1035 lines
27 KiB
JSON
1035 lines
27 KiB
JSON
{
|
|
"best_global_step": null,
|
|
"best_metric": null,
|
|
"best_model_checkpoint": null,
|
|
"epoch": 0.009984224924619102,
|
|
"eval_steps": 500,
|
|
"global_step": 500,
|
|
"is_hyper_param_search": false,
|
|
"is_local_process_zero": true,
|
|
"is_world_process_zero": true,
|
|
"log_history": [
|
|
{
|
|
"entropy": 10.692037105560303,
|
|
"epoch": 9.984224924619101e-05,
|
|
"grad_norm": 15.3125,
|
|
"learning_rate": 2e-06,
|
|
"loss": 10.6672,
|
|
"mean_token_accuracy": 0.0,
|
|
"num_tokens": 5422.0,
|
|
"step": 5
|
|
},
|
|
{
|
|
"entropy": 10.691985416412354,
|
|
"epoch": 0.00019968449849238202,
|
|
"grad_norm": 15.125,
|
|
"learning_rate": 4.5e-06,
|
|
"loss": 10.6253,
|
|
"mean_token_accuracy": 0.00012738853693008422,
|
|
"num_tokens": 11441.0,
|
|
"step": 10
|
|
},
|
|
{
|
|
"entropy": 10.690535736083984,
|
|
"epoch": 0.00029952674773857304,
|
|
"grad_norm": 12.375,
|
|
"learning_rate": 7e-06,
|
|
"loss": 10.3697,
|
|
"mean_token_accuracy": 0.03621518523432314,
|
|
"num_tokens": 17413.0,
|
|
"step": 15
|
|
},
|
|
{
|
|
"entropy": 10.663197803497315,
|
|
"epoch": 0.00039936899698476405,
|
|
"grad_norm": 7.4375,
|
|
"learning_rate": 9.5e-06,
|
|
"loss": 10.0374,
|
|
"mean_token_accuracy": 0.05610103793442249,
|
|
"num_tokens": 23280.0,
|
|
"step": 20
|
|
},
|
|
{
|
|
"entropy": 10.522790908813477,
|
|
"epoch": 0.0004992112462309551,
|
|
"grad_norm": 5.1875,
|
|
"learning_rate": 1.2e-05,
|
|
"loss": 9.7254,
|
|
"mean_token_accuracy": 0.05437180008739233,
|
|
"num_tokens": 28643.0,
|
|
"step": 25
|
|
},
|
|
{
|
|
"entropy": 10.49898328781128,
|
|
"epoch": 0.0005990534954771461,
|
|
"grad_norm": 4.21875,
|
|
"learning_rate": 1.4500000000000002e-05,
|
|
"loss": 9.5672,
|
|
"mean_token_accuracy": 0.059307342022657396,
|
|
"num_tokens": 34253.0,
|
|
"step": 30
|
|
},
|
|
{
|
|
"entropy": 10.442592239379882,
|
|
"epoch": 0.0006988957447233371,
|
|
"grad_norm": 3.609375,
|
|
"learning_rate": 1.7000000000000003e-05,
|
|
"loss": 9.5534,
|
|
"mean_token_accuracy": 0.053936305642127993,
|
|
"num_tokens": 39990.0,
|
|
"step": 35
|
|
},
|
|
{
|
|
"entropy": 10.463098812103272,
|
|
"epoch": 0.0007987379939695281,
|
|
"grad_norm": 3.0625,
|
|
"learning_rate": 1.95e-05,
|
|
"loss": 9.5075,
|
|
"mean_token_accuracy": 0.06420534774661064,
|
|
"num_tokens": 45837.0,
|
|
"step": 40
|
|
},
|
|
{
|
|
"entropy": 10.428625583648682,
|
|
"epoch": 0.0008985802432157192,
|
|
"grad_norm": 2.84375,
|
|
"learning_rate": 2.2e-05,
|
|
"loss": 9.4782,
|
|
"mean_token_accuracy": 0.07273747362196445,
|
|
"num_tokens": 51998.0,
|
|
"step": 45
|
|
},
|
|
{
|
|
"entropy": 10.35524263381958,
|
|
"epoch": 0.0009984224924619101,
|
|
"grad_norm": 2.546875,
|
|
"learning_rate": 2.4500000000000003e-05,
|
|
"loss": 9.3763,
|
|
"mean_token_accuracy": 0.07940605096518993,
|
|
"num_tokens": 58235.0,
|
|
"step": 50
|
|
},
|
|
{
|
|
"entropy": 10.194870853424073,
|
|
"epoch": 0.0010982647417081012,
|
|
"grad_norm": 2.71875,
|
|
"learning_rate": 2.7e-05,
|
|
"loss": 9.2541,
|
|
"mean_token_accuracy": 0.07720154039561748,
|
|
"num_tokens": 63553.0,
|
|
"step": 55
|
|
},
|
|
{
|
|
"entropy": 10.117318916320801,
|
|
"epoch": 0.0011981069909542921,
|
|
"grad_norm": 2.3125,
|
|
"learning_rate": 2.95e-05,
|
|
"loss": 9.3244,
|
|
"mean_token_accuracy": 0.07141049914062023,
|
|
"num_tokens": 69696.0,
|
|
"step": 60
|
|
},
|
|
{
|
|
"entropy": 10.213671588897705,
|
|
"epoch": 0.0012979492402004833,
|
|
"grad_norm": 2.234375,
|
|
"learning_rate": 3.2e-05,
|
|
"loss": 9.1874,
|
|
"mean_token_accuracy": 0.07848358415067196,
|
|
"num_tokens": 75580.0,
|
|
"step": 65
|
|
},
|
|
{
|
|
"entropy": 10.183634948730468,
|
|
"epoch": 0.0013977914894466742,
|
|
"grad_norm": 2.9375,
|
|
"learning_rate": 3.4500000000000005e-05,
|
|
"loss": 9.0286,
|
|
"mean_token_accuracy": 0.07954556941986084,
|
|
"num_tokens": 80808.0,
|
|
"step": 70
|
|
},
|
|
{
|
|
"entropy": 10.02001485824585,
|
|
"epoch": 0.0014976337386928653,
|
|
"grad_norm": 2.21875,
|
|
"learning_rate": 3.7e-05,
|
|
"loss": 9.1145,
|
|
"mean_token_accuracy": 0.07110607773065566,
|
|
"num_tokens": 86260.0,
|
|
"step": 75
|
|
},
|
|
{
|
|
"entropy": 10.129467391967774,
|
|
"epoch": 0.0015974759879390562,
|
|
"grad_norm": 2.34375,
|
|
"learning_rate": 3.95e-05,
|
|
"loss": 9.0261,
|
|
"mean_token_accuracy": 0.08673112615942954,
|
|
"num_tokens": 92099.0,
|
|
"step": 80
|
|
},
|
|
{
|
|
"entropy": 10.037476825714112,
|
|
"epoch": 0.0016973182371852473,
|
|
"grad_norm": 2.4375,
|
|
"learning_rate": 4.2000000000000004e-05,
|
|
"loss": 8.9278,
|
|
"mean_token_accuracy": 0.08497167229652405,
|
|
"num_tokens": 97720.0,
|
|
"step": 85
|
|
},
|
|
{
|
|
"entropy": 10.079257011413574,
|
|
"epoch": 0.0017971604864314384,
|
|
"grad_norm": 1.90625,
|
|
"learning_rate": 4.45e-05,
|
|
"loss": 8.9373,
|
|
"mean_token_accuracy": 0.07472074516117573,
|
|
"num_tokens": 103907.0,
|
|
"step": 90
|
|
},
|
|
{
|
|
"entropy": 10.07486629486084,
|
|
"epoch": 0.0018970027356776293,
|
|
"grad_norm": 1.8359375,
|
|
"learning_rate": 4.7000000000000004e-05,
|
|
"loss": 8.8796,
|
|
"mean_token_accuracy": 0.07506735101342202,
|
|
"num_tokens": 109981.0,
|
|
"step": 95
|
|
},
|
|
{
|
|
"entropy": 9.837776279449463,
|
|
"epoch": 0.0019968449849238202,
|
|
"grad_norm": 2.15625,
|
|
"learning_rate": 4.9500000000000004e-05,
|
|
"loss": 8.6173,
|
|
"mean_token_accuracy": 0.09049530178308487,
|
|
"num_tokens": 114997.0,
|
|
"step": 100
|
|
},
|
|
{
|
|
"entropy": 9.909902858734132,
|
|
"epoch": 0.0020966872341700114,
|
|
"grad_norm": 2.0625,
|
|
"learning_rate": 5.2e-05,
|
|
"loss": 8.6543,
|
|
"mean_token_accuracy": 0.08456408381462097,
|
|
"num_tokens": 120762.0,
|
|
"step": 105
|
|
},
|
|
{
|
|
"entropy": 9.8260329246521,
|
|
"epoch": 0.0021965294834162025,
|
|
"grad_norm": 2.328125,
|
|
"learning_rate": 5.45e-05,
|
|
"loss": 8.5515,
|
|
"mean_token_accuracy": 0.08778458833694458,
|
|
"num_tokens": 126113.0,
|
|
"step": 110
|
|
},
|
|
{
|
|
"entropy": 9.766840267181397,
|
|
"epoch": 0.0022963717326623936,
|
|
"grad_norm": 1.8046875,
|
|
"learning_rate": 5.7e-05,
|
|
"loss": 8.5581,
|
|
"mean_token_accuracy": 0.080570487678051,
|
|
"num_tokens": 132303.0,
|
|
"step": 115
|
|
},
|
|
{
|
|
"entropy": 9.73247833251953,
|
|
"epoch": 0.0023962139819085843,
|
|
"grad_norm": 1.7578125,
|
|
"learning_rate": 5.9499999999999996e-05,
|
|
"loss": 8.4172,
|
|
"mean_token_accuracy": 0.08937932215631009,
|
|
"num_tokens": 138642.0,
|
|
"step": 120
|
|
},
|
|
{
|
|
"entropy": 9.6115159034729,
|
|
"epoch": 0.0024960562311547754,
|
|
"grad_norm": 1.5234375,
|
|
"learning_rate": 6.2e-05,
|
|
"loss": 8.3259,
|
|
"mean_token_accuracy": 0.08444585129618645,
|
|
"num_tokens": 144420.0,
|
|
"step": 125
|
|
},
|
|
{
|
|
"entropy": 9.505456352233887,
|
|
"epoch": 0.0025958984804009665,
|
|
"grad_norm": 1.6953125,
|
|
"learning_rate": 6.450000000000001e-05,
|
|
"loss": 8.2552,
|
|
"mean_token_accuracy": 0.08594180978834629,
|
|
"num_tokens": 150126.0,
|
|
"step": 130
|
|
},
|
|
{
|
|
"entropy": 9.33946189880371,
|
|
"epoch": 0.0026957407296471577,
|
|
"grad_norm": 1.65625,
|
|
"learning_rate": 6.7e-05,
|
|
"loss": 8.1841,
|
|
"mean_token_accuracy": 0.08540923669934272,
|
|
"num_tokens": 156394.0,
|
|
"step": 135
|
|
},
|
|
{
|
|
"entropy": 9.154175662994385,
|
|
"epoch": 0.0027955829788933483,
|
|
"grad_norm": 1.7734375,
|
|
"learning_rate": 6.950000000000001e-05,
|
|
"loss": 8.0888,
|
|
"mean_token_accuracy": 0.08764742240309716,
|
|
"num_tokens": 161888.0,
|
|
"step": 140
|
|
},
|
|
{
|
|
"entropy": 9.103712558746338,
|
|
"epoch": 0.0028954252281395395,
|
|
"grad_norm": 1.703125,
|
|
"learning_rate": 7.2e-05,
|
|
"loss": 8.0941,
|
|
"mean_token_accuracy": 0.08791576996445656,
|
|
"num_tokens": 167479.0,
|
|
"step": 145
|
|
},
|
|
{
|
|
"entropy": 8.953013134002685,
|
|
"epoch": 0.0029952674773857306,
|
|
"grad_norm": 1.4140625,
|
|
"learning_rate": 7.45e-05,
|
|
"loss": 7.9641,
|
|
"mean_token_accuracy": 0.08592916205525399,
|
|
"num_tokens": 173013.0,
|
|
"step": 150
|
|
},
|
|
{
|
|
"entropy": 8.789963626861573,
|
|
"epoch": 0.0030951097266319217,
|
|
"grad_norm": 1.5625,
|
|
"learning_rate": 7.7e-05,
|
|
"loss": 7.9985,
|
|
"mean_token_accuracy": 0.08673654571175575,
|
|
"num_tokens": 178908.0,
|
|
"step": 155
|
|
},
|
|
{
|
|
"entropy": 8.74723539352417,
|
|
"epoch": 0.0031949519758781124,
|
|
"grad_norm": 1.2265625,
|
|
"learning_rate": 7.950000000000001e-05,
|
|
"loss": 7.8767,
|
|
"mean_token_accuracy": 0.09191061705350875,
|
|
"num_tokens": 184314.0,
|
|
"step": 160
|
|
},
|
|
{
|
|
"entropy": 8.576598930358887,
|
|
"epoch": 0.0032947942251243035,
|
|
"grad_norm": 1.46875,
|
|
"learning_rate": 8.2e-05,
|
|
"loss": 7.8791,
|
|
"mean_token_accuracy": 0.08847894445061684,
|
|
"num_tokens": 190217.0,
|
|
"step": 165
|
|
},
|
|
{
|
|
"entropy": 8.493614959716798,
|
|
"epoch": 0.0033946364743704946,
|
|
"grad_norm": 1.421875,
|
|
"learning_rate": 8.450000000000001e-05,
|
|
"loss": 7.8022,
|
|
"mean_token_accuracy": 0.09182382375001907,
|
|
"num_tokens": 196154.0,
|
|
"step": 170
|
|
},
|
|
{
|
|
"entropy": 8.42578411102295,
|
|
"epoch": 0.0034944787236166857,
|
|
"grad_norm": 1.609375,
|
|
"learning_rate": 8.7e-05,
|
|
"loss": 7.8988,
|
|
"mean_token_accuracy": 0.08841787204146385,
|
|
"num_tokens": 202043.0,
|
|
"step": 175
|
|
},
|
|
{
|
|
"entropy": 8.447116088867187,
|
|
"epoch": 0.003594320972862877,
|
|
"grad_norm": 1.40625,
|
|
"learning_rate": 8.95e-05,
|
|
"loss": 7.8551,
|
|
"mean_token_accuracy": 0.08922517634928226,
|
|
"num_tokens": 208843.0,
|
|
"step": 180
|
|
},
|
|
{
|
|
"entropy": 8.342639446258545,
|
|
"epoch": 0.0036941632221090676,
|
|
"grad_norm": 1.421875,
|
|
"learning_rate": 9.2e-05,
|
|
"loss": 7.8137,
|
|
"mean_token_accuracy": 0.09233534485101699,
|
|
"num_tokens": 213777.0,
|
|
"step": 185
|
|
},
|
|
{
|
|
"entropy": 8.31864881515503,
|
|
"epoch": 0.0037940054713552587,
|
|
"grad_norm": 1.3046875,
|
|
"learning_rate": 9.45e-05,
|
|
"loss": 7.7358,
|
|
"mean_token_accuracy": 0.0952209122478962,
|
|
"num_tokens": 220209.0,
|
|
"step": 190
|
|
},
|
|
{
|
|
"entropy": 8.204437160491944,
|
|
"epoch": 0.00389384772060145,
|
|
"grad_norm": 1.3828125,
|
|
"learning_rate": 9.7e-05,
|
|
"loss": 7.747,
|
|
"mean_token_accuracy": 0.0910322405397892,
|
|
"num_tokens": 225900.0,
|
|
"step": 195
|
|
},
|
|
{
|
|
"entropy": 8.255455112457275,
|
|
"epoch": 0.0039936899698476405,
|
|
"grad_norm": 1.4765625,
|
|
"learning_rate": 9.95e-05,
|
|
"loss": 7.8,
|
|
"mean_token_accuracy": 0.08869038000702859,
|
|
"num_tokens": 231781.0,
|
|
"step": 200
|
|
},
|
|
{
|
|
"entropy": 8.125996208190918,
|
|
"epoch": 0.004093532219093832,
|
|
"grad_norm": 1.3046875,
|
|
"learning_rate": 0.000102,
|
|
"loss": 7.7429,
|
|
"mean_token_accuracy": 0.08568662032485008,
|
|
"num_tokens": 236974.0,
|
|
"step": 205
|
|
},
|
|
{
|
|
"entropy": 8.232761573791503,
|
|
"epoch": 0.004193374468340023,
|
|
"grad_norm": 1.28125,
|
|
"learning_rate": 0.00010449999999999999,
|
|
"loss": 7.7475,
|
|
"mean_token_accuracy": 0.08775760419666767,
|
|
"num_tokens": 242588.0,
|
|
"step": 210
|
|
},
|
|
{
|
|
"entropy": 8.146185207366944,
|
|
"epoch": 0.004293216717586213,
|
|
"grad_norm": 1.3984375,
|
|
"learning_rate": 0.000107,
|
|
"loss": 7.6947,
|
|
"mean_token_accuracy": 0.0925628162920475,
|
|
"num_tokens": 248736.0,
|
|
"step": 215
|
|
},
|
|
{
|
|
"entropy": 8.136151313781738,
|
|
"epoch": 0.004393058966832405,
|
|
"grad_norm": 1.3203125,
|
|
"learning_rate": 0.0001095,
|
|
"loss": 7.7161,
|
|
"mean_token_accuracy": 0.08190489560365677,
|
|
"num_tokens": 254348.0,
|
|
"step": 220
|
|
},
|
|
{
|
|
"entropy": 8.15009708404541,
|
|
"epoch": 0.004492901216078596,
|
|
"grad_norm": 1.6171875,
|
|
"learning_rate": 0.000112,
|
|
"loss": 7.6301,
|
|
"mean_token_accuracy": 0.09184228777885436,
|
|
"num_tokens": 259415.0,
|
|
"step": 225
|
|
},
|
|
{
|
|
"entropy": 7.953672695159912,
|
|
"epoch": 0.004592743465324787,
|
|
"grad_norm": 1.125,
|
|
"learning_rate": 0.0001145,
|
|
"loss": 7.8223,
|
|
"mean_token_accuracy": 0.0824849434196949,
|
|
"num_tokens": 265073.0,
|
|
"step": 230
|
|
},
|
|
{
|
|
"entropy": 8.194051551818848,
|
|
"epoch": 0.004692585714570978,
|
|
"grad_norm": 1.328125,
|
|
"learning_rate": 0.00011700000000000001,
|
|
"loss": 7.7181,
|
|
"mean_token_accuracy": 0.08306673839688301,
|
|
"num_tokens": 270672.0,
|
|
"step": 235
|
|
},
|
|
{
|
|
"entropy": 7.986516904830933,
|
|
"epoch": 0.004792427963817169,
|
|
"grad_norm": 1.40625,
|
|
"learning_rate": 0.00011949999999999999,
|
|
"loss": 7.5703,
|
|
"mean_token_accuracy": 0.09606460258364677,
|
|
"num_tokens": 275971.0,
|
|
"step": 240
|
|
},
|
|
{
|
|
"entropy": 8.076891374588012,
|
|
"epoch": 0.00489227021306336,
|
|
"grad_norm": 1.453125,
|
|
"learning_rate": 0.000122,
|
|
"loss": 7.6701,
|
|
"mean_token_accuracy": 0.09255184978246689,
|
|
"num_tokens": 281554.0,
|
|
"step": 245
|
|
},
|
|
{
|
|
"entropy": 8.021392154693604,
|
|
"epoch": 0.004992112462309551,
|
|
"grad_norm": 1.4296875,
|
|
"learning_rate": 0.0001245,
|
|
"loss": 7.6684,
|
|
"mean_token_accuracy": 0.09083683714270592,
|
|
"num_tokens": 287733.0,
|
|
"step": 250
|
|
},
|
|
{
|
|
"entropy": 8.066862726211548,
|
|
"epoch": 0.0050919547115557415,
|
|
"grad_norm": 1.5234375,
|
|
"learning_rate": 0.000127,
|
|
"loss": 7.6995,
|
|
"mean_token_accuracy": 0.08480059802532196,
|
|
"num_tokens": 293958.0,
|
|
"step": 255
|
|
},
|
|
{
|
|
"entropy": 8.096820306777953,
|
|
"epoch": 0.005191796960801933,
|
|
"grad_norm": 1.5234375,
|
|
"learning_rate": 0.0001295,
|
|
"loss": 7.6813,
|
|
"mean_token_accuracy": 0.08337552659213543,
|
|
"num_tokens": 299615.0,
|
|
"step": 260
|
|
},
|
|
{
|
|
"entropy": 8.020620632171632,
|
|
"epoch": 0.005291639210048124,
|
|
"grad_norm": 1.5546875,
|
|
"learning_rate": 0.000132,
|
|
"loss": 7.6583,
|
|
"mean_token_accuracy": 0.09328315630555153,
|
|
"num_tokens": 305401.0,
|
|
"step": 265
|
|
},
|
|
{
|
|
"entropy": 8.00606279373169,
|
|
"epoch": 0.005391481459294315,
|
|
"grad_norm": 1.375,
|
|
"learning_rate": 0.00013450000000000002,
|
|
"loss": 7.5975,
|
|
"mean_token_accuracy": 0.09268495440483093,
|
|
"num_tokens": 310420.0,
|
|
"step": 270
|
|
},
|
|
{
|
|
"entropy": 7.975163745880127,
|
|
"epoch": 0.005491323708540506,
|
|
"grad_norm": 1.453125,
|
|
"learning_rate": 0.00013700000000000002,
|
|
"loss": 7.6291,
|
|
"mean_token_accuracy": 0.08745413720607757,
|
|
"num_tokens": 315937.0,
|
|
"step": 275
|
|
},
|
|
{
|
|
"entropy": 7.931376361846924,
|
|
"epoch": 0.005591165957786697,
|
|
"grad_norm": 1.34375,
|
|
"learning_rate": 0.0001395,
|
|
"loss": 7.6787,
|
|
"mean_token_accuracy": 0.09163259714841843,
|
|
"num_tokens": 321055.0,
|
|
"step": 280
|
|
},
|
|
{
|
|
"entropy": 8.027003908157349,
|
|
"epoch": 0.005691008207032888,
|
|
"grad_norm": 1.65625,
|
|
"learning_rate": 0.00014199999999999998,
|
|
"loss": 7.5598,
|
|
"mean_token_accuracy": 0.09943108633160591,
|
|
"num_tokens": 327446.0,
|
|
"step": 285
|
|
},
|
|
{
|
|
"entropy": 7.895185947418213,
|
|
"epoch": 0.005790850456279079,
|
|
"grad_norm": 1.8359375,
|
|
"learning_rate": 0.0001445,
|
|
"loss": 7.6409,
|
|
"mean_token_accuracy": 0.09412241280078888,
|
|
"num_tokens": 333363.0,
|
|
"step": 290
|
|
},
|
|
{
|
|
"entropy": 8.069699239730834,
|
|
"epoch": 0.0058906927055252705,
|
|
"grad_norm": 1.53125,
|
|
"learning_rate": 0.000147,
|
|
"loss": 7.6736,
|
|
"mean_token_accuracy": 0.09531420394778252,
|
|
"num_tokens": 339550.0,
|
|
"step": 295
|
|
},
|
|
{
|
|
"entropy": 8.054055500030518,
|
|
"epoch": 0.005990534954771461,
|
|
"grad_norm": 1.28125,
|
|
"learning_rate": 0.0001495,
|
|
"loss": 7.7013,
|
|
"mean_token_accuracy": 0.08208880387246609,
|
|
"num_tokens": 345472.0,
|
|
"step": 300
|
|
},
|
|
{
|
|
"entropy": 7.895511817932129,
|
|
"epoch": 0.006090377204017652,
|
|
"grad_norm": 1.1796875,
|
|
"learning_rate": 0.000152,
|
|
"loss": 7.6018,
|
|
"mean_token_accuracy": 0.08789039179682731,
|
|
"num_tokens": 350984.0,
|
|
"step": 305
|
|
},
|
|
{
|
|
"entropy": 7.9680193901062015,
|
|
"epoch": 0.006190219453263843,
|
|
"grad_norm": 1.4765625,
|
|
"learning_rate": 0.00015450000000000001,
|
|
"loss": 7.5038,
|
|
"mean_token_accuracy": 0.09207511395215988,
|
|
"num_tokens": 356054.0,
|
|
"step": 310
|
|
},
|
|
{
|
|
"entropy": 7.946393775939941,
|
|
"epoch": 0.006290061702510034,
|
|
"grad_norm": 1.5,
|
|
"learning_rate": 0.000157,
|
|
"loss": 7.6509,
|
|
"mean_token_accuracy": 0.1025794766843319,
|
|
"num_tokens": 361985.0,
|
|
"step": 315
|
|
},
|
|
{
|
|
"entropy": 7.917030382156372,
|
|
"epoch": 0.006389903951756225,
|
|
"grad_norm": 1.28125,
|
|
"learning_rate": 0.0001595,
|
|
"loss": 7.6035,
|
|
"mean_token_accuracy": 0.0876886785030365,
|
|
"num_tokens": 367784.0,
|
|
"step": 320
|
|
},
|
|
{
|
|
"entropy": 7.934617376327514,
|
|
"epoch": 0.006489746201002416,
|
|
"grad_norm": 1.15625,
|
|
"learning_rate": 0.000162,
|
|
"loss": 7.5577,
|
|
"mean_token_accuracy": 0.09602945894002915,
|
|
"num_tokens": 374012.0,
|
|
"step": 325
|
|
},
|
|
{
|
|
"entropy": 7.939635705947876,
|
|
"epoch": 0.006589588450248607,
|
|
"grad_norm": 1.4140625,
|
|
"learning_rate": 0.00016450000000000001,
|
|
"loss": 7.6332,
|
|
"mean_token_accuracy": 0.08928873538970947,
|
|
"num_tokens": 379623.0,
|
|
"step": 330
|
|
},
|
|
{
|
|
"entropy": 7.943338441848755,
|
|
"epoch": 0.006689430699494799,
|
|
"grad_norm": 1.28125,
|
|
"learning_rate": 0.00016700000000000002,
|
|
"loss": 7.5892,
|
|
"mean_token_accuracy": 0.09149550125002862,
|
|
"num_tokens": 384982.0,
|
|
"step": 335
|
|
},
|
|
{
|
|
"entropy": 7.994742679595947,
|
|
"epoch": 0.006789272948740989,
|
|
"grad_norm": 1.28125,
|
|
"learning_rate": 0.00016950000000000003,
|
|
"loss": 7.6425,
|
|
"mean_token_accuracy": 0.08843275494873523,
|
|
"num_tokens": 391107.0,
|
|
"step": 340
|
|
},
|
|
{
|
|
"entropy": 7.823350954055786,
|
|
"epoch": 0.00688911519798718,
|
|
"grad_norm": 2.125,
|
|
"learning_rate": 0.00017199999999999998,
|
|
"loss": 7.509,
|
|
"mean_token_accuracy": 0.09060701727867126,
|
|
"num_tokens": 396516.0,
|
|
"step": 345
|
|
},
|
|
{
|
|
"entropy": 7.931709098815918,
|
|
"epoch": 0.0069889574472333715,
|
|
"grad_norm": 1.109375,
|
|
"learning_rate": 0.00017449999999999999,
|
|
"loss": 7.6268,
|
|
"mean_token_accuracy": 0.08907039240002632,
|
|
"num_tokens": 403216.0,
|
|
"step": 350
|
|
},
|
|
{
|
|
"entropy": 7.828972816467285,
|
|
"epoch": 0.007088799696479562,
|
|
"grad_norm": 1.28125,
|
|
"learning_rate": 0.000177,
|
|
"loss": 7.5647,
|
|
"mean_token_accuracy": 0.09698711633682251,
|
|
"num_tokens": 409757.0,
|
|
"step": 355
|
|
},
|
|
{
|
|
"entropy": 7.901078701019287,
|
|
"epoch": 0.007188641945725754,
|
|
"grad_norm": 1.734375,
|
|
"learning_rate": 0.0001795,
|
|
"loss": 7.5776,
|
|
"mean_token_accuracy": 0.09460580721497536,
|
|
"num_tokens": 415071.0,
|
|
"step": 360
|
|
},
|
|
{
|
|
"entropy": 7.830478239059448,
|
|
"epoch": 0.007288484194971944,
|
|
"grad_norm": 1.3203125,
|
|
"learning_rate": 0.000182,
|
|
"loss": 7.5183,
|
|
"mean_token_accuracy": 0.09679247066378593,
|
|
"num_tokens": 420660.0,
|
|
"step": 365
|
|
},
|
|
{
|
|
"entropy": 7.885542583465576,
|
|
"epoch": 0.007388326444218135,
|
|
"grad_norm": 1.2734375,
|
|
"learning_rate": 0.0001845,
|
|
"loss": 7.5054,
|
|
"mean_token_accuracy": 0.09666672423481941,
|
|
"num_tokens": 427219.0,
|
|
"step": 370
|
|
},
|
|
{
|
|
"entropy": 7.882419490814209,
|
|
"epoch": 0.007488168693464327,
|
|
"grad_norm": 1.640625,
|
|
"learning_rate": 0.000187,
|
|
"loss": 7.4902,
|
|
"mean_token_accuracy": 0.0948188878595829,
|
|
"num_tokens": 432938.0,
|
|
"step": 375
|
|
},
|
|
{
|
|
"entropy": 7.615965557098389,
|
|
"epoch": 0.007588010942710517,
|
|
"grad_norm": 1.71875,
|
|
"learning_rate": 0.0001895,
|
|
"loss": 7.4774,
|
|
"mean_token_accuracy": 0.10164310932159423,
|
|
"num_tokens": 438223.0,
|
|
"step": 380
|
|
},
|
|
{
|
|
"entropy": 7.930799531936645,
|
|
"epoch": 0.007687853191956708,
|
|
"grad_norm": 1.15625,
|
|
"learning_rate": 0.000192,
|
|
"loss": 7.5398,
|
|
"mean_token_accuracy": 0.09940317496657372,
|
|
"num_tokens": 444432.0,
|
|
"step": 385
|
|
},
|
|
{
|
|
"entropy": 7.903547239303589,
|
|
"epoch": 0.0077876954412029,
|
|
"grad_norm": 1.40625,
|
|
"learning_rate": 0.0001945,
|
|
"loss": 7.5144,
|
|
"mean_token_accuracy": 0.08792076036334037,
|
|
"num_tokens": 450497.0,
|
|
"step": 390
|
|
},
|
|
{
|
|
"entropy": 7.71635160446167,
|
|
"epoch": 0.00788753769044909,
|
|
"grad_norm": 1.2109375,
|
|
"learning_rate": 0.00019700000000000002,
|
|
"loss": 7.5483,
|
|
"mean_token_accuracy": 0.09417223706841468,
|
|
"num_tokens": 456358.0,
|
|
"step": 395
|
|
},
|
|
{
|
|
"entropy": 7.832306432723999,
|
|
"epoch": 0.007987379939695281,
|
|
"grad_norm": 1.3984375,
|
|
"learning_rate": 0.00019950000000000002,
|
|
"loss": 7.4283,
|
|
"mean_token_accuracy": 0.09271656051278114,
|
|
"num_tokens": 461638.0,
|
|
"step": 400
|
|
},
|
|
{
|
|
"entropy": 7.71069369316101,
|
|
"epoch": 0.008087222188941472,
|
|
"grad_norm": 1.609375,
|
|
"learning_rate": 0.000202,
|
|
"loss": 7.2931,
|
|
"mean_token_accuracy": 0.10623703300952911,
|
|
"num_tokens": 466383.0,
|
|
"step": 405
|
|
},
|
|
{
|
|
"entropy": 7.75597825050354,
|
|
"epoch": 0.008187064438187664,
|
|
"grad_norm": 1.4140625,
|
|
"learning_rate": 0.00020449999999999998,
|
|
"loss": 7.5048,
|
|
"mean_token_accuracy": 0.09718195348978043,
|
|
"num_tokens": 472325.0,
|
|
"step": 410
|
|
},
|
|
{
|
|
"entropy": 7.676103734970093,
|
|
"epoch": 0.008286906687433855,
|
|
"grad_norm": 1.3359375,
|
|
"learning_rate": 0.000207,
|
|
"loss": 7.3461,
|
|
"mean_token_accuracy": 0.09879515469074249,
|
|
"num_tokens": 477810.0,
|
|
"step": 415
|
|
},
|
|
{
|
|
"entropy": 7.821485471725464,
|
|
"epoch": 0.008386748936680045,
|
|
"grad_norm": 1.3359375,
|
|
"learning_rate": 0.0002095,
|
|
"loss": 7.4209,
|
|
"mean_token_accuracy": 0.09962461516261101,
|
|
"num_tokens": 483375.0,
|
|
"step": 420
|
|
},
|
|
{
|
|
"entropy": 7.718852472305298,
|
|
"epoch": 0.008486591185926236,
|
|
"grad_norm": 1.09375,
|
|
"learning_rate": 0.000212,
|
|
"loss": 7.4184,
|
|
"mean_token_accuracy": 0.09406220316886901,
|
|
"num_tokens": 489826.0,
|
|
"step": 425
|
|
},
|
|
{
|
|
"entropy": 7.717019510269165,
|
|
"epoch": 0.008586433435172427,
|
|
"grad_norm": 1.4453125,
|
|
"learning_rate": 0.0002145,
|
|
"loss": 7.3222,
|
|
"mean_token_accuracy": 0.10805956795811653,
|
|
"num_tokens": 494797.0,
|
|
"step": 430
|
|
},
|
|
{
|
|
"entropy": 7.657623624801635,
|
|
"epoch": 0.00868627568441862,
|
|
"grad_norm": 1.546875,
|
|
"learning_rate": 0.00021700000000000002,
|
|
"loss": 7.3715,
|
|
"mean_token_accuracy": 0.11027861759066582,
|
|
"num_tokens": 499868.0,
|
|
"step": 435
|
|
},
|
|
{
|
|
"entropy": 7.766957855224609,
|
|
"epoch": 0.00878611793366481,
|
|
"grad_norm": 1.125,
|
|
"learning_rate": 0.0002195,
|
|
"loss": 7.489,
|
|
"mean_token_accuracy": 0.09478027820587158,
|
|
"num_tokens": 505777.0,
|
|
"step": 440
|
|
},
|
|
{
|
|
"entropy": 7.825216388702392,
|
|
"epoch": 0.008885960182911,
|
|
"grad_norm": 1.234375,
|
|
"learning_rate": 0.000222,
|
|
"loss": 7.4513,
|
|
"mean_token_accuracy": 0.09458905309438706,
|
|
"num_tokens": 511404.0,
|
|
"step": 445
|
|
},
|
|
{
|
|
"entropy": 7.681342935562133,
|
|
"epoch": 0.008985802432157191,
|
|
"grad_norm": 1.2890625,
|
|
"learning_rate": 0.0002245,
|
|
"loss": 7.3926,
|
|
"mean_token_accuracy": 0.10864230245351791,
|
|
"num_tokens": 517255.0,
|
|
"step": 450
|
|
},
|
|
{
|
|
"entropy": 7.636969661712646,
|
|
"epoch": 0.009085644681403382,
|
|
"grad_norm": 1.375,
|
|
"learning_rate": 0.00022700000000000002,
|
|
"loss": 7.418,
|
|
"mean_token_accuracy": 0.09724331274628639,
|
|
"num_tokens": 523299.0,
|
|
"step": 455
|
|
},
|
|
{
|
|
"entropy": 7.716694116592407,
|
|
"epoch": 0.009185486930649574,
|
|
"grad_norm": 1.2734375,
|
|
"learning_rate": 0.00022950000000000002,
|
|
"loss": 7.4159,
|
|
"mean_token_accuracy": 0.09504513517022133,
|
|
"num_tokens": 528700.0,
|
|
"step": 460
|
|
},
|
|
{
|
|
"entropy": 7.704607963562012,
|
|
"epoch": 0.009285329179895765,
|
|
"grad_norm": 1.3203125,
|
|
"learning_rate": 0.00023200000000000003,
|
|
"loss": 7.3906,
|
|
"mean_token_accuracy": 0.10305075645446778,
|
|
"num_tokens": 534456.0,
|
|
"step": 465
|
|
},
|
|
{
|
|
"entropy": 7.687975072860718,
|
|
"epoch": 0.009385171429141956,
|
|
"grad_norm": 1.3828125,
|
|
"learning_rate": 0.00023449999999999998,
|
|
"loss": 7.392,
|
|
"mean_token_accuracy": 0.09684277400374412,
|
|
"num_tokens": 541492.0,
|
|
"step": 470
|
|
},
|
|
{
|
|
"entropy": 7.614287185668945,
|
|
"epoch": 0.009485013678388146,
|
|
"grad_norm": 1.375,
|
|
"learning_rate": 0.000237,
|
|
"loss": 7.4688,
|
|
"mean_token_accuracy": 0.09323694482445717,
|
|
"num_tokens": 547126.0,
|
|
"step": 475
|
|
},
|
|
{
|
|
"entropy": 7.748909521102905,
|
|
"epoch": 0.009584855927634337,
|
|
"grad_norm": 1.40625,
|
|
"learning_rate": 0.0002395,
|
|
"loss": 7.4187,
|
|
"mean_token_accuracy": 0.09110049903392792,
|
|
"num_tokens": 553522.0,
|
|
"step": 480
|
|
},
|
|
{
|
|
"entropy": 7.599171686172485,
|
|
"epoch": 0.00968469817688053,
|
|
"grad_norm": 1.6015625,
|
|
"learning_rate": 0.000242,
|
|
"loss": 7.3032,
|
|
"mean_token_accuracy": 0.09761992767453194,
|
|
"num_tokens": 559487.0,
|
|
"step": 485
|
|
},
|
|
{
|
|
"entropy": 7.581656837463379,
|
|
"epoch": 0.00978454042612672,
|
|
"grad_norm": 1.3828125,
|
|
"learning_rate": 0.0002445,
|
|
"loss": 7.3578,
|
|
"mean_token_accuracy": 0.09209932833909988,
|
|
"num_tokens": 565483.0,
|
|
"step": 490
|
|
},
|
|
{
|
|
"entropy": 7.675942230224609,
|
|
"epoch": 0.009884382675372911,
|
|
"grad_norm": 1.578125,
|
|
"learning_rate": 0.000247,
|
|
"loss": 7.3958,
|
|
"mean_token_accuracy": 0.1011126957833767,
|
|
"num_tokens": 571609.0,
|
|
"step": 495
|
|
},
|
|
{
|
|
"entropy": 7.702158212661743,
|
|
"epoch": 0.009984224924619102,
|
|
"grad_norm": 1.25,
|
|
"learning_rate": 0.0002495,
|
|
"loss": 7.3743,
|
|
"mean_token_accuracy": 0.09817478843033314,
|
|
"num_tokens": 578310.0,
|
|
"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": 1021586522112000.0,
|
|
"train_batch_size": 16,
|
|
"trial_name": null,
|
|
"trial_params": null
|
|
}
|