{
  "best_global_step": 10026,
  "best_metric": 2.109537124633789,
  "best_model_checkpoint": "./sft_output_SimpleQA/checkpoint-10026",
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 10026,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.002992220227408737,
      "grad_norm": 215.0,
      "learning_rate": 4.9955116696588873e-05,
      "loss": 6.5961,
      "mean_token_accuracy": 0.14261817783117295,
      "num_tokens": 4118.0,
      "step": 10
    },
    {
      "epoch": 0.005984440454817474,
      "grad_norm": 28.5,
      "learning_rate": 4.990524635946539e-05,
      "loss": 6.8891,
      "mean_token_accuracy": 0.023234939482063056,
      "num_tokens": 10291.0,
      "step": 20
    },
    {
      "epoch": 0.008976660682226212,
      "grad_norm": 34.25,
      "learning_rate": 4.9855376022341915e-05,
      "loss": 6.4879,
      "mean_token_accuracy": 0.020476288627833128,
      "num_tokens": 14655.0,
      "step": 30
    },
    {
      "epoch": 0.011968880909634948,
      "grad_norm": 35.0,
      "learning_rate": 4.980550568521843e-05,
      "loss": 6.5951,
      "mean_token_accuracy": 0.021896280348300934,
      "num_tokens": 17680.0,
      "step": 40
    },
    {
      "epoch": 0.014961101137043686,
      "grad_norm": 40.75,
      "learning_rate": 4.975563534809496e-05,
      "loss": 6.5836,
      "mean_token_accuracy": 0.027274718414992093,
      "num_tokens": 20903.0,
      "step": 50
    },
    {
      "epoch": 0.017953321364452424,
      "grad_norm": 62.5,
      "learning_rate": 4.970576501097148e-05,
      "loss": 6.4953,
      "mean_token_accuracy": 0.036431427439674736,
      "num_tokens": 25889.0,
      "step": 60
    },
    {
      "epoch": 0.020945541591861162,
      "grad_norm": 20.75,
      "learning_rate": 4.9655894673848e-05,
      "loss": 6.3477,
      "mean_token_accuracy": 0.038724389113485816,
      "num_tokens": 29097.0,
      "step": 70
    },
    {
      "epoch": 0.023937761819269897,
      "grad_norm": 28.0,
      "learning_rate": 4.9606024336724516e-05,
      "loss": 6.3711,
      "mean_token_accuracy": 0.0324508004821837,
      "num_tokens": 33201.0,
      "step": 80
    },
    {
      "epoch": 0.026929982046678635,
      "grad_norm": 30.0,
      "learning_rate": 4.955615399960104e-05,
      "loss": 6.0164,
      "mean_token_accuracy": 0.049895737878978254,
      "num_tokens": 37509.0,
      "step": 90
    },
    {
      "epoch": 0.029922202274087373,
      "grad_norm": 14.875,
      "learning_rate": 4.950628366247756e-05,
      "loss": 6.218,
      "mean_token_accuracy": 0.0485645480453968,
      "num_tokens": 41623.0,
      "step": 100
    },
    {
      "epoch": 0.03291442250149611,
      "grad_norm": 25.375,
      "learning_rate": 4.945641332535408e-05,
      "loss": 6.014,
      "mean_token_accuracy": 0.05698083452880383,
      "num_tokens": 46601.0,
      "step": 110
    },
    {
      "epoch": 0.03590664272890485,
      "grad_norm": 20.5,
      "learning_rate": 4.9406542988230606e-05,
      "loss": 5.9919,
      "mean_token_accuracy": 0.05561460051685572,
      "num_tokens": 50342.0,
      "step": 120
    },
    {
      "epoch": 0.03889886295631358,
      "grad_norm": 20.875,
      "learning_rate": 4.935667265110712e-05,
      "loss": 6.166,
      "mean_token_accuracy": 0.04003935307264328,
      "num_tokens": 55021.0,
      "step": 130
    },
    {
      "epoch": 0.041891083183722325,
      "grad_norm": 22.375,
      "learning_rate": 4.930680231398365e-05,
      "loss": 6.0082,
      "mean_token_accuracy": 0.05440298151224852,
      "num_tokens": 58986.0,
      "step": 140
    },
    {
      "epoch": 0.04488330341113106,
      "grad_norm": 22.875,
      "learning_rate": 4.9256931976860165e-05,
      "loss": 5.9211,
      "mean_token_accuracy": 0.07408283576369286,
      "num_tokens": 62081.0,
      "step": 150
    },
    {
      "epoch": 0.047875523638539794,
      "grad_norm": 30.5,
      "learning_rate": 4.920706163973668e-05,
      "loss": 5.6951,
      "mean_token_accuracy": 0.10848762802779674,
      "num_tokens": 64439.0,
      "step": 160
    },
    {
      "epoch": 0.050867743865948535,
      "grad_norm": 15.4375,
      "learning_rate": 4.9157191302613206e-05,
      "loss": 5.9625,
      "mean_token_accuracy": 0.06950367838144303,
      "num_tokens": 68866.0,
      "step": 170
    },
    {
      "epoch": 0.05385996409335727,
      "grad_norm": 43.75,
      "learning_rate": 4.910732096548973e-05,
      "loss": 5.8368,
      "mean_token_accuracy": 0.07707072868943214,
      "num_tokens": 73615.0,
      "step": 180
    },
    {
      "epoch": 0.05685218432076601,
      "grad_norm": 32.75,
      "learning_rate": 4.9057450628366255e-05,
      "loss": 5.3259,
      "mean_token_accuracy": 0.1262487217783928,
      "num_tokens": 78851.0,
      "step": 190
    },
    {
      "epoch": 0.059844404548174746,
      "grad_norm": 28.125,
      "learning_rate": 4.900758029124277e-05,
      "loss": 5.7784,
      "mean_token_accuracy": 0.08831716366112233,
      "num_tokens": 82971.0,
      "step": 200
    },
    {
      "epoch": 0.06283662477558348,
      "grad_norm": 20.75,
      "learning_rate": 4.895770995411929e-05,
      "loss": 5.7123,
      "mean_token_accuracy": 0.0892865601927042,
      "num_tokens": 87170.0,
      "step": 210
    },
    {
      "epoch": 0.06582884500299221,
      "grad_norm": 37.5,
      "learning_rate": 4.8907839616995814e-05,
      "loss": 5.5482,
      "mean_token_accuracy": 0.09651436358690262,
      "num_tokens": 93105.0,
      "step": 220
    },
    {
      "epoch": 0.06882106523040096,
      "grad_norm": 22.125,
      "learning_rate": 4.885796927987233e-05,
      "loss": 5.1698,
      "mean_token_accuracy": 0.1460722029209137,
      "num_tokens": 97170.0,
      "step": 230
    },
    {
      "epoch": 0.0718132854578097,
      "grad_norm": 76.5,
      "learning_rate": 4.8808098942748855e-05,
      "loss": 5.244,
      "mean_token_accuracy": 0.12373836413025856,
      "num_tokens": 101830.0,
      "step": 240
    },
    {
      "epoch": 0.07480550568521843,
      "grad_norm": 16.875,
      "learning_rate": 4.875822860562538e-05,
      "loss": 5.207,
      "mean_token_accuracy": 0.12196040600538254,
      "num_tokens": 106063.0,
      "step": 250
    },
    {
      "epoch": 0.07779772591262717,
      "grad_norm": 21.25,
      "learning_rate": 4.87083582685019e-05,
      "loss": 5.2538,
      "mean_token_accuracy": 0.12483496218919754,
      "num_tokens": 110455.0,
      "step": 260
    },
    {
      "epoch": 0.0807899461400359,
      "grad_norm": 30.25,
      "learning_rate": 4.8658487931378414e-05,
      "loss": 5.0607,
      "mean_token_accuracy": 0.1395721070468426,
      "num_tokens": 114231.0,
      "step": 270
    },
    {
      "epoch": 0.08378216636744465,
      "grad_norm": 17.875,
      "learning_rate": 4.860861759425494e-05,
      "loss": 5.4699,
      "mean_token_accuracy": 0.11910050511360168,
      "num_tokens": 119683.0,
      "step": 280
    },
    {
      "epoch": 0.08677438659485338,
      "grad_norm": 28.25,
      "learning_rate": 4.855874725713146e-05,
      "loss": 5.1832,
      "mean_token_accuracy": 0.13222006037831308,
      "num_tokens": 123255.0,
      "step": 290
    },
    {
      "epoch": 0.08976660682226212,
      "grad_norm": 27.75,
      "learning_rate": 4.850887692000799e-05,
      "loss": 4.8555,
      "mean_token_accuracy": 0.14820684865117073,
      "num_tokens": 127933.0,
      "step": 300
    },
    {
      "epoch": 0.09275882704967085,
      "grad_norm": 23.0,
      "learning_rate": 4.8459006582884504e-05,
      "loss": 4.9349,
      "mean_token_accuracy": 0.15446224436163902,
      "num_tokens": 132134.0,
      "step": 310
    },
    {
      "epoch": 0.09575104727707959,
      "grad_norm": 26.25,
      "learning_rate": 4.840913624576102e-05,
      "loss": 5.1772,
      "mean_token_accuracy": 0.1377170652151108,
      "num_tokens": 136569.0,
      "step": 320
    },
    {
      "epoch": 0.09874326750448834,
      "grad_norm": 22.5,
      "learning_rate": 4.8359265908637546e-05,
      "loss": 5.3314,
      "mean_token_accuracy": 0.12001498714089394,
      "num_tokens": 140653.0,
      "step": 330
    },
    {
      "epoch": 0.10173548773189707,
      "grad_norm": 29.125,
      "learning_rate": 4.830939557151406e-05,
      "loss": 4.9681,
      "mean_token_accuracy": 0.135365604609251,
      "num_tokens": 145278.0,
      "step": 340
    },
    {
      "epoch": 0.1047277079593058,
      "grad_norm": 23.375,
      "learning_rate": 4.825952523439059e-05,
      "loss": 5.0282,
      "mean_token_accuracy": 0.15209860131144523,
      "num_tokens": 148698.0,
      "step": 350
    },
    {
      "epoch": 0.10771992818671454,
      "grad_norm": 21.0,
      "learning_rate": 4.820965489726711e-05,
      "loss": 5.1296,
      "mean_token_accuracy": 0.14033635556697846,
      "num_tokens": 152228.0,
      "step": 360
    },
    {
      "epoch": 0.11071214841412327,
      "grad_norm": 24.875,
      "learning_rate": 4.815978456014363e-05,
      "loss": 5.1174,
      "mean_token_accuracy": 0.14664285108447075,
      "num_tokens": 156663.0,
      "step": 370
    },
    {
      "epoch": 0.11370436864153202,
      "grad_norm": 21.625,
      "learning_rate": 4.8109914223020146e-05,
      "loss": 5.0752,
      "mean_token_accuracy": 0.16574262082576752,
      "num_tokens": 161542.0,
      "step": 380
    },
    {
      "epoch": 0.11669658886894076,
      "grad_norm": 21.625,
      "learning_rate": 4.806004388589667e-05,
      "loss": 4.8108,
      "mean_token_accuracy": 0.18466916903853417,
      "num_tokens": 165485.0,
      "step": 390
    },
    {
      "epoch": 0.11968880909634949,
      "grad_norm": 17.875,
      "learning_rate": 4.801017354877319e-05,
      "loss": 5.4408,
      "mean_token_accuracy": 0.12876051440834999,
      "num_tokens": 171666.0,
      "step": 400
    },
    {
      "epoch": 0.12268102932375823,
      "grad_norm": 19.125,
      "learning_rate": 4.796030321164971e-05,
      "loss": 5.2179,
      "mean_token_accuracy": 0.1634024627506733,
      "num_tokens": 177368.0,
      "step": 410
    },
    {
      "epoch": 0.12567324955116696,
      "grad_norm": 36.5,
      "learning_rate": 4.7910432874526236e-05,
      "loss": 5.0537,
      "mean_token_accuracy": 0.1408073700964451,
      "num_tokens": 182587.0,
      "step": 420
    },
    {
      "epoch": 0.1286654697785757,
      "grad_norm": 27.75,
      "learning_rate": 4.7860562537402754e-05,
      "loss": 4.9838,
      "mean_token_accuracy": 0.15434883609414102,
      "num_tokens": 187825.0,
      "step": 430
    },
    {
      "epoch": 0.13165769000598443,
      "grad_norm": 32.5,
      "learning_rate": 4.781069220027928e-05,
      "loss": 4.8094,
      "mean_token_accuracy": 0.1907724179327488,
      "num_tokens": 192757.0,
      "step": 440
    },
    {
      "epoch": 0.13464991023339318,
      "grad_norm": 30.75,
      "learning_rate": 4.7760821863155795e-05,
      "loss": 5.0489,
      "mean_token_accuracy": 0.14357075691223145,
      "num_tokens": 196335.0,
      "step": 450
    },
    {
      "epoch": 0.13764213046080193,
      "grad_norm": 33.75,
      "learning_rate": 4.771095152603231e-05,
      "loss": 5.0883,
      "mean_token_accuracy": 0.17117929756641387,
      "num_tokens": 200231.0,
      "step": 460
    },
    {
      "epoch": 0.14063435068821065,
      "grad_norm": 34.75,
      "learning_rate": 4.7661081188908844e-05,
      "loss": 4.7789,
      "mean_token_accuracy": 0.16136289834976197,
      "num_tokens": 204065.0,
      "step": 470
    },
    {
      "epoch": 0.1436265709156194,
      "grad_norm": 32.5,
      "learning_rate": 4.761121085178536e-05,
      "loss": 4.7431,
      "mean_token_accuracy": 0.1755128175020218,
      "num_tokens": 208508.0,
      "step": 480
    },
    {
      "epoch": 0.14661879114302812,
      "grad_norm": 23.125,
      "learning_rate": 4.7561340514661885e-05,
      "loss": 4.8372,
      "mean_token_accuracy": 0.18989472165703775,
      "num_tokens": 213285.0,
      "step": 490
    },
    {
      "epoch": 0.14961101137043686,
      "grad_norm": 30.875,
      "learning_rate": 4.75114701775384e-05,
      "loss": 4.6259,
      "mean_token_accuracy": 0.21336833536624908,
      "num_tokens": 218031.0,
      "step": 500
    },
    {
      "epoch": 0.1526032315978456,
      "grad_norm": 40.5,
      "learning_rate": 4.746159984041492e-05,
      "loss": 4.9149,
      "mean_token_accuracy": 0.16736119762063026,
      "num_tokens": 223481.0,
      "step": 510
    },
    {
      "epoch": 0.15559545182525433,
      "grad_norm": 20.125,
      "learning_rate": 4.7411729503291444e-05,
      "loss": 4.5618,
      "mean_token_accuracy": 0.20358971655368804,
      "num_tokens": 227469.0,
      "step": 520
    },
    {
      "epoch": 0.15858767205266308,
      "grad_norm": 32.0,
      "learning_rate": 4.736185916616797e-05,
      "loss": 4.7476,
      "mean_token_accuracy": 0.19662052094936372,
      "num_tokens": 232305.0,
      "step": 530
    },
    {
      "epoch": 0.1615798922800718,
      "grad_norm": 19.875,
      "learning_rate": 4.7311988829044486e-05,
      "loss": 4.6212,
      "mean_token_accuracy": 0.21569150388240815,
      "num_tokens": 235543.0,
      "step": 540
    },
    {
      "epoch": 0.16457211250748055,
      "grad_norm": 32.75,
      "learning_rate": 4.726211849192101e-05,
      "loss": 4.816,
      "mean_token_accuracy": 0.1801095575094223,
      "num_tokens": 239791.0,
      "step": 550
    },
    {
      "epoch": 0.1675643327348893,
      "grad_norm": 41.75,
      "learning_rate": 4.721224815479753e-05,
      "loss": 4.4197,
      "mean_token_accuracy": 0.2141128771007061,
      "num_tokens": 244335.0,
      "step": 560
    },
    {
      "epoch": 0.17055655296229802,
      "grad_norm": 33.0,
      "learning_rate": 4.7162377817674045e-05,
      "loss": 4.7224,
      "mean_token_accuracy": 0.18216264843940735,
      "num_tokens": 248915.0,
      "step": 570
    },
    {
      "epoch": 0.17354877318970677,
      "grad_norm": 18.375,
      "learning_rate": 4.711250748055057e-05,
      "loss": 4.6028,
      "mean_token_accuracy": 0.19473867118358612,
      "num_tokens": 253993.0,
      "step": 580
    },
    {
      "epoch": 0.1765409934171155,
      "grad_norm": 22.625,
      "learning_rate": 4.706263714342709e-05,
      "loss": 4.839,
      "mean_token_accuracy": 0.1866973988711834,
      "num_tokens": 258282.0,
      "step": 590
    },
    {
      "epoch": 0.17953321364452424,
      "grad_norm": 30.875,
      "learning_rate": 4.701276680630362e-05,
      "loss": 5.0221,
      "mean_token_accuracy": 0.16445840001106263,
      "num_tokens": 264153.0,
      "step": 600
    },
    {
      "epoch": 0.18252543387193298,
      "grad_norm": 30.25,
      "learning_rate": 4.6962896469180135e-05,
      "loss": 4.6525,
      "mean_token_accuracy": 0.18769880682229995,
      "num_tokens": 268110.0,
      "step": 610
    },
    {
      "epoch": 0.1855176540993417,
      "grad_norm": 31.375,
      "learning_rate": 4.691302613205665e-05,
      "loss": 4.7869,
      "mean_token_accuracy": 0.19825029596686364,
      "num_tokens": 272602.0,
      "step": 620
    },
    {
      "epoch": 0.18850987432675045,
      "grad_norm": 21.75,
      "learning_rate": 4.6863155794933176e-05,
      "loss": 4.8373,
      "mean_token_accuracy": 0.19757667630910875,
      "num_tokens": 276801.0,
      "step": 630
    },
    {
      "epoch": 0.19150209455415917,
      "grad_norm": 29.875,
      "learning_rate": 4.6813285457809694e-05,
      "loss": 4.4989,
      "mean_token_accuracy": 0.2317608743906021,
      "num_tokens": 281426.0,
      "step": 640
    },
    {
      "epoch": 0.19449431478156792,
      "grad_norm": 48.25,
      "learning_rate": 4.676341512068622e-05,
      "loss": 4.3297,
      "mean_token_accuracy": 0.2231058359146118,
      "num_tokens": 285040.0,
      "step": 650
    },
    {
      "epoch": 0.19748653500897667,
      "grad_norm": 19.75,
      "learning_rate": 4.671354478356274e-05,
      "loss": 4.6316,
      "mean_token_accuracy": 0.19868830889463424,
      "num_tokens": 288896.0,
      "step": 660
    },
    {
      "epoch": 0.2004787552363854,
      "grad_norm": 22.375,
      "learning_rate": 4.666367444643926e-05,
      "loss": 4.4535,
      "mean_token_accuracy": 0.21323003247380257,
      "num_tokens": 294355.0,
      "step": 670
    },
    {
      "epoch": 0.20347097546379414,
      "grad_norm": 22.25,
      "learning_rate": 4.661380410931578e-05,
      "loss": 4.5425,
      "mean_token_accuracy": 0.21277685910463334,
      "num_tokens": 297912.0,
      "step": 680
    },
    {
      "epoch": 0.20646319569120286,
      "grad_norm": 19.25,
      "learning_rate": 4.65639337721923e-05,
      "loss": 4.7785,
      "mean_token_accuracy": 0.21325334906578064,
      "num_tokens": 302327.0,
      "step": 690
    },
    {
      "epoch": 0.2094554159186116,
      "grad_norm": 46.0,
      "learning_rate": 4.6514063435068825e-05,
      "loss": 4.2218,
      "mean_token_accuracy": 0.2309760756790638,
      "num_tokens": 306067.0,
      "step": 700
    },
    {
      "epoch": 0.21244763614602036,
      "grad_norm": 17.875,
      "learning_rate": 4.646419309794535e-05,
      "loss": 4.6053,
      "mean_token_accuracy": 0.2088135614991188,
      "num_tokens": 310339.0,
      "step": 710
    },
    {
      "epoch": 0.21543985637342908,
      "grad_norm": 29.125,
      "learning_rate": 4.641432276082187e-05,
      "loss": 4.2837,
      "mean_token_accuracy": 0.239446784555912,
      "num_tokens": 315032.0,
      "step": 720
    },
    {
      "epoch": 0.21843207660083783,
      "grad_norm": 30.375,
      "learning_rate": 4.6364452423698384e-05,
      "loss": 4.3222,
      "mean_token_accuracy": 0.23795566260814666,
      "num_tokens": 318991.0,
      "step": 730
    },
    {
      "epoch": 0.22142429682824655,
      "grad_norm": 41.5,
      "learning_rate": 4.631458208657491e-05,
      "loss": 4.219,
      "mean_token_accuracy": 0.2595970630645752,
      "num_tokens": 321861.0,
      "step": 740
    },
    {
      "epoch": 0.2244165170556553,
      "grad_norm": 27.0,
      "learning_rate": 4.6264711749451426e-05,
      "loss": 4.6013,
      "mean_token_accuracy": 0.21915716528892518,
      "num_tokens": 326438.0,
      "step": 750
    },
    {
      "epoch": 0.22740873728306404,
      "grad_norm": 37.75,
      "learning_rate": 4.621484141232795e-05,
      "loss": 4.2543,
      "mean_token_accuracy": 0.23498550355434417,
      "num_tokens": 331305.0,
      "step": 760
    },
    {
      "epoch": 0.23040095751047276,
      "grad_norm": 25.625,
      "learning_rate": 4.6164971075204474e-05,
      "loss": 4.595,
      "mean_token_accuracy": 0.18473858758807182,
      "num_tokens": 336072.0,
      "step": 770
    },
    {
      "epoch": 0.2333931777378815,
      "grad_norm": 24.125,
      "learning_rate": 4.611510073808099e-05,
      "loss": 4.5353,
      "mean_token_accuracy": 0.20565996766090394,
      "num_tokens": 341423.0,
      "step": 780
    },
    {
      "epoch": 0.23638539796529023,
      "grad_norm": 34.75,
      "learning_rate": 4.6065230400957516e-05,
      "loss": 4.3319,
      "mean_token_accuracy": 0.25166238844394684,
      "num_tokens": 345172.0,
      "step": 790
    },
    {
      "epoch": 0.23937761819269898,
      "grad_norm": 34.75,
      "learning_rate": 4.601536006383403e-05,
      "loss": 3.9744,
      "mean_token_accuracy": 0.2654547780752182,
      "num_tokens": 349952.0,
      "step": 800
    },
    {
      "epoch": 0.24236983842010773,
      "grad_norm": 27.125,
      "learning_rate": 4.596548972671055e-05,
      "loss": 4.1817,
      "mean_token_accuracy": 0.2300293117761612,
      "num_tokens": 352347.0,
      "step": 810
    },
    {
      "epoch": 0.24536205864751645,
      "grad_norm": 24.0,
      "learning_rate": 4.5915619389587075e-05,
      "loss": 4.4951,
      "mean_token_accuracy": 0.20228217914700508,
      "num_tokens": 356974.0,
      "step": 820
    },
    {
      "epoch": 0.2483542788749252,
      "grad_norm": 28.25,
      "learning_rate": 4.58657490524636e-05,
      "loss": 4.025,
      "mean_token_accuracy": 0.2504382789134979,
      "num_tokens": 362547.0,
      "step": 830
    },
    {
      "epoch": 0.2513464991023339,
      "grad_norm": 30.125,
      "learning_rate": 4.5815878715340116e-05,
      "loss": 4.2457,
      "mean_token_accuracy": 0.24330067336559297,
      "num_tokens": 366551.0,
      "step": 840
    },
    {
      "epoch": 0.25433871932974267,
      "grad_norm": 28.75,
      "learning_rate": 4.576600837821664e-05,
      "loss": 3.977,
      "mean_token_accuracy": 0.2565977931022644,
      "num_tokens": 370062.0,
      "step": 850
    },
    {
      "epoch": 0.2573309395571514,
      "grad_norm": 23.0,
      "learning_rate": 4.571613804109316e-05,
      "loss": 4.3829,
      "mean_token_accuracy": 0.24670309722423553,
      "num_tokens": 373755.0,
      "step": 860
    },
    {
      "epoch": 0.26032315978456017,
      "grad_norm": 26.375,
      "learning_rate": 4.5666267703969675e-05,
      "loss": 3.9334,
      "mean_token_accuracy": 0.284832064807415,
      "num_tokens": 377036.0,
      "step": 870
    },
    {
      "epoch": 0.26331538001196886,
      "grad_norm": 25.25,
      "learning_rate": 4.5616397366846206e-05,
      "loss": 4.1585,
      "mean_token_accuracy": 0.2497795984148979,
      "num_tokens": 381584.0,
      "step": 880
    },
    {
      "epoch": 0.2663076002393776,
      "grad_norm": 28.375,
      "learning_rate": 4.5566527029722724e-05,
      "loss": 4.0316,
      "mean_token_accuracy": 0.2681065008044243,
      "num_tokens": 385316.0,
      "step": 890
    },
    {
      "epoch": 0.26929982046678635,
      "grad_norm": 19.25,
      "learning_rate": 4.551665669259925e-05,
      "loss": 4.3058,
      "mean_token_accuracy": 0.25345864295959475,
      "num_tokens": 389732.0,
      "step": 900
    },
    {
      "epoch": 0.2722920406941951,
      "grad_norm": 19.625,
      "learning_rate": 4.5466786355475765e-05,
      "loss": 4.5396,
      "mean_token_accuracy": 0.20461051017045975,
      "num_tokens": 393695.0,
      "step": 910
    },
    {
      "epoch": 0.27528426092160385,
      "grad_norm": 36.0,
      "learning_rate": 4.541691601835228e-05,
      "loss": 4.3035,
      "mean_token_accuracy": 0.28128461763262746,
      "num_tokens": 397792.0,
      "step": 920
    },
    {
      "epoch": 0.27827648114901254,
      "grad_norm": 31.625,
      "learning_rate": 4.536704568122881e-05,
      "loss": 4.368,
      "mean_token_accuracy": 0.2355639174580574,
      "num_tokens": 402363.0,
      "step": 930
    },
    {
      "epoch": 0.2812687013764213,
      "grad_norm": 25.375,
      "learning_rate": 4.531717534410533e-05,
      "loss": 4.1856,
      "mean_token_accuracy": 0.25618700236082076,
      "num_tokens": 406460.0,
      "step": 940
    },
    {
      "epoch": 0.28426092160383004,
      "grad_norm": 44.5,
      "learning_rate": 4.526730500698185e-05,
      "loss": 4.0633,
      "mean_token_accuracy": 0.27557541579008105,
      "num_tokens": 411194.0,
      "step": 950
    },
    {
      "epoch": 0.2872531418312388,
      "grad_norm": 23.125,
      "learning_rate": 4.521743466985837e-05,
      "loss": 3.8936,
      "mean_token_accuracy": 0.2943033441901207,
      "num_tokens": 415382.0,
      "step": 960
    },
    {
      "epoch": 0.29024536205864754,
      "grad_norm": 24.25,
      "learning_rate": 4.516756433273489e-05,
      "loss": 4.068,
      "mean_token_accuracy": 0.2729153633117676,
      "num_tokens": 419430.0,
      "step": 970
    },
    {
      "epoch": 0.29323758228605623,
      "grad_norm": 26.375,
      "learning_rate": 4.511769399561141e-05,
      "loss": 4.2647,
      "mean_token_accuracy": 0.237312476336956,
      "num_tokens": 423990.0,
      "step": 980
    },
    {
      "epoch": 0.296229802513465,
      "grad_norm": 22.0,
      "learning_rate": 4.506782365848793e-05,
      "loss": 3.8042,
      "mean_token_accuracy": 0.3054521858692169,
      "num_tokens": 427393.0,
      "step": 990
    },
    {
      "epoch": 0.2992220227408737,
      "grad_norm": 30.75,
      "learning_rate": 4.5017953321364456e-05,
      "loss": 4.6139,
      "mean_token_accuracy": 0.20095933154225348,
      "num_tokens": 432707.0,
      "step": 1000
    },
    {
      "epoch": 0.3022142429682825,
      "grad_norm": 46.5,
      "learning_rate": 4.496808298424098e-05,
      "loss": 3.9964,
      "mean_token_accuracy": 0.2647743716835976,
      "num_tokens": 437245.0,
      "step": 1010
    },
    {
      "epoch": 0.3052064631956912,
      "grad_norm": 34.0,
      "learning_rate": 4.49182126471175e-05,
      "loss": 3.9297,
      "mean_token_accuracy": 0.27339537590742113,
      "num_tokens": 440804.0,
      "step": 1020
    },
    {
      "epoch": 0.3081986834230999,
      "grad_norm": 27.5,
      "learning_rate": 4.4868342309994015e-05,
      "loss": 4.1647,
      "mean_token_accuracy": 0.28250799477100375,
      "num_tokens": 444520.0,
      "step": 1030
    },
    {
      "epoch": 0.31119090365050867,
      "grad_norm": 27.375,
      "learning_rate": 4.481847197287054e-05,
      "loss": 4.4008,
      "mean_token_accuracy": 0.24318356662988663,
      "num_tokens": 448536.0,
      "step": 1040
    },
    {
      "epoch": 0.3141831238779174,
      "grad_norm": 28.5,
      "learning_rate": 4.4768601635747056e-05,
      "loss": 4.2291,
      "mean_token_accuracy": 0.23109171837568282,
      "num_tokens": 452624.0,
      "step": 1050
    },
    {
      "epoch": 0.31717534410532616,
      "grad_norm": 21.25,
      "learning_rate": 4.471873129862358e-05,
      "loss": 4.0299,
      "mean_token_accuracy": 0.3011272594332695,
      "num_tokens": 457675.0,
      "step": 1060
    },
    {
      "epoch": 0.3201675643327349,
      "grad_norm": 19.75,
      "learning_rate": 4.4668860961500105e-05,
      "loss": 4.2009,
      "mean_token_accuracy": 0.24107901751995087,
      "num_tokens": 462649.0,
      "step": 1070
    },
    {
      "epoch": 0.3231597845601436,
      "grad_norm": 23.875,
      "learning_rate": 4.461899062437662e-05,
      "loss": 4.0349,
      "mean_token_accuracy": 0.29197061657905576,
      "num_tokens": 465830.0,
      "step": 1080
    },
    {
      "epoch": 0.32615200478755235,
      "grad_norm": 28.125,
      "learning_rate": 4.4569120287253146e-05,
      "loss": 3.7125,
      "mean_token_accuracy": 0.30826200395822523,
      "num_tokens": 470302.0,
      "step": 1090
    },
    {
      "epoch": 0.3291442250149611,
      "grad_norm": 24.625,
      "learning_rate": 4.4519249950129664e-05,
      "loss": 3.986,
      "mean_token_accuracy": 0.2726521834731102,
      "num_tokens": 474105.0,
      "step": 1100
    },
    {
      "epoch": 0.33213644524236985,
      "grad_norm": 28.75,
      "learning_rate": 4.446937961300619e-05,
      "loss": 4.4285,
      "mean_token_accuracy": 0.24719695299863814,
      "num_tokens": 479640.0,
      "step": 1110
    },
    {
      "epoch": 0.3351286654697786,
      "grad_norm": 24.5,
      "learning_rate": 4.441950927588271e-05,
      "loss": 4.2159,
      "mean_token_accuracy": 0.2708512395620346,
      "num_tokens": 484553.0,
      "step": 1120
    },
    {
      "epoch": 0.3381208856971873,
      "grad_norm": 36.25,
      "learning_rate": 4.436963893875923e-05,
      "loss": 4.2127,
      "mean_token_accuracy": 0.2593187764286995,
      "num_tokens": 489188.0,
      "step": 1130
    },
    {
      "epoch": 0.34111310592459604,
      "grad_norm": 27.5,
      "learning_rate": 4.431976860163575e-05,
      "loss": 3.6665,
      "mean_token_accuracy": 0.32944968044757844,
      "num_tokens": 491886.0,
      "step": 1140
    },
    {
      "epoch": 0.3441053261520048,
      "grad_norm": 24.0,
      "learning_rate": 4.426989826451227e-05,
      "loss": 3.9058,
      "mean_token_accuracy": 0.2709208592772484,
      "num_tokens": 497996.0,
      "step": 1150
    },
    {
      "epoch": 0.34709754637941354,
      "grad_norm": 16.5,
      "learning_rate": 4.422002792738879e-05,
      "loss": 3.6553,
      "mean_token_accuracy": 0.32792728394269943,
      "num_tokens": 502342.0,
      "step": 1160
    },
    {
      "epoch": 0.3500897666068223,
      "grad_norm": 23.5,
      "learning_rate": 4.417015759026531e-05,
      "loss": 4.2918,
      "mean_token_accuracy": 0.27489738166332245,
      "num_tokens": 507455.0,
      "step": 1170
    },
    {
      "epoch": 0.353081986834231,
      "grad_norm": 25.25,
      "learning_rate": 4.412028725314184e-05,
      "loss": 3.9952,
      "mean_token_accuracy": 0.28054697811603546,
      "num_tokens": 512936.0,
      "step": 1180
    },
    {
      "epoch": 0.3560742070616397,
      "grad_norm": 37.25,
      "learning_rate": 4.4070416916018354e-05,
      "loss": 3.9521,
      "mean_token_accuracy": 0.2756631106138229,
      "num_tokens": 517570.0,
      "step": 1190
    },
    {
      "epoch": 0.3590664272890485,
      "grad_norm": 23.75,
      "learning_rate": 4.402054657889488e-05,
      "loss": 3.8001,
      "mean_token_accuracy": 0.29615438580513,
      "num_tokens": 521641.0,
      "step": 1200
    },
    {
      "epoch": 0.3620586475164572,
      "grad_norm": 23.25,
      "learning_rate": 4.3970676241771396e-05,
      "loss": 3.934,
      "mean_token_accuracy": 0.30610962957143784,
      "num_tokens": 526036.0,
      "step": 1210
    },
    {
      "epoch": 0.36505086774386597,
      "grad_norm": 31.875,
      "learning_rate": 4.392080590464791e-05,
      "loss": 4.0934,
      "mean_token_accuracy": 0.27161254435777665,
      "num_tokens": 530945.0,
      "step": 1220
    },
    {
      "epoch": 0.36804308797127466,
      "grad_norm": 24.0,
      "learning_rate": 4.387093556752444e-05,
      "loss": 3.8777,
      "mean_token_accuracy": 0.2866261497139931,
      "num_tokens": 534340.0,
      "step": 1230
    },
    {
      "epoch": 0.3710353081986834,
      "grad_norm": 28.0,
      "learning_rate": 4.382106523040096e-05,
      "loss": 4.1483,
      "mean_token_accuracy": 0.2668736070394516,
      "num_tokens": 538433.0,
      "step": 1240
    },
    {
      "epoch": 0.37402752842609216,
      "grad_norm": 25.5,
      "learning_rate": 4.377119489327748e-05,
      "loss": 3.8482,
      "mean_token_accuracy": 0.27559166103601457,
      "num_tokens": 542653.0,
      "step": 1250
    },
    {
      "epoch": 0.3770197486535009,
      "grad_norm": 46.5,
      "learning_rate": 4.3721324556154e-05,
      "loss": 4.2931,
      "mean_token_accuracy": 0.25253031924366953,
      "num_tokens": 546820.0,
      "step": 1260
    },
    {
      "epoch": 0.38001196888090966,
      "grad_norm": 28.375,
      "learning_rate": 4.367145421903052e-05,
      "loss": 3.8307,
      "mean_token_accuracy": 0.3006118655204773,
      "num_tokens": 552012.0,
      "step": 1270
    },
    {
      "epoch": 0.38300418910831835,
      "grad_norm": 23.875,
      "learning_rate": 4.362158388190704e-05,
      "loss": 3.8406,
      "mean_token_accuracy": 0.31204836815595627,
      "num_tokens": 556352.0,
      "step": 1280
    },
    {
      "epoch": 0.3859964093357271,
      "grad_norm": 36.25,
      "learning_rate": 4.357171354478357e-05,
      "loss": 3.9484,
      "mean_token_accuracy": 0.27679614052176477,
      "num_tokens": 560781.0,
      "step": 1290
    },
    {
      "epoch": 0.38898862956313585,
      "grad_norm": 26.625,
      "learning_rate": 4.3521843207660086e-05,
      "loss": 3.7752,
      "mean_token_accuracy": 0.30054267644882204,
      "num_tokens": 563722.0,
      "step": 1300
    },
    {
      "epoch": 0.3919808497905446,
      "grad_norm": 63.75,
      "learning_rate": 4.347197287053661e-05,
      "loss": 3.7764,
      "mean_token_accuracy": 0.33363881558179853,
      "num_tokens": 567678.0,
      "step": 1310
    },
    {
      "epoch": 0.39497307001795334,
      "grad_norm": 23.5,
      "learning_rate": 4.342210253341313e-05,
      "loss": 3.8614,
      "mean_token_accuracy": 0.2934919849038124,
      "num_tokens": 571872.0,
      "step": 1320
    },
    {
      "epoch": 0.39796529024536204,
      "grad_norm": 37.5,
      "learning_rate": 4.3372232196289645e-05,
      "loss": 3.8984,
      "mean_token_accuracy": 0.2898759454488754,
      "num_tokens": 576076.0,
      "step": 1330
    },
    {
      "epoch": 0.4009575104727708,
      "grad_norm": 34.5,
      "learning_rate": 4.332236185916617e-05,
      "loss": 3.8018,
      "mean_token_accuracy": 0.30701966434717176,
      "num_tokens": 580652.0,
      "step": 1340
    },
    {
      "epoch": 0.40394973070017953,
      "grad_norm": 27.0,
      "learning_rate": 4.3272491522042694e-05,
      "loss": 4.0836,
      "mean_token_accuracy": 0.2806298568844795,
      "num_tokens": 584037.0,
      "step": 1350
    },
    {
      "epoch": 0.4069419509275883,
      "grad_norm": 26.875,
      "learning_rate": 4.322262118491921e-05,
      "loss": 3.3263,
      "mean_token_accuracy": 0.3484503984451294,
      "num_tokens": 587904.0,
      "step": 1360
    },
    {
      "epoch": 0.40993417115499703,
      "grad_norm": 31.625,
      "learning_rate": 4.3172750847795735e-05,
      "loss": 3.8748,
      "mean_token_accuracy": 0.28735994547605515,
      "num_tokens": 591506.0,
      "step": 1370
    },
    {
      "epoch": 0.4129263913824057,
      "grad_norm": 39.0,
      "learning_rate": 4.312288051067225e-05,
      "loss": 3.925,
      "mean_token_accuracy": 0.287045781314373,
      "num_tokens": 595632.0,
      "step": 1380
    },
    {
      "epoch": 0.41591861160981447,
      "grad_norm": 27.625,
      "learning_rate": 4.307301017354878e-05,
      "loss": 4.0082,
      "mean_token_accuracy": 0.2599979847669601,
      "num_tokens": 602507.0,
      "step": 1390
    },
    {
      "epoch": 0.4189108318372232,
      "grad_norm": 47.5,
      "learning_rate": 4.3023139836425294e-05,
      "loss": 4.1298,
      "mean_token_accuracy": 0.26573686897754667,
      "num_tokens": 606140.0,
      "step": 1400
    },
    {
      "epoch": 0.42190305206463197,
      "grad_norm": 24.0,
      "learning_rate": 4.297326949930182e-05,
      "loss": 3.883,
      "mean_token_accuracy": 0.2926565006375313,
      "num_tokens": 609882.0,
      "step": 1410
    },
    {
      "epoch": 0.4248952722920407,
      "grad_norm": 54.25,
      "learning_rate": 4.292339916217834e-05,
      "loss": 3.8369,
      "mean_token_accuracy": 0.3231379181146622,
      "num_tokens": 613987.0,
      "step": 1420
    },
    {
      "epoch": 0.4278874925194494,
      "grad_norm": 39.5,
      "learning_rate": 4.287352882505486e-05,
      "loss": 4.1388,
      "mean_token_accuracy": 0.28088164180517194,
      "num_tokens": 618639.0,
      "step": 1430
    },
    {
      "epoch": 0.43087971274685816,
      "grad_norm": 23.5,
      "learning_rate": 4.282365848793138e-05,
      "loss": 3.7898,
      "mean_token_accuracy": 0.3190530717372894,
      "num_tokens": 623238.0,
      "step": 1440
    },
    {
      "epoch": 0.4338719329742669,
      "grad_norm": 22.875,
      "learning_rate": 4.27737881508079e-05,
      "loss": 3.3278,
      "mean_token_accuracy": 0.3729147434234619,
      "num_tokens": 627865.0,
      "step": 1450
    },
    {
      "epoch": 0.43686415320167565,
      "grad_norm": 29.125,
      "learning_rate": 4.272391781368442e-05,
      "loss": 3.6825,
      "mean_token_accuracy": 0.3166591688990593,
      "num_tokens": 632426.0,
      "step": 1460
    },
    {
      "epoch": 0.4398563734290844,
      "grad_norm": 39.75,
      "learning_rate": 4.267404747656094e-05,
      "loss": 4.0308,
      "mean_token_accuracy": 0.2782156467437744,
      "num_tokens": 636085.0,
      "step": 1470
    },
    {
      "epoch": 0.4428485936564931,
      "grad_norm": 28.625,
      "learning_rate": 4.262417713943747e-05,
      "loss": 3.8042,
      "mean_token_accuracy": 0.3088426858186722,
      "num_tokens": 640123.0,
      "step": 1480
    },
    {
      "epoch": 0.44584081388390184,
      "grad_norm": 21.5,
      "learning_rate": 4.2574306802313985e-05,
      "loss": 3.3991,
      "mean_token_accuracy": 0.3608384057879448,
      "num_tokens": 644113.0,
      "step": 1490
    },
    {
      "epoch": 0.4488330341113106,
      "grad_norm": 35.25,
      "learning_rate": 4.252443646519051e-05,
      "loss": 3.973,
      "mean_token_accuracy": 0.28127337992191315,
      "num_tokens": 647802.0,
      "step": 1500
    },
    {
      "epoch": 0.45182525433871934,
      "grad_norm": 24.875,
      "learning_rate": 4.2474566128067026e-05,
      "loss": 3.7555,
      "mean_token_accuracy": 0.2975438117980957,
      "num_tokens": 651745.0,
      "step": 1510
    },
    {
      "epoch": 0.4548174745661281,
      "grad_norm": 29.875,
      "learning_rate": 4.242469579094355e-05,
      "loss": 3.9791,
      "mean_token_accuracy": 0.30450038611888885,
      "num_tokens": 655944.0,
      "step": 1520
    },
    {
      "epoch": 0.4578096947935368,
      "grad_norm": 39.75,
      "learning_rate": 4.2374825453820075e-05,
      "loss": 3.6477,
      "mean_token_accuracy": 0.3407430723309517,
      "num_tokens": 660460.0,
      "step": 1530
    },
    {
      "epoch": 0.46080191502094553,
      "grad_norm": 22.125,
      "learning_rate": 4.232495511669659e-05,
      "loss": 4.0353,
      "mean_token_accuracy": 0.3086406052112579,
      "num_tokens": 664503.0,
      "step": 1540
    },
    {
      "epoch": 0.4637941352483543,
      "grad_norm": 27.5,
      "learning_rate": 4.227508477957311e-05,
      "loss": 3.5857,
      "mean_token_accuracy": 0.34699532836675645,
      "num_tokens": 669506.0,
      "step": 1550
    },
    {
      "epoch": 0.466786355475763,
      "grad_norm": 57.5,
      "learning_rate": 4.2225214442449634e-05,
      "loss": 3.5541,
      "mean_token_accuracy": 0.3328308701515198,
      "num_tokens": 674486.0,
      "step": 1560
    },
    {
      "epoch": 0.4697785757031718,
      "grad_norm": 21.125,
      "learning_rate": 4.217534410532615e-05,
      "loss": 3.2879,
      "mean_token_accuracy": 0.3543445348739624,
      "num_tokens": 678563.0,
      "step": 1570
    },
    {
      "epoch": 0.47277079593058047,
      "grad_norm": 34.25,
      "learning_rate": 4.2125473768202675e-05,
      "loss": 3.6253,
      "mean_token_accuracy": 0.31252876818180086,
      "num_tokens": 682718.0,
      "step": 1580
    },
    {
      "epoch": 0.4757630161579892,
      "grad_norm": 26.875,
      "learning_rate": 4.20756034310792e-05,
      "loss": 4.02,
      "mean_token_accuracy": 0.3078885093331337,
      "num_tokens": 686239.0,
      "step": 1590
    },
    {
      "epoch": 0.47875523638539796,
      "grad_norm": 25.125,
      "learning_rate": 4.202573309395572e-05,
      "loss": 3.5034,
      "mean_token_accuracy": 0.3475412830710411,
      "num_tokens": 690265.0,
      "step": 1600
    },
    {
      "epoch": 0.4817474566128067,
      "grad_norm": 29.5,
      "learning_rate": 4.197586275683224e-05,
      "loss": 3.7491,
      "mean_token_accuracy": 0.3203135937452316,
      "num_tokens": 695201.0,
      "step": 1610
    },
    {
      "epoch": 0.48473967684021546,
      "grad_norm": 21.375,
      "learning_rate": 4.192599241970876e-05,
      "loss": 3.9912,
      "mean_token_accuracy": 0.2843865931034088,
      "num_tokens": 700042.0,
      "step": 1620
    },
    {
      "epoch": 0.48773189706762415,
      "grad_norm": 17.875,
      "learning_rate": 4.1876122082585276e-05,
      "loss": 4.4067,
      "mean_token_accuracy": 0.28071754723787307,
      "num_tokens": 703650.0,
      "step": 1630
    },
    {
      "epoch": 0.4907241172950329,
      "grad_norm": 18.5,
      "learning_rate": 4.18262517454618e-05,
      "loss": 3.5235,
      "mean_token_accuracy": 0.3489766031503677,
      "num_tokens": 708375.0,
      "step": 1640
    },
    {
      "epoch": 0.49371633752244165,
      "grad_norm": 26.375,
      "learning_rate": 4.1776381408338324e-05,
      "loss": 3.9863,
      "mean_token_accuracy": 0.31152606308460234,
      "num_tokens": 712864.0,
      "step": 1650
    },
    {
      "epoch": 0.4967085577498504,
      "grad_norm": 25.75,
      "learning_rate": 4.172651107121484e-05,
      "loss": 4.003,
      "mean_token_accuracy": 0.2858720526099205,
      "num_tokens": 717126.0,
      "step": 1660
    },
    {
      "epoch": 0.49970077797725915,
      "grad_norm": 45.25,
      "learning_rate": 4.1676640734091366e-05,
      "loss": 3.7429,
      "mean_token_accuracy": 0.3121335655450821,
      "num_tokens": 720139.0,
      "step": 1670
    },
    {
      "epoch": 0.5026929982046678,
      "grad_norm": 27.75,
      "learning_rate": 4.162677039696788e-05,
      "loss": 3.5136,
      "mean_token_accuracy": 0.32364853024482726,
      "num_tokens": 724517.0,
      "step": 1680
    },
    {
      "epoch": 0.5056852184320766,
      "grad_norm": 29.125,
      "learning_rate": 4.157690005984441e-05,
      "loss": 3.6384,
      "mean_token_accuracy": 0.3176667720079422,
      "num_tokens": 728613.0,
      "step": 1690
    },
    {
      "epoch": 0.5086774386594853,
      "grad_norm": 25.75,
      "learning_rate": 4.1527029722720925e-05,
      "loss": 3.4635,
      "mean_token_accuracy": 0.3570543646812439,
      "num_tokens": 733893.0,
      "step": 1700
    },
    {
      "epoch": 0.5116696588868941,
      "grad_norm": 35.25,
      "learning_rate": 4.147715938559745e-05,
      "loss": 4.0123,
      "mean_token_accuracy": 0.2949834749102592,
      "num_tokens": 737230.0,
      "step": 1710
    },
    {
      "epoch": 0.5146618791143028,
      "grad_norm": 36.5,
      "learning_rate": 4.142728904847397e-05,
      "loss": 3.8303,
      "mean_token_accuracy": 0.3259445518255234,
      "num_tokens": 740956.0,
      "step": 1720
    },
    {
      "epoch": 0.5176540993417116,
      "grad_norm": 26.0,
      "learning_rate": 4.137741871135049e-05,
      "loss": 3.5221,
      "mean_token_accuracy": 0.3586596347391605,
      "num_tokens": 744980.0,
      "step": 1730
    },
    {
      "epoch": 0.5206463195691203,
      "grad_norm": 42.25,
      "learning_rate": 4.132754837422701e-05,
      "loss": 3.945,
      "mean_token_accuracy": 0.31935873329639436,
      "num_tokens": 748174.0,
      "step": 1740
    },
    {
      "epoch": 0.5236385397965291,
      "grad_norm": 43.25,
      "learning_rate": 4.127767803710353e-05,
      "loss": 3.9663,
      "mean_token_accuracy": 0.31343400850892067,
      "num_tokens": 752551.0,
      "step": 1750
    },
    {
      "epoch": 0.5266307600239377,
      "grad_norm": 41.75,
      "learning_rate": 4.1227807699980056e-05,
      "loss": 3.5433,
      "mean_token_accuracy": 0.34673927128314974,
      "num_tokens": 757265.0,
      "step": 1760
    },
    {
      "epoch": 0.5296229802513465,
      "grad_norm": 22.125,
      "learning_rate": 4.1177937362856574e-05,
      "loss": 3.7775,
      "mean_token_accuracy": 0.3192309245467186,
      "num_tokens": 762186.0,
      "step": 1770
    },
    {
      "epoch": 0.5326152004787552,
      "grad_norm": 36.75,
      "learning_rate": 4.11280670257331e-05,
      "loss": 3.4575,
      "mean_token_accuracy": 0.3697234556078911,
      "num_tokens": 767136.0,
      "step": 1780
    },
    {
      "epoch": 0.535607420706164,
      "grad_norm": 55.75,
      "learning_rate": 4.1078196688609615e-05,
      "loss": 3.9442,
      "mean_token_accuracy": 0.306137615442276,
      "num_tokens": 772421.0,
      "step": 1790
    },
    {
      "epoch": 0.5385996409335727,
      "grad_norm": 30.25,
      "learning_rate": 4.102832635148614e-05,
      "loss": 3.4871,
      "mean_token_accuracy": 0.34432919770479203,
      "num_tokens": 777103.0,
      "step": 1800
    },
    {
      "epoch": 0.5415918611609815,
      "grad_norm": 31.375,
      "learning_rate": 4.097845601436266e-05,
      "loss": 4.0582,
      "mean_token_accuracy": 0.2766360715031624,
      "num_tokens": 781840.0,
      "step": 1810
    },
    {
      "epoch": 0.5445840813883902,
      "grad_norm": 22.125,
      "learning_rate": 4.092858567723918e-05,
      "loss": 3.6937,
      "mean_token_accuracy": 0.3213631004095078,
      "num_tokens": 787443.0,
      "step": 1820
    },
    {
      "epoch": 0.547576301615799,
      "grad_norm": 24.5,
      "learning_rate": 4.0878715340115705e-05,
      "loss": 3.2045,
      "mean_token_accuracy": 0.35475517213344576,
      "num_tokens": 792923.0,
      "step": 1830
    },
    {
      "epoch": 0.5505685218432077,
      "grad_norm": 23.625,
      "learning_rate": 4.082884500299222e-05,
      "loss": 3.5229,
      "mean_token_accuracy": 0.34978632628917694,
      "num_tokens": 797220.0,
      "step": 1840
    },
    {
      "epoch": 0.5535607420706165,
      "grad_norm": 31.75,
      "learning_rate": 4.077897466586874e-05,
      "loss": 3.7298,
      "mean_token_accuracy": 0.34635201692581175,
      "num_tokens": 801850.0,
      "step": 1850
    },
    {
      "epoch": 0.5565529622980251,
      "grad_norm": 38.5,
      "learning_rate": 4.0729104328745264e-05,
      "loss": 3.7174,
      "mean_token_accuracy": 0.30568271577358247,
      "num_tokens": 807033.0,
      "step": 1860
    },
    {
      "epoch": 0.5595451825254338,
      "grad_norm": 60.75,
      "learning_rate": 4.067923399162178e-05,
      "loss": 3.1007,
      "mean_token_accuracy": 0.3980128735303879,
      "num_tokens": 811446.0,
      "step": 1870
    },
    {
      "epoch": 0.5625374027528426,
      "grad_norm": 35.0,
      "learning_rate": 4.0629363654498306e-05,
      "loss": 3.7698,
      "mean_token_accuracy": 0.3161139450967312,
      "num_tokens": 817360.0,
      "step": 1880
    },
    {
      "epoch": 0.5655296229802513,
      "grad_norm": 22.0,
      "learning_rate": 4.057949331737483e-05,
      "loss": 3.7394,
      "mean_token_accuracy": 0.32581870555877684,
      "num_tokens": 821346.0,
      "step": 1890
    },
    {
      "epoch": 0.5685218432076601,
      "grad_norm": 40.75,
      "learning_rate": 4.052962298025135e-05,
      "loss": 3.2171,
      "mean_token_accuracy": 0.3967368185520172,
      "num_tokens": 824850.0,
      "step": 1900
    },
    {
      "epoch": 0.5715140634350688,
      "grad_norm": 39.75,
      "learning_rate": 4.047975264312787e-05,
      "loss": 3.7614,
      "mean_token_accuracy": 0.334186664968729,
      "num_tokens": 829049.0,
      "step": 1910
    },
    {
      "epoch": 0.5745062836624776,
      "grad_norm": 33.0,
      "learning_rate": 4.042988230600439e-05,
      "loss": 3.3675,
      "mean_token_accuracy": 0.3545887053012848,
      "num_tokens": 832992.0,
      "step": 1920
    },
    {
      "epoch": 0.5774985038898863,
      "grad_norm": 38.5,
      "learning_rate": 4.0380011968880906e-05,
      "loss": 4.1655,
      "mean_token_accuracy": 0.27635432183742525,
      "num_tokens": 837562.0,
      "step": 1930
    },
    {
      "epoch": 0.5804907241172951,
      "grad_norm": 23.625,
      "learning_rate": 4.033014163175744e-05,
      "loss": 3.3483,
      "mean_token_accuracy": 0.3648817211389542,
      "num_tokens": 842393.0,
      "step": 1940
    },
    {
      "epoch": 0.5834829443447038,
      "grad_norm": 28.125,
      "learning_rate": 4.0280271294633955e-05,
      "loss": 3.4962,
      "mean_token_accuracy": 0.34077437222003937,
      "num_tokens": 847798.0,
      "step": 1950
    },
    {
      "epoch": 0.5864751645721125,
      "grad_norm": 25.5,
      "learning_rate": 4.023040095751047e-05,
      "loss": 3.5223,
      "mean_token_accuracy": 0.3545456677675247,
      "num_tokens": 852416.0,
      "step": 1960
    },
    {
      "epoch": 0.5894673847995212,
      "grad_norm": 24.875,
      "learning_rate": 4.0180530620386996e-05,
      "loss": 3.2954,
      "mean_token_accuracy": 0.37952440828084943,
      "num_tokens": 856872.0,
      "step": 1970
    },
    {
      "epoch": 0.59245960502693,
      "grad_norm": 20.25,
      "learning_rate": 4.0130660283263514e-05,
      "loss": 3.214,
      "mean_token_accuracy": 0.3997642546892166,
      "num_tokens": 861808.0,
      "step": 1980
    },
    {
      "epoch": 0.5954518252543387,
      "grad_norm": 42.75,
      "learning_rate": 4.008078994614004e-05,
      "loss": 3.9122,
      "mean_token_accuracy": 0.31893026232719424,
      "num_tokens": 866080.0,
      "step": 1990
    },
    {
      "epoch": 0.5984440454817475,
      "grad_norm": 17.75,
      "learning_rate": 4.003091960901656e-05,
      "loss": 3.68,
      "mean_token_accuracy": 0.3326193943619728,
      "num_tokens": 870426.0,
      "step": 2000
    },
    {
      "epoch": 0.6014362657091562,
      "grad_norm": 29.5,
      "learning_rate": 3.998104927189308e-05,
      "loss": 4.0992,
      "mean_token_accuracy": 0.2667170882225037,
      "num_tokens": 875695.0,
      "step": 2010
    },
    {
      "epoch": 0.604428485936565,
      "grad_norm": 27.125,
      "learning_rate": 3.9931178934769604e-05,
      "loss": 3.7399,
      "mean_token_accuracy": 0.3234362483024597,
      "num_tokens": 878821.0,
      "step": 2020
    },
    {
      "epoch": 0.6074207061639737,
      "grad_norm": 22.625,
      "learning_rate": 3.988130859764612e-05,
      "loss": 3.2405,
      "mean_token_accuracy": 0.38044323176145556,
      "num_tokens": 882724.0,
      "step": 2030
    },
    {
      "epoch": 0.6104129263913824,
      "grad_norm": 25.375,
      "learning_rate": 3.983143826052264e-05,
      "loss": 3.5705,
      "mean_token_accuracy": 0.3319473534822464,
      "num_tokens": 888338.0,
      "step": 2040
    },
    {
      "epoch": 0.6134051466187912,
      "grad_norm": 23.125,
      "learning_rate": 3.978156792339916e-05,
      "loss": 3.475,
      "mean_token_accuracy": 0.3405942186713219,
      "num_tokens": 893344.0,
      "step": 2050
    },
    {
      "epoch": 0.6163973668461998,
      "grad_norm": 31.25,
      "learning_rate": 3.973169758627569e-05,
      "loss": 4.0281,
      "mean_token_accuracy": 0.29738594889640807,
      "num_tokens": 896526.0,
      "step": 2060
    },
    {
      "epoch": 0.6193895870736086,
      "grad_norm": 24.5,
      "learning_rate": 3.9681827249152204e-05,
      "loss": 3.1748,
      "mean_token_accuracy": 0.3818254411220551,
      "num_tokens": 900369.0,
      "step": 2070
    },
    {
      "epoch": 0.6223818073010173,
      "grad_norm": 36.0,
      "learning_rate": 3.963195691202873e-05,
      "loss": 3.3941,
      "mean_token_accuracy": 0.3613236457109451,
      "num_tokens": 904935.0,
      "step": 2080
    },
    {
      "epoch": 0.6253740275284261,
      "grad_norm": 63.25,
      "learning_rate": 3.9582086574905246e-05,
      "loss": 3.2628,
      "mean_token_accuracy": 0.3658029049634933,
      "num_tokens": 909555.0,
      "step": 2090
    },
    {
      "epoch": 0.6283662477558348,
      "grad_norm": 26.25,
      "learning_rate": 3.953221623778177e-05,
      "loss": 3.2202,
      "mean_token_accuracy": 0.38725468665361407,
      "num_tokens": 914294.0,
      "step": 2100
    },
    {
      "epoch": 0.6313584679832436,
      "grad_norm": 29.25,
      "learning_rate": 3.948234590065829e-05,
      "loss": 3.4529,
      "mean_token_accuracy": 0.3761670857667923,
      "num_tokens": 919096.0,
      "step": 2110
    },
    {
      "epoch": 0.6343506882106523,
      "grad_norm": 33.25,
      "learning_rate": 3.943247556353481e-05,
      "loss": 3.5437,
      "mean_token_accuracy": 0.3384189009666443,
      "num_tokens": 923520.0,
      "step": 2120
    },
    {
      "epoch": 0.6373429084380611,
      "grad_norm": 20.375,
      "learning_rate": 3.9382605226411336e-05,
      "loss": 3.5569,
      "mean_token_accuracy": 0.34821858182549476,
      "num_tokens": 928471.0,
      "step": 2130
    },
    {
      "epoch": 0.6403351286654698,
      "grad_norm": 29.0,
      "learning_rate": 3.933273488928785e-05,
      "loss": 3.6471,
      "mean_token_accuracy": 0.34270432889461516,
      "num_tokens": 932646.0,
      "step": 2140
    },
    {
      "epoch": 0.6433273488928786,
      "grad_norm": 30.625,
      "learning_rate": 3.928286455216437e-05,
      "loss": 2.7425,
      "mean_token_accuracy": 0.4409438192844391,
      "num_tokens": 936601.0,
      "step": 2150
    },
    {
      "epoch": 0.6463195691202872,
      "grad_norm": 21.875,
      "learning_rate": 3.9232994215040895e-05,
      "loss": 3.6186,
      "mean_token_accuracy": 0.33609611093997954,
      "num_tokens": 941591.0,
      "step": 2160
    },
    {
      "epoch": 0.649311789347696,
      "grad_norm": 26.25,
      "learning_rate": 3.918312387791742e-05,
      "loss": 2.9869,
      "mean_token_accuracy": 0.411373496055603,
      "num_tokens": 945657.0,
      "step": 2170
    },
    {
      "epoch": 0.6523040095751047,
      "grad_norm": 27.125,
      "learning_rate": 3.9133253540793936e-05,
      "loss": 4.0563,
      "mean_token_accuracy": 0.2994434475898743,
      "num_tokens": 951468.0,
      "step": 2180
    },
    {
      "epoch": 0.6552962298025135,
      "grad_norm": 28.375,
      "learning_rate": 3.908338320367046e-05,
      "loss": 3.1732,
      "mean_token_accuracy": 0.38346787095069884,
      "num_tokens": 954857.0,
      "step": 2190
    },
    {
      "epoch": 0.6582884500299222,
      "grad_norm": 37.0,
      "learning_rate": 3.903351286654698e-05,
      "loss": 3.3023,
      "mean_token_accuracy": 0.369389246404171,
      "num_tokens": 959539.0,
      "step": 2200
    },
    {
      "epoch": 0.661280670257331,
      "grad_norm": 20.875,
      "learning_rate": 3.89836425294235e-05,
      "loss": 3.7477,
      "mean_token_accuracy": 0.3186024323105812,
      "num_tokens": 964970.0,
      "step": 2210
    },
    {
      "epoch": 0.6642728904847397,
      "grad_norm": 25.25,
      "learning_rate": 3.893377219230002e-05,
      "loss": 3.6275,
      "mean_token_accuracy": 0.3292903244495392,
      "num_tokens": 970684.0,
      "step": 2220
    },
    {
      "epoch": 0.6672651107121484,
      "grad_norm": 27.125,
      "learning_rate": 3.8883901855176544e-05,
      "loss": 3.2893,
      "mean_token_accuracy": 0.3744408220052719,
      "num_tokens": 975251.0,
      "step": 2230
    },
    {
      "epoch": 0.6702573309395572,
      "grad_norm": 46.25,
      "learning_rate": 3.883403151805307e-05,
      "loss": 3.2116,
      "mean_token_accuracy": 0.40995657742023467,
      "num_tokens": 979714.0,
      "step": 2240
    },
    {
      "epoch": 0.6732495511669659,
      "grad_norm": 33.5,
      "learning_rate": 3.8784161180929585e-05,
      "loss": 3.7125,
      "mean_token_accuracy": 0.33205676227808,
      "num_tokens": 984789.0,
      "step": 2250
    },
    {
      "epoch": 0.6762417713943746,
      "grad_norm": 20.125,
      "learning_rate": 3.87342908438061e-05,
      "loss": 3.1216,
      "mean_token_accuracy": 0.39428712129592897,
      "num_tokens": 988232.0,
      "step": 2260
    },
    {
      "epoch": 0.6792339916217833,
      "grad_norm": 30.0,
      "learning_rate": 3.868442050668263e-05,
      "loss": 3.813,
      "mean_token_accuracy": 0.31533592492341994,
      "num_tokens": 994031.0,
      "step": 2270
    },
    {
      "epoch": 0.6822262118491921,
      "grad_norm": 37.25,
      "learning_rate": 3.8634550169559144e-05,
      "loss": 3.7227,
      "mean_token_accuracy": 0.32704337537288664,
      "num_tokens": 998035.0,
      "step": 2280
    },
    {
      "epoch": 0.6852184320766008,
      "grad_norm": 21.125,
      "learning_rate": 3.858467983243567e-05,
      "loss": 3.6412,
      "mean_token_accuracy": 0.34335303902626035,
      "num_tokens": 1002893.0,
      "step": 2290
    },
    {
      "epoch": 0.6882106523040096,
      "grad_norm": 28.75,
      "learning_rate": 3.853480949531219e-05,
      "loss": 3.3836,
      "mean_token_accuracy": 0.3560221940279007,
      "num_tokens": 1008036.0,
      "step": 2300
    },
    {
      "epoch": 0.6912028725314183,
      "grad_norm": 24.875,
      "learning_rate": 3.848493915818871e-05,
      "loss": 3.5464,
      "mean_token_accuracy": 0.36754521131515505,
      "num_tokens": 1012629.0,
      "step": 2310
    },
    {
      "epoch": 0.6941950927588271,
      "grad_norm": 32.25,
      "learning_rate": 3.8435068821065234e-05,
      "loss": 3.2238,
      "mean_token_accuracy": 0.38233456015586853,
      "num_tokens": 1017049.0,
      "step": 2320
    },
    {
      "epoch": 0.6971873129862358,
      "grad_norm": 34.25,
      "learning_rate": 3.838519848394175e-05,
      "loss": 3.1007,
      "mean_token_accuracy": 0.39890468716621397,
      "num_tokens": 1021908.0,
      "step": 2330
    },
    {
      "epoch": 0.7001795332136446,
      "grad_norm": 23.125,
      "learning_rate": 3.833532814681827e-05,
      "loss": 3.6741,
      "mean_token_accuracy": 0.3266971483826637,
      "num_tokens": 1027341.0,
      "step": 2340
    },
    {
      "epoch": 0.7031717534410533,
      "grad_norm": 43.5,
      "learning_rate": 3.82854578096948e-05,
      "loss": 3.1599,
      "mean_token_accuracy": 0.39910418093204497,
      "num_tokens": 1031079.0,
      "step": 2350
    },
    {
      "epoch": 0.706163973668462,
      "grad_norm": 27.75,
      "learning_rate": 3.823558747257132e-05,
      "loss": 3.055,
      "mean_token_accuracy": 0.41069386899471283,
      "num_tokens": 1035408.0,
      "step": 2360
    },
    {
      "epoch": 0.7091561938958707,
      "grad_norm": 25.5,
      "learning_rate": 3.8185717135447835e-05,
      "loss": 2.9248,
      "mean_token_accuracy": 0.4071251660585403,
      "num_tokens": 1039977.0,
      "step": 2370
    },
    {
      "epoch": 0.7121484141232794,
      "grad_norm": 35.75,
      "learning_rate": 3.813584679832436e-05,
      "loss": 3.1529,
      "mean_token_accuracy": 0.4023617923259735,
      "num_tokens": 1044329.0,
      "step": 2380
    },
    {
      "epoch": 0.7151406343506882,
      "grad_norm": 26.125,
      "learning_rate": 3.8085976461200877e-05,
      "loss": 3.2873,
      "mean_token_accuracy": 0.38149598240852356,
      "num_tokens": 1049211.0,
      "step": 2390
    },
    {
      "epoch": 0.718132854578097,
      "grad_norm": 27.625,
      "learning_rate": 3.80361061240774e-05,
      "loss": 2.9516,
      "mean_token_accuracy": 0.41472441554069517,
      "num_tokens": 1053849.0,
      "step": 2400
    },
    {
      "epoch": 0.7211250748055057,
      "grad_norm": 49.75,
      "learning_rate": 3.7986235786953925e-05,
      "loss": 3.0006,
      "mean_token_accuracy": 0.42310560047626494,
      "num_tokens": 1057578.0,
      "step": 2410
    },
    {
      "epoch": 0.7241172950329144,
      "grad_norm": 19.375,
      "learning_rate": 3.793636544983044e-05,
      "loss": 3.3444,
      "mean_token_accuracy": 0.38981711566448213,
      "num_tokens": 1061754.0,
      "step": 2420
    },
    {
      "epoch": 0.7271095152603232,
      "grad_norm": 31.875,
      "learning_rate": 3.7886495112706966e-05,
      "loss": 3.0919,
      "mean_token_accuracy": 0.3983312755823135,
      "num_tokens": 1066049.0,
      "step": 2430
    },
    {
      "epoch": 0.7301017354877319,
      "grad_norm": 39.75,
      "learning_rate": 3.7836624775583484e-05,
      "loss": 3.2329,
      "mean_token_accuracy": 0.36704597175121306,
      "num_tokens": 1070346.0,
      "step": 2440
    },
    {
      "epoch": 0.7330939557151407,
      "grad_norm": 20.875,
      "learning_rate": 3.778675443846e-05,
      "loss": 2.9394,
      "mean_token_accuracy": 0.44624441862106323,
      "num_tokens": 1073148.0,
      "step": 2450
    },
    {
      "epoch": 0.7360861759425493,
      "grad_norm": 28.875,
      "learning_rate": 3.7736884101336525e-05,
      "loss": 2.833,
      "mean_token_accuracy": 0.43321623355150224,
      "num_tokens": 1076646.0,
      "step": 2460
    },
    {
      "epoch": 0.7390783961699581,
      "grad_norm": 28.75,
      "learning_rate": 3.768701376421305e-05,
      "loss": 3.0669,
      "mean_token_accuracy": 0.4107959717512131,
      "num_tokens": 1080573.0,
      "step": 2470
    },
    {
      "epoch": 0.7420706163973668,
      "grad_norm": 18.125,
      "learning_rate": 3.763714342708957e-05,
      "loss": 2.5892,
      "mean_token_accuracy": 0.4758401960134506,
      "num_tokens": 1084686.0,
      "step": 2480
    },
    {
      "epoch": 0.7450628366247756,
      "grad_norm": 33.25,
      "learning_rate": 3.758727308996609e-05,
      "loss": 3.3653,
      "mean_token_accuracy": 0.37456787526607516,
      "num_tokens": 1090273.0,
      "step": 2490
    },
    {
      "epoch": 0.7480550568521843,
      "grad_norm": 24.875,
      "learning_rate": 3.753740275284261e-05,
      "loss": 3.0053,
      "mean_token_accuracy": 0.39563181102275846,
      "num_tokens": 1095363.0,
      "step": 2500
    },
    {
      "epoch": 0.7510472770795931,
      "grad_norm": 18.0,
      "learning_rate": 3.748753241571913e-05,
      "loss": 3.0128,
      "mean_token_accuracy": 0.4145811513066292,
      "num_tokens": 1100624.0,
      "step": 2510
    },
    {
      "epoch": 0.7540394973070018,
      "grad_norm": 32.5,
      "learning_rate": 3.743766207859565e-05,
      "loss": 3.2632,
      "mean_token_accuracy": 0.39147082567214964,
      "num_tokens": 1103844.0,
      "step": 2520
    },
    {
      "epoch": 0.7570317175344106,
      "grad_norm": 21.75,
      "learning_rate": 3.7387791741472174e-05,
      "loss": 3.4745,
      "mean_token_accuracy": 0.3503845572471619,
      "num_tokens": 1107559.0,
      "step": 2530
    },
    {
      "epoch": 0.7600239377618193,
      "grad_norm": 24.375,
      "learning_rate": 3.73379214043487e-05,
      "loss": 3.342,
      "mean_token_accuracy": 0.36978781670331956,
      "num_tokens": 1112153.0,
      "step": 2540
    },
    {
      "epoch": 0.7630161579892281,
      "grad_norm": 23.375,
      "learning_rate": 3.7288051067225216e-05,
      "loss": 3.0539,
      "mean_token_accuracy": 0.3902305573225021,
      "num_tokens": 1116743.0,
      "step": 2550
    },
    {
      "epoch": 0.7660083782166367,
      "grad_norm": 20.25,
      "learning_rate": 3.7238180730101733e-05,
      "loss": 2.8592,
      "mean_token_accuracy": 0.45265459418296816,
      "num_tokens": 1121737.0,
      "step": 2560
    },
    {
      "epoch": 0.7690005984440454,
      "grad_norm": 20.625,
      "learning_rate": 3.718831039297826e-05,
      "loss": 3.3017,
      "mean_token_accuracy": 0.38322472497820853,
      "num_tokens": 1127296.0,
      "step": 2570
    },
    {
      "epoch": 0.7719928186714542,
      "grad_norm": 25.5,
      "learning_rate": 3.713844005585478e-05,
      "loss": 2.7671,
      "mean_token_accuracy": 0.4317146301269531,
      "num_tokens": 1131376.0,
      "step": 2580
    },
    {
      "epoch": 0.7749850388988629,
      "grad_norm": 34.25,
      "learning_rate": 3.7088569718731306e-05,
      "loss": 3.0348,
      "mean_token_accuracy": 0.4127075791358948,
      "num_tokens": 1135501.0,
      "step": 2590
    },
    {
      "epoch": 0.7779772591262717,
      "grad_norm": 27.75,
      "learning_rate": 3.703869938160782e-05,
      "loss": 3.4994,
      "mean_token_accuracy": 0.37491829991340636,
      "num_tokens": 1139282.0,
      "step": 2600
    },
    {
      "epoch": 0.7809694793536804,
      "grad_norm": 30.875,
      "learning_rate": 3.698882904448434e-05,
      "loss": 3.1417,
      "mean_token_accuracy": 0.40053197741508484,
      "num_tokens": 1143542.0,
      "step": 2610
    },
    {
      "epoch": 0.7839616995810892,
      "grad_norm": 34.0,
      "learning_rate": 3.6938958707360865e-05,
      "loss": 3.1492,
      "mean_token_accuracy": 0.3982168883085251,
      "num_tokens": 1148007.0,
      "step": 2620
    },
    {
      "epoch": 0.7869539198084979,
      "grad_norm": 37.75,
      "learning_rate": 3.688908837023738e-05,
      "loss": 3.0793,
      "mean_token_accuracy": 0.3932459011673927,
      "num_tokens": 1151733.0,
      "step": 2630
    },
    {
      "epoch": 0.7899461400359067,
      "grad_norm": 25.875,
      "learning_rate": 3.6839218033113907e-05,
      "loss": 3.3654,
      "mean_token_accuracy": 0.37628018409013747,
      "num_tokens": 1155939.0,
      "step": 2640
    },
    {
      "epoch": 0.7929383602633154,
      "grad_norm": 130.0,
      "learning_rate": 3.678934769599043e-05,
      "loss": 2.9446,
      "mean_token_accuracy": 0.4227631092071533,
      "num_tokens": 1160646.0,
      "step": 2650
    },
    {
      "epoch": 0.7959305804907241,
      "grad_norm": 32.5,
      "learning_rate": 3.673947735886695e-05,
      "loss": 3.2153,
      "mean_token_accuracy": 0.3905955284833908,
      "num_tokens": 1164275.0,
      "step": 2660
    },
    {
      "epoch": 0.7989228007181328,
      "grad_norm": 24.125,
      "learning_rate": 3.6689607021743466e-05,
      "loss": 2.877,
      "mean_token_accuracy": 0.43215914219617846,
      "num_tokens": 1167441.0,
      "step": 2670
    },
    {
      "epoch": 0.8019150209455416,
      "grad_norm": 26.75,
      "learning_rate": 3.663973668461999e-05,
      "loss": 3.4118,
      "mean_token_accuracy": 0.3704531863331795,
      "num_tokens": 1171450.0,
      "step": 2680
    },
    {
      "epoch": 0.8049072411729503,
      "grad_norm": 39.25,
      "learning_rate": 3.658986634749651e-05,
      "loss": 3.6131,
      "mean_token_accuracy": 0.3625437140464783,
      "num_tokens": 1175198.0,
      "step": 2690
    },
    {
      "epoch": 0.8078994614003591,
      "grad_norm": 34.25,
      "learning_rate": 3.653999601037303e-05,
      "loss": 2.7747,
      "mean_token_accuracy": 0.4520352780818939,
      "num_tokens": 1178897.0,
      "step": 2700
    },
    {
      "epoch": 0.8108916816277678,
      "grad_norm": 28.625,
      "learning_rate": 3.6490125673249555e-05,
      "loss": 3.4897,
      "mean_token_accuracy": 0.3455218479037285,
      "num_tokens": 1183515.0,
      "step": 2710
    },
    {
      "epoch": 0.8138839018551766,
      "grad_norm": 27.0,
      "learning_rate": 3.644025533612607e-05,
      "loss": 3.6577,
      "mean_token_accuracy": 0.3339434161782265,
      "num_tokens": 1187074.0,
      "step": 2720
    },
    {
      "epoch": 0.8168761220825853,
      "grad_norm": 21.5,
      "learning_rate": 3.63903849990026e-05,
      "loss": 3.5652,
      "mean_token_accuracy": 0.34130973666906356,
      "num_tokens": 1192494.0,
      "step": 2730
    },
    {
      "epoch": 0.8198683423099941,
      "grad_norm": 28.125,
      "learning_rate": 3.6340514661879114e-05,
      "loss": 2.8965,
      "mean_token_accuracy": 0.42962965965270994,
      "num_tokens": 1195933.0,
      "step": 2740
    },
    {
      "epoch": 0.8228605625374028,
      "grad_norm": 32.75,
      "learning_rate": 3.629064432475563e-05,
      "loss": 2.9371,
      "mean_token_accuracy": 0.4335357487201691,
      "num_tokens": 1200582.0,
      "step": 2750
    },
    {
      "epoch": 0.8258527827648114,
      "grad_norm": 36.75,
      "learning_rate": 3.624077398763216e-05,
      "loss": 3.3401,
      "mean_token_accuracy": 0.38103923499584197,
      "num_tokens": 1206875.0,
      "step": 2760
    },
    {
      "epoch": 0.8288450029922202,
      "grad_norm": 26.0,
      "learning_rate": 3.619090365050868e-05,
      "loss": 2.6235,
      "mean_token_accuracy": 0.4731802463531494,
      "num_tokens": 1211109.0,
      "step": 2770
    },
    {
      "epoch": 0.8318372232196289,
      "grad_norm": 36.25,
      "learning_rate": 3.61410333133852e-05,
      "loss": 3.0112,
      "mean_token_accuracy": 0.4310316130518913,
      "num_tokens": 1215329.0,
      "step": 2780
    },
    {
      "epoch": 0.8348294434470377,
      "grad_norm": 26.75,
      "learning_rate": 3.609116297626172e-05,
      "loss": 2.9925,
      "mean_token_accuracy": 0.4128095477819443,
      "num_tokens": 1218332.0,
      "step": 2790
    },
    {
      "epoch": 0.8378216636744464,
      "grad_norm": 36.5,
      "learning_rate": 3.604129263913824e-05,
      "loss": 2.9043,
      "mean_token_accuracy": 0.4450998976826668,
      "num_tokens": 1221567.0,
      "step": 2800
    },
    {
      "epoch": 0.8408138839018552,
      "grad_norm": 23.75,
      "learning_rate": 3.5991422302014763e-05,
      "loss": 3.5922,
      "mean_token_accuracy": 0.3619121313095093,
      "num_tokens": 1226914.0,
      "step": 2810
    },
    {
      "epoch": 0.8438061041292639,
      "grad_norm": 31.875,
      "learning_rate": 3.594155196489129e-05,
      "loss": 2.9469,
      "mean_token_accuracy": 0.4257675677537918,
      "num_tokens": 1231099.0,
      "step": 2820
    },
    {
      "epoch": 0.8467983243566727,
      "grad_norm": 27.625,
      "learning_rate": 3.5891681627767805e-05,
      "loss": 3.5225,
      "mean_token_accuracy": 0.39053993821144106,
      "num_tokens": 1234858.0,
      "step": 2830
    },
    {
      "epoch": 0.8497905445840814,
      "grad_norm": 21.75,
      "learning_rate": 3.584181129064433e-05,
      "loss": 2.6881,
      "mean_token_accuracy": 0.4483701854944229,
      "num_tokens": 1239315.0,
      "step": 2840
    },
    {
      "epoch": 0.8527827648114902,
      "grad_norm": 32.5,
      "learning_rate": 3.5791940953520847e-05,
      "loss": 3.7535,
      "mean_token_accuracy": 0.3519248738884926,
      "num_tokens": 1244480.0,
      "step": 2850
    },
    {
      "epoch": 0.8557749850388988,
      "grad_norm": 46.25,
      "learning_rate": 3.5742070616397364e-05,
      "loss": 3.0964,
      "mean_token_accuracy": 0.4106554016470909,
      "num_tokens": 1249519.0,
      "step": 2860
    },
    {
      "epoch": 0.8587672052663076,
      "grad_norm": 35.25,
      "learning_rate": 3.569220027927389e-05,
      "loss": 3.2774,
      "mean_token_accuracy": 0.3928678810596466,
      "num_tokens": 1254161.0,
      "step": 2870
    },
    {
      "epoch": 0.8617594254937163,
      "grad_norm": 26.25,
      "learning_rate": 3.564232994215041e-05,
      "loss": 3.1824,
      "mean_token_accuracy": 0.39329985678195956,
      "num_tokens": 1257687.0,
      "step": 2880
    },
    {
      "epoch": 0.8647516457211251,
      "grad_norm": 23.875,
      "learning_rate": 3.5592459605026937e-05,
      "loss": 3.3218,
      "mean_token_accuracy": 0.40274937748908995,
      "num_tokens": 1261620.0,
      "step": 2890
    },
    {
      "epoch": 0.8677438659485338,
      "grad_norm": 36.0,
      "learning_rate": 3.5542589267903454e-05,
      "loss": 3.1617,
      "mean_token_accuracy": 0.386403800547123,
      "num_tokens": 1266555.0,
      "step": 2900
    },
    {
      "epoch": 0.8707360861759426,
      "grad_norm": 44.25,
      "learning_rate": 3.549271893077997e-05,
      "loss": 3.1503,
      "mean_token_accuracy": 0.40236749649047854,
      "num_tokens": 1270915.0,
      "step": 2910
    },
    {
      "epoch": 0.8737283064033513,
      "grad_norm": 31.875,
      "learning_rate": 3.5442848593656496e-05,
      "loss": 2.9395,
      "mean_token_accuracy": 0.4497757613658905,
      "num_tokens": 1274978.0,
      "step": 2920
    },
    {
      "epoch": 0.87672052663076,
      "grad_norm": 27.375,
      "learning_rate": 3.539297825653301e-05,
      "loss": 3.525,
      "mean_token_accuracy": 0.35292821377515793,
      "num_tokens": 1280176.0,
      "step": 2930
    },
    {
      "epoch": 0.8797127468581688,
      "grad_norm": 27.5,
      "learning_rate": 3.534310791940954e-05,
      "loss": 3.0989,
      "mean_token_accuracy": 0.39904648810625076,
      "num_tokens": 1284768.0,
      "step": 2940
    },
    {
      "epoch": 0.8827049670855776,
      "grad_norm": 22.75,
      "learning_rate": 3.529323758228606e-05,
      "loss": 2.5238,
      "mean_token_accuracy": 0.5021311789751053,
      "num_tokens": 1288030.0,
      "step": 2950
    },
    {
      "epoch": 0.8856971873129862,
      "grad_norm": 27.0,
      "learning_rate": 3.524336724516258e-05,
      "loss": 2.9962,
      "mean_token_accuracy": 0.4204836696386337,
      "num_tokens": 1292102.0,
      "step": 2960
    },
    {
      "epoch": 0.8886894075403949,
      "grad_norm": 31.625,
      "learning_rate": 3.5193496908039096e-05,
      "loss": 3.1125,
      "mean_token_accuracy": 0.4044138967990875,
      "num_tokens": 1296788.0,
      "step": 2970
    },
    {
      "epoch": 0.8916816277678037,
      "grad_norm": 23.125,
      "learning_rate": 3.514362657091562e-05,
      "loss": 3.3551,
      "mean_token_accuracy": 0.3792631521821022,
      "num_tokens": 1301386.0,
      "step": 2980
    },
    {
      "epoch": 0.8946738479952124,
      "grad_norm": 46.25,
      "learning_rate": 3.509375623379214e-05,
      "loss": 3.3053,
      "mean_token_accuracy": 0.39626367688179015,
      "num_tokens": 1306384.0,
      "step": 2990
    },
    {
      "epoch": 0.8976660682226212,
      "grad_norm": 41.0,
      "learning_rate": 3.504388589666867e-05,
      "loss": 3.2402,
      "mean_token_accuracy": 0.3704844668507576,
      "num_tokens": 1311546.0,
      "step": 3000
    },
    {
      "epoch": 0.9006582884500299,
      "grad_norm": 32.75,
      "learning_rate": 3.4994015559545186e-05,
      "loss": 2.9308,
      "mean_token_accuracy": 0.43605942130088804,
      "num_tokens": 1315636.0,
      "step": 3010
    },
    {
      "epoch": 0.9036505086774387,
      "grad_norm": 22.125,
      "learning_rate": 3.4944145222421703e-05,
      "loss": 3.0013,
      "mean_token_accuracy": 0.4220082849264145,
      "num_tokens": 1320955.0,
      "step": 3020
    },
    {
      "epoch": 0.9066427289048474,
      "grad_norm": 26.625,
      "learning_rate": 3.489427488529823e-05,
      "loss": 3.1008,
      "mean_token_accuracy": 0.42272399365901947,
      "num_tokens": 1324305.0,
      "step": 3030
    },
    {
      "epoch": 0.9096349491322562,
      "grad_norm": 21.75,
      "learning_rate": 3.4844404548174745e-05,
      "loss": 2.5731,
      "mean_token_accuracy": 0.4852191686630249,
      "num_tokens": 1327968.0,
      "step": 3040
    },
    {
      "epoch": 0.9126271693596649,
      "grad_norm": 23.0,
      "learning_rate": 3.479453421105127e-05,
      "loss": 2.9561,
      "mean_token_accuracy": 0.4218144237995148,
      "num_tokens": 1331005.0,
      "step": 3050
    },
    {
      "epoch": 0.9156193895870736,
      "grad_norm": 25.125,
      "learning_rate": 3.4744663873927793e-05,
      "loss": 3.4899,
      "mean_token_accuracy": 0.3722094401717186,
      "num_tokens": 1335408.0,
      "step": 3060
    },
    {
      "epoch": 0.9186116098144823,
      "grad_norm": 37.5,
      "learning_rate": 3.469479353680431e-05,
      "loss": 3.4433,
      "mean_token_accuracy": 0.358016200363636,
      "num_tokens": 1339542.0,
      "step": 3070
    },
    {
      "epoch": 0.9216038300418911,
      "grad_norm": 34.25,
      "learning_rate": 3.464492319968083e-05,
      "loss": 2.8409,
      "mean_token_accuracy": 0.421732684969902,
      "num_tokens": 1343054.0,
      "step": 3080
    },
    {
      "epoch": 0.9245960502692998,
      "grad_norm": 24.375,
      "learning_rate": 3.459505286255735e-05,
      "loss": 2.6645,
      "mean_token_accuracy": 0.47050471752882006,
      "num_tokens": 1348184.0,
      "step": 3090
    },
    {
      "epoch": 0.9275882704967086,
      "grad_norm": 30.5,
      "learning_rate": 3.454518252543387e-05,
      "loss": 2.9127,
      "mean_token_accuracy": 0.4304572999477386,
      "num_tokens": 1352551.0,
      "step": 3100
    },
    {
      "epoch": 0.9305804907241173,
      "grad_norm": 32.0,
      "learning_rate": 3.4495312188310394e-05,
      "loss": 3.4285,
      "mean_token_accuracy": 0.3730440601706505,
      "num_tokens": 1355944.0,
      "step": 3110
    },
    {
      "epoch": 0.933572710951526,
      "grad_norm": 23.625,
      "learning_rate": 3.444544185118692e-05,
      "loss": 3.1287,
      "mean_token_accuracy": 0.41855313777923586,
      "num_tokens": 1359442.0,
      "step": 3120
    },
    {
      "epoch": 0.9365649311789348,
      "grad_norm": 30.75,
      "learning_rate": 3.4395571514063436e-05,
      "loss": 2.7247,
      "mean_token_accuracy": 0.454880079627037,
      "num_tokens": 1364185.0,
      "step": 3130
    },
    {
      "epoch": 0.9395571514063435,
      "grad_norm": 33.0,
      "learning_rate": 3.434570117693996e-05,
      "loss": 3.4506,
      "mean_token_accuracy": 0.3528758123517036,
      "num_tokens": 1369430.0,
      "step": 3140
    },
    {
      "epoch": 0.9425493716337523,
      "grad_norm": 32.25,
      "learning_rate": 3.429583083981648e-05,
      "loss": 2.5508,
      "mean_token_accuracy": 0.5001329183578491,
      "num_tokens": 1374371.0,
      "step": 3150
    },
    {
      "epoch": 0.9455415918611609,
      "grad_norm": 33.75,
      "learning_rate": 3.4245960502692995e-05,
      "loss": 3.2475,
      "mean_token_accuracy": 0.38751300275325773,
      "num_tokens": 1379901.0,
      "step": 3160
    },
    {
      "epoch": 0.9485338120885697,
      "grad_norm": 29.875,
      "learning_rate": 3.419609016556952e-05,
      "loss": 3.3707,
      "mean_token_accuracy": 0.3779307246208191,
      "num_tokens": 1383740.0,
      "step": 3170
    },
    {
      "epoch": 0.9515260323159784,
      "grad_norm": 28.375,
      "learning_rate": 3.414621982844604e-05,
      "loss": 3.0298,
      "mean_token_accuracy": 0.4013781577348709,
      "num_tokens": 1388384.0,
      "step": 3180
    },
    {
      "epoch": 0.9545182525433872,
      "grad_norm": 31.625,
      "learning_rate": 3.409634949132257e-05,
      "loss": 2.6263,
      "mean_token_accuracy": 0.4786448389291763,
      "num_tokens": 1392084.0,
      "step": 3190
    },
    {
      "epoch": 0.9575104727707959,
      "grad_norm": 29.25,
      "learning_rate": 3.4046479154199085e-05,
      "loss": 3.0048,
      "mean_token_accuracy": 0.4301429718732834,
      "num_tokens": 1397018.0,
      "step": 3200
    },
    {
      "epoch": 0.9605026929982047,
      "grad_norm": 24.5,
      "learning_rate": 3.39966088170756e-05,
      "loss": 2.8107,
      "mean_token_accuracy": 0.4548857152462006,
      "num_tokens": 1403173.0,
      "step": 3210
    },
    {
      "epoch": 0.9634949132256134,
      "grad_norm": 23.625,
      "learning_rate": 3.3946738479952126e-05,
      "loss": 2.7295,
      "mean_token_accuracy": 0.430012971162796,
      "num_tokens": 1407652.0,
      "step": 3220
    },
    {
      "epoch": 0.9664871334530222,
      "grad_norm": 21.375,
      "learning_rate": 3.389686814282865e-05,
      "loss": 2.7726,
      "mean_token_accuracy": 0.4541109263896942,
      "num_tokens": 1412058.0,
      "step": 3230
    },
    {
      "epoch": 0.9694793536804309,
      "grad_norm": 28.375,
      "learning_rate": 3.384699780570517e-05,
      "loss": 3.0506,
      "mean_token_accuracy": 0.42236119508743286,
      "num_tokens": 1417616.0,
      "step": 3240
    },
    {
      "epoch": 0.9724715739078397,
      "grad_norm": 32.75,
      "learning_rate": 3.379712746858169e-05,
      "loss": 2.8471,
      "mean_token_accuracy": 0.4267496675252914,
      "num_tokens": 1422617.0,
      "step": 3250
    },
    {
      "epoch": 0.9754637941352483,
      "grad_norm": 31.875,
      "learning_rate": 3.374725713145821e-05,
      "loss": 3.2055,
      "mean_token_accuracy": 0.3901287063956261,
      "num_tokens": 1429066.0,
      "step": 3260
    },
    {
      "epoch": 0.9784560143626571,
      "grad_norm": 35.5,
      "learning_rate": 3.369738679433473e-05,
      "loss": 2.8184,
      "mean_token_accuracy": 0.43766441345214846,
      "num_tokens": 1434896.0,
      "step": 3270
    },
    {
      "epoch": 0.9814482345900658,
      "grad_norm": 46.5,
      "learning_rate": 3.364751645721125e-05,
      "loss": 2.9339,
      "mean_token_accuracy": 0.4233125030994415,
      "num_tokens": 1438794.0,
      "step": 3280
    },
    {
      "epoch": 0.9844404548174746,
      "grad_norm": 34.75,
      "learning_rate": 3.3597646120087775e-05,
      "loss": 2.8863,
      "mean_token_accuracy": 0.45387964099645617,
      "num_tokens": 1442559.0,
      "step": 3290
    },
    {
      "epoch": 0.9874326750448833,
      "grad_norm": 24.75,
      "learning_rate": 3.35477757829643e-05,
      "loss": 2.6063,
      "mean_token_accuracy": 0.4726902186870575,
      "num_tokens": 1447577.0,
      "step": 3300
    },
    {
      "epoch": 0.990424895272292,
      "grad_norm": 23.75,
      "learning_rate": 3.349790544584082e-05,
      "loss": 3.272,
      "mean_token_accuracy": 0.3847466349601746,
      "num_tokens": 1451271.0,
      "step": 3310
    },
    {
      "epoch": 0.9934171154997008,
      "grad_norm": 25.25,
      "learning_rate": 3.3448035108717334e-05,
      "loss": 3.2839,
      "mean_token_accuracy": 0.3943710759282112,
      "num_tokens": 1455126.0,
      "step": 3320
    },
    {
      "epoch": 0.9964093357271095,
      "grad_norm": 33.5,
      "learning_rate": 3.339816477159386e-05,
      "loss": 2.2571,
      "mean_token_accuracy": 0.5220259606838227,
      "num_tokens": 1457975.0,
      "step": 3330
    },
    {
      "epoch": 0.9994015559545183,
      "grad_norm": 28.625,
      "learning_rate": 3.3348294434470376e-05,
      "loss": 3.0659,
      "mean_token_accuracy": 0.4067483752965927,
      "num_tokens": 1463249.0,
      "step": 3340
    },
    {
      "epoch": 1.0,
      "eval_loss": 2.980771780014038,
      "eval_mean_token_accuracy": 0.42323312556968545,
      "eval_num_tokens": 1464244.0,
      "eval_runtime": 28.073,
      "eval_samples_per_second": 14.89,
      "eval_steps_per_second": 1.888,
      "step": 3342
    },
    {
      "epoch": 1.002393776181927,
      "grad_norm": 39.75,
      "learning_rate": 3.32984240973469e-05,
      "loss": 3.3756,
      "mean_token_accuracy": 0.35648047029972074,
      "num_tokens": 1467102.0,
      "step": 3350
    },
    {
      "epoch": 1.0053859964093357,
      "grad_norm": 22.625,
      "learning_rate": 3.3248553760223424e-05,
      "loss": 2.4231,
      "mean_token_accuracy": 0.49451322853565216,
      "num_tokens": 1472322.0,
      "step": 3360
    },
    {
      "epoch": 1.0083782166367445,
      "grad_norm": 21.875,
      "learning_rate": 3.319868342309994e-05,
      "loss": 2.4485,
      "mean_token_accuracy": 0.49648206532001493,
      "num_tokens": 1477093.0,
      "step": 3370
    },
    {
      "epoch": 1.0113704368641532,
      "grad_norm": 27.875,
      "learning_rate": 3.314881308597646e-05,
      "loss": 2.6138,
      "mean_token_accuracy": 0.4813783347606659,
      "num_tokens": 1480728.0,
      "step": 3380
    },
    {
      "epoch": 1.014362657091562,
      "grad_norm": 45.75,
      "learning_rate": 3.309894274885298e-05,
      "loss": 2.6418,
      "mean_token_accuracy": 0.4929414004087448,
      "num_tokens": 1485390.0,
      "step": 3390
    },
    {
      "epoch": 1.0173548773189707,
      "grad_norm": 27.875,
      "learning_rate": 3.30490724117295e-05,
      "loss": 2.4796,
      "mean_token_accuracy": 0.49346729218959806,
      "num_tokens": 1491122.0,
      "step": 3400
    },
    {
      "epoch": 1.0203470975463793,
      "grad_norm": 29.375,
      "learning_rate": 3.299920207460603e-05,
      "loss": 2.6773,
      "mean_token_accuracy": 0.4725665241479874,
      "num_tokens": 1495421.0,
      "step": 3410
    },
    {
      "epoch": 1.0233393177737882,
      "grad_norm": 43.5,
      "learning_rate": 3.294933173748255e-05,
      "loss": 3.0846,
      "mean_token_accuracy": 0.419183561205864,
      "num_tokens": 1499769.0,
      "step": 3420
    },
    {
      "epoch": 1.0263315380011968,
      "grad_norm": 33.75,
      "learning_rate": 3.2899461400359066e-05,
      "loss": 2.6298,
      "mean_token_accuracy": 0.4962582170963287,
      "num_tokens": 1504170.0,
      "step": 3430
    },
    {
      "epoch": 1.0293237582286057,
      "grad_norm": 20.5,
      "learning_rate": 3.284959106323559e-05,
      "loss": 2.5207,
      "mean_token_accuracy": 0.4708354324102402,
      "num_tokens": 1508664.0,
      "step": 3440
    },
    {
      "epoch": 1.0323159784560143,
      "grad_norm": 37.25,
      "learning_rate": 3.279972072611211e-05,
      "loss": 2.6655,
      "mean_token_accuracy": 0.4595662534236908,
      "num_tokens": 1514082.0,
      "step": 3450
    },
    {
      "epoch": 1.0353081986834232,
      "grad_norm": 24.875,
      "learning_rate": 3.274985038898863e-05,
      "loss": 2.8414,
      "mean_token_accuracy": 0.4344394147396088,
      "num_tokens": 1519310.0,
      "step": 3460
    },
    {
      "epoch": 1.0383004189108318,
      "grad_norm": 87.5,
      "learning_rate": 3.2699980051865156e-05,
      "loss": 2.5733,
      "mean_token_accuracy": 0.48141989260911944,
      "num_tokens": 1522961.0,
      "step": 3470
    },
    {
      "epoch": 1.0412926391382407,
      "grad_norm": 24.25,
      "learning_rate": 3.2650109714741674e-05,
      "loss": 2.6847,
      "mean_token_accuracy": 0.4702356606721878,
      "num_tokens": 1527447.0,
      "step": 3480
    },
    {
      "epoch": 1.0442848593656493,
      "grad_norm": 33.75,
      "learning_rate": 3.26002393776182e-05,
      "loss": 2.6967,
      "mean_token_accuracy": 0.4401080310344696,
      "num_tokens": 1532143.0,
      "step": 3490
    },
    {
      "epoch": 1.0472770795930582,
      "grad_norm": 39.75,
      "learning_rate": 3.2550369040494715e-05,
      "loss": 2.9048,
      "mean_token_accuracy": 0.4236243531107903,
      "num_tokens": 1535966.0,
      "step": 3500
    },
    {
      "epoch": 1.0502692998204668,
      "grad_norm": 32.25,
      "learning_rate": 3.250049870337123e-05,
      "loss": 2.8366,
      "mean_token_accuracy": 0.4745761215686798,
      "num_tokens": 1540192.0,
      "step": 3510
    },
    {
      "epoch": 1.0532615200478754,
      "grad_norm": 40.25,
      "learning_rate": 3.245062836624776e-05,
      "loss": 2.5117,
      "mean_token_accuracy": 0.4899686545133591,
      "num_tokens": 1544596.0,
      "step": 3520
    },
    {
      "epoch": 1.0562537402752843,
      "grad_norm": 26.25,
      "learning_rate": 3.240075802912428e-05,
      "loss": 2.6037,
      "mean_token_accuracy": 0.47316262423992156,
      "num_tokens": 1550210.0,
      "step": 3530
    },
    {
      "epoch": 1.059245960502693,
      "grad_norm": 22.0,
      "learning_rate": 3.23508876920008e-05,
      "loss": 2.8637,
      "mean_token_accuracy": 0.44064879715442656,
      "num_tokens": 1554533.0,
      "step": 3540
    },
    {
      "epoch": 1.0622381807301018,
      "grad_norm": 38.0,
      "learning_rate": 3.230101735487732e-05,
      "loss": 2.6214,
      "mean_token_accuracy": 0.47286906838417053,
      "num_tokens": 1560617.0,
      "step": 3550
    },
    {
      "epoch": 1.0652304009575104,
      "grad_norm": 27.875,
      "learning_rate": 3.225114701775384e-05,
      "loss": 2.8798,
      "mean_token_accuracy": 0.4531641513109207,
      "num_tokens": 1563908.0,
      "step": 3560
    },
    {
      "epoch": 1.0682226211849193,
      "grad_norm": 37.75,
      "learning_rate": 3.220127668063036e-05,
      "loss": 2.8749,
      "mean_token_accuracy": 0.43050872087478637,
      "num_tokens": 1568474.0,
      "step": 3570
    },
    {
      "epoch": 1.071214841412328,
      "grad_norm": 24.875,
      "learning_rate": 3.215140634350688e-05,
      "loss": 3.0744,
      "mean_token_accuracy": 0.4257981926202774,
      "num_tokens": 1572746.0,
      "step": 3580
    },
    {
      "epoch": 1.0742070616397368,
      "grad_norm": 40.5,
      "learning_rate": 3.2101536006383406e-05,
      "loss": 2.5789,
      "mean_token_accuracy": 0.47388690114021303,
      "num_tokens": 1578454.0,
      "step": 3590
    },
    {
      "epoch": 1.0771992818671454,
      "grad_norm": 29.75,
      "learning_rate": 3.205166566925993e-05,
      "loss": 2.9829,
      "mean_token_accuracy": 0.4279584616422653,
      "num_tokens": 1583489.0,
      "step": 3600
    },
    {
      "epoch": 1.080191502094554,
      "grad_norm": 37.0,
      "learning_rate": 3.200179533213645e-05,
      "loss": 2.7567,
      "mean_token_accuracy": 0.4380374073982239,
      "num_tokens": 1586955.0,
      "step": 3610
    },
    {
      "epoch": 1.083183722321963,
      "grad_norm": 21.75,
      "learning_rate": 3.1951924995012965e-05,
      "loss": 3.1002,
      "mean_token_accuracy": 0.40744185745716094,
      "num_tokens": 1592004.0,
      "step": 3620
    },
    {
      "epoch": 1.0861759425493716,
      "grad_norm": 50.75,
      "learning_rate": 3.190205465788949e-05,
      "loss": 2.9584,
      "mean_token_accuracy": 0.4398718744516373,
      "num_tokens": 1597537.0,
      "step": 3630
    },
    {
      "epoch": 1.0891681627767804,
      "grad_norm": 29.125,
      "learning_rate": 3.185218432076601e-05,
      "loss": 2.6612,
      "mean_token_accuracy": 0.4644526898860931,
      "num_tokens": 1601772.0,
      "step": 3640
    },
    {
      "epoch": 1.092160383004189,
      "grad_norm": 28.5,
      "learning_rate": 3.180231398364253e-05,
      "loss": 2.3019,
      "mean_token_accuracy": 0.5160307466983796,
      "num_tokens": 1605652.0,
      "step": 3650
    },
    {
      "epoch": 1.095152603231598,
      "grad_norm": 29.5,
      "learning_rate": 3.1752443646519055e-05,
      "loss": 3.1181,
      "mean_token_accuracy": 0.3980431482195854,
      "num_tokens": 1609800.0,
      "step": 3660
    },
    {
      "epoch": 1.0981448234590065,
      "grad_norm": 39.5,
      "learning_rate": 3.170257330939557e-05,
      "loss": 2.772,
      "mean_token_accuracy": 0.4531924784183502,
      "num_tokens": 1614274.0,
      "step": 3670
    },
    {
      "epoch": 1.1011370436864154,
      "grad_norm": 26.125,
      "learning_rate": 3.165270297227209e-05,
      "loss": 2.3661,
      "mean_token_accuracy": 0.514206713438034,
      "num_tokens": 1617770.0,
      "step": 3680
    },
    {
      "epoch": 1.104129263913824,
      "grad_norm": 27.875,
      "learning_rate": 3.1602832635148614e-05,
      "loss": 3.0596,
      "mean_token_accuracy": 0.4163840651512146,
      "num_tokens": 1621234.0,
      "step": 3690
    },
    {
      "epoch": 1.1071214841412327,
      "grad_norm": 53.5,
      "learning_rate": 3.155296229802514e-05,
      "loss": 2.6817,
      "mean_token_accuracy": 0.47163409888744356,
      "num_tokens": 1625009.0,
      "step": 3700
    },
    {
      "epoch": 1.1101137043686415,
      "grad_norm": 24.125,
      "learning_rate": 3.150309196090166e-05,
      "loss": 3.2935,
      "mean_token_accuracy": 0.3971044301986694,
      "num_tokens": 1630070.0,
      "step": 3710
    },
    {
      "epoch": 1.1131059245960502,
      "grad_norm": 24.375,
      "learning_rate": 3.145322162377818e-05,
      "loss": 2.5974,
      "mean_token_accuracy": 0.4851890727877617,
      "num_tokens": 1634810.0,
      "step": 3720
    },
    {
      "epoch": 1.116098144823459,
      "grad_norm": 40.5,
      "learning_rate": 3.14033512866547e-05,
      "loss": 2.5665,
      "mean_token_accuracy": 0.48123776614665986,
      "num_tokens": 1638819.0,
      "step": 3730
    },
    {
      "epoch": 1.1190903650508677,
      "grad_norm": 23.75,
      "learning_rate": 3.135348094953122e-05,
      "loss": 2.4931,
      "mean_token_accuracy": 0.49635804891586305,
      "num_tokens": 1644484.0,
      "step": 3740
    },
    {
      "epoch": 1.1220825852782765,
      "grad_norm": 36.25,
      "learning_rate": 3.130361061240774e-05,
      "loss": 2.9514,
      "mean_token_accuracy": 0.4365529015660286,
      "num_tokens": 1648231.0,
      "step": 3750
    },
    {
      "epoch": 1.1250748055056852,
      "grad_norm": 22.625,
      "learning_rate": 3.125374027528426e-05,
      "loss": 2.9057,
      "mean_token_accuracy": 0.4231098830699921,
      "num_tokens": 1653761.0,
      "step": 3760
    },
    {
      "epoch": 1.128067025733094,
      "grad_norm": 27.0,
      "learning_rate": 3.120386993816079e-05,
      "loss": 2.6079,
      "mean_token_accuracy": 0.5077326536178589,
      "num_tokens": 1657623.0,
      "step": 3770
    },
    {
      "epoch": 1.1310592459605027,
      "grad_norm": 24.625,
      "learning_rate": 3.1153999601037304e-05,
      "loss": 2.7906,
      "mean_token_accuracy": 0.4419715031981468,
      "num_tokens": 1661939.0,
      "step": 3780
    },
    {
      "epoch": 1.1340514661879113,
      "grad_norm": 28.125,
      "learning_rate": 3.110412926391383e-05,
      "loss": 2.4565,
      "mean_token_accuracy": 0.5014745116233825,
      "num_tokens": 1666447.0,
      "step": 3790
    },
    {
      "epoch": 1.1370436864153202,
      "grad_norm": 40.75,
      "learning_rate": 3.1054258926790346e-05,
      "loss": 2.3208,
      "mean_token_accuracy": 0.5103875666856765,
      "num_tokens": 1671267.0,
      "step": 3800
    },
    {
      "epoch": 1.140035906642729,
      "grad_norm": 31.375,
      "learning_rate": 3.100438858966686e-05,
      "loss": 3.0056,
      "mean_token_accuracy": 0.4169232577085495,
      "num_tokens": 1675208.0,
      "step": 3810
    },
    {
      "epoch": 1.1430281268701377,
      "grad_norm": 25.875,
      "learning_rate": 3.0954518252543394e-05,
      "loss": 2.7892,
      "mean_token_accuracy": 0.4438593596220016,
      "num_tokens": 1679847.0,
      "step": 3820
    },
    {
      "epoch": 1.1460203470975463,
      "grad_norm": 26.375,
      "learning_rate": 3.090464791541991e-05,
      "loss": 2.8558,
      "mean_token_accuracy": 0.4273435175418854,
      "num_tokens": 1684286.0,
      "step": 3830
    },
    {
      "epoch": 1.1490125673249552,
      "grad_norm": 69.0,
      "learning_rate": 3.085477757829643e-05,
      "loss": 2.9511,
      "mean_token_accuracy": 0.4328667402267456,
      "num_tokens": 1687901.0,
      "step": 3840
    },
    {
      "epoch": 1.1520047875523638,
      "grad_norm": 35.0,
      "learning_rate": 3.080490724117295e-05,
      "loss": 2.7366,
      "mean_token_accuracy": 0.45511636435985564,
      "num_tokens": 1691925.0,
      "step": 3850
    },
    {
      "epoch": 1.1549970077797727,
      "grad_norm": 32.75,
      "learning_rate": 3.075503690404947e-05,
      "loss": 2.3323,
      "mean_token_accuracy": 0.4906174510717392,
      "num_tokens": 1696413.0,
      "step": 3860
    },
    {
      "epoch": 1.1579892280071813,
      "grad_norm": 47.5,
      "learning_rate": 3.0705166566925995e-05,
      "loss": 3.2158,
      "mean_token_accuracy": 0.4027450352907181,
      "num_tokens": 1700064.0,
      "step": 3870
    },
    {
      "epoch": 1.1609814482345902,
      "grad_norm": 29.375,
      "learning_rate": 3.065529622980252e-05,
      "loss": 2.4488,
      "mean_token_accuracy": 0.49048165082931516,
      "num_tokens": 1704477.0,
      "step": 3880
    },
    {
      "epoch": 1.1639736684619988,
      "grad_norm": 23.125,
      "learning_rate": 3.0605425892679036e-05,
      "loss": 2.9511,
      "mean_token_accuracy": 0.41550472676753997,
      "num_tokens": 1709655.0,
      "step": 3890
    },
    {
      "epoch": 1.1669658886894076,
      "grad_norm": 24.5,
      "learning_rate": 3.055555555555556e-05,
      "loss": 2.4906,
      "mean_token_accuracy": 0.4970557034015656,
      "num_tokens": 1714138.0,
      "step": 3900
    },
    {
      "epoch": 1.1699581089168163,
      "grad_norm": 28.0,
      "learning_rate": 3.0505685218432078e-05,
      "loss": 2.1077,
      "mean_token_accuracy": 0.5406092315912246,
      "num_tokens": 1717129.0,
      "step": 3910
    },
    {
      "epoch": 1.172950329144225,
      "grad_norm": 27.75,
      "learning_rate": 3.04558148813086e-05,
      "loss": 2.7482,
      "mean_token_accuracy": 0.458242666721344,
      "num_tokens": 1722600.0,
      "step": 3920
    },
    {
      "epoch": 1.1759425493716338,
      "grad_norm": 24.5,
      "learning_rate": 3.0405944544185123e-05,
      "loss": 2.4023,
      "mean_token_accuracy": 0.49710969924926757,
      "num_tokens": 1727134.0,
      "step": 3930
    },
    {
      "epoch": 1.1789347695990424,
      "grad_norm": 35.75,
      "learning_rate": 3.035607420706164e-05,
      "loss": 2.7705,
      "mean_token_accuracy": 0.4731149792671204,
      "num_tokens": 1731436.0,
      "step": 3940
    },
    {
      "epoch": 1.1819269898264513,
      "grad_norm": 32.0,
      "learning_rate": 3.030620386993816e-05,
      "loss": 2.5216,
      "mean_token_accuracy": 0.49482154846191406,
      "num_tokens": 1734311.0,
      "step": 3950
    },
    {
      "epoch": 1.18491921005386,
      "grad_norm": 21.875,
      "learning_rate": 3.0256333532814685e-05,
      "loss": 2.6194,
      "mean_token_accuracy": 0.4716911166906357,
      "num_tokens": 1739795.0,
      "step": 3960
    },
    {
      "epoch": 1.1879114302812688,
      "grad_norm": 41.0,
      "learning_rate": 3.0206463195691203e-05,
      "loss": 2.5254,
      "mean_token_accuracy": 0.49075353145599365,
      "num_tokens": 1743697.0,
      "step": 3970
    },
    {
      "epoch": 1.1909036505086774,
      "grad_norm": 23.875,
      "learning_rate": 3.0156592858567723e-05,
      "loss": 2.7566,
      "mean_token_accuracy": 0.46060216426849365,
      "num_tokens": 1747837.0,
      "step": 3980
    },
    {
      "epoch": 1.1938958707360863,
      "grad_norm": 23.375,
      "learning_rate": 3.0106722521444248e-05,
      "loss": 2.4751,
      "mean_token_accuracy": 0.488738951086998,
      "num_tokens": 1752378.0,
      "step": 3990
    },
    {
      "epoch": 1.196888090963495,
      "grad_norm": 25.25,
      "learning_rate": 3.0056852184320765e-05,
      "loss": 2.7018,
      "mean_token_accuracy": 0.47838138192892077,
      "num_tokens": 1755508.0,
      "step": 4000
    },
    {
      "epoch": 1.1998803111909035,
      "grad_norm": 39.5,
      "learning_rate": 3.000698184719729e-05,
      "loss": 3.0874,
      "mean_token_accuracy": 0.42261676490306854,
      "num_tokens": 1760372.0,
      "step": 4010
    },
    {
      "epoch": 1.2028725314183124,
      "grad_norm": 33.5,
      "learning_rate": 2.995711151007381e-05,
      "loss": 2.0667,
      "mean_token_accuracy": 0.5681251019239426,
      "num_tokens": 1764165.0,
      "step": 4020
    },
    {
      "epoch": 1.205864751645721,
      "grad_norm": 28.625,
      "learning_rate": 2.9907241172950327e-05,
      "loss": 2.7279,
      "mean_token_accuracy": 0.46859243512153625,
      "num_tokens": 1769407.0,
      "step": 4030
    },
    {
      "epoch": 1.20885697187313,
      "grad_norm": 28.5,
      "learning_rate": 2.9857370835826855e-05,
      "loss": 2.3177,
      "mean_token_accuracy": 0.5130757451057434,
      "num_tokens": 1773247.0,
      "step": 4040
    },
    {
      "epoch": 1.2118491921005385,
      "grad_norm": 18.25,
      "learning_rate": 2.9807500498703372e-05,
      "loss": 2.614,
      "mean_token_accuracy": 0.5009590089321136,
      "num_tokens": 1778089.0,
      "step": 4050
    },
    {
      "epoch": 1.2148414123279474,
      "grad_norm": 50.0,
      "learning_rate": 2.975763016157989e-05,
      "loss": 2.474,
      "mean_token_accuracy": 0.5134980499744415,
      "num_tokens": 1781721.0,
      "step": 4060
    },
    {
      "epoch": 1.217833632555356,
      "grad_norm": 28.75,
      "learning_rate": 2.9707759824456417e-05,
      "loss": 2.7975,
      "mean_token_accuracy": 0.45258464217185973,
      "num_tokens": 1786756.0,
      "step": 4070
    },
    {
      "epoch": 1.220825852782765,
      "grad_norm": 34.5,
      "learning_rate": 2.9657889487332935e-05,
      "loss": 2.6269,
      "mean_token_accuracy": 0.46714275777339936,
      "num_tokens": 1791148.0,
      "step": 4080
    },
    {
      "epoch": 1.2238180730101735,
      "grad_norm": 17.25,
      "learning_rate": 2.960801915020946e-05,
      "loss": 2.8817,
      "mean_token_accuracy": 0.4374359607696533,
      "num_tokens": 1797159.0,
      "step": 4090
    },
    {
      "epoch": 1.2268102932375822,
      "grad_norm": 27.125,
      "learning_rate": 2.955814881308598e-05,
      "loss": 2.6874,
      "mean_token_accuracy": 0.4609710812568665,
      "num_tokens": 1802126.0,
      "step": 4100
    },
    {
      "epoch": 1.229802513464991,
      "grad_norm": 49.5,
      "learning_rate": 2.9508278475962497e-05,
      "loss": 2.8059,
      "mean_token_accuracy": 0.4560705676674843,
      "num_tokens": 1806970.0,
      "step": 4110
    },
    {
      "epoch": 1.2327947336923997,
      "grad_norm": 30.25,
      "learning_rate": 2.945840813883902e-05,
      "loss": 2.2505,
      "mean_token_accuracy": 0.5249142855405807,
      "num_tokens": 1810926.0,
      "step": 4120
    },
    {
      "epoch": 1.2357869539198085,
      "grad_norm": 39.25,
      "learning_rate": 2.9408537801715542e-05,
      "loss": 2.8528,
      "mean_token_accuracy": 0.4332689017057419,
      "num_tokens": 1814977.0,
      "step": 4130
    },
    {
      "epoch": 1.2387791741472172,
      "grad_norm": 26.75,
      "learning_rate": 2.935866746459206e-05,
      "loss": 2.786,
      "mean_token_accuracy": 0.4369647607207298,
      "num_tokens": 1819584.0,
      "step": 4140
    },
    {
      "epoch": 1.241771394374626,
      "grad_norm": 31.125,
      "learning_rate": 2.9308797127468584e-05,
      "loss": 2.5191,
      "mean_token_accuracy": 0.4885488286614418,
      "num_tokens": 1824482.0,
      "step": 4150
    },
    {
      "epoch": 1.2447636146020347,
      "grad_norm": 34.75,
      "learning_rate": 2.9258926790345104e-05,
      "loss": 2.6278,
      "mean_token_accuracy": 0.47676413506269455,
      "num_tokens": 1827908.0,
      "step": 4160
    },
    {
      "epoch": 1.2477558348294435,
      "grad_norm": 39.0,
      "learning_rate": 2.9209056453221622e-05,
      "loss": 2.7343,
      "mean_token_accuracy": 0.4897153973579407,
      "num_tokens": 1833197.0,
      "step": 4170
    },
    {
      "epoch": 1.2507480550568522,
      "grad_norm": 31.375,
      "learning_rate": 2.9159186116098146e-05,
      "loss": 2.9593,
      "mean_token_accuracy": 0.42748028337955474,
      "num_tokens": 1837651.0,
      "step": 4180
    },
    {
      "epoch": 1.2537402752842608,
      "grad_norm": 41.0,
      "learning_rate": 2.9109315778974667e-05,
      "loss": 2.5404,
      "mean_token_accuracy": 0.47800021767616274,
      "num_tokens": 1842555.0,
      "step": 4190
    },
    {
      "epoch": 1.2567324955116697,
      "grad_norm": 35.5,
      "learning_rate": 2.905944544185119e-05,
      "loss": 2.5295,
      "mean_token_accuracy": 0.47885517179965975,
      "num_tokens": 1846060.0,
      "step": 4200
    },
    {
      "epoch": 1.2597247157390785,
      "grad_norm": 33.5,
      "learning_rate": 2.900957510472771e-05,
      "loss": 2.7566,
      "mean_token_accuracy": 0.4646937608718872,
      "num_tokens": 1849781.0,
      "step": 4210
    },
    {
      "epoch": 1.2627169359664872,
      "grad_norm": 44.25,
      "learning_rate": 2.895970476760423e-05,
      "loss": 2.2948,
      "mean_token_accuracy": 0.5232088088989257,
      "num_tokens": 1853337.0,
      "step": 4220
    },
    {
      "epoch": 1.2657091561938958,
      "grad_norm": 26.125,
      "learning_rate": 2.8909834430480753e-05,
      "loss": 3.3095,
      "mean_token_accuracy": 0.38957217186689375,
      "num_tokens": 1857925.0,
      "step": 4230
    },
    {
      "epoch": 1.2687013764213046,
      "grad_norm": 33.0,
      "learning_rate": 2.885996409335727e-05,
      "loss": 2.3966,
      "mean_token_accuracy": 0.5166613161563873,
      "num_tokens": 1861656.0,
      "step": 4240
    },
    {
      "epoch": 1.2716935966487133,
      "grad_norm": 31.5,
      "learning_rate": 2.881009375623379e-05,
      "loss": 2.8243,
      "mean_token_accuracy": 0.4604656994342804,
      "num_tokens": 1866691.0,
      "step": 4250
    },
    {
      "epoch": 1.2746858168761221,
      "grad_norm": 32.75,
      "learning_rate": 2.8760223419110316e-05,
      "loss": 2.4936,
      "mean_token_accuracy": 0.4825888335704803,
      "num_tokens": 1871883.0,
      "step": 4260
    },
    {
      "epoch": 1.2776780371035308,
      "grad_norm": 24.5,
      "learning_rate": 2.8710353081986837e-05,
      "loss": 3.0457,
      "mean_token_accuracy": 0.44213865548372266,
      "num_tokens": 1876630.0,
      "step": 4270
    },
    {
      "epoch": 1.2806702573309394,
      "grad_norm": 31.875,
      "learning_rate": 2.8660482744863354e-05,
      "loss": 2.4731,
      "mean_token_accuracy": 0.5032434403896332,
      "num_tokens": 1882195.0,
      "step": 4280
    },
    {
      "epoch": 1.2836624775583483,
      "grad_norm": 45.75,
      "learning_rate": 2.8610612407739878e-05,
      "loss": 2.5616,
      "mean_token_accuracy": 0.4920288294553757,
      "num_tokens": 1886803.0,
      "step": 4290
    },
    {
      "epoch": 1.2866546977857571,
      "grad_norm": 29.375,
      "learning_rate": 2.85607420706164e-05,
      "loss": 2.7951,
      "mean_token_accuracy": 0.4569530040025711,
      "num_tokens": 1891033.0,
      "step": 4300
    },
    {
      "epoch": 1.2896469180131658,
      "grad_norm": 24.25,
      "learning_rate": 2.8510871733492923e-05,
      "loss": 2.6682,
      "mean_token_accuracy": 0.48598000407218933,
      "num_tokens": 1896433.0,
      "step": 4310
    },
    {
      "epoch": 1.2926391382405744,
      "grad_norm": 19.0,
      "learning_rate": 2.846100139636944e-05,
      "loss": 2.9857,
      "mean_token_accuracy": 0.3988271489739418,
      "num_tokens": 1901571.0,
      "step": 4320
    },
    {
      "epoch": 1.2956313584679833,
      "grad_norm": 30.25,
      "learning_rate": 2.841113105924596e-05,
      "loss": 1.8598,
      "mean_token_accuracy": 0.585317638516426,
      "num_tokens": 1906215.0,
      "step": 4330
    },
    {
      "epoch": 1.298623578695392,
      "grad_norm": 39.5,
      "learning_rate": 2.8361260722122485e-05,
      "loss": 2.6775,
      "mean_token_accuracy": 0.47187069356441497,
      "num_tokens": 1909121.0,
      "step": 4340
    },
    {
      "epoch": 1.3016157989228008,
      "grad_norm": 30.875,
      "learning_rate": 2.8311390384999003e-05,
      "loss": 3.2452,
      "mean_token_accuracy": 0.3938755810260773,
      "num_tokens": 1913371.0,
      "step": 4350
    },
    {
      "epoch": 1.3046080191502094,
      "grad_norm": 46.0,
      "learning_rate": 2.8261520047875524e-05,
      "loss": 2.6618,
      "mean_token_accuracy": 0.4829663932323456,
      "num_tokens": 1917072.0,
      "step": 4360
    },
    {
      "epoch": 1.3076002393776183,
      "grad_norm": 42.25,
      "learning_rate": 2.8211649710752048e-05,
      "loss": 2.636,
      "mean_token_accuracy": 0.48219700157642365,
      "num_tokens": 1919981.0,
      "step": 4370
    },
    {
      "epoch": 1.310592459605027,
      "grad_norm": 19.875,
      "learning_rate": 2.8161779373628565e-05,
      "loss": 2.2672,
      "mean_token_accuracy": 0.5360256373882294,
      "num_tokens": 1923903.0,
      "step": 4380
    },
    {
      "epoch": 1.3135846798324358,
      "grad_norm": 27.375,
      "learning_rate": 2.811190903650509e-05,
      "loss": 2.9116,
      "mean_token_accuracy": 0.4232723444700241,
      "num_tokens": 1927175.0,
      "step": 4390
    },
    {
      "epoch": 1.3165769000598444,
      "grad_norm": 24.875,
      "learning_rate": 2.806203869938161e-05,
      "loss": 2.5052,
      "mean_token_accuracy": 0.49001633524894717,
      "num_tokens": 1930883.0,
      "step": 4400
    },
    {
      "epoch": 1.319569120287253,
      "grad_norm": 29.875,
      "learning_rate": 2.8012168362258128e-05,
      "loss": 2.9911,
      "mean_token_accuracy": 0.44069938361644745,
      "num_tokens": 1934961.0,
      "step": 4410
    },
    {
      "epoch": 1.322561340514662,
      "grad_norm": 31.625,
      "learning_rate": 2.7962298025134652e-05,
      "loss": 2.9474,
      "mean_token_accuracy": 0.43861653506755827,
      "num_tokens": 1940103.0,
      "step": 4420
    },
    {
      "epoch": 1.3255535607420705,
      "grad_norm": 32.5,
      "learning_rate": 2.7912427688011173e-05,
      "loss": 2.6026,
      "mean_token_accuracy": 0.48938514590263366,
      "num_tokens": 1944793.0,
      "step": 4430
    },
    {
      "epoch": 1.3285457809694794,
      "grad_norm": 35.0,
      "learning_rate": 2.786255735088769e-05,
      "loss": 2.2731,
      "mean_token_accuracy": 0.517932477593422,
      "num_tokens": 1948864.0,
      "step": 4440
    },
    {
      "epoch": 1.331538001196888,
      "grad_norm": 28.875,
      "learning_rate": 2.7812687013764218e-05,
      "loss": 2.3692,
      "mean_token_accuracy": 0.5257283002138138,
      "num_tokens": 1951872.0,
      "step": 4450
    },
    {
      "epoch": 1.334530221424297,
      "grad_norm": 21.375,
      "learning_rate": 2.7762816676640735e-05,
      "loss": 2.2699,
      "mean_token_accuracy": 0.5130042910575867,
      "num_tokens": 1956690.0,
      "step": 4460
    },
    {
      "epoch": 1.3375224416517055,
      "grad_norm": 41.0,
      "learning_rate": 2.7712946339517252e-05,
      "loss": 2.6093,
      "mean_token_accuracy": 0.46930437386035917,
      "num_tokens": 1960310.0,
      "step": 4470
    },
    {
      "epoch": 1.3405146618791144,
      "grad_norm": 34.0,
      "learning_rate": 2.766307600239378e-05,
      "loss": 3.125,
      "mean_token_accuracy": 0.3979645177721977,
      "num_tokens": 1964518.0,
      "step": 4480
    },
    {
      "epoch": 1.343506882106523,
      "grad_norm": 41.5,
      "learning_rate": 2.7613205665270297e-05,
      "loss": 2.4804,
      "mean_token_accuracy": 0.5112906128168107,
      "num_tokens": 1967926.0,
      "step": 4490
    },
    {
      "epoch": 1.3464991023339317,
      "grad_norm": 19.25,
      "learning_rate": 2.756333532814682e-05,
      "loss": 2.3555,
      "mean_token_accuracy": 0.5107169628143311,
      "num_tokens": 1971907.0,
      "step": 4500
    },
    {
      "epoch": 1.3494913225613405,
      "grad_norm": 35.75,
      "learning_rate": 2.7513464991023342e-05,
      "loss": 2.7525,
      "mean_token_accuracy": 0.46286738514900205,
      "num_tokens": 1976005.0,
      "step": 4510
    },
    {
      "epoch": 1.3524835427887494,
      "grad_norm": 24.25,
      "learning_rate": 2.746359465389986e-05,
      "loss": 3.0099,
      "mean_token_accuracy": 0.4340349406003952,
      "num_tokens": 1980363.0,
      "step": 4520
    },
    {
      "epoch": 1.355475763016158,
      "grad_norm": 30.125,
      "learning_rate": 2.7413724316776384e-05,
      "loss": 2.4456,
      "mean_token_accuracy": 0.495415124297142,
      "num_tokens": 1984602.0,
      "step": 4530
    },
    {
      "epoch": 1.3584679832435667,
      "grad_norm": 18.75,
      "learning_rate": 2.7363853979652905e-05,
      "loss": 2.4915,
      "mean_token_accuracy": 0.49877011477947236,
      "num_tokens": 1988696.0,
      "step": 4540
    },
    {
      "epoch": 1.3614602034709755,
      "grad_norm": 35.5,
      "learning_rate": 2.7313983642529422e-05,
      "loss": 2.5525,
      "mean_token_accuracy": 0.46752736568450926,
      "num_tokens": 1993603.0,
      "step": 4550
    },
    {
      "epoch": 1.3644524236983842,
      "grad_norm": 26.25,
      "learning_rate": 2.7264113305405946e-05,
      "loss": 2.8927,
      "mean_token_accuracy": 0.4371457099914551,
      "num_tokens": 1998477.0,
      "step": 4560
    },
    {
      "epoch": 1.367444643925793,
      "grad_norm": 53.25,
      "learning_rate": 2.7214242968282467e-05,
      "loss": 2.874,
      "mean_token_accuracy": 0.4590120866894722,
      "num_tokens": 2002477.0,
      "step": 4570
    },
    {
      "epoch": 1.3704368641532017,
      "grad_norm": 37.75,
      "learning_rate": 2.7164372631158985e-05,
      "loss": 2.5264,
      "mean_token_accuracy": 0.4863176017999649,
      "num_tokens": 2006648.0,
      "step": 4580
    },
    {
      "epoch": 1.3734290843806103,
      "grad_norm": 21.625,
      "learning_rate": 2.711450229403551e-05,
      "loss": 2.5302,
      "mean_token_accuracy": 0.49698867201805114,
      "num_tokens": 2012013.0,
      "step": 4590
    },
    {
      "epoch": 1.3764213046080191,
      "grad_norm": 48.5,
      "learning_rate": 2.706463195691203e-05,
      "loss": 2.7545,
      "mean_token_accuracy": 0.4686457276344299,
      "num_tokens": 2016062.0,
      "step": 4600
    },
    {
      "epoch": 1.379413524835428,
      "grad_norm": 27.625,
      "learning_rate": 2.7014761619788554e-05,
      "loss": 2.7296,
      "mean_token_accuracy": 0.48929981589317323,
      "num_tokens": 2020561.0,
      "step": 4610
    },
    {
      "epoch": 1.3824057450628366,
      "grad_norm": 94.0,
      "learning_rate": 2.696489128266507e-05,
      "loss": 2.2092,
      "mean_token_accuracy": 0.5368007481098175,
      "num_tokens": 2023737.0,
      "step": 4620
    },
    {
      "epoch": 1.3853979652902453,
      "grad_norm": 27.375,
      "learning_rate": 2.6915020945541592e-05,
      "loss": 2.3728,
      "mean_token_accuracy": 0.5155722945928574,
      "num_tokens": 2028489.0,
      "step": 4630
    },
    {
      "epoch": 1.3883901855176541,
      "grad_norm": 48.75,
      "learning_rate": 2.6865150608418116e-05,
      "loss": 2.6691,
      "mean_token_accuracy": 0.463055419921875,
      "num_tokens": 2033078.0,
      "step": 4640
    },
    {
      "epoch": 1.3913824057450628,
      "grad_norm": 50.75,
      "learning_rate": 2.6815280271294633e-05,
      "loss": 2.1985,
      "mean_token_accuracy": 0.5305429100990295,
      "num_tokens": 2038096.0,
      "step": 4650
    },
    {
      "epoch": 1.3943746259724716,
      "grad_norm": 58.25,
      "learning_rate": 2.6765409934171154e-05,
      "loss": 2.6201,
      "mean_token_accuracy": 0.4828624278306961,
      "num_tokens": 2041923.0,
      "step": 4660
    },
    {
      "epoch": 1.3973668461998803,
      "grad_norm": 23.625,
      "learning_rate": 2.671553959704768e-05,
      "loss": 2.5054,
      "mean_token_accuracy": 0.5020751714706421,
      "num_tokens": 2047556.0,
      "step": 4670
    },
    {
      "epoch": 1.400359066427289,
      "grad_norm": 22.625,
      "learning_rate": 2.6665669259924196e-05,
      "loss": 2.5951,
      "mean_token_accuracy": 0.5069888651371002,
      "num_tokens": 2052067.0,
      "step": 4680
    },
    {
      "epoch": 1.4033512866546978,
      "grad_norm": 29.125,
      "learning_rate": 2.6615798922800723e-05,
      "loss": 2.668,
      "mean_token_accuracy": 0.4788062393665314,
      "num_tokens": 2056698.0,
      "step": 4690
    },
    {
      "epoch": 1.4063435068821066,
      "grad_norm": 30.75,
      "learning_rate": 2.656592858567724e-05,
      "loss": 2.4209,
      "mean_token_accuracy": 0.5212773263454438,
      "num_tokens": 2060376.0,
      "step": 4700
    },
    {
      "epoch": 1.4093357271095153,
      "grad_norm": 41.75,
      "learning_rate": 2.651605824855376e-05,
      "loss": 2.0899,
      "mean_token_accuracy": 0.580978724360466,
      "num_tokens": 2063694.0,
      "step": 4710
    },
    {
      "epoch": 1.412327947336924,
      "grad_norm": 36.75,
      "learning_rate": 2.6466187911430286e-05,
      "loss": 1.9738,
      "mean_token_accuracy": 0.5679178059101104,
      "num_tokens": 2068346.0,
      "step": 4720
    },
    {
      "epoch": 1.4153201675643328,
      "grad_norm": 34.0,
      "learning_rate": 2.6416317574306803e-05,
      "loss": 2.7249,
      "mean_token_accuracy": 0.4764808028936386,
      "num_tokens": 2071494.0,
      "step": 4730
    },
    {
      "epoch": 1.4183123877917414,
      "grad_norm": 27.125,
      "learning_rate": 2.6366447237183324e-05,
      "loss": 2.6907,
      "mean_token_accuracy": 0.46054676324129107,
      "num_tokens": 2076057.0,
      "step": 4740
    },
    {
      "epoch": 1.4213046080191503,
      "grad_norm": 49.25,
      "learning_rate": 2.6316576900059848e-05,
      "loss": 2.7199,
      "mean_token_accuracy": 0.4358877420425415,
      "num_tokens": 2080486.0,
      "step": 4750
    },
    {
      "epoch": 1.424296828246559,
      "grad_norm": 43.75,
      "learning_rate": 2.6266706562936366e-05,
      "loss": 2.1997,
      "mean_token_accuracy": 0.5404698431491852,
      "num_tokens": 2083740.0,
      "step": 4760
    },
    {
      "epoch": 1.4272890484739678,
      "grad_norm": 23.125,
      "learning_rate": 2.6216836225812886e-05,
      "loss": 2.9674,
      "mean_token_accuracy": 0.45190505385398866,
      "num_tokens": 2088920.0,
      "step": 4770
    },
    {
      "epoch": 1.4302812687013764,
      "grad_norm": 21.0,
      "learning_rate": 2.616696588868941e-05,
      "loss": 1.822,
      "mean_token_accuracy": 0.6119337141513824,
      "num_tokens": 2092746.0,
      "step": 4780
    },
    {
      "epoch": 1.4332734889287853,
      "grad_norm": 40.5,
      "learning_rate": 2.6117095551565928e-05,
      "loss": 2.81,
      "mean_token_accuracy": 0.4584572911262512,
      "num_tokens": 2096346.0,
      "step": 4790
    },
    {
      "epoch": 1.436265709156194,
      "grad_norm": 28.125,
      "learning_rate": 2.6067225214442452e-05,
      "loss": 2.5106,
      "mean_token_accuracy": 0.48858973383903503,
      "num_tokens": 2100465.0,
      "step": 4800
    },
    {
      "epoch": 1.4392579293836025,
      "grad_norm": 32.75,
      "learning_rate": 2.6017354877318973e-05,
      "loss": 2.519,
      "mean_token_accuracy": 0.4892947614192963,
      "num_tokens": 2105797.0,
      "step": 4810
    },
    {
      "epoch": 1.4422501496110114,
      "grad_norm": 34.0,
      "learning_rate": 2.596748454019549e-05,
      "loss": 2.2762,
      "mean_token_accuracy": 0.5497841536998749,
      "num_tokens": 2109547.0,
      "step": 4820
    },
    {
      "epoch": 1.44524236983842,
      "grad_norm": 28.0,
      "learning_rate": 2.5917614203072015e-05,
      "loss": 2.7829,
      "mean_token_accuracy": 0.44075029492378237,
      "num_tokens": 2114477.0,
      "step": 4830
    },
    {
      "epoch": 1.4482345900658289,
      "grad_norm": 32.75,
      "learning_rate": 2.5867743865948535e-05,
      "loss": 2.6167,
      "mean_token_accuracy": 0.49549303352832796,
      "num_tokens": 2119250.0,
      "step": 4840
    },
    {
      "epoch": 1.4512268102932375,
      "grad_norm": 27.375,
      "learning_rate": 2.5817873528825053e-05,
      "loss": 2.197,
      "mean_token_accuracy": 0.5380541890859604,
      "num_tokens": 2123856.0,
      "step": 4850
    },
    {
      "epoch": 1.4542190305206464,
      "grad_norm": 26.875,
      "learning_rate": 2.5768003191701577e-05,
      "loss": 2.7312,
      "mean_token_accuracy": 0.49555808305740356,
      "num_tokens": 2128744.0,
      "step": 4860
    },
    {
      "epoch": 1.457211250748055,
      "grad_norm": 23.375,
      "learning_rate": 2.5718132854578098e-05,
      "loss": 2.5342,
      "mean_token_accuracy": 0.48873263597488403,
      "num_tokens": 2134383.0,
      "step": 4870
    },
    {
      "epoch": 1.4602034709754639,
      "grad_norm": 36.5,
      "learning_rate": 2.5668262517454615e-05,
      "loss": 2.3537,
      "mean_token_accuracy": 0.5590060204267502,
      "num_tokens": 2138661.0,
      "step": 4880
    },
    {
      "epoch": 1.4631956912028725,
      "grad_norm": 44.25,
      "learning_rate": 2.5618392180331143e-05,
      "loss": 2.9716,
      "mean_token_accuracy": 0.4578070640563965,
      "num_tokens": 2143943.0,
      "step": 4890
    },
    {
      "epoch": 1.4661879114302812,
      "grad_norm": 18.25,
      "learning_rate": 2.556852184320766e-05,
      "loss": 2.4542,
      "mean_token_accuracy": 0.49858992397785185,
      "num_tokens": 2150346.0,
      "step": 4900
    },
    {
      "epoch": 1.46918013165769,
      "grad_norm": 43.5,
      "learning_rate": 2.5518651506084184e-05,
      "loss": 2.0952,
      "mean_token_accuracy": 0.5627575278282165,
      "num_tokens": 2155714.0,
      "step": 4910
    },
    {
      "epoch": 1.4721723518850989,
      "grad_norm": 24.375,
      "learning_rate": 2.5468781168960705e-05,
      "loss": 2.5803,
      "mean_token_accuracy": 0.4823772877454758,
      "num_tokens": 2160085.0,
      "step": 4920
    },
    {
      "epoch": 1.4751645721125075,
      "grad_norm": 21.125,
      "learning_rate": 2.5418910831837222e-05,
      "loss": 2.1136,
      "mean_token_accuracy": 0.5596380650997161,
      "num_tokens": 2164085.0,
      "step": 4930
    },
    {
      "epoch": 1.4781567923399161,
      "grad_norm": 29.625,
      "learning_rate": 2.5369040494713747e-05,
      "loss": 2.6144,
      "mean_token_accuracy": 0.46260478496551516,
      "num_tokens": 2167762.0,
      "step": 4940
    },
    {
      "epoch": 1.481149012567325,
      "grad_norm": 22.0,
      "learning_rate": 2.5319170157590267e-05,
      "loss": 2.2213,
      "mean_token_accuracy": 0.5482838392257691,
      "num_tokens": 2171910.0,
      "step": 4950
    },
    {
      "epoch": 1.4841412327947336,
      "grad_norm": 42.75,
      "learning_rate": 2.5269299820466785e-05,
      "loss": 2.7502,
      "mean_token_accuracy": 0.46187237948179244,
      "num_tokens": 2176244.0,
      "step": 4960
    },
    {
      "epoch": 1.4871334530221425,
      "grad_norm": 30.25,
      "learning_rate": 2.521942948334331e-05,
      "loss": 2.7146,
      "mean_token_accuracy": 0.48688181936740876,
      "num_tokens": 2181027.0,
      "step": 4970
    },
    {
      "epoch": 1.4901256732495511,
      "grad_norm": 29.25,
      "learning_rate": 2.516955914621983e-05,
      "loss": 2.1965,
      "mean_token_accuracy": 0.5618243396282196,
      "num_tokens": 2185783.0,
      "step": 4980
    },
    {
      "epoch": 1.4931178934769598,
      "grad_norm": 38.0,
      "learning_rate": 2.5119688809096354e-05,
      "loss": 2.5971,
      "mean_token_accuracy": 0.496805140376091,
      "num_tokens": 2190346.0,
      "step": 4990
    },
    {
      "epoch": 1.4961101137043686,
      "grad_norm": 48.5,
      "learning_rate": 2.506981847197287e-05,
      "loss": 2.5287,
      "mean_token_accuracy": 0.49744627475738523,
      "num_tokens": 2194540.0,
      "step": 5000
    },
    {
      "epoch": 1.4991023339317775,
      "grad_norm": 29.125,
      "learning_rate": 2.5019948134849392e-05,
      "loss": 2.4346,
      "mean_token_accuracy": 0.5162340372800827,
      "num_tokens": 2198738.0,
      "step": 5010
    },
    {
      "epoch": 1.5020945541591861,
      "grad_norm": 28.875,
      "learning_rate": 2.4970077797725913e-05,
      "loss": 2.4267,
      "mean_token_accuracy": 0.5139214277267456,
      "num_tokens": 2203672.0,
      "step": 5020
    },
    {
      "epoch": 1.5050867743865948,
      "grad_norm": 37.0,
      "learning_rate": 2.4920207460602434e-05,
      "loss": 2.1187,
      "mean_token_accuracy": 0.5651913076639176,
      "num_tokens": 2208322.0,
      "step": 5030
    },
    {
      "epoch": 1.5080789946140036,
      "grad_norm": 39.0,
      "learning_rate": 2.4870337123478958e-05,
      "loss": 2.0766,
      "mean_token_accuracy": 0.5604483515024186,
      "num_tokens": 2212748.0,
      "step": 5040
    },
    {
      "epoch": 1.5110712148414125,
      "grad_norm": 39.25,
      "learning_rate": 2.4820466786355475e-05,
      "loss": 2.437,
      "mean_token_accuracy": 0.5023006498813629,
      "num_tokens": 2216429.0,
      "step": 5050
    },
    {
      "epoch": 1.5140634350688211,
      "grad_norm": 22.375,
      "learning_rate": 2.4770596449231996e-05,
      "loss": 2.5274,
      "mean_token_accuracy": 0.5053725391626358,
      "num_tokens": 2220799.0,
      "step": 5060
    },
    {
      "epoch": 1.5170556552962298,
      "grad_norm": 32.0,
      "learning_rate": 2.472072611210852e-05,
      "loss": 2.1191,
      "mean_token_accuracy": 0.5730376541614532,
      "num_tokens": 2225612.0,
      "step": 5070
    },
    {
      "epoch": 1.5200478755236384,
      "grad_norm": 38.5,
      "learning_rate": 2.467085577498504e-05,
      "loss": 2.8519,
      "mean_token_accuracy": 0.4593438282608986,
      "num_tokens": 2231107.0,
      "step": 5080
    },
    {
      "epoch": 1.5230400957510473,
      "grad_norm": 29.125,
      "learning_rate": 2.462098543786156e-05,
      "loss": 2.1636,
      "mean_token_accuracy": 0.5632421642541885,
      "num_tokens": 2235169.0,
      "step": 5090
    },
    {
      "epoch": 1.5260323159784561,
      "grad_norm": 27.25,
      "learning_rate": 2.4571115100738083e-05,
      "loss": 2.2962,
      "mean_token_accuracy": 0.5278664976358414,
      "num_tokens": 2239382.0,
      "step": 5100
    },
    {
      "epoch": 1.5290245362058648,
      "grad_norm": 35.5,
      "learning_rate": 2.4521244763614604e-05,
      "loss": 2.8431,
      "mean_token_accuracy": 0.4671991243958473,
      "num_tokens": 2244088.0,
      "step": 5110
    },
    {
      "epoch": 1.5320167564332734,
      "grad_norm": 40.25,
      "learning_rate": 2.4471374426491124e-05,
      "loss": 2.3703,
      "mean_token_accuracy": 0.5311822235584259,
      "num_tokens": 2249483.0,
      "step": 5120
    },
    {
      "epoch": 1.5350089766606823,
      "grad_norm": 18.875,
      "learning_rate": 2.4421504089367645e-05,
      "loss": 2.1462,
      "mean_token_accuracy": 0.5573323130607605,
      "num_tokens": 2253313.0,
      "step": 5130
    },
    {
      "epoch": 1.5380011968880911,
      "grad_norm": 48.0,
      "learning_rate": 2.4371633752244166e-05,
      "loss": 2.4867,
      "mean_token_accuracy": 0.5295968234539032,
      "num_tokens": 2258004.0,
      "step": 5140
    },
    {
      "epoch": 1.5409934171154998,
      "grad_norm": 36.25,
      "learning_rate": 2.4321763415120687e-05,
      "loss": 2.4837,
      "mean_token_accuracy": 0.5265371382236481,
      "num_tokens": 2262216.0,
      "step": 5150
    },
    {
      "epoch": 1.5439856373429084,
      "grad_norm": 45.25,
      "learning_rate": 2.427189307799721e-05,
      "loss": 2.9704,
      "mean_token_accuracy": 0.43561657071113585,
      "num_tokens": 2266858.0,
      "step": 5160
    },
    {
      "epoch": 1.546977857570317,
      "grad_norm": 40.25,
      "learning_rate": 2.4222022740873728e-05,
      "loss": 2.759,
      "mean_token_accuracy": 0.47565021812915803,
      "num_tokens": 2270709.0,
      "step": 5170
    },
    {
      "epoch": 1.5499700777977259,
      "grad_norm": 34.25,
      "learning_rate": 2.417215240375025e-05,
      "loss": 2.1739,
      "mean_token_accuracy": 0.5407566428184509,
      "num_tokens": 2274979.0,
      "step": 5180
    },
    {
      "epoch": 1.5529622980251347,
      "grad_norm": 28.625,
      "learning_rate": 2.4122282066626773e-05,
      "loss": 2.789,
      "mean_token_accuracy": 0.4501289531588554,
      "num_tokens": 2278606.0,
      "step": 5190
    },
    {
      "epoch": 1.5559545182525434,
      "grad_norm": 20.75,
      "learning_rate": 2.407241172950329e-05,
      "loss": 2.3088,
      "mean_token_accuracy": 0.5458819806575775,
      "num_tokens": 2282960.0,
      "step": 5200
    },
    {
      "epoch": 1.558946738479952,
      "grad_norm": 22.625,
      "learning_rate": 2.4022541392379815e-05,
      "loss": 2.5838,
      "mean_token_accuracy": 0.4927642345428467,
      "num_tokens": 2287682.0,
      "step": 5210
    },
    {
      "epoch": 1.5619389587073609,
      "grad_norm": 27.625,
      "learning_rate": 2.3972671055256336e-05,
      "loss": 2.838,
      "mean_token_accuracy": 0.4325797975063324,
      "num_tokens": 2294115.0,
      "step": 5220
    },
    {
      "epoch": 1.5649311789347697,
      "grad_norm": 33.25,
      "learning_rate": 2.3922800718132856e-05,
      "loss": 2.0421,
      "mean_token_accuracy": 0.5684062302112579,
      "num_tokens": 2298189.0,
      "step": 5230
    },
    {
      "epoch": 1.5679233991621784,
      "grad_norm": 34.0,
      "learning_rate": 2.3872930381009377e-05,
      "loss": 2.2966,
      "mean_token_accuracy": 0.5321805104613304,
      "num_tokens": 2301941.0,
      "step": 5240
    },
    {
      "epoch": 1.570915619389587,
      "grad_norm": 28.125,
      "learning_rate": 2.3823060043885898e-05,
      "loss": 2.032,
      "mean_token_accuracy": 0.5710376381874085,
      "num_tokens": 2306666.0,
      "step": 5250
    },
    {
      "epoch": 1.5739078396169957,
      "grad_norm": 27.125,
      "learning_rate": 2.377318970676242e-05,
      "loss": 2.2798,
      "mean_token_accuracy": 0.5459753274917603,
      "num_tokens": 2310667.0,
      "step": 5260
    },
    {
      "epoch": 1.5769000598444045,
      "grad_norm": 34.0,
      "learning_rate": 2.372331936963894e-05,
      "loss": 2.193,
      "mean_token_accuracy": 0.5457642912864685,
      "num_tokens": 2315430.0,
      "step": 5270
    },
    {
      "epoch": 1.5798922800718134,
      "grad_norm": 28.125,
      "learning_rate": 2.367344903251546e-05,
      "loss": 2.7639,
      "mean_token_accuracy": 0.4813458412885666,
      "num_tokens": 2319719.0,
      "step": 5280
    },
    {
      "epoch": 1.582884500299222,
      "grad_norm": 24.5,
      "learning_rate": 2.362357869539198e-05,
      "loss": 2.7784,
      "mean_token_accuracy": 0.4711227312684059,
      "num_tokens": 2323940.0,
      "step": 5290
    },
    {
      "epoch": 1.5858767205266306,
      "grad_norm": 73.0,
      "learning_rate": 2.3573708358268502e-05,
      "loss": 2.5437,
      "mean_token_accuracy": 0.4930324018001556,
      "num_tokens": 2327459.0,
      "step": 5300
    },
    {
      "epoch": 1.5888689407540395,
      "grad_norm": 35.25,
      "learning_rate": 2.3523838021145026e-05,
      "loss": 2.4414,
      "mean_token_accuracy": 0.5069448441267014,
      "num_tokens": 2333115.0,
      "step": 5310
    },
    {
      "epoch": 1.5918611609814484,
      "grad_norm": 35.25,
      "learning_rate": 2.3473967684021544e-05,
      "loss": 2.5232,
      "mean_token_accuracy": 0.49047706127166746,
      "num_tokens": 2337354.0,
      "step": 5320
    },
    {
      "epoch": 1.594853381208857,
      "grad_norm": 37.0,
      "learning_rate": 2.3424097346898068e-05,
      "loss": 2.7698,
      "mean_token_accuracy": 0.46212276816368103,
      "num_tokens": 2342548.0,
      "step": 5330
    },
    {
      "epoch": 1.5978456014362656,
      "grad_norm": 30.25,
      "learning_rate": 2.337422700977459e-05,
      "loss": 2.2649,
      "mean_token_accuracy": 0.538226792216301,
      "num_tokens": 2347913.0,
      "step": 5340
    },
    {
      "epoch": 1.6008378216636745,
      "grad_norm": 29.375,
      "learning_rate": 2.3324356672651106e-05,
      "loss": 3.0188,
      "mean_token_accuracy": 0.4409749746322632,
      "num_tokens": 2351777.0,
      "step": 5350
    },
    {
      "epoch": 1.6038300418910831,
      "grad_norm": 31.0,
      "learning_rate": 2.327448633552763e-05,
      "loss": 2.4968,
      "mean_token_accuracy": 0.4966483414173126,
      "num_tokens": 2356737.0,
      "step": 5360
    },
    {
      "epoch": 1.606822262118492,
      "grad_norm": 35.25,
      "learning_rate": 2.322461599840415e-05,
      "loss": 1.9533,
      "mean_token_accuracy": 0.5769125640392303,
      "num_tokens": 2359790.0,
      "step": 5370
    },
    {
      "epoch": 1.6098144823459006,
      "grad_norm": 31.875,
      "learning_rate": 2.3174745661280672e-05,
      "loss": 2.5364,
      "mean_token_accuracy": 0.5159052804112434,
      "num_tokens": 2364322.0,
      "step": 5380
    },
    {
      "epoch": 1.6128067025733093,
      "grad_norm": 22.25,
      "learning_rate": 2.3124875324157193e-05,
      "loss": 2.3025,
      "mean_token_accuracy": 0.5282308667898178,
      "num_tokens": 2368854.0,
      "step": 5390
    },
    {
      "epoch": 1.6157989228007181,
      "grad_norm": 24.5,
      "learning_rate": 2.3075004987033713e-05,
      "loss": 2.1343,
      "mean_token_accuracy": 0.5515360802412033,
      "num_tokens": 2372689.0,
      "step": 5400
    },
    {
      "epoch": 1.618791143028127,
      "grad_norm": 25.0,
      "learning_rate": 2.3025134649910234e-05,
      "loss": 2.7252,
      "mean_token_accuracy": 0.4657323956489563,
      "num_tokens": 2377410.0,
      "step": 5410
    },
    {
      "epoch": 1.6217833632555356,
      "grad_norm": 22.625,
      "learning_rate": 2.2975264312786758e-05,
      "loss": 2.341,
      "mean_token_accuracy": 0.5442433267831802,
      "num_tokens": 2382107.0,
      "step": 5420
    },
    {
      "epoch": 1.6247755834829443,
      "grad_norm": 31.5,
      "learning_rate": 2.2925393975663276e-05,
      "loss": 2.3103,
      "mean_token_accuracy": 0.5383136540651321,
      "num_tokens": 2386352.0,
      "step": 5430
    },
    {
      "epoch": 1.6277678037103531,
      "grad_norm": 26.5,
      "learning_rate": 2.2875523638539796e-05,
      "loss": 2.6328,
      "mean_token_accuracy": 0.48615836501121523,
      "num_tokens": 2391189.0,
      "step": 5440
    },
    {
      "epoch": 1.630760023937762,
      "grad_norm": 24.0,
      "learning_rate": 2.282565330141632e-05,
      "loss": 2.0362,
      "mean_token_accuracy": 0.572376748919487,
      "num_tokens": 2394615.0,
      "step": 5450
    },
    {
      "epoch": 1.6337522441651706,
      "grad_norm": 34.5,
      "learning_rate": 2.277578296429284e-05,
      "loss": 2.1694,
      "mean_token_accuracy": 0.55944344997406,
      "num_tokens": 2398123.0,
      "step": 5460
    },
    {
      "epoch": 1.6367444643925793,
      "grad_norm": 27.625,
      "learning_rate": 2.272591262716936e-05,
      "loss": 2.253,
      "mean_token_accuracy": 0.5339977979660034,
      "num_tokens": 2402810.0,
      "step": 5470
    },
    {
      "epoch": 1.639736684619988,
      "grad_norm": 14.8125,
      "learning_rate": 2.2676042290045883e-05,
      "loss": 1.9026,
      "mean_token_accuracy": 0.5830067694187164,
      "num_tokens": 2407040.0,
      "step": 5480
    },
    {
      "epoch": 1.6427289048473968,
      "grad_norm": 22.125,
      "learning_rate": 2.2626171952922404e-05,
      "loss": 2.4653,
      "mean_token_accuracy": 0.5064762234687805,
      "num_tokens": 2411166.0,
      "step": 5490
    },
    {
      "epoch": 1.6457211250748056,
      "grad_norm": 24.25,
      "learning_rate": 2.257630161579892e-05,
      "loss": 2.4491,
      "mean_token_accuracy": 0.5069942772388458,
      "num_tokens": 2414282.0,
      "step": 5500
    },
    {
      "epoch": 1.6487133453022143,
      "grad_norm": 39.25,
      "learning_rate": 2.2526431278675445e-05,
      "loss": 2.5615,
      "mean_token_accuracy": 0.5000170230865478,
      "num_tokens": 2419330.0,
      "step": 5510
    },
    {
      "epoch": 1.6517055655296229,
      "grad_norm": 26.375,
      "learning_rate": 2.2476560941551966e-05,
      "loss": 1.7894,
      "mean_token_accuracy": 0.608027520775795,
      "num_tokens": 2423670.0,
      "step": 5520
    },
    {
      "epoch": 1.6546977857570317,
      "grad_norm": 23.375,
      "learning_rate": 2.2426690604428487e-05,
      "loss": 1.8739,
      "mean_token_accuracy": 0.5978760778903961,
      "num_tokens": 2428023.0,
      "step": 5530
    },
    {
      "epoch": 1.6576900059844406,
      "grad_norm": 26.0,
      "learning_rate": 2.2376820267305008e-05,
      "loss": 2.1767,
      "mean_token_accuracy": 0.5449935585260391,
      "num_tokens": 2432049.0,
      "step": 5540
    },
    {
      "epoch": 1.6606822262118492,
      "grad_norm": 39.75,
      "learning_rate": 2.232694993018153e-05,
      "loss": 1.9426,
      "mean_token_accuracy": 0.6162794232368469,
      "num_tokens": 2435742.0,
      "step": 5550
    },
    {
      "epoch": 1.6636744464392579,
      "grad_norm": 32.5,
      "learning_rate": 2.227707959305805e-05,
      "loss": 1.9529,
      "mean_token_accuracy": 0.5872822850942612,
      "num_tokens": 2440036.0,
      "step": 5560
    },
    {
      "epoch": 1.6666666666666665,
      "grad_norm": 32.75,
      "learning_rate": 2.2227209255934574e-05,
      "loss": 2.5299,
      "mean_token_accuracy": 0.5048285767436027,
      "num_tokens": 2443798.0,
      "step": 5570
    },
    {
      "epoch": 1.6696588868940754,
      "grad_norm": 26.875,
      "learning_rate": 2.217733891881109e-05,
      "loss": 2.2806,
      "mean_token_accuracy": 0.5388662964105606,
      "num_tokens": 2448264.0,
      "step": 5580
    },
    {
      "epoch": 1.6726511071214842,
      "grad_norm": 43.25,
      "learning_rate": 2.2127468581687612e-05,
      "loss": 2.751,
      "mean_token_accuracy": 0.4652885258197784,
      "num_tokens": 2453502.0,
      "step": 5590
    },
    {
      "epoch": 1.6756433273488929,
      "grad_norm": 15.6875,
      "learning_rate": 2.2077598244564136e-05,
      "loss": 2.5737,
      "mean_token_accuracy": 0.5215324550867081,
      "num_tokens": 2457629.0,
      "step": 5600
    },
    {
      "epoch": 1.6786355475763015,
      "grad_norm": 35.25,
      "learning_rate": 2.2027727907440657e-05,
      "loss": 2.275,
      "mean_token_accuracy": 0.5492411971092224,
      "num_tokens": 2463140.0,
      "step": 5610
    },
    {
      "epoch": 1.6816277678037104,
      "grad_norm": 23.125,
      "learning_rate": 2.1977857570317174e-05,
      "loss": 2.227,
      "mean_token_accuracy": 0.5448118776082993,
      "num_tokens": 2467905.0,
      "step": 5620
    },
    {
      "epoch": 1.6846199880311192,
      "grad_norm": 50.75,
      "learning_rate": 2.19279872331937e-05,
      "loss": 2.0872,
      "mean_token_accuracy": 0.5600731402635575,
      "num_tokens": 2472835.0,
      "step": 5630
    },
    {
      "epoch": 1.6876122082585279,
      "grad_norm": 22.0,
      "learning_rate": 2.187811689607022e-05,
      "loss": 2.7493,
      "mean_token_accuracy": 0.4741476386785507,
      "num_tokens": 2477281.0,
      "step": 5640
    },
    {
      "epoch": 1.6906044284859365,
      "grad_norm": 29.25,
      "learning_rate": 2.182824655894674e-05,
      "loss": 2.1632,
      "mean_token_accuracy": 0.5686624765396118,
      "num_tokens": 2481606.0,
      "step": 5650
    },
    {
      "epoch": 1.6935966487133451,
      "grad_norm": 43.5,
      "learning_rate": 2.177837622182326e-05,
      "loss": 2.656,
      "mean_token_accuracy": 0.47470551133155825,
      "num_tokens": 2485039.0,
      "step": 5660
    },
    {
      "epoch": 1.696588868940754,
      "grad_norm": 23.125,
      "learning_rate": 2.172850588469978e-05,
      "loss": 1.8427,
      "mean_token_accuracy": 0.625142702460289,
      "num_tokens": 2490589.0,
      "step": 5670
    },
    {
      "epoch": 1.6995810891681629,
      "grad_norm": 25.0,
      "learning_rate": 2.1678635547576302e-05,
      "loss": 2.5159,
      "mean_token_accuracy": 0.5088352680206298,
      "num_tokens": 2494052.0,
      "step": 5680
    },
    {
      "epoch": 1.7025733093955715,
      "grad_norm": 26.5,
      "learning_rate": 2.1628765210452823e-05,
      "loss": 2.5968,
      "mean_token_accuracy": 0.47904575169086455,
      "num_tokens": 2498952.0,
      "step": 5690
    },
    {
      "epoch": 1.7055655296229801,
      "grad_norm": 46.75,
      "learning_rate": 2.1578894873329344e-05,
      "loss": 3.29,
      "mean_token_accuracy": 0.4176428884267807,
      "num_tokens": 2503107.0,
      "step": 5700
    },
    {
      "epoch": 1.708557749850389,
      "grad_norm": 22.875,
      "learning_rate": 2.1529024536205865e-05,
      "loss": 1.9785,
      "mean_token_accuracy": 0.5829225838184356,
      "num_tokens": 2507193.0,
      "step": 5710
    },
    {
      "epoch": 1.7115499700777979,
      "grad_norm": 33.5,
      "learning_rate": 2.147915419908239e-05,
      "loss": 2.8111,
      "mean_token_accuracy": 0.4594015315175056,
      "num_tokens": 2511145.0,
      "step": 5720
    },
    {
      "epoch": 1.7145421903052065,
      "grad_norm": 39.5,
      "learning_rate": 2.1429283861958906e-05,
      "loss": 2.2739,
      "mean_token_accuracy": 0.5496653497219086,
      "num_tokens": 2516512.0,
      "step": 5730
    },
    {
      "epoch": 1.7175344105326151,
      "grad_norm": 23.625,
      "learning_rate": 2.137941352483543e-05,
      "loss": 2.4269,
      "mean_token_accuracy": 0.4974661499261856,
      "num_tokens": 2521653.0,
      "step": 5740
    },
    {
      "epoch": 1.720526630760024,
      "grad_norm": 25.125,
      "learning_rate": 2.132954318771195e-05,
      "loss": 2.8225,
      "mean_token_accuracy": 0.4561440125107765,
      "num_tokens": 2525305.0,
      "step": 5750
    },
    {
      "epoch": 1.7235188509874326,
      "grad_norm": 49.25,
      "learning_rate": 2.1279672850588472e-05,
      "loss": 2.308,
      "mean_token_accuracy": 0.5274252116680145,
      "num_tokens": 2528945.0,
      "step": 5760
    },
    {
      "epoch": 1.7265110712148415,
      "grad_norm": 21.625,
      "learning_rate": 2.1229802513464993e-05,
      "loss": 2.226,
      "mean_token_accuracy": 0.5421534836292267,
      "num_tokens": 2533763.0,
      "step": 5770
    },
    {
      "epoch": 1.7295032914422501,
      "grad_norm": 36.25,
      "learning_rate": 2.1179932176341514e-05,
      "loss": 2.2728,
      "mean_token_accuracy": 0.5372824460268021,
      "num_tokens": 2536930.0,
      "step": 5780
    },
    {
      "epoch": 1.7324955116696588,
      "grad_norm": 24.625,
      "learning_rate": 2.1130061839218034e-05,
      "loss": 2.1263,
      "mean_token_accuracy": 0.5437184333801269,
      "num_tokens": 2542163.0,
      "step": 5790
    },
    {
      "epoch": 1.7354877318970676,
      "grad_norm": 31.75,
      "learning_rate": 2.1080191502094555e-05,
      "loss": 2.5627,
      "mean_token_accuracy": 0.4753000020980835,
      "num_tokens": 2545587.0,
      "step": 5800
    },
    {
      "epoch": 1.7384799521244765,
      "grad_norm": 43.0,
      "learning_rate": 2.1030321164971076e-05,
      "loss": 2.1084,
      "mean_token_accuracy": 0.5742941737174988,
      "num_tokens": 2550025.0,
      "step": 5810
    },
    {
      "epoch": 1.7414721723518851,
      "grad_norm": 31.25,
      "learning_rate": 2.0980450827847597e-05,
      "loss": 2.0954,
      "mean_token_accuracy": 0.5463778793811798,
      "num_tokens": 2554373.0,
      "step": 5820
    },
    {
      "epoch": 1.7444643925792938,
      "grad_norm": 35.25,
      "learning_rate": 2.093058049072412e-05,
      "loss": 2.197,
      "mean_token_accuracy": 0.5357052236795425,
      "num_tokens": 2558090.0,
      "step": 5830
    },
    {
      "epoch": 1.7474566128067026,
      "grad_norm": 31.125,
      "learning_rate": 2.088071015360064e-05,
      "loss": 2.4045,
      "mean_token_accuracy": 0.5454428046941757,
      "num_tokens": 2562281.0,
      "step": 5840
    },
    {
      "epoch": 1.7504488330341115,
      "grad_norm": 17.0,
      "learning_rate": 2.083083981647716e-05,
      "loss": 2.6007,
      "mean_token_accuracy": 0.5035086557269096,
      "num_tokens": 2567346.0,
      "step": 5850
    },
    {
      "epoch": 1.75344105326152,
      "grad_norm": 18.5,
      "learning_rate": 2.0780969479353683e-05,
      "loss": 2.5044,
      "mean_token_accuracy": 0.5108116358518601,
      "num_tokens": 2571166.0,
      "step": 5860
    },
    {
      "epoch": 1.7564332734889287,
      "grad_norm": 24.125,
      "learning_rate": 2.0731099142230204e-05,
      "loss": 2.4193,
      "mean_token_accuracy": 0.5259674608707428,
      "num_tokens": 2577029.0,
      "step": 5870
    },
    {
      "epoch": 1.7594254937163374,
      "grad_norm": 48.5,
      "learning_rate": 2.068122880510672e-05,
      "loss": 2.5751,
      "mean_token_accuracy": 0.5132578700780869,
      "num_tokens": 2581099.0,
      "step": 5880
    },
    {
      "epoch": 1.7624177139437462,
      "grad_norm": 12.625,
      "learning_rate": 2.0631358467983246e-05,
      "loss": 2.575,
      "mean_token_accuracy": 0.528558287024498,
      "num_tokens": 2585381.0,
      "step": 5890
    },
    {
      "epoch": 1.765409934171155,
      "grad_norm": 63.75,
      "learning_rate": 2.0581488130859767e-05,
      "loss": 2.1645,
      "mean_token_accuracy": 0.558303925395012,
      "num_tokens": 2588571.0,
      "step": 5900
    },
    {
      "epoch": 1.7684021543985637,
      "grad_norm": 24.125,
      "learning_rate": 2.0531617793736287e-05,
      "loss": 2.2552,
      "mean_token_accuracy": 0.5411330461502075,
      "num_tokens": 2593493.0,
      "step": 5910
    },
    {
      "epoch": 1.7713943746259724,
      "grad_norm": 41.0,
      "learning_rate": 2.0481747456612808e-05,
      "loss": 2.4276,
      "mean_token_accuracy": 0.535361161828041,
      "num_tokens": 2597065.0,
      "step": 5920
    },
    {
      "epoch": 1.7743865948533812,
      "grad_norm": 36.5,
      "learning_rate": 2.043187711948933e-05,
      "loss": 2.292,
      "mean_token_accuracy": 0.534418734908104,
      "num_tokens": 2602326.0,
      "step": 5930
    },
    {
      "epoch": 1.77737881508079,
      "grad_norm": 37.75,
      "learning_rate": 2.038200678236585e-05,
      "loss": 2.5938,
      "mean_token_accuracy": 0.48518282175064087,
      "num_tokens": 2605919.0,
      "step": 5940
    },
    {
      "epoch": 1.7803710353081987,
      "grad_norm": 44.5,
      "learning_rate": 2.033213644524237e-05,
      "loss": 2.2369,
      "mean_token_accuracy": 0.5367230206727982,
      "num_tokens": 2609223.0,
      "step": 5950
    },
    {
      "epoch": 1.7833632555356074,
      "grad_norm": 25.25,
      "learning_rate": 2.028226610811889e-05,
      "loss": 2.0652,
      "mean_token_accuracy": 0.5805120468139648,
      "num_tokens": 2613758.0,
      "step": 5960
    },
    {
      "epoch": 1.786355475763016,
      "grad_norm": 23.25,
      "learning_rate": 2.0232395770995412e-05,
      "loss": 2.5217,
      "mean_token_accuracy": 0.48961347341537476,
      "num_tokens": 2618111.0,
      "step": 5970
    },
    {
      "epoch": 1.7893476959904249,
      "grad_norm": 11.5,
      "learning_rate": 2.0182525433871936e-05,
      "loss": 1.8521,
      "mean_token_accuracy": 0.6123942375183106,
      "num_tokens": 2623243.0,
      "step": 5980
    },
    {
      "epoch": 1.7923399162178337,
      "grad_norm": 32.25,
      "learning_rate": 2.0132655096748454e-05,
      "loss": 2.1077,
      "mean_token_accuracy": 0.5779639840126037,
      "num_tokens": 2627941.0,
      "step": 5990
    },
    {
      "epoch": 1.7953321364452424,
      "grad_norm": 40.75,
      "learning_rate": 2.0082784759624974e-05,
      "loss": 2.5294,
      "mean_token_accuracy": 0.5101372256875039,
      "num_tokens": 2632788.0,
      "step": 6000
    },
    {
      "epoch": 1.798324356672651,
      "grad_norm": 24.375,
      "learning_rate": 2.00329144225015e-05,
      "loss": 2.2084,
      "mean_token_accuracy": 0.56387078166008,
      "num_tokens": 2638259.0,
      "step": 6010
    },
    {
      "epoch": 1.8013165769000599,
      "grad_norm": 44.0,
      "learning_rate": 1.998304408537802e-05,
      "loss": 2.9581,
      "mean_token_accuracy": 0.42872937619686124,
      "num_tokens": 2642095.0,
      "step": 6020
    },
    {
      "epoch": 1.8043087971274687,
      "grad_norm": 25.5,
      "learning_rate": 1.9933173748254537e-05,
      "loss": 2.7151,
      "mean_token_accuracy": 0.4665300458669662,
      "num_tokens": 2647272.0,
      "step": 6030
    },
    {
      "epoch": 1.8073010173548774,
      "grad_norm": 30.375,
      "learning_rate": 1.988330341113106e-05,
      "loss": 2.6866,
      "mean_token_accuracy": 0.477365055680275,
      "num_tokens": 2650970.0,
      "step": 6040
    },
    {
      "epoch": 1.810293237582286,
      "grad_norm": 20.25,
      "learning_rate": 1.9833433074007582e-05,
      "loss": 2.5152,
      "mean_token_accuracy": 0.5005413472652436,
      "num_tokens": 2655297.0,
      "step": 6050
    },
    {
      "epoch": 1.8132854578096946,
      "grad_norm": 25.375,
      "learning_rate": 1.9783562736884103e-05,
      "loss": 2.0359,
      "mean_token_accuracy": 0.5744021505117416,
      "num_tokens": 2659789.0,
      "step": 6060
    },
    {
      "epoch": 1.8162776780371035,
      "grad_norm": 34.25,
      "learning_rate": 1.9733692399760623e-05,
      "loss": 2.5769,
      "mean_token_accuracy": 0.49441861510276797,
      "num_tokens": 2664682.0,
      "step": 6070
    },
    {
      "epoch": 1.8192698982645124,
      "grad_norm": 31.0,
      "learning_rate": 1.9683822062637144e-05,
      "loss": 2.8461,
      "mean_token_accuracy": 0.4470598191022873,
      "num_tokens": 2668051.0,
      "step": 6080
    },
    {
      "epoch": 1.822262118491921,
      "grad_norm": 33.75,
      "learning_rate": 1.9633951725513665e-05,
      "loss": 2.2639,
      "mean_token_accuracy": 0.5656882345676422,
      "num_tokens": 2673035.0,
      "step": 6090
    },
    {
      "epoch": 1.8252543387193296,
      "grad_norm": 33.5,
      "learning_rate": 1.9584081388390186e-05,
      "loss": 2.376,
      "mean_token_accuracy": 0.5445163309574127,
      "num_tokens": 2676665.0,
      "step": 6100
    },
    {
      "epoch": 1.8282465589467385,
      "grad_norm": 28.25,
      "learning_rate": 1.9534211051266707e-05,
      "loss": 2.2772,
      "mean_token_accuracy": 0.5421164140105248,
      "num_tokens": 2680076.0,
      "step": 6110
    },
    {
      "epoch": 1.8312387791741473,
      "grad_norm": 20.625,
      "learning_rate": 1.9484340714143227e-05,
      "loss": 2.0009,
      "mean_token_accuracy": 0.5641443014144898,
      "num_tokens": 2685756.0,
      "step": 6120
    },
    {
      "epoch": 1.834230999401556,
      "grad_norm": 13.0625,
      "learning_rate": 1.943447037701975e-05,
      "loss": 2.0111,
      "mean_token_accuracy": 0.6003489851951599,
      "num_tokens": 2690244.0,
      "step": 6130
    },
    {
      "epoch": 1.8372232196289646,
      "grad_norm": 20.25,
      "learning_rate": 1.938460003989627e-05,
      "loss": 2.4381,
      "mean_token_accuracy": 0.5133287519216537,
      "num_tokens": 2694430.0,
      "step": 6140
    },
    {
      "epoch": 1.8402154398563735,
      "grad_norm": 17.125,
      "learning_rate": 1.933472970277279e-05,
      "loss": 2.2444,
      "mean_token_accuracy": 0.5365898728370666,
      "num_tokens": 2700217.0,
      "step": 6150
    },
    {
      "epoch": 1.8432076600837821,
      "grad_norm": 31.0,
      "learning_rate": 1.9284859365649314e-05,
      "loss": 2.0314,
      "mean_token_accuracy": 0.5679359257221221,
      "num_tokens": 2702801.0,
      "step": 6160
    },
    {
      "epoch": 1.846199880311191,
      "grad_norm": 42.0,
      "learning_rate": 1.9234989028525835e-05,
      "loss": 2.7164,
      "mean_token_accuracy": 0.4744591027498245,
      "num_tokens": 2707886.0,
      "step": 6170
    },
    {
      "epoch": 1.8491921005385996,
      "grad_norm": 38.5,
      "learning_rate": 1.9185118691402356e-05,
      "loss": 2.7306,
      "mean_token_accuracy": 0.4657700195908546,
      "num_tokens": 2712105.0,
      "step": 6180
    },
    {
      "epoch": 1.8521843207660083,
      "grad_norm": 37.25,
      "learning_rate": 1.9135248354278876e-05,
      "loss": 2.7078,
      "mean_token_accuracy": 0.46402439177036287,
      "num_tokens": 2716450.0,
      "step": 6190
    },
    {
      "epoch": 1.8551765409934171,
      "grad_norm": 32.75,
      "learning_rate": 1.9085378017155397e-05,
      "loss": 2.1767,
      "mean_token_accuracy": 0.5529489815235138,
      "num_tokens": 2720919.0,
      "step": 6200
    },
    {
      "epoch": 1.858168761220826,
      "grad_norm": 24.875,
      "learning_rate": 1.9035507680031918e-05,
      "loss": 2.7361,
      "mean_token_accuracy": 0.4729373618960381,
      "num_tokens": 2724921.0,
      "step": 6210
    },
    {
      "epoch": 1.8611609814482346,
      "grad_norm": 28.0,
      "learning_rate": 1.898563734290844e-05,
      "loss": 2.3484,
      "mean_token_accuracy": 0.5261541813611984,
      "num_tokens": 2728812.0,
      "step": 6220
    },
    {
      "epoch": 1.8641532016756432,
      "grad_norm": 32.5,
      "learning_rate": 1.893576700578496e-05,
      "loss": 2.6856,
      "mean_token_accuracy": 0.48226538598537444,
      "num_tokens": 2734365.0,
      "step": 6230
    },
    {
      "epoch": 1.867145421903052,
      "grad_norm": 37.25,
      "learning_rate": 1.888589666866148e-05,
      "loss": 1.9228,
      "mean_token_accuracy": 0.6168141156435013,
      "num_tokens": 2738466.0,
      "step": 6240
    },
    {
      "epoch": 1.870137642130461,
      "grad_norm": 20.625,
      "learning_rate": 1.8836026331538e-05,
      "loss": 1.9347,
      "mean_token_accuracy": 0.5970963388681412,
      "num_tokens": 2742084.0,
      "step": 6250
    },
    {
      "epoch": 1.8731298623578696,
      "grad_norm": 30.875,
      "learning_rate": 1.8786155994414522e-05,
      "loss": 2.3609,
      "mean_token_accuracy": 0.5621721386909485,
      "num_tokens": 2747004.0,
      "step": 6260
    },
    {
      "epoch": 1.8761220825852782,
      "grad_norm": 24.75,
      "learning_rate": 1.8736285657291046e-05,
      "loss": 2.7582,
      "mean_token_accuracy": 0.4766868263483047,
      "num_tokens": 2751277.0,
      "step": 6270
    },
    {
      "epoch": 1.8791143028126869,
      "grad_norm": 31.125,
      "learning_rate": 1.8686415320167567e-05,
      "loss": 2.1709,
      "mean_token_accuracy": 0.5493841350078583,
      "num_tokens": 2754736.0,
      "step": 6280
    },
    {
      "epoch": 1.8821065230400957,
      "grad_norm": 29.5,
      "learning_rate": 1.8636544983044084e-05,
      "loss": 2.4101,
      "mean_token_accuracy": 0.5184233725070954,
      "num_tokens": 2759579.0,
      "step": 6290
    },
    {
      "epoch": 1.8850987432675046,
      "grad_norm": 43.25,
      "learning_rate": 1.858667464592061e-05,
      "loss": 2.7502,
      "mean_token_accuracy": 0.4661740347743034,
      "num_tokens": 2763344.0,
      "step": 6300
    },
    {
      "epoch": 1.8880909634949132,
      "grad_norm": 39.0,
      "learning_rate": 1.853680430879713e-05,
      "loss": 1.918,
      "mean_token_accuracy": 0.5811412990093231,
      "num_tokens": 2766322.0,
      "step": 6310
    },
    {
      "epoch": 1.8910831837223219,
      "grad_norm": 27.375,
      "learning_rate": 1.848693397167365e-05,
      "loss": 1.6647,
      "mean_token_accuracy": 0.6429639548063278,
      "num_tokens": 2770347.0,
      "step": 6320
    },
    {
      "epoch": 1.8940754039497307,
      "grad_norm": 34.0,
      "learning_rate": 1.843706363455017e-05,
      "loss": 2.7445,
      "mean_token_accuracy": 0.4729799941182137,
      "num_tokens": 2775124.0,
      "step": 6330
    },
    {
      "epoch": 1.8970676241771396,
      "grad_norm": 29.875,
      "learning_rate": 1.838719329742669e-05,
      "loss": 2.3141,
      "mean_token_accuracy": 0.5422122359275818,
      "num_tokens": 2779961.0,
      "step": 6340
    },
    {
      "epoch": 1.9000598444045482,
      "grad_norm": 23.5,
      "learning_rate": 1.8337322960303212e-05,
      "loss": 2.2344,
      "mean_token_accuracy": 0.5411304891109466,
      "num_tokens": 2783692.0,
      "step": 6350
    },
    {
      "epoch": 1.9030520646319569,
      "grad_norm": 30.625,
      "learning_rate": 1.8287452623179737e-05,
      "loss": 1.9572,
      "mean_token_accuracy": 0.5806999713182449,
      "num_tokens": 2788229.0,
      "step": 6360
    },
    {
      "epoch": 1.9060442848593655,
      "grad_norm": 45.75,
      "learning_rate": 1.8237582286056254e-05,
      "loss": 2.02,
      "mean_token_accuracy": 0.596290436387062,
      "num_tokens": 2792558.0,
      "step": 6370
    },
    {
      "epoch": 1.9090365050867744,
      "grad_norm": 40.0,
      "learning_rate": 1.8187711948932775e-05,
      "loss": 1.9781,
      "mean_token_accuracy": 0.5878248095512391,
      "num_tokens": 2797147.0,
      "step": 6380
    },
    {
      "epoch": 1.9120287253141832,
      "grad_norm": 20.125,
      "learning_rate": 1.81378416118093e-05,
      "loss": 2.3107,
      "mean_token_accuracy": 0.5291865646839142,
      "num_tokens": 2801640.0,
      "step": 6390
    },
    {
      "epoch": 1.9150209455415919,
      "grad_norm": 28.625,
      "learning_rate": 1.8087971274685816e-05,
      "loss": 1.8296,
      "mean_token_accuracy": 0.6036499828100205,
      "num_tokens": 2806291.0,
      "step": 6400
    },
    {
      "epoch": 1.9180131657690005,
      "grad_norm": 25.125,
      "learning_rate": 1.8038100937562337e-05,
      "loss": 1.9843,
      "mean_token_accuracy": 0.5820193707942962,
      "num_tokens": 2810023.0,
      "step": 6410
    },
    {
      "epoch": 1.9210053859964094,
      "grad_norm": 27.625,
      "learning_rate": 1.798823060043886e-05,
      "loss": 2.4543,
      "mean_token_accuracy": 0.5143000841140747,
      "num_tokens": 2814985.0,
      "step": 6420
    },
    {
      "epoch": 1.9239976062238182,
      "grad_norm": 30.0,
      "learning_rate": 1.7938360263315382e-05,
      "loss": 2.068,
      "mean_token_accuracy": 0.5631296753883361,
      "num_tokens": 2820578.0,
      "step": 6430
    },
    {
      "epoch": 1.9269898264512269,
      "grad_norm": 31.125,
      "learning_rate": 1.78884899261919e-05,
      "loss": 2.5761,
      "mean_token_accuracy": 0.48471910059452056,
      "num_tokens": 2824205.0,
      "step": 6440
    },
    {
      "epoch": 1.9299820466786355,
      "grad_norm": 26.5,
      "learning_rate": 1.7838619589068424e-05,
      "loss": 2.3299,
      "mean_token_accuracy": 0.5316969901323318,
      "num_tokens": 2828031.0,
      "step": 6450
    },
    {
      "epoch": 1.9329742669060441,
      "grad_norm": 43.75,
      "learning_rate": 1.7788749251944945e-05,
      "loss": 2.7208,
      "mean_token_accuracy": 0.49289258420467374,
      "num_tokens": 2831445.0,
      "step": 6460
    },
    {
      "epoch": 1.935966487133453,
      "grad_norm": 19.375,
      "learning_rate": 1.7738878914821465e-05,
      "loss": 1.7553,
      "mean_token_accuracy": 0.6282897502183914,
      "num_tokens": 2835771.0,
      "step": 6470
    },
    {
      "epoch": 1.9389587073608618,
      "grad_norm": 20.875,
      "learning_rate": 1.7689008577697986e-05,
      "loss": 2.1944,
      "mean_token_accuracy": 0.5652035892009735,
      "num_tokens": 2840434.0,
      "step": 6480
    },
    {
      "epoch": 1.9419509275882705,
      "grad_norm": 22.25,
      "learning_rate": 1.7639138240574507e-05,
      "loss": 2.1051,
      "mean_token_accuracy": 0.5805429667234421,
      "num_tokens": 2843900.0,
      "step": 6490
    },
    {
      "epoch": 1.9449431478156791,
      "grad_norm": 27.25,
      "learning_rate": 1.7589267903451028e-05,
      "loss": 2.7318,
      "mean_token_accuracy": 0.5218796044588089,
      "num_tokens": 2849886.0,
      "step": 6500
    },
    {
      "epoch": 1.947935368043088,
      "grad_norm": 33.25,
      "learning_rate": 1.7539397566327552e-05,
      "loss": 2.5077,
      "mean_token_accuracy": 0.5040024489164352,
      "num_tokens": 2852596.0,
      "step": 6510
    },
    {
      "epoch": 1.9509275882704968,
      "grad_norm": 35.5,
      "learning_rate": 1.748952722920407e-05,
      "loss": 1.7998,
      "mean_token_accuracy": 0.6108117759227752,
      "num_tokens": 2856341.0,
      "step": 6520
    },
    {
      "epoch": 1.9539198084979055,
      "grad_norm": 23.75,
      "learning_rate": 1.743965689208059e-05,
      "loss": 2.3724,
      "mean_token_accuracy": 0.5622206628322601,
      "num_tokens": 2859928.0,
      "step": 6530
    },
    {
      "epoch": 1.9569120287253141,
      "grad_norm": 26.375,
      "learning_rate": 1.7389786554957114e-05,
      "loss": 2.2704,
      "mean_token_accuracy": 0.549874222278595,
      "num_tokens": 2865567.0,
      "step": 6540
    },
    {
      "epoch": 1.959904248952723,
      "grad_norm": 46.75,
      "learning_rate": 1.733991621783363e-05,
      "loss": 2.0852,
      "mean_token_accuracy": 0.5803373664617538,
      "num_tokens": 2869227.0,
      "step": 6550
    },
    {
      "epoch": 1.9628964691801316,
      "grad_norm": 36.5,
      "learning_rate": 1.7290045880710152e-05,
      "loss": 2.8152,
      "mean_token_accuracy": 0.46767871379852294,
      "num_tokens": 2874440.0,
      "step": 6560
    },
    {
      "epoch": 1.9658886894075405,
      "grad_norm": 13.25,
      "learning_rate": 1.7240175543586677e-05,
      "loss": 1.9414,
      "mean_token_accuracy": 0.5902773320674897,
      "num_tokens": 2878799.0,
      "step": 6570
    },
    {
      "epoch": 1.968880909634949,
      "grad_norm": 38.0,
      "learning_rate": 1.7190305206463197e-05,
      "loss": 2.8453,
      "mean_token_accuracy": 0.4565995991230011,
      "num_tokens": 2882617.0,
      "step": 6580
    },
    {
      "epoch": 1.9718731298623577,
      "grad_norm": 27.75,
      "learning_rate": 1.7140434869339718e-05,
      "loss": 2.3815,
      "mean_token_accuracy": 0.5338667601346969,
      "num_tokens": 2887111.0,
      "step": 6590
    },
    {
      "epoch": 1.9748653500897666,
      "grad_norm": 39.75,
      "learning_rate": 1.709056453221624e-05,
      "loss": 2.4473,
      "mean_token_accuracy": 0.5243928819894791,
      "num_tokens": 2891051.0,
      "step": 6600
    },
    {
      "epoch": 1.9778575703171755,
      "grad_norm": 24.375,
      "learning_rate": 1.704069419509276e-05,
      "loss": 2.0229,
      "mean_token_accuracy": 0.5819611266255379,
      "num_tokens": 2895900.0,
      "step": 6610
    },
    {
      "epoch": 1.980849790544584,
      "grad_norm": 45.25,
      "learning_rate": 1.699082385796928e-05,
      "loss": 1.8578,
      "mean_token_accuracy": 0.6056030809879303,
      "num_tokens": 2901008.0,
      "step": 6620
    },
    {
      "epoch": 1.9838420107719927,
      "grad_norm": 29.625,
      "learning_rate": 1.69409535208458e-05,
      "loss": 2.2429,
      "mean_token_accuracy": 0.5784931719303131,
      "num_tokens": 2905651.0,
      "step": 6630
    },
    {
      "epoch": 1.9868342309994016,
      "grad_norm": 29.875,
      "learning_rate": 1.6891083183722322e-05,
      "loss": 2.0341,
      "mean_token_accuracy": 0.5632521450519562,
      "num_tokens": 2909307.0,
      "step": 6640
    },
    {
      "epoch": 1.9898264512268105,
      "grad_norm": 37.0,
      "learning_rate": 1.6841212846598843e-05,
      "loss": 2.0049,
      "mean_token_accuracy": 0.5811235994100571,
      "num_tokens": 2913081.0,
      "step": 6650
    },
    {
      "epoch": 1.992818671454219,
      "grad_norm": 28.5,
      "learning_rate": 1.6791342509475367e-05,
      "loss": 2.6649,
      "mean_token_accuracy": 0.4999321013689041,
      "num_tokens": 2918402.0,
      "step": 6660
    },
    {
      "epoch": 1.9958108916816277,
      "grad_norm": 29.0,
      "learning_rate": 1.6741472172351885e-05,
      "loss": 2.0443,
      "mean_token_accuracy": 0.5667910605669022,
      "num_tokens": 2923258.0,
      "step": 6670
    },
    {
      "epoch": 1.9988031119090364,
      "grad_norm": 26.875,
      "learning_rate": 1.6691601835228405e-05,
      "loss": 1.8138,
      "mean_token_accuracy": 0.629214882850647,
      "num_tokens": 2926579.0,
      "step": 6680
    },
    {
      "epoch": 2.0,
      "eval_loss": 2.242931842803955,
      "eval_mean_token_accuracy": 0.5522286700752547,
      "eval_num_tokens": 2928488.0,
      "eval_runtime": 28.1437,
      "eval_samples_per_second": 14.852,
      "eval_steps_per_second": 1.883,
      "step": 6684
    },
    {
      "epoch": 2.0017953321364454,
      "grad_norm": 31.125,
      "learning_rate": 1.664173149810493e-05,
      "loss": 1.2534,
      "mean_token_accuracy": 0.7216873943805695,
      "num_tokens": 2930316.0,
      "step": 6690
    },
    {
      "epoch": 2.004787552363854,
      "grad_norm": 19.875,
      "learning_rate": 1.6591861160981447e-05,
      "loss": 2.1657,
      "mean_token_accuracy": 0.5659398257732391,
      "num_tokens": 2935350.0,
      "step": 6700
    },
    {
      "epoch": 2.0077797725912627,
      "grad_norm": 25.125,
      "learning_rate": 1.654199082385797e-05,
      "loss": 1.9931,
      "mean_token_accuracy": 0.5885084450244904,
      "num_tokens": 2939609.0,
      "step": 6710
    },
    {
      "epoch": 2.0107719928186714,
      "grad_norm": 29.25,
      "learning_rate": 1.6492120486734492e-05,
      "loss": 2.232,
      "mean_token_accuracy": 0.5317733138799667,
      "num_tokens": 2944532.0,
      "step": 6720
    },
    {
      "epoch": 2.01376421304608,
      "grad_norm": 33.25,
      "learning_rate": 1.6442250149611013e-05,
      "loss": 2.0345,
      "mean_token_accuracy": 0.5766903549432755,
      "num_tokens": 2948542.0,
      "step": 6730
    },
    {
      "epoch": 2.016756433273489,
      "grad_norm": 26.5,
      "learning_rate": 1.6392379812487534e-05,
      "loss": 2.2125,
      "mean_token_accuracy": 0.5706226840615273,
      "num_tokens": 2953419.0,
      "step": 6740
    },
    {
      "epoch": 2.0197486535008977,
      "grad_norm": 33.5,
      "learning_rate": 1.6342509475364054e-05,
      "loss": 1.7447,
      "mean_token_accuracy": 0.633513942360878,
      "num_tokens": 2956668.0,
      "step": 6750
    },
    {
      "epoch": 2.0227408737283064,
      "grad_norm": 33.0,
      "learning_rate": 1.6292639138240575e-05,
      "loss": 1.6458,
      "mean_token_accuracy": 0.6448787659406662,
      "num_tokens": 2959587.0,
      "step": 6760
    },
    {
      "epoch": 2.025733093955715,
      "grad_norm": 27.125,
      "learning_rate": 1.6242768801117096e-05,
      "loss": 2.2686,
      "mean_token_accuracy": 0.5233890146017075,
      "num_tokens": 2964296.0,
      "step": 6770
    },
    {
      "epoch": 2.028725314183124,
      "grad_norm": 50.0,
      "learning_rate": 1.6192898463993617e-05,
      "loss": 1.961,
      "mean_token_accuracy": 0.5838585048913956,
      "num_tokens": 2968026.0,
      "step": 6780
    },
    {
      "epoch": 2.0317175344105327,
      "grad_norm": 14.0625,
      "learning_rate": 1.6143028126870137e-05,
      "loss": 2.2359,
      "mean_token_accuracy": 0.550393459200859,
      "num_tokens": 2974380.0,
      "step": 6790
    },
    {
      "epoch": 2.0347097546379413,
      "grad_norm": 37.0,
      "learning_rate": 1.609315778974666e-05,
      "loss": 1.8793,
      "mean_token_accuracy": 0.6059096574783325,
      "num_tokens": 2980634.0,
      "step": 6800
    },
    {
      "epoch": 2.03770197486535,
      "grad_norm": 18.75,
      "learning_rate": 1.604328745262318e-05,
      "loss": 1.7964,
      "mean_token_accuracy": 0.6337763130664825,
      "num_tokens": 2984793.0,
      "step": 6810
    },
    {
      "epoch": 2.0406941950927586,
      "grad_norm": 21.875,
      "learning_rate": 1.59934171154997e-05,
      "loss": 1.7028,
      "mean_token_accuracy": 0.6147032797336578,
      "num_tokens": 2989861.0,
      "step": 6820
    },
    {
      "epoch": 2.0436864153201677,
      "grad_norm": 34.0,
      "learning_rate": 1.5943546778376224e-05,
      "loss": 2.1306,
      "mean_token_accuracy": 0.5623955458402634,
      "num_tokens": 2995027.0,
      "step": 6830
    },
    {
      "epoch": 2.0466786355475763,
      "grad_norm": 44.25,
      "learning_rate": 1.5893676441252745e-05,
      "loss": 2.3411,
      "mean_token_accuracy": 0.5126163840293885,
      "num_tokens": 2999046.0,
      "step": 6840
    },
    {
      "epoch": 2.049670855774985,
      "grad_norm": 40.0,
      "learning_rate": 1.5843806104129262e-05,
      "loss": 2.2722,
      "mean_token_accuracy": 0.5325024276971817,
      "num_tokens": 3003260.0,
      "step": 6850
    },
    {
      "epoch": 2.0526630760023936,
      "grad_norm": 31.0,
      "learning_rate": 1.5793935767005786e-05,
      "loss": 2.1896,
      "mean_token_accuracy": 0.5521072208881378,
      "num_tokens": 3007547.0,
      "step": 6860
    },
    {
      "epoch": 2.0556552962298027,
      "grad_norm": 68.0,
      "learning_rate": 1.5744065429882307e-05,
      "loss": 2.0907,
      "mean_token_accuracy": 0.5546874791383744,
      "num_tokens": 3011215.0,
      "step": 6870
    },
    {
      "epoch": 2.0586475164572113,
      "grad_norm": 32.75,
      "learning_rate": 1.5694195092758828e-05,
      "loss": 2.5391,
      "mean_token_accuracy": 0.486826691031456,
      "num_tokens": 3015832.0,
      "step": 6880
    },
    {
      "epoch": 2.06163973668462,
      "grad_norm": 26.875,
      "learning_rate": 1.564432475563535e-05,
      "loss": 2.0941,
      "mean_token_accuracy": 0.5605701714754104,
      "num_tokens": 3020382.0,
      "step": 6890
    },
    {
      "epoch": 2.0646319569120286,
      "grad_norm": 27.75,
      "learning_rate": 1.559445441851187e-05,
      "loss": 1.8473,
      "mean_token_accuracy": 0.6215524405241013,
      "num_tokens": 3024830.0,
      "step": 6900
    },
    {
      "epoch": 2.0676241771394377,
      "grad_norm": 17.0,
      "learning_rate": 1.554458408138839e-05,
      "loss": 1.8716,
      "mean_token_accuracy": 0.5953522980213165,
      "num_tokens": 3029463.0,
      "step": 6910
    },
    {
      "epoch": 2.0706163973668463,
      "grad_norm": 30.375,
      "learning_rate": 1.5494713744264915e-05,
      "loss": 1.8055,
      "mean_token_accuracy": 0.6283937364816665,
      "num_tokens": 3033927.0,
      "step": 6920
    },
    {
      "epoch": 2.073608617594255,
      "grad_norm": 24.875,
      "learning_rate": 1.5444843407141432e-05,
      "loss": 1.8005,
      "mean_token_accuracy": 0.6029859393835068,
      "num_tokens": 3037770.0,
      "step": 6930
    },
    {
      "epoch": 2.0766008378216636,
      "grad_norm": 31.5,
      "learning_rate": 1.5394973070017953e-05,
      "loss": 1.9557,
      "mean_token_accuracy": 0.5977991700172425,
      "num_tokens": 3041979.0,
      "step": 6940
    },
    {
      "epoch": 2.0795930580490722,
      "grad_norm": 39.25,
      "learning_rate": 1.5345102732894477e-05,
      "loss": 2.1927,
      "mean_token_accuracy": 0.5491878598928451,
      "num_tokens": 3045573.0,
      "step": 6950
    },
    {
      "epoch": 2.0825852782764813,
      "grad_norm": 30.5,
      "learning_rate": 1.5295232395770994e-05,
      "loss": 2.5016,
      "mean_token_accuracy": 0.5196672089397907,
      "num_tokens": 3050255.0,
      "step": 6960
    },
    {
      "epoch": 2.08557749850389,
      "grad_norm": 30.0,
      "learning_rate": 1.5245362058647517e-05,
      "loss": 2.4885,
      "mean_token_accuracy": 0.49733306765556334,
      "num_tokens": 3055615.0,
      "step": 6970
    },
    {
      "epoch": 2.0885697187312986,
      "grad_norm": 27.875,
      "learning_rate": 1.5195491721524038e-05,
      "loss": 2.1265,
      "mean_token_accuracy": 0.5600803822278977,
      "num_tokens": 3060218.0,
      "step": 6980
    },
    {
      "epoch": 2.0915619389587072,
      "grad_norm": 36.5,
      "learning_rate": 1.514562138440056e-05,
      "loss": 2.8598,
      "mean_token_accuracy": 0.45551442801952363,
      "num_tokens": 3063650.0,
      "step": 6990
    },
    {
      "epoch": 2.0945541591861163,
      "grad_norm": 23.625,
      "learning_rate": 1.509575104727708e-05,
      "loss": 2.1769,
      "mean_token_accuracy": 0.5520025372505188,
      "num_tokens": 3067429.0,
      "step": 7000
    },
    {
      "epoch": 2.097546379413525,
      "grad_norm": 26.875,
      "learning_rate": 1.5045880710153602e-05,
      "loss": 2.2834,
      "mean_token_accuracy": 0.5472053974866867,
      "num_tokens": 3070937.0,
      "step": 7010
    },
    {
      "epoch": 2.1005385996409336,
      "grad_norm": 27.375,
      "learning_rate": 1.4996010373030123e-05,
      "loss": 2.2649,
      "mean_token_accuracy": 0.5259867966175079,
      "num_tokens": 3076459.0,
      "step": 7020
    },
    {
      "epoch": 2.1035308198683422,
      "grad_norm": 25.875,
      "learning_rate": 1.4946140035906645e-05,
      "loss": 2.0102,
      "mean_token_accuracy": 0.5812536478042603,
      "num_tokens": 3080734.0,
      "step": 7030
    },
    {
      "epoch": 2.106523040095751,
      "grad_norm": 43.5,
      "learning_rate": 1.4896269698783164e-05,
      "loss": 2.2867,
      "mean_token_accuracy": 0.5594320222735405,
      "num_tokens": 3084628.0,
      "step": 7040
    },
    {
      "epoch": 2.10951526032316,
      "grad_norm": 35.0,
      "learning_rate": 1.4846399361659685e-05,
      "loss": 2.0889,
      "mean_token_accuracy": 0.5646486788988113,
      "num_tokens": 3089221.0,
      "step": 7050
    },
    {
      "epoch": 2.1125074805505686,
      "grad_norm": 35.0,
      "learning_rate": 1.4796529024536207e-05,
      "loss": 1.4921,
      "mean_token_accuracy": 0.6738329261541367,
      "num_tokens": 3093896.0,
      "step": 7060
    },
    {
      "epoch": 2.115499700777977,
      "grad_norm": 30.5,
      "learning_rate": 1.4746658687412728e-05,
      "loss": 2.18,
      "mean_token_accuracy": 0.5686471343040467,
      "num_tokens": 3099829.0,
      "step": 7070
    },
    {
      "epoch": 2.118491921005386,
      "grad_norm": 25.0,
      "learning_rate": 1.4696788350289247e-05,
      "loss": 2.137,
      "mean_token_accuracy": 0.5568736657500267,
      "num_tokens": 3104969.0,
      "step": 7080
    },
    {
      "epoch": 2.121484141232795,
      "grad_norm": 36.25,
      "learning_rate": 1.464691801316577e-05,
      "loss": 1.5476,
      "mean_token_accuracy": 0.6572899341583252,
      "num_tokens": 3108655.0,
      "step": 7090
    },
    {
      "epoch": 2.1244763614602036,
      "grad_norm": 26.5,
      "learning_rate": 1.459704767604229e-05,
      "loss": 2.1881,
      "mean_token_accuracy": 0.5540028423070907,
      "num_tokens": 3114053.0,
      "step": 7100
    },
    {
      "epoch": 2.127468581687612,
      "grad_norm": 24.125,
      "learning_rate": 1.454717733891881e-05,
      "loss": 1.575,
      "mean_token_accuracy": 0.6586545079946518,
      "num_tokens": 3119135.0,
      "step": 7110
    },
    {
      "epoch": 2.130460801915021,
      "grad_norm": 45.75,
      "learning_rate": 1.4497307001795332e-05,
      "loss": 1.9054,
      "mean_token_accuracy": 0.6022154986858368,
      "num_tokens": 3123334.0,
      "step": 7120
    },
    {
      "epoch": 2.1334530221424295,
      "grad_norm": 26.75,
      "learning_rate": 1.4447436664671855e-05,
      "loss": 1.6995,
      "mean_token_accuracy": 0.6492284148931503,
      "num_tokens": 3128890.0,
      "step": 7130
    },
    {
      "epoch": 2.1364452423698386,
      "grad_norm": 5.46875,
      "learning_rate": 1.4397566327548375e-05,
      "loss": 2.4091,
      "mean_token_accuracy": 0.5174450129270554,
      "num_tokens": 3133639.0,
      "step": 7140
    },
    {
      "epoch": 2.139437462597247,
      "grad_norm": 28.75,
      "learning_rate": 1.4347695990424895e-05,
      "loss": 2.2288,
      "mean_token_accuracy": 0.5391303896903992,
      "num_tokens": 3137904.0,
      "step": 7150
    },
    {
      "epoch": 2.142429682824656,
      "grad_norm": 30.25,
      "learning_rate": 1.4297825653301417e-05,
      "loss": 1.524,
      "mean_token_accuracy": 0.6681494385004043,
      "num_tokens": 3143124.0,
      "step": 7160
    },
    {
      "epoch": 2.1454219030520645,
      "grad_norm": 30.0,
      "learning_rate": 1.4247955316177938e-05,
      "loss": 1.8881,
      "mean_token_accuracy": 0.608069297671318,
      "num_tokens": 3147219.0,
      "step": 7170
    },
    {
      "epoch": 2.1484141232794736,
      "grad_norm": 26.625,
      "learning_rate": 1.419808497905446e-05,
      "loss": 1.9308,
      "mean_token_accuracy": 0.5991872996091843,
      "num_tokens": 3151863.0,
      "step": 7180
    },
    {
      "epoch": 2.151406343506882,
      "grad_norm": 39.75,
      "learning_rate": 1.414821464193098e-05,
      "loss": 1.6427,
      "mean_token_accuracy": 0.6218064963817597,
      "num_tokens": 3155373.0,
      "step": 7190
    },
    {
      "epoch": 2.154398563734291,
      "grad_norm": 34.5,
      "learning_rate": 1.40983443048075e-05,
      "loss": 2.276,
      "mean_token_accuracy": 0.545945143699646,
      "num_tokens": 3159495.0,
      "step": 7200
    },
    {
      "epoch": 2.1573907839616995,
      "grad_norm": 26.25,
      "learning_rate": 1.4048473967684023e-05,
      "loss": 1.938,
      "mean_token_accuracy": 0.5874078899621964,
      "num_tokens": 3164398.0,
      "step": 7210
    },
    {
      "epoch": 2.160383004189108,
      "grad_norm": 23.25,
      "learning_rate": 1.3998603630560545e-05,
      "loss": 2.2729,
      "mean_token_accuracy": 0.5383491784334182,
      "num_tokens": 3166982.0,
      "step": 7220
    },
    {
      "epoch": 2.163375224416517,
      "grad_norm": 58.75,
      "learning_rate": 1.3948733293437064e-05,
      "loss": 1.6851,
      "mean_token_accuracy": 0.6350438117980957,
      "num_tokens": 3170364.0,
      "step": 7230
    },
    {
      "epoch": 2.166367444643926,
      "grad_norm": 33.25,
      "learning_rate": 1.3898862956313585e-05,
      "loss": 1.8358,
      "mean_token_accuracy": 0.6009310275316239,
      "num_tokens": 3173578.0,
      "step": 7240
    },
    {
      "epoch": 2.1693596648713345,
      "grad_norm": 26.625,
      "learning_rate": 1.3848992619190108e-05,
      "loss": 2.528,
      "mean_token_accuracy": 0.507240629196167,
      "num_tokens": 3178958.0,
      "step": 7250
    },
    {
      "epoch": 2.172351885098743,
      "grad_norm": 32.5,
      "learning_rate": 1.3799122282066627e-05,
      "loss": 1.7139,
      "mean_token_accuracy": 0.6379852175712586,
      "num_tokens": 3182457.0,
      "step": 7260
    },
    {
      "epoch": 2.175344105326152,
      "grad_norm": 73.0,
      "learning_rate": 1.3749251944943147e-05,
      "loss": 1.7549,
      "mean_token_accuracy": 0.6359315663576126,
      "num_tokens": 3186873.0,
      "step": 7270
    },
    {
      "epoch": 2.178336325553561,
      "grad_norm": 26.0,
      "learning_rate": 1.369938160781967e-05,
      "loss": 1.7649,
      "mean_token_accuracy": 0.6203795820474625,
      "num_tokens": 3190376.0,
      "step": 7280
    },
    {
      "epoch": 2.1813285457809695,
      "grad_norm": 15.75,
      "learning_rate": 1.364951127069619e-05,
      "loss": 1.8914,
      "mean_token_accuracy": 0.6036392211914062,
      "num_tokens": 3195318.0,
      "step": 7290
    },
    {
      "epoch": 2.184320766008378,
      "grad_norm": 49.0,
      "learning_rate": 1.359964093357271e-05,
      "loss": 1.9615,
      "mean_token_accuracy": 0.6003385275602341,
      "num_tokens": 3199389.0,
      "step": 7300
    },
    {
      "epoch": 2.1873129862357867,
      "grad_norm": 25.75,
      "learning_rate": 1.3549770596449232e-05,
      "loss": 2.3876,
      "mean_token_accuracy": 0.5285185039043426,
      "num_tokens": 3204839.0,
      "step": 7310
    },
    {
      "epoch": 2.190305206463196,
      "grad_norm": 30.25,
      "learning_rate": 1.3499900259325755e-05,
      "loss": 1.8095,
      "mean_token_accuracy": 0.6179131418466568,
      "num_tokens": 3208534.0,
      "step": 7320
    },
    {
      "epoch": 2.1932974266906045,
      "grad_norm": 48.0,
      "learning_rate": 1.3450029922202276e-05,
      "loss": 2.143,
      "mean_token_accuracy": 0.5466442197561264,
      "num_tokens": 3213107.0,
      "step": 7330
    },
    {
      "epoch": 2.196289646918013,
      "grad_norm": 19.0,
      "learning_rate": 1.3400159585078795e-05,
      "loss": 2.1544,
      "mean_token_accuracy": 0.5515126138925552,
      "num_tokens": 3217204.0,
      "step": 7340
    },
    {
      "epoch": 2.1992818671454217,
      "grad_norm": 30.25,
      "learning_rate": 1.3350289247955317e-05,
      "loss": 1.325,
      "mean_token_accuracy": 0.6974451810121536,
      "num_tokens": 3222040.0,
      "step": 7350
    },
    {
      "epoch": 2.202274087372831,
      "grad_norm": 27.125,
      "learning_rate": 1.3300418910831838e-05,
      "loss": 2.1716,
      "mean_token_accuracy": 0.5470889180898666,
      "num_tokens": 3227199.0,
      "step": 7360
    },
    {
      "epoch": 2.2052663076002395,
      "grad_norm": 27.375,
      "learning_rate": 1.325054857370836e-05,
      "loss": 2.3477,
      "mean_token_accuracy": 0.5298527553677559,
      "num_tokens": 3231808.0,
      "step": 7370
    },
    {
      "epoch": 2.208258527827648,
      "grad_norm": 46.25,
      "learning_rate": 1.320067823658488e-05,
      "loss": 1.5753,
      "mean_token_accuracy": 0.6611310541629791,
      "num_tokens": 3235862.0,
      "step": 7380
    },
    {
      "epoch": 2.2112507480550567,
      "grad_norm": 29.875,
      "learning_rate": 1.31508078994614e-05,
      "loss": 1.6568,
      "mean_token_accuracy": 0.6521678894758225,
      "num_tokens": 3239945.0,
      "step": 7390
    },
    {
      "epoch": 2.2142429682824654,
      "grad_norm": 54.25,
      "learning_rate": 1.3100937562337923e-05,
      "loss": 1.7918,
      "mean_token_accuracy": 0.6133479118347168,
      "num_tokens": 3243822.0,
      "step": 7400
    },
    {
      "epoch": 2.2172351885098744,
      "grad_norm": 43.0,
      "learning_rate": 1.3051067225214442e-05,
      "loss": 1.7599,
      "mean_token_accuracy": 0.6251044929027557,
      "num_tokens": 3248413.0,
      "step": 7410
    },
    {
      "epoch": 2.220227408737283,
      "grad_norm": 32.5,
      "learning_rate": 1.3001196888090963e-05,
      "loss": 2.4222,
      "mean_token_accuracy": 0.5013878315687179,
      "num_tokens": 3252864.0,
      "step": 7420
    },
    {
      "epoch": 2.2232196289646917,
      "grad_norm": 25.0,
      "learning_rate": 1.2951326550967485e-05,
      "loss": 1.9676,
      "mean_token_accuracy": 0.5924913167953492,
      "num_tokens": 3258877.0,
      "step": 7430
    },
    {
      "epoch": 2.2262118491921004,
      "grad_norm": 32.25,
      "learning_rate": 1.2901456213844008e-05,
      "loss": 2.1059,
      "mean_token_accuracy": 0.5556224346160888,
      "num_tokens": 3263966.0,
      "step": 7440
    },
    {
      "epoch": 2.2292040694195094,
      "grad_norm": 29.75,
      "learning_rate": 1.2851585876720527e-05,
      "loss": 1.6945,
      "mean_token_accuracy": 0.6180789291858673,
      "num_tokens": 3268942.0,
      "step": 7450
    },
    {
      "epoch": 2.232196289646918,
      "grad_norm": 18.75,
      "learning_rate": 1.2801715539597048e-05,
      "loss": 1.9998,
      "mean_token_accuracy": 0.5755996882915497,
      "num_tokens": 3273260.0,
      "step": 7460
    },
    {
      "epoch": 2.2351885098743267,
      "grad_norm": 38.0,
      "learning_rate": 1.275184520247357e-05,
      "loss": 2.0105,
      "mean_token_accuracy": 0.5683426201343537,
      "num_tokens": 3277591.0,
      "step": 7470
    },
    {
      "epoch": 2.2381807301017353,
      "grad_norm": 33.25,
      "learning_rate": 1.2701974865350091e-05,
      "loss": 2.1261,
      "mean_token_accuracy": 0.5513424545526504,
      "num_tokens": 3280618.0,
      "step": 7480
    },
    {
      "epoch": 2.2411729503291444,
      "grad_norm": 31.75,
      "learning_rate": 1.265210452822661e-05,
      "loss": 1.9736,
      "mean_token_accuracy": 0.58391652405262,
      "num_tokens": 3285301.0,
      "step": 7490
    },
    {
      "epoch": 2.244165170556553,
      "grad_norm": 26.5,
      "learning_rate": 1.2602234191103132e-05,
      "loss": 2.2189,
      "mean_token_accuracy": 0.5404590010643006,
      "num_tokens": 3290313.0,
      "step": 7500
    },
    {
      "epoch": 2.2471573907839617,
      "grad_norm": 32.0,
      "learning_rate": 1.2552363853979653e-05,
      "loss": 1.7485,
      "mean_token_accuracy": 0.6247122138738632,
      "num_tokens": 3294690.0,
      "step": 7510
    },
    {
      "epoch": 2.2501496110113703,
      "grad_norm": 18.0,
      "learning_rate": 1.2502493516856176e-05,
      "loss": 2.2453,
      "mean_token_accuracy": 0.544939911365509,
      "num_tokens": 3299559.0,
      "step": 7520
    },
    {
      "epoch": 2.253141831238779,
      "grad_norm": 26.5,
      "learning_rate": 1.2452623179732697e-05,
      "loss": 1.9581,
      "mean_token_accuracy": 0.6086787462234498,
      "num_tokens": 3305592.0,
      "step": 7530
    },
    {
      "epoch": 2.256134051466188,
      "grad_norm": 26.0,
      "learning_rate": 1.2402752842609217e-05,
      "loss": 1.8021,
      "mean_token_accuracy": 0.6123438656330109,
      "num_tokens": 3310369.0,
      "step": 7540
    },
    {
      "epoch": 2.2591262716935967,
      "grad_norm": 37.25,
      "learning_rate": 1.2352882505485738e-05,
      "loss": 2.3046,
      "mean_token_accuracy": 0.5472021579742432,
      "num_tokens": 3314477.0,
      "step": 7550
    },
    {
      "epoch": 2.2621184919210053,
      "grad_norm": 23.25,
      "learning_rate": 1.2303012168362259e-05,
      "loss": 2.0173,
      "mean_token_accuracy": 0.5664256989955903,
      "num_tokens": 3318450.0,
      "step": 7560
    },
    {
      "epoch": 2.265110712148414,
      "grad_norm": 14.9375,
      "learning_rate": 1.225314183123878e-05,
      "loss": 2.0863,
      "mean_token_accuracy": 0.5759825021028518,
      "num_tokens": 3323229.0,
      "step": 7570
    },
    {
      "epoch": 2.2681029323758226,
      "grad_norm": 17.875,
      "learning_rate": 1.22032714941153e-05,
      "loss": 2.7287,
      "mean_token_accuracy": 0.497147136926651,
      "num_tokens": 3327917.0,
      "step": 7580
    },
    {
      "epoch": 2.2710951526032317,
      "grad_norm": 12.1875,
      "learning_rate": 1.2153401156991821e-05,
      "loss": 1.8352,
      "mean_token_accuracy": 0.6011588901281357,
      "num_tokens": 3331229.0,
      "step": 7590
    },
    {
      "epoch": 2.2740873728306403,
      "grad_norm": 19.125,
      "learning_rate": 1.2103530819868344e-05,
      "loss": 1.8881,
      "mean_token_accuracy": 0.610486413538456,
      "num_tokens": 3335337.0,
      "step": 7600
    },
    {
      "epoch": 2.277079593058049,
      "grad_norm": 29.375,
      "learning_rate": 1.2053660482744863e-05,
      "loss": 2.0414,
      "mean_token_accuracy": 0.5699247926473617,
      "num_tokens": 3339984.0,
      "step": 7610
    },
    {
      "epoch": 2.280071813285458,
      "grad_norm": 46.25,
      "learning_rate": 1.2003790145621385e-05,
      "loss": 2.2427,
      "mean_token_accuracy": 0.5332239389419555,
      "num_tokens": 3343756.0,
      "step": 7620
    },
    {
      "epoch": 2.2830640335128667,
      "grad_norm": 40.5,
      "learning_rate": 1.1953919808497906e-05,
      "loss": 2.0443,
      "mean_token_accuracy": 0.5590986162424088,
      "num_tokens": 3347489.0,
      "step": 7630
    },
    {
      "epoch": 2.2860562537402753,
      "grad_norm": 29.75,
      "learning_rate": 1.1904049471374427e-05,
      "loss": 1.7902,
      "mean_token_accuracy": 0.621918460726738,
      "num_tokens": 3352828.0,
      "step": 7640
    },
    {
      "epoch": 2.289048473967684,
      "grad_norm": 37.75,
      "learning_rate": 1.1854179134250948e-05,
      "loss": 2.136,
      "mean_token_accuracy": 0.5378258466720581,
      "num_tokens": 3356433.0,
      "step": 7650
    },
    {
      "epoch": 2.2920406941950926,
      "grad_norm": 27.75,
      "learning_rate": 1.180430879712747e-05,
      "loss": 1.5546,
      "mean_token_accuracy": 0.6686955988407135,
      "num_tokens": 3362086.0,
      "step": 7660
    },
    {
      "epoch": 2.2950329144225017,
      "grad_norm": 39.5,
      "learning_rate": 1.175443846000399e-05,
      "loss": 1.9859,
      "mean_token_accuracy": 0.5787241399288178,
      "num_tokens": 3366370.0,
      "step": 7670
    },
    {
      "epoch": 2.2980251346499103,
      "grad_norm": 59.75,
      "learning_rate": 1.1704568122880512e-05,
      "loss": 1.5931,
      "mean_token_accuracy": 0.6556290119886399,
      "num_tokens": 3369998.0,
      "step": 7680
    },
    {
      "epoch": 2.301017354877319,
      "grad_norm": 64.0,
      "learning_rate": 1.1654697785757033e-05,
      "loss": 2.3505,
      "mean_token_accuracy": 0.530811858177185,
      "num_tokens": 3374677.0,
      "step": 7690
    },
    {
      "epoch": 2.3040095751047276,
      "grad_norm": 31.875,
      "learning_rate": 1.1604827448633553e-05,
      "loss": 2.0854,
      "mean_token_accuracy": 0.5914190322160721,
      "num_tokens": 3380055.0,
      "step": 7700
    },
    {
      "epoch": 2.3070017953321367,
      "grad_norm": 18.375,
      "learning_rate": 1.1554957111510074e-05,
      "loss": 1.8562,
      "mean_token_accuracy": 0.601532056927681,
      "num_tokens": 3383671.0,
      "step": 7710
    },
    {
      "epoch": 2.3099940155595453,
      "grad_norm": 30.875,
      "learning_rate": 1.1505086774386595e-05,
      "loss": 1.6012,
      "mean_token_accuracy": 0.6378510087728501,
      "num_tokens": 3387927.0,
      "step": 7720
    },
    {
      "epoch": 2.312986235786954,
      "grad_norm": 27.625,
      "learning_rate": 1.1455216437263116e-05,
      "loss": 2.6541,
      "mean_token_accuracy": 0.4840825363993645,
      "num_tokens": 3391989.0,
      "step": 7730
    },
    {
      "epoch": 2.3159784560143626,
      "grad_norm": 34.25,
      "learning_rate": 1.1405346100139637e-05,
      "loss": 1.9968,
      "mean_token_accuracy": 0.6017626792192459,
      "num_tokens": 3397104.0,
      "step": 7740
    },
    {
      "epoch": 2.3189706762417712,
      "grad_norm": 34.5,
      "learning_rate": 1.1355475763016159e-05,
      "loss": 1.9871,
      "mean_token_accuracy": 0.5988510757684707,
      "num_tokens": 3402355.0,
      "step": 7750
    },
    {
      "epoch": 2.3219628964691803,
      "grad_norm": 34.5,
      "learning_rate": 1.130560542589268e-05,
      "loss": 1.8842,
      "mean_token_accuracy": 0.6055288344621659,
      "num_tokens": 3405790.0,
      "step": 7760
    },
    {
      "epoch": 2.324955116696589,
      "grad_norm": 35.5,
      "learning_rate": 1.12557350887692e-05,
      "loss": 1.9169,
      "mean_token_accuracy": 0.595410618185997,
      "num_tokens": 3409449.0,
      "step": 7770
    },
    {
      "epoch": 2.3279473369239976,
      "grad_norm": 25.375,
      "learning_rate": 1.1205864751645721e-05,
      "loss": 1.7218,
      "mean_token_accuracy": 0.6117983996868134,
      "num_tokens": 3414393.0,
      "step": 7780
    },
    {
      "epoch": 2.330939557151406,
      "grad_norm": 32.5,
      "learning_rate": 1.1155994414522242e-05,
      "loss": 2.7433,
      "mean_token_accuracy": 0.46884743869304657,
      "num_tokens": 3418163.0,
      "step": 7790
    },
    {
      "epoch": 2.3339317773788153,
      "grad_norm": 44.25,
      "learning_rate": 1.1106124077398763e-05,
      "loss": 2.2331,
      "mean_token_accuracy": 0.5329806178808212,
      "num_tokens": 3422058.0,
      "step": 7800
    },
    {
      "epoch": 2.336923997606224,
      "grad_norm": 32.5,
      "learning_rate": 1.1056253740275286e-05,
      "loss": 2.0709,
      "mean_token_accuracy": 0.580215010046959,
      "num_tokens": 3426116.0,
      "step": 7810
    },
    {
      "epoch": 2.3399162178336326,
      "grad_norm": 21.625,
      "learning_rate": 1.1006383403151806e-05,
      "loss": 2.2106,
      "mean_token_accuracy": 0.5685110569000245,
      "num_tokens": 3430160.0,
      "step": 7820
    },
    {
      "epoch": 2.342908438061041,
      "grad_norm": 38.0,
      "learning_rate": 1.0956513066028327e-05,
      "loss": 2.1967,
      "mean_token_accuracy": 0.5463057830929756,
      "num_tokens": 3434026.0,
      "step": 7830
    },
    {
      "epoch": 2.34590065828845,
      "grad_norm": 61.25,
      "learning_rate": 1.0906642728904848e-05,
      "loss": 1.9145,
      "mean_token_accuracy": 0.5919240117073059,
      "num_tokens": 3438573.0,
      "step": 7840
    },
    {
      "epoch": 2.348892878515859,
      "grad_norm": 33.25,
      "learning_rate": 1.085677239178137e-05,
      "loss": 2.443,
      "mean_token_accuracy": 0.5153850048780442,
      "num_tokens": 3444090.0,
      "step": 7850
    },
    {
      "epoch": 2.3518850987432676,
      "grad_norm": 25.625,
      "learning_rate": 1.080690205465789e-05,
      "loss": 1.6518,
      "mean_token_accuracy": 0.6537412971258163,
      "num_tokens": 3447767.0,
      "step": 7860
    },
    {
      "epoch": 2.354877318970676,
      "grad_norm": 39.5,
      "learning_rate": 1.075703171753441e-05,
      "loss": 2.2954,
      "mean_token_accuracy": 0.5424281567335129,
      "num_tokens": 3451267.0,
      "step": 7870
    },
    {
      "epoch": 2.357869539198085,
      "grad_norm": 50.25,
      "learning_rate": 1.0707161380410933e-05,
      "loss": 2.0823,
      "mean_token_accuracy": 0.5709426254034042,
      "num_tokens": 3455425.0,
      "step": 7880
    },
    {
      "epoch": 2.360861759425494,
      "grad_norm": 25.625,
      "learning_rate": 1.0657291043287452e-05,
      "loss": 2.2418,
      "mean_token_accuracy": 0.5553388357162475,
      "num_tokens": 3459713.0,
      "step": 7890
    },
    {
      "epoch": 2.3638539796529026,
      "grad_norm": 23.875,
      "learning_rate": 1.0607420706163974e-05,
      "loss": 2.0082,
      "mean_token_accuracy": 0.6014505207538605,
      "num_tokens": 3463321.0,
      "step": 7900
    },
    {
      "epoch": 2.366846199880311,
      "grad_norm": 25.5,
      "learning_rate": 1.0557550369040495e-05,
      "loss": 1.8273,
      "mean_token_accuracy": 0.6049247473478317,
      "num_tokens": 3468143.0,
      "step": 7910
    },
    {
      "epoch": 2.36983842010772,
      "grad_norm": 22.625,
      "learning_rate": 1.0507680031917016e-05,
      "loss": 1.736,
      "mean_token_accuracy": 0.6522809594869614,
      "num_tokens": 3471017.0,
      "step": 7920
    },
    {
      "epoch": 2.3728306403351285,
      "grad_norm": 43.5,
      "learning_rate": 1.0457809694793537e-05,
      "loss": 1.9577,
      "mean_token_accuracy": 0.5918497428297996,
      "num_tokens": 3475374.0,
      "step": 7930
    },
    {
      "epoch": 2.3758228605625376,
      "grad_norm": 20.125,
      "learning_rate": 1.040793935767006e-05,
      "loss": 2.023,
      "mean_token_accuracy": 0.5746719360351562,
      "num_tokens": 3480701.0,
      "step": 7940
    },
    {
      "epoch": 2.378815080789946,
      "grad_norm": 27.5,
      "learning_rate": 1.0358069020546578e-05,
      "loss": 2.4225,
      "mean_token_accuracy": 0.5121300339698791,
      "num_tokens": 3484654.0,
      "step": 7950
    },
    {
      "epoch": 2.381807301017355,
      "grad_norm": 49.25,
      "learning_rate": 1.03081986834231e-05,
      "loss": 2.5751,
      "mean_token_accuracy": 0.5052287340164184,
      "num_tokens": 3488371.0,
      "step": 7960
    },
    {
      "epoch": 2.3847995212447635,
      "grad_norm": 51.75,
      "learning_rate": 1.0258328346299622e-05,
      "loss": 1.4107,
      "mean_token_accuracy": 0.6880755305290223,
      "num_tokens": 3493507.0,
      "step": 7970
    },
    {
      "epoch": 2.3877917414721725,
      "grad_norm": 17.375,
      "learning_rate": 1.0208458009176142e-05,
      "loss": 2.5548,
      "mean_token_accuracy": 0.4705794721841812,
      "num_tokens": 3498623.0,
      "step": 7980
    },
    {
      "epoch": 2.390783961699581,
      "grad_norm": 27.875,
      "learning_rate": 1.0158587672052663e-05,
      "loss": 1.909,
      "mean_token_accuracy": 0.6068336576223373,
      "num_tokens": 3502592.0,
      "step": 7990
    },
    {
      "epoch": 2.39377618192699,
      "grad_norm": 9.8125,
      "learning_rate": 1.0108717334929186e-05,
      "loss": 2.1661,
      "mean_token_accuracy": 0.5623240023851395,
      "num_tokens": 3507098.0,
      "step": 8000
    },
    {
      "epoch": 2.3967684021543985,
      "grad_norm": 33.0,
      "learning_rate": 1.0058846997805706e-05,
      "loss": 1.8429,
      "mean_token_accuracy": 0.6223942905664444,
      "num_tokens": 3511321.0,
      "step": 8010
    },
    {
      "epoch": 2.399760622381807,
      "grad_norm": 30.625,
      "learning_rate": 1.0008976660682226e-05,
      "loss": 1.8434,
      "mean_token_accuracy": 0.6259656488895416,
      "num_tokens": 3516657.0,
      "step": 8020
    },
    {
      "epoch": 2.402752842609216,
      "grad_norm": 27.25,
      "learning_rate": 9.959106323558748e-06,
      "loss": 1.9366,
      "mean_token_accuracy": 0.5986247807741165,
      "num_tokens": 3520858.0,
      "step": 8030
    },
    {
      "epoch": 2.405745062836625,
      "grad_norm": 19.0,
      "learning_rate": 9.909235986435269e-06,
      "loss": 1.5475,
      "mean_token_accuracy": 0.662953433394432,
      "num_tokens": 3525453.0,
      "step": 8040
    },
    {
      "epoch": 2.4087372830640335,
      "grad_norm": 51.25,
      "learning_rate": 9.85936564931179e-06,
      "loss": 1.894,
      "mean_token_accuracy": 0.6088249057531356,
      "num_tokens": 3529847.0,
      "step": 8050
    },
    {
      "epoch": 2.411729503291442,
      "grad_norm": 18.375,
      "learning_rate": 9.80949531218831e-06,
      "loss": 2.0747,
      "mean_token_accuracy": 0.5732029706239701,
      "num_tokens": 3535537.0,
      "step": 8060
    },
    {
      "epoch": 2.414721723518851,
      "grad_norm": 16.625,
      "learning_rate": 9.759624975064833e-06,
      "loss": 2.1334,
      "mean_token_accuracy": 0.5579861745238304,
      "num_tokens": 3539446.0,
      "step": 8070
    },
    {
      "epoch": 2.41771394374626,
      "grad_norm": 12.625,
      "learning_rate": 9.709754637941352e-06,
      "loss": 2.0315,
      "mean_token_accuracy": 0.5979110836982727,
      "num_tokens": 3542930.0,
      "step": 8080
    },
    {
      "epoch": 2.4207061639736684,
      "grad_norm": 19.25,
      "learning_rate": 9.659884300817875e-06,
      "loss": 1.7469,
      "mean_token_accuracy": 0.6296554863452911,
      "num_tokens": 3546756.0,
      "step": 8090
    },
    {
      "epoch": 2.423698384201077,
      "grad_norm": 37.0,
      "learning_rate": 9.610013963694395e-06,
      "loss": 1.7101,
      "mean_token_accuracy": 0.6294068902730942,
      "num_tokens": 3550808.0,
      "step": 8100
    },
    {
      "epoch": 2.4266906044284857,
      "grad_norm": 44.25,
      "learning_rate": 9.560143626570916e-06,
      "loss": 1.8353,
      "mean_token_accuracy": 0.6092359751462937,
      "num_tokens": 3554987.0,
      "step": 8110
    },
    {
      "epoch": 2.429682824655895,
      "grad_norm": 37.5,
      "learning_rate": 9.510273289447437e-06,
      "loss": 2.305,
      "mean_token_accuracy": 0.5346463292837143,
      "num_tokens": 3559276.0,
      "step": 8120
    },
    {
      "epoch": 2.4326750448833034,
      "grad_norm": 41.0,
      "learning_rate": 9.46040295232396e-06,
      "loss": 2.0357,
      "mean_token_accuracy": 0.5618094027042388,
      "num_tokens": 3563553.0,
      "step": 8130
    },
    {
      "epoch": 2.435667265110712,
      "grad_norm": 28.625,
      "learning_rate": 9.410532615200478e-06,
      "loss": 1.8299,
      "mean_token_accuracy": 0.6292552351951599,
      "num_tokens": 3567569.0,
      "step": 8140
    },
    {
      "epoch": 2.4386594853381207,
      "grad_norm": 27.375,
      "learning_rate": 9.360662278077001e-06,
      "loss": 1.6487,
      "mean_token_accuracy": 0.6505157589912415,
      "num_tokens": 3572902.0,
      "step": 8150
    },
    {
      "epoch": 2.44165170556553,
      "grad_norm": 42.0,
      "learning_rate": 9.310791940953522e-06,
      "loss": 1.9738,
      "mean_token_accuracy": 0.5727031499147415,
      "num_tokens": 3577788.0,
      "step": 8160
    },
    {
      "epoch": 2.4446439257929384,
      "grad_norm": 26.75,
      "learning_rate": 9.260921603830041e-06,
      "loss": 1.6068,
      "mean_token_accuracy": 0.6601596623659134,
      "num_tokens": 3582166.0,
      "step": 8170
    },
    {
      "epoch": 2.447636146020347,
      "grad_norm": 25.875,
      "learning_rate": 9.211051266706563e-06,
      "loss": 2.1705,
      "mean_token_accuracy": 0.5626641541719437,
      "num_tokens": 3587132.0,
      "step": 8180
    },
    {
      "epoch": 2.4506283662477557,
      "grad_norm": 33.75,
      "learning_rate": 9.161180929583084e-06,
      "loss": 1.7271,
      "mean_token_accuracy": 0.6523958176374436,
      "num_tokens": 3592763.0,
      "step": 8190
    },
    {
      "epoch": 2.4536205864751643,
      "grad_norm": 23.625,
      "learning_rate": 9.111310592459605e-06,
      "loss": 2.0387,
      "mean_token_accuracy": 0.595306721329689,
      "num_tokens": 3597038.0,
      "step": 8200
    },
    {
      "epoch": 2.4566128067025734,
      "grad_norm": 26.75,
      "learning_rate": 9.061440255336126e-06,
      "loss": 2.464,
      "mean_token_accuracy": 0.516951784491539,
      "num_tokens": 3602367.0,
      "step": 8210
    },
    {
      "epoch": 2.459605026929982,
      "grad_norm": 32.0,
      "learning_rate": 9.011569918212648e-06,
      "loss": 1.6953,
      "mean_token_accuracy": 0.6542330086231232,
      "num_tokens": 3606253.0,
      "step": 8220
    },
    {
      "epoch": 2.4625972471573907,
      "grad_norm": 17.75,
      "learning_rate": 8.961699581089169e-06,
      "loss": 1.2013,
      "mean_token_accuracy": 0.7238958954811097,
      "num_tokens": 3612181.0,
      "step": 8230
    },
    {
      "epoch": 2.4655894673847993,
      "grad_norm": 28.0,
      "learning_rate": 8.91182924396569e-06,
      "loss": 2.1341,
      "mean_token_accuracy": 0.5624263644218445,
      "num_tokens": 3616098.0,
      "step": 8240
    },
    {
      "epoch": 2.4685816876122084,
      "grad_norm": 43.0,
      "learning_rate": 8.86195890684221e-06,
      "loss": 2.1282,
      "mean_token_accuracy": 0.5523239612579346,
      "num_tokens": 3620981.0,
      "step": 8250
    },
    {
      "epoch": 2.471573907839617,
      "grad_norm": 29.25,
      "learning_rate": 8.812088569718731e-06,
      "loss": 2.1556,
      "mean_token_accuracy": 0.5922388851642608,
      "num_tokens": 3627070.0,
      "step": 8260
    },
    {
      "epoch": 2.4745661280670257,
      "grad_norm": 29.875,
      "learning_rate": 8.762218232595252e-06,
      "loss": 2.3435,
      "mean_token_accuracy": 0.5244899958372116,
      "num_tokens": 3631458.0,
      "step": 8270
    },
    {
      "epoch": 2.4775583482944343,
      "grad_norm": 41.75,
      "learning_rate": 8.712347895471775e-06,
      "loss": 2.0722,
      "mean_token_accuracy": 0.5771478354930878,
      "num_tokens": 3635862.0,
      "step": 8280
    },
    {
      "epoch": 2.480550568521843,
      "grad_norm": 29.75,
      "learning_rate": 8.662477558348295e-06,
      "loss": 1.9438,
      "mean_token_accuracy": 0.5872433006763458,
      "num_tokens": 3639631.0,
      "step": 8290
    },
    {
      "epoch": 2.483542788749252,
      "grad_norm": 27.125,
      "learning_rate": 8.612607221224816e-06,
      "loss": 1.4937,
      "mean_token_accuracy": 0.6888282895088196,
      "num_tokens": 3643726.0,
      "step": 8300
    },
    {
      "epoch": 2.4865350089766607,
      "grad_norm": 31.25,
      "learning_rate": 8.562736884101337e-06,
      "loss": 2.3963,
      "mean_token_accuracy": 0.5211231708526611,
      "num_tokens": 3647701.0,
      "step": 8310
    },
    {
      "epoch": 2.4895272292040693,
      "grad_norm": 74.5,
      "learning_rate": 8.512866546977858e-06,
      "loss": 1.8017,
      "mean_token_accuracy": 0.6249741375446319,
      "num_tokens": 3651854.0,
      "step": 8320
    },
    {
      "epoch": 2.492519449431478,
      "grad_norm": 20.375,
      "learning_rate": 8.462996209854379e-06,
      "loss": 2.1971,
      "mean_token_accuracy": 0.5514743745326995,
      "num_tokens": 3655610.0,
      "step": 8330
    },
    {
      "epoch": 2.495511669658887,
      "grad_norm": 60.25,
      "learning_rate": 8.4131258727309e-06,
      "loss": 2.1792,
      "mean_token_accuracy": 0.5621172040700912,
      "num_tokens": 3659872.0,
      "step": 8340
    },
    {
      "epoch": 2.4985038898862957,
      "grad_norm": 37.75,
      "learning_rate": 8.363255535607422e-06,
      "loss": 2.0568,
      "mean_token_accuracy": 0.5745551347732544,
      "num_tokens": 3664216.0,
      "step": 8350
    },
    {
      "epoch": 2.5014961101137043,
      "grad_norm": 33.5,
      "learning_rate": 8.313385198483941e-06,
      "loss": 1.8874,
      "mean_token_accuracy": 0.6061397522687912,
      "num_tokens": 3668880.0,
      "step": 8360
    },
    {
      "epoch": 2.504488330341113,
      "grad_norm": 32.75,
      "learning_rate": 8.263514861360464e-06,
      "loss": 1.9858,
      "mean_token_accuracy": 0.5864089488983154,
      "num_tokens": 3675294.0,
      "step": 8370
    },
    {
      "epoch": 2.5074805505685216,
      "grad_norm": 40.75,
      "learning_rate": 8.213644524236984e-06,
      "loss": 1.5634,
      "mean_token_accuracy": 0.6630536049604416,
      "num_tokens": 3680712.0,
      "step": 8380
    },
    {
      "epoch": 2.5104727707959307,
      "grad_norm": 28.5,
      "learning_rate": 8.163774187113505e-06,
      "loss": 1.8208,
      "mean_token_accuracy": 0.6154579162597656,
      "num_tokens": 3685294.0,
      "step": 8390
    },
    {
      "epoch": 2.5134649910233393,
      "grad_norm": 49.0,
      "learning_rate": 8.113903849990026e-06,
      "loss": 2.1475,
      "mean_token_accuracy": 0.562446939945221,
      "num_tokens": 3688792.0,
      "step": 8400
    },
    {
      "epoch": 2.516457211250748,
      "grad_norm": 31.625,
      "learning_rate": 8.064033512866548e-06,
      "loss": 2.1782,
      "mean_token_accuracy": 0.5494516760110855,
      "num_tokens": 3693374.0,
      "step": 8410
    },
    {
      "epoch": 2.519449431478157,
      "grad_norm": 42.75,
      "learning_rate": 8.014163175743067e-06,
      "loss": 1.516,
      "mean_token_accuracy": 0.6680077582597732,
      "num_tokens": 3697590.0,
      "step": 8420
    },
    {
      "epoch": 2.5224416517055657,
      "grad_norm": 36.25,
      "learning_rate": 7.96429283861959e-06,
      "loss": 2.8254,
      "mean_token_accuracy": 0.45352172702550886,
      "num_tokens": 3702193.0,
      "step": 8430
    },
    {
      "epoch": 2.5254338719329743,
      "grad_norm": 50.75,
      "learning_rate": 7.91442250149611e-06,
      "loss": 1.8653,
      "mean_token_accuracy": 0.6249462842941285,
      "num_tokens": 3706496.0,
      "step": 8440
    },
    {
      "epoch": 2.528426092160383,
      "grad_norm": 11.0,
      "learning_rate": 7.864552164372632e-06,
      "loss": 1.8116,
      "mean_token_accuracy": 0.6244892299175262,
      "num_tokens": 3710279.0,
      "step": 8450
    },
    {
      "epoch": 2.5314183123877916,
      "grad_norm": 38.75,
      "learning_rate": 7.814681827249152e-06,
      "loss": 1.8314,
      "mean_token_accuracy": 0.631192871928215,
      "num_tokens": 3713786.0,
      "step": 8460
    },
    {
      "epoch": 2.5344105326152,
      "grad_norm": 22.0,
      "learning_rate": 7.764811490125673e-06,
      "loss": 1.9399,
      "mean_token_accuracy": 0.5822267323732376,
      "num_tokens": 3718329.0,
      "step": 8470
    },
    {
      "epoch": 2.5374027528426093,
      "grad_norm": 22.0,
      "learning_rate": 7.714941153002194e-06,
      "loss": 2.0836,
      "mean_token_accuracy": 0.5822148770093918,
      "num_tokens": 3722215.0,
      "step": 8480
    },
    {
      "epoch": 2.540394973070018,
      "grad_norm": 22.875,
      "learning_rate": 7.665070815878715e-06,
      "loss": 1.8375,
      "mean_token_accuracy": 0.6172699481248856,
      "num_tokens": 3725939.0,
      "step": 8490
    },
    {
      "epoch": 2.5433871932974266,
      "grad_norm": 24.75,
      "learning_rate": 7.615200478755237e-06,
      "loss": 1.6583,
      "mean_token_accuracy": 0.6369618535041809,
      "num_tokens": 3730746.0,
      "step": 8500
    },
    {
      "epoch": 2.5463794135248357,
      "grad_norm": 40.25,
      "learning_rate": 7.565330141631757e-06,
      "loss": 2.1574,
      "mean_token_accuracy": 0.5527519911527634,
      "num_tokens": 3734695.0,
      "step": 8510
    },
    {
      "epoch": 2.5493716337522443,
      "grad_norm": 28.875,
      "learning_rate": 7.515459804508279e-06,
      "loss": 1.6126,
      "mean_token_accuracy": 0.650888466835022,
      "num_tokens": 3738514.0,
      "step": 8520
    },
    {
      "epoch": 2.552363853979653,
      "grad_norm": 21.125,
      "learning_rate": 7.4655894673848e-06,
      "loss": 2.3758,
      "mean_token_accuracy": 0.5187936812639237,
      "num_tokens": 3743566.0,
      "step": 8530
    },
    {
      "epoch": 2.5553560742070616,
      "grad_norm": 21.625,
      "learning_rate": 7.415719130261321e-06,
      "loss": 2.1145,
      "mean_token_accuracy": 0.5670811682939529,
      "num_tokens": 3747765.0,
      "step": 8540
    },
    {
      "epoch": 2.55834829443447,
      "grad_norm": 20.625,
      "learning_rate": 7.365848793137842e-06,
      "loss": 1.8283,
      "mean_token_accuracy": 0.6210377246141434,
      "num_tokens": 3752107.0,
      "step": 8550
    },
    {
      "epoch": 2.561340514661879,
      "grad_norm": 29.125,
      "learning_rate": 7.315978456014364e-06,
      "loss": 1.8758,
      "mean_token_accuracy": 0.6091048032045364,
      "num_tokens": 3756006.0,
      "step": 8560
    },
    {
      "epoch": 2.564332734889288,
      "grad_norm": 24.0,
      "learning_rate": 7.266108118890884e-06,
      "loss": 2.0391,
      "mean_token_accuracy": 0.576360335946083,
      "num_tokens": 3759916.0,
      "step": 8570
    },
    {
      "epoch": 2.5673249551166966,
      "grad_norm": 64.5,
      "learning_rate": 7.216237781767405e-06,
      "loss": 2.1935,
      "mean_token_accuracy": 0.5540457516908646,
      "num_tokens": 3763119.0,
      "step": 8580
    },
    {
      "epoch": 2.570317175344105,
      "grad_norm": 29.0,
      "learning_rate": 7.166367444643926e-06,
      "loss": 1.4691,
      "mean_token_accuracy": 0.6710379928350448,
      "num_tokens": 3768397.0,
      "step": 8590
    },
    {
      "epoch": 2.5733093955715143,
      "grad_norm": 19.375,
      "learning_rate": 7.116497107520448e-06,
      "loss": 1.8466,
      "mean_token_accuracy": 0.6215602785348893,
      "num_tokens": 3772852.0,
      "step": 8600
    },
    {
      "epoch": 2.576301615798923,
      "grad_norm": 29.125,
      "learning_rate": 7.0666267703969685e-06,
      "loss": 1.796,
      "mean_token_accuracy": 0.6303971827030181,
      "num_tokens": 3777531.0,
      "step": 8610
    },
    {
      "epoch": 2.5792938360263316,
      "grad_norm": 39.25,
      "learning_rate": 7.0167564332734884e-06,
      "loss": 1.7612,
      "mean_token_accuracy": 0.6404202848672866,
      "num_tokens": 3781420.0,
      "step": 8620
    },
    {
      "epoch": 2.58228605625374,
      "grad_norm": 23.125,
      "learning_rate": 6.96688609615001e-06,
      "loss": 2.5314,
      "mean_token_accuracy": 0.49140324890613557,
      "num_tokens": 3783768.0,
      "step": 8630
    },
    {
      "epoch": 2.585278276481149,
      "grad_norm": 29.5,
      "learning_rate": 6.917015759026531e-06,
      "loss": 2.1956,
      "mean_token_accuracy": 0.5796375393867492,
      "num_tokens": 3786827.0,
      "step": 8640
    },
    {
      "epoch": 2.588270496708558,
      "grad_norm": 35.0,
      "learning_rate": 6.8671454219030525e-06,
      "loss": 2.2291,
      "mean_token_accuracy": 0.5234993904829025,
      "num_tokens": 3790613.0,
      "step": 8650
    },
    {
      "epoch": 2.5912627169359665,
      "grad_norm": 25.0,
      "learning_rate": 6.817275084779573e-06,
      "loss": 2.9253,
      "mean_token_accuracy": 0.44142225980758665,
      "num_tokens": 3796033.0,
      "step": 8660
    },
    {
      "epoch": 2.594254937163375,
      "grad_norm": 33.25,
      "learning_rate": 6.767404747656095e-06,
      "loss": 2.8558,
      "mean_token_accuracy": 0.46484392285346987,
      "num_tokens": 3799913.0,
      "step": 8670
    },
    {
      "epoch": 2.597247157390784,
      "grad_norm": 25.0,
      "learning_rate": 6.717534410532615e-06,
      "loss": 2.0318,
      "mean_token_accuracy": 0.5812970370054245,
      "num_tokens": 3804951.0,
      "step": 8680
    },
    {
      "epoch": 2.600239377618193,
      "grad_norm": 37.75,
      "learning_rate": 6.6676640734091365e-06,
      "loss": 1.9003,
      "mean_token_accuracy": 0.6071935892105103,
      "num_tokens": 3808152.0,
      "step": 8690
    },
    {
      "epoch": 2.6032315978456015,
      "grad_norm": 24.75,
      "learning_rate": 6.617793736285657e-06,
      "loss": 1.9307,
      "mean_token_accuracy": 0.6070375055074692,
      "num_tokens": 3811382.0,
      "step": 8700
    },
    {
      "epoch": 2.60622381807301,
      "grad_norm": 25.625,
      "learning_rate": 6.567923399162179e-06,
      "loss": 1.6272,
      "mean_token_accuracy": 0.6634323447942734,
      "num_tokens": 3816044.0,
      "step": 8710
    },
    {
      "epoch": 2.609216038300419,
      "grad_norm": 29.625,
      "learning_rate": 6.5180530620387e-06,
      "loss": 2.0368,
      "mean_token_accuracy": 0.5823567092418671,
      "num_tokens": 3820355.0,
      "step": 8720
    },
    {
      "epoch": 2.6122082585278275,
      "grad_norm": 28.375,
      "learning_rate": 6.468182724915221e-06,
      "loss": 2.2438,
      "mean_token_accuracy": 0.5450930893421173,
      "num_tokens": 3824214.0,
      "step": 8730
    },
    {
      "epoch": 2.6152004787552365,
      "grad_norm": 36.75,
      "learning_rate": 6.418312387791741e-06,
      "loss": 1.9044,
      "mean_token_accuracy": 0.6274087101221084,
      "num_tokens": 3829582.0,
      "step": 8740
    },
    {
      "epoch": 2.618192698982645,
      "grad_norm": 34.25,
      "learning_rate": 6.368442050668264e-06,
      "loss": 2.3823,
      "mean_token_accuracy": 0.522585716843605,
      "num_tokens": 3834310.0,
      "step": 8750
    },
    {
      "epoch": 2.621184919210054,
      "grad_norm": 14.0625,
      "learning_rate": 6.318571713544784e-06,
      "loss": 2.1984,
      "mean_token_accuracy": 0.5534607917070389,
      "num_tokens": 3841092.0,
      "step": 8760
    },
    {
      "epoch": 2.6241771394374624,
      "grad_norm": 36.0,
      "learning_rate": 6.268701376421305e-06,
      "loss": 1.5606,
      "mean_token_accuracy": 0.6726154178380966,
      "num_tokens": 3846284.0,
      "step": 8770
    },
    {
      "epoch": 2.6271693596648715,
      "grad_norm": 31.625,
      "learning_rate": 6.218831039297826e-06,
      "loss": 1.9837,
      "mean_token_accuracy": 0.5853519588708878,
      "num_tokens": 3849669.0,
      "step": 8780
    },
    {
      "epoch": 2.63016157989228,
      "grad_norm": 33.5,
      "learning_rate": 6.168960702174347e-06,
      "loss": 1.4509,
      "mean_token_accuracy": 0.6898014098405838,
      "num_tokens": 3852892.0,
      "step": 8790
    },
    {
      "epoch": 2.633153800119689,
      "grad_norm": 33.5,
      "learning_rate": 6.119090365050869e-06,
      "loss": 1.6132,
      "mean_token_accuracy": 0.6505110949277878,
      "num_tokens": 3856161.0,
      "step": 8800
    },
    {
      "epoch": 2.6361460203470974,
      "grad_norm": 16.25,
      "learning_rate": 6.0692200279273894e-06,
      "loss": 2.2723,
      "mean_token_accuracy": 0.5464581429958344,
      "num_tokens": 3860799.0,
      "step": 8810
    },
    {
      "epoch": 2.639138240574506,
      "grad_norm": 12.125,
      "learning_rate": 6.01934969080391e-06,
      "loss": 2.0183,
      "mean_token_accuracy": 0.574837502837181,
      "num_tokens": 3864947.0,
      "step": 8820
    },
    {
      "epoch": 2.642130460801915,
      "grad_norm": 22.875,
      "learning_rate": 5.969479353680431e-06,
      "loss": 1.6786,
      "mean_token_accuracy": 0.6457695811986923,
      "num_tokens": 3869804.0,
      "step": 8830
    },
    {
      "epoch": 2.645122681029324,
      "grad_norm": 27.75,
      "learning_rate": 5.919609016556952e-06,
      "loss": 2.1758,
      "mean_token_accuracy": 0.5874395340681076,
      "num_tokens": 3874553.0,
      "step": 8840
    },
    {
      "epoch": 2.6481149012567324,
      "grad_norm": 32.5,
      "learning_rate": 5.869738679433473e-06,
      "loss": 1.8416,
      "mean_token_accuracy": 0.6180979579687118,
      "num_tokens": 3878177.0,
      "step": 8850
    },
    {
      "epoch": 2.651107121484141,
      "grad_norm": 25.875,
      "learning_rate": 5.819868342309994e-06,
      "loss": 1.9093,
      "mean_token_accuracy": 0.6036896035075188,
      "num_tokens": 3881704.0,
      "step": 8860
    },
    {
      "epoch": 2.65409934171155,
      "grad_norm": 48.75,
      "learning_rate": 5.769998005186515e-06,
      "loss": 1.7578,
      "mean_token_accuracy": 0.6388221472501755,
      "num_tokens": 3886023.0,
      "step": 8870
    },
    {
      "epoch": 2.657091561938959,
      "grad_norm": 51.25,
      "learning_rate": 5.720127668063036e-06,
      "loss": 2.7133,
      "mean_token_accuracy": 0.47807915806770324,
      "num_tokens": 3890025.0,
      "step": 8880
    },
    {
      "epoch": 2.6600837821663674,
      "grad_norm": 25.875,
      "learning_rate": 5.6702573309395575e-06,
      "loss": 1.9606,
      "mean_token_accuracy": 0.5969976544380188,
      "num_tokens": 3893625.0,
      "step": 8890
    },
    {
      "epoch": 2.663076002393776,
      "grad_norm": 14.5,
      "learning_rate": 5.620386993816078e-06,
      "loss": 2.481,
      "mean_token_accuracy": 0.5044433563947678,
      "num_tokens": 3899249.0,
      "step": 8900
    },
    {
      "epoch": 2.6660682226211847,
      "grad_norm": 38.0,
      "learning_rate": 5.5705166566926e-06,
      "loss": 1.7464,
      "mean_token_accuracy": 0.6185297697782517,
      "num_tokens": 3903839.0,
      "step": 8910
    },
    {
      "epoch": 2.669060442848594,
      "grad_norm": 36.75,
      "learning_rate": 5.520646319569121e-06,
      "loss": 2.305,
      "mean_token_accuracy": 0.5210951715707779,
      "num_tokens": 3906730.0,
      "step": 8920
    },
    {
      "epoch": 2.6720526630760024,
      "grad_norm": 34.0,
      "learning_rate": 5.4707759824456415e-06,
      "loss": 1.1306,
      "mean_token_accuracy": 0.7620425939559936,
      "num_tokens": 3910169.0,
      "step": 8930
    },
    {
      "epoch": 2.675044883303411,
      "grad_norm": 28.0,
      "learning_rate": 5.420905645322163e-06,
      "loss": 1.8178,
      "mean_token_accuracy": 0.6252069532871246,
      "num_tokens": 3914875.0,
      "step": 8940
    },
    {
      "epoch": 2.67803710353082,
      "grad_norm": 13.625,
      "learning_rate": 5.371035308198684e-06,
      "loss": 1.6462,
      "mean_token_accuracy": 0.6465720534324646,
      "num_tokens": 3918847.0,
      "step": 8950
    },
    {
      "epoch": 2.6810293237582288,
      "grad_norm": 44.0,
      "learning_rate": 5.321164971075205e-06,
      "loss": 1.9698,
      "mean_token_accuracy": 0.6158938229084014,
      "num_tokens": 3921966.0,
      "step": 8960
    },
    {
      "epoch": 2.6840215439856374,
      "grad_norm": 30.875,
      "learning_rate": 5.271294633951726e-06,
      "loss": 2.3922,
      "mean_token_accuracy": 0.5279428198933601,
      "num_tokens": 3926081.0,
      "step": 8970
    },
    {
      "epoch": 2.687013764213046,
      "grad_norm": 30.75,
      "learning_rate": 5.221424296828246e-06,
      "loss": 2.0273,
      "mean_token_accuracy": 0.5862969070672989,
      "num_tokens": 3931317.0,
      "step": 8980
    },
    {
      "epoch": 2.6900059844404547,
      "grad_norm": 31.875,
      "learning_rate": 5.171553959704768e-06,
      "loss": 2.6512,
      "mean_token_accuracy": 0.4965745806694031,
      "num_tokens": 3935938.0,
      "step": 8990
    },
    {
      "epoch": 2.6929982046678633,
      "grad_norm": 29.875,
      "learning_rate": 5.121683622581289e-06,
      "loss": 2.0997,
      "mean_token_accuracy": 0.5695976376533508,
      "num_tokens": 3941232.0,
      "step": 9000
    },
    {
      "epoch": 2.6959904248952724,
      "grad_norm": 22.25,
      "learning_rate": 5.0718132854578096e-06,
      "loss": 1.9799,
      "mean_token_accuracy": 0.5941878288984299,
      "num_tokens": 3944101.0,
      "step": 9010
    },
    {
      "epoch": 2.698982645122681,
      "grad_norm": 22.375,
      "learning_rate": 5.021942948334331e-06,
      "loss": 1.1311,
      "mean_token_accuracy": 0.7389927178621292,
      "num_tokens": 3947397.0,
      "step": 9020
    },
    {
      "epoch": 2.7019748653500897,
      "grad_norm": 27.0,
      "learning_rate": 4.972072611210852e-06,
      "loss": 1.8437,
      "mean_token_accuracy": 0.620136833190918,
      "num_tokens": 3953487.0,
      "step": 9030
    },
    {
      "epoch": 2.7049670855774988,
      "grad_norm": 53.0,
      "learning_rate": 4.922202274087373e-06,
      "loss": 2.3224,
      "mean_token_accuracy": 0.5359113857150077,
      "num_tokens": 3957159.0,
      "step": 9040
    },
    {
      "epoch": 2.7079593058049074,
      "grad_norm": 29.0,
      "learning_rate": 4.872331936963894e-06,
      "loss": 1.6784,
      "mean_token_accuracy": 0.6427404046058655,
      "num_tokens": 3961444.0,
      "step": 9050
    },
    {
      "epoch": 2.710951526032316,
      "grad_norm": 47.25,
      "learning_rate": 4.822461599840415e-06,
      "loss": 2.3031,
      "mean_token_accuracy": 0.5322034955024719,
      "num_tokens": 3966270.0,
      "step": 9060
    },
    {
      "epoch": 2.7139437462597247,
      "grad_norm": 35.0,
      "learning_rate": 4.772591262716936e-06,
      "loss": 1.7478,
      "mean_token_accuracy": 0.6326006144285202,
      "num_tokens": 3970793.0,
      "step": 9070
    },
    {
      "epoch": 2.7169359664871333,
      "grad_norm": 21.25,
      "learning_rate": 4.722720925593458e-06,
      "loss": 1.8017,
      "mean_token_accuracy": 0.6345465123653412,
      "num_tokens": 3974977.0,
      "step": 9080
    },
    {
      "epoch": 2.719928186714542,
      "grad_norm": 34.25,
      "learning_rate": 4.6728505884699784e-06,
      "loss": 1.8966,
      "mean_token_accuracy": 0.6168614506721497,
      "num_tokens": 3980230.0,
      "step": 9090
    },
    {
      "epoch": 2.722920406941951,
      "grad_norm": 30.0,
      "learning_rate": 4.622980251346499e-06,
      "loss": 1.9168,
      "mean_token_accuracy": 0.5983828186988831,
      "num_tokens": 3985529.0,
      "step": 9100
    },
    {
      "epoch": 2.7259126271693597,
      "grad_norm": 41.5,
      "learning_rate": 4.573109914223021e-06,
      "loss": 1.7793,
      "mean_token_accuracy": 0.6030765622854233,
      "num_tokens": 3989337.0,
      "step": 9110
    },
    {
      "epoch": 2.7289048473967683,
      "grad_norm": 46.0,
      "learning_rate": 4.523239577099542e-06,
      "loss": 2.2499,
      "mean_token_accuracy": 0.5526516705751419,
      "num_tokens": 3995079.0,
      "step": 9120
    },
    {
      "epoch": 2.7318970676241774,
      "grad_norm": 27.0,
      "learning_rate": 4.4733692399760625e-06,
      "loss": 1.8144,
      "mean_token_accuracy": 0.6078069090843201,
      "num_tokens": 3999875.0,
      "step": 9130
    },
    {
      "epoch": 2.734889287851586,
      "grad_norm": 35.0,
      "learning_rate": 4.423498902852583e-06,
      "loss": 1.819,
      "mean_token_accuracy": 0.6298864781856537,
      "num_tokens": 4004317.0,
      "step": 9140
    },
    {
      "epoch": 2.7378815080789947,
      "grad_norm": 29.25,
      "learning_rate": 4.373628565729104e-06,
      "loss": 2.062,
      "mean_token_accuracy": 0.5830421209335327,
      "num_tokens": 4009760.0,
      "step": 9150
    },
    {
      "epoch": 2.7408737283064033,
      "grad_norm": 27.375,
      "learning_rate": 4.323758228605626e-06,
      "loss": 2.1911,
      "mean_token_accuracy": 0.5540030539035797,
      "num_tokens": 4014489.0,
      "step": 9160
    },
    {
      "epoch": 2.743865948533812,
      "grad_norm": 23.25,
      "learning_rate": 4.2738878914821465e-06,
      "loss": 1.8985,
      "mean_token_accuracy": 0.5950787127017975,
      "num_tokens": 4018655.0,
      "step": 9170
    },
    {
      "epoch": 2.7468581687612206,
      "grad_norm": 25.125,
      "learning_rate": 4.224017554358667e-06,
      "loss": 1.5748,
      "mean_token_accuracy": 0.6524001300334931,
      "num_tokens": 4023175.0,
      "step": 9180
    },
    {
      "epoch": 2.7498503889886297,
      "grad_norm": 18.875,
      "learning_rate": 4.174147217235189e-06,
      "loss": 1.598,
      "mean_token_accuracy": 0.6746624201536179,
      "num_tokens": 4028084.0,
      "step": 9190
    },
    {
      "epoch": 2.7528426092160383,
      "grad_norm": 41.0,
      "learning_rate": 4.12427688011171e-06,
      "loss": 1.7713,
      "mean_token_accuracy": 0.6220902055501938,
      "num_tokens": 4032253.0,
      "step": 9200
    },
    {
      "epoch": 2.755834829443447,
      "grad_norm": 29.0,
      "learning_rate": 4.0744065429882305e-06,
      "loss": 1.8611,
      "mean_token_accuracy": 0.6254755735397339,
      "num_tokens": 4035717.0,
      "step": 9210
    },
    {
      "epoch": 2.758827049670856,
      "grad_norm": 34.75,
      "learning_rate": 4.024536205864752e-06,
      "loss": 1.786,
      "mean_token_accuracy": 0.6216672778129577,
      "num_tokens": 4039463.0,
      "step": 9220
    },
    {
      "epoch": 2.7618192698982647,
      "grad_norm": 17.125,
      "learning_rate": 3.974665868741273e-06,
      "loss": 1.4427,
      "mean_token_accuracy": 0.6828666806221009,
      "num_tokens": 4044969.0,
      "step": 9230
    },
    {
      "epoch": 2.7648114901256733,
      "grad_norm": 39.0,
      "learning_rate": 3.924795531617794e-06,
      "loss": 2.1013,
      "mean_token_accuracy": 0.5784855902194976,
      "num_tokens": 4049817.0,
      "step": 9240
    },
    {
      "epoch": 2.767803710353082,
      "grad_norm": 29.25,
      "learning_rate": 3.874925194494315e-06,
      "loss": 2.1812,
      "mean_token_accuracy": 0.5448796033859253,
      "num_tokens": 4054405.0,
      "step": 9250
    },
    {
      "epoch": 2.7707959305804906,
      "grad_norm": 30.25,
      "learning_rate": 3.825054857370836e-06,
      "loss": 2.0633,
      "mean_token_accuracy": 0.5805003941059113,
      "num_tokens": 4059175.0,
      "step": 9260
    },
    {
      "epoch": 2.773788150807899,
      "grad_norm": 34.5,
      "learning_rate": 3.7751845202473574e-06,
      "loss": 1.5614,
      "mean_token_accuracy": 0.6646808028221131,
      "num_tokens": 4062650.0,
      "step": 9270
    },
    {
      "epoch": 2.7767803710353083,
      "grad_norm": 40.75,
      "learning_rate": 3.7253141831238778e-06,
      "loss": 2.1483,
      "mean_token_accuracy": 0.5612222149968147,
      "num_tokens": 4066947.0,
      "step": 9280
    },
    {
      "epoch": 2.779772591262717,
      "grad_norm": 17.375,
      "learning_rate": 3.675443846000399e-06,
      "loss": 2.0027,
      "mean_token_accuracy": 0.5933429479599,
      "num_tokens": 4071052.0,
      "step": 9290
    },
    {
      "epoch": 2.7827648114901256,
      "grad_norm": 31.0,
      "learning_rate": 3.6255735088769198e-06,
      "loss": 1.4943,
      "mean_token_accuracy": 0.6853374063968658,
      "num_tokens": 4075442.0,
      "step": 9300
    },
    {
      "epoch": 2.7857570317175346,
      "grad_norm": 12.375,
      "learning_rate": 3.575703171753441e-06,
      "loss": 1.8588,
      "mean_token_accuracy": 0.600016513466835,
      "num_tokens": 4080536.0,
      "step": 9310
    },
    {
      "epoch": 2.7887492519449433,
      "grad_norm": 34.25,
      "learning_rate": 3.525832834629962e-06,
      "loss": 1.9483,
      "mean_token_accuracy": 0.5879528015851975,
      "num_tokens": 4084930.0,
      "step": 9320
    },
    {
      "epoch": 2.791741472172352,
      "grad_norm": 29.625,
      "learning_rate": 3.475962497506483e-06,
      "loss": 1.6104,
      "mean_token_accuracy": 0.660100382566452,
      "num_tokens": 4088724.0,
      "step": 9330
    },
    {
      "epoch": 2.7947336923997605,
      "grad_norm": 39.0,
      "learning_rate": 3.426092160383004e-06,
      "loss": 2.4462,
      "mean_token_accuracy": 0.5218830987811088,
      "num_tokens": 4092698.0,
      "step": 9340
    },
    {
      "epoch": 2.797725912627169,
      "grad_norm": 38.75,
      "learning_rate": 3.3762218232595254e-06,
      "loss": 1.7577,
      "mean_token_accuracy": 0.6282264947891235,
      "num_tokens": 4097093.0,
      "step": 9350
    },
    {
      "epoch": 2.800718132854578,
      "grad_norm": 82.5,
      "learning_rate": 3.3263514861360466e-06,
      "loss": 2.0219,
      "mean_token_accuracy": 0.5767106354236603,
      "num_tokens": 4101442.0,
      "step": 9360
    },
    {
      "epoch": 2.803710353081987,
      "grad_norm": 30.0,
      "learning_rate": 3.2764811490125674e-06,
      "loss": 1.3033,
      "mean_token_accuracy": 0.7084575563669204,
      "num_tokens": 4104841.0,
      "step": 9370
    },
    {
      "epoch": 2.8067025733093955,
      "grad_norm": 32.5,
      "learning_rate": 3.2266108118890887e-06,
      "loss": 1.5835,
      "mean_token_accuracy": 0.6537139505147934,
      "num_tokens": 4108121.0,
      "step": 9380
    },
    {
      "epoch": 2.809694793536804,
      "grad_norm": 25.875,
      "learning_rate": 3.17674047476561e-06,
      "loss": 1.6108,
      "mean_token_accuracy": 0.6637275129556656,
      "num_tokens": 4112032.0,
      "step": 9390
    },
    {
      "epoch": 2.8126870137642133,
      "grad_norm": 32.75,
      "learning_rate": 3.1268701376421307e-06,
      "loss": 1.8535,
      "mean_token_accuracy": 0.6086684614419937,
      "num_tokens": 4115984.0,
      "step": 9400
    },
    {
      "epoch": 2.815679233991622,
      "grad_norm": 35.25,
      "learning_rate": 3.0769998005186515e-06,
      "loss": 1.8364,
      "mean_token_accuracy": 0.6273092567920685,
      "num_tokens": 4120579.0,
      "step": 9410
    },
    {
      "epoch": 2.8186714542190305,
      "grad_norm": 33.75,
      "learning_rate": 3.0271294633951727e-06,
      "loss": 1.7187,
      "mean_token_accuracy": 0.6125760763883591,
      "num_tokens": 4123676.0,
      "step": 9420
    },
    {
      "epoch": 2.821663674446439,
      "grad_norm": 32.25,
      "learning_rate": 2.977259126271694e-06,
      "loss": 1.3378,
      "mean_token_accuracy": 0.7154994040727616,
      "num_tokens": 4128761.0,
      "step": 9430
    },
    {
      "epoch": 2.824655894673848,
      "grad_norm": 33.0,
      "learning_rate": 2.9273887891482147e-06,
      "loss": 1.4772,
      "mean_token_accuracy": 0.6775191038846969,
      "num_tokens": 4132492.0,
      "step": 9440
    },
    {
      "epoch": 2.827648114901257,
      "grad_norm": 41.5,
      "learning_rate": 2.877518452024736e-06,
      "loss": 2.1733,
      "mean_token_accuracy": 0.5554867714643479,
      "num_tokens": 4137709.0,
      "step": 9450
    },
    {
      "epoch": 2.8306403351286655,
      "grad_norm": 22.125,
      "learning_rate": 2.827648114901257e-06,
      "loss": 1.8694,
      "mean_token_accuracy": 0.5886508136987686,
      "num_tokens": 4141086.0,
      "step": 9460
    },
    {
      "epoch": 2.833632555356074,
      "grad_norm": 31.625,
      "learning_rate": 2.777777777777778e-06,
      "loss": 1.431,
      "mean_token_accuracy": 0.6860785514116288,
      "num_tokens": 4146674.0,
      "step": 9470
    },
    {
      "epoch": 2.836624775583483,
      "grad_norm": 27.125,
      "learning_rate": 2.7279074406542987e-06,
      "loss": 2.2025,
      "mean_token_accuracy": 0.554258507490158,
      "num_tokens": 4150892.0,
      "step": 9480
    },
    {
      "epoch": 2.839616995810892,
      "grad_norm": 45.0,
      "learning_rate": 2.67803710353082e-06,
      "loss": 2.46,
      "mean_token_accuracy": 0.500794404745102,
      "num_tokens": 4156396.0,
      "step": 9490
    },
    {
      "epoch": 2.8426092160383005,
      "grad_norm": 17.25,
      "learning_rate": 2.628166766407341e-06,
      "loss": 2.0167,
      "mean_token_accuracy": 0.5807638555765152,
      "num_tokens": 4160340.0,
      "step": 9500
    },
    {
      "epoch": 2.845601436265709,
      "grad_norm": 37.75,
      "learning_rate": 2.578296429283862e-06,
      "loss": 2.3393,
      "mean_token_accuracy": 0.5255331352353096,
      "num_tokens": 4164764.0,
      "step": 9510
    },
    {
      "epoch": 2.848593656493118,
      "grad_norm": 44.25,
      "learning_rate": 2.528426092160383e-06,
      "loss": 1.9644,
      "mean_token_accuracy": 0.6037229418754577,
      "num_tokens": 4168859.0,
      "step": 9520
    },
    {
      "epoch": 2.8515858767205264,
      "grad_norm": 26.5,
      "learning_rate": 2.4785557550369044e-06,
      "loss": 1.3245,
      "mean_token_accuracy": 0.7033275246620179,
      "num_tokens": 4174167.0,
      "step": 9530
    },
    {
      "epoch": 2.8545780969479355,
      "grad_norm": 14.4375,
      "learning_rate": 2.4286854179134256e-06,
      "loss": 1.2336,
      "mean_token_accuracy": 0.732453578710556,
      "num_tokens": 4178884.0,
      "step": 9540
    },
    {
      "epoch": 2.857570317175344,
      "grad_norm": 18.625,
      "learning_rate": 2.378815080789946e-06,
      "loss": 1.5867,
      "mean_token_accuracy": 0.6451264679431915,
      "num_tokens": 4182310.0,
      "step": 9550
    },
    {
      "epoch": 2.860562537402753,
      "grad_norm": 14.0625,
      "learning_rate": 2.328944743666467e-06,
      "loss": 1.7747,
      "mean_token_accuracy": 0.6581346958875656,
      "num_tokens": 4187444.0,
      "step": 9560
    },
    {
      "epoch": 2.8635547576301614,
      "grad_norm": 57.0,
      "learning_rate": 2.2790744065429884e-06,
      "loss": 1.9725,
      "mean_token_accuracy": 0.6179261445999146,
      "num_tokens": 4192217.0,
      "step": 9570
    },
    {
      "epoch": 2.8665469778575705,
      "grad_norm": 50.75,
      "learning_rate": 2.229204069419509e-06,
      "loss": 2.4587,
      "mean_token_accuracy": 0.4987264096736908,
      "num_tokens": 4195921.0,
      "step": 9580
    },
    {
      "epoch": 2.869539198084979,
      "grad_norm": 25.0,
      "learning_rate": 2.1793337322960304e-06,
      "loss": 1.3648,
      "mean_token_accuracy": 0.7000836312770844,
      "num_tokens": 4200450.0,
      "step": 9590
    },
    {
      "epoch": 2.872531418312388,
      "grad_norm": 61.75,
      "learning_rate": 2.1294633951725516e-06,
      "loss": 1.72,
      "mean_token_accuracy": 0.6363197565078735,
      "num_tokens": 4204745.0,
      "step": 9600
    },
    {
      "epoch": 2.8755236385397964,
      "grad_norm": 39.5,
      "learning_rate": 2.079593058049073e-06,
      "loss": 1.6018,
      "mean_token_accuracy": 0.6612495809793473,
      "num_tokens": 4208026.0,
      "step": 9610
    },
    {
      "epoch": 2.878515858767205,
      "grad_norm": 52.75,
      "learning_rate": 2.029722720925593e-06,
      "loss": 1.902,
      "mean_token_accuracy": 0.6017478436231614,
      "num_tokens": 4212714.0,
      "step": 9620
    },
    {
      "epoch": 2.881508078994614,
      "grad_norm": 23.25,
      "learning_rate": 1.9798523838021144e-06,
      "loss": 2.171,
      "mean_token_accuracy": 0.5636053562164307,
      "num_tokens": 4217985.0,
      "step": 9630
    },
    {
      "epoch": 2.884500299222023,
      "grad_norm": 21.25,
      "learning_rate": 1.9299820466786356e-06,
      "loss": 2.16,
      "mean_token_accuracy": 0.5612272381782532,
      "num_tokens": 4222542.0,
      "step": 9640
    },
    {
      "epoch": 2.8874925194494314,
      "grad_norm": 8.875,
      "learning_rate": 1.8801117095551566e-06,
      "loss": 2.0578,
      "mean_token_accuracy": 0.5891127914190293,
      "num_tokens": 4227617.0,
      "step": 9650
    },
    {
      "epoch": 2.89048473967684,
      "grad_norm": 30.375,
      "learning_rate": 1.8302413724316779e-06,
      "loss": 2.158,
      "mean_token_accuracy": 0.5701712220907211,
      "num_tokens": 4232468.0,
      "step": 9660
    },
    {
      "epoch": 2.893476959904249,
      "grad_norm": 50.0,
      "learning_rate": 1.7803710353081989e-06,
      "loss": 1.9668,
      "mean_token_accuracy": 0.5937556356191636,
      "num_tokens": 4236640.0,
      "step": 9670
    },
    {
      "epoch": 2.8964691801316578,
      "grad_norm": 35.5,
      "learning_rate": 1.7305006981847199e-06,
      "loss": 2.0277,
      "mean_token_accuracy": 0.5735794514417648,
      "num_tokens": 4242045.0,
      "step": 9680
    },
    {
      "epoch": 2.8994614003590664,
      "grad_norm": 50.0,
      "learning_rate": 1.680630361061241e-06,
      "loss": 2.0398,
      "mean_token_accuracy": 0.5936207205057145,
      "num_tokens": 4245908.0,
      "step": 9690
    },
    {
      "epoch": 2.902453620586475,
      "grad_norm": 51.75,
      "learning_rate": 1.6307600239377617e-06,
      "loss": 2.1149,
      "mean_token_accuracy": 0.5694210886955261,
      "num_tokens": 4250424.0,
      "step": 9700
    },
    {
      "epoch": 2.9054458408138837,
      "grad_norm": 32.5,
      "learning_rate": 1.5808896868142829e-06,
      "loss": 2.1492,
      "mean_token_accuracy": 0.5405183464288712,
      "num_tokens": 4254490.0,
      "step": 9710
    },
    {
      "epoch": 2.9084380610412928,
      "grad_norm": 17.75,
      "learning_rate": 1.5310193496908039e-06,
      "loss": 1.8224,
      "mean_token_accuracy": 0.6002204239368438,
      "num_tokens": 4258253.0,
      "step": 9720
    },
    {
      "epoch": 2.9114302812687014,
      "grad_norm": 47.5,
      "learning_rate": 1.4811490125673251e-06,
      "loss": 1.8376,
      "mean_token_accuracy": 0.610488024353981,
      "num_tokens": 4261515.0,
      "step": 9730
    },
    {
      "epoch": 2.91442250149611,
      "grad_norm": 27.25,
      "learning_rate": 1.4312786754438461e-06,
      "loss": 2.3097,
      "mean_token_accuracy": 0.5365430325269699,
      "num_tokens": 4268730.0,
      "step": 9740
    },
    {
      "epoch": 2.917414721723519,
      "grad_norm": 18.5,
      "learning_rate": 1.3814083383203671e-06,
      "loss": 1.8928,
      "mean_token_accuracy": 0.5975118815898895,
      "num_tokens": 4272857.0,
      "step": 9750
    },
    {
      "epoch": 2.9204069419509278,
      "grad_norm": 42.75,
      "learning_rate": 1.3315380011968881e-06,
      "loss": 2.1786,
      "mean_token_accuracy": 0.5498067557811737,
      "num_tokens": 4277174.0,
      "step": 9760
    },
    {
      "epoch": 2.9233991621783364,
      "grad_norm": 23.125,
      "learning_rate": 1.2816676640734091e-06,
      "loss": 1.3384,
      "mean_token_accuracy": 0.6992694318294526,
      "num_tokens": 4281157.0,
      "step": 9770
    },
    {
      "epoch": 2.926391382405745,
      "grad_norm": 20.75,
      "learning_rate": 1.2317973269499303e-06,
      "loss": 2.176,
      "mean_token_accuracy": 0.5774380654096604,
      "num_tokens": 4286810.0,
      "step": 9780
    },
    {
      "epoch": 2.9293836026331537,
      "grad_norm": 17.0,
      "learning_rate": 1.1819269898264511e-06,
      "loss": 1.7546,
      "mean_token_accuracy": 0.6253076940774918,
      "num_tokens": 4291078.0,
      "step": 9790
    },
    {
      "epoch": 2.9323758228605623,
      "grad_norm": 32.75,
      "learning_rate": 1.1320566527029724e-06,
      "loss": 1.8533,
      "mean_token_accuracy": 0.6021129012107849,
      "num_tokens": 4294902.0,
      "step": 9800
    },
    {
      "epoch": 2.9353680430879714,
      "grad_norm": 55.75,
      "learning_rate": 1.0821863155794934e-06,
      "loss": 1.7355,
      "mean_token_accuracy": 0.6272025167942047,
      "num_tokens": 4299523.0,
      "step": 9810
    },
    {
      "epoch": 2.93836026331538,
      "grad_norm": 55.75,
      "learning_rate": 1.0323159784560146e-06,
      "loss": 1.8426,
      "mean_token_accuracy": 0.6203331828117371,
      "num_tokens": 4305206.0,
      "step": 9820
    },
    {
      "epoch": 2.9413524835427887,
      "grad_norm": 33.5,
      "learning_rate": 9.824456413325354e-07,
      "loss": 2.3632,
      "mean_token_accuracy": 0.5340803563594818,
      "num_tokens": 4308887.0,
      "step": 9830
    },
    {
      "epoch": 2.9443447037701977,
      "grad_norm": 32.25,
      "learning_rate": 9.325753042090565e-07,
      "loss": 1.1291,
      "mean_token_accuracy": 0.7302363455295563,
      "num_tokens": 4312388.0,
      "step": 9840
    },
    {
      "epoch": 2.9473369239976064,
      "grad_norm": 41.0,
      "learning_rate": 8.827049670855776e-07,
      "loss": 1.5903,
      "mean_token_accuracy": 0.6539588749408722,
      "num_tokens": 4317350.0,
      "step": 9850
    },
    {
      "epoch": 2.950329144225015,
      "grad_norm": 24.25,
      "learning_rate": 8.328346299620985e-07,
      "loss": 1.9191,
      "mean_token_accuracy": 0.6088144451379776,
      "num_tokens": 4321927.0,
      "step": 9860
    },
    {
      "epoch": 2.9533213644524237,
      "grad_norm": 63.5,
      "learning_rate": 7.829642928386196e-07,
      "loss": 2.393,
      "mean_token_accuracy": 0.5242181986570358,
      "num_tokens": 4326242.0,
      "step": 9870
    },
    {
      "epoch": 2.9563135846798323,
      "grad_norm": 30.625,
      "learning_rate": 7.330939557151407e-07,
      "loss": 1.9247,
      "mean_token_accuracy": 0.5961373239755631,
      "num_tokens": 4330126.0,
      "step": 9880
    },
    {
      "epoch": 2.959305804907241,
      "grad_norm": 69.0,
      "learning_rate": 6.832236185916617e-07,
      "loss": 1.4818,
      "mean_token_accuracy": 0.6732258468866348,
      "num_tokens": 4333877.0,
      "step": 9890
    },
    {
      "epoch": 2.96229802513465,
      "grad_norm": 32.25,
      "learning_rate": 6.333532814681827e-07,
      "loss": 1.7031,
      "mean_token_accuracy": 0.6317979246377945,
      "num_tokens": 4338761.0,
      "step": 9900
    },
    {
      "epoch": 2.9652902453620587,
      "grad_norm": 41.5,
      "learning_rate": 5.834829443447037e-07,
      "loss": 2.1448,
      "mean_token_accuracy": 0.5809542566537857,
      "num_tokens": 4343273.0,
      "step": 9910
    },
    {
      "epoch": 2.9682824655894673,
      "grad_norm": 29.125,
      "learning_rate": 5.336126072212248e-07,
      "loss": 1.8744,
      "mean_token_accuracy": 0.6038303703069687,
      "num_tokens": 4348215.0,
      "step": 9920
    },
    {
      "epoch": 2.9712746858168764,
      "grad_norm": 21.25,
      "learning_rate": 4.837422700977458e-07,
      "loss": 1.7846,
      "mean_token_accuracy": 0.6139202952384949,
      "num_tokens": 4351873.0,
      "step": 9930
    },
    {
      "epoch": 2.974266906044285,
      "grad_norm": 26.375,
      "learning_rate": 4.3387193297426696e-07,
      "loss": 1.8021,
      "mean_token_accuracy": 0.6420397758483887,
      "num_tokens": 4355736.0,
      "step": 9940
    },
    {
      "epoch": 2.9772591262716936,
      "grad_norm": 29.375,
      "learning_rate": 3.8400159585078796e-07,
      "loss": 1.7598,
      "mean_token_accuracy": 0.6427461594343186,
      "num_tokens": 4360600.0,
      "step": 9950
    },
    {
      "epoch": 2.9802513464991023,
      "grad_norm": 19.375,
      "learning_rate": 3.34131258727309e-07,
      "loss": 1.9815,
      "mean_token_accuracy": 0.6036760896444321,
      "num_tokens": 4366937.0,
      "step": 9960
    },
    {
      "epoch": 2.983243566726511,
      "grad_norm": 50.5,
      "learning_rate": 2.8426092160383003e-07,
      "loss": 1.7909,
      "mean_token_accuracy": 0.6144560068845749,
      "num_tokens": 4370286.0,
      "step": 9970
    },
    {
      "epoch": 2.9862357869539196,
      "grad_norm": 28.375,
      "learning_rate": 2.3439058448035108e-07,
      "loss": 1.7324,
      "mean_token_accuracy": 0.6381353378295899,
      "num_tokens": 4374860.0,
      "step": 9980
    },
    {
      "epoch": 2.9892280071813286,
      "grad_norm": 20.625,
      "learning_rate": 1.8452024735687214e-07,
      "loss": 1.4505,
      "mean_token_accuracy": 0.6755415856838226,
      "num_tokens": 4379167.0,
      "step": 9990
    },
    {
      "epoch": 2.9922202274087373,
      "grad_norm": 31.625,
      "learning_rate": 1.346499102333932e-07,
      "loss": 1.9699,
      "mean_token_accuracy": 0.589902862906456,
      "num_tokens": 4382621.0,
      "step": 10000
    },
    {
      "epoch": 2.995212447636146,
      "grad_norm": 29.125,
      "learning_rate": 8.477957310991422e-08,
      "loss": 1.7263,
      "mean_token_accuracy": 0.6296306014060974,
      "num_tokens": 4386420.0,
      "step": 10010
    },
    {
      "epoch": 2.998204667863555,
      "grad_norm": 34.75,
      "learning_rate": 3.490923598643527e-08,
      "loss": 1.5662,
      "mean_token_accuracy": 0.6742969185113907,
      "num_tokens": 4390095.0,
      "step": 10020
    },
    {
      "epoch": 3.0,
      "eval_loss": 2.109537124633789,
      "eval_mean_token_accuracy": 0.5822922750464026,
      "eval_num_tokens": 4392732.0,
      "eval_runtime": 28.1736,
      "eval_samples_per_second": 14.837,
      "eval_steps_per_second": 1.881,
      "step": 10026
    }
  ],
  "logging_steps": 10,
  "max_steps": 10026,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.8809891524214784e+17,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}