1046 lines
28 KiB
JSON
1046 lines
28 KiB
JSON
{
|
|
"best_global_step": null,
|
|
"best_metric": null,
|
|
"best_model_checkpoint": null,
|
|
"epoch": 0.39184952978056425,
|
|
"eval_steps": 500,
|
|
"global_step": 500,
|
|
"is_hyper_param_search": false,
|
|
"is_local_process_zero": true,
|
|
"is_world_process_zero": true,
|
|
"log_history": [
|
|
{
|
|
"entropy": 10.692053604125977,
|
|
"epoch": 0.003918495297805642,
|
|
"grad_norm": 13.875,
|
|
"learning_rate": 2e-06,
|
|
"loss": 10.821,
|
|
"mean_token_accuracy": 0.00013404826167970896,
|
|
"num_tokens": 9174.0,
|
|
"step": 5
|
|
},
|
|
{
|
|
"entropy": 10.691976261138915,
|
|
"epoch": 0.007836990595611285,
|
|
"grad_norm": 13.25,
|
|
"learning_rate": 4.5e-06,
|
|
"loss": 10.7352,
|
|
"mean_token_accuracy": 0.0,
|
|
"num_tokens": 17790.0,
|
|
"step": 10
|
|
},
|
|
{
|
|
"entropy": 10.6909686088562,
|
|
"epoch": 0.011755485893416929,
|
|
"grad_norm": 9.5,
|
|
"learning_rate": 7e-06,
|
|
"loss": 10.4692,
|
|
"mean_token_accuracy": 0.024429761280771345,
|
|
"num_tokens": 25850.0,
|
|
"step": 15
|
|
},
|
|
{
|
|
"entropy": 10.677701091766357,
|
|
"epoch": 0.01567398119122257,
|
|
"grad_norm": 6.59375,
|
|
"learning_rate": 9.5e-06,
|
|
"loss": 10.2158,
|
|
"mean_token_accuracy": 0.04213750921189785,
|
|
"num_tokens": 35194.0,
|
|
"step": 20
|
|
},
|
|
{
|
|
"entropy": 10.598451328277587,
|
|
"epoch": 0.019592476489028215,
|
|
"grad_norm": 4.375,
|
|
"learning_rate": 1.2e-05,
|
|
"loss": 9.9693,
|
|
"mean_token_accuracy": 0.042495733872056005,
|
|
"num_tokens": 44218.0,
|
|
"step": 25
|
|
},
|
|
{
|
|
"entropy": 10.572938823699952,
|
|
"epoch": 0.023510971786833857,
|
|
"grad_norm": 3.421875,
|
|
"learning_rate": 1.4500000000000002e-05,
|
|
"loss": 9.8009,
|
|
"mean_token_accuracy": 0.04398789629340172,
|
|
"num_tokens": 53397.0,
|
|
"step": 30
|
|
},
|
|
{
|
|
"entropy": 10.539127254486084,
|
|
"epoch": 0.0274294670846395,
|
|
"grad_norm": 2.921875,
|
|
"learning_rate": 1.7000000000000003e-05,
|
|
"loss": 9.6844,
|
|
"mean_token_accuracy": 0.04925155565142632,
|
|
"num_tokens": 62749.0,
|
|
"step": 35
|
|
},
|
|
{
|
|
"entropy": 10.50910243988037,
|
|
"epoch": 0.03134796238244514,
|
|
"grad_norm": 2.59375,
|
|
"learning_rate": 1.95e-05,
|
|
"loss": 9.6464,
|
|
"mean_token_accuracy": 0.04322206266224384,
|
|
"num_tokens": 71721.0,
|
|
"step": 40
|
|
},
|
|
{
|
|
"entropy": 10.538900089263915,
|
|
"epoch": 0.03526645768025078,
|
|
"grad_norm": 2.765625,
|
|
"learning_rate": 2.2e-05,
|
|
"loss": 9.5818,
|
|
"mean_token_accuracy": 0.04916957952082157,
|
|
"num_tokens": 79844.0,
|
|
"step": 45
|
|
},
|
|
{
|
|
"entropy": 10.527848815917968,
|
|
"epoch": 0.03918495297805643,
|
|
"grad_norm": 2.421875,
|
|
"learning_rate": 2.4500000000000003e-05,
|
|
"loss": 9.567,
|
|
"mean_token_accuracy": 0.05607832483947277,
|
|
"num_tokens": 88866.0,
|
|
"step": 50
|
|
},
|
|
{
|
|
"entropy": 10.449817848205566,
|
|
"epoch": 0.04310344827586207,
|
|
"grad_norm": 2.265625,
|
|
"learning_rate": 2.7e-05,
|
|
"loss": 9.4999,
|
|
"mean_token_accuracy": 0.06071024388074875,
|
|
"num_tokens": 97918.0,
|
|
"step": 55
|
|
},
|
|
{
|
|
"entropy": 10.289769554138184,
|
|
"epoch": 0.047021943573667714,
|
|
"grad_norm": 2.203125,
|
|
"learning_rate": 2.95e-05,
|
|
"loss": 9.42,
|
|
"mean_token_accuracy": 0.06239944435656071,
|
|
"num_tokens": 107043.0,
|
|
"step": 60
|
|
},
|
|
{
|
|
"entropy": 10.339974212646485,
|
|
"epoch": 0.050940438871473356,
|
|
"grad_norm": 2.328125,
|
|
"learning_rate": 3.2e-05,
|
|
"loss": 9.3525,
|
|
"mean_token_accuracy": 0.06377544328570366,
|
|
"num_tokens": 116000.0,
|
|
"step": 65
|
|
},
|
|
{
|
|
"entropy": 10.358237171173096,
|
|
"epoch": 0.054858934169279,
|
|
"grad_norm": 2.15625,
|
|
"learning_rate": 3.4500000000000005e-05,
|
|
"loss": 9.2722,
|
|
"mean_token_accuracy": 0.06179721914231777,
|
|
"num_tokens": 124559.0,
|
|
"step": 70
|
|
},
|
|
{
|
|
"entropy": 10.306263256072999,
|
|
"epoch": 0.05877742946708464,
|
|
"grad_norm": 2.015625,
|
|
"learning_rate": 3.7e-05,
|
|
"loss": 9.1558,
|
|
"mean_token_accuracy": 0.0636992421001196,
|
|
"num_tokens": 132868.0,
|
|
"step": 75
|
|
},
|
|
{
|
|
"entropy": 10.310817909240722,
|
|
"epoch": 0.06269592476489028,
|
|
"grad_norm": 2.09375,
|
|
"learning_rate": 3.95e-05,
|
|
"loss": 9.1437,
|
|
"mean_token_accuracy": 0.06109033823013306,
|
|
"num_tokens": 141286.0,
|
|
"step": 80
|
|
},
|
|
{
|
|
"entropy": 10.31660270690918,
|
|
"epoch": 0.06661442006269593,
|
|
"grad_norm": 1.8828125,
|
|
"learning_rate": 4.2000000000000004e-05,
|
|
"loss": 9.0017,
|
|
"mean_token_accuracy": 0.06191765181720257,
|
|
"num_tokens": 150406.0,
|
|
"step": 85
|
|
},
|
|
{
|
|
"entropy": 10.19018259048462,
|
|
"epoch": 0.07053291536050156,
|
|
"grad_norm": 1.9140625,
|
|
"learning_rate": 4.45e-05,
|
|
"loss": 8.9406,
|
|
"mean_token_accuracy": 0.06189325153827667,
|
|
"num_tokens": 158770.0,
|
|
"step": 90
|
|
},
|
|
{
|
|
"entropy": 10.224696159362793,
|
|
"epoch": 0.07445141065830721,
|
|
"grad_norm": 1.8515625,
|
|
"learning_rate": 4.7000000000000004e-05,
|
|
"loss": 8.8329,
|
|
"mean_token_accuracy": 0.06557972468435765,
|
|
"num_tokens": 167763.0,
|
|
"step": 95
|
|
},
|
|
{
|
|
"entropy": 10.152040767669678,
|
|
"epoch": 0.07836990595611286,
|
|
"grad_norm": 1.5859375,
|
|
"learning_rate": 4.9500000000000004e-05,
|
|
"loss": 8.6853,
|
|
"mean_token_accuracy": 0.06438785269856453,
|
|
"num_tokens": 177306.0,
|
|
"step": 100
|
|
},
|
|
{
|
|
"entropy": 10.053633880615234,
|
|
"epoch": 0.0822884012539185,
|
|
"grad_norm": 1.59375,
|
|
"learning_rate": 5.2e-05,
|
|
"loss": 8.5993,
|
|
"mean_token_accuracy": 0.06454790569841862,
|
|
"num_tokens": 186014.0,
|
|
"step": 105
|
|
},
|
|
{
|
|
"entropy": 9.987529850006103,
|
|
"epoch": 0.08620689655172414,
|
|
"grad_norm": 1.625,
|
|
"learning_rate": 5.45e-05,
|
|
"loss": 8.4292,
|
|
"mean_token_accuracy": 0.06873111054301262,
|
|
"num_tokens": 194122.0,
|
|
"step": 110
|
|
},
|
|
{
|
|
"entropy": 9.858163261413575,
|
|
"epoch": 0.09012539184952978,
|
|
"grad_norm": 1.703125,
|
|
"learning_rate": 5.7e-05,
|
|
"loss": 8.3325,
|
|
"mean_token_accuracy": 0.06681315004825591,
|
|
"num_tokens": 203097.0,
|
|
"step": 115
|
|
},
|
|
{
|
|
"entropy": 9.74828462600708,
|
|
"epoch": 0.09404388714733543,
|
|
"grad_norm": 1.6015625,
|
|
"learning_rate": 5.9499999999999996e-05,
|
|
"loss": 8.2069,
|
|
"mean_token_accuracy": 0.06950863264501095,
|
|
"num_tokens": 211413.0,
|
|
"step": 120
|
|
},
|
|
{
|
|
"entropy": 9.593859004974366,
|
|
"epoch": 0.09796238244514106,
|
|
"grad_norm": 1.3828125,
|
|
"learning_rate": 6.2e-05,
|
|
"loss": 8.1517,
|
|
"mean_token_accuracy": 0.06460827700793743,
|
|
"num_tokens": 220550.0,
|
|
"step": 125
|
|
},
|
|
{
|
|
"entropy": 9.457711124420166,
|
|
"epoch": 0.10188087774294671,
|
|
"grad_norm": 1.3828125,
|
|
"learning_rate": 6.450000000000001e-05,
|
|
"loss": 7.9844,
|
|
"mean_token_accuracy": 0.072696752846241,
|
|
"num_tokens": 229197.0,
|
|
"step": 130
|
|
},
|
|
{
|
|
"entropy": 9.283110237121582,
|
|
"epoch": 0.10579937304075235,
|
|
"grad_norm": 1.359375,
|
|
"learning_rate": 6.7e-05,
|
|
"loss": 7.9816,
|
|
"mean_token_accuracy": 0.06841139532625676,
|
|
"num_tokens": 238479.0,
|
|
"step": 135
|
|
},
|
|
{
|
|
"entropy": 9.081107711791992,
|
|
"epoch": 0.109717868338558,
|
|
"grad_norm": 1.515625,
|
|
"learning_rate": 6.950000000000001e-05,
|
|
"loss": 7.8324,
|
|
"mean_token_accuracy": 0.07079579271376132,
|
|
"num_tokens": 246318.0,
|
|
"step": 140
|
|
},
|
|
{
|
|
"entropy": 8.931202793121338,
|
|
"epoch": 0.11363636363636363,
|
|
"grad_norm": 1.3984375,
|
|
"learning_rate": 7.2e-05,
|
|
"loss": 7.8567,
|
|
"mean_token_accuracy": 0.06792247369885444,
|
|
"num_tokens": 254783.0,
|
|
"step": 145
|
|
},
|
|
{
|
|
"entropy": 8.767581272125245,
|
|
"epoch": 0.11755485893416928,
|
|
"grad_norm": 1.375,
|
|
"learning_rate": 7.45e-05,
|
|
"loss": 7.6935,
|
|
"mean_token_accuracy": 0.0719048798084259,
|
|
"num_tokens": 263416.0,
|
|
"step": 150
|
|
},
|
|
{
|
|
"entropy": 8.662985897064209,
|
|
"epoch": 0.12147335423197492,
|
|
"grad_norm": 1.671875,
|
|
"learning_rate": 7.7e-05,
|
|
"loss": 7.712,
|
|
"mean_token_accuracy": 0.07060935273766518,
|
|
"num_tokens": 271930.0,
|
|
"step": 155
|
|
},
|
|
{
|
|
"entropy": 8.506498527526855,
|
|
"epoch": 0.12539184952978055,
|
|
"grad_norm": 1.3671875,
|
|
"learning_rate": 7.950000000000001e-05,
|
|
"loss": 7.619,
|
|
"mean_token_accuracy": 0.07228946462273597,
|
|
"num_tokens": 280546.0,
|
|
"step": 160
|
|
},
|
|
{
|
|
"entropy": 8.340063381195069,
|
|
"epoch": 0.12931034482758622,
|
|
"grad_norm": 1.359375,
|
|
"learning_rate": 8.2e-05,
|
|
"loss": 7.6031,
|
|
"mean_token_accuracy": 0.07542856186628341,
|
|
"num_tokens": 288813.0,
|
|
"step": 165
|
|
},
|
|
{
|
|
"entropy": 8.317859268188476,
|
|
"epoch": 0.13322884012539185,
|
|
"grad_norm": 1.2265625,
|
|
"learning_rate": 8.450000000000001e-05,
|
|
"loss": 7.5757,
|
|
"mean_token_accuracy": 0.07388538345694542,
|
|
"num_tokens": 297966.0,
|
|
"step": 170
|
|
},
|
|
{
|
|
"entropy": 8.287030601501465,
|
|
"epoch": 0.1371473354231975,
|
|
"grad_norm": 1.2421875,
|
|
"learning_rate": 8.7e-05,
|
|
"loss": 7.6313,
|
|
"mean_token_accuracy": 0.07094714678823948,
|
|
"num_tokens": 307135.0,
|
|
"step": 175
|
|
},
|
|
{
|
|
"entropy": 8.15151834487915,
|
|
"epoch": 0.14106583072100312,
|
|
"grad_norm": 1.1875,
|
|
"learning_rate": 8.95e-05,
|
|
"loss": 7.543,
|
|
"mean_token_accuracy": 0.07589463070034981,
|
|
"num_tokens": 315546.0,
|
|
"step": 180
|
|
},
|
|
{
|
|
"entropy": 8.161028385162354,
|
|
"epoch": 0.14498432601880878,
|
|
"grad_norm": 1.2265625,
|
|
"learning_rate": 9.2e-05,
|
|
"loss": 7.5249,
|
|
"mean_token_accuracy": 0.0759074218571186,
|
|
"num_tokens": 323930.0,
|
|
"step": 185
|
|
},
|
|
{
|
|
"entropy": 8.058754682540894,
|
|
"epoch": 0.14890282131661442,
|
|
"grad_norm": 1.2578125,
|
|
"learning_rate": 9.45e-05,
|
|
"loss": 7.392,
|
|
"mean_token_accuracy": 0.08224981352686882,
|
|
"num_tokens": 332419.0,
|
|
"step": 190
|
|
},
|
|
{
|
|
"entropy": 8.08242244720459,
|
|
"epoch": 0.15282131661442006,
|
|
"grad_norm": 1.265625,
|
|
"learning_rate": 9.7e-05,
|
|
"loss": 7.4781,
|
|
"mean_token_accuracy": 0.07802417948842048,
|
|
"num_tokens": 341362.0,
|
|
"step": 195
|
|
},
|
|
{
|
|
"entropy": 7.985074710845947,
|
|
"epoch": 0.15673981191222572,
|
|
"grad_norm": 1.234375,
|
|
"learning_rate": 9.95e-05,
|
|
"loss": 7.3972,
|
|
"mean_token_accuracy": 0.07822780273854732,
|
|
"num_tokens": 349395.0,
|
|
"step": 200
|
|
},
|
|
{
|
|
"entropy": 8.01458387374878,
|
|
"epoch": 0.16065830721003135,
|
|
"grad_norm": 1.546875,
|
|
"learning_rate": 0.000102,
|
|
"loss": 7.4184,
|
|
"mean_token_accuracy": 0.07329343855381013,
|
|
"num_tokens": 358413.0,
|
|
"step": 205
|
|
},
|
|
{
|
|
"entropy": 7.919571828842163,
|
|
"epoch": 0.164576802507837,
|
|
"grad_norm": 1.484375,
|
|
"learning_rate": 0.00010449999999999999,
|
|
"loss": 7.5032,
|
|
"mean_token_accuracy": 0.07889928370714187,
|
|
"num_tokens": 366489.0,
|
|
"step": 210
|
|
},
|
|
{
|
|
"entropy": 7.988800716400147,
|
|
"epoch": 0.16849529780564262,
|
|
"grad_norm": 1.265625,
|
|
"learning_rate": 0.000107,
|
|
"loss": 7.3554,
|
|
"mean_token_accuracy": 0.07564577832818031,
|
|
"num_tokens": 375335.0,
|
|
"step": 215
|
|
},
|
|
{
|
|
"entropy": 7.991494131088257,
|
|
"epoch": 0.1724137931034483,
|
|
"grad_norm": 1.1953125,
|
|
"learning_rate": 0.0001095,
|
|
"loss": 7.407,
|
|
"mean_token_accuracy": 0.08004131242632866,
|
|
"num_tokens": 384276.0,
|
|
"step": 220
|
|
},
|
|
{
|
|
"entropy": 7.928954362869263,
|
|
"epoch": 0.17633228840125392,
|
|
"grad_norm": 1.21875,
|
|
"learning_rate": 0.000112,
|
|
"loss": 7.447,
|
|
"mean_token_accuracy": 0.07608951888978481,
|
|
"num_tokens": 393571.0,
|
|
"step": 225
|
|
},
|
|
{
|
|
"entropy": 7.913915538787842,
|
|
"epoch": 0.18025078369905956,
|
|
"grad_norm": 1.421875,
|
|
"learning_rate": 0.0001145,
|
|
"loss": 7.3764,
|
|
"mean_token_accuracy": 0.08197141736745835,
|
|
"num_tokens": 401865.0,
|
|
"step": 230
|
|
},
|
|
{
|
|
"entropy": 7.777705907821655,
|
|
"epoch": 0.1841692789968652,
|
|
"grad_norm": 1.15625,
|
|
"learning_rate": 0.00011700000000000001,
|
|
"loss": 7.3381,
|
|
"mean_token_accuracy": 0.08282902240753173,
|
|
"num_tokens": 410518.0,
|
|
"step": 235
|
|
},
|
|
{
|
|
"entropy": 7.857710313796997,
|
|
"epoch": 0.18808777429467086,
|
|
"grad_norm": 1.21875,
|
|
"learning_rate": 0.00011949999999999999,
|
|
"loss": 7.356,
|
|
"mean_token_accuracy": 0.08076885864138603,
|
|
"num_tokens": 419769.0,
|
|
"step": 240
|
|
},
|
|
{
|
|
"entropy": 7.7723448276519775,
|
|
"epoch": 0.1920062695924765,
|
|
"grad_norm": 1.265625,
|
|
"learning_rate": 0.000122,
|
|
"loss": 7.2799,
|
|
"mean_token_accuracy": 0.08509492203593254,
|
|
"num_tokens": 428204.0,
|
|
"step": 245
|
|
},
|
|
{
|
|
"entropy": 7.8825091361999515,
|
|
"epoch": 0.19592476489028213,
|
|
"grad_norm": 1.2421875,
|
|
"learning_rate": 0.0001245,
|
|
"loss": 7.3833,
|
|
"mean_token_accuracy": 0.07587559074163437,
|
|
"num_tokens": 436594.0,
|
|
"step": 250
|
|
},
|
|
{
|
|
"entropy": 7.804588508605957,
|
|
"epoch": 0.19984326018808776,
|
|
"grad_norm": 1.4609375,
|
|
"learning_rate": 0.000127,
|
|
"loss": 7.1619,
|
|
"mean_token_accuracy": 0.08455468490719795,
|
|
"num_tokens": 444645.0,
|
|
"step": 255
|
|
},
|
|
{
|
|
"entropy": 7.734504652023316,
|
|
"epoch": 0.20376175548589343,
|
|
"grad_norm": 1.4453125,
|
|
"learning_rate": 0.0001295,
|
|
"loss": 7.2047,
|
|
"mean_token_accuracy": 0.0890711173415184,
|
|
"num_tokens": 453016.0,
|
|
"step": 260
|
|
},
|
|
{
|
|
"entropy": 7.689095687866211,
|
|
"epoch": 0.20768025078369906,
|
|
"grad_norm": 1.3828125,
|
|
"learning_rate": 0.000132,
|
|
"loss": 7.2874,
|
|
"mean_token_accuracy": 0.08078921884298325,
|
|
"num_tokens": 462116.0,
|
|
"step": 265
|
|
},
|
|
{
|
|
"entropy": 7.745668983459472,
|
|
"epoch": 0.2115987460815047,
|
|
"grad_norm": 1.265625,
|
|
"learning_rate": 0.00013450000000000002,
|
|
"loss": 7.2359,
|
|
"mean_token_accuracy": 0.08409795872867107,
|
|
"num_tokens": 470807.0,
|
|
"step": 270
|
|
},
|
|
{
|
|
"entropy": 7.6965230941772464,
|
|
"epoch": 0.21551724137931033,
|
|
"grad_norm": 1.546875,
|
|
"learning_rate": 0.00013700000000000002,
|
|
"loss": 7.3247,
|
|
"mean_token_accuracy": 0.08189421519637108,
|
|
"num_tokens": 479592.0,
|
|
"step": 275
|
|
},
|
|
{
|
|
"entropy": 7.784852504730225,
|
|
"epoch": 0.219435736677116,
|
|
"grad_norm": 1.25,
|
|
"learning_rate": 0.0001395,
|
|
"loss": 7.2423,
|
|
"mean_token_accuracy": 0.08114428222179412,
|
|
"num_tokens": 488107.0,
|
|
"step": 280
|
|
},
|
|
{
|
|
"entropy": 7.663507890701294,
|
|
"epoch": 0.22335423197492163,
|
|
"grad_norm": 1.359375,
|
|
"learning_rate": 0.00014199999999999998,
|
|
"loss": 7.2547,
|
|
"mean_token_accuracy": 0.08439974337816239,
|
|
"num_tokens": 496775.0,
|
|
"step": 285
|
|
},
|
|
{
|
|
"entropy": 7.740933275222778,
|
|
"epoch": 0.22727272727272727,
|
|
"grad_norm": 1.4609375,
|
|
"learning_rate": 0.0001445,
|
|
"loss": 7.2816,
|
|
"mean_token_accuracy": 0.08367991298437119,
|
|
"num_tokens": 505415.0,
|
|
"step": 290
|
|
},
|
|
{
|
|
"entropy": 7.680744886398315,
|
|
"epoch": 0.23119122257053293,
|
|
"grad_norm": 1.28125,
|
|
"learning_rate": 0.000147,
|
|
"loss": 7.2656,
|
|
"mean_token_accuracy": 0.08193187937140464,
|
|
"num_tokens": 514447.0,
|
|
"step": 295
|
|
},
|
|
{
|
|
"entropy": 7.735220909118652,
|
|
"epoch": 0.23510971786833856,
|
|
"grad_norm": 1.234375,
|
|
"learning_rate": 0.0001495,
|
|
"loss": 7.2519,
|
|
"mean_token_accuracy": 0.086246570199728,
|
|
"num_tokens": 522998.0,
|
|
"step": 300
|
|
},
|
|
{
|
|
"entropy": 7.693396520614624,
|
|
"epoch": 0.2390282131661442,
|
|
"grad_norm": 1.34375,
|
|
"learning_rate": 0.000152,
|
|
"loss": 7.2173,
|
|
"mean_token_accuracy": 0.08238770589232444,
|
|
"num_tokens": 531542.0,
|
|
"step": 305
|
|
},
|
|
{
|
|
"entropy": 7.604292917251587,
|
|
"epoch": 0.24294670846394983,
|
|
"grad_norm": 1.3359375,
|
|
"learning_rate": 0.00015450000000000001,
|
|
"loss": 7.1457,
|
|
"mean_token_accuracy": 0.08177541941404343,
|
|
"num_tokens": 540143.0,
|
|
"step": 310
|
|
},
|
|
{
|
|
"entropy": 7.716537094116211,
|
|
"epoch": 0.2468652037617555,
|
|
"grad_norm": 1.2421875,
|
|
"learning_rate": 0.000157,
|
|
"loss": 7.2444,
|
|
"mean_token_accuracy": 0.08475765213370323,
|
|
"num_tokens": 549156.0,
|
|
"step": 315
|
|
},
|
|
{
|
|
"entropy": 7.560835886001587,
|
|
"epoch": 0.2507836990595611,
|
|
"grad_norm": 1.4140625,
|
|
"learning_rate": 0.0001595,
|
|
"loss": 7.0897,
|
|
"mean_token_accuracy": 0.08665165156126023,
|
|
"num_tokens": 557522.0,
|
|
"step": 320
|
|
},
|
|
{
|
|
"entropy": 7.5509428024292,
|
|
"epoch": 0.2547021943573668,
|
|
"grad_norm": 1.3125,
|
|
"learning_rate": 0.000162,
|
|
"loss": 7.1604,
|
|
"mean_token_accuracy": 0.0852648138999939,
|
|
"num_tokens": 566650.0,
|
|
"step": 325
|
|
},
|
|
{
|
|
"entropy": 7.643481254577637,
|
|
"epoch": 0.25862068965517243,
|
|
"grad_norm": 1.375,
|
|
"learning_rate": 0.00016450000000000001,
|
|
"loss": 7.1157,
|
|
"mean_token_accuracy": 0.08511288538575172,
|
|
"num_tokens": 576219.0,
|
|
"step": 330
|
|
},
|
|
{
|
|
"entropy": 7.535575675964355,
|
|
"epoch": 0.26253918495297807,
|
|
"grad_norm": 1.6328125,
|
|
"learning_rate": 0.00016700000000000002,
|
|
"loss": 7.2172,
|
|
"mean_token_accuracy": 0.08702815398573875,
|
|
"num_tokens": 584304.0,
|
|
"step": 335
|
|
},
|
|
{
|
|
"entropy": 7.477091073989868,
|
|
"epoch": 0.2664576802507837,
|
|
"grad_norm": 1.296875,
|
|
"learning_rate": 0.00016950000000000003,
|
|
"loss": 7.1379,
|
|
"mean_token_accuracy": 0.08963485434651375,
|
|
"num_tokens": 593163.0,
|
|
"step": 340
|
|
},
|
|
{
|
|
"entropy": 7.577557039260864,
|
|
"epoch": 0.27037617554858934,
|
|
"grad_norm": 1.4921875,
|
|
"learning_rate": 0.00017199999999999998,
|
|
"loss": 7.0509,
|
|
"mean_token_accuracy": 0.08813998103141785,
|
|
"num_tokens": 602077.0,
|
|
"step": 345
|
|
},
|
|
{
|
|
"entropy": 7.504067659378052,
|
|
"epoch": 0.274294670846395,
|
|
"grad_norm": 1.4765625,
|
|
"learning_rate": 0.00017449999999999999,
|
|
"loss": 7.1645,
|
|
"mean_token_accuracy": 0.08567056804895401,
|
|
"num_tokens": 610009.0,
|
|
"step": 350
|
|
},
|
|
{
|
|
"entropy": 7.615056896209717,
|
|
"epoch": 0.2782131661442006,
|
|
"grad_norm": 1.171875,
|
|
"learning_rate": 0.000177,
|
|
"loss": 7.1942,
|
|
"mean_token_accuracy": 0.08498035669326783,
|
|
"num_tokens": 619282.0,
|
|
"step": 355
|
|
},
|
|
{
|
|
"entropy": 7.465681743621826,
|
|
"epoch": 0.28213166144200624,
|
|
"grad_norm": 1.4453125,
|
|
"learning_rate": 0.0001795,
|
|
"loss": 7.1301,
|
|
"mean_token_accuracy": 0.08965396955609321,
|
|
"num_tokens": 628138.0,
|
|
"step": 360
|
|
},
|
|
{
|
|
"entropy": 7.515298175811767,
|
|
"epoch": 0.28605015673981193,
|
|
"grad_norm": 1.375,
|
|
"learning_rate": 0.000182,
|
|
"loss": 6.9813,
|
|
"mean_token_accuracy": 0.0954482302069664,
|
|
"num_tokens": 637021.0,
|
|
"step": 365
|
|
},
|
|
{
|
|
"entropy": 7.504141473770142,
|
|
"epoch": 0.28996865203761757,
|
|
"grad_norm": 1.1796875,
|
|
"learning_rate": 0.0001845,
|
|
"loss": 7.1534,
|
|
"mean_token_accuracy": 0.0818020910024643,
|
|
"num_tokens": 646703.0,
|
|
"step": 370
|
|
},
|
|
{
|
|
"entropy": 7.471045875549317,
|
|
"epoch": 0.2938871473354232,
|
|
"grad_norm": 1.3359375,
|
|
"learning_rate": 0.000187,
|
|
"loss": 7.0346,
|
|
"mean_token_accuracy": 0.09074857905507087,
|
|
"num_tokens": 655616.0,
|
|
"step": 375
|
|
},
|
|
{
|
|
"entropy": 7.431417989730835,
|
|
"epoch": 0.29780564263322884,
|
|
"grad_norm": 1.34375,
|
|
"learning_rate": 0.0001895,
|
|
"loss": 7.0272,
|
|
"mean_token_accuracy": 0.09865856096148491,
|
|
"num_tokens": 663783.0,
|
|
"step": 380
|
|
},
|
|
{
|
|
"entropy": 7.422271394729615,
|
|
"epoch": 0.3017241379310345,
|
|
"grad_norm": 1.2890625,
|
|
"learning_rate": 0.000192,
|
|
"loss": 7.0382,
|
|
"mean_token_accuracy": 0.09664822444319725,
|
|
"num_tokens": 671855.0,
|
|
"step": 385
|
|
},
|
|
{
|
|
"entropy": 7.441711044311523,
|
|
"epoch": 0.3056426332288401,
|
|
"grad_norm": 1.4921875,
|
|
"learning_rate": 0.0001945,
|
|
"loss": 7.0574,
|
|
"mean_token_accuracy": 0.0905517354607582,
|
|
"num_tokens": 680981.0,
|
|
"step": 390
|
|
},
|
|
{
|
|
"entropy": 7.409027004241944,
|
|
"epoch": 0.30956112852664575,
|
|
"grad_norm": 1.328125,
|
|
"learning_rate": 0.00019700000000000002,
|
|
"loss": 6.9713,
|
|
"mean_token_accuracy": 0.09158209860324859,
|
|
"num_tokens": 689294.0,
|
|
"step": 395
|
|
},
|
|
{
|
|
"entropy": 7.345873880386352,
|
|
"epoch": 0.31347962382445144,
|
|
"grad_norm": 1.2421875,
|
|
"learning_rate": 0.00019950000000000002,
|
|
"loss": 6.936,
|
|
"mean_token_accuracy": 0.09149299785494805,
|
|
"num_tokens": 697703.0,
|
|
"step": 400
|
|
},
|
|
{
|
|
"entropy": 7.485394239425659,
|
|
"epoch": 0.31739811912225707,
|
|
"grad_norm": 1.2109375,
|
|
"learning_rate": 0.000202,
|
|
"loss": 7.0726,
|
|
"mean_token_accuracy": 0.09204713851213456,
|
|
"num_tokens": 706792.0,
|
|
"step": 405
|
|
},
|
|
{
|
|
"entropy": 7.409369468688965,
|
|
"epoch": 0.3213166144200627,
|
|
"grad_norm": 1.3203125,
|
|
"learning_rate": 0.00020449999999999998,
|
|
"loss": 6.9865,
|
|
"mean_token_accuracy": 0.09500117674469948,
|
|
"num_tokens": 715864.0,
|
|
"step": 410
|
|
},
|
|
{
|
|
"entropy": 7.2893208980560305,
|
|
"epoch": 0.32523510971786834,
|
|
"grad_norm": 1.40625,
|
|
"learning_rate": 0.000207,
|
|
"loss": 7.0287,
|
|
"mean_token_accuracy": 0.0940815657377243,
|
|
"num_tokens": 723921.0,
|
|
"step": 415
|
|
},
|
|
{
|
|
"entropy": 7.437120723724365,
|
|
"epoch": 0.329153605015674,
|
|
"grad_norm": 1.609375,
|
|
"learning_rate": 0.0002095,
|
|
"loss": 7.0524,
|
|
"mean_token_accuracy": 0.08924909234046936,
|
|
"num_tokens": 732797.0,
|
|
"step": 420
|
|
},
|
|
{
|
|
"entropy": 7.3198113441467285,
|
|
"epoch": 0.3330721003134796,
|
|
"grad_norm": 1.3203125,
|
|
"learning_rate": 0.000212,
|
|
"loss": 6.8983,
|
|
"mean_token_accuracy": 0.10327747613191604,
|
|
"num_tokens": 741248.0,
|
|
"step": 425
|
|
},
|
|
{
|
|
"entropy": 7.3485208511352536,
|
|
"epoch": 0.33699059561128525,
|
|
"grad_norm": 1.359375,
|
|
"learning_rate": 0.0002145,
|
|
"loss": 7.0034,
|
|
"mean_token_accuracy": 0.0921932153403759,
|
|
"num_tokens": 749766.0,
|
|
"step": 430
|
|
},
|
|
{
|
|
"entropy": 7.331899881362915,
|
|
"epoch": 0.3409090909090909,
|
|
"grad_norm": 1.296875,
|
|
"learning_rate": 0.00021700000000000002,
|
|
"loss": 6.9561,
|
|
"mean_token_accuracy": 0.09300489202141762,
|
|
"num_tokens": 758695.0,
|
|
"step": 435
|
|
},
|
|
{
|
|
"entropy": 7.339229249954224,
|
|
"epoch": 0.3448275862068966,
|
|
"grad_norm": 1.375,
|
|
"learning_rate": 0.0002195,
|
|
"loss": 6.9566,
|
|
"mean_token_accuracy": 0.09157029166817665,
|
|
"num_tokens": 767624.0,
|
|
"step": 440
|
|
},
|
|
{
|
|
"entropy": 7.343198585510254,
|
|
"epoch": 0.3487460815047022,
|
|
"grad_norm": 1.4296875,
|
|
"learning_rate": 0.000222,
|
|
"loss": 6.9819,
|
|
"mean_token_accuracy": 0.09128969609737396,
|
|
"num_tokens": 776616.0,
|
|
"step": 445
|
|
},
|
|
{
|
|
"entropy": 7.2496236801147464,
|
|
"epoch": 0.35266457680250785,
|
|
"grad_norm": 1.328125,
|
|
"learning_rate": 0.0002245,
|
|
"loss": 6.9895,
|
|
"mean_token_accuracy": 0.0921534925699234,
|
|
"num_tokens": 786147.0,
|
|
"step": 450
|
|
},
|
|
{
|
|
"entropy": 7.356827306747436,
|
|
"epoch": 0.3565830721003135,
|
|
"grad_norm": 1.3359375,
|
|
"learning_rate": 0.00022700000000000002,
|
|
"loss": 7.0808,
|
|
"mean_token_accuracy": 0.08939703032374383,
|
|
"num_tokens": 795213.0,
|
|
"step": 455
|
|
},
|
|
{
|
|
"entropy": 7.332914733886719,
|
|
"epoch": 0.3605015673981191,
|
|
"grad_norm": 1.4453125,
|
|
"learning_rate": 0.00022950000000000002,
|
|
"loss": 6.8956,
|
|
"mean_token_accuracy": 0.09629729315638542,
|
|
"num_tokens": 803118.0,
|
|
"step": 460
|
|
},
|
|
{
|
|
"entropy": 7.346728849411011,
|
|
"epoch": 0.36442006269592475,
|
|
"grad_norm": 1.2109375,
|
|
"learning_rate": 0.00023200000000000003,
|
|
"loss": 6.8641,
|
|
"mean_token_accuracy": 0.10322200134396553,
|
|
"num_tokens": 811358.0,
|
|
"step": 465
|
|
},
|
|
{
|
|
"entropy": 7.147189569473267,
|
|
"epoch": 0.3683385579937304,
|
|
"grad_norm": 1.4140625,
|
|
"learning_rate": 0.00023449999999999998,
|
|
"loss": 6.8718,
|
|
"mean_token_accuracy": 0.09597681015729904,
|
|
"num_tokens": 819653.0,
|
|
"step": 470
|
|
},
|
|
{
|
|
"entropy": 7.195614671707153,
|
|
"epoch": 0.3722570532915361,
|
|
"grad_norm": 1.390625,
|
|
"learning_rate": 0.000237,
|
|
"loss": 6.9088,
|
|
"mean_token_accuracy": 0.09375306144356728,
|
|
"num_tokens": 828462.0,
|
|
"step": 475
|
|
},
|
|
{
|
|
"entropy": 7.361067628860473,
|
|
"epoch": 0.3761755485893417,
|
|
"grad_norm": 1.265625,
|
|
"learning_rate": 0.0002395,
|
|
"loss": 6.9995,
|
|
"mean_token_accuracy": 0.09758970066905022,
|
|
"num_tokens": 836204.0,
|
|
"step": 480
|
|
},
|
|
{
|
|
"entropy": 7.163541269302368,
|
|
"epoch": 0.38009404388714735,
|
|
"grad_norm": 1.4609375,
|
|
"learning_rate": 0.000242,
|
|
"loss": 6.8697,
|
|
"mean_token_accuracy": 0.1009294331073761,
|
|
"num_tokens": 845032.0,
|
|
"step": 485
|
|
},
|
|
{
|
|
"entropy": 7.267932987213134,
|
|
"epoch": 0.384012539184953,
|
|
"grad_norm": 1.40625,
|
|
"learning_rate": 0.0002445,
|
|
"loss": 6.8906,
|
|
"mean_token_accuracy": 0.10076733604073525,
|
|
"num_tokens": 853324.0,
|
|
"step": 490
|
|
},
|
|
{
|
|
"entropy": 7.26249213218689,
|
|
"epoch": 0.3879310344827586,
|
|
"grad_norm": 1.4609375,
|
|
"learning_rate": 0.000247,
|
|
"loss": 6.8639,
|
|
"mean_token_accuracy": 0.1036192737519741,
|
|
"num_tokens": 861640.0,
|
|
"step": 495
|
|
},
|
|
{
|
|
"entropy": 7.264737176895141,
|
|
"epoch": 0.39184952978056425,
|
|
"grad_norm": 1.3828125,
|
|
"learning_rate": 0.0002495,
|
|
"loss": 7.047,
|
|
"mean_token_accuracy": 0.08870344087481499,
|
|
"num_tokens": 870758.0,
|
|
"step": 500
|
|
},
|
|
{
|
|
"epoch": 0.39184952978056425,
|
|
"eval_entropy": 7.3145306674084924,
|
|
"eval_loss": 7.030722141265869,
|
|
"eval_mean_token_accuracy": 0.09521551475708567,
|
|
"eval_num_tokens": 870758.0,
|
|
"eval_runtime": 2.8925,
|
|
"eval_samples_per_second": 1425.051,
|
|
"eval_steps_per_second": 178.391,
|
|
"step": 500
|
|
}
|
|
],
|
|
"logging_steps": 5,
|
|
"max_steps": 12750,
|
|
"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": 1156331556864000.0,
|
|
"train_batch_size": 16,
|
|
"trial_name": null,
|
|
"trial_params": null
|
|
}
|