564147949,
       "learning_rate": 5.660948536831484e-05,
-      "loss": 7.1661,
+      "loss": 7.1416,
       "step": 31500
     },
     {
       "epoch": 13.43408900083963,
-      "grad_norm": 4.086568355560303,
+      "grad_norm": 3.9904205799102783,
       "learning_rate": 5.5600403632694254e-05,
-      "loss": 7.1789,
+      "loss": 7.1558,
       "step": 32000
     },
     {
       "epoch": 13.64399664147775,
-      "grad_norm": 3.866208553314209,
+      "grad_norm": 3.377580165863037,
       "learning_rate": 5.4591321897073666e-05,
-      "loss": 7.1964,
+      "loss": 7.1784,
       "step": 32500
     },
     {
       "epoch": 13.853904282115868,
-      "grad_norm": 4.327902317047119,
+      "grad_norm": 3.7697737216949463,
       "learning_rate": 5.358224016145308e-05,
-      "loss": 7.2045,
+      "loss": 7.1809,
       "step": 33000
     },
     {
       "epoch": 14.063811922753988,
-      "grad_norm": 3.7905776500701904,
+      "grad_norm": 4.175858020782471,
       "learning_rate": 5.257315842583249e-05,
-      "loss": 7.1609,
+      "loss": 7.1405,
       "step": 33500
     },
     {
       "epoch": 14.273719563392108,
-      "grad_norm": 3.391533613204956,
+      "grad_norm": 4.2780656814575195,
       "learning_rate": 5.15640766902119e-05,
-      "loss": 7.0648,
+      "loss": 7.0469,
       "step": 34000
     },
     {
       "epoch": 14.483627204030226,
-      "grad_norm": 3.62021541595459,
+      "grad_norm": 5.313665866851807,
       "learning_rate": 5.055499495459133e-05,
-      "loss": 7.0855,
+      "loss": 7.0634,
       "step": 34500
     },
     {
       "epoch": 14.693534844668346,
-      "grad_norm": 3.3568079471588135,
+      "grad_norm": 4.02442741394043,
       "learning_rate": 4.954591321897074e-05,
-      "loss": 7.0978,
+      "loss": 7.0702,
       "step": 35000
     },
     {
       "epoch": 14.903442485306465,
-      "grad_norm": 5.139899730682373,
+      "grad_norm": 4.080717086791992,
       "learning_rate": 4.853683148335015e-05,
-      "loss": 7.1073,
+      "loss": 7.0794,
       "step": 35500
     },
     {
       "epoch": 15.113350125944585,
-      "grad_norm": 4.743819236755371,
+      "grad_norm": 3.703134775161743,
       "learning_rate": 4.7527749747729566e-05,
-      "loss": 7.023,
+      "loss": 7.0012,
       "step": 36000
     },
     {
       "epoch": 15.323257766582703,
-      "grad_norm": 3.7411317825317383,
+      "grad_norm": 4.541247367858887,
       "learning_rate": 4.6518668012108985e-05,
-      "loss": 6.9746,
+      "loss": 6.9505,
       "step": 36500
     },
     {
       "epoch": 15.533165407220823,
-      "grad_norm": 4.443670749664307,
+      "grad_norm": 4.306236267089844,
       "learning_rate": 4.55095862764884e-05,
-      "loss": 6.9832,
+      "loss": 6.9656,
       "step": 37000
     },
     {
       "epoch": 15.743073047858942,
-      "grad_norm": 3.6280035972595215,
+      "grad_norm": 5.425760746002197,
       "learning_rate": 4.450050454086781e-05,
-      "loss": 7.0011,
+      "loss": 6.976,
       "step": 37500
     },
     {
       "epoch": 15.95298068849706,
-      "grad_norm": 5.191313743591309,
+      "grad_norm": 4.601401329040527,
       "learning_rate": 4.349142280524723e-05,
-      "loss": 7.0189,
+      "loss": 6.9847,
       "step": 38000
     },
     {
       "epoch": 16.162888329135182,
-      "grad_norm": 4.786304950714111,
+      "grad_norm": 4.355700492858887,
       "learning_rate": 4.248234106962664e-05,
-      "loss": 6.9044,
+      "loss": 6.8707,
       "step": 38500
     },
     {
       "epoch": 16.3727959697733,
-      "grad_norm": 6.1741623878479,
+      "grad_norm": 5.1258320808410645,
       "learning_rate": 4.147325933400605e-05,
-      "loss": 6.8875,
+      "loss": 6.8587,
       "step": 39000
     },
     {
       "epoch": 16.582703610411418,
-      "grad_norm": 5.1629862785339355,
+      "grad_norm": 5.246161937713623,
       "learning_rate": 4.046417759838547e-05,
-      "loss": 6.8963,
+      "loss": 6.8724,
       "step": 39500
     },
     {
       "epoch": 16.79261125104954,
-      "grad_norm": 4.76313591003418,
+      "grad_norm": 3.921492338180542,
       "learning_rate": 3.9455095862764884e-05,
-      "loss": 6.9106,
+      "loss": 6.8741,
       "step": 40000
     },
     {
       "epoch": 16.79261125104954,
-      "eval_accuracy": 0.15173264550857912,
-      "eval_loss": 8.4091157913208,
-      "eval_runtime": 0.7016,
-      "eval_samples_per_second": 682.738,
-      "eval_steps_per_second": 7.127,
+      "eval_accuracy": 0.15178871817875966,
+      "eval_loss": 8.307951927185059,
+      "eval_runtime": 0.6831,
+      "eval_samples_per_second": 701.25,
+      "eval_steps_per_second": 7.32,
       "step": 40000
     },
     {
       "epoch": 17.002518891687657,
-      "grad_norm": 4.705944061279297,
+      "grad_norm": 5.0183210372924805,
       "learning_rate": 3.84460141271443e-05,
-      "loss": 6.9156,
+      "loss": 6.8901,
       "step": 40500
     },
     {
       "epoch": 17.212426532325775,
-      "grad_norm": 6.631649971008301,
+      "grad_norm": 5.882933616638184,
       "learning_rate": 3.7436932391523715e-05,
-      "loss": 6.7843,
+      "loss": 6.7504,
       "step": 41000
     },
     {
       "epoch": 17.422334172963897,
-      "grad_norm": 6.722867488861084,
+      "grad_norm": 6.301322937011719,
       "learning_rate": 3.642785065590313e-05,
-      "loss": 6.7971,
+      "loss": 6.7699,
       "step": 41500
     },
     {
       "epoch": 17.632241813602015,
-      "grad_norm": 5.879517555236816,
+      "grad_norm": 5.363762855529785,
       "learning_rate": 3.541876892028255e-05,
-      "loss": 6.8151,
+      "loss": 6.7827,
       "step": 42000
     },
     {
       "epoch": 17.842149454240133,
-      "grad_norm": 5.177842617034912,
+      "grad_norm": 4.860353469848633,
       "learning_rate": 3.440968718466196e-05,
-      "loss": 6.8234,
+      "loss": 6.7947,
       "step": 42500
     },
     {
       "epoch": 18.052057094878254,
-      "grad_norm": 6.357394695281982,
+      "grad_norm": 6.5538201332092285,
       "learning_rate": 3.340060544904137e-05,
-      "loss": 6.7893,
+      "loss": 6.7655,
       "step": 43000
     },
     {
       "epoch": 18.261964735516372,
-      "grad_norm": 6.905867099761963,
+      "grad_norm": 6.815134525299072,
       "learning_rate": 3.239152371342079e-05,
-      "loss": 6.7071,
+      "loss": 6.6763,
       "step": 43500
     },
     {
       "epoch": 18.471872376154494,
-      "grad_norm": 5.510581970214844,
+      "grad_norm": 6.133576393127441,
       "learning_rate": 3.13824419778002e-05,
-      "loss": 6.7232,
+      "loss": 6.6939,
       "step": 44000
     },
     {
       "epoch": 18.68178001679261,
-      "grad_norm": 6.023885726928711,
+      "grad_norm": 5.66119909286499,
       "learning_rate": 3.0373360242179615e-05,
-      "loss": 6.7343,
+      "loss": 6.7095,
       "step": 44500
     },
     {
       "epoch": 18.89168765743073,
-      "grad_norm": 6.117109775543213,
+      "grad_norm": 4.890359401702881,
       "learning_rate": 2.9364278506559034e-05,
-      "loss": 6.7425,
+      "loss": 6.7215,
       "step": 45000
     },
     {
       "epoch": 19.10159529806885,
-      "grad_norm": 6.974370002746582,
+      "grad_norm": 5.522354602813721,
       "learning_rate": 2.8355196770938446e-05,
-      "loss": 6.6831,
+      "loss": 6.6595,
       "step": 45500
     },
     {
       "epoch": 19.31150293870697,
-      "grad_norm": 5.8680806159973145,
+      "grad_norm": 7.090081214904785,
       "learning_rate": 2.734611503531786e-05,
-      "loss": 6.6358,
+      "loss": 6.6073,
       "step": 46000
     },
     {
       "epoch": 19.521410579345087,
-      "grad_norm": 6.429953098297119,
+      "grad_norm": 7.082339286804199,
       "learning_rate": 2.6337033299697278e-05,
-      "loss": 6.6515,
+      "loss": 6.6206,
       "step": 46500
     },
     {
       "epoch": 19.73131821998321,
-      "grad_norm": 6.423466682434082,
+      "grad_norm": 7.417139530181885,
       "learning_rate": 2.532795156407669e-05,
-      "loss": 6.6567,
+      "loss": 6.6238,
       "step": 47000
     },
     {
       "epoch": 19.941225860621326,
-      "grad_norm": 5.939815044403076,
+      "grad_norm": 6.693517208099365,
       "learning_rate": 2.4318869828456106e-05,
-      "loss": 6.6605,
+      "loss": 6.6367,
       "step": 47500
     },
     {
       "epoch": 20.151133501259444,
-      "grad_norm": 8.216270446777344,
+      "grad_norm": 7.777519702911377,
       "learning_rate": 2.330978809283552e-05,
-      "loss": 6.6019,
+      "loss": 6.573,
       "step": 48000
     },
     {
       "epoch": 20.361041141897566,
-      "grad_norm": 7.790245056152344,
+      "grad_norm": 5.900769233703613,
       "learning_rate": 2.2300706357214934e-05,
-      "loss": 6.5774,
+      "loss": 6.5524,
       "step": 48500
     },
     {
       "epoch": 20.570948782535684,
-      "grad_norm": 5.645440578460693,
+      "grad_norm": 4.83951473236084,
       "learning_rate": 2.129162462159435e-05,
-      "loss": 6.5835,
+      "loss": 6.5599,
       "step": 49000
     },
     {
       "epoch": 20.7808564231738,
-      "grad_norm": 7.451056003570557,
+      "grad_norm": 6.502433776855469,
       "learning_rate": 2.0282542885973765e-05,
-      "loss": 6.5889,
+      "loss": 6.5616,
       "step": 49500
     },
     {
       "epoch": 20.990764063811923,
-      "grad_norm": 6.5177459716796875,
+      "grad_norm": 6.9227681159973145,
       "learning_rate": 1.927346115035318e-05,
-      "loss": 6.6002,
+      "loss": 6.5784,
       "step": 50000
     },
     {
       "epoch": 20.990764063811923,
-      "eval_accuracy": 0.1508957860206914,
-      "eval_loss": 8.99544620513916,
-      "eval_runtime": 0.7124,
-      "eval_samples_per_second": 672.34,
-      "eval_steps_per_second": 7.018,
+      "eval_accuracy": 0.1510640087475818,
+      "eval_loss": 8.965774536132812,
+      "eval_runtime": 0.6883,
+      "eval_samples_per_second": 695.961,
+      "eval_steps_per_second": 7.265,
       "step": 50000
     },
     {
       "epoch": 21.20067170445004,
-      "grad_norm": 7.058279514312744,
+      "grad_norm": 5.205214023590088,
       "learning_rate": 1.8264379414732593e-05,
-      "loss": 6.5173,
+      "loss": 6.4946,
       "step": 50500
     },
     {
       "epoch": 21.410579345088163,
-      "grad_norm": 7.122002601623535,
+      "grad_norm": 7.3034892082214355,
       "learning_rate": 1.725529767911201e-05,
-      "loss": 6.5287,
+      "loss": 6.4962,
       "step": 51000
     },
     {
       "epoch": 21.62048698572628,
-      "grad_norm": 8.928396224975586,
+      "grad_norm": 8.07140064239502,
       "learning_rate": 1.6246215943491424e-05,
-      "loss": 6.5322,
+      "loss": 6.5092,
       "step": 51500
     },
     {
       "epoch": 21.8303946263644,
-      "grad_norm": 7.159752368927002,
+      "grad_norm": 6.4861273765563965,
       "learning_rate": 1.5237134207870837e-05,
-      "loss": 6.5348,
+      "loss": 6.5071,
       "step": 52000
     },
     {
       "epoch": 22.04030226700252,
-      "grad_norm": 5.921533107757568,
+      "grad_norm": 6.208216190338135,
       "learning_rate": 1.4228052472250252e-05,
-      "loss": 6.52,
+      "loss": 6.4984,
       "step": 52500
     },
     {
       "epoch": 22.250209907640638,
-      "grad_norm": 5.727647304534912,
+      "grad_norm": 5.834749698638916,
       "learning_rate": 1.3218970736629668e-05,
-      "loss": 6.4699,
+      "loss": 6.4485,
       "step": 53000
     },
     {
       "epoch": 22.460117548278756,
-      "grad_norm": 6.089862823486328,
+      "grad_norm": 5.528692722320557,
       "learning_rate": 1.2209889001009082e-05,
-      "loss": 6.4751,
+      "loss": 6.4489,
       "step": 53500
     },
     {
       "epoch": 22.670025188916878,
-      "grad_norm": 6.925287246704102,
+      "grad_norm": 6.912316799163818,
       "learning_rate": 1.1200807265388496e-05,
-      "loss": 6.4883,
+      "loss": 6.4613,
       "step": 54000
     },
     {
       "epoch": 22.879932829554996,
-      "grad_norm": 5.46176290512085,
+      "grad_norm": 7.528624057769775,
       "learning_rate": 1.0191725529767912e-05,
-      "loss": 6.4806,
+      "loss": 6.4543,
       "step": 54500
     },
     {
       "epoch": 23.089840470193113,
-      "grad_norm": 8.658884048461914,
+      "grad_norm": 6.200753211975098,
       "learning_rate": 9.182643794147326e-06,
-      "loss": 6.4588,
+      "loss": 6.4392,
       "step": 55000
     },
     {
       "epoch": 23.299748110831235,
-      "grad_norm": 6.634410381317139,
+      "grad_norm": 7.027346611022949,
       "learning_rate": 8.173562058526741e-06,
-      "loss": 6.4317,
+      "loss": 6.4098,
       "step": 55500
     },
     {
       "epoch": 23.509655751469353,
-      "grad_norm": 7.97455358505249,
+      "grad_norm": 6.891469478607178,
       "learning_rate": 7.164480322906156e-06,
-      "loss": 6.439,
+      "loss": 6.4122,
       "step": 56000
     },
     {
       "epoch": 23.719563392107474,
-      "grad_norm": 7.158724784851074,
+      "grad_norm": 7.570326328277588,
       "learning_rate": 6.155398587285571e-06,
-      "loss": 6.4375,
+      "loss": 6.4116,
       "step": 56500
     },
     {
       "epoch": 23.929471032745592,
-      "grad_norm": 7.277324676513672,
+      "grad_norm": 6.899569034576416,
       "learning_rate": 5.146316851664986e-06,
-      "loss": 6.4374,
+      "loss": 6.4178,
       "step": 57000
     },
     {
       "epoch": 24.13937867338371,
-      "grad_norm": 7.925325393676758,
+      "grad_norm": 6.469606876373291,
       "learning_rate": 4.1372351160444e-06,
-      "loss": 6.4215,
+      "loss": 6.3977,
       "step": 57500
     },
     {
       "epoch": 24.349286314021832,
-      "grad_norm": 6.896994590759277,
+      "grad_norm": 4.759333610534668,
       "learning_rate": 3.1281533804238145e-06,
-      "loss": 6.4126,
+      "loss": 6.3958,
       "step": 58000
     },
     {
       "epoch": 24.55919395465995,
-      "grad_norm": 7.394350051879883,
+      "grad_norm": 5.711739540100098,
       "learning_rate": 2.1190716448032293e-06,
-      "loss": 6.4088,
+      "loss": 6.3952,
       "step": 58500
     },
     {
       "epoch": 24.769101595298068,
-      "grad_norm": 7.3497843742370605,
+      "grad_norm": 6.582256317138672,
       "learning_rate": 1.109989909182644e-06,
-      "loss": 6.4091,
+      "loss": 6.3862,
       "step": 59000
     },
     {
       "epoch": 24.97900923593619,
-      "grad_norm": 6.837725639343262,
+      "grad_norm": 6.575337886810303,
       "learning_rate": 1.0090817356205853e-07,
-      "loss": 6.4054,
+      "loss": 6.3807,
       "step": 59500
     },
     {
       "epoch": 25.0,
       "step": 59550,
       "total_flos": 3.049796905526016e+17,
-      "train_loss": 7.280221478184364,
-      "train_runtime": 11794.2322,
-      "train_samples_per_second": 484.642,
-      "train_steps_per_second": 5.049
+      "train_loss": 7.265635145365141,
+      "train_runtime": 11576.6001,
+      "train_samples_per_second": 493.752,
+      "train_steps_per_second": 5.144
     }
   ],
   "logging_steps": 500,