| { |
| "best_global_step": 1000, |
| "best_metric": 1.9334253933452872, |
| "best_model_checkpoint": "/content/cot-split2/checkpoint-1000", |
| "epoch": 0.7831466833737959, |
| "eval_steps": 250, |
| "global_step": 1000, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0, |
| "eval_bpd": 5.449506205976702, |
| "eval_entropy": 3.8259646681880883, |
| "eval_loss": 3.7819511890411377, |
| "eval_mean_token_accuracy": 0.3298281561563004, |
| "eval_nll": 3.777309862116675, |
| "eval_num_tokens": 0.0, |
| "eval_ppl": 43.69832894461198, |
| "eval_runtime": 318.4675, |
| "eval_samples_per_second": 570.259, |
| "eval_steps_per_second": 8.911, |
| "step": 0 |
| }, |
| { |
| "entropy": 4.035161377492861, |
| "epoch": 0.0007831466833737959, |
| "grad_norm": 2.9375, |
| "learning_rate": 0.0, |
| "loss": 3.6375203132629395, |
| "mean_token_accuracy": 0.3242907148120855, |
| "num_tokens": 156431.0, |
| "step": 1 |
| }, |
| { |
| "entropy": 4.157279884560115, |
| "epoch": 0.019578667084344897, |
| "grad_norm": 1.8515625, |
| "learning_rate": 9.375e-05, |
| "loss": 3.7293694814046225, |
| "mean_token_accuracy": 0.3223868495477843, |
| "num_tokens": 3882398.0, |
| "step": 25 |
| }, |
| { |
| "entropy": 4.069613211207327, |
| "epoch": 0.03915733416868979, |
| "grad_norm": 6.65625, |
| "learning_rate": 0.00019140625, |
| "loss": 3.286188659667969, |
| "mean_token_accuracy": 0.3470086962011918, |
| "num_tokens": 7771579.0, |
| "step": 50 |
| }, |
| { |
| "entropy": 3.712915897718112, |
| "epoch": 0.05873600125303469, |
| "grad_norm": 1.078125, |
| "learning_rate": 0.0002890625, |
| "loss": 2.9255511474609377, |
| "mean_token_accuracy": 0.37689020381328076, |
| "num_tokens": 11677887.0, |
| "step": 75 |
| }, |
| { |
| "entropy": 3.4658741959475385, |
| "epoch": 0.07831466833737959, |
| "grad_norm": 1.671875, |
| "learning_rate": 0.00038671875, |
| "loss": 2.5379095458984375, |
| "mean_token_accuracy": 0.41490777290923164, |
| "num_tokens": 15584019.0, |
| "step": 100 |
| }, |
| { |
| "entropy": 3.295554994185651, |
| "epoch": 0.09789333542172449, |
| "grad_norm": 0.59765625, |
| "learning_rate": 0.000484375, |
| "loss": 2.405628204345703, |
| "mean_token_accuracy": 0.4313112262648087, |
| "num_tokens": 19484337.0, |
| "step": 125 |
| }, |
| { |
| "entropy": 3.1617948360015533, |
| "epoch": 0.11747200250606939, |
| "grad_norm": 0.80078125, |
| "learning_rate": 0.0004999075642455791, |
| "loss": 2.3207965087890625, |
| "mean_token_accuracy": 0.4436994497174724, |
| "num_tokens": 23377294.0, |
| "step": 150 |
| }, |
| { |
| "entropy": 3.0947463825081942, |
| "epoch": 0.1370506695904143, |
| "grad_norm": 0.6640625, |
| "learning_rate": 0.0004995565798713207, |
| "loss": 2.2583859252929686, |
| "mean_token_accuracy": 0.45017372579314874, |
| "num_tokens": 27287783.0, |
| "step": 175 |
| }, |
| { |
| "entropy": 3.045135160417696, |
| "epoch": 0.15662933667475917, |
| "grad_norm": 0.65625, |
| "learning_rate": 0.0004989440608306114, |
| "loss": 2.230653076171875, |
| "mean_token_accuracy": 0.4551180636124264, |
| "num_tokens": 31205249.0, |
| "step": 200 |
| }, |
| { |
| "entropy": 3.0288844077617805, |
| "epoch": 0.1762080037591041, |
| "grad_norm": 0.625, |
| "learning_rate": 0.0004980706490418613, |
| "loss": 2.214619140625, |
| "mean_token_accuracy": 0.4571791829436394, |
| "num_tokens": 35081817.0, |
| "step": 225 |
| }, |
| { |
| "entropy": 2.9857863094922594, |
| "epoch": 0.19578667084344897, |
| "grad_norm": 0.6015625, |
| "learning_rate": 0.0004969372598384225, |
| "loss": 2.183477783203125, |
| "mean_token_accuracy": 0.4623076980301137, |
| "num_tokens": 38961791.0, |
| "step": 250 |
| }, |
| { |
| "epoch": 0.19578667084344897, |
| "eval_bpd": 3.0542670212451286, |
| "eval_entropy": 2.831413693293467, |
| "eval_loss": 2.114036798477173, |
| "eval_mean_token_accuracy": 0.471896302096762, |
| "eval_nll": 2.1170565744532834, |
| "eval_num_tokens": 38961791.0, |
| "eval_ppl": 8.306651458853036, |
| "eval_runtime": 317.7783, |
| "eval_samples_per_second": 571.496, |
| "eval_steps_per_second": 8.931, |
| "step": 250 |
| }, |
| { |
| "entropy": 2.960300336004756, |
| "epoch": 0.2153653379277939, |
| "grad_norm": 0.59765625, |
| "learning_rate": 0.0004955450810093227, |
| "loss": 2.1587040710449217, |
| "mean_token_accuracy": 0.46429396638467557, |
| "num_tokens": 42868157.0, |
| "step": 275 |
| }, |
| { |
| "entropy": 2.9496937473585083, |
| "epoch": 0.23494400501213877, |
| "grad_norm": 0.6953125, |
| "learning_rate": 0.0004938955715544644, |
| "loss": 2.1702507019042967, |
| "mean_token_accuracy": 0.4644734879386833, |
| "num_tokens": 46774567.0, |
| "step": 300 |
| }, |
| { |
| "entropy": 2.925756508647974, |
| "epoch": 0.25452267209648366, |
| "grad_norm": 0.61328125, |
| "learning_rate": 0.0004919904601555969, |
| "loss": 2.1320841979980467, |
| "mean_token_accuracy": 0.4678374269200783, |
| "num_tokens": 50651509.0, |
| "step": 325 |
| }, |
| { |
| "entropy": 2.916845726644516, |
| "epoch": 0.2741013391808286, |
| "grad_norm": 0.60546875, |
| "learning_rate": 0.0004898317433646626, |
| "loss": 2.149468994140625, |
| "mean_token_accuracy": 0.46948018886239345, |
| "num_tokens": 54545113.0, |
| "step": 350 |
| }, |
| { |
| "entropy": 2.8672960236041836, |
| "epoch": 0.2936800062651735, |
| "grad_norm": 0.59765625, |
| "learning_rate": 0.00048742168351141525, |
| "loss": 2.095373077392578, |
| "mean_token_accuracy": 0.47492304637997923, |
| "num_tokens": 58450259.0, |
| "step": 375 |
| }, |
| { |
| "entropy": 2.866450867223391, |
| "epoch": 0.31325867334951835, |
| "grad_norm": 0.609375, |
| "learning_rate": 0.00048476280633250404, |
| "loss": 2.093819122314453, |
| "mean_token_accuracy": 0.4762016071737522, |
| "num_tokens": 62334830.0, |
| "step": 400 |
| }, |
| { |
| "entropy": 2.8444264384733913, |
| "epoch": 0.33283734043386326, |
| "grad_norm": 0.671875, |
| "learning_rate": 0.00048185789832450813, |
| "loss": 2.07142822265625, |
| "mean_token_accuracy": 0.47914267030296726, |
| "num_tokens": 66254709.0, |
| "step": 425 |
| }, |
| { |
| "entropy": 2.8643093723566215, |
| "epoch": 0.3524160075182082, |
| "grad_norm": 0.578125, |
| "learning_rate": 0.00047871000382369524, |
| "loss": 2.0907044982910157, |
| "mean_token_accuracy": 0.475262880935016, |
| "num_tokens": 70149264.0, |
| "step": 450 |
| }, |
| { |
| "entropy": 2.8425666356253885, |
| "epoch": 0.37199467460255303, |
| "grad_norm": 1.453125, |
| "learning_rate": 0.0004753224218155648, |
| "loss": 2.0714340209960938, |
| "mean_token_accuracy": 0.47639239522332366, |
| "num_tokens": 74036195.0, |
| "step": 475 |
| }, |
| { |
| "entropy": 2.8495917513303857, |
| "epoch": 0.39157334168689795, |
| "grad_norm": 0.58203125, |
| "learning_rate": 0.0004716987024775197, |
| "loss": 2.0957228088378907, |
| "mean_token_accuracy": 0.4775797151536097, |
| "num_tokens": 77964744.0, |
| "step": 500 |
| }, |
| { |
| "epoch": 0.39157334168689795, |
| "eval_bpd": 2.9013035889897933, |
| "eval_entropy": 2.6662824092598547, |
| "eval_loss": 2.007856607437134, |
| "eval_mean_token_accuracy": 0.48950937525505117, |
| "eval_nll": 2.0110304026567256, |
| "eval_num_tokens": 77964744.0, |
| "eval_ppl": 7.471011532715633, |
| "eval_runtime": 317.7246, |
| "eval_samples_per_second": 571.593, |
| "eval_steps_per_second": 8.932, |
| "step": 500 |
| }, |
| { |
| "entropy": 2.8462999362059764, |
| "epoch": 0.41115200877124286, |
| "grad_norm": 0.59765625, |
| "learning_rate": 0.000467842643458289, |
| "loss": 2.0840892028808593, |
| "mean_token_accuracy": 0.47706405281698494, |
| "num_tokens": 81861726.0, |
| "step": 525 |
| }, |
| { |
| "entropy": 2.8082084615239284, |
| "epoch": 0.4307306758555878, |
| "grad_norm": 0.62109375, |
| "learning_rate": 0.0004637582858980017, |
| "loss": 2.041808624267578, |
| "mean_token_accuracy": 0.4826572669515401, |
| "num_tokens": 85760943.0, |
| "step": 550 |
| }, |
| { |
| "entropy": 2.8388380295514866, |
| "epoch": 0.45030934293993263, |
| "grad_norm": 0.55859375, |
| "learning_rate": 0.0004594499101930825, |
| "loss": 2.078563385009766, |
| "mean_token_accuracy": 0.4769605679393806, |
| "num_tokens": 89653726.0, |
| "step": 575 |
| }, |
| { |
| "entropy": 2.8302025972271743, |
| "epoch": 0.46988801002427755, |
| "grad_norm": 0.578125, |
| "learning_rate": 0.00045492203151040655, |
| "loss": 2.0540911865234377, |
| "mean_token_accuracy": 0.478844059310469, |
| "num_tokens": 93570215.0, |
| "step": 600 |
| }, |
| { |
| "entropy": 2.792682902628425, |
| "epoch": 0.48946667710862246, |
| "grad_norm": 0.69140625, |
| "learning_rate": 0.00045017939505541626, |
| "loss": 2.0480902099609377, |
| "mean_token_accuracy": 0.4848035732412953, |
| "num_tokens": 97477693.0, |
| "step": 625 |
| }, |
| { |
| "entropy": 2.7777595483840716, |
| "epoch": 0.5090453441929673, |
| "grad_norm": 0.66796875, |
| "learning_rate": 0.00044522697109915813, |
| "loss": 2.0343661499023438, |
| "mean_token_accuracy": 0.48608020488521336, |
| "num_tokens": 101394752.0, |
| "step": 650 |
| }, |
| { |
| "entropy": 2.742625050496606, |
| "epoch": 0.5286240112773123, |
| "grad_norm": 0.65625, |
| "learning_rate": 0.0004400699497694506, |
| "loss": 2.0052694702148437, |
| "mean_token_accuracy": 0.4903364669038516, |
| "num_tokens": 105270418.0, |
| "step": 675 |
| }, |
| { |
| "entropy": 2.8160673456556586, |
| "epoch": 0.5482026783616571, |
| "grad_norm": 0.65234375, |
| "learning_rate": 0.00043471373561164396, |
| "loss": 2.0440855407714844, |
| "mean_token_accuracy": 0.4817835272308287, |
| "num_tokens": 109173756.0, |
| "step": 700 |
| }, |
| { |
| "entropy": 2.731675431270922, |
| "epoch": 0.567781345446002, |
| "grad_norm": 0.625, |
| "learning_rate": 0.00042916394192466997, |
| "loss": 1.9994474792480468, |
| "mean_token_accuracy": 0.49231067122440925, |
| "num_tokens": 113075567.0, |
| "step": 725 |
| }, |
| { |
| "entropy": 2.761901786283376, |
| "epoch": 0.587360012530347, |
| "grad_norm": 0.5859375, |
| "learning_rate": 0.00042342638487831874, |
| "loss": 2.0215072631835938, |
| "mean_token_accuracy": 0.48973995306839846, |
| "num_tokens": 116976712.0, |
| "step": 750 |
| }, |
| { |
| "epoch": 0.587360012530347, |
| "eval_bpd": 2.8346400005221386, |
| "eval_entropy": 2.564649322232512, |
| "eval_loss": 1.9616204500198364, |
| "eval_mean_token_accuracy": 0.4976015657363103, |
| "eval_nll": 1.964822724264362, |
| "eval_num_tokens": 116976712.0, |
| "eval_ppl": 7.133647855387014, |
| "eval_runtime": 317.7915, |
| "eval_samples_per_second": 571.472, |
| "eval_steps_per_second": 8.93, |
| "step": 750 |
| }, |
| { |
| "entropy": 2.772676199080625, |
| "epoch": 0.6069386796146918, |
| "grad_norm": 0.6484375, |
| "learning_rate": 0.0004175070774179078, |
| "loss": 2.0193243408203125, |
| "mean_token_accuracy": 0.4871398110581337, |
| "num_tokens": 120856970.0, |
| "step": 775 |
| }, |
| { |
| "entropy": 2.7522809489804128, |
| "epoch": 0.6265173466990367, |
| "grad_norm": 0.51953125, |
| "learning_rate": 0.0004114122229627302, |
| "loss": 2.0201119995117187, |
| "mean_token_accuracy": 0.4884969423881558, |
| "num_tokens": 124747012.0, |
| "step": 800 |
| }, |
| { |
| "entropy": 2.728323356258852, |
| "epoch": 0.6460960137833817, |
| "grad_norm": 0.59765625, |
| "learning_rate": 0.00040514820890488664, |
| "loss": 2.000582427978516, |
| "mean_token_accuracy": 0.49145337661979344, |
| "num_tokens": 128633902.0, |
| "step": 825 |
| }, |
| { |
| "entropy": 2.7372595225009375, |
| "epoch": 0.6656746808677265, |
| "grad_norm": 0.59765625, |
| "learning_rate": 0.00039872159991531533, |
| "loss": 2.00209716796875, |
| "mean_token_accuracy": 0.4912540972003919, |
| "num_tokens": 132518194.0, |
| "step": 850 |
| }, |
| { |
| "entropy": 2.738841588033162, |
| "epoch": 0.6852533479520714, |
| "grad_norm": 0.55859375, |
| "learning_rate": 0.0003921391310640325, |
| "loss": 1.994276580810547, |
| "mean_token_accuracy": 0.491252567341933, |
| "num_tokens": 136385199.0, |
| "step": 875 |
| }, |
| { |
| "entropy": 2.7295696415657966, |
| "epoch": 0.7048320150364163, |
| "grad_norm": 0.53515625, |
| "learning_rate": 0.0003854077007617969, |
| "loss": 1.9865512084960937, |
| "mean_token_accuracy": 0.49203635280876573, |
| "num_tokens": 140293961.0, |
| "step": 900 |
| }, |
| { |
| "entropy": 2.7351411749253978, |
| "epoch": 0.7244106821207612, |
| "grad_norm": 0.5703125, |
| "learning_rate": 0.0003785343635305919, |
| "loss": 1.9895498657226562, |
| "mean_token_accuracy": 0.49209919251085, |
| "num_tokens": 144186204.0, |
| "step": 925 |
| }, |
| { |
| "entropy": 2.7418933621077652, |
| "epoch": 0.7439893492051061, |
| "grad_norm": 0.63671875, |
| "learning_rate": 0.00037152632261050483, |
| "loss": 2.0013661193847656, |
| "mean_token_accuracy": 0.49214851600044945, |
| "num_tokens": 148108105.0, |
| "step": 950 |
| }, |
| { |
| "entropy": 2.7243121079421706, |
| "epoch": 0.763568016289451, |
| "grad_norm": 0.640625, |
| "learning_rate": 0.0003643909224107492, |
| "loss": 1.991259307861328, |
| "mean_token_accuracy": 0.4933121377369521, |
| "num_tokens": 152015615.0, |
| "step": 975 |
| }, |
| { |
| "entropy": 2.778145840960037, |
| "epoch": 0.7831466833737959, |
| "grad_norm": 0.58984375, |
| "learning_rate": 0.00035713564081274257, |
| "loss": 2.0278318786621092, |
| "mean_token_accuracy": 0.48574800621666386, |
| "num_tokens": 155927990.0, |
| "step": 1000 |
| }, |
| { |
| "epoch": 0.7831466833737959, |
| "eval_bpd": 2.789343226908039, |
| "eval_entropy": 2.525867688934465, |
| "eval_loss": 1.9301713705062866, |
| "eval_mean_token_accuracy": 0.502722952650009, |
| "eval_nll": 1.9334253933452872, |
| "eval_num_tokens": 155927990.0, |
| "eval_ppl": 6.913149988716967, |
| "eval_runtime": 317.7734, |
| "eval_samples_per_second": 571.505, |
| "eval_steps_per_second": 8.931, |
| "step": 1000 |
| } |
| ], |
| "logging_steps": 25, |
| "max_steps": 2554, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 2, |
| "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": 5.0707419727233024e+17, |
| "train_batch_size": 64, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
