1046 lines
28 KiB
JSON
1046 lines
28 KiB
JSON
{
|
|
"best_global_step": null,
|
|
"best_metric": null,
|
|
"best_model_checkpoint": null,
|
|
"epoch": 0.5408328826392644,
|
|
"eval_steps": 500,
|
|
"global_step": 500,
|
|
"is_hyper_param_search": false,
|
|
"is_local_process_zero": true,
|
|
"is_world_process_zero": true,
|
|
"log_history": [
|
|
{
|
|
"entropy": 10.692022037506103,
|
|
"epoch": 0.005408328826392645,
|
|
"grad_norm": 14.3125,
|
|
"learning_rate": 2e-06,
|
|
"loss": 10.8462,
|
|
"mean_token_accuracy": 0.0,
|
|
"num_tokens": 10942.0,
|
|
"step": 5
|
|
},
|
|
{
|
|
"entropy": 10.691925811767579,
|
|
"epoch": 0.01081665765278529,
|
|
"grad_norm": 12.5625,
|
|
"learning_rate": 4.5e-06,
|
|
"loss": 10.7663,
|
|
"mean_token_accuracy": 0.00011025358689948916,
|
|
"num_tokens": 21887.0,
|
|
"step": 10
|
|
},
|
|
{
|
|
"entropy": 10.69113998413086,
|
|
"epoch": 0.016224986479177934,
|
|
"grad_norm": 9.9375,
|
|
"learning_rate": 7e-06,
|
|
"loss": 10.515,
|
|
"mean_token_accuracy": 0.014385890925768762,
|
|
"num_tokens": 34345.0,
|
|
"step": 15
|
|
},
|
|
{
|
|
"entropy": 10.684020137786865,
|
|
"epoch": 0.02163331530557058,
|
|
"grad_norm": 6.46875,
|
|
"learning_rate": 9.5e-06,
|
|
"loss": 10.1559,
|
|
"mean_token_accuracy": 0.0367370892316103,
|
|
"num_tokens": 45445.0,
|
|
"step": 20
|
|
},
|
|
{
|
|
"entropy": 10.650034141540527,
|
|
"epoch": 0.02704164413196322,
|
|
"grad_norm": 3.828125,
|
|
"learning_rate": 1.2e-05,
|
|
"loss": 9.8592,
|
|
"mean_token_accuracy": 0.039543523266911504,
|
|
"num_tokens": 58237.0,
|
|
"step": 25
|
|
},
|
|
{
|
|
"entropy": 10.601274299621583,
|
|
"epoch": 0.03244997295835587,
|
|
"grad_norm": 3.34375,
|
|
"learning_rate": 1.4500000000000002e-05,
|
|
"loss": 9.6444,
|
|
"mean_token_accuracy": 0.04297022223472595,
|
|
"num_tokens": 68098.0,
|
|
"step": 30
|
|
},
|
|
{
|
|
"entropy": 10.589245128631593,
|
|
"epoch": 0.03785830178474851,
|
|
"grad_norm": 2.75,
|
|
"learning_rate": 1.7000000000000003e-05,
|
|
"loss": 9.597,
|
|
"mean_token_accuracy": 0.040016197971999644,
|
|
"num_tokens": 79867.0,
|
|
"step": 35
|
|
},
|
|
{
|
|
"entropy": 10.5987380027771,
|
|
"epoch": 0.04326663061114116,
|
|
"grad_norm": 2.71875,
|
|
"learning_rate": 1.95e-05,
|
|
"loss": 9.5915,
|
|
"mean_token_accuracy": 0.04129943773150444,
|
|
"num_tokens": 91257.0,
|
|
"step": 40
|
|
},
|
|
{
|
|
"entropy": 10.589694404602051,
|
|
"epoch": 0.048674959437533805,
|
|
"grad_norm": 2.40625,
|
|
"learning_rate": 2.2e-05,
|
|
"loss": 9.5887,
|
|
"mean_token_accuracy": 0.04088671933859587,
|
|
"num_tokens": 103474.0,
|
|
"step": 45
|
|
},
|
|
{
|
|
"entropy": 10.584661674499511,
|
|
"epoch": 0.05408328826392644,
|
|
"grad_norm": 2.65625,
|
|
"learning_rate": 2.4500000000000003e-05,
|
|
"loss": 9.4849,
|
|
"mean_token_accuracy": 0.039065463840961455,
|
|
"num_tokens": 114830.0,
|
|
"step": 50
|
|
},
|
|
{
|
|
"entropy": 10.57274866104126,
|
|
"epoch": 0.05949161709031909,
|
|
"grad_norm": 2.375,
|
|
"learning_rate": 2.7e-05,
|
|
"loss": 9.4147,
|
|
"mean_token_accuracy": 0.042727841436862944,
|
|
"num_tokens": 126017.0,
|
|
"step": 55
|
|
},
|
|
{
|
|
"entropy": 10.554320049285888,
|
|
"epoch": 0.06489994591671173,
|
|
"grad_norm": 2.453125,
|
|
"learning_rate": 2.95e-05,
|
|
"loss": 9.3348,
|
|
"mean_token_accuracy": 0.048937696032226086,
|
|
"num_tokens": 138509.0,
|
|
"step": 60
|
|
},
|
|
{
|
|
"entropy": 10.518039512634278,
|
|
"epoch": 0.07030827474310439,
|
|
"grad_norm": 2.34375,
|
|
"learning_rate": 3.2e-05,
|
|
"loss": 9.2417,
|
|
"mean_token_accuracy": 0.06088445819914341,
|
|
"num_tokens": 148353.0,
|
|
"step": 65
|
|
},
|
|
{
|
|
"entropy": 10.38691234588623,
|
|
"epoch": 0.07571660356949703,
|
|
"grad_norm": 2.34375,
|
|
"learning_rate": 3.4500000000000005e-05,
|
|
"loss": 9.2043,
|
|
"mean_token_accuracy": 0.05567694678902626,
|
|
"num_tokens": 160269.0,
|
|
"step": 70
|
|
},
|
|
{
|
|
"entropy": 10.452953910827636,
|
|
"epoch": 0.08112493239588967,
|
|
"grad_norm": 2.359375,
|
|
"learning_rate": 3.7e-05,
|
|
"loss": 9.0443,
|
|
"mean_token_accuracy": 0.059686141461133956,
|
|
"num_tokens": 171568.0,
|
|
"step": 75
|
|
},
|
|
{
|
|
"entropy": 10.435920333862304,
|
|
"epoch": 0.08653326122228232,
|
|
"grad_norm": 2.375,
|
|
"learning_rate": 3.95e-05,
|
|
"loss": 8.9471,
|
|
"mean_token_accuracy": 0.06421410515904427,
|
|
"num_tokens": 184243.0,
|
|
"step": 80
|
|
},
|
|
{
|
|
"entropy": 10.38515739440918,
|
|
"epoch": 0.09194159004867496,
|
|
"grad_norm": 2.359375,
|
|
"learning_rate": 4.2000000000000004e-05,
|
|
"loss": 8.849,
|
|
"mean_token_accuracy": 0.06825026907026768,
|
|
"num_tokens": 194887.0,
|
|
"step": 85
|
|
},
|
|
{
|
|
"entropy": 10.27032928466797,
|
|
"epoch": 0.09734991887506761,
|
|
"grad_norm": 2.53125,
|
|
"learning_rate": 4.45e-05,
|
|
"loss": 8.7532,
|
|
"mean_token_accuracy": 0.07215963676571846,
|
|
"num_tokens": 206476.0,
|
|
"step": 90
|
|
},
|
|
{
|
|
"entropy": 10.199292469024659,
|
|
"epoch": 0.10275824770146025,
|
|
"grad_norm": 2.578125,
|
|
"learning_rate": 4.7000000000000004e-05,
|
|
"loss": 8.6005,
|
|
"mean_token_accuracy": 0.07982434406876564,
|
|
"num_tokens": 216578.0,
|
|
"step": 95
|
|
},
|
|
{
|
|
"entropy": 10.170300388336182,
|
|
"epoch": 0.10816657652785289,
|
|
"grad_norm": 2.078125,
|
|
"learning_rate": 4.9500000000000004e-05,
|
|
"loss": 8.506,
|
|
"mean_token_accuracy": 0.0810505922883749,
|
|
"num_tokens": 226941.0,
|
|
"step": 100
|
|
},
|
|
{
|
|
"entropy": 10.120428371429444,
|
|
"epoch": 0.11357490535424554,
|
|
"grad_norm": 2.140625,
|
|
"learning_rate": 5.2e-05,
|
|
"loss": 8.4082,
|
|
"mean_token_accuracy": 0.08101015165448189,
|
|
"num_tokens": 238060.0,
|
|
"step": 105
|
|
},
|
|
{
|
|
"entropy": 10.04278860092163,
|
|
"epoch": 0.11898323418063818,
|
|
"grad_norm": 2.1875,
|
|
"learning_rate": 5.45e-05,
|
|
"loss": 8.2557,
|
|
"mean_token_accuracy": 0.08279627189040184,
|
|
"num_tokens": 249137.0,
|
|
"step": 110
|
|
},
|
|
{
|
|
"entropy": 9.902883434295655,
|
|
"epoch": 0.12439156300703083,
|
|
"grad_norm": 1.8984375,
|
|
"learning_rate": 5.7e-05,
|
|
"loss": 8.1556,
|
|
"mean_token_accuracy": 0.08368096873164177,
|
|
"num_tokens": 260712.0,
|
|
"step": 115
|
|
},
|
|
{
|
|
"entropy": 9.842740917205811,
|
|
"epoch": 0.12979989183342347,
|
|
"grad_norm": 2.109375,
|
|
"learning_rate": 5.9499999999999996e-05,
|
|
"loss": 7.9778,
|
|
"mean_token_accuracy": 0.08551804721355438,
|
|
"num_tokens": 271039.0,
|
|
"step": 120
|
|
},
|
|
{
|
|
"entropy": 9.667721843719482,
|
|
"epoch": 0.1352082206598161,
|
|
"grad_norm": 1.8203125,
|
|
"learning_rate": 6.2e-05,
|
|
"loss": 7.8296,
|
|
"mean_token_accuracy": 0.088394845277071,
|
|
"num_tokens": 281353.0,
|
|
"step": 125
|
|
},
|
|
{
|
|
"entropy": 9.477235412597656,
|
|
"epoch": 0.14061654948620878,
|
|
"grad_norm": 1.96875,
|
|
"learning_rate": 6.450000000000001e-05,
|
|
"loss": 7.8005,
|
|
"mean_token_accuracy": 0.08695972561836243,
|
|
"num_tokens": 292221.0,
|
|
"step": 130
|
|
},
|
|
{
|
|
"entropy": 9.307325172424317,
|
|
"epoch": 0.1460248783126014,
|
|
"grad_norm": 1.625,
|
|
"learning_rate": 6.7e-05,
|
|
"loss": 7.6751,
|
|
"mean_token_accuracy": 0.08949057161808013,
|
|
"num_tokens": 303353.0,
|
|
"step": 135
|
|
},
|
|
{
|
|
"entropy": 9.106245803833009,
|
|
"epoch": 0.15143320713899405,
|
|
"grad_norm": 1.578125,
|
|
"learning_rate": 6.950000000000001e-05,
|
|
"loss": 7.6111,
|
|
"mean_token_accuracy": 0.09538959562778473,
|
|
"num_tokens": 314165.0,
|
|
"step": 140
|
|
},
|
|
{
|
|
"entropy": 8.896690082550048,
|
|
"epoch": 0.1568415359653867,
|
|
"grad_norm": 1.5546875,
|
|
"learning_rate": 7.2e-05,
|
|
"loss": 7.4219,
|
|
"mean_token_accuracy": 0.09479932934045791,
|
|
"num_tokens": 326387.0,
|
|
"step": 145
|
|
},
|
|
{
|
|
"entropy": 8.698641109466553,
|
|
"epoch": 0.16224986479177933,
|
|
"grad_norm": 1.53125,
|
|
"learning_rate": 7.45e-05,
|
|
"loss": 7.4523,
|
|
"mean_token_accuracy": 0.08837239667773247,
|
|
"num_tokens": 339155.0,
|
|
"step": 150
|
|
},
|
|
{
|
|
"entropy": 8.527585792541505,
|
|
"epoch": 0.167658193618172,
|
|
"grad_norm": 1.640625,
|
|
"learning_rate": 7.7e-05,
|
|
"loss": 7.3469,
|
|
"mean_token_accuracy": 0.09304244369268418,
|
|
"num_tokens": 350643.0,
|
|
"step": 155
|
|
},
|
|
{
|
|
"entropy": 8.301559352874756,
|
|
"epoch": 0.17306652244456464,
|
|
"grad_norm": 1.46875,
|
|
"learning_rate": 7.950000000000001e-05,
|
|
"loss": 7.257,
|
|
"mean_token_accuracy": 0.09016378521919251,
|
|
"num_tokens": 361977.0,
|
|
"step": 160
|
|
},
|
|
{
|
|
"entropy": 8.236385345458984,
|
|
"epoch": 0.17847485127095727,
|
|
"grad_norm": 1.4140625,
|
|
"learning_rate": 8.2e-05,
|
|
"loss": 7.2952,
|
|
"mean_token_accuracy": 0.09531197249889374,
|
|
"num_tokens": 373604.0,
|
|
"step": 165
|
|
},
|
|
{
|
|
"entropy": 8.092022514343261,
|
|
"epoch": 0.1838831800973499,
|
|
"grad_norm": 1.4453125,
|
|
"learning_rate": 8.450000000000001e-05,
|
|
"loss": 7.2732,
|
|
"mean_token_accuracy": 0.09200240671634674,
|
|
"num_tokens": 384306.0,
|
|
"step": 170
|
|
},
|
|
{
|
|
"entropy": 7.969397687911988,
|
|
"epoch": 0.18929150892374255,
|
|
"grad_norm": 1.375,
|
|
"learning_rate": 8.7e-05,
|
|
"loss": 7.169,
|
|
"mean_token_accuracy": 0.1011082261800766,
|
|
"num_tokens": 396479.0,
|
|
"step": 175
|
|
},
|
|
{
|
|
"entropy": 7.91082444190979,
|
|
"epoch": 0.19469983775013522,
|
|
"grad_norm": 1.515625,
|
|
"learning_rate": 8.95e-05,
|
|
"loss": 7.1033,
|
|
"mean_token_accuracy": 0.10111359134316444,
|
|
"num_tokens": 408489.0,
|
|
"step": 180
|
|
},
|
|
{
|
|
"entropy": 7.861333703994751,
|
|
"epoch": 0.20010816657652786,
|
|
"grad_norm": 1.7109375,
|
|
"learning_rate": 9.2e-05,
|
|
"loss": 7.1182,
|
|
"mean_token_accuracy": 0.10262049958109856,
|
|
"num_tokens": 419339.0,
|
|
"step": 185
|
|
},
|
|
{
|
|
"entropy": 7.818029165267944,
|
|
"epoch": 0.2055164954029205,
|
|
"grad_norm": 1.25,
|
|
"learning_rate": 9.45e-05,
|
|
"loss": 7.0883,
|
|
"mean_token_accuracy": 0.10314406752586365,
|
|
"num_tokens": 430486.0,
|
|
"step": 190
|
|
},
|
|
{
|
|
"entropy": 7.71181092262268,
|
|
"epoch": 0.21092482422931313,
|
|
"grad_norm": 1.46875,
|
|
"learning_rate": 9.7e-05,
|
|
"loss": 7.1069,
|
|
"mean_token_accuracy": 0.10316698178648949,
|
|
"num_tokens": 440414.0,
|
|
"step": 195
|
|
},
|
|
{
|
|
"entropy": 7.695619297027588,
|
|
"epoch": 0.21633315305570577,
|
|
"grad_norm": 1.4140625,
|
|
"learning_rate": 9.95e-05,
|
|
"loss": 7.0605,
|
|
"mean_token_accuracy": 0.1009691059589386,
|
|
"num_tokens": 451475.0,
|
|
"step": 200
|
|
},
|
|
{
|
|
"entropy": 7.618443393707276,
|
|
"epoch": 0.22174148188209844,
|
|
"grad_norm": 1.234375,
|
|
"learning_rate": 0.000102,
|
|
"loss": 7.126,
|
|
"mean_token_accuracy": 0.1053870052099228,
|
|
"num_tokens": 462053.0,
|
|
"step": 205
|
|
},
|
|
{
|
|
"entropy": 7.729849576950073,
|
|
"epoch": 0.22714981070849108,
|
|
"grad_norm": 1.3203125,
|
|
"learning_rate": 0.00010449999999999999,
|
|
"loss": 7.0694,
|
|
"mean_token_accuracy": 0.10431502312421799,
|
|
"num_tokens": 473207.0,
|
|
"step": 210
|
|
},
|
|
{
|
|
"entropy": 7.642491388320923,
|
|
"epoch": 0.23255813953488372,
|
|
"grad_norm": 1.3671875,
|
|
"learning_rate": 0.000107,
|
|
"loss": 7.0593,
|
|
"mean_token_accuracy": 0.10497235506772995,
|
|
"num_tokens": 484999.0,
|
|
"step": 215
|
|
},
|
|
{
|
|
"entropy": 7.669218826293945,
|
|
"epoch": 0.23796646836127636,
|
|
"grad_norm": 1.25,
|
|
"learning_rate": 0.0001095,
|
|
"loss": 7.0536,
|
|
"mean_token_accuracy": 0.10244816839694977,
|
|
"num_tokens": 496772.0,
|
|
"step": 220
|
|
},
|
|
{
|
|
"entropy": 7.611874437332153,
|
|
"epoch": 0.24337479718766902,
|
|
"grad_norm": 1.6484375,
|
|
"learning_rate": 0.000112,
|
|
"loss": 7.0209,
|
|
"mean_token_accuracy": 0.10049235001206398,
|
|
"num_tokens": 507782.0,
|
|
"step": 225
|
|
},
|
|
{
|
|
"entropy": 7.446110391616822,
|
|
"epoch": 0.24878312601406166,
|
|
"grad_norm": 1.4296875,
|
|
"learning_rate": 0.0001145,
|
|
"loss": 6.9319,
|
|
"mean_token_accuracy": 0.10533562302589417,
|
|
"num_tokens": 518950.0,
|
|
"step": 230
|
|
},
|
|
{
|
|
"entropy": 7.5693611145019535,
|
|
"epoch": 0.2541914548404543,
|
|
"grad_norm": 1.25,
|
|
"learning_rate": 0.00011700000000000001,
|
|
"loss": 7.0201,
|
|
"mean_token_accuracy": 0.1056381680071354,
|
|
"num_tokens": 530795.0,
|
|
"step": 235
|
|
},
|
|
{
|
|
"entropy": 7.5523681640625,
|
|
"epoch": 0.25959978366684694,
|
|
"grad_norm": 1.5625,
|
|
"learning_rate": 0.00011949999999999999,
|
|
"loss": 7.027,
|
|
"mean_token_accuracy": 0.10181101113557815,
|
|
"num_tokens": 542878.0,
|
|
"step": 240
|
|
},
|
|
{
|
|
"entropy": 7.477112722396851,
|
|
"epoch": 0.2650081124932396,
|
|
"grad_norm": 1.140625,
|
|
"learning_rate": 0.000122,
|
|
"loss": 6.9276,
|
|
"mean_token_accuracy": 0.10972334071993828,
|
|
"num_tokens": 553682.0,
|
|
"step": 245
|
|
},
|
|
{
|
|
"entropy": 7.378021717071533,
|
|
"epoch": 0.2704164413196322,
|
|
"grad_norm": 1.1953125,
|
|
"learning_rate": 0.0001245,
|
|
"loss": 6.9296,
|
|
"mean_token_accuracy": 0.11474497318267822,
|
|
"num_tokens": 564433.0,
|
|
"step": 250
|
|
},
|
|
{
|
|
"entropy": 7.477735948562622,
|
|
"epoch": 0.27582477014602486,
|
|
"grad_norm": 1.203125,
|
|
"learning_rate": 0.000127,
|
|
"loss": 6.9645,
|
|
"mean_token_accuracy": 0.10399928912520409,
|
|
"num_tokens": 576001.0,
|
|
"step": 255
|
|
},
|
|
{
|
|
"entropy": 7.369494915008545,
|
|
"epoch": 0.28123309897241755,
|
|
"grad_norm": 1.21875,
|
|
"learning_rate": 0.0001295,
|
|
"loss": 6.8768,
|
|
"mean_token_accuracy": 0.11646146923303605,
|
|
"num_tokens": 587138.0,
|
|
"step": 260
|
|
},
|
|
{
|
|
"entropy": 7.286924839019775,
|
|
"epoch": 0.2866414277988102,
|
|
"grad_norm": 1.3828125,
|
|
"learning_rate": 0.000132,
|
|
"loss": 6.851,
|
|
"mean_token_accuracy": 0.1115494042634964,
|
|
"num_tokens": 597572.0,
|
|
"step": 265
|
|
},
|
|
{
|
|
"entropy": 7.285667753219604,
|
|
"epoch": 0.2920497566252028,
|
|
"grad_norm": 1.34375,
|
|
"learning_rate": 0.00013450000000000002,
|
|
"loss": 6.8139,
|
|
"mean_token_accuracy": 0.11849615946412087,
|
|
"num_tokens": 608879.0,
|
|
"step": 270
|
|
},
|
|
{
|
|
"entropy": 7.397619819641113,
|
|
"epoch": 0.29745808545159547,
|
|
"grad_norm": 1.3203125,
|
|
"learning_rate": 0.00013700000000000002,
|
|
"loss": 6.8427,
|
|
"mean_token_accuracy": 0.10948013365268708,
|
|
"num_tokens": 620439.0,
|
|
"step": 275
|
|
},
|
|
{
|
|
"entropy": 7.3241359233856205,
|
|
"epoch": 0.3028664142779881,
|
|
"grad_norm": 1.2265625,
|
|
"learning_rate": 0.0001395,
|
|
"loss": 6.9064,
|
|
"mean_token_accuracy": 0.10746880844235421,
|
|
"num_tokens": 632543.0,
|
|
"step": 280
|
|
},
|
|
{
|
|
"entropy": 7.305241250991822,
|
|
"epoch": 0.30827474310438074,
|
|
"grad_norm": 1.1953125,
|
|
"learning_rate": 0.00014199999999999998,
|
|
"loss": 6.9523,
|
|
"mean_token_accuracy": 0.11276887506246566,
|
|
"num_tokens": 643369.0,
|
|
"step": 285
|
|
},
|
|
{
|
|
"entropy": 7.290642261505127,
|
|
"epoch": 0.3136830719307734,
|
|
"grad_norm": 1.03125,
|
|
"learning_rate": 0.0001445,
|
|
"loss": 6.8175,
|
|
"mean_token_accuracy": 0.11630758419632911,
|
|
"num_tokens": 655328.0,
|
|
"step": 290
|
|
},
|
|
{
|
|
"entropy": 7.304603433609008,
|
|
"epoch": 0.319091400757166,
|
|
"grad_norm": 1.3984375,
|
|
"learning_rate": 0.000147,
|
|
"loss": 6.8837,
|
|
"mean_token_accuracy": 0.1105578251183033,
|
|
"num_tokens": 665780.0,
|
|
"step": 295
|
|
},
|
|
{
|
|
"entropy": 7.252665376663208,
|
|
"epoch": 0.32449972958355866,
|
|
"grad_norm": 1.3046875,
|
|
"learning_rate": 0.0001495,
|
|
"loss": 6.7811,
|
|
"mean_token_accuracy": 0.11430338621139527,
|
|
"num_tokens": 676566.0,
|
|
"step": 300
|
|
},
|
|
{
|
|
"entropy": 7.213497543334961,
|
|
"epoch": 0.3299080584099513,
|
|
"grad_norm": 1.2265625,
|
|
"learning_rate": 0.000152,
|
|
"loss": 6.7701,
|
|
"mean_token_accuracy": 0.115247243642807,
|
|
"num_tokens": 687407.0,
|
|
"step": 305
|
|
},
|
|
{
|
|
"entropy": 7.21800651550293,
|
|
"epoch": 0.335316387236344,
|
|
"grad_norm": 1.28125,
|
|
"learning_rate": 0.00015450000000000001,
|
|
"loss": 6.7086,
|
|
"mean_token_accuracy": 0.11728021055459976,
|
|
"num_tokens": 698162.0,
|
|
"step": 310
|
|
},
|
|
{
|
|
"entropy": 7.1137049198150635,
|
|
"epoch": 0.34072471606273663,
|
|
"grad_norm": 1.3515625,
|
|
"learning_rate": 0.000157,
|
|
"loss": 6.6813,
|
|
"mean_token_accuracy": 0.11424923315644264,
|
|
"num_tokens": 708743.0,
|
|
"step": 315
|
|
},
|
|
{
|
|
"entropy": 7.287932205200195,
|
|
"epoch": 0.34613304488912927,
|
|
"grad_norm": 1.203125,
|
|
"learning_rate": 0.0001595,
|
|
"loss": 6.9224,
|
|
"mean_token_accuracy": 0.11441284045577049,
|
|
"num_tokens": 719828.0,
|
|
"step": 320
|
|
},
|
|
{
|
|
"entropy": 7.240033435821533,
|
|
"epoch": 0.3515413737155219,
|
|
"grad_norm": 1.1953125,
|
|
"learning_rate": 0.000162,
|
|
"loss": 6.8683,
|
|
"mean_token_accuracy": 0.11263540089130401,
|
|
"num_tokens": 732514.0,
|
|
"step": 325
|
|
},
|
|
{
|
|
"entropy": 7.1317566394805905,
|
|
"epoch": 0.35694970254191455,
|
|
"grad_norm": 1.3046875,
|
|
"learning_rate": 0.00016450000000000001,
|
|
"loss": 6.6928,
|
|
"mean_token_accuracy": 0.11701238229870796,
|
|
"num_tokens": 742391.0,
|
|
"step": 330
|
|
},
|
|
{
|
|
"entropy": 7.036797666549683,
|
|
"epoch": 0.3623580313683072,
|
|
"grad_norm": 1.3671875,
|
|
"learning_rate": 0.00016700000000000002,
|
|
"loss": 6.6964,
|
|
"mean_token_accuracy": 0.11464602425694466,
|
|
"num_tokens": 753638.0,
|
|
"step": 335
|
|
},
|
|
{
|
|
"entropy": 7.286917877197266,
|
|
"epoch": 0.3677663601946998,
|
|
"grad_norm": 1.21875,
|
|
"learning_rate": 0.00016950000000000003,
|
|
"loss": 6.9188,
|
|
"mean_token_accuracy": 0.11068255305290223,
|
|
"num_tokens": 764850.0,
|
|
"step": 340
|
|
},
|
|
{
|
|
"entropy": 7.178950262069702,
|
|
"epoch": 0.37317468902109246,
|
|
"grad_norm": 1.2578125,
|
|
"learning_rate": 0.00017199999999999998,
|
|
"loss": 6.6992,
|
|
"mean_token_accuracy": 0.1204465426504612,
|
|
"num_tokens": 776192.0,
|
|
"step": 345
|
|
},
|
|
{
|
|
"entropy": 6.9970947265625,
|
|
"epoch": 0.3785830178474851,
|
|
"grad_norm": 1.2265625,
|
|
"learning_rate": 0.00017449999999999999,
|
|
"loss": 6.6463,
|
|
"mean_token_accuracy": 0.12572311908006667,
|
|
"num_tokens": 787051.0,
|
|
"step": 350
|
|
},
|
|
{
|
|
"entropy": 7.054988002777099,
|
|
"epoch": 0.3839913466738778,
|
|
"grad_norm": 1.515625,
|
|
"learning_rate": 0.000177,
|
|
"loss": 6.6818,
|
|
"mean_token_accuracy": 0.1151166208088398,
|
|
"num_tokens": 799281.0,
|
|
"step": 355
|
|
},
|
|
{
|
|
"entropy": 7.076799201965332,
|
|
"epoch": 0.38939967550027044,
|
|
"grad_norm": 1.1171875,
|
|
"learning_rate": 0.0001795,
|
|
"loss": 6.7153,
|
|
"mean_token_accuracy": 0.11469916179776192,
|
|
"num_tokens": 810003.0,
|
|
"step": 360
|
|
},
|
|
{
|
|
"entropy": 6.988044166564942,
|
|
"epoch": 0.3948080043266631,
|
|
"grad_norm": 1.3984375,
|
|
"learning_rate": 0.000182,
|
|
"loss": 6.6715,
|
|
"mean_token_accuracy": 0.11684603095054627,
|
|
"num_tokens": 819749.0,
|
|
"step": 365
|
|
},
|
|
{
|
|
"entropy": 7.086754989624024,
|
|
"epoch": 0.4002163331530557,
|
|
"grad_norm": 1.328125,
|
|
"learning_rate": 0.0001845,
|
|
"loss": 6.669,
|
|
"mean_token_accuracy": 0.1218308962881565,
|
|
"num_tokens": 830552.0,
|
|
"step": 370
|
|
},
|
|
{
|
|
"entropy": 7.108803462982178,
|
|
"epoch": 0.40562466197944835,
|
|
"grad_norm": 1.328125,
|
|
"learning_rate": 0.000187,
|
|
"loss": 6.7743,
|
|
"mean_token_accuracy": 0.12103541642427444,
|
|
"num_tokens": 840495.0,
|
|
"step": 375
|
|
},
|
|
{
|
|
"entropy": 6.959353446960449,
|
|
"epoch": 0.411032990805841,
|
|
"grad_norm": 1.265625,
|
|
"learning_rate": 0.0001895,
|
|
"loss": 6.7243,
|
|
"mean_token_accuracy": 0.11957930251955987,
|
|
"num_tokens": 852915.0,
|
|
"step": 380
|
|
},
|
|
{
|
|
"entropy": 7.139857006072998,
|
|
"epoch": 0.41644131963223363,
|
|
"grad_norm": 1.25,
|
|
"learning_rate": 0.000192,
|
|
"loss": 6.6824,
|
|
"mean_token_accuracy": 0.11836005300283432,
|
|
"num_tokens": 862762.0,
|
|
"step": 385
|
|
},
|
|
{
|
|
"entropy": 6.925649166107178,
|
|
"epoch": 0.42184964845862627,
|
|
"grad_norm": 1.3671875,
|
|
"learning_rate": 0.0001945,
|
|
"loss": 6.6169,
|
|
"mean_token_accuracy": 0.11832203418016433,
|
|
"num_tokens": 873023.0,
|
|
"step": 390
|
|
},
|
|
{
|
|
"entropy": 7.009290456771851,
|
|
"epoch": 0.4272579772850189,
|
|
"grad_norm": 1.1640625,
|
|
"learning_rate": 0.00019700000000000002,
|
|
"loss": 6.7184,
|
|
"mean_token_accuracy": 0.11268399432301521,
|
|
"num_tokens": 885281.0,
|
|
"step": 395
|
|
},
|
|
{
|
|
"entropy": 7.027568912506103,
|
|
"epoch": 0.43266630611141155,
|
|
"grad_norm": 1.265625,
|
|
"learning_rate": 0.00019950000000000002,
|
|
"loss": 6.7191,
|
|
"mean_token_accuracy": 0.11810961216688157,
|
|
"num_tokens": 896210.0,
|
|
"step": 400
|
|
},
|
|
{
|
|
"entropy": 6.915582704544067,
|
|
"epoch": 0.43807463493780424,
|
|
"grad_norm": 1.109375,
|
|
"learning_rate": 0.000202,
|
|
"loss": 6.6367,
|
|
"mean_token_accuracy": 0.12111475020647049,
|
|
"num_tokens": 909028.0,
|
|
"step": 405
|
|
},
|
|
{
|
|
"entropy": 6.9255085468292235,
|
|
"epoch": 0.4434829637641969,
|
|
"grad_norm": 1.3984375,
|
|
"learning_rate": 0.00020449999999999998,
|
|
"loss": 6.5633,
|
|
"mean_token_accuracy": 0.12149589061737061,
|
|
"num_tokens": 919710.0,
|
|
"step": 410
|
|
},
|
|
{
|
|
"entropy": 7.012827968597412,
|
|
"epoch": 0.4488912925905895,
|
|
"grad_norm": 1.1015625,
|
|
"learning_rate": 0.000207,
|
|
"loss": 6.633,
|
|
"mean_token_accuracy": 0.1167762778699398,
|
|
"num_tokens": 930019.0,
|
|
"step": 415
|
|
},
|
|
{
|
|
"entropy": 6.988396644592285,
|
|
"epoch": 0.45429962141698216,
|
|
"grad_norm": 1.2109375,
|
|
"learning_rate": 0.0002095,
|
|
"loss": 6.5543,
|
|
"mean_token_accuracy": 0.12245730757713318,
|
|
"num_tokens": 940535.0,
|
|
"step": 420
|
|
},
|
|
{
|
|
"entropy": 6.710291481018066,
|
|
"epoch": 0.4597079502433748,
|
|
"grad_norm": 1.2578125,
|
|
"learning_rate": 0.000212,
|
|
"loss": 6.5134,
|
|
"mean_token_accuracy": 0.12176155820488929,
|
|
"num_tokens": 951390.0,
|
|
"step": 425
|
|
},
|
|
{
|
|
"entropy": 6.93554916381836,
|
|
"epoch": 0.46511627906976744,
|
|
"grad_norm": 1.328125,
|
|
"learning_rate": 0.0002145,
|
|
"loss": 6.5595,
|
|
"mean_token_accuracy": 0.12662623450160027,
|
|
"num_tokens": 963010.0,
|
|
"step": 430
|
|
},
|
|
{
|
|
"entropy": 6.83835015296936,
|
|
"epoch": 0.4705246078961601,
|
|
"grad_norm": 1.21875,
|
|
"learning_rate": 0.00021700000000000002,
|
|
"loss": 6.5177,
|
|
"mean_token_accuracy": 0.12881121560931205,
|
|
"num_tokens": 974032.0,
|
|
"step": 435
|
|
},
|
|
{
|
|
"entropy": 6.867484664916992,
|
|
"epoch": 0.4759329367225527,
|
|
"grad_norm": 1.390625,
|
|
"learning_rate": 0.0002195,
|
|
"loss": 6.5658,
|
|
"mean_token_accuracy": 0.12559181153774263,
|
|
"num_tokens": 983954.0,
|
|
"step": 440
|
|
},
|
|
{
|
|
"entropy": 6.875189256668091,
|
|
"epoch": 0.48134126554894535,
|
|
"grad_norm": 1.4765625,
|
|
"learning_rate": 0.000222,
|
|
"loss": 6.6071,
|
|
"mean_token_accuracy": 0.12600242644548415,
|
|
"num_tokens": 994135.0,
|
|
"step": 445
|
|
},
|
|
{
|
|
"entropy": 6.954481601715088,
|
|
"epoch": 0.48674959437533805,
|
|
"grad_norm": 1.2421875,
|
|
"learning_rate": 0.0002245,
|
|
"loss": 6.5621,
|
|
"mean_token_accuracy": 0.12190107628703117,
|
|
"num_tokens": 1006275.0,
|
|
"step": 450
|
|
},
|
|
{
|
|
"entropy": 6.6844987869262695,
|
|
"epoch": 0.4921579232017307,
|
|
"grad_norm": 1.296875,
|
|
"learning_rate": 0.00022700000000000002,
|
|
"loss": 6.3918,
|
|
"mean_token_accuracy": 0.13092931881546974,
|
|
"num_tokens": 1017767.0,
|
|
"step": 455
|
|
},
|
|
{
|
|
"entropy": 6.859748840332031,
|
|
"epoch": 0.4975662520281233,
|
|
"grad_norm": 1.328125,
|
|
"learning_rate": 0.00022950000000000002,
|
|
"loss": 6.5152,
|
|
"mean_token_accuracy": 0.1274128295481205,
|
|
"num_tokens": 1028286.0,
|
|
"step": 460
|
|
},
|
|
{
|
|
"entropy": 6.7749512672424315,
|
|
"epoch": 0.502974580854516,
|
|
"grad_norm": 1.3203125,
|
|
"learning_rate": 0.00023200000000000003,
|
|
"loss": 6.5196,
|
|
"mean_token_accuracy": 0.12368692606687545,
|
|
"num_tokens": 1040555.0,
|
|
"step": 465
|
|
},
|
|
{
|
|
"entropy": 6.847295665740967,
|
|
"epoch": 0.5083829096809086,
|
|
"grad_norm": 1.1640625,
|
|
"learning_rate": 0.00023449999999999998,
|
|
"loss": 6.6368,
|
|
"mean_token_accuracy": 0.12430229112505913,
|
|
"num_tokens": 1053668.0,
|
|
"step": 470
|
|
},
|
|
{
|
|
"entropy": 6.735847425460816,
|
|
"epoch": 0.5137912385073012,
|
|
"grad_norm": 1.421875,
|
|
"learning_rate": 0.000237,
|
|
"loss": 6.5764,
|
|
"mean_token_accuracy": 0.12110921442508697,
|
|
"num_tokens": 1066042.0,
|
|
"step": 475
|
|
},
|
|
{
|
|
"entropy": 6.780033254623413,
|
|
"epoch": 0.5191995673336939,
|
|
"grad_norm": 1.203125,
|
|
"learning_rate": 0.0002395,
|
|
"loss": 6.5933,
|
|
"mean_token_accuracy": 0.12213548719882965,
|
|
"num_tokens": 1077843.0,
|
|
"step": 480
|
|
},
|
|
{
|
|
"entropy": 6.815031719207764,
|
|
"epoch": 0.5246078961600865,
|
|
"grad_norm": 1.3046875,
|
|
"learning_rate": 0.000242,
|
|
"loss": 6.479,
|
|
"mean_token_accuracy": 0.12793946266174316,
|
|
"num_tokens": 1089448.0,
|
|
"step": 485
|
|
},
|
|
{
|
|
"entropy": 6.753958320617675,
|
|
"epoch": 0.5300162249864792,
|
|
"grad_norm": 1.1796875,
|
|
"learning_rate": 0.0002445,
|
|
"loss": 6.5871,
|
|
"mean_token_accuracy": 0.12737660706043244,
|
|
"num_tokens": 1100989.0,
|
|
"step": 490
|
|
},
|
|
{
|
|
"entropy": 6.929710721969604,
|
|
"epoch": 0.5354245538128718,
|
|
"grad_norm": 1.140625,
|
|
"learning_rate": 0.000247,
|
|
"loss": 6.6087,
|
|
"mean_token_accuracy": 0.12573689594864845,
|
|
"num_tokens": 1113012.0,
|
|
"step": 495
|
|
},
|
|
{
|
|
"entropy": 6.811409521102905,
|
|
"epoch": 0.5408328826392644,
|
|
"grad_norm": 1.125,
|
|
"learning_rate": 0.0002495,
|
|
"loss": 6.5551,
|
|
"mean_token_accuracy": 0.12299831435084344,
|
|
"num_tokens": 1123551.0,
|
|
"step": 500
|
|
},
|
|
{
|
|
"epoch": 0.5408328826392644,
|
|
"eval_entropy": 6.899274277687073,
|
|
"eval_loss": 6.698337554931641,
|
|
"eval_mean_token_accuracy": 0.12483972003683448,
|
|
"eval_num_tokens": 1123551.0,
|
|
"eval_runtime": 2.6565,
|
|
"eval_samples_per_second": 1323.152,
|
|
"eval_steps_per_second": 165.629,
|
|
"step": 500
|
|
}
|
|
],
|
|
"logging_steps": 5,
|
|
"max_steps": 9240,
|
|
"num_input_tokens_seen": 0,
|
|
"num_train_epochs": 10,
|
|
"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": 1751032221696000.0,
|
|
"train_batch_size": 16,
|
|
"trial_name": null,
|
|
"trial_params": null
|
|
}
|