1035 lines
27 KiB
JSON
1035 lines
27 KiB
JSON
{
|
|
"best_global_step": null,
|
|
"best_metric": null,
|
|
"best_model_checkpoint": null,
|
|
"epoch": 0.027977506085107573,
|
|
"eval_steps": 500,
|
|
"global_step": 500,
|
|
"is_hyper_param_search": false,
|
|
"is_local_process_zero": true,
|
|
"is_world_process_zero": true,
|
|
"log_history": [
|
|
{
|
|
"entropy": 10.691998958587646,
|
|
"epoch": 0.0002797750608510757,
|
|
"grad_norm": 15.75,
|
|
"learning_rate": 2e-06,
|
|
"loss": 10.7863,
|
|
"mean_token_accuracy": 0.0001396648003719747,
|
|
"num_tokens": 8714.0,
|
|
"step": 5
|
|
},
|
|
{
|
|
"entropy": 10.691917991638183,
|
|
"epoch": 0.0005595501217021514,
|
|
"grad_norm": 12.6875,
|
|
"learning_rate": 4.5e-06,
|
|
"loss": 10.7427,
|
|
"mean_token_accuracy": 6.393861840479076e-05,
|
|
"num_tokens": 18630.0,
|
|
"step": 10
|
|
},
|
|
{
|
|
"entropy": 10.690977668762207,
|
|
"epoch": 0.0008393251825532272,
|
|
"grad_norm": 9.8125,
|
|
"learning_rate": 7e-06,
|
|
"loss": 10.4865,
|
|
"mean_token_accuracy": 0.026551677379757166,
|
|
"num_tokens": 28134.0,
|
|
"step": 15
|
|
},
|
|
{
|
|
"entropy": 10.678603363037109,
|
|
"epoch": 0.0011191002434043028,
|
|
"grad_norm": 6.4375,
|
|
"learning_rate": 9.5e-06,
|
|
"loss": 10.188,
|
|
"mean_token_accuracy": 0.04146423395723105,
|
|
"num_tokens": 37326.0,
|
|
"step": 20
|
|
},
|
|
{
|
|
"entropy": 10.619599628448487,
|
|
"epoch": 0.0013988753042553786,
|
|
"grad_norm": 4.15625,
|
|
"learning_rate": 1.2e-05,
|
|
"loss": 9.8709,
|
|
"mean_token_accuracy": 0.04364684820175171,
|
|
"num_tokens": 45813.0,
|
|
"step": 25
|
|
},
|
|
{
|
|
"entropy": 10.597146987915039,
|
|
"epoch": 0.0016786503651064545,
|
|
"grad_norm": 3.265625,
|
|
"learning_rate": 1.4500000000000002e-05,
|
|
"loss": 9.7719,
|
|
"mean_token_accuracy": 0.03872578963637352,
|
|
"num_tokens": 54746.0,
|
|
"step": 30
|
|
},
|
|
{
|
|
"entropy": 10.614226627349854,
|
|
"epoch": 0.0019584254259575303,
|
|
"grad_norm": 2.734375,
|
|
"learning_rate": 1.7000000000000003e-05,
|
|
"loss": 9.6616,
|
|
"mean_token_accuracy": 0.04101139903068542,
|
|
"num_tokens": 64386.0,
|
|
"step": 35
|
|
},
|
|
{
|
|
"entropy": 10.594823837280273,
|
|
"epoch": 0.0022382004868086057,
|
|
"grad_norm": 2.71875,
|
|
"learning_rate": 1.95e-05,
|
|
"loss": 9.6762,
|
|
"mean_token_accuracy": 0.03911035992205143,
|
|
"num_tokens": 73648.0,
|
|
"step": 40
|
|
},
|
|
{
|
|
"entropy": 10.59369068145752,
|
|
"epoch": 0.0025179755476596815,
|
|
"grad_norm": 2.546875,
|
|
"learning_rate": 2.2e-05,
|
|
"loss": 9.5708,
|
|
"mean_token_accuracy": 0.04052439667284489,
|
|
"num_tokens": 82108.0,
|
|
"step": 45
|
|
},
|
|
{
|
|
"entropy": 10.583232975006103,
|
|
"epoch": 0.0027977506085107573,
|
|
"grad_norm": 2.546875,
|
|
"learning_rate": 2.4500000000000003e-05,
|
|
"loss": 9.5017,
|
|
"mean_token_accuracy": 0.047141891345381735,
|
|
"num_tokens": 92109.0,
|
|
"step": 50
|
|
},
|
|
{
|
|
"entropy": 10.565110969543458,
|
|
"epoch": 0.003077525669361833,
|
|
"grad_norm": 2.703125,
|
|
"learning_rate": 2.7e-05,
|
|
"loss": 9.4323,
|
|
"mean_token_accuracy": 0.052879580110311505,
|
|
"num_tokens": 101365.0,
|
|
"step": 55
|
|
},
|
|
{
|
|
"entropy": 10.425063991546631,
|
|
"epoch": 0.003357300730212909,
|
|
"grad_norm": 2.546875,
|
|
"learning_rate": 2.95e-05,
|
|
"loss": 9.3995,
|
|
"mean_token_accuracy": 0.05232988931238651,
|
|
"num_tokens": 110432.0,
|
|
"step": 60
|
|
},
|
|
{
|
|
"entropy": 10.407742786407471,
|
|
"epoch": 0.0036370757910639847,
|
|
"grad_norm": 2.34375,
|
|
"learning_rate": 3.2e-05,
|
|
"loss": 9.3122,
|
|
"mean_token_accuracy": 0.051964789256453515,
|
|
"num_tokens": 120191.0,
|
|
"step": 65
|
|
},
|
|
{
|
|
"entropy": 10.429958152770997,
|
|
"epoch": 0.0039168508519150606,
|
|
"grad_norm": 2.3125,
|
|
"learning_rate": 3.4500000000000005e-05,
|
|
"loss": 9.1579,
|
|
"mean_token_accuracy": 0.06370341181755065,
|
|
"num_tokens": 129580.0,
|
|
"step": 70
|
|
},
|
|
{
|
|
"entropy": 10.423612117767334,
|
|
"epoch": 0.004196625912766136,
|
|
"grad_norm": 2.375,
|
|
"learning_rate": 3.7e-05,
|
|
"loss": 9.1589,
|
|
"mean_token_accuracy": 0.05433790199458599,
|
|
"num_tokens": 139083.0,
|
|
"step": 75
|
|
},
|
|
{
|
|
"entropy": 10.42996587753296,
|
|
"epoch": 0.004476400973617211,
|
|
"grad_norm": 2.359375,
|
|
"learning_rate": 3.95e-05,
|
|
"loss": 8.9628,
|
|
"mean_token_accuracy": 0.05919392295181751,
|
|
"num_tokens": 148183.0,
|
|
"step": 80
|
|
},
|
|
{
|
|
"entropy": 10.377639865875244,
|
|
"epoch": 0.004756176034468288,
|
|
"grad_norm": 2.359375,
|
|
"learning_rate": 4.2000000000000004e-05,
|
|
"loss": 8.8812,
|
|
"mean_token_accuracy": 0.060510285571217536,
|
|
"num_tokens": 157015.0,
|
|
"step": 85
|
|
},
|
|
{
|
|
"entropy": 10.363342475891113,
|
|
"epoch": 0.005035951095319363,
|
|
"grad_norm": 2.921875,
|
|
"learning_rate": 4.45e-05,
|
|
"loss": 8.8142,
|
|
"mean_token_accuracy": 0.06536102406680584,
|
|
"num_tokens": 166314.0,
|
|
"step": 90
|
|
},
|
|
{
|
|
"entropy": 10.275917053222656,
|
|
"epoch": 0.005315726156170439,
|
|
"grad_norm": 2.203125,
|
|
"learning_rate": 4.7000000000000004e-05,
|
|
"loss": 8.6269,
|
|
"mean_token_accuracy": 0.07039711400866508,
|
|
"num_tokens": 175975.0,
|
|
"step": 95
|
|
},
|
|
{
|
|
"entropy": 10.235202980041503,
|
|
"epoch": 0.005595501217021515,
|
|
"grad_norm": 2.734375,
|
|
"learning_rate": 4.9500000000000004e-05,
|
|
"loss": 8.574,
|
|
"mean_token_accuracy": 0.06645569279789924,
|
|
"num_tokens": 185976.0,
|
|
"step": 100
|
|
},
|
|
{
|
|
"entropy": 10.220190620422363,
|
|
"epoch": 0.005875276277872591,
|
|
"grad_norm": 2.171875,
|
|
"learning_rate": 5.2e-05,
|
|
"loss": 8.4802,
|
|
"mean_token_accuracy": 0.06686681807041168,
|
|
"num_tokens": 195112.0,
|
|
"step": 105
|
|
},
|
|
{
|
|
"entropy": 10.113536167144776,
|
|
"epoch": 0.006155051338723666,
|
|
"grad_norm": 1.984375,
|
|
"learning_rate": 5.45e-05,
|
|
"loss": 8.4636,
|
|
"mean_token_accuracy": 0.05900593213737011,
|
|
"num_tokens": 205951.0,
|
|
"step": 110
|
|
},
|
|
{
|
|
"entropy": 10.075279331207275,
|
|
"epoch": 0.006434826399574742,
|
|
"grad_norm": 2.25,
|
|
"learning_rate": 5.7e-05,
|
|
"loss": 8.294,
|
|
"mean_token_accuracy": 0.07148804180324078,
|
|
"num_tokens": 214744.0,
|
|
"step": 115
|
|
},
|
|
{
|
|
"entropy": 9.955604076385498,
|
|
"epoch": 0.006714601460425818,
|
|
"grad_norm": 2.25,
|
|
"learning_rate": 5.9499999999999996e-05,
|
|
"loss": 8.1343,
|
|
"mean_token_accuracy": 0.0751000139862299,
|
|
"num_tokens": 223279.0,
|
|
"step": 120
|
|
},
|
|
{
|
|
"entropy": 9.898994541168213,
|
|
"epoch": 0.006994376521276893,
|
|
"grad_norm": 1.796875,
|
|
"learning_rate": 6.2e-05,
|
|
"loss": 8.0915,
|
|
"mean_token_accuracy": 0.07411009185016156,
|
|
"num_tokens": 232677.0,
|
|
"step": 125
|
|
},
|
|
{
|
|
"entropy": 9.685633563995362,
|
|
"epoch": 0.0072741515821279695,
|
|
"grad_norm": 2.015625,
|
|
"learning_rate": 6.450000000000001e-05,
|
|
"loss": 7.9322,
|
|
"mean_token_accuracy": 0.07425446137785911,
|
|
"num_tokens": 242890.0,
|
|
"step": 130
|
|
},
|
|
{
|
|
"entropy": 9.561828994750977,
|
|
"epoch": 0.007553926642979045,
|
|
"grad_norm": 1.78125,
|
|
"learning_rate": 6.7e-05,
|
|
"loss": 7.7431,
|
|
"mean_token_accuracy": 0.07636365555226803,
|
|
"num_tokens": 252490.0,
|
|
"step": 135
|
|
},
|
|
{
|
|
"entropy": 9.356704235076904,
|
|
"epoch": 0.007833701703830121,
|
|
"grad_norm": 1.5703125,
|
|
"learning_rate": 6.950000000000001e-05,
|
|
"loss": 7.6948,
|
|
"mean_token_accuracy": 0.08176706209778786,
|
|
"num_tokens": 261280.0,
|
|
"step": 140
|
|
},
|
|
{
|
|
"entropy": 9.167966556549072,
|
|
"epoch": 0.008113476764681197,
|
|
"grad_norm": 1.65625,
|
|
"learning_rate": 7.2e-05,
|
|
"loss": 7.65,
|
|
"mean_token_accuracy": 0.08069443143904209,
|
|
"num_tokens": 270000.0,
|
|
"step": 145
|
|
},
|
|
{
|
|
"entropy": 8.9812726020813,
|
|
"epoch": 0.008393251825532272,
|
|
"grad_norm": 2.453125,
|
|
"learning_rate": 7.45e-05,
|
|
"loss": 7.458,
|
|
"mean_token_accuracy": 0.08502581343054771,
|
|
"num_tokens": 279709.0,
|
|
"step": 150
|
|
},
|
|
{
|
|
"entropy": 8.729528427124023,
|
|
"epoch": 0.008673026886383347,
|
|
"grad_norm": 1.4453125,
|
|
"learning_rate": 7.7e-05,
|
|
"loss": 7.5225,
|
|
"mean_token_accuracy": 0.082467582821846,
|
|
"num_tokens": 288343.0,
|
|
"step": 155
|
|
},
|
|
{
|
|
"entropy": 8.595497226715088,
|
|
"epoch": 0.008952801947234423,
|
|
"grad_norm": 1.5234375,
|
|
"learning_rate": 7.950000000000001e-05,
|
|
"loss": 7.3566,
|
|
"mean_token_accuracy": 0.08212127052247524,
|
|
"num_tokens": 297387.0,
|
|
"step": 160
|
|
},
|
|
{
|
|
"entropy": 8.476513195037843,
|
|
"epoch": 0.0092325770080855,
|
|
"grad_norm": 1.625,
|
|
"learning_rate": 8.2e-05,
|
|
"loss": 7.3103,
|
|
"mean_token_accuracy": 0.08278237357735634,
|
|
"num_tokens": 307175.0,
|
|
"step": 165
|
|
},
|
|
{
|
|
"entropy": 8.29503517150879,
|
|
"epoch": 0.009512352068936575,
|
|
"grad_norm": 1.53125,
|
|
"learning_rate": 8.450000000000001e-05,
|
|
"loss": 7.2308,
|
|
"mean_token_accuracy": 0.08270871713757515,
|
|
"num_tokens": 316033.0,
|
|
"step": 170
|
|
},
|
|
{
|
|
"entropy": 8.15572385787964,
|
|
"epoch": 0.00979212712978765,
|
|
"grad_norm": 1.3984375,
|
|
"learning_rate": 8.7e-05,
|
|
"loss": 7.3076,
|
|
"mean_token_accuracy": 0.08210290595889091,
|
|
"num_tokens": 325740.0,
|
|
"step": 175
|
|
},
|
|
{
|
|
"entropy": 8.114148664474488,
|
|
"epoch": 0.010071902190638726,
|
|
"grad_norm": 1.375,
|
|
"learning_rate": 8.95e-05,
|
|
"loss": 7.2221,
|
|
"mean_token_accuracy": 0.09068928360939026,
|
|
"num_tokens": 334637.0,
|
|
"step": 180
|
|
},
|
|
{
|
|
"entropy": 8.006746578216553,
|
|
"epoch": 0.010351677251489803,
|
|
"grad_norm": 1.765625,
|
|
"learning_rate": 9.2e-05,
|
|
"loss": 7.2562,
|
|
"mean_token_accuracy": 0.0831844873726368,
|
|
"num_tokens": 344020.0,
|
|
"step": 185
|
|
},
|
|
{
|
|
"entropy": 7.939169883728027,
|
|
"epoch": 0.010631452312340878,
|
|
"grad_norm": 1.5703125,
|
|
"learning_rate": 9.45e-05,
|
|
"loss": 7.1292,
|
|
"mean_token_accuracy": 0.09121446087956428,
|
|
"num_tokens": 352924.0,
|
|
"step": 190
|
|
},
|
|
{
|
|
"entropy": 7.8939591407775875,
|
|
"epoch": 0.010911227373191954,
|
|
"grad_norm": 1.390625,
|
|
"learning_rate": 9.7e-05,
|
|
"loss": 7.149,
|
|
"mean_token_accuracy": 0.08797023706138134,
|
|
"num_tokens": 361755.0,
|
|
"step": 195
|
|
},
|
|
{
|
|
"entropy": 7.845557069778442,
|
|
"epoch": 0.01119100243404303,
|
|
"grad_norm": 1.2265625,
|
|
"learning_rate": 9.95e-05,
|
|
"loss": 7.1833,
|
|
"mean_token_accuracy": 0.08474454060196876,
|
|
"num_tokens": 370605.0,
|
|
"step": 200
|
|
},
|
|
{
|
|
"entropy": 7.757431316375732,
|
|
"epoch": 0.011470777494894105,
|
|
"grad_norm": 1.8125,
|
|
"learning_rate": 0.000102,
|
|
"loss": 7.0648,
|
|
"mean_token_accuracy": 0.08956050127744675,
|
|
"num_tokens": 379723.0,
|
|
"step": 205
|
|
},
|
|
{
|
|
"entropy": 7.683956050872803,
|
|
"epoch": 0.011750552555745182,
|
|
"grad_norm": 1.4140625,
|
|
"learning_rate": 0.00010449999999999999,
|
|
"loss": 7.0111,
|
|
"mean_token_accuracy": 0.09118850380182267,
|
|
"num_tokens": 389502.0,
|
|
"step": 210
|
|
},
|
|
{
|
|
"entropy": 7.613946056365966,
|
|
"epoch": 0.012030327616596257,
|
|
"grad_norm": 1.375,
|
|
"learning_rate": 0.000107,
|
|
"loss": 6.9865,
|
|
"mean_token_accuracy": 0.09210970029234886,
|
|
"num_tokens": 399116.0,
|
|
"step": 215
|
|
},
|
|
{
|
|
"entropy": 7.6888175964355465,
|
|
"epoch": 0.012310102677447332,
|
|
"grad_norm": 1.5390625,
|
|
"learning_rate": 0.0001095,
|
|
"loss": 7.0522,
|
|
"mean_token_accuracy": 0.08312384895980358,
|
|
"num_tokens": 408239.0,
|
|
"step": 220
|
|
},
|
|
{
|
|
"entropy": 7.582952213287354,
|
|
"epoch": 0.012589877738298408,
|
|
"grad_norm": 1.2109375,
|
|
"learning_rate": 0.000112,
|
|
"loss": 7.0558,
|
|
"mean_token_accuracy": 0.08601429760456085,
|
|
"num_tokens": 418801.0,
|
|
"step": 225
|
|
},
|
|
{
|
|
"entropy": 7.63437852859497,
|
|
"epoch": 0.012869652799149483,
|
|
"grad_norm": 1.3359375,
|
|
"learning_rate": 0.0001145,
|
|
"loss": 7.0975,
|
|
"mean_token_accuracy": 0.08165879175066948,
|
|
"num_tokens": 427584.0,
|
|
"step": 230
|
|
},
|
|
{
|
|
"entropy": 7.647107696533203,
|
|
"epoch": 0.01314942786000056,
|
|
"grad_norm": 1.4453125,
|
|
"learning_rate": 0.00011700000000000001,
|
|
"loss": 7.0398,
|
|
"mean_token_accuracy": 0.08437106013298035,
|
|
"num_tokens": 437345.0,
|
|
"step": 235
|
|
},
|
|
{
|
|
"entropy": 7.595585727691651,
|
|
"epoch": 0.013429202920851636,
|
|
"grad_norm": 1.4453125,
|
|
"learning_rate": 0.00011949999999999999,
|
|
"loss": 7.0693,
|
|
"mean_token_accuracy": 0.09033490121364593,
|
|
"num_tokens": 446550.0,
|
|
"step": 240
|
|
},
|
|
{
|
|
"entropy": 7.530496215820312,
|
|
"epoch": 0.013708977981702711,
|
|
"grad_norm": 1.1875,
|
|
"learning_rate": 0.000122,
|
|
"loss": 7.0332,
|
|
"mean_token_accuracy": 0.08355908021330834,
|
|
"num_tokens": 456394.0,
|
|
"step": 245
|
|
},
|
|
{
|
|
"entropy": 7.545537424087525,
|
|
"epoch": 0.013988753042553786,
|
|
"grad_norm": 1.3671875,
|
|
"learning_rate": 0.0001245,
|
|
"loss": 6.9254,
|
|
"mean_token_accuracy": 0.09105066359043121,
|
|
"num_tokens": 465467.0,
|
|
"step": 250
|
|
},
|
|
{
|
|
"entropy": 7.4903782367706295,
|
|
"epoch": 0.014268528103404862,
|
|
"grad_norm": 1.296875,
|
|
"learning_rate": 0.000127,
|
|
"loss": 6.9994,
|
|
"mean_token_accuracy": 0.09321118667721748,
|
|
"num_tokens": 474838.0,
|
|
"step": 255
|
|
},
|
|
{
|
|
"entropy": 7.407280015945434,
|
|
"epoch": 0.014548303164255939,
|
|
"grad_norm": 1.2109375,
|
|
"learning_rate": 0.0001295,
|
|
"loss": 6.8929,
|
|
"mean_token_accuracy": 0.09139523953199387,
|
|
"num_tokens": 483414.0,
|
|
"step": 260
|
|
},
|
|
{
|
|
"entropy": 7.481622314453125,
|
|
"epoch": 0.014828078225107014,
|
|
"grad_norm": 1.2578125,
|
|
"learning_rate": 0.000132,
|
|
"loss": 6.9248,
|
|
"mean_token_accuracy": 0.08975184708833694,
|
|
"num_tokens": 491802.0,
|
|
"step": 265
|
|
},
|
|
{
|
|
"entropy": 7.453138542175293,
|
|
"epoch": 0.01510785328595809,
|
|
"grad_norm": 1.2734375,
|
|
"learning_rate": 0.00013450000000000002,
|
|
"loss": 6.9211,
|
|
"mean_token_accuracy": 0.08993491381406785,
|
|
"num_tokens": 502007.0,
|
|
"step": 270
|
|
},
|
|
{
|
|
"entropy": 7.3567157745361325,
|
|
"epoch": 0.015387628346809165,
|
|
"grad_norm": 1.328125,
|
|
"learning_rate": 0.00013700000000000002,
|
|
"loss": 6.9517,
|
|
"mean_token_accuracy": 0.09103975892066955,
|
|
"num_tokens": 510806.0,
|
|
"step": 275
|
|
},
|
|
{
|
|
"entropy": 7.387433815002441,
|
|
"epoch": 0.015667403407660242,
|
|
"grad_norm": 1.3203125,
|
|
"learning_rate": 0.0001395,
|
|
"loss": 6.8927,
|
|
"mean_token_accuracy": 0.09123391136527062,
|
|
"num_tokens": 519526.0,
|
|
"step": 280
|
|
},
|
|
{
|
|
"entropy": 7.356798839569092,
|
|
"epoch": 0.015947178468511316,
|
|
"grad_norm": 1.2109375,
|
|
"learning_rate": 0.00014199999999999998,
|
|
"loss": 6.8069,
|
|
"mean_token_accuracy": 0.09702679142355919,
|
|
"num_tokens": 529036.0,
|
|
"step": 285
|
|
},
|
|
{
|
|
"entropy": 7.348542070388794,
|
|
"epoch": 0.016226953529362393,
|
|
"grad_norm": 1.53125,
|
|
"learning_rate": 0.0001445,
|
|
"loss": 6.8877,
|
|
"mean_token_accuracy": 0.09220745116472244,
|
|
"num_tokens": 537758.0,
|
|
"step": 290
|
|
},
|
|
{
|
|
"entropy": 7.252509832382202,
|
|
"epoch": 0.016506728590213467,
|
|
"grad_norm": 1.25,
|
|
"learning_rate": 0.000147,
|
|
"loss": 6.797,
|
|
"mean_token_accuracy": 0.0996977336704731,
|
|
"num_tokens": 546401.0,
|
|
"step": 295
|
|
},
|
|
{
|
|
"entropy": 7.282312059402466,
|
|
"epoch": 0.016786503651064544,
|
|
"grad_norm": 1.1796875,
|
|
"learning_rate": 0.0001495,
|
|
"loss": 6.8229,
|
|
"mean_token_accuracy": 0.09258124381303787,
|
|
"num_tokens": 555839.0,
|
|
"step": 300
|
|
},
|
|
{
|
|
"entropy": 7.447148847579956,
|
|
"epoch": 0.01706627871191562,
|
|
"grad_norm": 1.6875,
|
|
"learning_rate": 0.000152,
|
|
"loss": 6.9703,
|
|
"mean_token_accuracy": 0.09544998183846473,
|
|
"num_tokens": 565150.0,
|
|
"step": 305
|
|
},
|
|
{
|
|
"entropy": 7.174850606918335,
|
|
"epoch": 0.017346053772766695,
|
|
"grad_norm": 1.3984375,
|
|
"learning_rate": 0.00015450000000000001,
|
|
"loss": 6.8185,
|
|
"mean_token_accuracy": 0.09337743893265724,
|
|
"num_tokens": 574376.0,
|
|
"step": 310
|
|
},
|
|
{
|
|
"entropy": 7.367434978485107,
|
|
"epoch": 0.01762582883361777,
|
|
"grad_norm": 1.2890625,
|
|
"learning_rate": 0.000157,
|
|
"loss": 6.9439,
|
|
"mean_token_accuracy": 0.09078950360417366,
|
|
"num_tokens": 583075.0,
|
|
"step": 315
|
|
},
|
|
{
|
|
"entropy": 7.410228586196899,
|
|
"epoch": 0.017905603894468845,
|
|
"grad_norm": 1.234375,
|
|
"learning_rate": 0.0001595,
|
|
"loss": 6.9607,
|
|
"mean_token_accuracy": 0.09146012365818024,
|
|
"num_tokens": 593003.0,
|
|
"step": 320
|
|
},
|
|
{
|
|
"entropy": 7.167853116989136,
|
|
"epoch": 0.018185378955319922,
|
|
"grad_norm": 1.3515625,
|
|
"learning_rate": 0.000162,
|
|
"loss": 6.7824,
|
|
"mean_token_accuracy": 0.10193281322717666,
|
|
"num_tokens": 602273.0,
|
|
"step": 325
|
|
},
|
|
{
|
|
"entropy": 7.272089242935181,
|
|
"epoch": 0.018465154016171,
|
|
"grad_norm": 1.3671875,
|
|
"learning_rate": 0.00016450000000000001,
|
|
"loss": 6.7087,
|
|
"mean_token_accuracy": 0.09642531052231788,
|
|
"num_tokens": 610364.0,
|
|
"step": 330
|
|
},
|
|
{
|
|
"entropy": 7.081447410583496,
|
|
"epoch": 0.018744929077022073,
|
|
"grad_norm": 1.4140625,
|
|
"learning_rate": 0.00016700000000000002,
|
|
"loss": 6.7188,
|
|
"mean_token_accuracy": 0.0957895465195179,
|
|
"num_tokens": 618753.0,
|
|
"step": 335
|
|
},
|
|
{
|
|
"entropy": 7.264566278457641,
|
|
"epoch": 0.01902470413787315,
|
|
"grad_norm": 1.421875,
|
|
"learning_rate": 0.00016950000000000003,
|
|
"loss": 6.8352,
|
|
"mean_token_accuracy": 0.0967351920902729,
|
|
"num_tokens": 627697.0,
|
|
"step": 340
|
|
},
|
|
{
|
|
"entropy": 7.207143497467041,
|
|
"epoch": 0.019304479198724227,
|
|
"grad_norm": 1.4296875,
|
|
"learning_rate": 0.00017199999999999998,
|
|
"loss": 6.7869,
|
|
"mean_token_accuracy": 0.093389343470335,
|
|
"num_tokens": 636499.0,
|
|
"step": 345
|
|
},
|
|
{
|
|
"entropy": 7.1959278106689455,
|
|
"epoch": 0.0195842542595753,
|
|
"grad_norm": 1.2890625,
|
|
"learning_rate": 0.00017449999999999999,
|
|
"loss": 6.8033,
|
|
"mean_token_accuracy": 0.08842543438076973,
|
|
"num_tokens": 646426.0,
|
|
"step": 350
|
|
},
|
|
{
|
|
"entropy": 7.133398866653442,
|
|
"epoch": 0.019864029320426378,
|
|
"grad_norm": 1.125,
|
|
"learning_rate": 0.000177,
|
|
"loss": 6.7289,
|
|
"mean_token_accuracy": 0.09562224373221398,
|
|
"num_tokens": 656593.0,
|
|
"step": 355
|
|
},
|
|
{
|
|
"entropy": 7.104539823532105,
|
|
"epoch": 0.020143804381277452,
|
|
"grad_norm": 1.265625,
|
|
"learning_rate": 0.0001795,
|
|
"loss": 6.7823,
|
|
"mean_token_accuracy": 0.0965128093957901,
|
|
"num_tokens": 665631.0,
|
|
"step": 360
|
|
},
|
|
{
|
|
"entropy": 7.133638668060303,
|
|
"epoch": 0.02042357944212853,
|
|
"grad_norm": 1.21875,
|
|
"learning_rate": 0.000182,
|
|
"loss": 6.6893,
|
|
"mean_token_accuracy": 0.09800055772066116,
|
|
"num_tokens": 674943.0,
|
|
"step": 365
|
|
},
|
|
{
|
|
"entropy": 6.991700029373169,
|
|
"epoch": 0.020703354502979606,
|
|
"grad_norm": 1.4375,
|
|
"learning_rate": 0.0001845,
|
|
"loss": 6.6059,
|
|
"mean_token_accuracy": 0.10655041188001632,
|
|
"num_tokens": 683500.0,
|
|
"step": 370
|
|
},
|
|
{
|
|
"entropy": 7.158477783203125,
|
|
"epoch": 0.02098312956383068,
|
|
"grad_norm": 1.3359375,
|
|
"learning_rate": 0.000187,
|
|
"loss": 6.7946,
|
|
"mean_token_accuracy": 0.09516055136919022,
|
|
"num_tokens": 692147.0,
|
|
"step": 375
|
|
},
|
|
{
|
|
"entropy": 7.127016592025757,
|
|
"epoch": 0.021262904624681757,
|
|
"grad_norm": 1.328125,
|
|
"learning_rate": 0.0001895,
|
|
"loss": 6.7322,
|
|
"mean_token_accuracy": 0.09688446819782257,
|
|
"num_tokens": 701226.0,
|
|
"step": 380
|
|
},
|
|
{
|
|
"entropy": 7.150118589401245,
|
|
"epoch": 0.02154267968553283,
|
|
"grad_norm": 1.421875,
|
|
"learning_rate": 0.000192,
|
|
"loss": 6.7553,
|
|
"mean_token_accuracy": 0.10244472473859786,
|
|
"num_tokens": 710886.0,
|
|
"step": 385
|
|
},
|
|
{
|
|
"entropy": 7.099573230743408,
|
|
"epoch": 0.021822454746383908,
|
|
"grad_norm": 1.5078125,
|
|
"learning_rate": 0.0001945,
|
|
"loss": 6.6973,
|
|
"mean_token_accuracy": 0.09484406411647797,
|
|
"num_tokens": 719209.0,
|
|
"step": 390
|
|
},
|
|
{
|
|
"entropy": 6.925067281723022,
|
|
"epoch": 0.022102229807234985,
|
|
"grad_norm": 1.3046875,
|
|
"learning_rate": 0.00019700000000000002,
|
|
"loss": 6.5703,
|
|
"mean_token_accuracy": 0.10873782187700272,
|
|
"num_tokens": 728376.0,
|
|
"step": 395
|
|
},
|
|
{
|
|
"entropy": 7.04664626121521,
|
|
"epoch": 0.02238200486808606,
|
|
"grad_norm": 1.1171875,
|
|
"learning_rate": 0.00019950000000000002,
|
|
"loss": 6.6328,
|
|
"mean_token_accuracy": 0.09875192418694496,
|
|
"num_tokens": 737110.0,
|
|
"step": 400
|
|
},
|
|
{
|
|
"entropy": 7.108943176269531,
|
|
"epoch": 0.022661779928937135,
|
|
"grad_norm": 1.3984375,
|
|
"learning_rate": 0.000202,
|
|
"loss": 6.6957,
|
|
"mean_token_accuracy": 0.09161964356899262,
|
|
"num_tokens": 746133.0,
|
|
"step": 405
|
|
},
|
|
{
|
|
"entropy": 6.9601562976837155,
|
|
"epoch": 0.02294155498978821,
|
|
"grad_norm": 1.296875,
|
|
"learning_rate": 0.00020449999999999998,
|
|
"loss": 6.6511,
|
|
"mean_token_accuracy": 0.10639781132340431,
|
|
"num_tokens": 755994.0,
|
|
"step": 410
|
|
},
|
|
{
|
|
"entropy": 6.884982681274414,
|
|
"epoch": 0.023221330050639286,
|
|
"grad_norm": 1.296875,
|
|
"learning_rate": 0.000207,
|
|
"loss": 6.5945,
|
|
"mean_token_accuracy": 0.10252679362893105,
|
|
"num_tokens": 764630.0,
|
|
"step": 415
|
|
},
|
|
{
|
|
"entropy": 7.020522737503052,
|
|
"epoch": 0.023501105111490363,
|
|
"grad_norm": 1.28125,
|
|
"learning_rate": 0.0002095,
|
|
"loss": 6.6171,
|
|
"mean_token_accuracy": 0.10485479906201363,
|
|
"num_tokens": 773577.0,
|
|
"step": 420
|
|
},
|
|
{
|
|
"entropy": 6.976370239257813,
|
|
"epoch": 0.023780880172341437,
|
|
"grad_norm": 1.234375,
|
|
"learning_rate": 0.000212,
|
|
"loss": 6.5175,
|
|
"mean_token_accuracy": 0.10637958571314812,
|
|
"num_tokens": 782405.0,
|
|
"step": 425
|
|
},
|
|
{
|
|
"entropy": 6.990795373916626,
|
|
"epoch": 0.024060655233192514,
|
|
"grad_norm": 1.2109375,
|
|
"learning_rate": 0.0002145,
|
|
"loss": 6.679,
|
|
"mean_token_accuracy": 0.1029113069176674,
|
|
"num_tokens": 790940.0,
|
|
"step": 430
|
|
},
|
|
{
|
|
"entropy": 6.904673862457275,
|
|
"epoch": 0.024340430294043588,
|
|
"grad_norm": 1.359375,
|
|
"learning_rate": 0.00021700000000000002,
|
|
"loss": 6.6626,
|
|
"mean_token_accuracy": 0.09815835878252983,
|
|
"num_tokens": 800023.0,
|
|
"step": 435
|
|
},
|
|
{
|
|
"entropy": 7.0055601596832275,
|
|
"epoch": 0.024620205354894665,
|
|
"grad_norm": 1.203125,
|
|
"learning_rate": 0.0002195,
|
|
"loss": 6.5512,
|
|
"mean_token_accuracy": 0.11006631255149842,
|
|
"num_tokens": 809146.0,
|
|
"step": 440
|
|
},
|
|
{
|
|
"entropy": 6.848315715789795,
|
|
"epoch": 0.024899980415745742,
|
|
"grad_norm": 1.2265625,
|
|
"learning_rate": 0.000222,
|
|
"loss": 6.5253,
|
|
"mean_token_accuracy": 0.10747256800532341,
|
|
"num_tokens": 818300.0,
|
|
"step": 445
|
|
},
|
|
{
|
|
"entropy": 6.93147349357605,
|
|
"epoch": 0.025179755476596816,
|
|
"grad_norm": 1.2421875,
|
|
"learning_rate": 0.0002245,
|
|
"loss": 6.5748,
|
|
"mean_token_accuracy": 0.10701234936714173,
|
|
"num_tokens": 826991.0,
|
|
"step": 450
|
|
},
|
|
{
|
|
"entropy": 6.828905248641968,
|
|
"epoch": 0.025459530537447893,
|
|
"grad_norm": 1.265625,
|
|
"learning_rate": 0.00022700000000000002,
|
|
"loss": 6.5695,
|
|
"mean_token_accuracy": 0.10557045862078666,
|
|
"num_tokens": 836530.0,
|
|
"step": 455
|
|
},
|
|
{
|
|
"entropy": 6.996236515045166,
|
|
"epoch": 0.025739305598298966,
|
|
"grad_norm": 1.1953125,
|
|
"learning_rate": 0.00022950000000000002,
|
|
"loss": 6.6996,
|
|
"mean_token_accuracy": 0.10405678376555443,
|
|
"num_tokens": 845754.0,
|
|
"step": 460
|
|
},
|
|
{
|
|
"entropy": 6.909709072113037,
|
|
"epoch": 0.026019080659150044,
|
|
"grad_norm": 1.234375,
|
|
"learning_rate": 0.00023200000000000003,
|
|
"loss": 6.5825,
|
|
"mean_token_accuracy": 0.10672992318868638,
|
|
"num_tokens": 854778.0,
|
|
"step": 465
|
|
},
|
|
{
|
|
"entropy": 6.931979942321777,
|
|
"epoch": 0.02629885572000112,
|
|
"grad_norm": 1.328125,
|
|
"learning_rate": 0.00023449999999999998,
|
|
"loss": 6.5878,
|
|
"mean_token_accuracy": 0.10348478704690933,
|
|
"num_tokens": 863057.0,
|
|
"step": 470
|
|
},
|
|
{
|
|
"entropy": 6.977036380767823,
|
|
"epoch": 0.026578630780852194,
|
|
"grad_norm": 1.3125,
|
|
"learning_rate": 0.000237,
|
|
"loss": 6.6828,
|
|
"mean_token_accuracy": 0.10146789252758026,
|
|
"num_tokens": 872781.0,
|
|
"step": 475
|
|
},
|
|
{
|
|
"entropy": 6.804804468154908,
|
|
"epoch": 0.02685840584170327,
|
|
"grad_norm": 1.3125,
|
|
"learning_rate": 0.0002395,
|
|
"loss": 6.4354,
|
|
"mean_token_accuracy": 0.10844895020127296,
|
|
"num_tokens": 881815.0,
|
|
"step": 480
|
|
},
|
|
{
|
|
"entropy": 6.778733253479004,
|
|
"epoch": 0.027138180902554345,
|
|
"grad_norm": 1.2109375,
|
|
"learning_rate": 0.000242,
|
|
"loss": 6.5231,
|
|
"mean_token_accuracy": 0.10757277086377144,
|
|
"num_tokens": 890745.0,
|
|
"step": 485
|
|
},
|
|
{
|
|
"entropy": 6.819875049591064,
|
|
"epoch": 0.027417955963405422,
|
|
"grad_norm": 1.1171875,
|
|
"learning_rate": 0.0002445,
|
|
"loss": 6.5657,
|
|
"mean_token_accuracy": 0.10660690292716027,
|
|
"num_tokens": 900315.0,
|
|
"step": 490
|
|
},
|
|
{
|
|
"entropy": 6.791854763031006,
|
|
"epoch": 0.0276977310242565,
|
|
"grad_norm": 1.2734375,
|
|
"learning_rate": 0.000247,
|
|
"loss": 6.5405,
|
|
"mean_token_accuracy": 0.10886701792478562,
|
|
"num_tokens": 909367.0,
|
|
"step": 495
|
|
},
|
|
{
|
|
"entropy": 6.921594429016113,
|
|
"epoch": 0.027977506085107573,
|
|
"grad_norm": 1.234375,
|
|
"learning_rate": 0.0002495,
|
|
"loss": 6.6413,
|
|
"mean_token_accuracy": 0.10079527497291565,
|
|
"num_tokens": 919180.0,
|
|
"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": 1366924769280000.0,
|
|
"train_batch_size": 16,
|
|
"trial_name": null,
|
|
"trial_params": null
|
|
}
|