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": 10.691921043395997,
|
|
"epoch": 0.00027957951241333037,
|
|
"grad_norm": 12.625,
|
|
"learning_rate": 2e-06,
|
|
"loss": 10.8124,
|
|
"mean_token_accuracy": 0.0,
|
|
"num_tokens": 9861.0,
|
|
"step": 5
|
|
},
|
|
{
|
|
"entropy": 10.691890811920166,
|
|
"epoch": 0.0005591590248266607,
|
|
"grad_norm": 13.375,
|
|
"learning_rate": 4.5e-06,
|
|
"loss": 10.7399,
|
|
"mean_token_accuracy": 0.000473622081335634,
|
|
"num_tokens": 19476.0,
|
|
"step": 10
|
|
},
|
|
{
|
|
"entropy": 10.691054916381836,
|
|
"epoch": 0.0008387385372399911,
|
|
"grad_norm": 9.625,
|
|
"learning_rate": 7e-06,
|
|
"loss": 10.506,
|
|
"mean_token_accuracy": 0.02479663831181824,
|
|
"num_tokens": 29731.0,
|
|
"step": 15
|
|
},
|
|
{
|
|
"entropy": 10.682694625854491,
|
|
"epoch": 0.0011183180496533215,
|
|
"grad_norm": 6.4375,
|
|
"learning_rate": 9.5e-06,
|
|
"loss": 10.1591,
|
|
"mean_token_accuracy": 0.03802054561674595,
|
|
"num_tokens": 38643.0,
|
|
"step": 20
|
|
},
|
|
{
|
|
"entropy": 10.634562492370605,
|
|
"epoch": 0.0013978975620666518,
|
|
"grad_norm": 4.28125,
|
|
"learning_rate": 1.2e-05,
|
|
"loss": 9.8669,
|
|
"mean_token_accuracy": 0.04109327364712954,
|
|
"num_tokens": 47707.0,
|
|
"step": 25
|
|
},
|
|
{
|
|
"entropy": 10.554908943176269,
|
|
"epoch": 0.0016774770744799822,
|
|
"grad_norm": 3.328125,
|
|
"learning_rate": 1.4500000000000002e-05,
|
|
"loss": 9.7299,
|
|
"mean_token_accuracy": 0.04091411866247654,
|
|
"num_tokens": 57219.0,
|
|
"step": 30
|
|
},
|
|
{
|
|
"entropy": 10.570738887786865,
|
|
"epoch": 0.0019570565868933126,
|
|
"grad_norm": 2.90625,
|
|
"learning_rate": 1.7000000000000003e-05,
|
|
"loss": 9.6768,
|
|
"mean_token_accuracy": 0.03916576430201531,
|
|
"num_tokens": 66318.0,
|
|
"step": 35
|
|
},
|
|
{
|
|
"entropy": 10.571315002441406,
|
|
"epoch": 0.002236636099306643,
|
|
"grad_norm": 2.890625,
|
|
"learning_rate": 1.95e-05,
|
|
"loss": 9.5213,
|
|
"mean_token_accuracy": 0.03872227370738983,
|
|
"num_tokens": 74602.0,
|
|
"step": 40
|
|
},
|
|
{
|
|
"entropy": 10.566377449035645,
|
|
"epoch": 0.0025162156117199733,
|
|
"grad_norm": 2.640625,
|
|
"learning_rate": 2.2e-05,
|
|
"loss": 9.5088,
|
|
"mean_token_accuracy": 0.04258945845067501,
|
|
"num_tokens": 84666.0,
|
|
"step": 45
|
|
},
|
|
{
|
|
"entropy": 10.548401355743408,
|
|
"epoch": 0.0027957951241333037,
|
|
"grad_norm": 2.671875,
|
|
"learning_rate": 2.4500000000000003e-05,
|
|
"loss": 9.4353,
|
|
"mean_token_accuracy": 0.0500976238399744,
|
|
"num_tokens": 94488.0,
|
|
"step": 50
|
|
},
|
|
{
|
|
"entropy": 10.471971893310547,
|
|
"epoch": 0.003075374636546634,
|
|
"grad_norm": 2.5,
|
|
"learning_rate": 2.7e-05,
|
|
"loss": 9.352,
|
|
"mean_token_accuracy": 0.056226722151041034,
|
|
"num_tokens": 104083.0,
|
|
"step": 55
|
|
},
|
|
{
|
|
"entropy": 10.36090316772461,
|
|
"epoch": 0.0033549541489599644,
|
|
"grad_norm": 2.453125,
|
|
"learning_rate": 2.95e-05,
|
|
"loss": 9.3357,
|
|
"mean_token_accuracy": 0.0560699962079525,
|
|
"num_tokens": 113536.0,
|
|
"step": 60
|
|
},
|
|
{
|
|
"entropy": 10.390652847290038,
|
|
"epoch": 0.0036345336613732944,
|
|
"grad_norm": 2.859375,
|
|
"learning_rate": 3.2e-05,
|
|
"loss": 9.2217,
|
|
"mean_token_accuracy": 0.056196273863315584,
|
|
"num_tokens": 123048.0,
|
|
"step": 65
|
|
},
|
|
{
|
|
"entropy": 10.416264057159424,
|
|
"epoch": 0.003914113173786625,
|
|
"grad_norm": 2.90625,
|
|
"learning_rate": 3.4500000000000005e-05,
|
|
"loss": 9.0594,
|
|
"mean_token_accuracy": 0.06116741932928562,
|
|
"num_tokens": 131550.0,
|
|
"step": 70
|
|
},
|
|
{
|
|
"entropy": 10.374309062957764,
|
|
"epoch": 0.004193692686199955,
|
|
"grad_norm": 3.0,
|
|
"learning_rate": 3.7e-05,
|
|
"loss": 9.033,
|
|
"mean_token_accuracy": 0.061420274525880815,
|
|
"num_tokens": 141235.0,
|
|
"step": 75
|
|
},
|
|
{
|
|
"entropy": 10.379674530029297,
|
|
"epoch": 0.004473272198613286,
|
|
"grad_norm": 2.265625,
|
|
"learning_rate": 3.95e-05,
|
|
"loss": 8.9435,
|
|
"mean_token_accuracy": 0.0640605840831995,
|
|
"num_tokens": 150681.0,
|
|
"step": 80
|
|
},
|
|
{
|
|
"entropy": 10.345459938049316,
|
|
"epoch": 0.004752851711026616,
|
|
"grad_norm": 2.65625,
|
|
"learning_rate": 4.2000000000000004e-05,
|
|
"loss": 8.8576,
|
|
"mean_token_accuracy": 0.06255088299512863,
|
|
"num_tokens": 160506.0,
|
|
"step": 85
|
|
},
|
|
{
|
|
"entropy": 10.31331491470337,
|
|
"epoch": 0.005032431223439947,
|
|
"grad_norm": 2.46875,
|
|
"learning_rate": 4.45e-05,
|
|
"loss": 8.7311,
|
|
"mean_token_accuracy": 0.06503418013453484,
|
|
"num_tokens": 169975.0,
|
|
"step": 90
|
|
},
|
|
{
|
|
"entropy": 10.259163188934327,
|
|
"epoch": 0.005312010735853277,
|
|
"grad_norm": 2.21875,
|
|
"learning_rate": 4.7000000000000004e-05,
|
|
"loss": 8.6912,
|
|
"mean_token_accuracy": 0.06565042100846767,
|
|
"num_tokens": 178593.0,
|
|
"step": 95
|
|
},
|
|
{
|
|
"entropy": 10.210173034667969,
|
|
"epoch": 0.005591590248266607,
|
|
"grad_norm": 2.125,
|
|
"learning_rate": 4.9500000000000004e-05,
|
|
"loss": 8.5786,
|
|
"mean_token_accuracy": 0.0654834620654583,
|
|
"num_tokens": 188472.0,
|
|
"step": 100
|
|
},
|
|
{
|
|
"entropy": 10.204802703857421,
|
|
"epoch": 0.005871169760679937,
|
|
"grad_norm": 2.3125,
|
|
"learning_rate": 5.2e-05,
|
|
"loss": 8.4289,
|
|
"mean_token_accuracy": 0.06512044295668602,
|
|
"num_tokens": 197362.0,
|
|
"step": 105
|
|
},
|
|
{
|
|
"entropy": 10.070987224578857,
|
|
"epoch": 0.006150749273093268,
|
|
"grad_norm": 2.3125,
|
|
"learning_rate": 5.45e-05,
|
|
"loss": 8.3881,
|
|
"mean_token_accuracy": 0.07190057411789894,
|
|
"num_tokens": 207551.0,
|
|
"step": 110
|
|
},
|
|
{
|
|
"entropy": 10.01644868850708,
|
|
"epoch": 0.006430328785506598,
|
|
"grad_norm": 2.15625,
|
|
"learning_rate": 5.7e-05,
|
|
"loss": 8.1562,
|
|
"mean_token_accuracy": 0.07526783086359501,
|
|
"num_tokens": 218498.0,
|
|
"step": 115
|
|
},
|
|
{
|
|
"entropy": 9.895391368865967,
|
|
"epoch": 0.006709908297919929,
|
|
"grad_norm": 2.03125,
|
|
"learning_rate": 5.9499999999999996e-05,
|
|
"loss": 8.0826,
|
|
"mean_token_accuracy": 0.07997448220849038,
|
|
"num_tokens": 228254.0,
|
|
"step": 120
|
|
},
|
|
{
|
|
"entropy": 9.790292835235595,
|
|
"epoch": 0.006989487810333259,
|
|
"grad_norm": 1.921875,
|
|
"learning_rate": 6.2e-05,
|
|
"loss": 7.9707,
|
|
"mean_token_accuracy": 0.07530777603387832,
|
|
"num_tokens": 237478.0,
|
|
"step": 125
|
|
},
|
|
{
|
|
"entropy": 9.68728494644165,
|
|
"epoch": 0.007269067322746589,
|
|
"grad_norm": 1.875,
|
|
"learning_rate": 6.450000000000001e-05,
|
|
"loss": 7.9089,
|
|
"mean_token_accuracy": 0.07417170405387878,
|
|
"num_tokens": 246342.0,
|
|
"step": 130
|
|
},
|
|
{
|
|
"entropy": 9.525614833831787,
|
|
"epoch": 0.0075486468351599195,
|
|
"grad_norm": 1.484375,
|
|
"learning_rate": 6.7e-05,
|
|
"loss": 7.8067,
|
|
"mean_token_accuracy": 0.07382837124168873,
|
|
"num_tokens": 257289.0,
|
|
"step": 135
|
|
},
|
|
{
|
|
"entropy": 9.321121501922608,
|
|
"epoch": 0.00782822634757325,
|
|
"grad_norm": 1.6875,
|
|
"learning_rate": 6.950000000000001e-05,
|
|
"loss": 7.6453,
|
|
"mean_token_accuracy": 0.07587397061288356,
|
|
"num_tokens": 266862.0,
|
|
"step": 140
|
|
},
|
|
{
|
|
"entropy": 9.158805179595948,
|
|
"epoch": 0.00810780585998658,
|
|
"grad_norm": 1.78125,
|
|
"learning_rate": 7.2e-05,
|
|
"loss": 7.5626,
|
|
"mean_token_accuracy": 0.07878116257488728,
|
|
"num_tokens": 276787.0,
|
|
"step": 145
|
|
},
|
|
{
|
|
"entropy": 8.895209884643554,
|
|
"epoch": 0.00838738537239991,
|
|
"grad_norm": 1.453125,
|
|
"learning_rate": 7.45e-05,
|
|
"loss": 7.4282,
|
|
"mean_token_accuracy": 0.08020649701356888,
|
|
"num_tokens": 286991.0,
|
|
"step": 150
|
|
},
|
|
{
|
|
"entropy": 8.710405635833741,
|
|
"epoch": 0.008666964884813241,
|
|
"grad_norm": 1.5703125,
|
|
"learning_rate": 7.7e-05,
|
|
"loss": 7.3875,
|
|
"mean_token_accuracy": 0.08550404459238052,
|
|
"num_tokens": 296517.0,
|
|
"step": 155
|
|
},
|
|
{
|
|
"entropy": 8.512088584899903,
|
|
"epoch": 0.008946544397226572,
|
|
"grad_norm": 1.2265625,
|
|
"learning_rate": 7.950000000000001e-05,
|
|
"loss": 7.3397,
|
|
"mean_token_accuracy": 0.0807737536728382,
|
|
"num_tokens": 306476.0,
|
|
"step": 160
|
|
},
|
|
{
|
|
"entropy": 8.354952907562256,
|
|
"epoch": 0.0092261239096399,
|
|
"grad_norm": 1.4296875,
|
|
"learning_rate": 8.2e-05,
|
|
"loss": 7.2449,
|
|
"mean_token_accuracy": 0.0858085073530674,
|
|
"num_tokens": 315372.0,
|
|
"step": 165
|
|
},
|
|
{
|
|
"entropy": 8.175031757354736,
|
|
"epoch": 0.009505703422053232,
|
|
"grad_norm": 1.21875,
|
|
"learning_rate": 8.450000000000001e-05,
|
|
"loss": 7.2093,
|
|
"mean_token_accuracy": 0.08027389757335186,
|
|
"num_tokens": 325361.0,
|
|
"step": 170
|
|
},
|
|
{
|
|
"entropy": 8.012962198257446,
|
|
"epoch": 0.009785282934466562,
|
|
"grad_norm": 1.453125,
|
|
"learning_rate": 8.7e-05,
|
|
"loss": 7.127,
|
|
"mean_token_accuracy": 0.09110699072480202,
|
|
"num_tokens": 335099.0,
|
|
"step": 175
|
|
},
|
|
{
|
|
"entropy": 7.874845790863037,
|
|
"epoch": 0.010064862446879893,
|
|
"grad_norm": 1.40625,
|
|
"learning_rate": 8.95e-05,
|
|
"loss": 7.037,
|
|
"mean_token_accuracy": 0.08859180435538291,
|
|
"num_tokens": 343850.0,
|
|
"step": 180
|
|
},
|
|
{
|
|
"entropy": 7.876893758773804,
|
|
"epoch": 0.010344441959293222,
|
|
"grad_norm": 1.4375,
|
|
"learning_rate": 9.2e-05,
|
|
"loss": 7.172,
|
|
"mean_token_accuracy": 0.08158803507685661,
|
|
"num_tokens": 353580.0,
|
|
"step": 185
|
|
},
|
|
{
|
|
"entropy": 7.905035734176636,
|
|
"epoch": 0.010624021471706553,
|
|
"grad_norm": 1.6875,
|
|
"learning_rate": 9.45e-05,
|
|
"loss": 7.1346,
|
|
"mean_token_accuracy": 0.07959189042448997,
|
|
"num_tokens": 362915.0,
|
|
"step": 190
|
|
},
|
|
{
|
|
"entropy": 7.806015872955323,
|
|
"epoch": 0.010903600984119884,
|
|
"grad_norm": 1.71875,
|
|
"learning_rate": 9.7e-05,
|
|
"loss": 7.0652,
|
|
"mean_token_accuracy": 0.0788392372429371,
|
|
"num_tokens": 373384.0,
|
|
"step": 195
|
|
},
|
|
{
|
|
"entropy": 7.610120105743408,
|
|
"epoch": 0.011183180496533215,
|
|
"grad_norm": 1.578125,
|
|
"learning_rate": 9.95e-05,
|
|
"loss": 7.0387,
|
|
"mean_token_accuracy": 0.08487512543797493,
|
|
"num_tokens": 382764.0,
|
|
"step": 200
|
|
},
|
|
{
|
|
"entropy": 7.788971900939941,
|
|
"epoch": 0.011462760008946544,
|
|
"grad_norm": 1.6015625,
|
|
"learning_rate": 0.000102,
|
|
"loss": 7.1369,
|
|
"mean_token_accuracy": 0.08739647641777992,
|
|
"num_tokens": 391243.0,
|
|
"step": 205
|
|
},
|
|
{
|
|
"entropy": 7.615197801589966,
|
|
"epoch": 0.011742339521359875,
|
|
"grad_norm": 1.3125,
|
|
"learning_rate": 0.00010449999999999999,
|
|
"loss": 7.034,
|
|
"mean_token_accuracy": 0.09340409561991692,
|
|
"num_tokens": 400983.0,
|
|
"step": 210
|
|
},
|
|
{
|
|
"entropy": 7.660063695907593,
|
|
"epoch": 0.012021919033773205,
|
|
"grad_norm": 1.4453125,
|
|
"learning_rate": 0.000107,
|
|
"loss": 6.8764,
|
|
"mean_token_accuracy": 0.09217005968093872,
|
|
"num_tokens": 410196.0,
|
|
"step": 215
|
|
},
|
|
{
|
|
"entropy": 7.543980884552002,
|
|
"epoch": 0.012301498546186536,
|
|
"grad_norm": 1.546875,
|
|
"learning_rate": 0.0001095,
|
|
"loss": 6.9262,
|
|
"mean_token_accuracy": 0.09098378121852875,
|
|
"num_tokens": 419344.0,
|
|
"step": 220
|
|
},
|
|
{
|
|
"entropy": 7.577476549148559,
|
|
"epoch": 0.012581078058599865,
|
|
"grad_norm": 1.4921875,
|
|
"learning_rate": 0.000112,
|
|
"loss": 7.0198,
|
|
"mean_token_accuracy": 0.08162087500095368,
|
|
"num_tokens": 429557.0,
|
|
"step": 225
|
|
},
|
|
{
|
|
"entropy": 7.595211410522461,
|
|
"epoch": 0.012860657571013196,
|
|
"grad_norm": 1.109375,
|
|
"learning_rate": 0.0001145,
|
|
"loss": 7.038,
|
|
"mean_token_accuracy": 0.07889171466231346,
|
|
"num_tokens": 440474.0,
|
|
"step": 230
|
|
},
|
|
{
|
|
"entropy": 7.516703319549561,
|
|
"epoch": 0.013140237083426527,
|
|
"grad_norm": 1.2890625,
|
|
"learning_rate": 0.00011700000000000001,
|
|
"loss": 6.8554,
|
|
"mean_token_accuracy": 0.09746571034193038,
|
|
"num_tokens": 450552.0,
|
|
"step": 235
|
|
},
|
|
{
|
|
"entropy": 7.418551683425903,
|
|
"epoch": 0.013419816595839858,
|
|
"grad_norm": 1.34375,
|
|
"learning_rate": 0.00011949999999999999,
|
|
"loss": 6.8744,
|
|
"mean_token_accuracy": 0.08559218943119049,
|
|
"num_tokens": 459501.0,
|
|
"step": 240
|
|
},
|
|
{
|
|
"entropy": 7.459112453460693,
|
|
"epoch": 0.013699396108253187,
|
|
"grad_norm": 1.328125,
|
|
"learning_rate": 0.000122,
|
|
"loss": 6.9676,
|
|
"mean_token_accuracy": 0.08793454244732857,
|
|
"num_tokens": 468807.0,
|
|
"step": 245
|
|
},
|
|
{
|
|
"entropy": 7.494116115570068,
|
|
"epoch": 0.013978975620666518,
|
|
"grad_norm": 1.5625,
|
|
"learning_rate": 0.0001245,
|
|
"loss": 6.9437,
|
|
"mean_token_accuracy": 0.08950937315821647,
|
|
"num_tokens": 477386.0,
|
|
"step": 250
|
|
},
|
|
{
|
|
"entropy": 7.427541923522949,
|
|
"epoch": 0.014258555133079848,
|
|
"grad_norm": 1.21875,
|
|
"learning_rate": 0.000127,
|
|
"loss": 6.8711,
|
|
"mean_token_accuracy": 0.08891022279858589,
|
|
"num_tokens": 485894.0,
|
|
"step": 255
|
|
},
|
|
{
|
|
"entropy": 7.419974088668823,
|
|
"epoch": 0.014538134645493177,
|
|
"grad_norm": 1.359375,
|
|
"learning_rate": 0.0001295,
|
|
"loss": 7.0094,
|
|
"mean_token_accuracy": 0.08291822224855423,
|
|
"num_tokens": 495083.0,
|
|
"step": 260
|
|
},
|
|
{
|
|
"entropy": 7.431387090682984,
|
|
"epoch": 0.014817714157906508,
|
|
"grad_norm": 1.5546875,
|
|
"learning_rate": 0.000132,
|
|
"loss": 6.9432,
|
|
"mean_token_accuracy": 0.08412053138017654,
|
|
"num_tokens": 504052.0,
|
|
"step": 265
|
|
},
|
|
{
|
|
"entropy": 7.427925777435303,
|
|
"epoch": 0.015097293670319839,
|
|
"grad_norm": 1.328125,
|
|
"learning_rate": 0.00013450000000000002,
|
|
"loss": 6.9054,
|
|
"mean_token_accuracy": 0.08690859079360962,
|
|
"num_tokens": 514054.0,
|
|
"step": 270
|
|
},
|
|
{
|
|
"entropy": 7.391806268692017,
|
|
"epoch": 0.01537687318273317,
|
|
"grad_norm": 1.2265625,
|
|
"learning_rate": 0.00013700000000000002,
|
|
"loss": 6.9026,
|
|
"mean_token_accuracy": 0.08920464739203453,
|
|
"num_tokens": 523135.0,
|
|
"step": 275
|
|
},
|
|
{
|
|
"entropy": 7.383348894119263,
|
|
"epoch": 0.0156564526951465,
|
|
"grad_norm": 1.328125,
|
|
"learning_rate": 0.0001395,
|
|
"loss": 6.8545,
|
|
"mean_token_accuracy": 0.08667187243700028,
|
|
"num_tokens": 532440.0,
|
|
"step": 280
|
|
},
|
|
{
|
|
"entropy": 7.335691356658936,
|
|
"epoch": 0.01593603220755983,
|
|
"grad_norm": 1.34375,
|
|
"learning_rate": 0.00014199999999999998,
|
|
"loss": 6.8642,
|
|
"mean_token_accuracy": 0.0919418253004551,
|
|
"num_tokens": 541707.0,
|
|
"step": 285
|
|
},
|
|
{
|
|
"entropy": 7.2934582233428955,
|
|
"epoch": 0.01621561171997316,
|
|
"grad_norm": 1.234375,
|
|
"learning_rate": 0.0001445,
|
|
"loss": 6.8127,
|
|
"mean_token_accuracy": 0.0894357517361641,
|
|
"num_tokens": 551263.0,
|
|
"step": 290
|
|
},
|
|
{
|
|
"entropy": 7.366137933731079,
|
|
"epoch": 0.01649519123238649,
|
|
"grad_norm": 1.2265625,
|
|
"learning_rate": 0.000147,
|
|
"loss": 6.8343,
|
|
"mean_token_accuracy": 0.09377153888344765,
|
|
"num_tokens": 560217.0,
|
|
"step": 295
|
|
},
|
|
{
|
|
"entropy": 7.273529672622681,
|
|
"epoch": 0.01677477074479982,
|
|
"grad_norm": 1.2890625,
|
|
"learning_rate": 0.0001495,
|
|
"loss": 6.7378,
|
|
"mean_token_accuracy": 0.08633734658360481,
|
|
"num_tokens": 569840.0,
|
|
"step": 300
|
|
},
|
|
{
|
|
"entropy": 7.181650400161743,
|
|
"epoch": 0.017054350257213153,
|
|
"grad_norm": 1.3671875,
|
|
"learning_rate": 0.000152,
|
|
"loss": 6.6951,
|
|
"mean_token_accuracy": 0.09873108789324761,
|
|
"num_tokens": 579117.0,
|
|
"step": 305
|
|
},
|
|
{
|
|
"entropy": 7.249513053894043,
|
|
"epoch": 0.017333929769626482,
|
|
"grad_norm": 1.421875,
|
|
"learning_rate": 0.00015450000000000001,
|
|
"loss": 6.8331,
|
|
"mean_token_accuracy": 0.09003806784749031,
|
|
"num_tokens": 588961.0,
|
|
"step": 310
|
|
},
|
|
{
|
|
"entropy": 7.089381504058838,
|
|
"epoch": 0.01761350928203981,
|
|
"grad_norm": 1.1953125,
|
|
"learning_rate": 0.000157,
|
|
"loss": 6.6727,
|
|
"mean_token_accuracy": 0.09398194402456284,
|
|
"num_tokens": 597906.0,
|
|
"step": 315
|
|
},
|
|
{
|
|
"entropy": 7.242098236083985,
|
|
"epoch": 0.017893088794453144,
|
|
"grad_norm": 1.328125,
|
|
"learning_rate": 0.0001595,
|
|
"loss": 6.822,
|
|
"mean_token_accuracy": 0.09356562569737434,
|
|
"num_tokens": 607015.0,
|
|
"step": 320
|
|
},
|
|
{
|
|
"entropy": 7.1716405868530275,
|
|
"epoch": 0.018172668306866473,
|
|
"grad_norm": 1.2578125,
|
|
"learning_rate": 0.000162,
|
|
"loss": 6.6879,
|
|
"mean_token_accuracy": 0.09605172201991082,
|
|
"num_tokens": 615878.0,
|
|
"step": 325
|
|
},
|
|
{
|
|
"entropy": 7.05837836265564,
|
|
"epoch": 0.0184522478192798,
|
|
"grad_norm": 1.453125,
|
|
"learning_rate": 0.00016450000000000001,
|
|
"loss": 6.7008,
|
|
"mean_token_accuracy": 0.09321764335036278,
|
|
"num_tokens": 625542.0,
|
|
"step": 330
|
|
},
|
|
{
|
|
"entropy": 7.229507350921631,
|
|
"epoch": 0.018731827331693134,
|
|
"grad_norm": 1.375,
|
|
"learning_rate": 0.00016700000000000002,
|
|
"loss": 6.7562,
|
|
"mean_token_accuracy": 0.08855808004736901,
|
|
"num_tokens": 634482.0,
|
|
"step": 335
|
|
},
|
|
{
|
|
"entropy": 7.227516841888428,
|
|
"epoch": 0.019011406844106463,
|
|
"grad_norm": 1.265625,
|
|
"learning_rate": 0.00016950000000000003,
|
|
"loss": 6.7065,
|
|
"mean_token_accuracy": 0.08909320756793022,
|
|
"num_tokens": 643282.0,
|
|
"step": 340
|
|
},
|
|
{
|
|
"entropy": 7.060443782806397,
|
|
"epoch": 0.019290986356519796,
|
|
"grad_norm": 1.3828125,
|
|
"learning_rate": 0.00017199999999999998,
|
|
"loss": 6.6112,
|
|
"mean_token_accuracy": 0.09391226842999459,
|
|
"num_tokens": 652613.0,
|
|
"step": 345
|
|
},
|
|
{
|
|
"entropy": 6.943112230300903,
|
|
"epoch": 0.019570565868933125,
|
|
"grad_norm": 1.3828125,
|
|
"learning_rate": 0.00017449999999999999,
|
|
"loss": 6.5652,
|
|
"mean_token_accuracy": 0.10253061279654503,
|
|
"num_tokens": 662111.0,
|
|
"step": 350
|
|
},
|
|
{
|
|
"entropy": 7.038354539871216,
|
|
"epoch": 0.019850145381346454,
|
|
"grad_norm": 1.34375,
|
|
"learning_rate": 0.000177,
|
|
"loss": 6.7651,
|
|
"mean_token_accuracy": 0.09167603999376298,
|
|
"num_tokens": 670807.0,
|
|
"step": 355
|
|
},
|
|
{
|
|
"entropy": 7.153097629547119,
|
|
"epoch": 0.020129724893759787,
|
|
"grad_norm": 1.21875,
|
|
"learning_rate": 0.0001795,
|
|
"loss": 6.7888,
|
|
"mean_token_accuracy": 0.0894574448466301,
|
|
"num_tokens": 680447.0,
|
|
"step": 360
|
|
},
|
|
{
|
|
"entropy": 7.086006307601929,
|
|
"epoch": 0.020409304406173116,
|
|
"grad_norm": 1.265625,
|
|
"learning_rate": 0.000182,
|
|
"loss": 6.7339,
|
|
"mean_token_accuracy": 0.08920840471982956,
|
|
"num_tokens": 690818.0,
|
|
"step": 365
|
|
},
|
|
{
|
|
"entropy": 7.078864479064942,
|
|
"epoch": 0.020688883918586445,
|
|
"grad_norm": 1.28125,
|
|
"learning_rate": 0.0001845,
|
|
"loss": 6.7043,
|
|
"mean_token_accuracy": 0.09464118853211403,
|
|
"num_tokens": 700014.0,
|
|
"step": 370
|
|
},
|
|
{
|
|
"entropy": 7.129553127288818,
|
|
"epoch": 0.020968463430999777,
|
|
"grad_norm": 1.3203125,
|
|
"learning_rate": 0.000187,
|
|
"loss": 6.7369,
|
|
"mean_token_accuracy": 0.09654519632458687,
|
|
"num_tokens": 710390.0,
|
|
"step": 375
|
|
},
|
|
{
|
|
"entropy": 6.971775245666504,
|
|
"epoch": 0.021248042943413106,
|
|
"grad_norm": 1.2734375,
|
|
"learning_rate": 0.0001895,
|
|
"loss": 6.5676,
|
|
"mean_token_accuracy": 0.1039146564900875,
|
|
"num_tokens": 719907.0,
|
|
"step": 380
|
|
},
|
|
{
|
|
"entropy": 6.9949382781982425,
|
|
"epoch": 0.021527622455826435,
|
|
"grad_norm": 1.4296875,
|
|
"learning_rate": 0.000192,
|
|
"loss": 6.6302,
|
|
"mean_token_accuracy": 0.09534877464175225,
|
|
"num_tokens": 730641.0,
|
|
"step": 385
|
|
},
|
|
{
|
|
"entropy": 6.961507129669189,
|
|
"epoch": 0.021807201968239768,
|
|
"grad_norm": 1.2578125,
|
|
"learning_rate": 0.0001945,
|
|
"loss": 6.5664,
|
|
"mean_token_accuracy": 0.10222888961434365,
|
|
"num_tokens": 739638.0,
|
|
"step": 390
|
|
},
|
|
{
|
|
"entropy": 6.8918750286102295,
|
|
"epoch": 0.022086781480653097,
|
|
"grad_norm": 1.234375,
|
|
"learning_rate": 0.00019700000000000002,
|
|
"loss": 6.5968,
|
|
"mean_token_accuracy": 0.09721470028162002,
|
|
"num_tokens": 748638.0,
|
|
"step": 395
|
|
},
|
|
{
|
|
"entropy": 7.056669282913208,
|
|
"epoch": 0.02236636099306643,
|
|
"grad_norm": 1.2734375,
|
|
"learning_rate": 0.00019950000000000002,
|
|
"loss": 6.6367,
|
|
"mean_token_accuracy": 0.09450221732258797,
|
|
"num_tokens": 758705.0,
|
|
"step": 400
|
|
},
|
|
{
|
|
"entropy": 7.024502992630005,
|
|
"epoch": 0.02264594050547976,
|
|
"grad_norm": 1.21875,
|
|
"learning_rate": 0.000202,
|
|
"loss": 6.6539,
|
|
"mean_token_accuracy": 0.09404293596744537,
|
|
"num_tokens": 768527.0,
|
|
"step": 405
|
|
},
|
|
{
|
|
"entropy": 6.913692283630371,
|
|
"epoch": 0.022925520017893088,
|
|
"grad_norm": 1.296875,
|
|
"learning_rate": 0.00020449999999999998,
|
|
"loss": 6.5867,
|
|
"mean_token_accuracy": 0.09760746508836746,
|
|
"num_tokens": 778533.0,
|
|
"step": 410
|
|
},
|
|
{
|
|
"entropy": 6.921697902679443,
|
|
"epoch": 0.02320509953030642,
|
|
"grad_norm": 1.4375,
|
|
"learning_rate": 0.000207,
|
|
"loss": 6.5677,
|
|
"mean_token_accuracy": 0.09978625848889351,
|
|
"num_tokens": 788171.0,
|
|
"step": 415
|
|
},
|
|
{
|
|
"entropy": 6.961454296112061,
|
|
"epoch": 0.02348467904271975,
|
|
"grad_norm": 1.234375,
|
|
"learning_rate": 0.0002095,
|
|
"loss": 6.6712,
|
|
"mean_token_accuracy": 0.09329927712678909,
|
|
"num_tokens": 797928.0,
|
|
"step": 420
|
|
},
|
|
{
|
|
"entropy": 6.977083206176758,
|
|
"epoch": 0.02376425855513308,
|
|
"grad_norm": 1.2890625,
|
|
"learning_rate": 0.000212,
|
|
"loss": 6.5859,
|
|
"mean_token_accuracy": 0.0953036293387413,
|
|
"num_tokens": 807565.0,
|
|
"step": 425
|
|
},
|
|
{
|
|
"entropy": 7.022476720809936,
|
|
"epoch": 0.02404383806754641,
|
|
"grad_norm": 1.1015625,
|
|
"learning_rate": 0.0002145,
|
|
"loss": 6.6095,
|
|
"mean_token_accuracy": 0.09925682097673416,
|
|
"num_tokens": 817018.0,
|
|
"step": 430
|
|
},
|
|
{
|
|
"entropy": 6.906362009048462,
|
|
"epoch": 0.02432341757995974,
|
|
"grad_norm": 1.4140625,
|
|
"learning_rate": 0.00021700000000000002,
|
|
"loss": 6.5497,
|
|
"mean_token_accuracy": 0.1020013377070427,
|
|
"num_tokens": 826850.0,
|
|
"step": 435
|
|
},
|
|
{
|
|
"entropy": 6.890042304992676,
|
|
"epoch": 0.024602997092373072,
|
|
"grad_norm": 1.3203125,
|
|
"learning_rate": 0.0002195,
|
|
"loss": 6.5654,
|
|
"mean_token_accuracy": 0.10329198613762855,
|
|
"num_tokens": 835097.0,
|
|
"step": 440
|
|
},
|
|
{
|
|
"entropy": 6.845738363265991,
|
|
"epoch": 0.0248825766047864,
|
|
"grad_norm": 1.15625,
|
|
"learning_rate": 0.000222,
|
|
"loss": 6.5229,
|
|
"mean_token_accuracy": 0.10356307923793792,
|
|
"num_tokens": 844645.0,
|
|
"step": 445
|
|
},
|
|
{
|
|
"entropy": 6.889461040496826,
|
|
"epoch": 0.02516215611719973,
|
|
"grad_norm": 1.3828125,
|
|
"learning_rate": 0.0002245,
|
|
"loss": 6.5038,
|
|
"mean_token_accuracy": 0.10186199322342873,
|
|
"num_tokens": 853912.0,
|
|
"step": 450
|
|
},
|
|
{
|
|
"entropy": 6.83258090019226,
|
|
"epoch": 0.025441735629613063,
|
|
"grad_norm": 1.2578125,
|
|
"learning_rate": 0.00022700000000000002,
|
|
"loss": 6.5741,
|
|
"mean_token_accuracy": 0.10296021774411201,
|
|
"num_tokens": 863312.0,
|
|
"step": 455
|
|
},
|
|
{
|
|
"entropy": 6.889059400558471,
|
|
"epoch": 0.025721315142026392,
|
|
"grad_norm": 1.1796875,
|
|
"learning_rate": 0.00022950000000000002,
|
|
"loss": 6.5453,
|
|
"mean_token_accuracy": 0.09688765034079552,
|
|
"num_tokens": 873075.0,
|
|
"step": 460
|
|
},
|
|
{
|
|
"entropy": 6.824527597427368,
|
|
"epoch": 0.02600089465443972,
|
|
"grad_norm": 1.3203125,
|
|
"learning_rate": 0.00023200000000000003,
|
|
"loss": 6.4318,
|
|
"mean_token_accuracy": 0.10380328297615052,
|
|
"num_tokens": 880964.0,
|
|
"step": 465
|
|
},
|
|
{
|
|
"entropy": 6.830307388305664,
|
|
"epoch": 0.026280474166853054,
|
|
"grad_norm": 1.3203125,
|
|
"learning_rate": 0.00023449999999999998,
|
|
"loss": 6.5615,
|
|
"mean_token_accuracy": 0.09817187637090682,
|
|
"num_tokens": 891148.0,
|
|
"step": 470
|
|
},
|
|
{
|
|
"entropy": 6.774664688110351,
|
|
"epoch": 0.026560053679266383,
|
|
"grad_norm": 1.3203125,
|
|
"learning_rate": 0.000237,
|
|
"loss": 6.4576,
|
|
"mean_token_accuracy": 0.09947997033596039,
|
|
"num_tokens": 900635.0,
|
|
"step": 475
|
|
},
|
|
{
|
|
"entropy": 6.78422474861145,
|
|
"epoch": 0.026839633191679715,
|
|
"grad_norm": 1.34375,
|
|
"learning_rate": 0.0002395,
|
|
"loss": 6.4409,
|
|
"mean_token_accuracy": 0.10720937103033065,
|
|
"num_tokens": 910454.0,
|
|
"step": 480
|
|
},
|
|
{
|
|
"entropy": 6.737676954269409,
|
|
"epoch": 0.027119212704093044,
|
|
"grad_norm": 1.3984375,
|
|
"learning_rate": 0.000242,
|
|
"loss": 6.456,
|
|
"mean_token_accuracy": 0.10532223731279373,
|
|
"num_tokens": 919504.0,
|
|
"step": 485
|
|
},
|
|
{
|
|
"entropy": 6.7197558879852295,
|
|
"epoch": 0.027398792216506374,
|
|
"grad_norm": 1.1875,
|
|
"learning_rate": 0.0002445,
|
|
"loss": 6.4703,
|
|
"mean_token_accuracy": 0.10164613649249077,
|
|
"num_tokens": 929255.0,
|
|
"step": 490
|
|
},
|
|
{
|
|
"entropy": 6.818446922302246,
|
|
"epoch": 0.027678371728919706,
|
|
"grad_norm": 1.296875,
|
|
"learning_rate": 0.000247,
|
|
"loss": 6.5297,
|
|
"mean_token_accuracy": 0.09887348264455795,
|
|
"num_tokens": 939144.0,
|
|
"step": 495
|
|
},
|
|
{
|
|
"entropy": 6.689020252227783,
|
|
"epoch": 0.027957951241333035,
|
|
"grad_norm": 1.34375,
|
|
"learning_rate": 0.0002495,
|
|
"loss": 6.3954,
|
|
"mean_token_accuracy": 0.10200168937444687,
|
|
"num_tokens": 947345.0,
|
|
"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": 1387526012928000.0,
|
|
"train_batch_size": 16,
|
|
"trial_name": null,
|
|
"trial_params": null
|
|
}
|