1035 lines
27 KiB
JSON
1035 lines
27 KiB
JSON
{
|
|
"best_global_step": null,
|
|
"best_metric": null,
|
|
"best_model_checkpoint": null,
|
|
"epoch": 0.027957951241333035,
|
|
"eval_steps": 500,
|
|
"global_step": 500,
|
|
"is_hyper_param_search": false,
|
|
"is_local_process_zero": true,
|
|
"is_world_process_zero": true,
|
|
"log_history": [
|
|
{
|
|
"entropy": 10.692024612426758,
|
|
"epoch": 0.00027957951241333037,
|
|
"grad_norm": 13.5625,
|
|
"learning_rate": 2e-06,
|
|
"loss": 10.8007,
|
|
"mean_token_accuracy": 0.00010405827779322863,
|
|
"num_tokens": 8997.0,
|
|
"step": 5
|
|
},
|
|
{
|
|
"entropy": 10.691960906982422,
|
|
"epoch": 0.0005591590248266607,
|
|
"grad_norm": 13.875,
|
|
"learning_rate": 4.5e-06,
|
|
"loss": 10.7307,
|
|
"mean_token_accuracy": 0.0001552795059978962,
|
|
"num_tokens": 18058.0,
|
|
"step": 10
|
|
},
|
|
{
|
|
"entropy": 10.691014766693115,
|
|
"epoch": 0.0008387385372399911,
|
|
"grad_norm": 9.125,
|
|
"learning_rate": 7e-06,
|
|
"loss": 10.5,
|
|
"mean_token_accuracy": 0.023502371355425565,
|
|
"num_tokens": 27988.0,
|
|
"step": 15
|
|
},
|
|
{
|
|
"entropy": 10.680759143829345,
|
|
"epoch": 0.0011183180496533215,
|
|
"grad_norm": 6.28125,
|
|
"learning_rate": 9.5e-06,
|
|
"loss": 10.1358,
|
|
"mean_token_accuracy": 0.046215491555631164,
|
|
"num_tokens": 36654.0,
|
|
"step": 20
|
|
},
|
|
{
|
|
"entropy": 10.621339321136475,
|
|
"epoch": 0.0013978975620666518,
|
|
"grad_norm": 4.09375,
|
|
"learning_rate": 1.2e-05,
|
|
"loss": 9.9217,
|
|
"mean_token_accuracy": 0.04288874380290508,
|
|
"num_tokens": 45807.0,
|
|
"step": 25
|
|
},
|
|
{
|
|
"entropy": 10.59019603729248,
|
|
"epoch": 0.0016774770744799822,
|
|
"grad_norm": 3.65625,
|
|
"learning_rate": 1.4500000000000002e-05,
|
|
"loss": 9.7316,
|
|
"mean_token_accuracy": 0.04100009612739086,
|
|
"num_tokens": 55528.0,
|
|
"step": 30
|
|
},
|
|
{
|
|
"entropy": 10.609690189361572,
|
|
"epoch": 0.0019570565868933126,
|
|
"grad_norm": 3.078125,
|
|
"learning_rate": 1.7000000000000003e-05,
|
|
"loss": 9.7034,
|
|
"mean_token_accuracy": 0.041670043021440506,
|
|
"num_tokens": 66262.0,
|
|
"step": 35
|
|
},
|
|
{
|
|
"entropy": 10.595404052734375,
|
|
"epoch": 0.002236636099306643,
|
|
"grad_norm": 2.734375,
|
|
"learning_rate": 1.95e-05,
|
|
"loss": 9.6042,
|
|
"mean_token_accuracy": 0.03849245011806488,
|
|
"num_tokens": 75145.0,
|
|
"step": 40
|
|
},
|
|
{
|
|
"entropy": 10.588605785369873,
|
|
"epoch": 0.0025162156117199733,
|
|
"grad_norm": 2.59375,
|
|
"learning_rate": 2.2e-05,
|
|
"loss": 9.5456,
|
|
"mean_token_accuracy": 0.04283268712460995,
|
|
"num_tokens": 84519.0,
|
|
"step": 45
|
|
},
|
|
{
|
|
"entropy": 10.56249475479126,
|
|
"epoch": 0.0027957951241333037,
|
|
"grad_norm": 2.484375,
|
|
"learning_rate": 2.4500000000000003e-05,
|
|
"loss": 9.4937,
|
|
"mean_token_accuracy": 0.04488439559936523,
|
|
"num_tokens": 93968.0,
|
|
"step": 50
|
|
},
|
|
{
|
|
"entropy": 10.560630321502686,
|
|
"epoch": 0.003075374636546634,
|
|
"grad_norm": 2.5625,
|
|
"learning_rate": 2.7e-05,
|
|
"loss": 9.4295,
|
|
"mean_token_accuracy": 0.05186801627278328,
|
|
"num_tokens": 103228.0,
|
|
"step": 55
|
|
},
|
|
{
|
|
"entropy": 10.478568172454834,
|
|
"epoch": 0.0033549541489599644,
|
|
"grad_norm": 2.78125,
|
|
"learning_rate": 2.95e-05,
|
|
"loss": 9.3103,
|
|
"mean_token_accuracy": 0.05836689881980419,
|
|
"num_tokens": 112382.0,
|
|
"step": 60
|
|
},
|
|
{
|
|
"entropy": 10.372736740112305,
|
|
"epoch": 0.0036345336613732944,
|
|
"grad_norm": 2.21875,
|
|
"learning_rate": 3.2e-05,
|
|
"loss": 9.3096,
|
|
"mean_token_accuracy": 0.050225290656089785,
|
|
"num_tokens": 123006.0,
|
|
"step": 65
|
|
},
|
|
{
|
|
"entropy": 10.493916034698486,
|
|
"epoch": 0.003914113173786625,
|
|
"grad_norm": 2.34375,
|
|
"learning_rate": 3.4500000000000005e-05,
|
|
"loss": 9.2174,
|
|
"mean_token_accuracy": 0.054478041082620624,
|
|
"num_tokens": 132323.0,
|
|
"step": 70
|
|
},
|
|
{
|
|
"entropy": 10.459541130065919,
|
|
"epoch": 0.004193692686199955,
|
|
"grad_norm": 2.421875,
|
|
"learning_rate": 3.7e-05,
|
|
"loss": 9.0473,
|
|
"mean_token_accuracy": 0.06052669584751129,
|
|
"num_tokens": 142668.0,
|
|
"step": 75
|
|
},
|
|
{
|
|
"entropy": 10.37304391860962,
|
|
"epoch": 0.004473272198613286,
|
|
"grad_norm": 3.1875,
|
|
"learning_rate": 3.95e-05,
|
|
"loss": 8.9207,
|
|
"mean_token_accuracy": 0.06333019770681858,
|
|
"num_tokens": 152291.0,
|
|
"step": 80
|
|
},
|
|
{
|
|
"entropy": 10.375545501708984,
|
|
"epoch": 0.004752851711026616,
|
|
"grad_norm": 2.796875,
|
|
"learning_rate": 4.2000000000000004e-05,
|
|
"loss": 8.8031,
|
|
"mean_token_accuracy": 0.06564208678901196,
|
|
"num_tokens": 162443.0,
|
|
"step": 85
|
|
},
|
|
{
|
|
"entropy": 10.33843536376953,
|
|
"epoch": 0.005032431223439947,
|
|
"grad_norm": 2.453125,
|
|
"learning_rate": 4.45e-05,
|
|
"loss": 8.8512,
|
|
"mean_token_accuracy": 0.06008858419954777,
|
|
"num_tokens": 172011.0,
|
|
"step": 90
|
|
},
|
|
{
|
|
"entropy": 10.306972694396972,
|
|
"epoch": 0.005312010735853277,
|
|
"grad_norm": 2.46875,
|
|
"learning_rate": 4.7000000000000004e-05,
|
|
"loss": 8.6374,
|
|
"mean_token_accuracy": 0.0602202158421278,
|
|
"num_tokens": 180882.0,
|
|
"step": 95
|
|
},
|
|
{
|
|
"entropy": 10.262969303131104,
|
|
"epoch": 0.005591590248266607,
|
|
"grad_norm": 2.171875,
|
|
"learning_rate": 4.9500000000000004e-05,
|
|
"loss": 8.5492,
|
|
"mean_token_accuracy": 0.06696949824690819,
|
|
"num_tokens": 190318.0,
|
|
"step": 100
|
|
},
|
|
{
|
|
"entropy": 10.25103530883789,
|
|
"epoch": 0.005871169760679937,
|
|
"grad_norm": 2.8125,
|
|
"learning_rate": 5.2e-05,
|
|
"loss": 8.4281,
|
|
"mean_token_accuracy": 0.06335118524730206,
|
|
"num_tokens": 199369.0,
|
|
"step": 105
|
|
},
|
|
{
|
|
"entropy": 10.153358459472656,
|
|
"epoch": 0.006150749273093268,
|
|
"grad_norm": 2.359375,
|
|
"learning_rate": 5.45e-05,
|
|
"loss": 8.3094,
|
|
"mean_token_accuracy": 0.06752052381634713,
|
|
"num_tokens": 208848.0,
|
|
"step": 110
|
|
},
|
|
{
|
|
"entropy": 10.075451278686524,
|
|
"epoch": 0.006430328785506598,
|
|
"grad_norm": 2.21875,
|
|
"learning_rate": 5.7e-05,
|
|
"loss": 8.2362,
|
|
"mean_token_accuracy": 0.07109953351318836,
|
|
"num_tokens": 218285.0,
|
|
"step": 115
|
|
},
|
|
{
|
|
"entropy": 9.987942695617676,
|
|
"epoch": 0.006709908297919929,
|
|
"grad_norm": 2.28125,
|
|
"learning_rate": 5.9499999999999996e-05,
|
|
"loss": 8.08,
|
|
"mean_token_accuracy": 0.07430721223354339,
|
|
"num_tokens": 227033.0,
|
|
"step": 120
|
|
},
|
|
{
|
|
"entropy": 9.844741344451904,
|
|
"epoch": 0.006989487810333259,
|
|
"grad_norm": 1.9609375,
|
|
"learning_rate": 6.2e-05,
|
|
"loss": 7.8837,
|
|
"mean_token_accuracy": 0.0787685714662075,
|
|
"num_tokens": 236020.0,
|
|
"step": 125
|
|
},
|
|
{
|
|
"entropy": 9.748367309570312,
|
|
"epoch": 0.007269067322746589,
|
|
"grad_norm": 1.890625,
|
|
"learning_rate": 6.450000000000001e-05,
|
|
"loss": 7.8756,
|
|
"mean_token_accuracy": 0.0757671982049942,
|
|
"num_tokens": 244996.0,
|
|
"step": 130
|
|
},
|
|
{
|
|
"entropy": 9.509360885620117,
|
|
"epoch": 0.0075486468351599195,
|
|
"grad_norm": 1.578125,
|
|
"learning_rate": 6.7e-05,
|
|
"loss": 7.7588,
|
|
"mean_token_accuracy": 0.07781498581171035,
|
|
"num_tokens": 254721.0,
|
|
"step": 135
|
|
},
|
|
{
|
|
"entropy": 9.336756896972656,
|
|
"epoch": 0.00782822634757325,
|
|
"grad_norm": 1.9140625,
|
|
"learning_rate": 6.950000000000001e-05,
|
|
"loss": 7.5597,
|
|
"mean_token_accuracy": 0.07736790589988232,
|
|
"num_tokens": 264203.0,
|
|
"step": 140
|
|
},
|
|
{
|
|
"entropy": 9.150464534759521,
|
|
"epoch": 0.00810780585998658,
|
|
"grad_norm": 1.5703125,
|
|
"learning_rate": 7.2e-05,
|
|
"loss": 7.5539,
|
|
"mean_token_accuracy": 0.07673263326287269,
|
|
"num_tokens": 274136.0,
|
|
"step": 145
|
|
},
|
|
{
|
|
"entropy": 8.9904709815979,
|
|
"epoch": 0.00838738537239991,
|
|
"grad_norm": 1.6484375,
|
|
"learning_rate": 7.45e-05,
|
|
"loss": 7.5468,
|
|
"mean_token_accuracy": 0.07521282956004142,
|
|
"num_tokens": 284089.0,
|
|
"step": 150
|
|
},
|
|
{
|
|
"entropy": 8.798522186279296,
|
|
"epoch": 0.008666964884813241,
|
|
"grad_norm": 1.5546875,
|
|
"learning_rate": 7.7e-05,
|
|
"loss": 7.3354,
|
|
"mean_token_accuracy": 0.08436919301748276,
|
|
"num_tokens": 292975.0,
|
|
"step": 155
|
|
},
|
|
{
|
|
"entropy": 8.52847204208374,
|
|
"epoch": 0.008946544397226572,
|
|
"grad_norm": 1.5078125,
|
|
"learning_rate": 7.950000000000001e-05,
|
|
"loss": 7.3229,
|
|
"mean_token_accuracy": 0.08412722572684288,
|
|
"num_tokens": 301900.0,
|
|
"step": 160
|
|
},
|
|
{
|
|
"entropy": 8.361807441711425,
|
|
"epoch": 0.0092261239096399,
|
|
"grad_norm": 1.5546875,
|
|
"learning_rate": 8.2e-05,
|
|
"loss": 7.3062,
|
|
"mean_token_accuracy": 0.07742807678878308,
|
|
"num_tokens": 312052.0,
|
|
"step": 165
|
|
},
|
|
{
|
|
"entropy": 8.169294548034667,
|
|
"epoch": 0.009505703422053232,
|
|
"grad_norm": 1.4296875,
|
|
"learning_rate": 8.450000000000001e-05,
|
|
"loss": 7.0957,
|
|
"mean_token_accuracy": 0.08770816624164582,
|
|
"num_tokens": 321834.0,
|
|
"step": 170
|
|
},
|
|
{
|
|
"entropy": 8.13689374923706,
|
|
"epoch": 0.009785282934466562,
|
|
"grad_norm": 1.453125,
|
|
"learning_rate": 8.7e-05,
|
|
"loss": 7.0995,
|
|
"mean_token_accuracy": 0.08651999905705451,
|
|
"num_tokens": 331806.0,
|
|
"step": 175
|
|
},
|
|
{
|
|
"entropy": 7.894659185409546,
|
|
"epoch": 0.010064862446879893,
|
|
"grad_norm": 1.7421875,
|
|
"learning_rate": 8.95e-05,
|
|
"loss": 7.0891,
|
|
"mean_token_accuracy": 0.08311521410942077,
|
|
"num_tokens": 341015.0,
|
|
"step": 180
|
|
},
|
|
{
|
|
"entropy": 7.842967796325683,
|
|
"epoch": 0.010344441959293222,
|
|
"grad_norm": 1.1796875,
|
|
"learning_rate": 9.2e-05,
|
|
"loss": 7.1195,
|
|
"mean_token_accuracy": 0.085103327780962,
|
|
"num_tokens": 350653.0,
|
|
"step": 185
|
|
},
|
|
{
|
|
"entropy": 7.891383028030395,
|
|
"epoch": 0.010624021471706553,
|
|
"grad_norm": 1.5625,
|
|
"learning_rate": 9.45e-05,
|
|
"loss": 7.2382,
|
|
"mean_token_accuracy": 0.07780816406011581,
|
|
"num_tokens": 360090.0,
|
|
"step": 190
|
|
},
|
|
{
|
|
"entropy": 7.868162822723389,
|
|
"epoch": 0.010903600984119884,
|
|
"grad_norm": 1.3359375,
|
|
"learning_rate": 9.7e-05,
|
|
"loss": 6.9814,
|
|
"mean_token_accuracy": 0.0890000157058239,
|
|
"num_tokens": 370265.0,
|
|
"step": 195
|
|
},
|
|
{
|
|
"entropy": 7.779269695281982,
|
|
"epoch": 0.011183180496533215,
|
|
"grad_norm": 2.140625,
|
|
"learning_rate": 9.95e-05,
|
|
"loss": 7.0828,
|
|
"mean_token_accuracy": 0.08332997038960457,
|
|
"num_tokens": 379060.0,
|
|
"step": 200
|
|
},
|
|
{
|
|
"entropy": 7.6772924900054935,
|
|
"epoch": 0.011462760008946544,
|
|
"grad_norm": 1.234375,
|
|
"learning_rate": 0.000102,
|
|
"loss": 7.07,
|
|
"mean_token_accuracy": 0.08489679098129273,
|
|
"num_tokens": 388621.0,
|
|
"step": 205
|
|
},
|
|
{
|
|
"entropy": 7.725631523132324,
|
|
"epoch": 0.011742339521359875,
|
|
"grad_norm": 1.3984375,
|
|
"learning_rate": 0.00010449999999999999,
|
|
"loss": 7.0195,
|
|
"mean_token_accuracy": 0.08262372985482216,
|
|
"num_tokens": 397708.0,
|
|
"step": 210
|
|
},
|
|
{
|
|
"entropy": 7.685599374771118,
|
|
"epoch": 0.012021919033773205,
|
|
"grad_norm": 1.34375,
|
|
"learning_rate": 0.000107,
|
|
"loss": 7.0448,
|
|
"mean_token_accuracy": 0.08679846599698067,
|
|
"num_tokens": 406979.0,
|
|
"step": 215
|
|
},
|
|
{
|
|
"entropy": 7.672210216522217,
|
|
"epoch": 0.012301498546186536,
|
|
"grad_norm": 1.484375,
|
|
"learning_rate": 0.0001095,
|
|
"loss": 7.0787,
|
|
"mean_token_accuracy": 0.08454066962003708,
|
|
"num_tokens": 416150.0,
|
|
"step": 220
|
|
},
|
|
{
|
|
"entropy": 7.62153000831604,
|
|
"epoch": 0.012581078058599865,
|
|
"grad_norm": 1.3046875,
|
|
"learning_rate": 0.000112,
|
|
"loss": 6.9668,
|
|
"mean_token_accuracy": 0.08243566080927849,
|
|
"num_tokens": 425052.0,
|
|
"step": 225
|
|
},
|
|
{
|
|
"entropy": 7.601655292510986,
|
|
"epoch": 0.012860657571013196,
|
|
"grad_norm": 1.3046875,
|
|
"learning_rate": 0.0001145,
|
|
"loss": 7.0291,
|
|
"mean_token_accuracy": 0.089765153080225,
|
|
"num_tokens": 433950.0,
|
|
"step": 230
|
|
},
|
|
{
|
|
"entropy": 7.578113460540772,
|
|
"epoch": 0.013140237083426527,
|
|
"grad_norm": 1.375,
|
|
"learning_rate": 0.00011700000000000001,
|
|
"loss": 6.9937,
|
|
"mean_token_accuracy": 0.08947755619883538,
|
|
"num_tokens": 443400.0,
|
|
"step": 235
|
|
},
|
|
{
|
|
"entropy": 7.512868022918701,
|
|
"epoch": 0.013419816595839858,
|
|
"grad_norm": 2.046875,
|
|
"learning_rate": 0.00011949999999999999,
|
|
"loss": 7.0135,
|
|
"mean_token_accuracy": 0.08865847960114479,
|
|
"num_tokens": 452965.0,
|
|
"step": 240
|
|
},
|
|
{
|
|
"entropy": 7.477978706359863,
|
|
"epoch": 0.013699396108253187,
|
|
"grad_norm": 1.484375,
|
|
"learning_rate": 0.000122,
|
|
"loss": 6.9064,
|
|
"mean_token_accuracy": 0.08826079666614532,
|
|
"num_tokens": 461926.0,
|
|
"step": 245
|
|
},
|
|
{
|
|
"entropy": 7.451308345794677,
|
|
"epoch": 0.013978975620666518,
|
|
"grad_norm": 1.1796875,
|
|
"learning_rate": 0.0001245,
|
|
"loss": 6.886,
|
|
"mean_token_accuracy": 0.0926682025194168,
|
|
"num_tokens": 471013.0,
|
|
"step": 250
|
|
},
|
|
{
|
|
"entropy": 7.453368616104126,
|
|
"epoch": 0.014258555133079848,
|
|
"grad_norm": 1.4140625,
|
|
"learning_rate": 0.000127,
|
|
"loss": 6.8576,
|
|
"mean_token_accuracy": 0.09064204692840576,
|
|
"num_tokens": 480800.0,
|
|
"step": 255
|
|
},
|
|
{
|
|
"entropy": 7.39466347694397,
|
|
"epoch": 0.014538134645493177,
|
|
"grad_norm": 1.46875,
|
|
"learning_rate": 0.0001295,
|
|
"loss": 6.8605,
|
|
"mean_token_accuracy": 0.08739916756749153,
|
|
"num_tokens": 489410.0,
|
|
"step": 260
|
|
},
|
|
{
|
|
"entropy": 7.416737365722656,
|
|
"epoch": 0.014817714157906508,
|
|
"grad_norm": 1.28125,
|
|
"learning_rate": 0.000132,
|
|
"loss": 6.9196,
|
|
"mean_token_accuracy": 0.09045738354325294,
|
|
"num_tokens": 498568.0,
|
|
"step": 265
|
|
},
|
|
{
|
|
"entropy": 7.417768335342407,
|
|
"epoch": 0.015097293670319839,
|
|
"grad_norm": 1.234375,
|
|
"learning_rate": 0.00013450000000000002,
|
|
"loss": 6.8301,
|
|
"mean_token_accuracy": 0.09208849593997001,
|
|
"num_tokens": 508412.0,
|
|
"step": 270
|
|
},
|
|
{
|
|
"entropy": 7.365527629852295,
|
|
"epoch": 0.01537687318273317,
|
|
"grad_norm": 1.265625,
|
|
"learning_rate": 0.00013700000000000002,
|
|
"loss": 6.7434,
|
|
"mean_token_accuracy": 0.09199867770075798,
|
|
"num_tokens": 516447.0,
|
|
"step": 275
|
|
},
|
|
{
|
|
"entropy": 7.340102052688598,
|
|
"epoch": 0.0156564526951465,
|
|
"grad_norm": 1.4609375,
|
|
"learning_rate": 0.0001395,
|
|
"loss": 6.8042,
|
|
"mean_token_accuracy": 0.09145343601703644,
|
|
"num_tokens": 525352.0,
|
|
"step": 280
|
|
},
|
|
{
|
|
"entropy": 7.316589975357056,
|
|
"epoch": 0.01593603220755983,
|
|
"grad_norm": 1.515625,
|
|
"learning_rate": 0.00014199999999999998,
|
|
"loss": 6.823,
|
|
"mean_token_accuracy": 0.09479128122329712,
|
|
"num_tokens": 534827.0,
|
|
"step": 285
|
|
},
|
|
{
|
|
"entropy": 7.141154193878174,
|
|
"epoch": 0.01621561171997316,
|
|
"grad_norm": 1.328125,
|
|
"learning_rate": 0.0001445,
|
|
"loss": 6.7454,
|
|
"mean_token_accuracy": 0.0958427369594574,
|
|
"num_tokens": 543666.0,
|
|
"step": 290
|
|
},
|
|
{
|
|
"entropy": 7.3483844757080075,
|
|
"epoch": 0.01649519123238649,
|
|
"grad_norm": 1.2109375,
|
|
"learning_rate": 0.000147,
|
|
"loss": 6.8936,
|
|
"mean_token_accuracy": 0.08978378623723984,
|
|
"num_tokens": 553296.0,
|
|
"step": 295
|
|
},
|
|
{
|
|
"entropy": 7.285057926177979,
|
|
"epoch": 0.01677477074479982,
|
|
"grad_norm": 1.40625,
|
|
"learning_rate": 0.0001495,
|
|
"loss": 6.7804,
|
|
"mean_token_accuracy": 0.09113546311855317,
|
|
"num_tokens": 562053.0,
|
|
"step": 300
|
|
},
|
|
{
|
|
"entropy": 7.211677932739258,
|
|
"epoch": 0.017054350257213153,
|
|
"grad_norm": 1.296875,
|
|
"learning_rate": 0.000152,
|
|
"loss": 6.6528,
|
|
"mean_token_accuracy": 0.09451861307024956,
|
|
"num_tokens": 571090.0,
|
|
"step": 305
|
|
},
|
|
{
|
|
"entropy": 7.283277893066407,
|
|
"epoch": 0.017333929769626482,
|
|
"grad_norm": 1.1484375,
|
|
"learning_rate": 0.00015450000000000001,
|
|
"loss": 6.8128,
|
|
"mean_token_accuracy": 0.08825366944074631,
|
|
"num_tokens": 580432.0,
|
|
"step": 310
|
|
},
|
|
{
|
|
"entropy": 7.061421537399292,
|
|
"epoch": 0.01761350928203981,
|
|
"grad_norm": 1.3046875,
|
|
"learning_rate": 0.000157,
|
|
"loss": 6.6328,
|
|
"mean_token_accuracy": 0.09620391875505448,
|
|
"num_tokens": 589977.0,
|
|
"step": 315
|
|
},
|
|
{
|
|
"entropy": 7.243497753143311,
|
|
"epoch": 0.017893088794453144,
|
|
"grad_norm": 1.5234375,
|
|
"learning_rate": 0.0001595,
|
|
"loss": 6.8048,
|
|
"mean_token_accuracy": 0.09028137400746346,
|
|
"num_tokens": 599676.0,
|
|
"step": 320
|
|
},
|
|
{
|
|
"entropy": 7.100687837600708,
|
|
"epoch": 0.018172668306866473,
|
|
"grad_norm": 1.3203125,
|
|
"learning_rate": 0.000162,
|
|
"loss": 6.6098,
|
|
"mean_token_accuracy": 0.09744876846671105,
|
|
"num_tokens": 609037.0,
|
|
"step": 325
|
|
},
|
|
{
|
|
"entropy": 7.148408699035644,
|
|
"epoch": 0.0184522478192798,
|
|
"grad_norm": 1.2890625,
|
|
"learning_rate": 0.00016450000000000001,
|
|
"loss": 6.7023,
|
|
"mean_token_accuracy": 0.09769300967454911,
|
|
"num_tokens": 618191.0,
|
|
"step": 330
|
|
},
|
|
{
|
|
"entropy": 7.204887104034424,
|
|
"epoch": 0.018731827331693134,
|
|
"grad_norm": 1.4765625,
|
|
"learning_rate": 0.00016700000000000002,
|
|
"loss": 6.707,
|
|
"mean_token_accuracy": 0.09259288236498833,
|
|
"num_tokens": 627463.0,
|
|
"step": 335
|
|
},
|
|
{
|
|
"entropy": 7.125395584106445,
|
|
"epoch": 0.019011406844106463,
|
|
"grad_norm": 1.28125,
|
|
"learning_rate": 0.00016950000000000003,
|
|
"loss": 6.7114,
|
|
"mean_token_accuracy": 0.09360760822892189,
|
|
"num_tokens": 636421.0,
|
|
"step": 340
|
|
},
|
|
{
|
|
"entropy": 7.173436880111694,
|
|
"epoch": 0.019290986356519796,
|
|
"grad_norm": 1.2421875,
|
|
"learning_rate": 0.00017199999999999998,
|
|
"loss": 6.7133,
|
|
"mean_token_accuracy": 0.0982758454978466,
|
|
"num_tokens": 646378.0,
|
|
"step": 345
|
|
},
|
|
{
|
|
"entropy": 7.1207976818084715,
|
|
"epoch": 0.019570565868933125,
|
|
"grad_norm": 1.2578125,
|
|
"learning_rate": 0.00017449999999999999,
|
|
"loss": 6.7165,
|
|
"mean_token_accuracy": 0.09481776431202889,
|
|
"num_tokens": 655706.0,
|
|
"step": 350
|
|
},
|
|
{
|
|
"entropy": 7.053719902038575,
|
|
"epoch": 0.019850145381346454,
|
|
"grad_norm": 1.3671875,
|
|
"learning_rate": 0.000177,
|
|
"loss": 6.6744,
|
|
"mean_token_accuracy": 0.09270476400852204,
|
|
"num_tokens": 664903.0,
|
|
"step": 355
|
|
},
|
|
{
|
|
"entropy": 7.0642224788665775,
|
|
"epoch": 0.020129724893759787,
|
|
"grad_norm": 1.3515625,
|
|
"learning_rate": 0.0001795,
|
|
"loss": 6.6401,
|
|
"mean_token_accuracy": 0.09566896334290505,
|
|
"num_tokens": 674144.0,
|
|
"step": 360
|
|
},
|
|
{
|
|
"entropy": 7.060983371734619,
|
|
"epoch": 0.020409304406173116,
|
|
"grad_norm": 1.6640625,
|
|
"learning_rate": 0.000182,
|
|
"loss": 6.6035,
|
|
"mean_token_accuracy": 0.10136298835277557,
|
|
"num_tokens": 682427.0,
|
|
"step": 365
|
|
},
|
|
{
|
|
"entropy": 7.124999570846557,
|
|
"epoch": 0.020688883918586445,
|
|
"grad_norm": 1.265625,
|
|
"learning_rate": 0.0001845,
|
|
"loss": 6.6969,
|
|
"mean_token_accuracy": 0.09326310679316521,
|
|
"num_tokens": 691388.0,
|
|
"step": 370
|
|
},
|
|
{
|
|
"entropy": 7.0904834270477295,
|
|
"epoch": 0.020968463430999777,
|
|
"grad_norm": 1.3046875,
|
|
"learning_rate": 0.000187,
|
|
"loss": 6.6528,
|
|
"mean_token_accuracy": 0.09667944237589836,
|
|
"num_tokens": 701182.0,
|
|
"step": 375
|
|
},
|
|
{
|
|
"entropy": 7.035393142700196,
|
|
"epoch": 0.021248042943413106,
|
|
"grad_norm": 1.4453125,
|
|
"learning_rate": 0.0001895,
|
|
"loss": 6.6434,
|
|
"mean_token_accuracy": 0.0958517998456955,
|
|
"num_tokens": 710348.0,
|
|
"step": 380
|
|
},
|
|
{
|
|
"entropy": 6.9457237243652346,
|
|
"epoch": 0.021527622455826435,
|
|
"grad_norm": 1.4453125,
|
|
"learning_rate": 0.000192,
|
|
"loss": 6.6466,
|
|
"mean_token_accuracy": 0.09273892566561699,
|
|
"num_tokens": 720718.0,
|
|
"step": 385
|
|
},
|
|
{
|
|
"entropy": 7.007018089294434,
|
|
"epoch": 0.021807201968239768,
|
|
"grad_norm": 1.3203125,
|
|
"learning_rate": 0.0001945,
|
|
"loss": 6.5941,
|
|
"mean_token_accuracy": 0.10153442919254303,
|
|
"num_tokens": 729450.0,
|
|
"step": 390
|
|
},
|
|
{
|
|
"entropy": 7.075080108642578,
|
|
"epoch": 0.022086781480653097,
|
|
"grad_norm": 1.359375,
|
|
"learning_rate": 0.00019700000000000002,
|
|
"loss": 6.617,
|
|
"mean_token_accuracy": 0.10214217454195022,
|
|
"num_tokens": 738474.0,
|
|
"step": 395
|
|
},
|
|
{
|
|
"entropy": 6.958080196380616,
|
|
"epoch": 0.02236636099306643,
|
|
"grad_norm": 1.5546875,
|
|
"learning_rate": 0.00019950000000000002,
|
|
"loss": 6.6101,
|
|
"mean_token_accuracy": 0.09967887029051781,
|
|
"num_tokens": 748305.0,
|
|
"step": 400
|
|
},
|
|
{
|
|
"entropy": 6.9478137493133545,
|
|
"epoch": 0.02264594050547976,
|
|
"grad_norm": 1.140625,
|
|
"learning_rate": 0.000202,
|
|
"loss": 6.6537,
|
|
"mean_token_accuracy": 0.0958751380443573,
|
|
"num_tokens": 758173.0,
|
|
"step": 405
|
|
},
|
|
{
|
|
"entropy": 6.986770677566528,
|
|
"epoch": 0.022925520017893088,
|
|
"grad_norm": 1.359375,
|
|
"learning_rate": 0.00020449999999999998,
|
|
"loss": 6.7001,
|
|
"mean_token_accuracy": 0.09021164998412132,
|
|
"num_tokens": 768082.0,
|
|
"step": 410
|
|
},
|
|
{
|
|
"entropy": 7.0162006378173825,
|
|
"epoch": 0.02320509953030642,
|
|
"grad_norm": 1.25,
|
|
"learning_rate": 0.000207,
|
|
"loss": 6.6369,
|
|
"mean_token_accuracy": 0.0928497076034546,
|
|
"num_tokens": 778575.0,
|
|
"step": 415
|
|
},
|
|
{
|
|
"entropy": 6.848886346817016,
|
|
"epoch": 0.02348467904271975,
|
|
"grad_norm": 1.2890625,
|
|
"learning_rate": 0.0002095,
|
|
"loss": 6.553,
|
|
"mean_token_accuracy": 0.09735222682356834,
|
|
"num_tokens": 788164.0,
|
|
"step": 420
|
|
},
|
|
{
|
|
"entropy": 6.993223476409912,
|
|
"epoch": 0.02376425855513308,
|
|
"grad_norm": 1.1171875,
|
|
"learning_rate": 0.000212,
|
|
"loss": 6.524,
|
|
"mean_token_accuracy": 0.10375597029924392,
|
|
"num_tokens": 798028.0,
|
|
"step": 425
|
|
},
|
|
{
|
|
"entropy": 6.933980226516724,
|
|
"epoch": 0.02404383806754641,
|
|
"grad_norm": 1.359375,
|
|
"learning_rate": 0.0002145,
|
|
"loss": 6.5493,
|
|
"mean_token_accuracy": 0.09785934314131736,
|
|
"num_tokens": 807540.0,
|
|
"step": 430
|
|
},
|
|
{
|
|
"entropy": 6.838881254196167,
|
|
"epoch": 0.02432341757995974,
|
|
"grad_norm": 1.2421875,
|
|
"learning_rate": 0.00021700000000000002,
|
|
"loss": 6.5584,
|
|
"mean_token_accuracy": 0.09884036779403686,
|
|
"num_tokens": 817649.0,
|
|
"step": 435
|
|
},
|
|
{
|
|
"entropy": 6.971255254745484,
|
|
"epoch": 0.024602997092373072,
|
|
"grad_norm": 1.421875,
|
|
"learning_rate": 0.0002195,
|
|
"loss": 6.5859,
|
|
"mean_token_accuracy": 0.10443184226751327,
|
|
"num_tokens": 826410.0,
|
|
"step": 440
|
|
},
|
|
{
|
|
"entropy": 6.933424615859986,
|
|
"epoch": 0.0248825766047864,
|
|
"grad_norm": 1.3125,
|
|
"learning_rate": 0.000222,
|
|
"loss": 6.6334,
|
|
"mean_token_accuracy": 0.09922882169485092,
|
|
"num_tokens": 836299.0,
|
|
"step": 445
|
|
},
|
|
{
|
|
"entropy": 6.831998491287232,
|
|
"epoch": 0.02516215611719973,
|
|
"grad_norm": 1.4296875,
|
|
"learning_rate": 0.0002245,
|
|
"loss": 6.4622,
|
|
"mean_token_accuracy": 0.10322369784116744,
|
|
"num_tokens": 845199.0,
|
|
"step": 450
|
|
},
|
|
{
|
|
"entropy": 6.814246892929077,
|
|
"epoch": 0.025441735629613063,
|
|
"grad_norm": 1.2421875,
|
|
"learning_rate": 0.00022700000000000002,
|
|
"loss": 6.4667,
|
|
"mean_token_accuracy": 0.09550682380795479,
|
|
"num_tokens": 854235.0,
|
|
"step": 455
|
|
},
|
|
{
|
|
"entropy": 6.9324622631073,
|
|
"epoch": 0.025721315142026392,
|
|
"grad_norm": 1.3671875,
|
|
"learning_rate": 0.00022950000000000002,
|
|
"loss": 6.5738,
|
|
"mean_token_accuracy": 0.09983837381005287,
|
|
"num_tokens": 862540.0,
|
|
"step": 460
|
|
},
|
|
{
|
|
"entropy": 6.814285469055176,
|
|
"epoch": 0.02600089465443972,
|
|
"grad_norm": 1.2734375,
|
|
"learning_rate": 0.00023200000000000003,
|
|
"loss": 6.4353,
|
|
"mean_token_accuracy": 0.10586772412061692,
|
|
"num_tokens": 871352.0,
|
|
"step": 465
|
|
},
|
|
{
|
|
"entropy": 6.827783298492432,
|
|
"epoch": 0.026280474166853054,
|
|
"grad_norm": 1.265625,
|
|
"learning_rate": 0.00023449999999999998,
|
|
"loss": 6.4619,
|
|
"mean_token_accuracy": 0.10347421318292618,
|
|
"num_tokens": 880004.0,
|
|
"step": 470
|
|
},
|
|
{
|
|
"entropy": 6.814178466796875,
|
|
"epoch": 0.026560053679266383,
|
|
"grad_norm": 1.078125,
|
|
"learning_rate": 0.000237,
|
|
"loss": 6.46,
|
|
"mean_token_accuracy": 0.09918070882558823,
|
|
"num_tokens": 890392.0,
|
|
"step": 475
|
|
},
|
|
{
|
|
"entropy": 6.833171796798706,
|
|
"epoch": 0.026839633191679715,
|
|
"grad_norm": 1.21875,
|
|
"learning_rate": 0.0002395,
|
|
"loss": 6.5195,
|
|
"mean_token_accuracy": 0.09971475973725319,
|
|
"num_tokens": 899442.0,
|
|
"step": 480
|
|
},
|
|
{
|
|
"entropy": 6.814929580688476,
|
|
"epoch": 0.027119212704093044,
|
|
"grad_norm": 1.3984375,
|
|
"learning_rate": 0.000242,
|
|
"loss": 6.4525,
|
|
"mean_token_accuracy": 0.10721626579761505,
|
|
"num_tokens": 907440.0,
|
|
"step": 485
|
|
},
|
|
{
|
|
"entropy": 6.789797449111939,
|
|
"epoch": 0.027398792216506374,
|
|
"grad_norm": 1.328125,
|
|
"learning_rate": 0.0002445,
|
|
"loss": 6.6721,
|
|
"mean_token_accuracy": 0.09315093532204628,
|
|
"num_tokens": 917734.0,
|
|
"step": 490
|
|
},
|
|
{
|
|
"entropy": 6.8634199619293215,
|
|
"epoch": 0.027678371728919706,
|
|
"grad_norm": 1.4296875,
|
|
"learning_rate": 0.000247,
|
|
"loss": 6.5902,
|
|
"mean_token_accuracy": 0.10036376044154167,
|
|
"num_tokens": 927717.0,
|
|
"step": 495
|
|
},
|
|
{
|
|
"entropy": 6.807837009429932,
|
|
"epoch": 0.027957951241333035,
|
|
"grad_norm": 1.203125,
|
|
"learning_rate": 0.0002495,
|
|
"loss": 6.5431,
|
|
"mean_token_accuracy": 0.09580114185810089,
|
|
"num_tokens": 937385.0,
|
|
"step": 500
|
|
}
|
|
],
|
|
"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": 1344061716480000.0,
|
|
"train_batch_size": 16,
|
|
"trial_name": null,
|
|
"trial_params": null
|
|
}
|