1035 lines
27 KiB
JSON
1035 lines
27 KiB
JSON
{
|
|
"best_global_step": null,
|
|
"best_metric": null,
|
|
"best_model_checkpoint": null,
|
|
"epoch": 0.011172310545543924,
|
|
"eval_steps": 500,
|
|
"global_step": 500,
|
|
"is_hyper_param_search": false,
|
|
"is_local_process_zero": true,
|
|
"is_world_process_zero": true,
|
|
"log_history": [
|
|
{
|
|
"entropy": 10.692034435272216,
|
|
"epoch": 0.00011172310545543924,
|
|
"grad_norm": 16.25,
|
|
"learning_rate": 2e-06,
|
|
"loss": 10.6039,
|
|
"mean_token_accuracy": 0.0,
|
|
"num_tokens": 4250.0,
|
|
"step": 5
|
|
},
|
|
{
|
|
"entropy": 10.692024230957031,
|
|
"epoch": 0.00022344621091087847,
|
|
"grad_norm": 17.125,
|
|
"learning_rate": 4.5e-06,
|
|
"loss": 10.5084,
|
|
"mean_token_accuracy": 0.0,
|
|
"num_tokens": 8228.0,
|
|
"step": 10
|
|
},
|
|
{
|
|
"entropy": 10.691641712188721,
|
|
"epoch": 0.0003351693163663177,
|
|
"grad_norm": 13.125,
|
|
"learning_rate": 7e-06,
|
|
"loss": 10.2399,
|
|
"mean_token_accuracy": 0.01585084570106119,
|
|
"num_tokens": 12209.0,
|
|
"step": 15
|
|
},
|
|
{
|
|
"entropy": 10.6864333152771,
|
|
"epoch": 0.00044689242182175694,
|
|
"grad_norm": 8.875,
|
|
"learning_rate": 9.5e-06,
|
|
"loss": 9.9231,
|
|
"mean_token_accuracy": 0.04045895729213953,
|
|
"num_tokens": 16225.0,
|
|
"step": 20
|
|
},
|
|
{
|
|
"entropy": 10.658738136291504,
|
|
"epoch": 0.0005586155272771962,
|
|
"grad_norm": 5.71875,
|
|
"learning_rate": 1.2e-05,
|
|
"loss": 9.6193,
|
|
"mean_token_accuracy": 0.05384383201599121,
|
|
"num_tokens": 20212.0,
|
|
"step": 25
|
|
},
|
|
{
|
|
"entropy": 10.617362594604492,
|
|
"epoch": 0.0006703386327326354,
|
|
"grad_norm": 4.46875,
|
|
"learning_rate": 1.4500000000000002e-05,
|
|
"loss": 9.3987,
|
|
"mean_token_accuracy": 0.06703231446444988,
|
|
"num_tokens": 24445.0,
|
|
"step": 30
|
|
},
|
|
{
|
|
"entropy": 10.580356121063232,
|
|
"epoch": 0.0007820617381880747,
|
|
"grad_norm": 4.09375,
|
|
"learning_rate": 1.7000000000000003e-05,
|
|
"loss": 9.2682,
|
|
"mean_token_accuracy": 0.06737326122820378,
|
|
"num_tokens": 28365.0,
|
|
"step": 35
|
|
},
|
|
{
|
|
"entropy": 10.578466510772705,
|
|
"epoch": 0.0008937848436435139,
|
|
"grad_norm": 3.9375,
|
|
"learning_rate": 1.95e-05,
|
|
"loss": 9.3685,
|
|
"mean_token_accuracy": 0.05515027567744255,
|
|
"num_tokens": 33055.0,
|
|
"step": 40
|
|
},
|
|
{
|
|
"entropy": 10.575786018371582,
|
|
"epoch": 0.0010055079490989532,
|
|
"grad_norm": 3.5625,
|
|
"learning_rate": 2.2e-05,
|
|
"loss": 9.2537,
|
|
"mean_token_accuracy": 0.0627420760691166,
|
|
"num_tokens": 37599.0,
|
|
"step": 45
|
|
},
|
|
{
|
|
"entropy": 10.472075748443604,
|
|
"epoch": 0.0011172310545543925,
|
|
"grad_norm": 3.5,
|
|
"learning_rate": 2.4500000000000003e-05,
|
|
"loss": 9.1143,
|
|
"mean_token_accuracy": 0.07111909165978432,
|
|
"num_tokens": 41934.0,
|
|
"step": 50
|
|
},
|
|
{
|
|
"entropy": 10.315015411376953,
|
|
"epoch": 0.0012289541600098315,
|
|
"grad_norm": 3.21875,
|
|
"learning_rate": 2.7e-05,
|
|
"loss": 9.0802,
|
|
"mean_token_accuracy": 0.06700236462056637,
|
|
"num_tokens": 46178.0,
|
|
"step": 55
|
|
},
|
|
{
|
|
"entropy": 10.27945785522461,
|
|
"epoch": 0.0013406772654652708,
|
|
"grad_norm": 3.546875,
|
|
"learning_rate": 2.95e-05,
|
|
"loss": 8.9726,
|
|
"mean_token_accuracy": 0.0730497345328331,
|
|
"num_tokens": 50513.0,
|
|
"step": 60
|
|
},
|
|
{
|
|
"entropy": 10.237419605255127,
|
|
"epoch": 0.0014524003709207101,
|
|
"grad_norm": 2.859375,
|
|
"learning_rate": 3.2e-05,
|
|
"loss": 8.9616,
|
|
"mean_token_accuracy": 0.07224201261997223,
|
|
"num_tokens": 54924.0,
|
|
"step": 65
|
|
},
|
|
{
|
|
"entropy": 10.185122776031495,
|
|
"epoch": 0.0015641234763761494,
|
|
"grad_norm": 3.125,
|
|
"learning_rate": 3.4500000000000005e-05,
|
|
"loss": 8.8671,
|
|
"mean_token_accuracy": 0.06654836386442184,
|
|
"num_tokens": 59083.0,
|
|
"step": 70
|
|
},
|
|
{
|
|
"entropy": 10.172991943359374,
|
|
"epoch": 0.0016758465818315885,
|
|
"grad_norm": 3.9375,
|
|
"learning_rate": 3.7e-05,
|
|
"loss": 8.7827,
|
|
"mean_token_accuracy": 0.06427431814372539,
|
|
"num_tokens": 63324.0,
|
|
"step": 75
|
|
},
|
|
{
|
|
"entropy": 10.14193344116211,
|
|
"epoch": 0.0017875696872870278,
|
|
"grad_norm": 2.640625,
|
|
"learning_rate": 3.95e-05,
|
|
"loss": 8.8298,
|
|
"mean_token_accuracy": 0.07168133780360222,
|
|
"num_tokens": 67738.0,
|
|
"step": 80
|
|
},
|
|
{
|
|
"entropy": 10.173482418060303,
|
|
"epoch": 0.001899292792742467,
|
|
"grad_norm": 2.359375,
|
|
"learning_rate": 4.2000000000000004e-05,
|
|
"loss": 8.7488,
|
|
"mean_token_accuracy": 0.0716201275587082,
|
|
"num_tokens": 72305.0,
|
|
"step": 85
|
|
},
|
|
{
|
|
"entropy": 10.12889585494995,
|
|
"epoch": 0.0020110158981979064,
|
|
"grad_norm": 3.046875,
|
|
"learning_rate": 4.45e-05,
|
|
"loss": 8.5249,
|
|
"mean_token_accuracy": 0.0720902569591999,
|
|
"num_tokens": 76579.0,
|
|
"step": 90
|
|
},
|
|
{
|
|
"entropy": 10.03916826248169,
|
|
"epoch": 0.0021227390036533456,
|
|
"grad_norm": 2.203125,
|
|
"learning_rate": 4.7000000000000004e-05,
|
|
"loss": 8.5006,
|
|
"mean_token_accuracy": 0.07380848824977874,
|
|
"num_tokens": 80812.0,
|
|
"step": 95
|
|
},
|
|
{
|
|
"entropy": 9.937397384643555,
|
|
"epoch": 0.002234462109108785,
|
|
"grad_norm": 1.953125,
|
|
"learning_rate": 4.9500000000000004e-05,
|
|
"loss": 8.4135,
|
|
"mean_token_accuracy": 0.07211422435939312,
|
|
"num_tokens": 85090.0,
|
|
"step": 100
|
|
},
|
|
{
|
|
"entropy": 9.917206573486329,
|
|
"epoch": 0.0023461852145642242,
|
|
"grad_norm": 2.53125,
|
|
"learning_rate": 5.2e-05,
|
|
"loss": 8.344,
|
|
"mean_token_accuracy": 0.07151557318866253,
|
|
"num_tokens": 89578.0,
|
|
"step": 105
|
|
},
|
|
{
|
|
"entropy": 9.800644588470458,
|
|
"epoch": 0.002457908320019663,
|
|
"grad_norm": 1.9140625,
|
|
"learning_rate": 5.45e-05,
|
|
"loss": 8.2336,
|
|
"mean_token_accuracy": 0.07592308931052685,
|
|
"num_tokens": 94117.0,
|
|
"step": 110
|
|
},
|
|
{
|
|
"entropy": 9.670217800140382,
|
|
"epoch": 0.0025696314254751024,
|
|
"grad_norm": 2.1875,
|
|
"learning_rate": 5.7e-05,
|
|
"loss": 8.0337,
|
|
"mean_token_accuracy": 0.08074008114635944,
|
|
"num_tokens": 98082.0,
|
|
"step": 115
|
|
},
|
|
{
|
|
"entropy": 9.533781147003173,
|
|
"epoch": 0.0026813545309305417,
|
|
"grad_norm": 1.875,
|
|
"learning_rate": 5.9499999999999996e-05,
|
|
"loss": 8.0226,
|
|
"mean_token_accuracy": 0.07543974481523037,
|
|
"num_tokens": 102327.0,
|
|
"step": 120
|
|
},
|
|
{
|
|
"entropy": 9.384562969207764,
|
|
"epoch": 0.002793077636385981,
|
|
"grad_norm": 1.9296875,
|
|
"learning_rate": 6.2e-05,
|
|
"loss": 7.8842,
|
|
"mean_token_accuracy": 0.07998752668499946,
|
|
"num_tokens": 106567.0,
|
|
"step": 125
|
|
},
|
|
{
|
|
"entropy": 9.304523754119874,
|
|
"epoch": 0.0029048007418414202,
|
|
"grad_norm": 1.9140625,
|
|
"learning_rate": 6.450000000000001e-05,
|
|
"loss": 7.7414,
|
|
"mean_token_accuracy": 0.07796499505639076,
|
|
"num_tokens": 110654.0,
|
|
"step": 130
|
|
},
|
|
{
|
|
"entropy": 9.14549970626831,
|
|
"epoch": 0.0030165238472968595,
|
|
"grad_norm": 2.140625,
|
|
"learning_rate": 6.7e-05,
|
|
"loss": 7.7396,
|
|
"mean_token_accuracy": 0.0759395856410265,
|
|
"num_tokens": 114679.0,
|
|
"step": 135
|
|
},
|
|
{
|
|
"entropy": 8.918598842620849,
|
|
"epoch": 0.003128246952752299,
|
|
"grad_norm": 1.7578125,
|
|
"learning_rate": 6.950000000000001e-05,
|
|
"loss": 7.7208,
|
|
"mean_token_accuracy": 0.0734277281910181,
|
|
"num_tokens": 118817.0,
|
|
"step": 140
|
|
},
|
|
{
|
|
"entropy": 8.880987071990967,
|
|
"epoch": 0.003239970058207738,
|
|
"grad_norm": 1.640625,
|
|
"learning_rate": 7.2e-05,
|
|
"loss": 7.63,
|
|
"mean_token_accuracy": 0.08075562492012978,
|
|
"num_tokens": 123188.0,
|
|
"step": 145
|
|
},
|
|
{
|
|
"entropy": 8.665369319915772,
|
|
"epoch": 0.003351693163663177,
|
|
"grad_norm": 1.96875,
|
|
"learning_rate": 7.45e-05,
|
|
"loss": 7.571,
|
|
"mean_token_accuracy": 0.0803100574761629,
|
|
"num_tokens": 127767.0,
|
|
"step": 150
|
|
},
|
|
{
|
|
"entropy": 8.561956405639648,
|
|
"epoch": 0.0034634162691186163,
|
|
"grad_norm": 1.8125,
|
|
"learning_rate": 7.7e-05,
|
|
"loss": 7.3994,
|
|
"mean_token_accuracy": 0.0842840738594532,
|
|
"num_tokens": 131837.0,
|
|
"step": 155
|
|
},
|
|
{
|
|
"entropy": 8.48619909286499,
|
|
"epoch": 0.0035751393745740555,
|
|
"grad_norm": 1.515625,
|
|
"learning_rate": 7.950000000000001e-05,
|
|
"loss": 7.4235,
|
|
"mean_token_accuracy": 0.08158716633915901,
|
|
"num_tokens": 136247.0,
|
|
"step": 160
|
|
},
|
|
{
|
|
"entropy": 8.290351867675781,
|
|
"epoch": 0.003686862480029495,
|
|
"grad_norm": 2.203125,
|
|
"learning_rate": 8.2e-05,
|
|
"loss": 7.264,
|
|
"mean_token_accuracy": 0.0918136142194271,
|
|
"num_tokens": 140170.0,
|
|
"step": 165
|
|
},
|
|
{
|
|
"entropy": 8.1635573387146,
|
|
"epoch": 0.003798585585484934,
|
|
"grad_norm": 1.546875,
|
|
"learning_rate": 8.450000000000001e-05,
|
|
"loss": 7.3583,
|
|
"mean_token_accuracy": 0.09022970125079155,
|
|
"num_tokens": 144139.0,
|
|
"step": 170
|
|
},
|
|
{
|
|
"entropy": 8.112592220306396,
|
|
"epoch": 0.003910308690940373,
|
|
"grad_norm": 1.5859375,
|
|
"learning_rate": 8.7e-05,
|
|
"loss": 7.3255,
|
|
"mean_token_accuracy": 0.08751063793897629,
|
|
"num_tokens": 148792.0,
|
|
"step": 175
|
|
},
|
|
{
|
|
"entropy": 8.065679502487182,
|
|
"epoch": 0.004022031796395813,
|
|
"grad_norm": 2.1875,
|
|
"learning_rate": 8.95e-05,
|
|
"loss": 7.2901,
|
|
"mean_token_accuracy": 0.08218263126909733,
|
|
"num_tokens": 152844.0,
|
|
"step": 180
|
|
},
|
|
{
|
|
"entropy": 8.083935976028442,
|
|
"epoch": 0.0041337549018512516,
|
|
"grad_norm": 1.7265625,
|
|
"learning_rate": 9.2e-05,
|
|
"loss": 7.3158,
|
|
"mean_token_accuracy": 0.08304233588278294,
|
|
"num_tokens": 157366.0,
|
|
"step": 185
|
|
},
|
|
{
|
|
"entropy": 7.946479368209839,
|
|
"epoch": 0.004245478007306691,
|
|
"grad_norm": 2.015625,
|
|
"learning_rate": 9.45e-05,
|
|
"loss": 7.1574,
|
|
"mean_token_accuracy": 0.093728469312191,
|
|
"num_tokens": 161348.0,
|
|
"step": 190
|
|
},
|
|
{
|
|
"entropy": 7.886520957946777,
|
|
"epoch": 0.00435720111276213,
|
|
"grad_norm": 1.8046875,
|
|
"learning_rate": 9.7e-05,
|
|
"loss": 7.1876,
|
|
"mean_token_accuracy": 0.10094638615846634,
|
|
"num_tokens": 165647.0,
|
|
"step": 195
|
|
},
|
|
{
|
|
"entropy": 7.8108118057250975,
|
|
"epoch": 0.00446892421821757,
|
|
"grad_norm": 1.8515625,
|
|
"learning_rate": 9.95e-05,
|
|
"loss": 7.1137,
|
|
"mean_token_accuracy": 0.09259389042854309,
|
|
"num_tokens": 169521.0,
|
|
"step": 200
|
|
},
|
|
{
|
|
"entropy": 7.7959239959716795,
|
|
"epoch": 0.004580647323673009,
|
|
"grad_norm": 1.8671875,
|
|
"learning_rate": 0.000102,
|
|
"loss": 7.1787,
|
|
"mean_token_accuracy": 0.09544989094138145,
|
|
"num_tokens": 173466.0,
|
|
"step": 205
|
|
},
|
|
{
|
|
"entropy": 7.768160486221314,
|
|
"epoch": 0.0046923704291284484,
|
|
"grad_norm": 1.734375,
|
|
"learning_rate": 0.00010449999999999999,
|
|
"loss": 7.1178,
|
|
"mean_token_accuracy": 0.08695701211690902,
|
|
"num_tokens": 177663.0,
|
|
"step": 210
|
|
},
|
|
{
|
|
"entropy": 7.66025218963623,
|
|
"epoch": 0.004804093534583887,
|
|
"grad_norm": 1.8671875,
|
|
"learning_rate": 0.000107,
|
|
"loss": 7.1533,
|
|
"mean_token_accuracy": 0.08611816093325615,
|
|
"num_tokens": 181778.0,
|
|
"step": 215
|
|
},
|
|
{
|
|
"entropy": 7.759128427505493,
|
|
"epoch": 0.004915816640039326,
|
|
"grad_norm": 2.046875,
|
|
"learning_rate": 0.0001095,
|
|
"loss": 7.1497,
|
|
"mean_token_accuracy": 0.09380927979946137,
|
|
"num_tokens": 185525.0,
|
|
"step": 220
|
|
},
|
|
{
|
|
"entropy": 7.712249279022217,
|
|
"epoch": 0.005027539745494766,
|
|
"grad_norm": 1.8359375,
|
|
"learning_rate": 0.000112,
|
|
"loss": 7.0063,
|
|
"mean_token_accuracy": 0.10432608500123024,
|
|
"num_tokens": 189418.0,
|
|
"step": 225
|
|
},
|
|
{
|
|
"entropy": 7.597976541519165,
|
|
"epoch": 0.005139262850950205,
|
|
"grad_norm": 2.0625,
|
|
"learning_rate": 0.0001145,
|
|
"loss": 7.0884,
|
|
"mean_token_accuracy": 0.10287595391273499,
|
|
"num_tokens": 193494.0,
|
|
"step": 230
|
|
},
|
|
{
|
|
"entropy": 7.700113201141358,
|
|
"epoch": 0.0052509859564056445,
|
|
"grad_norm": 1.53125,
|
|
"learning_rate": 0.00011700000000000001,
|
|
"loss": 7.1323,
|
|
"mean_token_accuracy": 0.09622926861047745,
|
|
"num_tokens": 198018.0,
|
|
"step": 235
|
|
},
|
|
{
|
|
"entropy": 7.695586109161377,
|
|
"epoch": 0.005362709061861083,
|
|
"grad_norm": 1.78125,
|
|
"learning_rate": 0.00011949999999999999,
|
|
"loss": 7.0591,
|
|
"mean_token_accuracy": 0.09503638371825218,
|
|
"num_tokens": 202296.0,
|
|
"step": 240
|
|
},
|
|
{
|
|
"entropy": 7.5984455108642575,
|
|
"epoch": 0.005474432167316523,
|
|
"grad_norm": 1.515625,
|
|
"learning_rate": 0.000122,
|
|
"loss": 7.0644,
|
|
"mean_token_accuracy": 0.09454682543873787,
|
|
"num_tokens": 206694.0,
|
|
"step": 245
|
|
},
|
|
{
|
|
"entropy": 7.6198304176330565,
|
|
"epoch": 0.005586155272771962,
|
|
"grad_norm": 2.03125,
|
|
"learning_rate": 0.0001245,
|
|
"loss": 7.0863,
|
|
"mean_token_accuracy": 0.10327245965600014,
|
|
"num_tokens": 210810.0,
|
|
"step": 250
|
|
},
|
|
{
|
|
"entropy": 7.671643877029419,
|
|
"epoch": 0.005697878378227402,
|
|
"grad_norm": 1.5078125,
|
|
"learning_rate": 0.000127,
|
|
"loss": 7.0545,
|
|
"mean_token_accuracy": 0.09972267299890518,
|
|
"num_tokens": 215044.0,
|
|
"step": 255
|
|
},
|
|
{
|
|
"entropy": 7.580405235290527,
|
|
"epoch": 0.0058096014836828405,
|
|
"grad_norm": 1.734375,
|
|
"learning_rate": 0.0001295,
|
|
"loss": 7.1836,
|
|
"mean_token_accuracy": 0.09796730950474739,
|
|
"num_tokens": 219235.0,
|
|
"step": 260
|
|
},
|
|
{
|
|
"entropy": 7.618637609481811,
|
|
"epoch": 0.005921324589138279,
|
|
"grad_norm": 1.5390625,
|
|
"learning_rate": 0.000132,
|
|
"loss": 6.9848,
|
|
"mean_token_accuracy": 0.10349723398685455,
|
|
"num_tokens": 223639.0,
|
|
"step": 265
|
|
},
|
|
{
|
|
"entropy": 7.4419373035430905,
|
|
"epoch": 0.006033047694593719,
|
|
"grad_norm": 1.75,
|
|
"learning_rate": 0.00013450000000000002,
|
|
"loss": 7.0238,
|
|
"mean_token_accuracy": 0.10185197666287422,
|
|
"num_tokens": 227873.0,
|
|
"step": 270
|
|
},
|
|
{
|
|
"entropy": 7.483202838897705,
|
|
"epoch": 0.006144770800049158,
|
|
"grad_norm": 1.5859375,
|
|
"learning_rate": 0.00013700000000000002,
|
|
"loss": 6.9487,
|
|
"mean_token_accuracy": 0.11078577488660812,
|
|
"num_tokens": 232147.0,
|
|
"step": 275
|
|
},
|
|
{
|
|
"entropy": 7.483991575241089,
|
|
"epoch": 0.006256493905504598,
|
|
"grad_norm": 1.8203125,
|
|
"learning_rate": 0.0001395,
|
|
"loss": 7.0367,
|
|
"mean_token_accuracy": 0.1014625109732151,
|
|
"num_tokens": 236456.0,
|
|
"step": 280
|
|
},
|
|
{
|
|
"entropy": 7.457619524002075,
|
|
"epoch": 0.0063682170109600365,
|
|
"grad_norm": 1.46875,
|
|
"learning_rate": 0.00014199999999999998,
|
|
"loss": 7.0498,
|
|
"mean_token_accuracy": 0.10427944287657738,
|
|
"num_tokens": 241039.0,
|
|
"step": 285
|
|
},
|
|
{
|
|
"entropy": 7.737891387939453,
|
|
"epoch": 0.006479940116415476,
|
|
"grad_norm": 1.8046875,
|
|
"learning_rate": 0.0001445,
|
|
"loss": 7.097,
|
|
"mean_token_accuracy": 0.09794913604855537,
|
|
"num_tokens": 245132.0,
|
|
"step": 290
|
|
},
|
|
{
|
|
"entropy": 7.422194004058838,
|
|
"epoch": 0.006591663221870915,
|
|
"grad_norm": 1.71875,
|
|
"learning_rate": 0.000147,
|
|
"loss": 7.0166,
|
|
"mean_token_accuracy": 0.1079220362007618,
|
|
"num_tokens": 249152.0,
|
|
"step": 295
|
|
},
|
|
{
|
|
"entropy": 7.482765865325928,
|
|
"epoch": 0.006703386327326354,
|
|
"grad_norm": 1.5234375,
|
|
"learning_rate": 0.0001495,
|
|
"loss": 6.9395,
|
|
"mean_token_accuracy": 0.10692858770489692,
|
|
"num_tokens": 253439.0,
|
|
"step": 300
|
|
},
|
|
{
|
|
"entropy": 7.519821882247925,
|
|
"epoch": 0.006815109432781794,
|
|
"grad_norm": 1.78125,
|
|
"learning_rate": 0.000152,
|
|
"loss": 6.9877,
|
|
"mean_token_accuracy": 0.10241441354155541,
|
|
"num_tokens": 258066.0,
|
|
"step": 305
|
|
},
|
|
{
|
|
"entropy": 7.4823202133178714,
|
|
"epoch": 0.0069268325382372325,
|
|
"grad_norm": 1.890625,
|
|
"learning_rate": 0.00015450000000000001,
|
|
"loss": 6.9284,
|
|
"mean_token_accuracy": 0.1084162712097168,
|
|
"num_tokens": 261954.0,
|
|
"step": 310
|
|
},
|
|
{
|
|
"entropy": 7.320609712600708,
|
|
"epoch": 0.007038555643692672,
|
|
"grad_norm": 1.8046875,
|
|
"learning_rate": 0.000157,
|
|
"loss": 6.9852,
|
|
"mean_token_accuracy": 0.10961857140064239,
|
|
"num_tokens": 266650.0,
|
|
"step": 315
|
|
},
|
|
{
|
|
"entropy": 7.539988470077515,
|
|
"epoch": 0.007150278749148111,
|
|
"grad_norm": 1.6640625,
|
|
"learning_rate": 0.0001595,
|
|
"loss": 7.0467,
|
|
"mean_token_accuracy": 0.10235239341855049,
|
|
"num_tokens": 271069.0,
|
|
"step": 320
|
|
},
|
|
{
|
|
"entropy": 7.4312175750732425,
|
|
"epoch": 0.007262001854603551,
|
|
"grad_norm": 1.7734375,
|
|
"learning_rate": 0.000162,
|
|
"loss": 6.9195,
|
|
"mean_token_accuracy": 0.10588012635707855,
|
|
"num_tokens": 275084.0,
|
|
"step": 325
|
|
},
|
|
{
|
|
"entropy": 7.405316781997681,
|
|
"epoch": 0.00737372496005899,
|
|
"grad_norm": 1.7734375,
|
|
"learning_rate": 0.00016450000000000001,
|
|
"loss": 6.9708,
|
|
"mean_token_accuracy": 0.11100029796361924,
|
|
"num_tokens": 279721.0,
|
|
"step": 330
|
|
},
|
|
{
|
|
"entropy": 7.45182785987854,
|
|
"epoch": 0.007485448065514429,
|
|
"grad_norm": 1.546875,
|
|
"learning_rate": 0.00016700000000000002,
|
|
"loss": 6.9745,
|
|
"mean_token_accuracy": 0.11023727878928184,
|
|
"num_tokens": 284317.0,
|
|
"step": 335
|
|
},
|
|
{
|
|
"entropy": 7.376049518585205,
|
|
"epoch": 0.007597171170969868,
|
|
"grad_norm": 1.5078125,
|
|
"learning_rate": 0.00016950000000000003,
|
|
"loss": 6.9627,
|
|
"mean_token_accuracy": 0.11432994231581688,
|
|
"num_tokens": 288870.0,
|
|
"step": 340
|
|
},
|
|
{
|
|
"entropy": 7.399447011947632,
|
|
"epoch": 0.007708894276425307,
|
|
"grad_norm": 1.65625,
|
|
"learning_rate": 0.00017199999999999998,
|
|
"loss": 6.8883,
|
|
"mean_token_accuracy": 0.11889545023441314,
|
|
"num_tokens": 292996.0,
|
|
"step": 345
|
|
},
|
|
{
|
|
"entropy": 7.417701625823975,
|
|
"epoch": 0.007820617381880746,
|
|
"grad_norm": 1.6796875,
|
|
"learning_rate": 0.00017449999999999999,
|
|
"loss": 6.8748,
|
|
"mean_token_accuracy": 0.11174583658576012,
|
|
"num_tokens": 297238.0,
|
|
"step": 350
|
|
},
|
|
{
|
|
"entropy": 7.387735509872437,
|
|
"epoch": 0.007932340487336187,
|
|
"grad_norm": 1.890625,
|
|
"learning_rate": 0.000177,
|
|
"loss": 6.8957,
|
|
"mean_token_accuracy": 0.1190306767821312,
|
|
"num_tokens": 301453.0,
|
|
"step": 355
|
|
},
|
|
{
|
|
"entropy": 7.298361539840698,
|
|
"epoch": 0.008044063592791625,
|
|
"grad_norm": 1.8984375,
|
|
"learning_rate": 0.0001795,
|
|
"loss": 6.8362,
|
|
"mean_token_accuracy": 0.11696951016783715,
|
|
"num_tokens": 305949.0,
|
|
"step": 360
|
|
},
|
|
{
|
|
"entropy": 7.339731693267822,
|
|
"epoch": 0.008155786698247064,
|
|
"grad_norm": 1.734375,
|
|
"learning_rate": 0.000182,
|
|
"loss": 6.8502,
|
|
"mean_token_accuracy": 0.12272856459021568,
|
|
"num_tokens": 310097.0,
|
|
"step": 365
|
|
},
|
|
{
|
|
"entropy": 7.378586864471435,
|
|
"epoch": 0.008267509803702503,
|
|
"grad_norm": 1.8046875,
|
|
"learning_rate": 0.0001845,
|
|
"loss": 6.8746,
|
|
"mean_token_accuracy": 0.1073436789214611,
|
|
"num_tokens": 314567.0,
|
|
"step": 370
|
|
},
|
|
{
|
|
"entropy": 7.258294773101807,
|
|
"epoch": 0.008379232909157944,
|
|
"grad_norm": 1.828125,
|
|
"learning_rate": 0.000187,
|
|
"loss": 6.9539,
|
|
"mean_token_accuracy": 0.1148001030087471,
|
|
"num_tokens": 319166.0,
|
|
"step": 375
|
|
},
|
|
{
|
|
"entropy": 7.30250129699707,
|
|
"epoch": 0.008490956014613383,
|
|
"grad_norm": 1.6953125,
|
|
"learning_rate": 0.0001895,
|
|
"loss": 6.8336,
|
|
"mean_token_accuracy": 0.11580941900610923,
|
|
"num_tokens": 323682.0,
|
|
"step": 380
|
|
},
|
|
{
|
|
"entropy": 7.259005832672119,
|
|
"epoch": 0.008602679120068821,
|
|
"grad_norm": 1.6328125,
|
|
"learning_rate": 0.000192,
|
|
"loss": 6.7239,
|
|
"mean_token_accuracy": 0.12464385628700256,
|
|
"num_tokens": 327994.0,
|
|
"step": 385
|
|
},
|
|
{
|
|
"entropy": 7.200961637496948,
|
|
"epoch": 0.00871440222552426,
|
|
"grad_norm": 1.6328125,
|
|
"learning_rate": 0.0001945,
|
|
"loss": 6.8242,
|
|
"mean_token_accuracy": 0.11740145906805992,
|
|
"num_tokens": 332026.0,
|
|
"step": 390
|
|
},
|
|
{
|
|
"entropy": 7.481966304779053,
|
|
"epoch": 0.0088261253309797,
|
|
"grad_norm": 3.078125,
|
|
"learning_rate": 0.00019700000000000002,
|
|
"loss": 6.8603,
|
|
"mean_token_accuracy": 0.12176617532968521,
|
|
"num_tokens": 336552.0,
|
|
"step": 395
|
|
},
|
|
{
|
|
"entropy": 7.178363084793091,
|
|
"epoch": 0.00893784843643514,
|
|
"grad_norm": 1.6796875,
|
|
"learning_rate": 0.00019950000000000002,
|
|
"loss": 6.8005,
|
|
"mean_token_accuracy": 0.11857323870062828,
|
|
"num_tokens": 340643.0,
|
|
"step": 400
|
|
},
|
|
{
|
|
"entropy": 7.28194899559021,
|
|
"epoch": 0.009049571541890579,
|
|
"grad_norm": 1.65625,
|
|
"learning_rate": 0.000202,
|
|
"loss": 6.7174,
|
|
"mean_token_accuracy": 0.12609147280454636,
|
|
"num_tokens": 344886.0,
|
|
"step": 405
|
|
},
|
|
{
|
|
"entropy": 7.182888126373291,
|
|
"epoch": 0.009161294647346017,
|
|
"grad_norm": 1.8125,
|
|
"learning_rate": 0.00020449999999999998,
|
|
"loss": 6.7831,
|
|
"mean_token_accuracy": 0.12051703482866287,
|
|
"num_tokens": 349115.0,
|
|
"step": 410
|
|
},
|
|
{
|
|
"entropy": 7.436312055587768,
|
|
"epoch": 0.009273017752801456,
|
|
"grad_norm": 1.71875,
|
|
"learning_rate": 0.000207,
|
|
"loss": 6.7448,
|
|
"mean_token_accuracy": 0.12795047089457512,
|
|
"num_tokens": 353368.0,
|
|
"step": 415
|
|
},
|
|
{
|
|
"entropy": 7.033882093429566,
|
|
"epoch": 0.009384740858256897,
|
|
"grad_norm": 1.6484375,
|
|
"learning_rate": 0.0002095,
|
|
"loss": 6.7951,
|
|
"mean_token_accuracy": 0.11617444008588791,
|
|
"num_tokens": 357382.0,
|
|
"step": 420
|
|
},
|
|
{
|
|
"entropy": 7.290721368789673,
|
|
"epoch": 0.009496463963712336,
|
|
"grad_norm": 1.8359375,
|
|
"learning_rate": 0.000212,
|
|
"loss": 6.7152,
|
|
"mean_token_accuracy": 0.12455029636621476,
|
|
"num_tokens": 361542.0,
|
|
"step": 425
|
|
},
|
|
{
|
|
"entropy": 7.23753170967102,
|
|
"epoch": 0.009608187069167775,
|
|
"grad_norm": 1.5,
|
|
"learning_rate": 0.0002145,
|
|
"loss": 6.7497,
|
|
"mean_token_accuracy": 0.12337451875209808,
|
|
"num_tokens": 366006.0,
|
|
"step": 430
|
|
},
|
|
{
|
|
"entropy": 7.083537578582764,
|
|
"epoch": 0.009719910174623213,
|
|
"grad_norm": 1.6796875,
|
|
"learning_rate": 0.00021700000000000002,
|
|
"loss": 6.6902,
|
|
"mean_token_accuracy": 0.1326639696955681,
|
|
"num_tokens": 370021.0,
|
|
"step": 435
|
|
},
|
|
{
|
|
"entropy": 7.248496341705322,
|
|
"epoch": 0.009831633280078652,
|
|
"grad_norm": 1.6796875,
|
|
"learning_rate": 0.0002195,
|
|
"loss": 6.805,
|
|
"mean_token_accuracy": 0.12277387753129006,
|
|
"num_tokens": 374434.0,
|
|
"step": 440
|
|
},
|
|
{
|
|
"entropy": 7.251986265182495,
|
|
"epoch": 0.009943356385534093,
|
|
"grad_norm": 1.5859375,
|
|
"learning_rate": 0.000222,
|
|
"loss": 6.8176,
|
|
"mean_token_accuracy": 0.11638971790671349,
|
|
"num_tokens": 378634.0,
|
|
"step": 445
|
|
},
|
|
{
|
|
"entropy": 7.187275457382202,
|
|
"epoch": 0.010055079490989532,
|
|
"grad_norm": 1.6171875,
|
|
"learning_rate": 0.0002245,
|
|
"loss": 6.6952,
|
|
"mean_token_accuracy": 0.12703127264976502,
|
|
"num_tokens": 382904.0,
|
|
"step": 450
|
|
},
|
|
{
|
|
"entropy": 7.244022417068481,
|
|
"epoch": 0.01016680259644497,
|
|
"grad_norm": 1.65625,
|
|
"learning_rate": 0.00022700000000000002,
|
|
"loss": 6.6803,
|
|
"mean_token_accuracy": 0.12515532448887826,
|
|
"num_tokens": 386970.0,
|
|
"step": 455
|
|
},
|
|
{
|
|
"entropy": 7.082424831390381,
|
|
"epoch": 0.01027852570190041,
|
|
"grad_norm": 1.59375,
|
|
"learning_rate": 0.00022950000000000002,
|
|
"loss": 6.6572,
|
|
"mean_token_accuracy": 0.13228979632258414,
|
|
"num_tokens": 391043.0,
|
|
"step": 460
|
|
},
|
|
{
|
|
"entropy": 7.1885302543640135,
|
|
"epoch": 0.01039024880735585,
|
|
"grad_norm": 1.390625,
|
|
"learning_rate": 0.00023200000000000003,
|
|
"loss": 6.6428,
|
|
"mean_token_accuracy": 0.13703735694289207,
|
|
"num_tokens": 395413.0,
|
|
"step": 465
|
|
},
|
|
{
|
|
"entropy": 7.166730117797852,
|
|
"epoch": 0.010501971912811289,
|
|
"grad_norm": 1.9296875,
|
|
"learning_rate": 0.00023449999999999998,
|
|
"loss": 6.7767,
|
|
"mean_token_accuracy": 0.12424237206578255,
|
|
"num_tokens": 399821.0,
|
|
"step": 470
|
|
},
|
|
{
|
|
"entropy": 7.0755476474761965,
|
|
"epoch": 0.010613695018266728,
|
|
"grad_norm": 1.59375,
|
|
"learning_rate": 0.000237,
|
|
"loss": 6.5815,
|
|
"mean_token_accuracy": 0.13624327406287193,
|
|
"num_tokens": 404043.0,
|
|
"step": 475
|
|
},
|
|
{
|
|
"entropy": 7.202919864654541,
|
|
"epoch": 0.010725418123722167,
|
|
"grad_norm": 1.9921875,
|
|
"learning_rate": 0.0002395,
|
|
"loss": 6.7277,
|
|
"mean_token_accuracy": 0.1173846572637558,
|
|
"num_tokens": 408339.0,
|
|
"step": 480
|
|
},
|
|
{
|
|
"entropy": 7.004957008361816,
|
|
"epoch": 0.010837141229177606,
|
|
"grad_norm": 1.8671875,
|
|
"learning_rate": 0.000242,
|
|
"loss": 6.6233,
|
|
"mean_token_accuracy": 0.1330234356224537,
|
|
"num_tokens": 412384.0,
|
|
"step": 485
|
|
},
|
|
{
|
|
"entropy": 7.170296764373779,
|
|
"epoch": 0.010948864334633046,
|
|
"grad_norm": 1.7421875,
|
|
"learning_rate": 0.0002445,
|
|
"loss": 6.7419,
|
|
"mean_token_accuracy": 0.122147136926651,
|
|
"num_tokens": 416891.0,
|
|
"step": 490
|
|
},
|
|
{
|
|
"entropy": 7.086709833145141,
|
|
"epoch": 0.011060587440088485,
|
|
"grad_norm": 1.6875,
|
|
"learning_rate": 0.000247,
|
|
"loss": 6.6048,
|
|
"mean_token_accuracy": 0.13360075876116753,
|
|
"num_tokens": 420980.0,
|
|
"step": 495
|
|
},
|
|
{
|
|
"entropy": 6.981275844573974,
|
|
"epoch": 0.011172310545543924,
|
|
"grad_norm": 1.921875,
|
|
"learning_rate": 0.0002495,
|
|
"loss": 6.6585,
|
|
"mean_token_accuracy": 0.12793864756822587,
|
|
"num_tokens": 425454.0,
|
|
"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": 617629040640000.0,
|
|
"train_batch_size": 16,
|
|
"trial_name": null,
|
|
"trial_params": null
|
|
}
|