2035 lines
54 KiB
JSON
2035 lines
54 KiB
JSON
{
|
|
"best_global_step": null,
|
|
"best_metric": null,
|
|
"best_model_checkpoint": null,
|
|
"epoch": 0.06597611664577423,
|
|
"eval_steps": 500,
|
|
"global_step": 1000,
|
|
"is_hyper_param_search": false,
|
|
"is_local_process_zero": true,
|
|
"is_world_process_zero": true,
|
|
"log_history": [
|
|
{
|
|
"entropy": 10.74258794784546,
|
|
"epoch": 0.00032988058322887113,
|
|
"grad_norm": 4.90625,
|
|
"learning_rate": 2e-06,
|
|
"loss": 10.7786,
|
|
"mean_token_accuracy": 0.0,
|
|
"num_tokens": 11020.0,
|
|
"step": 5
|
|
},
|
|
{
|
|
"entropy": 10.742591094970702,
|
|
"epoch": 0.0006597611664577423,
|
|
"grad_norm": 5.0,
|
|
"learning_rate": 4.5e-06,
|
|
"loss": 10.7947,
|
|
"mean_token_accuracy": 7.716049440205097e-05,
|
|
"num_tokens": 23411.0,
|
|
"step": 10
|
|
},
|
|
{
|
|
"entropy": 10.742575454711915,
|
|
"epoch": 0.0009896417496866134,
|
|
"grad_norm": 4.78125,
|
|
"learning_rate": 7e-06,
|
|
"loss": 10.7696,
|
|
"mean_token_accuracy": 0.0,
|
|
"num_tokens": 35592.0,
|
|
"step": 15
|
|
},
|
|
{
|
|
"entropy": 10.742617797851562,
|
|
"epoch": 0.0013195223329154845,
|
|
"grad_norm": 4.6875,
|
|
"learning_rate": 9.5e-06,
|
|
"loss": 10.6972,
|
|
"mean_token_accuracy": 0.0001264222548343241,
|
|
"num_tokens": 46417.0,
|
|
"step": 20
|
|
},
|
|
{
|
|
"entropy": 10.74265775680542,
|
|
"epoch": 0.0016494029161443558,
|
|
"grad_norm": 4.40625,
|
|
"learning_rate": 1.2e-05,
|
|
"loss": 10.6062,
|
|
"mean_token_accuracy": 0.0033247150597162543,
|
|
"num_tokens": 57261.0,
|
|
"step": 25
|
|
},
|
|
{
|
|
"entropy": 10.742412281036376,
|
|
"epoch": 0.001979283499373227,
|
|
"grad_norm": 3.578125,
|
|
"learning_rate": 1.4500000000000002e-05,
|
|
"loss": 10.4921,
|
|
"mean_token_accuracy": 0.023236998729407786,
|
|
"num_tokens": 68120.0,
|
|
"step": 30
|
|
},
|
|
{
|
|
"entropy": 10.741452598571778,
|
|
"epoch": 0.002309164082602098,
|
|
"grad_norm": 3.171875,
|
|
"learning_rate": 1.7000000000000003e-05,
|
|
"loss": 10.3687,
|
|
"mean_token_accuracy": 0.037138034030795096,
|
|
"num_tokens": 80384.0,
|
|
"step": 35
|
|
},
|
|
{
|
|
"entropy": 10.738858604431153,
|
|
"epoch": 0.002639044665830969,
|
|
"grad_norm": 2.4375,
|
|
"learning_rate": 1.95e-05,
|
|
"loss": 10.2112,
|
|
"mean_token_accuracy": 0.0408711526542902,
|
|
"num_tokens": 90416.0,
|
|
"step": 40
|
|
},
|
|
{
|
|
"entropy": 10.734870719909669,
|
|
"epoch": 0.0029689252490598403,
|
|
"grad_norm": 2.546875,
|
|
"learning_rate": 2.2e-05,
|
|
"loss": 10.1103,
|
|
"mean_token_accuracy": 0.037869381718337536,
|
|
"num_tokens": 101944.0,
|
|
"step": 45
|
|
},
|
|
{
|
|
"entropy": 10.73185043334961,
|
|
"epoch": 0.0032988058322887116,
|
|
"grad_norm": 2.0,
|
|
"learning_rate": 2.4500000000000003e-05,
|
|
"loss": 10.0493,
|
|
"mean_token_accuracy": 0.042053837701678275,
|
|
"num_tokens": 113439.0,
|
|
"step": 50
|
|
},
|
|
{
|
|
"entropy": 10.729621887207031,
|
|
"epoch": 0.0036286864155175825,
|
|
"grad_norm": 1.84375,
|
|
"learning_rate": 2.7e-05,
|
|
"loss": 9.9855,
|
|
"mean_token_accuracy": 0.039308933168649675,
|
|
"num_tokens": 124817.0,
|
|
"step": 55
|
|
},
|
|
{
|
|
"entropy": 10.72744541168213,
|
|
"epoch": 0.003958566998746454,
|
|
"grad_norm": 1.84375,
|
|
"learning_rate": 2.95e-05,
|
|
"loss": 9.9358,
|
|
"mean_token_accuracy": 0.036126206070184706,
|
|
"num_tokens": 136784.0,
|
|
"step": 60
|
|
},
|
|
{
|
|
"entropy": 10.726340103149415,
|
|
"epoch": 0.004288447581975325,
|
|
"grad_norm": 1.8359375,
|
|
"learning_rate": 3.2e-05,
|
|
"loss": 9.871,
|
|
"mean_token_accuracy": 0.037863926775753495,
|
|
"num_tokens": 148412.0,
|
|
"step": 65
|
|
},
|
|
{
|
|
"entropy": 10.724691486358642,
|
|
"epoch": 0.004618328165204196,
|
|
"grad_norm": 1.8359375,
|
|
"learning_rate": 3.4500000000000005e-05,
|
|
"loss": 9.8069,
|
|
"mean_token_accuracy": 0.03520041722804308,
|
|
"num_tokens": 159001.0,
|
|
"step": 70
|
|
},
|
|
{
|
|
"entropy": 10.722445201873779,
|
|
"epoch": 0.004948208748433067,
|
|
"grad_norm": 1.8046875,
|
|
"learning_rate": 3.7e-05,
|
|
"loss": 9.7519,
|
|
"mean_token_accuracy": 0.03713920433074236,
|
|
"num_tokens": 170301.0,
|
|
"step": 75
|
|
},
|
|
{
|
|
"entropy": 10.718492603302002,
|
|
"epoch": 0.005278089331661938,
|
|
"grad_norm": 1.6640625,
|
|
"learning_rate": 3.95e-05,
|
|
"loss": 9.6685,
|
|
"mean_token_accuracy": 0.03926718235015869,
|
|
"num_tokens": 180966.0,
|
|
"step": 80
|
|
},
|
|
{
|
|
"entropy": 10.711327266693115,
|
|
"epoch": 0.00560796991489081,
|
|
"grad_norm": 1.75,
|
|
"learning_rate": 4.2000000000000004e-05,
|
|
"loss": 9.6346,
|
|
"mean_token_accuracy": 0.04008835144340992,
|
|
"num_tokens": 192843.0,
|
|
"step": 85
|
|
},
|
|
{
|
|
"entropy": 10.702139282226563,
|
|
"epoch": 0.005937850498119681,
|
|
"grad_norm": 1.8125,
|
|
"learning_rate": 4.45e-05,
|
|
"loss": 9.5538,
|
|
"mean_token_accuracy": 0.03782005812972784,
|
|
"num_tokens": 203167.0,
|
|
"step": 90
|
|
},
|
|
{
|
|
"entropy": 10.690580654144288,
|
|
"epoch": 0.0062677310813485515,
|
|
"grad_norm": 1.6640625,
|
|
"learning_rate": 4.7000000000000004e-05,
|
|
"loss": 9.4665,
|
|
"mean_token_accuracy": 0.039251597598195076,
|
|
"num_tokens": 214704.0,
|
|
"step": 95
|
|
},
|
|
{
|
|
"entropy": 10.676863384246825,
|
|
"epoch": 0.006597611664577423,
|
|
"grad_norm": 1.671875,
|
|
"learning_rate": 4.9500000000000004e-05,
|
|
"loss": 9.3669,
|
|
"mean_token_accuracy": 0.04201584011316299,
|
|
"num_tokens": 224848.0,
|
|
"step": 100
|
|
},
|
|
{
|
|
"entropy": 10.663097476959228,
|
|
"epoch": 0.006927492247806294,
|
|
"grad_norm": 1.5625,
|
|
"learning_rate": 5.2e-05,
|
|
"loss": 9.2966,
|
|
"mean_token_accuracy": 0.046305606700479986,
|
|
"num_tokens": 236213.0,
|
|
"step": 105
|
|
},
|
|
{
|
|
"entropy": 10.64140272140503,
|
|
"epoch": 0.007257372831035165,
|
|
"grad_norm": 1.75,
|
|
"learning_rate": 5.45e-05,
|
|
"loss": 9.1314,
|
|
"mean_token_accuracy": 0.0530491016805172,
|
|
"num_tokens": 246918.0,
|
|
"step": 110
|
|
},
|
|
{
|
|
"entropy": 10.607359409332275,
|
|
"epoch": 0.007587253414264037,
|
|
"grad_norm": 1.8125,
|
|
"learning_rate": 5.7e-05,
|
|
"loss": 9.006,
|
|
"mean_token_accuracy": 0.04945136085152626,
|
|
"num_tokens": 256940.0,
|
|
"step": 115
|
|
},
|
|
{
|
|
"entropy": 10.56920051574707,
|
|
"epoch": 0.007917133997492908,
|
|
"grad_norm": 1.59375,
|
|
"learning_rate": 5.9499999999999996e-05,
|
|
"loss": 8.944,
|
|
"mean_token_accuracy": 0.05116630271077156,
|
|
"num_tokens": 267317.0,
|
|
"step": 120
|
|
},
|
|
{
|
|
"entropy": 10.51912956237793,
|
|
"epoch": 0.008247014580721778,
|
|
"grad_norm": 1.5625,
|
|
"learning_rate": 6.2e-05,
|
|
"loss": 8.8397,
|
|
"mean_token_accuracy": 0.04565899148583412,
|
|
"num_tokens": 277636.0,
|
|
"step": 125
|
|
},
|
|
{
|
|
"entropy": 10.449232959747315,
|
|
"epoch": 0.00857689516395065,
|
|
"grad_norm": 1.4453125,
|
|
"learning_rate": 6.450000000000001e-05,
|
|
"loss": 8.7503,
|
|
"mean_token_accuracy": 0.0457552932202816,
|
|
"num_tokens": 288733.0,
|
|
"step": 130
|
|
},
|
|
{
|
|
"entropy": 10.36309232711792,
|
|
"epoch": 0.008906775747179522,
|
|
"grad_norm": 1.375,
|
|
"learning_rate": 6.7e-05,
|
|
"loss": 8.6193,
|
|
"mean_token_accuracy": 0.04694342277944088,
|
|
"num_tokens": 300288.0,
|
|
"step": 135
|
|
},
|
|
{
|
|
"entropy": 10.278613567352295,
|
|
"epoch": 0.009236656330408393,
|
|
"grad_norm": 1.3984375,
|
|
"learning_rate": 6.950000000000001e-05,
|
|
"loss": 8.4825,
|
|
"mean_token_accuracy": 0.047248030081391335,
|
|
"num_tokens": 311399.0,
|
|
"step": 140
|
|
},
|
|
{
|
|
"entropy": 10.169164276123047,
|
|
"epoch": 0.009566536913637264,
|
|
"grad_norm": 1.453125,
|
|
"learning_rate": 7.2e-05,
|
|
"loss": 8.3299,
|
|
"mean_token_accuracy": 0.04917343035340309,
|
|
"num_tokens": 323432.0,
|
|
"step": 145
|
|
},
|
|
{
|
|
"entropy": 10.025929737091065,
|
|
"epoch": 0.009896417496866134,
|
|
"grad_norm": 1.3203125,
|
|
"learning_rate": 7.45e-05,
|
|
"loss": 8.2228,
|
|
"mean_token_accuracy": 0.05003512389957905,
|
|
"num_tokens": 334825.0,
|
|
"step": 150
|
|
},
|
|
{
|
|
"entropy": 9.882170677185059,
|
|
"epoch": 0.010226298080095005,
|
|
"grad_norm": 1.1953125,
|
|
"learning_rate": 7.7e-05,
|
|
"loss": 8.1172,
|
|
"mean_token_accuracy": 0.052342257276177404,
|
|
"num_tokens": 347005.0,
|
|
"step": 155
|
|
},
|
|
{
|
|
"entropy": 9.701091003417968,
|
|
"epoch": 0.010556178663323876,
|
|
"grad_norm": 1.140625,
|
|
"learning_rate": 7.950000000000001e-05,
|
|
"loss": 8.0662,
|
|
"mean_token_accuracy": 0.0490269985049963,
|
|
"num_tokens": 358362.0,
|
|
"step": 160
|
|
},
|
|
{
|
|
"entropy": 9.532548332214356,
|
|
"epoch": 0.010886059246552749,
|
|
"grad_norm": 1.09375,
|
|
"learning_rate": 8.2e-05,
|
|
"loss": 7.9874,
|
|
"mean_token_accuracy": 0.05272770524024963,
|
|
"num_tokens": 370076.0,
|
|
"step": 165
|
|
},
|
|
{
|
|
"entropy": 9.28233880996704,
|
|
"epoch": 0.01121593982978162,
|
|
"grad_norm": 0.94140625,
|
|
"learning_rate": 8.450000000000001e-05,
|
|
"loss": 7.8741,
|
|
"mean_token_accuracy": 0.054555711522698405,
|
|
"num_tokens": 381231.0,
|
|
"step": 170
|
|
},
|
|
{
|
|
"entropy": 9.080228996276855,
|
|
"epoch": 0.01154582041301049,
|
|
"grad_norm": 0.84375,
|
|
"learning_rate": 8.7e-05,
|
|
"loss": 7.7861,
|
|
"mean_token_accuracy": 0.05426693670451641,
|
|
"num_tokens": 392818.0,
|
|
"step": 175
|
|
},
|
|
{
|
|
"entropy": 8.890596771240235,
|
|
"epoch": 0.011875700996239361,
|
|
"grad_norm": 0.90234375,
|
|
"learning_rate": 8.95e-05,
|
|
"loss": 7.6905,
|
|
"mean_token_accuracy": 0.06160585395991802,
|
|
"num_tokens": 403013.0,
|
|
"step": 180
|
|
},
|
|
{
|
|
"entropy": 8.703719329833984,
|
|
"epoch": 0.012205581579468232,
|
|
"grad_norm": 1.3515625,
|
|
"learning_rate": 9.2e-05,
|
|
"loss": 7.6488,
|
|
"mean_token_accuracy": 0.05801261067390442,
|
|
"num_tokens": 415665.0,
|
|
"step": 185
|
|
},
|
|
{
|
|
"entropy": 8.534394073486329,
|
|
"epoch": 0.012535462162697103,
|
|
"grad_norm": 0.7890625,
|
|
"learning_rate": 9.45e-05,
|
|
"loss": 7.6955,
|
|
"mean_token_accuracy": 0.060597951710224154,
|
|
"num_tokens": 427732.0,
|
|
"step": 190
|
|
},
|
|
{
|
|
"entropy": 8.45839786529541,
|
|
"epoch": 0.012865342745925976,
|
|
"grad_norm": 1.0,
|
|
"learning_rate": 9.7e-05,
|
|
"loss": 7.5648,
|
|
"mean_token_accuracy": 0.06318297423422337,
|
|
"num_tokens": 438081.0,
|
|
"step": 195
|
|
},
|
|
{
|
|
"entropy": 8.366796493530273,
|
|
"epoch": 0.013195223329154846,
|
|
"grad_norm": 0.98828125,
|
|
"learning_rate": 9.95e-05,
|
|
"loss": 7.5754,
|
|
"mean_token_accuracy": 0.06246511675417423,
|
|
"num_tokens": 449592.0,
|
|
"step": 200
|
|
},
|
|
{
|
|
"entropy": 8.335751819610596,
|
|
"epoch": 0.013525103912383717,
|
|
"grad_norm": 0.85546875,
|
|
"learning_rate": 0.000102,
|
|
"loss": 7.5236,
|
|
"mean_token_accuracy": 0.06291199699044228,
|
|
"num_tokens": 459192.0,
|
|
"step": 205
|
|
},
|
|
{
|
|
"entropy": 8.298966217041016,
|
|
"epoch": 0.013854984495612588,
|
|
"grad_norm": 0.84765625,
|
|
"learning_rate": 0.00010449999999999999,
|
|
"loss": 7.5824,
|
|
"mean_token_accuracy": 0.057720037549734114,
|
|
"num_tokens": 469973.0,
|
|
"step": 210
|
|
},
|
|
{
|
|
"entropy": 8.246175575256348,
|
|
"epoch": 0.014184865078841459,
|
|
"grad_norm": 0.88671875,
|
|
"learning_rate": 0.000107,
|
|
"loss": 7.4915,
|
|
"mean_token_accuracy": 0.06254454962909221,
|
|
"num_tokens": 480690.0,
|
|
"step": 215
|
|
},
|
|
{
|
|
"entropy": 8.183988761901855,
|
|
"epoch": 0.01451474566207033,
|
|
"grad_norm": 1.109375,
|
|
"learning_rate": 0.0001095,
|
|
"loss": 7.4713,
|
|
"mean_token_accuracy": 0.0652155488729477,
|
|
"num_tokens": 493081.0,
|
|
"step": 220
|
|
},
|
|
{
|
|
"entropy": 8.148521518707275,
|
|
"epoch": 0.014844626245299202,
|
|
"grad_norm": 0.77734375,
|
|
"learning_rate": 0.000112,
|
|
"loss": 7.4634,
|
|
"mean_token_accuracy": 0.07105538100004197,
|
|
"num_tokens": 504880.0,
|
|
"step": 225
|
|
},
|
|
{
|
|
"entropy": 8.087314224243164,
|
|
"epoch": 0.015174506828528073,
|
|
"grad_norm": 0.85546875,
|
|
"learning_rate": 0.0001145,
|
|
"loss": 7.3533,
|
|
"mean_token_accuracy": 0.07126509137451649,
|
|
"num_tokens": 515151.0,
|
|
"step": 230
|
|
},
|
|
{
|
|
"entropy": 8.10634479522705,
|
|
"epoch": 0.015504387411756944,
|
|
"grad_norm": 0.9375,
|
|
"learning_rate": 0.00011700000000000001,
|
|
"loss": 7.4316,
|
|
"mean_token_accuracy": 0.06604659669101239,
|
|
"num_tokens": 526547.0,
|
|
"step": 235
|
|
},
|
|
{
|
|
"entropy": 8.09265594482422,
|
|
"epoch": 0.015834267994985815,
|
|
"grad_norm": 0.92578125,
|
|
"learning_rate": 0.00011949999999999999,
|
|
"loss": 7.4037,
|
|
"mean_token_accuracy": 0.06221759617328644,
|
|
"num_tokens": 539456.0,
|
|
"step": 240
|
|
},
|
|
{
|
|
"entropy": 8.080215978622437,
|
|
"epoch": 0.016164148578214688,
|
|
"grad_norm": 1.5,
|
|
"learning_rate": 0.000122,
|
|
"loss": 7.513,
|
|
"mean_token_accuracy": 0.06819579191505909,
|
|
"num_tokens": 553097.0,
|
|
"step": 245
|
|
},
|
|
{
|
|
"entropy": 8.054958295822143,
|
|
"epoch": 0.016494029161443557,
|
|
"grad_norm": 1.0,
|
|
"learning_rate": 0.0001245,
|
|
"loss": 7.3445,
|
|
"mean_token_accuracy": 0.06898633353412151,
|
|
"num_tokens": 563352.0,
|
|
"step": 250
|
|
},
|
|
{
|
|
"entropy": 8.045195722579956,
|
|
"epoch": 0.01682390974467243,
|
|
"grad_norm": 0.76171875,
|
|
"learning_rate": 0.000127,
|
|
"loss": 7.3903,
|
|
"mean_token_accuracy": 0.07358907461166382,
|
|
"num_tokens": 576709.0,
|
|
"step": 255
|
|
},
|
|
{
|
|
"entropy": 8.005211210250854,
|
|
"epoch": 0.0171537903279013,
|
|
"grad_norm": 0.984375,
|
|
"learning_rate": 0.0001295,
|
|
"loss": 7.351,
|
|
"mean_token_accuracy": 0.07183001339435577,
|
|
"num_tokens": 587943.0,
|
|
"step": 260
|
|
},
|
|
{
|
|
"entropy": 7.958833646774292,
|
|
"epoch": 0.01748367091113017,
|
|
"grad_norm": 0.9765625,
|
|
"learning_rate": 0.000132,
|
|
"loss": 7.3205,
|
|
"mean_token_accuracy": 0.07523220106959343,
|
|
"num_tokens": 598751.0,
|
|
"step": 265
|
|
},
|
|
{
|
|
"entropy": 7.940512800216675,
|
|
"epoch": 0.017813551494359044,
|
|
"grad_norm": 1.0234375,
|
|
"learning_rate": 0.00013450000000000002,
|
|
"loss": 7.3491,
|
|
"mean_token_accuracy": 0.06923789866268634,
|
|
"num_tokens": 610403.0,
|
|
"step": 270
|
|
},
|
|
{
|
|
"entropy": 7.9217428207397464,
|
|
"epoch": 0.018143432077587913,
|
|
"grad_norm": 1.0703125,
|
|
"learning_rate": 0.00013700000000000002,
|
|
"loss": 7.3099,
|
|
"mean_token_accuracy": 0.06510985493659974,
|
|
"num_tokens": 620524.0,
|
|
"step": 275
|
|
},
|
|
{
|
|
"entropy": 7.901014757156372,
|
|
"epoch": 0.018473312660816785,
|
|
"grad_norm": 1.25,
|
|
"learning_rate": 0.0001395,
|
|
"loss": 7.2537,
|
|
"mean_token_accuracy": 0.07578651644289494,
|
|
"num_tokens": 631821.0,
|
|
"step": 280
|
|
},
|
|
{
|
|
"entropy": 7.807597589492798,
|
|
"epoch": 0.018803193244045655,
|
|
"grad_norm": 0.9453125,
|
|
"learning_rate": 0.00014199999999999998,
|
|
"loss": 7.3151,
|
|
"mean_token_accuracy": 0.0787085548043251,
|
|
"num_tokens": 643721.0,
|
|
"step": 285
|
|
},
|
|
{
|
|
"entropy": 7.939801359176636,
|
|
"epoch": 0.019133073827274527,
|
|
"grad_norm": 1.1328125,
|
|
"learning_rate": 0.0001445,
|
|
"loss": 7.3286,
|
|
"mean_token_accuracy": 0.06894742250442505,
|
|
"num_tokens": 654474.0,
|
|
"step": 290
|
|
},
|
|
{
|
|
"entropy": 7.8524211883544925,
|
|
"epoch": 0.019462954410503396,
|
|
"grad_norm": 1.0234375,
|
|
"learning_rate": 0.000147,
|
|
"loss": 7.2327,
|
|
"mean_token_accuracy": 0.07615564242005349,
|
|
"num_tokens": 664918.0,
|
|
"step": 295
|
|
},
|
|
{
|
|
"entropy": 7.847302389144898,
|
|
"epoch": 0.01979283499373227,
|
|
"grad_norm": 0.953125,
|
|
"learning_rate": 0.0001495,
|
|
"loss": 7.2281,
|
|
"mean_token_accuracy": 0.07257858999073505,
|
|
"num_tokens": 674533.0,
|
|
"step": 300
|
|
},
|
|
{
|
|
"entropy": 7.838640356063843,
|
|
"epoch": 0.02012271557696114,
|
|
"grad_norm": 0.93359375,
|
|
"learning_rate": 0.000152,
|
|
"loss": 7.2107,
|
|
"mean_token_accuracy": 0.07706848718225956,
|
|
"num_tokens": 685394.0,
|
|
"step": 305
|
|
},
|
|
{
|
|
"entropy": 7.811349105834961,
|
|
"epoch": 0.02045259616019001,
|
|
"grad_norm": 0.98046875,
|
|
"learning_rate": 0.00015450000000000001,
|
|
"loss": 7.2874,
|
|
"mean_token_accuracy": 0.07056003175675869,
|
|
"num_tokens": 698698.0,
|
|
"step": 310
|
|
},
|
|
{
|
|
"entropy": 7.894655036926269,
|
|
"epoch": 0.020782476743418883,
|
|
"grad_norm": 0.7734375,
|
|
"learning_rate": 0.000157,
|
|
"loss": 7.2127,
|
|
"mean_token_accuracy": 0.07129954807460308,
|
|
"num_tokens": 709874.0,
|
|
"step": 315
|
|
},
|
|
{
|
|
"entropy": 7.816462707519531,
|
|
"epoch": 0.021112357326647752,
|
|
"grad_norm": 0.9296875,
|
|
"learning_rate": 0.0001595,
|
|
"loss": 7.2176,
|
|
"mean_token_accuracy": 0.07924212142825127,
|
|
"num_tokens": 720121.0,
|
|
"step": 320
|
|
},
|
|
{
|
|
"entropy": 7.711975002288819,
|
|
"epoch": 0.021442237909876625,
|
|
"grad_norm": 0.83984375,
|
|
"learning_rate": 0.000162,
|
|
"loss": 7.1539,
|
|
"mean_token_accuracy": 0.07983077466487884,
|
|
"num_tokens": 731027.0,
|
|
"step": 325
|
|
},
|
|
{
|
|
"entropy": 7.772567462921143,
|
|
"epoch": 0.021772118493105497,
|
|
"grad_norm": 0.9296875,
|
|
"learning_rate": 0.00016450000000000001,
|
|
"loss": 7.0939,
|
|
"mean_token_accuracy": 0.07682515569031238,
|
|
"num_tokens": 741589.0,
|
|
"step": 330
|
|
},
|
|
{
|
|
"entropy": 7.709491157531739,
|
|
"epoch": 0.022101999076334367,
|
|
"grad_norm": 0.8984375,
|
|
"learning_rate": 0.00016700000000000002,
|
|
"loss": 7.1547,
|
|
"mean_token_accuracy": 0.0771139271557331,
|
|
"num_tokens": 752199.0,
|
|
"step": 335
|
|
},
|
|
{
|
|
"entropy": 7.7034478187561035,
|
|
"epoch": 0.02243187965956324,
|
|
"grad_norm": 1.015625,
|
|
"learning_rate": 0.00016950000000000003,
|
|
"loss": 7.1658,
|
|
"mean_token_accuracy": 0.07423071078956127,
|
|
"num_tokens": 762272.0,
|
|
"step": 340
|
|
},
|
|
{
|
|
"entropy": 7.713706302642822,
|
|
"epoch": 0.02276176024279211,
|
|
"grad_norm": 0.8125,
|
|
"learning_rate": 0.00017199999999999998,
|
|
"loss": 7.1245,
|
|
"mean_token_accuracy": 0.07617058306932449,
|
|
"num_tokens": 773985.0,
|
|
"step": 345
|
|
},
|
|
{
|
|
"entropy": 7.666186141967773,
|
|
"epoch": 0.02309164082602098,
|
|
"grad_norm": 0.95703125,
|
|
"learning_rate": 0.00017449999999999999,
|
|
"loss": 7.1073,
|
|
"mean_token_accuracy": 0.07593936771154404,
|
|
"num_tokens": 785112.0,
|
|
"step": 350
|
|
},
|
|
{
|
|
"entropy": 7.651518201828003,
|
|
"epoch": 0.02342152140924985,
|
|
"grad_norm": 0.9296875,
|
|
"learning_rate": 0.000177,
|
|
"loss": 7.0753,
|
|
"mean_token_accuracy": 0.07759405225515366,
|
|
"num_tokens": 795541.0,
|
|
"step": 355
|
|
},
|
|
{
|
|
"entropy": 7.735073232650757,
|
|
"epoch": 0.023751401992478723,
|
|
"grad_norm": 0.97265625,
|
|
"learning_rate": 0.0001795,
|
|
"loss": 7.1378,
|
|
"mean_token_accuracy": 0.07670942842960357,
|
|
"num_tokens": 807143.0,
|
|
"step": 360
|
|
},
|
|
{
|
|
"entropy": 7.705486726760864,
|
|
"epoch": 0.024081282575707595,
|
|
"grad_norm": 1.03125,
|
|
"learning_rate": 0.000182,
|
|
"loss": 7.1476,
|
|
"mean_token_accuracy": 0.07829145677387714,
|
|
"num_tokens": 819095.0,
|
|
"step": 365
|
|
},
|
|
{
|
|
"entropy": 7.56443305015564,
|
|
"epoch": 0.024411163158936464,
|
|
"grad_norm": 1.234375,
|
|
"learning_rate": 0.0001845,
|
|
"loss": 7.0475,
|
|
"mean_token_accuracy": 0.08632016405463219,
|
|
"num_tokens": 830380.0,
|
|
"step": 370
|
|
},
|
|
{
|
|
"entropy": 7.582523441314697,
|
|
"epoch": 0.024741043742165337,
|
|
"grad_norm": 1.0234375,
|
|
"learning_rate": 0.000187,
|
|
"loss": 7.0863,
|
|
"mean_token_accuracy": 0.07546742670238019,
|
|
"num_tokens": 841405.0,
|
|
"step": 375
|
|
},
|
|
{
|
|
"entropy": 7.61822395324707,
|
|
"epoch": 0.025070924325394206,
|
|
"grad_norm": 0.8984375,
|
|
"learning_rate": 0.0001895,
|
|
"loss": 7.0332,
|
|
"mean_token_accuracy": 0.07516690790653228,
|
|
"num_tokens": 852529.0,
|
|
"step": 380
|
|
},
|
|
{
|
|
"entropy": 7.600083303451538,
|
|
"epoch": 0.02540080490862308,
|
|
"grad_norm": 1.0546875,
|
|
"learning_rate": 0.000192,
|
|
"loss": 7.0249,
|
|
"mean_token_accuracy": 0.08358373120427132,
|
|
"num_tokens": 865100.0,
|
|
"step": 385
|
|
},
|
|
{
|
|
"entropy": 7.598215484619141,
|
|
"epoch": 0.02573068549185195,
|
|
"grad_norm": 1.2578125,
|
|
"learning_rate": 0.0001945,
|
|
"loss": 7.0693,
|
|
"mean_token_accuracy": 0.08309284597635269,
|
|
"num_tokens": 876385.0,
|
|
"step": 390
|
|
},
|
|
{
|
|
"entropy": 7.477734994888306,
|
|
"epoch": 0.02606056607508082,
|
|
"grad_norm": 0.90234375,
|
|
"learning_rate": 0.00019700000000000002,
|
|
"loss": 7.0131,
|
|
"mean_token_accuracy": 0.07874878086149692,
|
|
"num_tokens": 887207.0,
|
|
"step": 395
|
|
},
|
|
{
|
|
"entropy": 7.606724977493286,
|
|
"epoch": 0.026390446658309693,
|
|
"grad_norm": 0.8203125,
|
|
"learning_rate": 0.00019950000000000002,
|
|
"loss": 6.9847,
|
|
"mean_token_accuracy": 0.0870030753314495,
|
|
"num_tokens": 900124.0,
|
|
"step": 400
|
|
},
|
|
{
|
|
"entropy": 7.569502305984497,
|
|
"epoch": 0.026720327241538562,
|
|
"grad_norm": 1.046875,
|
|
"learning_rate": 0.000202,
|
|
"loss": 6.9792,
|
|
"mean_token_accuracy": 0.0829981118440628,
|
|
"num_tokens": 911177.0,
|
|
"step": 405
|
|
},
|
|
{
|
|
"entropy": 7.464511060714722,
|
|
"epoch": 0.027050207824767435,
|
|
"grad_norm": 1.15625,
|
|
"learning_rate": 0.00020449999999999998,
|
|
"loss": 6.9924,
|
|
"mean_token_accuracy": 0.08479403480887412,
|
|
"num_tokens": 922524.0,
|
|
"step": 410
|
|
},
|
|
{
|
|
"entropy": 7.63841438293457,
|
|
"epoch": 0.027380088407996304,
|
|
"grad_norm": 0.9921875,
|
|
"learning_rate": 0.000207,
|
|
"loss": 7.0118,
|
|
"mean_token_accuracy": 0.07898868545889855,
|
|
"num_tokens": 933484.0,
|
|
"step": 415
|
|
},
|
|
{
|
|
"entropy": 7.524638557434082,
|
|
"epoch": 0.027709968991225176,
|
|
"grad_norm": 0.9375,
|
|
"learning_rate": 0.0002095,
|
|
"loss": 7.0145,
|
|
"mean_token_accuracy": 0.07973532900214195,
|
|
"num_tokens": 944996.0,
|
|
"step": 420
|
|
},
|
|
{
|
|
"entropy": 7.463073682785034,
|
|
"epoch": 0.02803984957445405,
|
|
"grad_norm": 1.0546875,
|
|
"learning_rate": 0.000212,
|
|
"loss": 6.9025,
|
|
"mean_token_accuracy": 0.08155579566955566,
|
|
"num_tokens": 955943.0,
|
|
"step": 425
|
|
},
|
|
{
|
|
"entropy": 7.434240293502808,
|
|
"epoch": 0.028369730157682918,
|
|
"grad_norm": 0.921875,
|
|
"learning_rate": 0.0002145,
|
|
"loss": 6.9737,
|
|
"mean_token_accuracy": 0.08236184045672416,
|
|
"num_tokens": 967721.0,
|
|
"step": 430
|
|
},
|
|
{
|
|
"entropy": 7.47903618812561,
|
|
"epoch": 0.02869961074091179,
|
|
"grad_norm": 1.09375,
|
|
"learning_rate": 0.00021700000000000002,
|
|
"loss": 6.883,
|
|
"mean_token_accuracy": 0.08318156823515892,
|
|
"num_tokens": 978976.0,
|
|
"step": 435
|
|
},
|
|
{
|
|
"entropy": 7.432163333892822,
|
|
"epoch": 0.02902949132414066,
|
|
"grad_norm": 1.0,
|
|
"learning_rate": 0.0002195,
|
|
"loss": 6.9556,
|
|
"mean_token_accuracy": 0.08390010446310044,
|
|
"num_tokens": 990703.0,
|
|
"step": 440
|
|
},
|
|
{
|
|
"entropy": 7.600602722167968,
|
|
"epoch": 0.029359371907369532,
|
|
"grad_norm": 1.1328125,
|
|
"learning_rate": 0.000222,
|
|
"loss": 7.0618,
|
|
"mean_token_accuracy": 0.08305796831846238,
|
|
"num_tokens": 1002011.0,
|
|
"step": 445
|
|
},
|
|
{
|
|
"entropy": 7.397836685180664,
|
|
"epoch": 0.029689252490598405,
|
|
"grad_norm": 0.94140625,
|
|
"learning_rate": 0.0002245,
|
|
"loss": 6.8248,
|
|
"mean_token_accuracy": 0.09162046760320663,
|
|
"num_tokens": 1011843.0,
|
|
"step": 450
|
|
},
|
|
{
|
|
"entropy": 7.350779676437378,
|
|
"epoch": 0.030019133073827274,
|
|
"grad_norm": 0.9453125,
|
|
"learning_rate": 0.00022700000000000002,
|
|
"loss": 6.8199,
|
|
"mean_token_accuracy": 0.0903800368309021,
|
|
"num_tokens": 1023472.0,
|
|
"step": 455
|
|
},
|
|
{
|
|
"entropy": 7.454651403427124,
|
|
"epoch": 0.030349013657056147,
|
|
"grad_norm": 1.09375,
|
|
"learning_rate": 0.00022950000000000002,
|
|
"loss": 6.9027,
|
|
"mean_token_accuracy": 0.08405964896082878,
|
|
"num_tokens": 1034106.0,
|
|
"step": 460
|
|
},
|
|
{
|
|
"entropy": 7.382388973236084,
|
|
"epoch": 0.030678894240285016,
|
|
"grad_norm": 0.93359375,
|
|
"learning_rate": 0.00023200000000000003,
|
|
"loss": 6.8711,
|
|
"mean_token_accuracy": 0.09321296438574792,
|
|
"num_tokens": 1045877.0,
|
|
"step": 465
|
|
},
|
|
{
|
|
"entropy": 7.400975370407105,
|
|
"epoch": 0.03100877482351389,
|
|
"grad_norm": 0.8984375,
|
|
"learning_rate": 0.00023449999999999998,
|
|
"loss": 6.9807,
|
|
"mean_token_accuracy": 0.0839313305914402,
|
|
"num_tokens": 1058738.0,
|
|
"step": 470
|
|
},
|
|
{
|
|
"entropy": 7.334363412857056,
|
|
"epoch": 0.03133865540674276,
|
|
"grad_norm": 1.1796875,
|
|
"learning_rate": 0.000237,
|
|
"loss": 6.8072,
|
|
"mean_token_accuracy": 0.08514157012104988,
|
|
"num_tokens": 1069415.0,
|
|
"step": 475
|
|
},
|
|
{
|
|
"entropy": 7.370652484893799,
|
|
"epoch": 0.03166853598997163,
|
|
"grad_norm": 1.03125,
|
|
"learning_rate": 0.0002395,
|
|
"loss": 6.8433,
|
|
"mean_token_accuracy": 0.09043429493904113,
|
|
"num_tokens": 1079979.0,
|
|
"step": 480
|
|
},
|
|
{
|
|
"entropy": 7.424137449264526,
|
|
"epoch": 0.0319984165732005,
|
|
"grad_norm": 1.0546875,
|
|
"learning_rate": 0.000242,
|
|
"loss": 6.937,
|
|
"mean_token_accuracy": 0.08223749808967114,
|
|
"num_tokens": 1090946.0,
|
|
"step": 485
|
|
},
|
|
{
|
|
"entropy": 7.299629068374633,
|
|
"epoch": 0.032328297156429375,
|
|
"grad_norm": 1.0,
|
|
"learning_rate": 0.0002445,
|
|
"loss": 6.7989,
|
|
"mean_token_accuracy": 0.090982224047184,
|
|
"num_tokens": 1100926.0,
|
|
"step": 490
|
|
},
|
|
{
|
|
"entropy": 7.324662590026856,
|
|
"epoch": 0.03265817773965824,
|
|
"grad_norm": 0.921875,
|
|
"learning_rate": 0.000247,
|
|
"loss": 6.8114,
|
|
"mean_token_accuracy": 0.08763068690896034,
|
|
"num_tokens": 1112169.0,
|
|
"step": 495
|
|
},
|
|
{
|
|
"entropy": 7.444135046005249,
|
|
"epoch": 0.032988058322887114,
|
|
"grad_norm": 1.1015625,
|
|
"learning_rate": 0.0002495,
|
|
"loss": 6.9175,
|
|
"mean_token_accuracy": 0.08510688915848733,
|
|
"num_tokens": 1123425.0,
|
|
"step": 500
|
|
},
|
|
{
|
|
"entropy": 7.269032144546509,
|
|
"epoch": 0.033317938906115986,
|
|
"grad_norm": 0.99609375,
|
|
"learning_rate": 0.000252,
|
|
"loss": 6.7995,
|
|
"mean_token_accuracy": 0.08869035243988037,
|
|
"num_tokens": 1134918.0,
|
|
"step": 505
|
|
},
|
|
{
|
|
"entropy": 7.33376898765564,
|
|
"epoch": 0.03364781948934486,
|
|
"grad_norm": 1.0703125,
|
|
"learning_rate": 0.0002545,
|
|
"loss": 6.7655,
|
|
"mean_token_accuracy": 0.09362135753035546,
|
|
"num_tokens": 1145710.0,
|
|
"step": 510
|
|
},
|
|
{
|
|
"entropy": 7.278433465957642,
|
|
"epoch": 0.03397770007257373,
|
|
"grad_norm": 0.9609375,
|
|
"learning_rate": 0.000257,
|
|
"loss": 6.8243,
|
|
"mean_token_accuracy": 0.08989340886473655,
|
|
"num_tokens": 1156386.0,
|
|
"step": 515
|
|
},
|
|
{
|
|
"entropy": 7.300722551345825,
|
|
"epoch": 0.0343075806558026,
|
|
"grad_norm": 1.140625,
|
|
"learning_rate": 0.0002595,
|
|
"loss": 6.7518,
|
|
"mean_token_accuracy": 0.0956637755036354,
|
|
"num_tokens": 1167641.0,
|
|
"step": 520
|
|
},
|
|
{
|
|
"entropy": 7.245179462432861,
|
|
"epoch": 0.03463746123903147,
|
|
"grad_norm": 1.265625,
|
|
"learning_rate": 0.000262,
|
|
"loss": 6.8259,
|
|
"mean_token_accuracy": 0.08638799116015435,
|
|
"num_tokens": 1178716.0,
|
|
"step": 525
|
|
},
|
|
{
|
|
"entropy": 7.320114231109619,
|
|
"epoch": 0.03496734182226034,
|
|
"grad_norm": 0.9921875,
|
|
"learning_rate": 0.00026450000000000003,
|
|
"loss": 6.8509,
|
|
"mean_token_accuracy": 0.08957468420267105,
|
|
"num_tokens": 1190717.0,
|
|
"step": 530
|
|
},
|
|
{
|
|
"entropy": 7.258654689788818,
|
|
"epoch": 0.035297222405489215,
|
|
"grad_norm": 1.1796875,
|
|
"learning_rate": 0.00026700000000000004,
|
|
"loss": 6.7686,
|
|
"mean_token_accuracy": 0.09318572208285332,
|
|
"num_tokens": 1201381.0,
|
|
"step": 535
|
|
},
|
|
{
|
|
"entropy": 7.2403817653656,
|
|
"epoch": 0.03562710298871809,
|
|
"grad_norm": 0.94921875,
|
|
"learning_rate": 0.00026950000000000005,
|
|
"loss": 6.7925,
|
|
"mean_token_accuracy": 0.08640593588352204,
|
|
"num_tokens": 1212903.0,
|
|
"step": 540
|
|
},
|
|
{
|
|
"entropy": 7.328101110458374,
|
|
"epoch": 0.03595698357194695,
|
|
"grad_norm": 0.93359375,
|
|
"learning_rate": 0.00027200000000000005,
|
|
"loss": 6.7223,
|
|
"mean_token_accuracy": 0.09287714958190918,
|
|
"num_tokens": 1224990.0,
|
|
"step": 545
|
|
},
|
|
{
|
|
"entropy": 7.165910434722901,
|
|
"epoch": 0.036286864155175826,
|
|
"grad_norm": 0.96875,
|
|
"learning_rate": 0.0002745,
|
|
"loss": 6.7881,
|
|
"mean_token_accuracy": 0.08515067920088767,
|
|
"num_tokens": 1236331.0,
|
|
"step": 550
|
|
},
|
|
{
|
|
"entropy": 7.1720420837402346,
|
|
"epoch": 0.0366167447384047,
|
|
"grad_norm": 0.9921875,
|
|
"learning_rate": 0.000277,
|
|
"loss": 6.6695,
|
|
"mean_token_accuracy": 0.09367566034197808,
|
|
"num_tokens": 1248337.0,
|
|
"step": 555
|
|
},
|
|
{
|
|
"entropy": 7.139011192321777,
|
|
"epoch": 0.03694662532163357,
|
|
"grad_norm": 1.015625,
|
|
"learning_rate": 0.0002795,
|
|
"loss": 6.6922,
|
|
"mean_token_accuracy": 0.09301456287503243,
|
|
"num_tokens": 1261063.0,
|
|
"step": 560
|
|
},
|
|
{
|
|
"entropy": 7.250580167770385,
|
|
"epoch": 0.037276505904862436,
|
|
"grad_norm": 1.0390625,
|
|
"learning_rate": 0.00028199999999999997,
|
|
"loss": 6.7142,
|
|
"mean_token_accuracy": 0.0939611129462719,
|
|
"num_tokens": 1272890.0,
|
|
"step": 565
|
|
},
|
|
{
|
|
"entropy": 7.252413749694824,
|
|
"epoch": 0.03760638648809131,
|
|
"grad_norm": 1.015625,
|
|
"learning_rate": 0.0002845,
|
|
"loss": 6.7064,
|
|
"mean_token_accuracy": 0.09456135258078575,
|
|
"num_tokens": 1283551.0,
|
|
"step": 570
|
|
},
|
|
{
|
|
"entropy": 7.06690616607666,
|
|
"epoch": 0.03793626707132018,
|
|
"grad_norm": 0.97265625,
|
|
"learning_rate": 0.000287,
|
|
"loss": 6.6083,
|
|
"mean_token_accuracy": 0.09540683031082153,
|
|
"num_tokens": 1294800.0,
|
|
"step": 575
|
|
},
|
|
{
|
|
"entropy": 7.184878826141357,
|
|
"epoch": 0.038266147654549054,
|
|
"grad_norm": 0.953125,
|
|
"learning_rate": 0.0002895,
|
|
"loss": 6.7872,
|
|
"mean_token_accuracy": 0.0963341936469078,
|
|
"num_tokens": 1307045.0,
|
|
"step": 580
|
|
},
|
|
{
|
|
"entropy": 7.174728441238403,
|
|
"epoch": 0.03859602823777793,
|
|
"grad_norm": 1.2421875,
|
|
"learning_rate": 0.000292,
|
|
"loss": 6.7593,
|
|
"mean_token_accuracy": 0.08541202917695045,
|
|
"num_tokens": 1319111.0,
|
|
"step": 585
|
|
},
|
|
{
|
|
"entropy": 7.226782894134521,
|
|
"epoch": 0.03892590882100679,
|
|
"grad_norm": 1.125,
|
|
"learning_rate": 0.0002945,
|
|
"loss": 6.6702,
|
|
"mean_token_accuracy": 0.09444425702095031,
|
|
"num_tokens": 1330234.0,
|
|
"step": 590
|
|
},
|
|
{
|
|
"entropy": 7.027184295654297,
|
|
"epoch": 0.039255789404235665,
|
|
"grad_norm": 0.94140625,
|
|
"learning_rate": 0.000297,
|
|
"loss": 6.7098,
|
|
"mean_token_accuracy": 0.0925945483148098,
|
|
"num_tokens": 1341287.0,
|
|
"step": 595
|
|
},
|
|
{
|
|
"entropy": 7.157588958740234,
|
|
"epoch": 0.03958566998746454,
|
|
"grad_norm": 0.9296875,
|
|
"learning_rate": 0.0002995,
|
|
"loss": 6.6625,
|
|
"mean_token_accuracy": 0.09391765892505646,
|
|
"num_tokens": 1354502.0,
|
|
"step": 600
|
|
},
|
|
{
|
|
"entropy": 7.0550346851348875,
|
|
"epoch": 0.03991555057069341,
|
|
"grad_norm": 1.03125,
|
|
"learning_rate": 0.000302,
|
|
"loss": 6.5588,
|
|
"mean_token_accuracy": 0.09691971093416214,
|
|
"num_tokens": 1364700.0,
|
|
"step": 605
|
|
},
|
|
{
|
|
"entropy": 7.096816539764404,
|
|
"epoch": 0.04024543115392228,
|
|
"grad_norm": 1.140625,
|
|
"learning_rate": 0.0003045,
|
|
"loss": 6.5775,
|
|
"mean_token_accuracy": 0.10072705522179604,
|
|
"num_tokens": 1375042.0,
|
|
"step": 610
|
|
},
|
|
{
|
|
"entropy": 6.993506813049317,
|
|
"epoch": 0.04057531173715115,
|
|
"grad_norm": 0.9765625,
|
|
"learning_rate": 0.000307,
|
|
"loss": 6.7,
|
|
"mean_token_accuracy": 0.08952248021960259,
|
|
"num_tokens": 1386782.0,
|
|
"step": 615
|
|
},
|
|
{
|
|
"entropy": 7.189109134674072,
|
|
"epoch": 0.04090519232038002,
|
|
"grad_norm": 0.89453125,
|
|
"learning_rate": 0.0003095,
|
|
"loss": 6.7194,
|
|
"mean_token_accuracy": 0.09634155109524727,
|
|
"num_tokens": 1397502.0,
|
|
"step": 620
|
|
},
|
|
{
|
|
"entropy": 7.136458206176758,
|
|
"epoch": 0.041235072903608894,
|
|
"grad_norm": 1.0234375,
|
|
"learning_rate": 0.000312,
|
|
"loss": 6.7027,
|
|
"mean_token_accuracy": 0.0920153796672821,
|
|
"num_tokens": 1408291.0,
|
|
"step": 625
|
|
},
|
|
{
|
|
"entropy": 6.90060567855835,
|
|
"epoch": 0.041564953486837766,
|
|
"grad_norm": 1.03125,
|
|
"learning_rate": 0.0003145,
|
|
"loss": 6.4931,
|
|
"mean_token_accuracy": 0.10356115698814392,
|
|
"num_tokens": 1420557.0,
|
|
"step": 630
|
|
},
|
|
{
|
|
"entropy": 7.038266181945801,
|
|
"epoch": 0.04189483407006664,
|
|
"grad_norm": 1.1484375,
|
|
"learning_rate": 0.000317,
|
|
"loss": 6.6219,
|
|
"mean_token_accuracy": 0.09532114937901497,
|
|
"num_tokens": 1430991.0,
|
|
"step": 635
|
|
},
|
|
{
|
|
"entropy": 7.113650941848755,
|
|
"epoch": 0.042224714653295504,
|
|
"grad_norm": 1.0,
|
|
"learning_rate": 0.0003195,
|
|
"loss": 6.6942,
|
|
"mean_token_accuracy": 0.08879134431481361,
|
|
"num_tokens": 1442114.0,
|
|
"step": 640
|
|
},
|
|
{
|
|
"entropy": 7.066236972808838,
|
|
"epoch": 0.04255459523652438,
|
|
"grad_norm": 0.9609375,
|
|
"learning_rate": 0.000322,
|
|
"loss": 6.645,
|
|
"mean_token_accuracy": 0.09683249071240425,
|
|
"num_tokens": 1453881.0,
|
|
"step": 645
|
|
},
|
|
{
|
|
"entropy": 7.095011949539185,
|
|
"epoch": 0.04288447581975325,
|
|
"grad_norm": 1.03125,
|
|
"learning_rate": 0.00032450000000000003,
|
|
"loss": 6.7115,
|
|
"mean_token_accuracy": 0.09116270616650582,
|
|
"num_tokens": 1466544.0,
|
|
"step": 650
|
|
},
|
|
{
|
|
"entropy": 7.081717681884766,
|
|
"epoch": 0.04321435640298212,
|
|
"grad_norm": 1.0703125,
|
|
"learning_rate": 0.00032700000000000003,
|
|
"loss": 6.6304,
|
|
"mean_token_accuracy": 0.09193866774439811,
|
|
"num_tokens": 1477245.0,
|
|
"step": 655
|
|
},
|
|
{
|
|
"entropy": 7.021490907669067,
|
|
"epoch": 0.043544236986210995,
|
|
"grad_norm": 1.0703125,
|
|
"learning_rate": 0.00032950000000000004,
|
|
"loss": 6.6149,
|
|
"mean_token_accuracy": 0.09211188033223153,
|
|
"num_tokens": 1487722.0,
|
|
"step": 660
|
|
},
|
|
{
|
|
"entropy": 6.972258710861206,
|
|
"epoch": 0.04387411756943986,
|
|
"grad_norm": 0.859375,
|
|
"learning_rate": 0.00033200000000000005,
|
|
"loss": 6.5751,
|
|
"mean_token_accuracy": 0.09731654524803161,
|
|
"num_tokens": 1498953.0,
|
|
"step": 665
|
|
},
|
|
{
|
|
"entropy": 7.111265420913696,
|
|
"epoch": 0.04420399815266873,
|
|
"grad_norm": 0.8828125,
|
|
"learning_rate": 0.00033450000000000005,
|
|
"loss": 6.6805,
|
|
"mean_token_accuracy": 0.08997858166694642,
|
|
"num_tokens": 1511883.0,
|
|
"step": 670
|
|
},
|
|
{
|
|
"entropy": 6.981727504730225,
|
|
"epoch": 0.044533878735897606,
|
|
"grad_norm": 0.94921875,
|
|
"learning_rate": 0.000337,
|
|
"loss": 6.6041,
|
|
"mean_token_accuracy": 0.09806737676262856,
|
|
"num_tokens": 1524639.0,
|
|
"step": 675
|
|
},
|
|
{
|
|
"entropy": 6.944048118591309,
|
|
"epoch": 0.04486375931912648,
|
|
"grad_norm": 1.03125,
|
|
"learning_rate": 0.0003395,
|
|
"loss": 6.4714,
|
|
"mean_token_accuracy": 0.1029288925230503,
|
|
"num_tokens": 1536123.0,
|
|
"step": 680
|
|
},
|
|
{
|
|
"entropy": 6.9569906234741214,
|
|
"epoch": 0.045193639902355344,
|
|
"grad_norm": 1.1328125,
|
|
"learning_rate": 0.000342,
|
|
"loss": 6.4986,
|
|
"mean_token_accuracy": 0.1026460200548172,
|
|
"num_tokens": 1547237.0,
|
|
"step": 685
|
|
},
|
|
{
|
|
"entropy": 6.901335096359253,
|
|
"epoch": 0.04552352048558422,
|
|
"grad_norm": 1.125,
|
|
"learning_rate": 0.00034449999999999997,
|
|
"loss": 6.4209,
|
|
"mean_token_accuracy": 0.10071871280670167,
|
|
"num_tokens": 1557527.0,
|
|
"step": 690
|
|
},
|
|
{
|
|
"entropy": 6.940638017654419,
|
|
"epoch": 0.04585340106881309,
|
|
"grad_norm": 0.88671875,
|
|
"learning_rate": 0.000347,
|
|
"loss": 6.5884,
|
|
"mean_token_accuracy": 0.10167076736688614,
|
|
"num_tokens": 1569867.0,
|
|
"step": 695
|
|
},
|
|
{
|
|
"entropy": 6.92997374534607,
|
|
"epoch": 0.04618328165204196,
|
|
"grad_norm": 1.171875,
|
|
"learning_rate": 0.0003495,
|
|
"loss": 6.4871,
|
|
"mean_token_accuracy": 0.10178036540746689,
|
|
"num_tokens": 1580729.0,
|
|
"step": 700
|
|
},
|
|
{
|
|
"entropy": 6.9461352825164795,
|
|
"epoch": 0.046513162235270834,
|
|
"grad_norm": 0.98828125,
|
|
"learning_rate": 0.000352,
|
|
"loss": 6.6135,
|
|
"mean_token_accuracy": 0.0916481964290142,
|
|
"num_tokens": 1592737.0,
|
|
"step": 705
|
|
},
|
|
{
|
|
"entropy": 7.001119470596313,
|
|
"epoch": 0.0468430428184997,
|
|
"grad_norm": 1.1640625,
|
|
"learning_rate": 0.0003545,
|
|
"loss": 6.5773,
|
|
"mean_token_accuracy": 0.09536935016512871,
|
|
"num_tokens": 1603347.0,
|
|
"step": 710
|
|
},
|
|
{
|
|
"entropy": 6.909364700317383,
|
|
"epoch": 0.04717292340172857,
|
|
"grad_norm": 1.015625,
|
|
"learning_rate": 0.000357,
|
|
"loss": 6.5847,
|
|
"mean_token_accuracy": 0.09743436947464942,
|
|
"num_tokens": 1614570.0,
|
|
"step": 715
|
|
},
|
|
{
|
|
"entropy": 6.923969364166259,
|
|
"epoch": 0.047502803984957445,
|
|
"grad_norm": 1.0078125,
|
|
"learning_rate": 0.0003595,
|
|
"loss": 6.4595,
|
|
"mean_token_accuracy": 0.10219181180000306,
|
|
"num_tokens": 1624796.0,
|
|
"step": 720
|
|
},
|
|
{
|
|
"entropy": 6.8807957649230955,
|
|
"epoch": 0.04783268456818632,
|
|
"grad_norm": 0.8515625,
|
|
"learning_rate": 0.000362,
|
|
"loss": 6.5337,
|
|
"mean_token_accuracy": 0.09934688806533813,
|
|
"num_tokens": 1637486.0,
|
|
"step": 725
|
|
},
|
|
{
|
|
"entropy": 6.961328601837158,
|
|
"epoch": 0.04816256515141519,
|
|
"grad_norm": 1.0390625,
|
|
"learning_rate": 0.0003645,
|
|
"loss": 6.558,
|
|
"mean_token_accuracy": 0.09534360244870185,
|
|
"num_tokens": 1648612.0,
|
|
"step": 730
|
|
},
|
|
{
|
|
"entropy": 6.844017457962036,
|
|
"epoch": 0.048492445734644056,
|
|
"grad_norm": 1.0703125,
|
|
"learning_rate": 0.000367,
|
|
"loss": 6.4521,
|
|
"mean_token_accuracy": 0.10778806954622269,
|
|
"num_tokens": 1658164.0,
|
|
"step": 735
|
|
},
|
|
{
|
|
"entropy": 6.809712696075439,
|
|
"epoch": 0.04882232631787293,
|
|
"grad_norm": 0.9921875,
|
|
"learning_rate": 0.0003695,
|
|
"loss": 6.4179,
|
|
"mean_token_accuracy": 0.10334590896964073,
|
|
"num_tokens": 1669126.0,
|
|
"step": 740
|
|
},
|
|
{
|
|
"entropy": 6.9402045726776125,
|
|
"epoch": 0.0491522069011018,
|
|
"grad_norm": 1.046875,
|
|
"learning_rate": 0.000372,
|
|
"loss": 6.5369,
|
|
"mean_token_accuracy": 0.10299055650830269,
|
|
"num_tokens": 1681075.0,
|
|
"step": 745
|
|
},
|
|
{
|
|
"entropy": 6.924366044998169,
|
|
"epoch": 0.049482087484330674,
|
|
"grad_norm": 1.0234375,
|
|
"learning_rate": 0.0003745,
|
|
"loss": 6.538,
|
|
"mean_token_accuracy": 0.09764362126588821,
|
|
"num_tokens": 1692733.0,
|
|
"step": 750
|
|
},
|
|
{
|
|
"entropy": 6.876445484161377,
|
|
"epoch": 0.049811968067559546,
|
|
"grad_norm": 0.91015625,
|
|
"learning_rate": 0.000377,
|
|
"loss": 6.4472,
|
|
"mean_token_accuracy": 0.1055051825940609,
|
|
"num_tokens": 1704532.0,
|
|
"step": 755
|
|
},
|
|
{
|
|
"entropy": 6.732907676696778,
|
|
"epoch": 0.05014184865078841,
|
|
"grad_norm": 1.0234375,
|
|
"learning_rate": 0.0003795,
|
|
"loss": 6.5222,
|
|
"mean_token_accuracy": 0.10377345606684685,
|
|
"num_tokens": 1716726.0,
|
|
"step": 760
|
|
},
|
|
{
|
|
"entropy": 6.961489915847778,
|
|
"epoch": 0.050471729234017285,
|
|
"grad_norm": 1.0625,
|
|
"learning_rate": 0.000382,
|
|
"loss": 6.4784,
|
|
"mean_token_accuracy": 0.10288533419370652,
|
|
"num_tokens": 1726720.0,
|
|
"step": 765
|
|
},
|
|
{
|
|
"entropy": 6.75892333984375,
|
|
"epoch": 0.05080160981724616,
|
|
"grad_norm": 0.99609375,
|
|
"learning_rate": 0.0003845,
|
|
"loss": 6.4747,
|
|
"mean_token_accuracy": 0.10111702308058738,
|
|
"num_tokens": 1738267.0,
|
|
"step": 770
|
|
},
|
|
{
|
|
"entropy": 6.826756715774536,
|
|
"epoch": 0.05113149040047503,
|
|
"grad_norm": 0.91015625,
|
|
"learning_rate": 0.00038700000000000003,
|
|
"loss": 6.3798,
|
|
"mean_token_accuracy": 0.10712268501520157,
|
|
"num_tokens": 1749094.0,
|
|
"step": 775
|
|
},
|
|
{
|
|
"entropy": 6.865867471694946,
|
|
"epoch": 0.0514613709837039,
|
|
"grad_norm": 1.078125,
|
|
"learning_rate": 0.00038950000000000003,
|
|
"loss": 6.5161,
|
|
"mean_token_accuracy": 0.09913287609815598,
|
|
"num_tokens": 1760066.0,
|
|
"step": 780
|
|
},
|
|
{
|
|
"entropy": 6.775571203231811,
|
|
"epoch": 0.05179125156693277,
|
|
"grad_norm": 0.90234375,
|
|
"learning_rate": 0.00039200000000000004,
|
|
"loss": 6.4364,
|
|
"mean_token_accuracy": 0.10061402916908264,
|
|
"num_tokens": 1771404.0,
|
|
"step": 785
|
|
},
|
|
{
|
|
"entropy": 6.756534242630005,
|
|
"epoch": 0.05212113215016164,
|
|
"grad_norm": 0.86328125,
|
|
"learning_rate": 0.00039450000000000005,
|
|
"loss": 6.4757,
|
|
"mean_token_accuracy": 0.10254075974225998,
|
|
"num_tokens": 1783265.0,
|
|
"step": 790
|
|
},
|
|
{
|
|
"entropy": 6.865955018997193,
|
|
"epoch": 0.05245101273339051,
|
|
"grad_norm": 0.859375,
|
|
"learning_rate": 0.00039700000000000005,
|
|
"loss": 6.4364,
|
|
"mean_token_accuracy": 0.10619740709662437,
|
|
"num_tokens": 1795062.0,
|
|
"step": 795
|
|
},
|
|
{
|
|
"entropy": 6.7756916046142575,
|
|
"epoch": 0.052780893316619386,
|
|
"grad_norm": 0.87890625,
|
|
"learning_rate": 0.0003995,
|
|
"loss": 6.4869,
|
|
"mean_token_accuracy": 0.09984813183546067,
|
|
"num_tokens": 1807307.0,
|
|
"step": 800
|
|
},
|
|
{
|
|
"entropy": 6.806865167617798,
|
|
"epoch": 0.05311077389984825,
|
|
"grad_norm": 1.046875,
|
|
"learning_rate": 0.000402,
|
|
"loss": 6.4058,
|
|
"mean_token_accuracy": 0.1037381909787655,
|
|
"num_tokens": 1818621.0,
|
|
"step": 805
|
|
},
|
|
{
|
|
"entropy": 6.7584771633148195,
|
|
"epoch": 0.053440654483077124,
|
|
"grad_norm": 1.0390625,
|
|
"learning_rate": 0.0004045,
|
|
"loss": 6.4676,
|
|
"mean_token_accuracy": 0.10668170377612114,
|
|
"num_tokens": 1829586.0,
|
|
"step": 810
|
|
},
|
|
{
|
|
"entropy": 6.816446971893311,
|
|
"epoch": 0.053770535066306,
|
|
"grad_norm": 1.078125,
|
|
"learning_rate": 0.00040699999999999997,
|
|
"loss": 6.4232,
|
|
"mean_token_accuracy": 0.10452193915843963,
|
|
"num_tokens": 1842204.0,
|
|
"step": 815
|
|
},
|
|
{
|
|
"entropy": 6.782231283187866,
|
|
"epoch": 0.05410041564953487,
|
|
"grad_norm": 1.0078125,
|
|
"learning_rate": 0.0004095,
|
|
"loss": 6.4303,
|
|
"mean_token_accuracy": 0.10292760878801346,
|
|
"num_tokens": 1854808.0,
|
|
"step": 820
|
|
},
|
|
{
|
|
"entropy": 6.896180295944214,
|
|
"epoch": 0.05443029623276374,
|
|
"grad_norm": 1.015625,
|
|
"learning_rate": 0.000412,
|
|
"loss": 6.5556,
|
|
"mean_token_accuracy": 0.09649628847837448,
|
|
"num_tokens": 1865896.0,
|
|
"step": 825
|
|
},
|
|
{
|
|
"entropy": 6.704454660415649,
|
|
"epoch": 0.05476017681599261,
|
|
"grad_norm": 0.984375,
|
|
"learning_rate": 0.0004145,
|
|
"loss": 6.4313,
|
|
"mean_token_accuracy": 0.10251373648643494,
|
|
"num_tokens": 1877137.0,
|
|
"step": 830
|
|
},
|
|
{
|
|
"entropy": 6.768334531784058,
|
|
"epoch": 0.05509005739922148,
|
|
"grad_norm": 0.87109375,
|
|
"learning_rate": 0.000417,
|
|
"loss": 6.4446,
|
|
"mean_token_accuracy": 0.10653403997421265,
|
|
"num_tokens": 1888442.0,
|
|
"step": 835
|
|
},
|
|
{
|
|
"entropy": 6.714272165298462,
|
|
"epoch": 0.05541993798245035,
|
|
"grad_norm": 1.0234375,
|
|
"learning_rate": 0.0004195,
|
|
"loss": 6.3687,
|
|
"mean_token_accuracy": 0.11127936169505119,
|
|
"num_tokens": 1899943.0,
|
|
"step": 840
|
|
},
|
|
{
|
|
"entropy": 6.777819633483887,
|
|
"epoch": 0.055749818565679225,
|
|
"grad_norm": 0.96875,
|
|
"learning_rate": 0.000422,
|
|
"loss": 6.4617,
|
|
"mean_token_accuracy": 0.10296945869922638,
|
|
"num_tokens": 1910725.0,
|
|
"step": 845
|
|
},
|
|
{
|
|
"entropy": 6.739258670806885,
|
|
"epoch": 0.0560796991489081,
|
|
"grad_norm": 0.94921875,
|
|
"learning_rate": 0.0004245,
|
|
"loss": 6.402,
|
|
"mean_token_accuracy": 0.10930126681923866,
|
|
"num_tokens": 1922194.0,
|
|
"step": 850
|
|
},
|
|
{
|
|
"entropy": 6.722884321212769,
|
|
"epoch": 0.056409579732136964,
|
|
"grad_norm": 0.90234375,
|
|
"learning_rate": 0.000427,
|
|
"loss": 6.3871,
|
|
"mean_token_accuracy": 0.10791410580277443,
|
|
"num_tokens": 1932985.0,
|
|
"step": 855
|
|
},
|
|
{
|
|
"entropy": 6.811666202545166,
|
|
"epoch": 0.056739460315365836,
|
|
"grad_norm": 1.109375,
|
|
"learning_rate": 0.0004295,
|
|
"loss": 6.4753,
|
|
"mean_token_accuracy": 0.10246776863932609,
|
|
"num_tokens": 1944560.0,
|
|
"step": 860
|
|
},
|
|
{
|
|
"entropy": 6.694799852371216,
|
|
"epoch": 0.05706934089859471,
|
|
"grad_norm": 0.96484375,
|
|
"learning_rate": 0.000432,
|
|
"loss": 6.3872,
|
|
"mean_token_accuracy": 0.10333370044827461,
|
|
"num_tokens": 1954460.0,
|
|
"step": 865
|
|
},
|
|
{
|
|
"entropy": 6.6748510837554935,
|
|
"epoch": 0.05739922148182358,
|
|
"grad_norm": 1.0390625,
|
|
"learning_rate": 0.0004345,
|
|
"loss": 6.395,
|
|
"mean_token_accuracy": 0.10444966927170754,
|
|
"num_tokens": 1966856.0,
|
|
"step": 870
|
|
},
|
|
{
|
|
"entropy": 6.788928270339966,
|
|
"epoch": 0.057729102065052454,
|
|
"grad_norm": 1.109375,
|
|
"learning_rate": 0.000437,
|
|
"loss": 6.448,
|
|
"mean_token_accuracy": 0.10007872432470322,
|
|
"num_tokens": 1978301.0,
|
|
"step": 875
|
|
},
|
|
{
|
|
"entropy": 6.734203243255616,
|
|
"epoch": 0.05805898264828132,
|
|
"grad_norm": 0.83984375,
|
|
"learning_rate": 0.0004395,
|
|
"loss": 6.4429,
|
|
"mean_token_accuracy": 0.09999567046761512,
|
|
"num_tokens": 1990345.0,
|
|
"step": 880
|
|
},
|
|
{
|
|
"entropy": 6.668420028686524,
|
|
"epoch": 0.05838886323151019,
|
|
"grad_norm": 1.03125,
|
|
"learning_rate": 0.000442,
|
|
"loss": 6.3568,
|
|
"mean_token_accuracy": 0.10593720152974129,
|
|
"num_tokens": 2002261.0,
|
|
"step": 885
|
|
},
|
|
{
|
|
"entropy": 6.645977258682251,
|
|
"epoch": 0.058718743814739065,
|
|
"grad_norm": 0.875,
|
|
"learning_rate": 0.0004445,
|
|
"loss": 6.2904,
|
|
"mean_token_accuracy": 0.11104925945401192,
|
|
"num_tokens": 2013154.0,
|
|
"step": 890
|
|
},
|
|
{
|
|
"entropy": 6.74332070350647,
|
|
"epoch": 0.05904862439796794,
|
|
"grad_norm": 1.0546875,
|
|
"learning_rate": 0.000447,
|
|
"loss": 6.2642,
|
|
"mean_token_accuracy": 0.11385588869452476,
|
|
"num_tokens": 2023222.0,
|
|
"step": 895
|
|
},
|
|
{
|
|
"entropy": 6.569306468963623,
|
|
"epoch": 0.05937850498119681,
|
|
"grad_norm": 0.90234375,
|
|
"learning_rate": 0.00044950000000000003,
|
|
"loss": 6.3664,
|
|
"mean_token_accuracy": 0.10586040914058685,
|
|
"num_tokens": 2034163.0,
|
|
"step": 900
|
|
},
|
|
{
|
|
"entropy": 6.647012186050415,
|
|
"epoch": 0.059708385564425676,
|
|
"grad_norm": 1.03125,
|
|
"learning_rate": 0.00045200000000000004,
|
|
"loss": 6.3483,
|
|
"mean_token_accuracy": 0.10745750218629838,
|
|
"num_tokens": 2045722.0,
|
|
"step": 905
|
|
},
|
|
{
|
|
"entropy": 6.613310384750366,
|
|
"epoch": 0.06003826614765455,
|
|
"grad_norm": 0.96875,
|
|
"learning_rate": 0.00045450000000000004,
|
|
"loss": 6.3857,
|
|
"mean_token_accuracy": 0.10500504523515701,
|
|
"num_tokens": 2057116.0,
|
|
"step": 910
|
|
},
|
|
{
|
|
"entropy": 6.669482088088989,
|
|
"epoch": 0.06036814673088342,
|
|
"grad_norm": 0.98046875,
|
|
"learning_rate": 0.00045700000000000005,
|
|
"loss": 6.3459,
|
|
"mean_token_accuracy": 0.10842561721801758,
|
|
"num_tokens": 2068105.0,
|
|
"step": 915
|
|
},
|
|
{
|
|
"entropy": 6.643321228027344,
|
|
"epoch": 0.06069802731411229,
|
|
"grad_norm": 1.046875,
|
|
"learning_rate": 0.00045950000000000006,
|
|
"loss": 6.3553,
|
|
"mean_token_accuracy": 0.10748406276106834,
|
|
"num_tokens": 2079562.0,
|
|
"step": 920
|
|
},
|
|
{
|
|
"entropy": 6.633617401123047,
|
|
"epoch": 0.061027907897341166,
|
|
"grad_norm": 0.91796875,
|
|
"learning_rate": 0.000462,
|
|
"loss": 6.3111,
|
|
"mean_token_accuracy": 0.11069138199090958,
|
|
"num_tokens": 2090020.0,
|
|
"step": 925
|
|
},
|
|
{
|
|
"entropy": 6.670710563659668,
|
|
"epoch": 0.06135778848057003,
|
|
"grad_norm": 1.0546875,
|
|
"learning_rate": 0.0004645,
|
|
"loss": 6.3965,
|
|
"mean_token_accuracy": 0.10716222003102302,
|
|
"num_tokens": 2101432.0,
|
|
"step": 930
|
|
},
|
|
{
|
|
"entropy": 6.629409217834473,
|
|
"epoch": 0.061687669063798904,
|
|
"grad_norm": 0.9296875,
|
|
"learning_rate": 0.000467,
|
|
"loss": 6.4048,
|
|
"mean_token_accuracy": 0.10834496244788169,
|
|
"num_tokens": 2113196.0,
|
|
"step": 935
|
|
},
|
|
{
|
|
"entropy": 6.649847459793091,
|
|
"epoch": 0.06201754964702778,
|
|
"grad_norm": 0.9140625,
|
|
"learning_rate": 0.0004695,
|
|
"loss": 6.3363,
|
|
"mean_token_accuracy": 0.10928401499986648,
|
|
"num_tokens": 2123664.0,
|
|
"step": 940
|
|
},
|
|
{
|
|
"entropy": 6.699871063232422,
|
|
"epoch": 0.06234743023025665,
|
|
"grad_norm": 1.046875,
|
|
"learning_rate": 0.000472,
|
|
"loss": 6.4351,
|
|
"mean_token_accuracy": 0.10690466091036796,
|
|
"num_tokens": 2133973.0,
|
|
"step": 945
|
|
},
|
|
{
|
|
"entropy": 6.686114501953125,
|
|
"epoch": 0.06267731081348552,
|
|
"grad_norm": 1.1328125,
|
|
"learning_rate": 0.0004745,
|
|
"loss": 6.4187,
|
|
"mean_token_accuracy": 0.10491651818156242,
|
|
"num_tokens": 2146957.0,
|
|
"step": 950
|
|
},
|
|
{
|
|
"entropy": 6.661528491973877,
|
|
"epoch": 0.06300719139671439,
|
|
"grad_norm": 1.09375,
|
|
"learning_rate": 0.000477,
|
|
"loss": 6.3329,
|
|
"mean_token_accuracy": 0.11448214650154113,
|
|
"num_tokens": 2156442.0,
|
|
"step": 955
|
|
},
|
|
{
|
|
"entropy": 6.652226686477661,
|
|
"epoch": 0.06333707197994326,
|
|
"grad_norm": 1.0078125,
|
|
"learning_rate": 0.0004795,
|
|
"loss": 6.3445,
|
|
"mean_token_accuracy": 0.10994396656751633,
|
|
"num_tokens": 2167385.0,
|
|
"step": 960
|
|
},
|
|
{
|
|
"entropy": 6.648888683319091,
|
|
"epoch": 0.06366695256317213,
|
|
"grad_norm": 0.94140625,
|
|
"learning_rate": 0.000482,
|
|
"loss": 6.3536,
|
|
"mean_token_accuracy": 0.11058482751250268,
|
|
"num_tokens": 2178537.0,
|
|
"step": 965
|
|
},
|
|
{
|
|
"entropy": 6.561604452133179,
|
|
"epoch": 0.063996833146401,
|
|
"grad_norm": 0.9296875,
|
|
"learning_rate": 0.0004845,
|
|
"loss": 6.2966,
|
|
"mean_token_accuracy": 0.1082301378250122,
|
|
"num_tokens": 2190927.0,
|
|
"step": 970
|
|
},
|
|
{
|
|
"entropy": 6.6591477394104,
|
|
"epoch": 0.06432671372962988,
|
|
"grad_norm": 0.95703125,
|
|
"learning_rate": 0.000487,
|
|
"loss": 6.426,
|
|
"mean_token_accuracy": 0.10577302202582359,
|
|
"num_tokens": 2203395.0,
|
|
"step": 975
|
|
},
|
|
{
|
|
"entropy": 6.554022645950317,
|
|
"epoch": 0.06465659431285875,
|
|
"grad_norm": 0.87109375,
|
|
"learning_rate": 0.0004895,
|
|
"loss": 6.2469,
|
|
"mean_token_accuracy": 0.1093327522277832,
|
|
"num_tokens": 2214582.0,
|
|
"step": 980
|
|
},
|
|
{
|
|
"entropy": 6.6766685962677,
|
|
"epoch": 0.06498647489608762,
|
|
"grad_norm": 1.0859375,
|
|
"learning_rate": 0.000492,
|
|
"loss": 6.3288,
|
|
"mean_token_accuracy": 0.10834297239780426,
|
|
"num_tokens": 2223849.0,
|
|
"step": 985
|
|
},
|
|
{
|
|
"entropy": 6.505914926528931,
|
|
"epoch": 0.06531635547931648,
|
|
"grad_norm": 0.97265625,
|
|
"learning_rate": 0.0004945,
|
|
"loss": 6.2819,
|
|
"mean_token_accuracy": 0.11325727254152299,
|
|
"num_tokens": 2235327.0,
|
|
"step": 990
|
|
},
|
|
{
|
|
"entropy": 6.573624134063721,
|
|
"epoch": 0.06564623606254535,
|
|
"grad_norm": 0.99609375,
|
|
"learning_rate": 0.000497,
|
|
"loss": 6.3135,
|
|
"mean_token_accuracy": 0.1097367025911808,
|
|
"num_tokens": 2247684.0,
|
|
"step": 995
|
|
},
|
|
{
|
|
"entropy": 6.526281309127808,
|
|
"epoch": 0.06597611664577423,
|
|
"grad_norm": 0.90625,
|
|
"learning_rate": 0.0004995,
|
|
"loss": 6.2919,
|
|
"mean_token_accuracy": 0.10900129601359368,
|
|
"num_tokens": 2258406.0,
|
|
"step": 1000
|
|
}
|
|
],
|
|
"logging_steps": 5,
|
|
"max_steps": 4000,
|
|
"num_input_tokens_seen": 0,
|
|
"num_train_epochs": 1,
|
|
"save_steps": 500,
|
|
"stateful_callbacks": {
|
|
"TrainerControl": {
|
|
"args": {
|
|
"should_epoch_stop": false,
|
|
"should_evaluate": false,
|
|
"should_log": false,
|
|
"should_save": true,
|
|
"should_training_stop": false
|
|
},
|
|
"attributes": {}
|
|
}
|
|
},
|
|
"total_flos": 561419709972480.0,
|
|
"train_batch_size": 16,
|
|
"trial_name": null,
|
|
"trial_params": null
|
|
}
|