495 lines
14 KiB
JSON
495 lines
14 KiB
JSON
{
|
|
"best_global_step": null,
|
|
"best_metric": null,
|
|
"best_model_checkpoint": null,
|
|
"epoch": 1.0,
|
|
"eval_steps": 500,
|
|
"global_step": 2331,
|
|
"is_hyper_param_search": false,
|
|
"is_local_process_zero": true,
|
|
"is_world_process_zero": true,
|
|
"log_history": [
|
|
{
|
|
"entropy": 1.7801356130838395,
|
|
"epoch": 0.021458076283461188,
|
|
"grad_norm": 2.984375,
|
|
"learning_rate": 6.282051282051282e-06,
|
|
"loss": 2.7148,
|
|
"mean_token_accuracy": 0.5160968991369009,
|
|
"num_tokens": 13027304.0,
|
|
"step": 50
|
|
},
|
|
{
|
|
"entropy": 2.12295062482357,
|
|
"epoch": 0.042916152566922376,
|
|
"grad_norm": 0.7890625,
|
|
"learning_rate": 1.2692307692307693e-05,
|
|
"loss": 2.1312,
|
|
"mean_token_accuracy": 0.5675665412843227,
|
|
"num_tokens": 26060152.0,
|
|
"step": 100
|
|
},
|
|
{
|
|
"entropy": 1.844578797519207,
|
|
"epoch": 0.06437422885038356,
|
|
"grad_norm": 0.82421875,
|
|
"learning_rate": 1.9102564102564103e-05,
|
|
"loss": 1.844,
|
|
"mean_token_accuracy": 0.6109572112560272,
|
|
"num_tokens": 39093818.0,
|
|
"step": 150
|
|
},
|
|
{
|
|
"entropy": 1.6819258108735085,
|
|
"epoch": 0.08583230513384475,
|
|
"grad_norm": 0.8515625,
|
|
"learning_rate": 2.551282051282051e-05,
|
|
"loss": 1.6802,
|
|
"mean_token_accuracy": 0.6382600109279156,
|
|
"num_tokens": 52123695.0,
|
|
"step": 200
|
|
},
|
|
{
|
|
"entropy": 1.5457392731308937,
|
|
"epoch": 0.10729038141730594,
|
|
"grad_norm": 0.80078125,
|
|
"learning_rate": 2.9996212711496212e-05,
|
|
"loss": 1.5424,
|
|
"mean_token_accuracy": 0.6616944916546345,
|
|
"num_tokens": 65156301.0,
|
|
"step": 250
|
|
},
|
|
{
|
|
"entropy": 1.4541309079527855,
|
|
"epoch": 0.12874845770076712,
|
|
"grad_norm": 0.83203125,
|
|
"learning_rate": 2.992893632784019e-05,
|
|
"loss": 1.4502,
|
|
"mean_token_accuracy": 0.6769822642207146,
|
|
"num_tokens": 78189155.0,
|
|
"step": 300
|
|
},
|
|
{
|
|
"entropy": 1.3839471048116685,
|
|
"epoch": 0.1502065339842283,
|
|
"grad_norm": 0.90234375,
|
|
"learning_rate": 2.9777932337106776e-05,
|
|
"loss": 1.3795,
|
|
"mean_token_accuracy": 0.6891845127940178,
|
|
"num_tokens": 91225387.0,
|
|
"step": 350
|
|
},
|
|
{
|
|
"entropy": 1.3477001741528511,
|
|
"epoch": 0.1716646102676895,
|
|
"grad_norm": 0.8046875,
|
|
"learning_rate": 2.95440476317018e-05,
|
|
"loss": 1.3427,
|
|
"mean_token_accuracy": 0.6955448819696903,
|
|
"num_tokens": 104264395.0,
|
|
"step": 400
|
|
},
|
|
{
|
|
"entropy": 1.3151711079478263,
|
|
"epoch": 0.1931226865511507,
|
|
"grad_norm": 0.79296875,
|
|
"learning_rate": 2.9228593933122738e-05,
|
|
"loss": 1.3096,
|
|
"mean_token_accuracy": 0.7014366894960403,
|
|
"num_tokens": 117304006.0,
|
|
"step": 450
|
|
},
|
|
{
|
|
"entropy": 1.286205082833767,
|
|
"epoch": 0.2145807628346119,
|
|
"grad_norm": 0.76953125,
|
|
"learning_rate": 2.8833340435285732e-05,
|
|
"loss": 1.281,
|
|
"mean_token_accuracy": 0.7059694881737232,
|
|
"num_tokens": 130330369.0,
|
|
"step": 500
|
|
},
|
|
{
|
|
"entropy": 1.256286238282919,
|
|
"epoch": 0.23603883911807305,
|
|
"grad_norm": 0.76953125,
|
|
"learning_rate": 2.8360503882159413e-05,
|
|
"loss": 1.2508,
|
|
"mean_token_accuracy": 0.711535804271698,
|
|
"num_tokens": 143357759.0,
|
|
"step": 550
|
|
},
|
|
{
|
|
"entropy": 1.2380341781675817,
|
|
"epoch": 0.25749691540153424,
|
|
"grad_norm": 0.78125,
|
|
"learning_rate": 2.7812736135354192e-05,
|
|
"loss": 1.2326,
|
|
"mean_token_accuracy": 0.7147769713401795,
|
|
"num_tokens": 156390933.0,
|
|
"step": 600
|
|
},
|
|
{
|
|
"entropy": 1.2227300041913987,
|
|
"epoch": 0.27895499168499543,
|
|
"grad_norm": 0.75,
|
|
"learning_rate": 2.7193109301393063e-05,
|
|
"loss": 1.2166,
|
|
"mean_token_accuracy": 0.7174056661128998,
|
|
"num_tokens": 169421819.0,
|
|
"step": 650
|
|
},
|
|
{
|
|
"entropy": 1.2046894371509551,
|
|
"epoch": 0.3004130679684566,
|
|
"grad_norm": 0.73828125,
|
|
"learning_rate": 2.6505098502076353e-05,
|
|
"loss": 1.1997,
|
|
"mean_token_accuracy": 0.7204455882310867,
|
|
"num_tokens": 182458556.0,
|
|
"step": 700
|
|
},
|
|
{
|
|
"entropy": 1.1964470529556275,
|
|
"epoch": 0.3218711442519178,
|
|
"grad_norm": 0.7734375,
|
|
"learning_rate": 2.575256238457122e-05,
|
|
"loss": 1.1904,
|
|
"mean_token_accuracy": 0.7221927911043167,
|
|
"num_tokens": 195498872.0,
|
|
"step": 750
|
|
},
|
|
{
|
|
"entropy": 1.1912988266348838,
|
|
"epoch": 0.343329220535379,
|
|
"grad_norm": 0.7578125,
|
|
"learning_rate": 2.4939721480533394e-05,
|
|
"loss": 1.1855,
|
|
"mean_token_accuracy": 0.7228270301222801,
|
|
"num_tokens": 208529244.0,
|
|
"step": 800
|
|
},
|
|
{
|
|
"entropy": 1.1789171296358107,
|
|
"epoch": 0.3647872968188402,
|
|
"grad_norm": 0.74609375,
|
|
"learning_rate": 2.4071134535632042e-05,
|
|
"loss": 1.1735,
|
|
"mean_token_accuracy": 0.7249331058561802,
|
|
"num_tokens": 221559622.0,
|
|
"step": 850
|
|
},
|
|
{
|
|
"entropy": 1.168346780538559,
|
|
"epoch": 0.3862453731023014,
|
|
"grad_norm": 0.734375,
|
|
"learning_rate": 2.3151672942231506e-05,
|
|
"loss": 1.1623,
|
|
"mean_token_accuracy": 0.7268652066588401,
|
|
"num_tokens": 234596401.0,
|
|
"step": 900
|
|
},
|
|
{
|
|
"entropy": 1.1605928602814675,
|
|
"epoch": 0.4077034493857626,
|
|
"grad_norm": 0.765625,
|
|
"learning_rate": 2.218649341862198e-05,
|
|
"loss": 1.1548,
|
|
"mean_token_accuracy": 0.728048648238182,
|
|
"num_tokens": 247619938.0,
|
|
"step": 950
|
|
},
|
|
{
|
|
"entropy": 1.1611860990524292,
|
|
"epoch": 0.4291615256692238,
|
|
"grad_norm": 0.76171875,
|
|
"learning_rate": 2.1181009088025206e-05,
|
|
"loss": 1.1556,
|
|
"mean_token_accuracy": 0.7282680490612984,
|
|
"num_tokens": 260654172.0,
|
|
"step": 1000
|
|
},
|
|
{
|
|
"entropy": 1.1592345923185348,
|
|
"epoch": 0.45061960195268497,
|
|
"grad_norm": 0.75,
|
|
"learning_rate": 2.014085911957592e-05,
|
|
"loss": 1.1534,
|
|
"mean_token_accuracy": 0.7286559711396694,
|
|
"num_tokens": 273684574.0,
|
|
"step": 1050
|
|
},
|
|
{
|
|
"entropy": 1.1588025417923928,
|
|
"epoch": 0.4720776782361461,
|
|
"grad_norm": 0.71875,
|
|
"learning_rate": 1.907187710154509e-05,
|
|
"loss": 1.1524,
|
|
"mean_token_accuracy": 0.7285710147023201,
|
|
"num_tokens": 286722109.0,
|
|
"step": 1100
|
|
},
|
|
{
|
|
"entropy": 1.147208172529936,
|
|
"epoch": 0.4935357545196073,
|
|
"grad_norm": 0.70703125,
|
|
"learning_rate": 1.7980058324180654e-05,
|
|
"loss": 1.1415,
|
|
"mean_token_accuracy": 0.7308723784983158,
|
|
"num_tokens": 299759049.0,
|
|
"step": 1150
|
|
},
|
|
{
|
|
"entropy": 1.1459918165206908,
|
|
"epoch": 0.5149938308030685,
|
|
"grad_norm": 0.71875,
|
|
"learning_rate": 1.687152615565724e-05,
|
|
"loss": 1.1394,
|
|
"mean_token_accuracy": 0.7309729683399201,
|
|
"num_tokens": 312785989.0,
|
|
"step": 1200
|
|
},
|
|
{
|
|
"entropy": 1.138493761718273,
|
|
"epoch": 0.5364519070865297,
|
|
"grad_norm": 0.703125,
|
|
"learning_rate": 1.5752497699712207e-05,
|
|
"loss": 1.1326,
|
|
"mean_token_accuracy": 0.7323851896822453,
|
|
"num_tokens": 325818163.0,
|
|
"step": 1250
|
|
},
|
|
{
|
|
"entropy": 1.142863957285881,
|
|
"epoch": 0.5579099833699909,
|
|
"grad_norm": 0.734375,
|
|
"learning_rate": 1.4629248927574217e-05,
|
|
"loss": 1.1361,
|
|
"mean_token_accuracy": 0.7315937775373459,
|
|
"num_tokens": 338855461.0,
|
|
"step": 1300
|
|
},
|
|
{
|
|
"entropy": 1.1367172484099866,
|
|
"epoch": 0.5793680596534521,
|
|
"grad_norm": 0.703125,
|
|
"learning_rate": 1.3508079479738632e-05,
|
|
"loss": 1.1308,
|
|
"mean_token_accuracy": 0.7327514922618866,
|
|
"num_tokens": 351888193.0,
|
|
"step": 1350
|
|
},
|
|
{
|
|
"entropy": 1.138921956717968,
|
|
"epoch": 0.6008261359369133,
|
|
"grad_norm": 0.6796875,
|
|
"learning_rate": 1.2395277334996045e-05,
|
|
"loss": 1.1327,
|
|
"mean_token_accuracy": 0.732325647175312,
|
|
"num_tokens": 364920249.0,
|
|
"step": 1400
|
|
},
|
|
{
|
|
"entropy": 1.1359232895076274,
|
|
"epoch": 0.6222842122203744,
|
|
"grad_norm": 0.6953125,
|
|
"learning_rate": 1.1297083544864386e-05,
|
|
"loss": 1.1293,
|
|
"mean_token_accuracy": 0.7327671211957931,
|
|
"num_tokens": 377954848.0,
|
|
"step": 1450
|
|
},
|
|
{
|
|
"entropy": 1.133522237688303,
|
|
"epoch": 0.6437422885038356,
|
|
"grad_norm": 0.6875,
|
|
"learning_rate": 1.0219657231208455e-05,
|
|
"loss": 1.1274,
|
|
"mean_token_accuracy": 0.7333970531821251,
|
|
"num_tokens": 390982547.0,
|
|
"step": 1500
|
|
},
|
|
{
|
|
"entropy": 1.1293988834321498,
|
|
"epoch": 0.6652003647872968,
|
|
"grad_norm": 0.67578125,
|
|
"learning_rate": 9.1690410433546e-06,
|
|
"loss": 1.1229,
|
|
"mean_token_accuracy": 0.7344232758879662,
|
|
"num_tokens": 404012732.0,
|
|
"step": 1550
|
|
},
|
|
{
|
|
"entropy": 1.1354819835722447,
|
|
"epoch": 0.686658441070758,
|
|
"grad_norm": 0.6640625,
|
|
"learning_rate": 8.151127268431263e-06,
|
|
"loss": 1.1294,
|
|
"mean_token_accuracy": 0.7327083775401115,
|
|
"num_tokens": 417050367.0,
|
|
"step": 1600
|
|
},
|
|
{
|
|
"entropy": 1.1320132564008236,
|
|
"epoch": 0.7081165173542192,
|
|
"grad_norm": 0.67578125,
|
|
"learning_rate": 7.171624785002444e-06,
|
|
"loss": 1.1255,
|
|
"mean_token_accuracy": 0.7334317502379417,
|
|
"num_tokens": 430085128.0,
|
|
"step": 1650
|
|
},
|
|
{
|
|
"entropy": 1.1373174653947353,
|
|
"epoch": 0.7295745936376804,
|
|
"grad_norm": 0.66796875,
|
|
"learning_rate": 6.236027045331792e-06,
|
|
"loss": 1.1309,
|
|
"mean_token_accuracy": 0.7325193387269974,
|
|
"num_tokens": 443116587.0,
|
|
"step": 1700
|
|
},
|
|
{
|
|
"entropy": 1.1353763638436793,
|
|
"epoch": 0.7510326699211416,
|
|
"grad_norm": 0.6484375,
|
|
"learning_rate": 5.349581265845804e-06,
|
|
"loss": 1.1285,
|
|
"mean_token_accuracy": 0.7330000093579292,
|
|
"num_tokens": 456150190.0,
|
|
"step": 1750
|
|
},
|
|
{
|
|
"entropy": 1.1333225139975547,
|
|
"epoch": 0.7724907462046028,
|
|
"grad_norm": 0.6796875,
|
|
"learning_rate": 4.517258998588792e-06,
|
|
"loss": 1.1267,
|
|
"mean_token_accuracy": 0.7335374882817268,
|
|
"num_tokens": 469187645.0,
|
|
"step": 1800
|
|
},
|
|
{
|
|
"entropy": 1.1265139281749725,
|
|
"epoch": 0.793948822488064,
|
|
"grad_norm": 0.66796875,
|
|
"learning_rate": 3.743728248716951e-06,
|
|
"loss": 1.1194,
|
|
"mean_token_accuracy": 0.7348724040389061,
|
|
"num_tokens": 482216628.0,
|
|
"step": 1850
|
|
},
|
|
{
|
|
"entropy": 1.1329601185023785,
|
|
"epoch": 0.8154068987715252,
|
|
"grad_norm": 0.70703125,
|
|
"learning_rate": 3.033327294407988e-06,
|
|
"loss": 1.1266,
|
|
"mean_token_accuracy": 0.7334246894717217,
|
|
"num_tokens": 495240948.0,
|
|
"step": 1900
|
|
},
|
|
{
|
|
"entropy": 1.1300819149613381,
|
|
"epoch": 0.8368649750549864,
|
|
"grad_norm": 0.65234375,
|
|
"learning_rate": 2.390040356015151e-06,
|
|
"loss": 1.1238,
|
|
"mean_token_accuracy": 0.7344046778976917,
|
|
"num_tokens": 508270012.0,
|
|
"step": 1950
|
|
},
|
|
{
|
|
"entropy": 1.128454978466034,
|
|
"epoch": 0.8583230513384476,
|
|
"grad_norm": 0.671875,
|
|
"learning_rate": 1.817475250922957e-06,
|
|
"loss": 1.1214,
|
|
"mean_token_accuracy": 0.7345199052989483,
|
|
"num_tokens": 521303590.0,
|
|
"step": 2000
|
|
},
|
|
{
|
|
"entropy": 1.1328057833015919,
|
|
"epoch": 0.8797811276219087,
|
|
"grad_norm": 0.66796875,
|
|
"learning_rate": 1.318843159425332e-06,
|
|
"loss": 1.1269,
|
|
"mean_token_accuracy": 0.733275038599968,
|
|
"num_tokens": 534336739.0,
|
|
"step": 2050
|
|
},
|
|
{
|
|
"entropy": 1.1365984900295734,
|
|
"epoch": 0.9012392039053699,
|
|
"grad_norm": 0.67578125,
|
|
"learning_rate": 8.969406151075494e-07,
|
|
"loss": 1.1311,
|
|
"mean_token_accuracy": 0.7326788820326329,
|
|
"num_tokens": 547357092.0,
|
|
"step": 2100
|
|
},
|
|
{
|
|
"entropy": 1.132674039453268,
|
|
"epoch": 0.9226972801888311,
|
|
"grad_norm": 0.66796875,
|
|
"learning_rate": 5.541338207371605e-07,
|
|
"loss": 1.126,
|
|
"mean_token_accuracy": 0.7336283141374588,
|
|
"num_tokens": 560392664.0,
|
|
"step": 2150
|
|
},
|
|
{
|
|
"entropy": 1.1360688427090644,
|
|
"epoch": 0.9441553564722922,
|
|
"grad_norm": 0.6484375,
|
|
"learning_rate": 2.923453776269558e-07,
|
|
"loss": 1.1291,
|
|
"mean_token_accuracy": 0.7327949374914169,
|
|
"num_tokens": 573423859.0,
|
|
"step": 2200
|
|
},
|
|
{
|
|
"entropy": 1.1402589020133018,
|
|
"epoch": 0.9656134327557534,
|
|
"grad_norm": 0.69921875,
|
|
"learning_rate": 1.130435028970872e-07,
|
|
"loss": 1.133,
|
|
"mean_token_accuracy": 0.7320883478224277,
|
|
"num_tokens": 586459673.0,
|
|
"step": 2250
|
|
},
|
|
{
|
|
"entropy": 1.129337106347084,
|
|
"epoch": 0.9870715090392146,
|
|
"grad_norm": 0.6875,
|
|
"learning_rate": 1.7233795110367935e-08,
|
|
"loss": 1.1221,
|
|
"mean_token_accuracy": 0.734157979041338,
|
|
"num_tokens": 599492881.0,
|
|
"step": 2300
|
|
}
|
|
],
|
|
"logging_steps": 50,
|
|
"max_steps": 2331,
|
|
"num_input_tokens_seen": 0,
|
|
"num_train_epochs": 1,
|
|
"save_steps": 50000,
|
|
"stateful_callbacks": {
|
|
"TrainerControl": {
|
|
"args": {
|
|
"should_epoch_stop": false,
|
|
"should_evaluate": false,
|
|
"should_log": false,
|
|
"should_save": true,
|
|
"should_training_stop": true
|
|
},
|
|
"attributes": {}
|
|
}
|
|
},
|
|
"total_flos": 5.135994653597762e+18,
|
|
"train_batch_size": 2,
|
|
"trial_name": null,
|
|
"trial_params": null
|
|
}
|