Training in progress, step 1900, checkpoint

checkpoint-1900/adapter_config.json

@@ -23,13 +23,13 @@
   "rank_pattern": {},
   "revision": null,
   "target_modules": [
-    "up_proj",
+    "o_proj",
     "down_proj",
     "gate_proj",
+    "k_proj",
     "q_proj",
     "v_proj",
-    "k_proj",
-    "o_proj"
+    "up_proj"
   ],
   "task_type": "CAUSAL_LM",
   "use_dora": false,

checkpoint-1900/adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c79b9f3cf43774f906ed44e39b277a12a900dce3bcc471687c3453db8f147a5d
+oid sha256:fdb50d6643791a33c8b7fef5454caaab311b3ec34b4dc05fbafde334317a4c30
 size 145287696

checkpoint-1900/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:24fe0e6def55478f1961be544f62440cb93ee2d0535eb9c90bf138abe32c3ef7
+oid sha256:623844d7d79711c67ab14993a5e7f796eaf8fa529368d5b78efc252916c86e94
 size 290833618

checkpoint-1900/tokenizer_config.json

@@ -1865,7 +1865,7 @@
   "legacy": true,
   "model_max_length": 4096,
   "pad_token": "<|endoftext|>",
-  "padding_side": "right",
+  "padding_side": "left",
   "sp_model_kwargs": {},
   "spaces_between_special_tokens": false,
   "tokenizer_class": "LlamaTokenizer",

checkpoint-1900/trainer_state.json

@@ -10,1636 +10,1636 @@
   "log_history": [
     {
       "epoch": 0.010869565217391304,
-      "grad_norm": 2.0792834758758545,
+      "grad_norm": 2.120478868484497,
       "learning_rate": 4.981884057971015e-05,
-      "loss": 2.3996,
+      "loss": 2.4608,
       "step": 10
     },
     {
       "epoch": 0.021739130434782608,
-      "grad_norm": 1.335732102394104,
+      "grad_norm": 1.4271776676177979,
       "learning_rate": 4.963768115942029e-05,
-      "loss": 1.25,
+      "loss": 1.2741,
       "step": 20
     },
     {
       "epoch": 0.03260869565217391,
-      "grad_norm": 1.0561087131500244,
+      "grad_norm": 0.9902746677398682,
       "learning_rate": 4.945652173913044e-05,
-      "loss": 0.9325,
+      "loss": 0.9403,
       "step": 30
     },
     {
       "epoch": 0.043478260869565216,
-      "grad_norm": 1.0983027219772339,
+      "grad_norm": 1.0730223655700684,
       "learning_rate": 4.9275362318840584e-05,
-      "loss": 0.75,
+      "loss": 0.7517,
       "step": 40
     },
     {
       "epoch": 0.05434782608695652,
-      "grad_norm": 0.8451186418533325,
+      "grad_norm": 0.8592391610145569,
       "learning_rate": 4.909420289855073e-05,
-      "loss": 0.6671,
+      "loss": 0.6695,
       "step": 50
     },
     {
       "epoch": 0.05434782608695652,
-      "eval_loss": 0.8509634733200073,
-      "eval_runtime": 10.3656,
-      "eval_samples_per_second": 46.307,
-      "eval_steps_per_second": 2.894,
+      "eval_loss": 0.8473024964332581,
+      "eval_runtime": 10.3597,
+      "eval_samples_per_second": 46.334,
+      "eval_steps_per_second": 2.896,
       "step": 50
     },
     {
       "epoch": 0.06521739130434782,
-      "grad_norm": 0.7693590521812439,
+      "grad_norm": 0.7817952632904053,
       "learning_rate": 4.891304347826087e-05,
-      "loss": 0.6309,
+      "loss": 0.6325,
       "step": 60
     },
     {
       "epoch": 0.07608695652173914,
-      "grad_norm": 0.8172665238380432,
+      "grad_norm": 0.8163846135139465,
       "learning_rate": 4.873188405797102e-05,
-      "loss": 0.5979,
+      "loss": 0.5994,
       "step": 70
     },
     {
       "epoch": 0.08695652173913043,
-      "grad_norm": 0.7351016402244568,
+      "grad_norm": 0.7358843088150024,
       "learning_rate": 4.855072463768116e-05,
-      "loss": 0.6149,
+      "loss": 0.6158,
       "step": 80
     },
     {
       "epoch": 0.09782608695652174,
-      "grad_norm": 0.8038097023963928,
+      "grad_norm": 0.7856508493423462,
       "learning_rate": 4.836956521739131e-05,
-      "loss": 0.5969,
+      "loss": 0.5983,
       "step": 90
     },
     {
       "epoch": 0.10869565217391304,
-      "grad_norm": 0.8729331493377686,
+      "grad_norm": 0.8615956902503967,
       "learning_rate": 4.818840579710145e-05,
-      "loss": 0.5847,
+      "loss": 0.5858,
       "step": 100
     },
     {
       "epoch": 0.10869565217391304,
-      "eval_loss": 0.7441961765289307,
-      "eval_runtime": 10.4341,
-      "eval_samples_per_second": 46.003,
-      "eval_steps_per_second": 2.875,
+      "eval_loss": 0.7398766875267029,
+      "eval_runtime": 10.4317,
+      "eval_samples_per_second": 46.014,
+      "eval_steps_per_second": 2.876,
       "step": 100
     },
     {
       "epoch": 0.11956521739130435,
-      "grad_norm": 0.7882421016693115,
+      "grad_norm": 0.7915664315223694,
       "learning_rate": 4.80072463768116e-05,
-      "loss": 0.622,
+      "loss": 0.6228,
       "step": 110
     },
     {
       "epoch": 0.13043478260869565,
-      "grad_norm": 0.8385023474693298,
+      "grad_norm": 0.8432077765464783,
       "learning_rate": 4.782608695652174e-05,
-      "loss": 0.5637,
+      "loss": 0.5649,
       "step": 120
     },
     {
       "epoch": 0.14130434782608695,
-      "grad_norm": 0.8384169340133667,
+      "grad_norm": 0.8230385780334473,
       "learning_rate": 4.764492753623189e-05,
-      "loss": 0.5363,
+      "loss": 0.5368,
       "step": 130
     },
     {
       "epoch": 0.15217391304347827,
-      "grad_norm": 0.8619622588157654,
+      "grad_norm": 0.8497728109359741,
       "learning_rate": 4.746376811594203e-05,
-      "loss": 0.544,
+      "loss": 0.5447,
       "step": 140
     },
     {
       "epoch": 0.16304347826086957,
-      "grad_norm": 0.9462847709655762,
+      "grad_norm": 0.9394277930259705,
       "learning_rate": 4.7282608695652177e-05,
-      "loss": 0.5512,
+      "loss": 0.5515,
       "step": 150
     },
     {
       "epoch": 0.16304347826086957,
-      "eval_loss": 0.6991452574729919,
-      "eval_runtime": 10.4072,
-      "eval_samples_per_second": 46.122,
-      "eval_steps_per_second": 2.883,
+      "eval_loss": 0.694922149181366,
+      "eval_runtime": 10.4348,
+      "eval_samples_per_second": 46.0,
+      "eval_steps_per_second": 2.875,
       "step": 150
     },
     {
       "epoch": 0.17391304347826086,
-      "grad_norm": 0.7737827897071838,
+      "grad_norm": 0.7665875554084778,
       "learning_rate": 4.710144927536232e-05,
-      "loss": 0.5623,
+      "loss": 0.5632,
       "step": 160
     },
     {
       "epoch": 0.18478260869565216,
-      "grad_norm": 0.8910329937934875,
+      "grad_norm": 0.8912480473518372,
       "learning_rate": 4.6920289855072464e-05,
-      "loss": 0.5254,
+      "loss": 0.5259,
       "step": 170
     },
     {
       "epoch": 0.1956521739130435,
-      "grad_norm": 0.8883982300758362,
+      "grad_norm": 0.8959771990776062,
       "learning_rate": 4.673913043478261e-05,
-      "loss": 0.5302,
+      "loss": 0.5307,
       "step": 180
     },
     {
       "epoch": 0.20652173913043478,
-      "grad_norm": 0.855445921421051,
+      "grad_norm": 0.8595960140228271,
       "learning_rate": 4.655797101449276e-05,
-      "loss": 0.5157,
+      "loss": 0.5167,
       "step": 190
     },
     {
       "epoch": 0.21739130434782608,
-      "grad_norm": 0.8524109721183777,
+      "grad_norm": 0.8418941497802734,
       "learning_rate": 4.63768115942029e-05,
-      "loss": 0.5102,
+      "loss": 0.5106,
       "step": 200
     },
     {
       "epoch": 0.21739130434782608,
-      "eval_loss": 0.6746214628219604,
-      "eval_runtime": 10.3921,
-      "eval_samples_per_second": 46.189,
-      "eval_steps_per_second": 2.887,
+      "eval_loss": 0.6705857515335083,
+      "eval_runtime": 10.4259,
+      "eval_samples_per_second": 46.039,
+      "eval_steps_per_second": 2.877,
       "step": 200
     },
     {
       "epoch": 0.22826086956521738,
-      "grad_norm": 0.9038218855857849,
+      "grad_norm": 0.8937019109725952,
       "learning_rate": 4.6195652173913046e-05,
-      "loss": 0.5321,
+      "loss": 0.5329,
       "step": 210
     },
     {
       "epoch": 0.2391304347826087,
-      "grad_norm": 0.9139806628227234,
+      "grad_norm": 0.9121237397193909,
       "learning_rate": 4.601449275362319e-05,
-      "loss": 0.5337,
+      "loss": 0.5341,
       "step": 220
     },
     {
       "epoch": 0.25,
-      "grad_norm": 0.8306450247764587,
+      "grad_norm": 0.8253990411758423,
       "learning_rate": 4.5833333333333334e-05,
-      "loss": 0.5324,
+      "loss": 0.5328,
       "step": 230
     },
     {
       "epoch": 0.2608695652173913,
-      "grad_norm": 0.9253225326538086,
+      "grad_norm": 0.9218540787696838,
       "learning_rate": 4.565217391304348e-05,
-      "loss": 0.4702,
+      "loss": 0.4706,
       "step": 240
     },
     {
       "epoch": 0.2717391304347826,
-      "grad_norm": 0.9208872318267822,
+      "grad_norm": 0.9071463346481323,
       "learning_rate": 4.547101449275363e-05,
       "loss": 0.5012,
       "step": 250
     },
     {
       "epoch": 0.2717391304347826,
-      "eval_loss": 0.6523857116699219,
-      "eval_runtime": 10.3866,
-      "eval_samples_per_second": 46.214,
-      "eval_steps_per_second": 2.888,
+      "eval_loss": 0.6487293839454651,
+      "eval_runtime": 10.4286,
+      "eval_samples_per_second": 46.027,
+      "eval_steps_per_second": 2.877,
       "step": 250
     },
     {
       "epoch": 0.2826086956521739,
-      "grad_norm": 0.9187518954277039,
+      "grad_norm": 0.9131689071655273,
       "learning_rate": 4.528985507246377e-05,
-      "loss": 0.4811,
+      "loss": 0.4819,
       "step": 260
     },
     {
       "epoch": 0.29347826086956524,
-      "grad_norm": 0.849318265914917,
+      "grad_norm": 0.8460387587547302,
       "learning_rate": 4.5108695652173916e-05,
-      "loss": 0.494,
+      "loss": 0.4951,
       "step": 270
     },
     {
       "epoch": 0.30434782608695654,
-      "grad_norm": 0.8616385459899902,
+      "grad_norm": 0.8639910817146301,
       "learning_rate": 4.492753623188406e-05,
-      "loss": 0.4934,
+      "loss": 0.4941,
       "step": 280
     },
     {
       "epoch": 0.31521739130434784,
-      "grad_norm": 1.0425682067871094,
+      "grad_norm": 1.0238360166549683,
       "learning_rate": 4.4746376811594203e-05,
-      "loss": 0.4835,
+      "loss": 0.4844,
       "step": 290
     },
     {
       "epoch": 0.32608695652173914,
-      "grad_norm": 0.8278887271881104,
+      "grad_norm": 0.8217597603797913,
       "learning_rate": 4.456521739130435e-05,
-      "loss": 0.4961,
+      "loss": 0.497,
       "step": 300
     },
     {
       "epoch": 0.32608695652173914,
-      "eval_loss": 0.639061689376831,
-      "eval_runtime": 10.3771,
-      "eval_samples_per_second": 46.256,
-      "eval_steps_per_second": 2.891,
+      "eval_loss": 0.6351093053817749,
+      "eval_runtime": 10.4278,
+      "eval_samples_per_second": 46.031,
+      "eval_steps_per_second": 2.877,
       "step": 300
     },
     {
       "epoch": 0.33695652173913043,
-      "grad_norm": 0.8740423321723938,
+      "grad_norm": 0.8742851614952087,
       "learning_rate": 4.438405797101449e-05,
-      "loss": 0.5131,
+      "loss": 0.5134,
       "step": 310
     },
     {
       "epoch": 0.34782608695652173,
-      "grad_norm": 0.8640497326850891,
+      "grad_norm": 0.8606206774711609,
       "learning_rate": 4.4202898550724645e-05,
-      "loss": 0.4576,
+      "loss": 0.4577,
       "step": 320
     },
     {
       "epoch": 0.358695652173913,
-      "grad_norm": 1.0350477695465088,
+      "grad_norm": 1.030929446220398,
       "learning_rate": 4.4021739130434786e-05,
-      "loss": 0.4621,
+      "loss": 0.4624,
       "step": 330
     },
     {
       "epoch": 0.3695652173913043,
-      "grad_norm": 0.8288527727127075,
+      "grad_norm": 0.8211489915847778,
       "learning_rate": 4.384057971014493e-05,
-      "loss": 0.4929,
+      "loss": 0.4942,
       "step": 340
     },
     {
       "epoch": 0.3804347826086957,
-      "grad_norm": 1.0214691162109375,
+      "grad_norm": 1.0272365808486938,
       "learning_rate": 4.365942028985507e-05,
       "loss": 0.4923,
       "step": 350
     },
     {
       "epoch": 0.3804347826086957,
-      "eval_loss": 0.6334717869758606,
-      "eval_runtime": 10.3968,
-      "eval_samples_per_second": 46.168,
-      "eval_steps_per_second": 2.885,
+      "eval_loss": 0.6294742226600647,
+      "eval_runtime": 10.439,
+      "eval_samples_per_second": 45.981,
+      "eval_steps_per_second": 2.874,
       "step": 350
     },
     {
       "epoch": 0.391304347826087,
-      "grad_norm": 0.9189789295196533,
+      "grad_norm": 0.9275714159011841,
       "learning_rate": 4.347826086956522e-05,
-      "loss": 0.4781,
+      "loss": 0.4784,
       "step": 360
     },
     {
       "epoch": 0.40217391304347827,
-      "grad_norm": 0.9531463980674744,
+      "grad_norm": 0.9278607368469238,
       "learning_rate": 4.329710144927536e-05,
-      "loss": 0.4651,
+      "loss": 0.465,
       "step": 370
     },
     {
       "epoch": 0.41304347826086957,
-      "grad_norm": 1.0185341835021973,
+      "grad_norm": 1.009921669960022,
       "learning_rate": 4.3115942028985515e-05,
-      "loss": 0.477,
+      "loss": 0.4776,
       "step": 380
     },
     {
       "epoch": 0.42391304347826086,
-      "grad_norm": 0.9705609679222107,
+      "grad_norm": 0.9694483280181885,
       "learning_rate": 4.2934782608695655e-05,
-      "loss": 0.4954,
+      "loss": 0.4956,
       "step": 390
     },
     {
       "epoch": 0.43478260869565216,
-      "grad_norm": 0.8557504415512085,
+      "grad_norm": 0.8650519251823425,
       "learning_rate": 4.27536231884058e-05,
-      "loss": 0.4602,
+      "loss": 0.4607,
       "step": 400
     },
     {
       "epoch": 0.43478260869565216,
-      "eval_loss": 0.6238241791725159,
-      "eval_runtime": 10.4192,
-      "eval_samples_per_second": 46.069,
+      "eval_loss": 0.6202508807182312,
+      "eval_runtime": 10.4217,
+      "eval_samples_per_second": 46.058,
       "eval_steps_per_second": 2.879,
       "step": 400
     },
     {
       "epoch": 0.44565217391304346,
-      "grad_norm": 1.0175505876541138,
+      "grad_norm": 1.0009912252426147,
       "learning_rate": 4.257246376811594e-05,
-      "loss": 0.4682,
+      "loss": 0.4695,
       "step": 410
     },
     {
       "epoch": 0.45652173913043476,
-      "grad_norm": 0.9760918617248535,
+      "grad_norm": 0.9651597738265991,
       "learning_rate": 4.239130434782609e-05,
-      "loss": 0.4731,
+      "loss": 0.4733,
       "step": 420
     },
     {
       "epoch": 0.4673913043478261,
-      "grad_norm": 0.9057648181915283,
+      "grad_norm": 0.9021621346473694,
       "learning_rate": 4.221014492753623e-05,
-      "loss": 0.4537,
+      "loss": 0.4543,
       "step": 430
     },
     {
       "epoch": 0.4782608695652174,
-      "grad_norm": 0.9714989066123962,
+      "grad_norm": 0.9730608463287354,
       "learning_rate": 4.202898550724638e-05,
-      "loss": 0.446,
+      "loss": 0.4461,
       "step": 440
     },
     {
       "epoch": 0.4891304347826087,
-      "grad_norm": 0.950578510761261,
+      "grad_norm": 0.9395557641983032,
       "learning_rate": 4.1847826086956525e-05,
-      "loss": 0.4444,
+      "loss": 0.4448,
       "step": 450
     },
     {
       "epoch": 0.4891304347826087,
-      "eval_loss": 0.6164894104003906,
-      "eval_runtime": 10.3782,
-      "eval_samples_per_second": 46.251,
-      "eval_steps_per_second": 2.891,
+      "eval_loss": 0.6122747659683228,
+      "eval_runtime": 10.4358,
+      "eval_samples_per_second": 45.995,
+      "eval_steps_per_second": 2.875,
       "step": 450
     },
     {
       "epoch": 0.5,
-      "grad_norm": 0.9109216928482056,
+      "grad_norm": 0.9062746167182922,
       "learning_rate": 4.166666666666667e-05,
-      "loss": 0.465,
+      "loss": 0.4653,
       "step": 460
     },
     {
       "epoch": 0.5108695652173914,
-      "grad_norm": 0.9473273158073425,
+      "grad_norm": 0.9474561214447021,
       "learning_rate": 4.148550724637681e-05,
-      "loss": 0.458,
+      "loss": 0.4571,
       "step": 470
     },
     {
       "epoch": 0.5217391304347826,
-      "grad_norm": 0.921631932258606,
+      "grad_norm": 0.9197052717208862,
       "learning_rate": 4.130434782608696e-05,
       "loss": 0.4416,
       "step": 480
     },
     {
       "epoch": 0.532608695652174,
-      "grad_norm": 0.994552493095398,
+      "grad_norm": 0.9986901879310608,
       "learning_rate": 4.11231884057971e-05,
-      "loss": 0.4441,
+      "loss": 0.4446,
       "step": 490
     },
     {
       "epoch": 0.5434782608695652,
-      "grad_norm": 0.9409403800964355,
+      "grad_norm": 0.9315592050552368,
       "learning_rate": 4.094202898550725e-05,
-      "loss": 0.45,
+      "loss": 0.4501,
       "step": 500
     },
     {
       "epoch": 0.5434782608695652,
-      "eval_loss": 0.6097907423973083,
-      "eval_runtime": 10.3702,
-      "eval_samples_per_second": 46.286,
-      "eval_steps_per_second": 2.893,
+      "eval_loss": 0.6061260104179382,
+      "eval_runtime": 10.4306,
+      "eval_samples_per_second": 46.018,
+      "eval_steps_per_second": 2.876,
       "step": 500
     },
     {
       "epoch": 0.5543478260869565,
-      "grad_norm": 0.9720210433006287,
+      "grad_norm": 0.9725834131240845,
       "learning_rate": 4.076086956521739e-05,
-      "loss": 0.4468,
+      "loss": 0.4465,
       "step": 510
     },
     {
       "epoch": 0.5652173913043478,
-      "grad_norm": 0.9678219556808472,
+      "grad_norm": 0.9681097865104675,
       "learning_rate": 4.057971014492754e-05,
-      "loss": 0.4741,
+      "loss": 0.4735,
       "step": 520
     },
     {
       "epoch": 0.5760869565217391,
-      "grad_norm": 0.9966796636581421,
+      "grad_norm": 1.0040682554244995,
       "learning_rate": 4.039855072463768e-05,
-      "loss": 0.4742,
+      "loss": 0.4743,
       "step": 530
     },
     {
       "epoch": 0.5869565217391305,
-      "grad_norm": 0.9896467328071594,
+      "grad_norm": 0.9864315986633301,
       "learning_rate": 4.021739130434783e-05,
-      "loss": 0.4499,
+      "loss": 0.45,
       "step": 540
     },
     {
       "epoch": 0.5978260869565217,
-      "grad_norm": 0.9556202292442322,
+      "grad_norm": 0.954130232334137,
       "learning_rate": 4.003623188405797e-05,
-      "loss": 0.4398,
+      "loss": 0.439,
       "step": 550
     },
     {
       "epoch": 0.5978260869565217,
-      "eval_loss": 0.6025962829589844,
-      "eval_runtime": 10.3801,
-      "eval_samples_per_second": 46.242,
-      "eval_steps_per_second": 2.89,
+      "eval_loss": 0.5990516543388367,
+      "eval_runtime": 10.427,
+      "eval_samples_per_second": 46.035,
+      "eval_steps_per_second": 2.877,
       "step": 550
     },
     {
       "epoch": 0.6086956521739131,
-      "grad_norm": 1.0646259784698486,
+      "grad_norm": 1.150810718536377,
       "learning_rate": 3.985507246376812e-05,
-      "loss": 0.4596,
+      "loss": 0.4603,
       "step": 560
     },
     {
       "epoch": 0.6195652173913043,
-      "grad_norm": 0.9542304873466492,
+      "grad_norm": 0.9530227184295654,
       "learning_rate": 3.9673913043478264e-05,
       "loss": 0.447,
       "step": 570
     },
     {
       "epoch": 0.6304347826086957,
-      "grad_norm": 1.153639316558838,
+      "grad_norm": 1.1645703315734863,
       "learning_rate": 3.9492753623188405e-05,
-      "loss": 0.4194,
+      "loss": 0.4195,
       "step": 580
     },
     {
       "epoch": 0.6413043478260869,
-      "grad_norm": 1.0100998878479004,
+      "grad_norm": 1.0095150470733643,
       "learning_rate": 3.931159420289855e-05,
-      "loss": 0.4344,
+      "loss": 0.434,
       "step": 590
     },
     {
       "epoch": 0.6521739130434783,
-      "grad_norm": 0.8966252207756042,
+      "grad_norm": 0.8818724155426025,
       "learning_rate": 3.91304347826087e-05,
-      "loss": 0.4387,
+      "loss": 0.4385,
       "step": 600
     },
     {
       "epoch": 0.6521739130434783,
-      "eval_loss": 0.5988693237304688,
-      "eval_runtime": 10.3834,
-      "eval_samples_per_second": 46.228,
-      "eval_steps_per_second": 2.889,
+      "eval_loss": 0.5951303839683533,
+      "eval_runtime": 10.4035,
+      "eval_samples_per_second": 46.138,
+      "eval_steps_per_second": 2.884,
       "step": 600
     },
     {
       "epoch": 0.6630434782608695,
-      "grad_norm": 1.0731533765792847,
+      "grad_norm": 1.0614160299301147,
       "learning_rate": 3.8949275362318846e-05,
-      "loss": 0.4127,
+      "loss": 0.4128,
       "step": 610
     },
     {
       "epoch": 0.6739130434782609,
-      "grad_norm": 0.8809463977813721,
+      "grad_norm": 0.883694589138031,
       "learning_rate": 3.876811594202899e-05,
-      "loss": 0.4663,
+      "loss": 0.4664,
       "step": 620
     },
     {
       "epoch": 0.6847826086956522,
-      "grad_norm": 0.9806007742881775,
+      "grad_norm": 0.97767573595047,
       "learning_rate": 3.8586956521739134e-05,
-      "loss": 0.4146,
+      "loss": 0.4149,
       "step": 630
     },
     {
       "epoch": 0.6956521739130435,
-      "grad_norm": 0.9644826650619507,
+      "grad_norm": 0.9714617133140564,
       "learning_rate": 3.8405797101449274e-05,
-      "loss": 0.4378,
+      "loss": 0.4369,
       "step": 640
     },
     {
       "epoch": 0.7065217391304348,
-      "grad_norm": 0.9650328755378723,
+      "grad_norm": 0.9593505263328552,
       "learning_rate": 3.822463768115942e-05,
-      "loss": 0.4269,
+      "loss": 0.4268,
       "step": 650
     },
     {
       "epoch": 0.7065217391304348,
-      "eval_loss": 0.5938716530799866,
-      "eval_runtime": 10.3772,
-      "eval_samples_per_second": 46.255,
+      "eval_loss": 0.5908556580543518,
+      "eval_runtime": 10.3762,
+      "eval_samples_per_second": 46.26,
       "eval_steps_per_second": 2.891,
       "step": 650
     },
     {
       "epoch": 0.717391304347826,
-      "grad_norm": 0.9935194849967957,
+      "grad_norm": 0.9807693362236023,
       "learning_rate": 3.804347826086957e-05,
-      "loss": 0.4223,
+      "loss": 0.4224,
       "step": 660
     },
     {
       "epoch": 0.7282608695652174,
-      "grad_norm": 1.0365718603134155,
+      "grad_norm": 1.0408517122268677,
       "learning_rate": 3.7862318840579716e-05,
       "loss": 0.41,
       "step": 670
     },
     {
       "epoch": 0.7391304347826086,
-      "grad_norm": 0.9572827219963074,
+      "grad_norm": 0.9578678607940674,
       "learning_rate": 3.7681159420289856e-05,
-      "loss": 0.42,
+      "loss": 0.4207,
       "step": 680
     },
     {
       "epoch": 0.75,
-      "grad_norm": 1.034522294998169,
+      "grad_norm": 1.0341449975967407,
       "learning_rate": 3.7500000000000003e-05,
-      "loss": 0.4432,
+      "loss": 0.4435,
       "step": 690
     },
     {
       "epoch": 0.7608695652173914,
-      "grad_norm": 0.9503633379936218,
+      "grad_norm": 0.9478809833526611,
       "learning_rate": 3.7318840579710144e-05,
-      "loss": 0.4464,
+      "loss": 0.4463,
       "step": 700
     },
     {
       "epoch": 0.7608695652173914,
-      "eval_loss": 0.5888856053352356,
-      "eval_runtime": 10.3948,
-      "eval_samples_per_second": 46.177,
-      "eval_steps_per_second": 2.886,
+      "eval_loss": 0.585399866104126,
+      "eval_runtime": 10.3883,
+      "eval_samples_per_second": 46.206,
+      "eval_steps_per_second": 2.888,
       "step": 700
     },
     {
       "epoch": 0.7717391304347826,
-      "grad_norm": 1.038071870803833,
+      "grad_norm": 1.0197566747665405,
       "learning_rate": 3.713768115942029e-05,
       "loss": 0.4234,
       "step": 710
     },
     {
       "epoch": 0.782608695652174,
-      "grad_norm": 1.0283771753311157,
+      "grad_norm": 1.0265061855316162,
       "learning_rate": 3.695652173913043e-05,
-      "loss": 0.4019,
+      "loss": 0.4027,
       "step": 720
     },
     {
       "epoch": 0.7934782608695652,
-      "grad_norm": 0.9672926068305969,
+      "grad_norm": 0.9763724207878113,
       "learning_rate": 3.6775362318840586e-05,
-      "loss": 0.4286,
+      "loss": 0.4284,
       "step": 730
     },
     {
       "epoch": 0.8043478260869565,
-      "grad_norm": 1.0848443508148193,
+      "grad_norm": 1.1108958721160889,
       "learning_rate": 3.6594202898550726e-05,
-      "loss": 0.4196,
+      "loss": 0.4197,
       "step": 740
     },
     {
       "epoch": 0.8152173913043478,
-      "grad_norm": 0.9358295798301697,
+      "grad_norm": 0.938008189201355,
       "learning_rate": 3.641304347826087e-05,
-      "loss": 0.4127,
+      "loss": 0.4133,
       "step": 750
     },
     {
       "epoch": 0.8152173913043478,
-      "eval_loss": 0.5893052220344543,
-      "eval_runtime": 10.3778,
-      "eval_samples_per_second": 46.252,
+      "eval_loss": 0.5860960483551025,
+      "eval_runtime": 10.3772,
+      "eval_samples_per_second": 46.255,
       "eval_steps_per_second": 2.891,
       "step": 750
     },
     {
       "epoch": 0.8260869565217391,
-      "grad_norm": 1.1050411462783813,
+      "grad_norm": 1.106451392173767,
       "learning_rate": 3.6231884057971014e-05,
-      "loss": 0.4335,
+      "loss": 0.4342,
       "step": 760
     },
     {
       "epoch": 0.8369565217391305,
-      "grad_norm": 1.1128437519073486,
+      "grad_norm": 1.1099311113357544,
       "learning_rate": 3.605072463768116e-05,
-      "loss": 0.434,
+      "loss": 0.4345,
       "step": 770
     },
     {
       "epoch": 0.8478260869565217,
-      "grad_norm": 1.0179415941238403,
+      "grad_norm": 1.0185613632202148,
       "learning_rate": 3.58695652173913e-05,
-      "loss": 0.419,
+      "loss": 0.4196,
       "step": 780
     },
     {
       "epoch": 0.8586956521739131,
-      "grad_norm": 0.9764388799667358,
+      "grad_norm": 0.9686940908432007,
       "learning_rate": 3.568840579710145e-05,
-      "loss": 0.4061,
+      "loss": 0.4063,
       "step": 790
     },
     {
       "epoch": 0.8695652173913043,
-      "grad_norm": 0.9722176194190979,
+      "grad_norm": 0.9732295274734497,
       "learning_rate": 3.5507246376811596e-05,
-      "loss": 0.4359,
+      "loss": 0.4365,
       "step": 800
     },
     {
       "epoch": 0.8695652173913043,
-      "eval_loss": 0.5821022391319275,
-      "eval_runtime": 10.3721,
-      "eval_samples_per_second": 46.278,
-      "eval_steps_per_second": 2.892,
+      "eval_loss": 0.5782727003097534,
+      "eval_runtime": 10.4122,
+      "eval_samples_per_second": 46.1,
+      "eval_steps_per_second": 2.881,
       "step": 800
     },
     {
       "epoch": 0.8804347826086957,
-      "grad_norm": 1.1173466444015503,
+      "grad_norm": 1.1204928159713745,
       "learning_rate": 3.532608695652174e-05,
-      "loss": 0.416,
+      "loss": 0.4165,
       "step": 810
     },
     {
       "epoch": 0.8913043478260869,
-      "grad_norm": 1.0256565809249878,
+      "grad_norm": 1.0249269008636475,
       "learning_rate": 3.514492753623188e-05,
-      "loss": 0.4335,
+      "loss": 0.4339,
       "step": 820
     },
     {
       "epoch": 0.9021739130434783,
-      "grad_norm": 0.9599832892417908,
+      "grad_norm": 0.9561118483543396,
       "learning_rate": 3.496376811594203e-05,
-      "loss": 0.4072,
+      "loss": 0.4077,
       "step": 830
     },
     {
       "epoch": 0.9130434782608695,
-      "grad_norm": 1.0554136037826538,
+      "grad_norm": 1.0603339672088623,
       "learning_rate": 3.478260869565218e-05,
-      "loss": 0.4173,
+      "loss": 0.4176,
       "step": 840
     },
     {
       "epoch": 0.9239130434782609,
-      "grad_norm": 0.9580271244049072,
+      "grad_norm": 0.9610790610313416,
       "learning_rate": 3.460144927536232e-05,
-      "loss": 0.418,
+      "loss": 0.4179,
       "step": 850
     },
     {
       "epoch": 0.9239130434782609,
-      "eval_loss": 0.5831677317619324,
-      "eval_runtime": 10.3923,
-      "eval_samples_per_second": 46.188,
-      "eval_steps_per_second": 2.887,
+      "eval_loss": 0.5790796875953674,
+      "eval_runtime": 10.4182,
+      "eval_samples_per_second": 46.073,
+      "eval_steps_per_second": 2.88,
       "step": 850
     },
     {
       "epoch": 0.9347826086956522,
-      "grad_norm": 1.0673909187316895,
+      "grad_norm": 1.0670075416564941,
       "learning_rate": 3.4420289855072465e-05,
-      "loss": 0.4115,
+      "loss": 0.4114,
       "step": 860
     },
     {
       "epoch": 0.9456521739130435,
-      "grad_norm": 1.0771151781082153,
+      "grad_norm": 1.070837140083313,
       "learning_rate": 3.423913043478261e-05,
-      "loss": 0.4038,
+      "loss": 0.4034,
       "step": 870
     },
     {
       "epoch": 0.9565217391304348,
-      "grad_norm": 1.0083266496658325,
+      "grad_norm": 1.0078990459442139,
       "learning_rate": 3.405797101449276e-05,
-      "loss": 0.399,
+      "loss": 0.3985,
       "step": 880
     },
     {
       "epoch": 0.967391304347826,
-      "grad_norm": 1.0740772485733032,
+      "grad_norm": 1.0645533800125122,
       "learning_rate": 3.38768115942029e-05,
-      "loss": 0.4231,
+      "loss": 0.4238,
       "step": 890
     },
     {
       "epoch": 0.9782608695652174,
-      "grad_norm": 0.9597785472869873,
+      "grad_norm": 0.9449084401130676,
       "learning_rate": 3.369565217391305e-05,
-      "loss": 0.4151,
+      "loss": 0.4146,
       "step": 900
     },
     {
       "epoch": 0.9782608695652174,
-      "eval_loss": 0.5803946852684021,
-      "eval_runtime": 10.3767,
-      "eval_samples_per_second": 46.258,
-      "eval_steps_per_second": 2.891,
+      "eval_loss": 0.5765789151191711,
+      "eval_runtime": 10.4319,
+      "eval_samples_per_second": 46.013,
+      "eval_steps_per_second": 2.876,
       "step": 900
     },
     {
       "epoch": 0.9891304347826086,
-      "grad_norm": 1.1659215688705444,
+      "grad_norm": 1.1554139852523804,
       "learning_rate": 3.351449275362319e-05,
-      "loss": 0.397,
+      "loss": 0.3972,
       "step": 910
     },
     {
       "epoch": 1.0,
-      "grad_norm": 2.251577138900757,
+      "grad_norm": 2.253143787384033,
       "learning_rate": 3.3333333333333335e-05,
-      "loss": 0.4355,
+      "loss": 0.4359,
       "step": 920
     },
     {
       "epoch": 1.0108695652173914,
-      "grad_norm": 1.1286661624908447,
+      "grad_norm": 1.124770164489746,
       "learning_rate": 3.3152173913043475e-05,
-      "loss": 0.3536,
+      "loss": 0.3533,
       "step": 930
     },
     {
       "epoch": 1.0217391304347827,
-      "grad_norm": 1.1189433336257935,
+      "grad_norm": 1.1178290843963623,
       "learning_rate": 3.297101449275363e-05,
-      "loss": 0.3596,
+      "loss": 0.3599,
       "step": 940
     },
     {
       "epoch": 1.0326086956521738,
-      "grad_norm": 1.0220948457717896,
+      "grad_norm": 1.0256346464157104,
       "learning_rate": 3.278985507246377e-05,
       "loss": 0.3632,
       "step": 950
     },
     {
       "epoch": 1.0326086956521738,
-      "eval_loss": 0.5912913680076599,
-      "eval_runtime": 10.3768,
-      "eval_samples_per_second": 46.257,
-      "eval_steps_per_second": 2.891,
+      "eval_loss": 0.5881566405296326,
+      "eval_runtime": 10.4627,
+      "eval_samples_per_second": 45.877,
+      "eval_steps_per_second": 2.867,
       "step": 950
     },
     {
       "epoch": 1.0434782608695652,
-      "grad_norm": 1.0051358938217163,
+      "grad_norm": 1.0145591497421265,
       "learning_rate": 3.260869565217392e-05,
-      "loss": 0.3648,
+      "loss": 0.3647,
       "step": 960
     },
     {
       "epoch": 1.0543478260869565,
-      "grad_norm": 1.1821051836013794,
+      "grad_norm": 1.2014787197113037,
       "learning_rate": 3.242753623188406e-05,
-      "loss": 0.3705,
+      "loss": 0.37,
       "step": 970
     },
     {
       "epoch": 1.065217391304348,
-      "grad_norm": 1.1006765365600586,
+      "grad_norm": 1.0973201990127563,
       "learning_rate": 3.2246376811594205e-05,
-      "loss": 0.3606,
+      "loss": 0.3608,
       "step": 980
     },
     {
       "epoch": 1.0760869565217392,
-      "grad_norm": 1.1193490028381348,
+      "grad_norm": 1.120807409286499,
       "learning_rate": 3.2065217391304345e-05,
-      "loss": 0.3471,
+      "loss": 0.3466,
       "step": 990
     },
     {
       "epoch": 1.0869565217391304,
-      "grad_norm": 1.1935309171676636,
+      "grad_norm": 1.1913529634475708,
       "learning_rate": 3.188405797101449e-05,
-      "loss": 0.3667,
+      "loss": 0.3671,
       "step": 1000
     },
     {
       "epoch": 1.0869565217391304,
-      "eval_loss": 0.5953035950660706,
-      "eval_runtime": 10.3685,
-      "eval_samples_per_second": 46.294,
-      "eval_steps_per_second": 2.893,
+      "eval_loss": 0.591897189617157,
+      "eval_runtime": 10.4218,
+      "eval_samples_per_second": 46.057,
+      "eval_steps_per_second": 2.879,
       "step": 1000
     },
     {
       "epoch": 1.0978260869565217,
-      "grad_norm": 1.1882245540618896,
+      "grad_norm": 1.1925430297851562,
       "learning_rate": 3.170289855072464e-05,
-      "loss": 0.3731,
+      "loss": 0.3729,
       "step": 1010
     },
     {
       "epoch": 1.108695652173913,
-      "grad_norm": 1.0312139987945557,
+      "grad_norm": 1.0378402471542358,
       "learning_rate": 3.152173913043479e-05,
-      "loss": 0.3606,
+      "loss": 0.3615,
       "step": 1020
     },
     {
       "epoch": 1.1195652173913044,
-      "grad_norm": 1.2470943927764893,
+      "grad_norm": 1.2244843244552612,
       "learning_rate": 3.134057971014493e-05,
-      "loss": 0.3508,
+      "loss": 0.3507,
       "step": 1030
     },
     {
       "epoch": 1.1304347826086956,
-      "grad_norm": 1.2272464036941528,
+      "grad_norm": 1.211162805557251,
       "learning_rate": 3.1159420289855074e-05,
-      "loss": 0.3609,
+      "loss": 0.361,
       "step": 1040
     },
     {
       "epoch": 1.141304347826087,
-      "grad_norm": 1.1007187366485596,
+      "grad_norm": 1.1001977920532227,
       "learning_rate": 3.0978260869565215e-05,
       "loss": 0.3396,
       "step": 1050
     },
     {
       "epoch": 1.141304347826087,
-      "eval_loss": 0.5945261716842651,
-      "eval_runtime": 10.374,
-      "eval_samples_per_second": 46.269,
-      "eval_steps_per_second": 2.892,
+      "eval_loss": 0.5917331576347351,
+      "eval_runtime": 10.4223,
+      "eval_samples_per_second": 46.055,
+      "eval_steps_per_second": 2.878,
       "step": 1050
     },
     {
       "epoch": 1.1521739130434783,
-      "grad_norm": 1.1125508546829224,
+      "grad_norm": 1.1034841537475586,
       "learning_rate": 3.079710144927536e-05,
       "loss": 0.3718,
       "step": 1060
     },
     {
       "epoch": 1.1630434782608696,
-      "grad_norm": 1.1342809200286865,
+      "grad_norm": 1.1438697576522827,
       "learning_rate": 3.061594202898551e-05,
-      "loss": 0.3433,
+      "loss": 0.3436,
       "step": 1070
     },
     {
       "epoch": 1.1739130434782608,
-      "grad_norm": 1.1907680034637451,
+      "grad_norm": 1.1911267042160034,
       "learning_rate": 3.0434782608695656e-05,
-      "loss": 0.357,
+      "loss": 0.3569,
       "step": 1080
     },
     {
       "epoch": 1.184782608695652,
-      "grad_norm": 1.1469812393188477,
+      "grad_norm": 1.1565200090408325,
       "learning_rate": 3.02536231884058e-05,
-      "loss": 0.3632,
+      "loss": 0.3639,
       "step": 1090
     },
     {
       "epoch": 1.1956521739130435,
-      "grad_norm": 1.2115755081176758,
+      "grad_norm": 1.199948787689209,
       "learning_rate": 3.0072463768115944e-05,
-      "loss": 0.3578,
+      "loss": 0.3588,
       "step": 1100
     },
     {
       "epoch": 1.1956521739130435,
-      "eval_loss": 0.5987864136695862,
-      "eval_runtime": 10.3646,
-      "eval_samples_per_second": 46.311,
-      "eval_steps_per_second": 2.894,
+      "eval_loss": 0.5949474573135376,
+      "eval_runtime": 10.4416,
+      "eval_samples_per_second": 45.97,
+      "eval_steps_per_second": 2.873,
       "step": 1100
     },
     {
       "epoch": 1.2065217391304348,
-      "grad_norm": 1.139615535736084,
+      "grad_norm": 1.1408541202545166,
       "learning_rate": 2.9891304347826088e-05,
-      "loss": 0.3476,
+      "loss": 0.3479,
       "step": 1110
     },
     {
       "epoch": 1.2173913043478262,
-      "grad_norm": 1.1213746070861816,
+      "grad_norm": 1.1276154518127441,
       "learning_rate": 2.971014492753623e-05,
-      "loss": 0.3494,
+      "loss": 0.3498,
       "step": 1120
     },
     {
       "epoch": 1.2282608695652173,
-      "grad_norm": 1.3232239484786987,
+      "grad_norm": 1.3294016122817993,
       "learning_rate": 2.9528985507246375e-05,
-      "loss": 0.3483,
+      "loss": 0.3492,
       "step": 1130
     },
     {
       "epoch": 1.2391304347826086,
-      "grad_norm": 1.3333462476730347,
+      "grad_norm": 1.3786628246307373,
       "learning_rate": 2.9347826086956526e-05,
-      "loss": 0.3571,
+      "loss": 0.3578,
       "step": 1140
     },
     {
       "epoch": 1.25,
-      "grad_norm": 1.1727484464645386,
+      "grad_norm": 1.1772373914718628,
       "learning_rate": 2.916666666666667e-05,
-      "loss": 0.3574,
+      "loss": 0.3576,
       "step": 1150
     },
     {
       "epoch": 1.25,
-      "eval_loss": 0.5970553159713745,
-      "eval_runtime": 10.3653,
-      "eval_samples_per_second": 46.308,
-      "eval_steps_per_second": 2.894,
+      "eval_loss": 0.5932352542877197,
+      "eval_runtime": 10.4231,
+      "eval_samples_per_second": 46.052,
+      "eval_steps_per_second": 2.878,
       "step": 1150
     },
     {
       "epoch": 1.2608695652173914,
-      "grad_norm": 1.3021057844161987,
+      "grad_norm": 1.2964465618133545,
       "learning_rate": 2.8985507246376814e-05,
-      "loss": 0.3541,
+      "loss": 0.3549,
       "step": 1160
     },
     {
       "epoch": 1.2717391304347827,
-      "grad_norm": 1.1331459283828735,
+      "grad_norm": 1.1511178016662598,
       "learning_rate": 2.8804347826086957e-05,
-      "loss": 0.3403,
+      "loss": 0.3404,
       "step": 1170
     },
     {
       "epoch": 1.2826086956521738,
-      "grad_norm": 1.0406198501586914,
+      "grad_norm": 1.028828501701355,
       "learning_rate": 2.86231884057971e-05,
-      "loss": 0.3454,
+      "loss": 0.3457,
       "step": 1180
     },
     {
       "epoch": 1.2934782608695652,
-      "grad_norm": 1.2329171895980835,
+      "grad_norm": 1.2443244457244873,
       "learning_rate": 2.8442028985507245e-05,
-      "loss": 0.3389,
+      "loss": 0.3388,
       "step": 1190
     },
     {
       "epoch": 1.3043478260869565,
-      "grad_norm": 1.1532611846923828,
+      "grad_norm": 1.1428587436676025,
       "learning_rate": 2.826086956521739e-05,
-      "loss": 0.3347,
+      "loss": 0.3348,
       "step": 1200
     },
     {
       "epoch": 1.3043478260869565,
-      "eval_loss": 0.6043132543563843,
-      "eval_runtime": 10.3783,
-      "eval_samples_per_second": 46.25,
-      "eval_steps_per_second": 2.891,
+      "eval_loss": 0.6009512543678284,
+      "eval_runtime": 10.4138,
+      "eval_samples_per_second": 46.093,
+      "eval_steps_per_second": 2.881,
       "step": 1200
     },
     {
       "epoch": 1.315217391304348,
-      "grad_norm": 1.2748662233352661,
+      "grad_norm": 1.2716634273529053,
       "learning_rate": 2.807971014492754e-05,
-      "loss": 0.3421,
+      "loss": 0.3419,
       "step": 1210
     },
     {
       "epoch": 1.3260869565217392,
-      "grad_norm": 1.2641656398773193,
+      "grad_norm": 1.2348277568817139,
       "learning_rate": 2.7898550724637683e-05,
-      "loss": 0.3545,
+      "loss": 0.3541,
       "step": 1220
     },
     {
       "epoch": 1.3369565217391304,
-      "grad_norm": 1.2328797578811646,
+      "grad_norm": 1.2257776260375977,
       "learning_rate": 2.7717391304347827e-05,
-      "loss": 0.3466,
+      "loss": 0.3464,
       "step": 1230
     },
     {
       "epoch": 1.3478260869565217,
-      "grad_norm": 1.4312686920166016,
+      "grad_norm": 1.3694759607315063,
       "learning_rate": 2.753623188405797e-05,
-      "loss": 0.3442,
+      "loss": 0.3439,
       "step": 1240
     },
     {
       "epoch": 1.358695652173913,
-      "grad_norm": 1.24868905544281,
+      "grad_norm": 1.243914246559143,
       "learning_rate": 2.7355072463768118e-05,
-      "loss": 0.3442,
+      "loss": 0.3439,
       "step": 1250
     },
     {
       "epoch": 1.358695652173913,
-      "eval_loss": 0.596993625164032,
-      "eval_runtime": 10.3794,
-      "eval_samples_per_second": 46.245,
+      "eval_loss": 0.5941683650016785,
+      "eval_runtime": 10.3819,
+      "eval_samples_per_second": 46.234,
       "eval_steps_per_second": 2.89,
       "step": 1250
     },
     {
       "epoch": 1.3695652173913042,
-      "grad_norm": 1.4520926475524902,
+      "grad_norm": 1.4521154165267944,
       "learning_rate": 2.7173913043478262e-05,
-      "loss": 0.3379,
+      "loss": 0.3383,
       "step": 1260
     },
     {
       "epoch": 1.3804347826086958,
-      "grad_norm": 1.1851826906204224,
+      "grad_norm": 1.1729083061218262,
       "learning_rate": 2.6992753623188406e-05,
       "loss": 0.3578,
       "step": 1270
     },
     {
       "epoch": 1.391304347826087,
-      "grad_norm": 1.172242522239685,
+      "grad_norm": 1.1995911598205566,
       "learning_rate": 2.6811594202898553e-05,
-      "loss": 0.3298,
+      "loss": 0.3304,
       "step": 1280
     },
     {
       "epoch": 1.4021739130434783,
-      "grad_norm": 1.0888515710830688,
+      "grad_norm": 1.0937422513961792,
       "learning_rate": 2.66304347826087e-05,
-      "loss": 0.329,
+      "loss": 0.3303,
       "step": 1290
     },
     {
       "epoch": 1.4130434782608696,
-      "grad_norm": 1.2789475917816162,
+      "grad_norm": 1.291451334953308,
       "learning_rate": 2.6449275362318844e-05,
-      "loss": 0.3404,
+      "loss": 0.3399,
       "step": 1300
     },
     {
       "epoch": 1.4130434782608696,
-      "eval_loss": 0.6022054553031921,
-      "eval_runtime": 10.3983,
-      "eval_samples_per_second": 46.162,
-      "eval_steps_per_second": 2.885,
+      "eval_loss": 0.5981742143630981,
+      "eval_runtime": 10.3912,
+      "eval_samples_per_second": 46.193,
+      "eval_steps_per_second": 2.887,
       "step": 1300
     },
     {
       "epoch": 1.4239130434782608,
-      "grad_norm": 1.3096864223480225,
+      "grad_norm": 1.3131028413772583,
       "learning_rate": 2.6268115942028988e-05,
-      "loss": 0.342,
+      "loss": 0.3414,
       "step": 1310
     },
     {
       "epoch": 1.434782608695652,
-      "grad_norm": 1.1879781484603882,
+      "grad_norm": 1.1776825189590454,
       "learning_rate": 2.608695652173913e-05,
-      "loss": 0.3295,
+      "loss": 0.3297,
       "step": 1320
     },
     {
       "epoch": 1.4456521739130435,
-      "grad_norm": 1.3001481294631958,
+      "grad_norm": 1.3045450448989868,
       "learning_rate": 2.5905797101449275e-05,
-      "loss": 0.3325,
+      "loss": 0.3331,
       "step": 1330
     },
     {
       "epoch": 1.4565217391304348,
-      "grad_norm": 1.2701698541641235,
+      "grad_norm": 1.265739917755127,
       "learning_rate": 2.572463768115942e-05,
-      "loss": 0.3441,
+      "loss": 0.3443,
       "step": 1340
     },
     {
       "epoch": 1.4673913043478262,
-      "grad_norm": 1.180391550064087,
+      "grad_norm": 1.1657543182373047,
       "learning_rate": 2.554347826086957e-05,
-      "loss": 0.3278,
+      "loss": 0.3287,
       "step": 1350
     },
     {
       "epoch": 1.4673913043478262,
-      "eval_loss": 0.6003170609474182,
-      "eval_runtime": 10.4048,
-      "eval_samples_per_second": 46.132,
-      "eval_steps_per_second": 2.883,
+      "eval_loss": 0.5955829620361328,
+      "eval_runtime": 10.363,
+      "eval_samples_per_second": 46.319,
+      "eval_steps_per_second": 2.895,
       "step": 1350
     },
     {
       "epoch": 1.4782608695652173,
-      "grad_norm": 1.4365824460983276,
+      "grad_norm": 1.4320471286773682,
       "learning_rate": 2.5362318840579714e-05,
-      "loss": 0.328,
+      "loss": 0.3282,
       "step": 1360
     },
     {
       "epoch": 1.4891304347826086,
-      "grad_norm": 1.1897424459457397,
+      "grad_norm": 1.20095694065094,
       "learning_rate": 2.5181159420289857e-05,
-      "loss": 0.35,
+      "loss": 0.351,
       "step": 1370
     },
     {
       "epoch": 1.5,
-      "grad_norm": 1.2596850395202637,
+      "grad_norm": 1.2566215991973877,
       "learning_rate": 2.5e-05,
-      "loss": 0.362,
+      "loss": 0.3627,
       "step": 1380
     },
     {
       "epoch": 1.5108695652173914,
-      "grad_norm": 1.1584769487380981,
+      "grad_norm": 1.1641753911972046,
       "learning_rate": 2.4818840579710145e-05,
-      "loss": 0.332,
+      "loss": 0.3312,
       "step": 1390
     },
     {
       "epoch": 1.5217391304347827,
-      "grad_norm": 1.2620126008987427,
+      "grad_norm": 1.2655019760131836,
       "learning_rate": 2.4637681159420292e-05,
-      "loss": 0.3317,
+      "loss": 0.3319,
       "step": 1400
     },
     {
       "epoch": 1.5217391304347827,
-      "eval_loss": 0.6029025912284851,
-      "eval_runtime": 10.3897,
-      "eval_samples_per_second": 46.2,
-      "eval_steps_per_second": 2.887,
+      "eval_loss": 0.5990059971809387,
+      "eval_runtime": 10.3839,
+      "eval_samples_per_second": 46.225,
+      "eval_steps_per_second": 2.889,
       "step": 1400
     },
     {
       "epoch": 1.5326086956521738,
-      "grad_norm": 1.4026389122009277,
+      "grad_norm": 1.402687430381775,
       "learning_rate": 2.4456521739130436e-05,
-      "loss": 0.3345,
+      "loss": 0.3348,
       "step": 1410
     },
     {
       "epoch": 1.5434782608695652,
-      "grad_norm": 1.2000699043273926,
+      "grad_norm": 1.1932505369186401,
       "learning_rate": 2.427536231884058e-05,
-      "loss": 0.328,
+      "loss": 0.3282,
       "step": 1420
     },
     {
       "epoch": 1.5543478260869565,
-      "grad_norm": 1.1770294904708862,
+      "grad_norm": 1.1783493757247925,
       "learning_rate": 2.4094202898550724e-05,
-      "loss": 0.3226,
+      "loss": 0.3219,
       "step": 1430
     },
     {
       "epoch": 1.5652173913043477,
-      "grad_norm": 1.472050666809082,
+      "grad_norm": 1.4830645322799683,
       "learning_rate": 2.391304347826087e-05,
-      "loss": 0.3261,
+      "loss": 0.3267,
       "step": 1440
     },
     {
       "epoch": 1.5760869565217392,
-      "grad_norm": 1.2399369478225708,
+      "grad_norm": 1.2483339309692383,
       "learning_rate": 2.3731884057971015e-05,
-      "loss": 0.3259,
+      "loss": 0.3256,
       "step": 1450
     },
     {
       "epoch": 1.5760869565217392,
-      "eval_loss": 0.6003721356391907,
-      "eval_runtime": 10.3879,
-      "eval_samples_per_second": 46.207,
-      "eval_steps_per_second": 2.888,
+      "eval_loss": 0.5974804162979126,
+      "eval_runtime": 10.4146,
+      "eval_samples_per_second": 46.089,
+      "eval_steps_per_second": 2.881,
       "step": 1450
     },
     {
       "epoch": 1.5869565217391304,
-      "grad_norm": 1.2852166891098022,
+      "grad_norm": 1.2872872352600098,
       "learning_rate": 2.355072463768116e-05,
-      "loss": 0.3352,
+      "loss": 0.3356,
       "step": 1460
     },
     {
       "epoch": 1.5978260869565217,
-      "grad_norm": 1.3454734086990356,
+      "grad_norm": 1.3353910446166992,
       "learning_rate": 2.3369565217391306e-05,
-      "loss": 0.3244,
+      "loss": 0.324,
       "step": 1470
     },
     {
       "epoch": 1.608695652173913,
-      "grad_norm": 1.4294050931930542,
+      "grad_norm": 1.4367629289627075,
       "learning_rate": 2.318840579710145e-05,
-      "loss": 0.3224,
+      "loss": 0.3218,
       "step": 1480
     },
     {
       "epoch": 1.6195652173913042,
-      "grad_norm": 1.2381614446640015,
+      "grad_norm": 1.2451330423355103,
       "learning_rate": 2.3007246376811593e-05,
-      "loss": 0.3369,
+      "loss": 0.3379,
       "step": 1490
     },
     {
       "epoch": 1.6304347826086958,
-      "grad_norm": 1.2527271509170532,
+      "grad_norm": 1.2464361190795898,
       "learning_rate": 2.282608695652174e-05,
       "loss": 0.3366,
       "step": 1500
     },
     {
       "epoch": 1.6304347826086958,
-      "eval_loss": 0.602122962474823,
-      "eval_runtime": 10.3547,
-      "eval_samples_per_second": 46.356,
-      "eval_steps_per_second": 2.897,
+      "eval_loss": 0.5991224646568298,
+      "eval_runtime": 10.469,
+      "eval_samples_per_second": 45.85,
+      "eval_steps_per_second": 2.866,
       "step": 1500
     },
     {
       "epoch": 1.641304347826087,
-      "grad_norm": 1.1952861547470093,
+      "grad_norm": 1.1958304643630981,
       "learning_rate": 2.2644927536231884e-05,
-      "loss": 0.3304,
+      "loss": 0.3308,
       "step": 1510
     },
     {
       "epoch": 1.6521739130434783,
-      "grad_norm": 1.198760747909546,
+      "grad_norm": 1.2368805408477783,
       "learning_rate": 2.246376811594203e-05,
       "loss": 0.3239,
       "step": 1520
     },
     {
       "epoch": 1.6630434782608696,
-      "grad_norm": 1.2754151821136475,
+      "grad_norm": 1.3021612167358398,
       "learning_rate": 2.2282608695652175e-05,
-      "loss": 0.3243,
+      "loss": 0.3247,
       "step": 1530
     },
     {
       "epoch": 1.6739130434782608,
-      "grad_norm": 1.3670730590820312,
+      "grad_norm": 1.3558640480041504,
       "learning_rate": 2.2101449275362323e-05,
-      "loss": 0.3212,
+      "loss": 0.3207,
       "step": 1540
     },
     {
       "epoch": 1.6847826086956523,
-      "grad_norm": 1.2905199527740479,
+      "grad_norm": 1.2872494459152222,
       "learning_rate": 2.1920289855072466e-05,
-      "loss": 0.3356,
+      "loss": 0.335,
       "step": 1550
     },
     {
       "epoch": 1.6847826086956523,
-      "eval_loss": 0.6040811538696289,
-      "eval_runtime": 10.345,
-      "eval_samples_per_second": 46.399,
-      "eval_steps_per_second": 2.9,
+      "eval_loss": 0.6015117764472961,
+      "eval_runtime": 10.4359,
+      "eval_samples_per_second": 45.995,
+      "eval_steps_per_second": 2.875,
       "step": 1550
     },
     {
       "epoch": 1.6956521739130435,
-      "grad_norm": 1.2227987051010132,
+      "grad_norm": 1.2460740804672241,
       "learning_rate": 2.173913043478261e-05,
-      "loss": 0.32,
+      "loss": 0.3205,
       "step": 1560
     },
     {
       "epoch": 1.7065217391304348,
-      "grad_norm": 1.4033117294311523,
+      "grad_norm": 1.516802430152893,
       "learning_rate": 2.1557971014492757e-05,
-      "loss": 0.3389,
+      "loss": 0.3394,
       "step": 1570
     },
     {
       "epoch": 1.7173913043478262,
-      "grad_norm": 1.192590594291687,
+      "grad_norm": 1.2020652294158936,
       "learning_rate": 2.13768115942029e-05,
-      "loss": 0.318,
+      "loss": 0.3194,
       "step": 1580
     },
     {
       "epoch": 1.7282608695652173,
-      "grad_norm": 1.4258476495742798,
+      "grad_norm": 1.426993727684021,
       "learning_rate": 2.1195652173913045e-05,
-      "loss": 0.3223,
+      "loss": 0.3218,
       "step": 1590
     },
     {
       "epoch": 1.7391304347826086,
-      "grad_norm": 1.2608294486999512,
+      "grad_norm": 1.2477283477783203,
       "learning_rate": 2.101449275362319e-05,
-      "loss": 0.3245,
+      "loss": 0.3252,
       "step": 1600
     },
     {
       "epoch": 1.7391304347826086,
-      "eval_loss": 0.6098019480705261,
-      "eval_runtime": 10.349,
-      "eval_samples_per_second": 46.381,
-      "eval_steps_per_second": 2.899,
+      "eval_loss": 0.6045735478401184,
+      "eval_runtime": 10.4421,
+      "eval_samples_per_second": 45.968,
+      "eval_steps_per_second": 2.873,
       "step": 1600
     },
     {
       "epoch": 1.75,
-      "grad_norm": 1.2881015539169312,
+      "grad_norm": 1.2844955921173096,
       "learning_rate": 2.0833333333333336e-05,
-      "loss": 0.2941,
+      "loss": 0.2933,
       "step": 1610
     },
     {
       "epoch": 1.7608695652173914,
-      "grad_norm": 1.3477824926376343,
+      "grad_norm": 1.3252885341644287,
       "learning_rate": 2.065217391304348e-05,
-      "loss": 0.3086,
+      "loss": 0.3081,
       "step": 1620
     },
     {
       "epoch": 1.7717391304347827,
-      "grad_norm": 1.5044372081756592,
+      "grad_norm": 1.5310088396072388,
       "learning_rate": 2.0471014492753624e-05,
-      "loss": 0.3111,
+      "loss": 0.3121,
       "step": 1630
     },
     {
       "epoch": 1.7826086956521738,
-      "grad_norm": 1.378635048866272,
+      "grad_norm": 1.3772246837615967,
       "learning_rate": 2.028985507246377e-05,
-      "loss": 0.3182,
+      "loss": 0.3186,
       "step": 1640
     },
     {
       "epoch": 1.7934782608695652,
-      "grad_norm": 1.3590328693389893,
+      "grad_norm": 1.3666644096374512,
       "learning_rate": 2.0108695652173915e-05,
-      "loss": 0.3255,
+      "loss": 0.3262,
       "step": 1650
     },
     {
       "epoch": 1.7934782608695652,
-      "eval_loss": 0.6043348908424377,
-      "eval_runtime": 10.3437,
-      "eval_samples_per_second": 46.405,
-      "eval_steps_per_second": 2.9,
+      "eval_loss": 0.6000680923461914,
+      "eval_runtime": 10.4274,
+      "eval_samples_per_second": 46.033,
+      "eval_steps_per_second": 2.877,
       "step": 1650
     },
     {
       "epoch": 1.8043478260869565,
-      "grad_norm": 1.3664731979370117,
+      "grad_norm": 1.369613528251648,
       "learning_rate": 1.992753623188406e-05,
-      "loss": 0.3164,
+      "loss": 0.3157,
       "step": 1660
     },
     {
       "epoch": 1.8152173913043477,
-      "grad_norm": 1.4544974565505981,
+      "grad_norm": 1.4644464254379272,
       "learning_rate": 1.9746376811594202e-05,
-      "loss": 0.3263,
+      "loss": 0.3258,
       "step": 1670
     },
     {
       "epoch": 1.8260869565217392,
-      "grad_norm": 1.548271656036377,
+      "grad_norm": 1.5432722568511963,
       "learning_rate": 1.956521739130435e-05,
-      "loss": 0.2951,
+      "loss": 0.2959,
       "step": 1680
     },
     {
       "epoch": 1.8369565217391304,
-      "grad_norm": 1.4348987340927124,
+      "grad_norm": 1.4335534572601318,
       "learning_rate": 1.9384057971014493e-05,
-      "loss": 0.3357,
+      "loss": 0.3365,
       "step": 1690
     },
     {
       "epoch": 1.8478260869565217,
-      "grad_norm": 1.4023321866989136,
+      "grad_norm": 1.4229481220245361,
       "learning_rate": 1.9202898550724637e-05,
-      "loss": 0.3119,
+      "loss": 0.3129,
       "step": 1700
     },
     {
       "epoch": 1.8478260869565217,
-      "eval_loss": 0.6111882328987122,
-      "eval_runtime": 10.3465,
-      "eval_samples_per_second": 46.393,
-      "eval_steps_per_second": 2.9,
+      "eval_loss": 0.6060259342193604,
+      "eval_runtime": 10.4283,
+      "eval_samples_per_second": 46.028,
+      "eval_steps_per_second": 2.877,
       "step": 1700
     },
     {
       "epoch": 1.858695652173913,
-      "grad_norm": 1.2122089862823486,
+      "grad_norm": 1.2169373035430908,
       "learning_rate": 1.9021739130434784e-05,
-      "loss": 0.3174,
+      "loss": 0.3166,
       "step": 1710
     },
     {
       "epoch": 1.8695652173913042,
-      "grad_norm": 1.4489362239837646,
+      "grad_norm": 1.4397977590560913,
       "learning_rate": 1.8840579710144928e-05,
-      "loss": 0.3106,
+      "loss": 0.3113,
       "step": 1720
     },
     {
       "epoch": 1.8804347826086958,
-      "grad_norm": 1.290483832359314,
+      "grad_norm": 1.2902106046676636,
       "learning_rate": 1.8659420289855072e-05,
-      "loss": 0.3081,
+      "loss": 0.3073,
       "step": 1730
     },
     {
       "epoch": 1.891304347826087,
-      "grad_norm": 1.4748475551605225,
+      "grad_norm": 1.5002868175506592,
       "learning_rate": 1.8478260869565216e-05,
-      "loss": 0.3128,
+      "loss": 0.313,
       "step": 1740
     },
     {
       "epoch": 1.9021739130434783,
-      "grad_norm": 1.3447638750076294,
+      "grad_norm": 1.3328644037246704,
       "learning_rate": 1.8297101449275363e-05,
-      "loss": 0.3228,
+      "loss": 0.3235,
       "step": 1750
     },
     {
       "epoch": 1.9021739130434783,
-      "eval_loss": 0.613942563533783,
-      "eval_runtime": 10.3587,
-      "eval_samples_per_second": 46.338,
-      "eval_steps_per_second": 2.896,
+      "eval_loss": 0.6095381379127502,
+      "eval_runtime": 10.4147,
+      "eval_samples_per_second": 46.089,
+      "eval_steps_per_second": 2.881,
       "step": 1750
     },
     {
       "epoch": 1.9130434782608696,
-      "grad_norm": 1.367020606994629,
+      "grad_norm": 1.3620116710662842,
       "learning_rate": 1.8115942028985507e-05,
-      "loss": 0.303,
+      "loss": 0.3024,
       "step": 1760
     },
     {
       "epoch": 1.9239130434782608,
-      "grad_norm": 1.382914423942566,
+      "grad_norm": 1.3795499801635742,
       "learning_rate": 1.793478260869565e-05,
-      "loss": 0.3162,
+      "loss": 0.317,
       "step": 1770
     },
     {
       "epoch": 1.9347826086956523,
-      "grad_norm": 1.361312985420227,
+      "grad_norm": 1.3702704906463623,
       "learning_rate": 1.7753623188405798e-05,
-      "loss": 0.319,
+      "loss": 0.3193,
       "step": 1780
     },
     {
       "epoch": 1.9456521739130435,
-      "grad_norm": 1.3756697177886963,
+      "grad_norm": 1.372766137123108,
       "learning_rate": 1.757246376811594e-05,
-      "loss": 0.2994,
+      "loss": 0.2997,
       "step": 1790
     },
     {
       "epoch": 1.9565217391304348,
-      "grad_norm": 1.2937418222427368,
+      "grad_norm": 1.3204894065856934,
       "learning_rate": 1.739130434782609e-05,
-      "loss": 0.3101,
+      "loss": 0.3104,
       "step": 1800
     },
     {
       "epoch": 1.9565217391304348,
-      "eval_loss": 0.6147708892822266,
-      "eval_runtime": 10.368,
-      "eval_samples_per_second": 46.296,
-      "eval_steps_per_second": 2.894,
+      "eval_loss": 0.6124075651168823,
+      "eval_runtime": 10.3987,
+      "eval_samples_per_second": 46.16,
+      "eval_steps_per_second": 2.885,
       "step": 1800
     },
     {
       "epoch": 1.9673913043478262,
-      "grad_norm": 1.356997013092041,
+      "grad_norm": 1.3335340023040771,
       "learning_rate": 1.7210144927536233e-05,
-      "loss": 0.2974,
+      "loss": 0.2981,
       "step": 1810
     },
     {
       "epoch": 1.9782608695652173,
-      "grad_norm": 1.4485703706741333,
+      "grad_norm": 1.455363154411316,
       "learning_rate": 1.702898550724638e-05,
-      "loss": 0.3008,
+      "loss": 0.3011,
       "step": 1820
     },
     {
       "epoch": 1.9891304347826086,
-      "grad_norm": 1.4469425678253174,
+      "grad_norm": 1.4685810804367065,
       "learning_rate": 1.6847826086956524e-05,
-      "loss": 0.3061,
+      "loss": 0.3068,
       "step": 1830
     },
     {
       "epoch": 2.0,
-      "grad_norm": 3.6113595962524414,
+      "grad_norm": 3.5936105251312256,
       "learning_rate": 1.6666666666666667e-05,
-      "loss": 0.3213,
+      "loss": 0.321,
       "step": 1840
     },
     {
       "epoch": 2.010869565217391,
-      "grad_norm": 1.3155393600463867,
+      "grad_norm": 1.3150142431259155,
       "learning_rate": 1.6485507246376815e-05,
-      "loss": 0.28,
+      "loss": 0.2801,
       "step": 1850
     },
     {
       "epoch": 2.010869565217391,
-      "eval_loss": 0.6274634003639221,
-      "eval_runtime": 10.386,
-      "eval_samples_per_second": 46.216,
-      "eval_steps_per_second": 2.889,
+      "eval_loss": 0.6250590085983276,
+      "eval_runtime": 10.374,
+      "eval_samples_per_second": 46.269,
+      "eval_steps_per_second": 2.892,
       "step": 1850
     },
     {
       "epoch": 2.0217391304347827,
-      "grad_norm": 1.4573590755462646,
+      "grad_norm": 1.4412224292755127,
       "learning_rate": 1.630434782608696e-05,
-      "loss": 0.2688,
+      "loss": 0.2687,
       "step": 1860
     },
     {
       "epoch": 2.032608695652174,
-      "grad_norm": 1.3323371410369873,
+      "grad_norm": 1.3438794612884521,
       "learning_rate": 1.6123188405797102e-05,
-      "loss": 0.2653,
+      "loss": 0.2651,
       "step": 1870
     },
     {
       "epoch": 2.0434782608695654,
-      "grad_norm": 1.4544154405593872,
+      "grad_norm": 1.4525725841522217,
       "learning_rate": 1.5942028985507246e-05,
-      "loss": 0.2729,
+      "loss": 0.2734,
       "step": 1880
     },
     {
       "epoch": 2.0543478260869565,
-      "grad_norm": 1.453372597694397,
+      "grad_norm": 1.4576469659805298,
       "learning_rate": 1.5760869565217393e-05,
-      "loss": 0.2531,
+      "loss": 0.2527,
       "step": 1890
     },
     {
       "epoch": 2.0652173913043477,
-      "grad_norm": 1.5393754243850708,
+      "grad_norm": 1.5585767030715942,
       "learning_rate": 1.5579710144927537e-05,
-      "loss": 0.2799,
+      "loss": 0.2797,
       "step": 1900
     },
     {
       "epoch": 2.0652173913043477,
-      "eval_loss": 0.654403805732727,
-      "eval_runtime": 10.3912,
-      "eval_samples_per_second": 46.193,
-      "eval_steps_per_second": 2.887,
+      "eval_loss": 0.6516745090484619,
+      "eval_runtime": 10.3702,
+      "eval_samples_per_second": 46.287,
+      "eval_steps_per_second": 2.893,
       "step": 1900
     }
   ],

checkpoint-1900/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:40ebe6ec892e02f02c5df3d2f37d3f9302b8a2381547d3bdf4d52f5e045e3cdc
+oid sha256:a3712c5fdd6e5241e8ea26ae62b28612fc514ec88e8a03bc6c1b3a46acfc611a
 size 5816