| { |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 2.961038961038961, |
| "eval_steps": 500, |
| "global_step": 57, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.05194805194805195, |
| "grad_norm": 4.959436893463135, |
| "learning_rate": 2e-05, |
| "loss": 2.0897, |
| "step": 1 |
| }, |
| { |
| "epoch": 0.1038961038961039, |
| "grad_norm": 3.942784547805786, |
| "learning_rate": 4e-05, |
| "loss": 2.3866, |
| "step": 2 |
| }, |
| { |
| "epoch": 0.15584415584415584, |
| "grad_norm": 4.221513748168945, |
| "learning_rate": 6e-05, |
| "loss": 2.4587, |
| "step": 3 |
| }, |
| { |
| "epoch": 0.2077922077922078, |
| "grad_norm": 3.284791946411133, |
| "learning_rate": 8e-05, |
| "loss": 2.1799, |
| "step": 4 |
| }, |
| { |
| "epoch": 0.2597402597402597, |
| "grad_norm": 5.71047306060791, |
| "learning_rate": 0.0001, |
| "loss": 2.1779, |
| "step": 5 |
| }, |
| { |
| "epoch": 0.3116883116883117, |
| "grad_norm": 3.1526741981506348, |
| "learning_rate": 0.00012, |
| "loss": 2.1932, |
| "step": 6 |
| }, |
| { |
| "epoch": 0.36363636363636365, |
| "grad_norm": 3.1342685222625732, |
| "learning_rate": 0.00014, |
| "loss": 1.4442, |
| "step": 7 |
| }, |
| { |
| "epoch": 0.4155844155844156, |
| "grad_norm": 5.338088035583496, |
| "learning_rate": 0.00016, |
| "loss": 2.0, |
| "step": 8 |
| }, |
| { |
| "epoch": 0.4675324675324675, |
| "grad_norm": 4.116346836090088, |
| "learning_rate": 0.00018, |
| "loss": 1.5286, |
| "step": 9 |
| }, |
| { |
| "epoch": 0.5194805194805194, |
| "grad_norm": 2.5058088302612305, |
| "learning_rate": 0.0002, |
| "loss": 1.4238, |
| "step": 10 |
| }, |
| { |
| "epoch": 0.5714285714285714, |
| "grad_norm": 4.636209487915039, |
| "learning_rate": 0.0001999317060143023, |
| "loss": 1.511, |
| "step": 11 |
| }, |
| { |
| "epoch": 0.6233766233766234, |
| "grad_norm": 3.1061031818389893, |
| "learning_rate": 0.00019972691733857883, |
| "loss": 1.3241, |
| "step": 12 |
| }, |
| { |
| "epoch": 0.6753246753246753, |
| "grad_norm": 2.3636441230773926, |
| "learning_rate": 0.0001993859136895274, |
| "loss": 1.044, |
| "step": 13 |
| }, |
| { |
| "epoch": 0.7272727272727273, |
| "grad_norm": 2.5445024967193604, |
| "learning_rate": 0.0001989091608371146, |
| "loss": 0.9548, |
| "step": 14 |
| }, |
| { |
| "epoch": 0.7792207792207793, |
| "grad_norm": 2.371872901916504, |
| "learning_rate": 0.0001982973099683902, |
| "loss": 1.0614, |
| "step": 15 |
| }, |
| { |
| "epoch": 0.8311688311688312, |
| "grad_norm": 5.0257039070129395, |
| "learning_rate": 0.00019755119679804367, |
| "loss": 1.1408, |
| "step": 16 |
| }, |
| { |
| "epoch": 0.8831168831168831, |
| "grad_norm": 1.9052318334579468, |
| "learning_rate": 0.00019667184042691875, |
| "loss": 0.8993, |
| "step": 17 |
| }, |
| { |
| "epoch": 0.935064935064935, |
| "grad_norm": 2.2592344284057617, |
| "learning_rate": 0.0001956604419500441, |
| "loss": 1.0319, |
| "step": 18 |
| }, |
| { |
| "epoch": 0.987012987012987, |
| "grad_norm": 2.264328718185425, |
| "learning_rate": 0.00019451838281608197, |
| "loss": 1.483, |
| "step": 19 |
| }, |
| { |
| "epoch": 1.0389610389610389, |
| "grad_norm": 4.711090564727783, |
| "learning_rate": 0.00019324722294043558, |
| "loss": 1.4984, |
| "step": 20 |
| }, |
| { |
| "epoch": 1.0909090909090908, |
| "grad_norm": 1.5374799966812134, |
| "learning_rate": 0.00019184869857459232, |
| "loss": 0.7076, |
| "step": 21 |
| }, |
| { |
| "epoch": 1.1428571428571428, |
| "grad_norm": 2.26493501663208, |
| "learning_rate": 0.0001903247199346129, |
| "loss": 1.1543, |
| "step": 22 |
| }, |
| { |
| "epoch": 1.1948051948051948, |
| "grad_norm": 1.51936674118042, |
| "learning_rate": 0.0001886773685920062, |
| "loss": 0.7064, |
| "step": 23 |
| }, |
| { |
| "epoch": 1.2467532467532467, |
| "grad_norm": 2.1054816246032715, |
| "learning_rate": 0.00018690889463055283, |
| "loss": 0.8401, |
| "step": 24 |
| }, |
| { |
| "epoch": 1.2987012987012987, |
| "grad_norm": 2.4671549797058105, |
| "learning_rate": 0.00018502171357296144, |
| "loss": 1.0385, |
| "step": 25 |
| }, |
| { |
| "epoch": 1.3506493506493507, |
| "grad_norm": 1.6618045568466187, |
| "learning_rate": 0.00018301840308155507, |
| "loss": 0.5543, |
| "step": 26 |
| }, |
| { |
| "epoch": 1.4025974025974026, |
| "grad_norm": 1.7877517938613892, |
| "learning_rate": 0.00018090169943749476, |
| "loss": 0.5464, |
| "step": 27 |
| }, |
| { |
| "epoch": 1.4545454545454546, |
| "grad_norm": 3.1368205547332764, |
| "learning_rate": 0.00017867449380334834, |
| "loss": 1.2296, |
| "step": 28 |
| }, |
| { |
| "epoch": 1.5064935064935066, |
| "grad_norm": 1.902140498161316, |
| "learning_rate": 0.00017633982827411032, |
| "loss": 0.8562, |
| "step": 29 |
| }, |
| { |
| "epoch": 1.5584415584415585, |
| "grad_norm": 1.6555312871932983, |
| "learning_rate": 0.00017390089172206592, |
| "loss": 0.7186, |
| "step": 30 |
| }, |
| { |
| "epoch": 1.6103896103896105, |
| "grad_norm": 1.526365041732788, |
| "learning_rate": 0.00017136101544117525, |
| "loss": 0.6501, |
| "step": 31 |
| }, |
| { |
| "epoch": 1.6623376623376624, |
| "grad_norm": 1.7592672109603882, |
| "learning_rate": 0.00016872366859692627, |
| "loss": 0.9721, |
| "step": 32 |
| }, |
| { |
| "epoch": 1.7142857142857144, |
| "grad_norm": 1.7448221445083618, |
| "learning_rate": 0.0001659924534878723, |
| "loss": 0.6916, |
| "step": 33 |
| }, |
| { |
| "epoch": 1.7662337662337664, |
| "grad_norm": 2.216275930404663, |
| "learning_rate": 0.0001631711006253251, |
| "loss": 0.8664, |
| "step": 34 |
| }, |
| { |
| "epoch": 1.8181818181818183, |
| "grad_norm": 1.5474787950515747, |
| "learning_rate": 0.00016026346363792567, |
| "loss": 0.5205, |
| "step": 35 |
| }, |
| { |
| "epoch": 1.87012987012987, |
| "grad_norm": 2.3516626358032227, |
| "learning_rate": 0.00015727351400805052, |
| "loss": 1.0247, |
| "step": 36 |
| }, |
| { |
| "epoch": 1.922077922077922, |
| "grad_norm": 1.921420931816101, |
| "learning_rate": 0.00015420533564724495, |
| "loss": 0.9075, |
| "step": 37 |
| }, |
| { |
| "epoch": 1.974025974025974, |
| "grad_norm": 1.9142687320709229, |
| "learning_rate": 0.0001510631193180907, |
| "loss": 0.8456, |
| "step": 38 |
| }, |
| { |
| "epoch": 2.0259740259740258, |
| "grad_norm": 2.5155398845672607, |
| "learning_rate": 0.00014785115691012864, |
| "loss": 0.7963, |
| "step": 39 |
| }, |
| { |
| "epoch": 2.0779220779220777, |
| "grad_norm": 1.6434123516082764, |
| "learning_rate": 0.00014457383557765386, |
| "loss": 0.7432, |
| "step": 40 |
| }, |
| { |
| "epoch": 2.1298701298701297, |
| "grad_norm": 1.6011264324188232, |
| "learning_rate": 0.00014123563174739037, |
| "loss": 0.5455, |
| "step": 41 |
| }, |
| { |
| "epoch": 2.1818181818181817, |
| "grad_norm": 1.7301688194274902, |
| "learning_rate": 0.00013784110500423104, |
| "loss": 0.4415, |
| "step": 42 |
| }, |
| { |
| "epoch": 2.2337662337662336, |
| "grad_norm": 1.334765911102295, |
| "learning_rate": 0.00013439489186339282, |
| "loss": 0.3515, |
| "step": 43 |
| }, |
| { |
| "epoch": 2.2857142857142856, |
| "grad_norm": 1.9331822395324707, |
| "learning_rate": 0.00013090169943749476, |
| "loss": 0.6585, |
| "step": 44 |
| }, |
| { |
| "epoch": 2.3376623376623376, |
| "grad_norm": 1.5188448429107666, |
| "learning_rate": 0.0001273662990072083, |
| "loss": 0.3202, |
| "step": 45 |
| }, |
| { |
| "epoch": 2.3896103896103895, |
| "grad_norm": 2.288816213607788, |
| "learning_rate": 0.00012379351950426187, |
| "loss": 0.6965, |
| "step": 46 |
| }, |
| { |
| "epoch": 2.4415584415584415, |
| "grad_norm": 1.356819987297058, |
| "learning_rate": 0.00012018824091570103, |
| "loss": 0.2039, |
| "step": 47 |
| }, |
| { |
| "epoch": 2.4935064935064934, |
| "grad_norm": 2.0376572608947754, |
| "learning_rate": 0.000116555387618413, |
| "loss": 0.4097, |
| "step": 48 |
| }, |
| { |
| "epoch": 2.5454545454545454, |
| "grad_norm": 1.3941322565078735, |
| "learning_rate": 0.00011289992165302035, |
| "loss": 0.318, |
| "step": 49 |
| }, |
| { |
| "epoch": 2.5974025974025974, |
| "grad_norm": 1.9851269721984863, |
| "learning_rate": 0.00010922683594633021, |
| "loss": 0.4971, |
| "step": 50 |
| }, |
| { |
| "epoch": 2.6493506493506493, |
| "grad_norm": 1.670423984527588, |
| "learning_rate": 0.000105541147491597, |
| "loss": 0.3796, |
| "step": 51 |
| }, |
| { |
| "epoch": 2.7012987012987013, |
| "grad_norm": 1.6468473672866821, |
| "learning_rate": 0.00010184789049591299, |
| "loss": 0.5085, |
| "step": 52 |
| }, |
| { |
| "epoch": 2.7532467532467533, |
| "grad_norm": 1.2429389953613281, |
| "learning_rate": 9.815210950408704e-05, |
| "loss": 0.2477, |
| "step": 53 |
| }, |
| { |
| "epoch": 2.8051948051948052, |
| "grad_norm": 2.051650047302246, |
| "learning_rate": 9.4458852508403e-05, |
| "loss": 0.6422, |
| "step": 54 |
| }, |
| { |
| "epoch": 2.857142857142857, |
| "grad_norm": 1.2772984504699707, |
| "learning_rate": 9.077316405366981e-05, |
| "loss": 0.2495, |
| "step": 55 |
| }, |
| { |
| "epoch": 2.909090909090909, |
| "grad_norm": 1.7530468702316284, |
| "learning_rate": 8.710007834697969e-05, |
| "loss": 0.4244, |
| "step": 56 |
| }, |
| { |
| "epoch": 2.961038961038961, |
| "grad_norm": 1.568015694618225, |
| "learning_rate": 8.344461238158699e-05, |
| "loss": 0.3102, |
| "step": 57 |
| } |
| ], |
| "logging_steps": 1, |
| "max_steps": 95, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 5, |
| "save_steps": 19, |
| "stateful_callbacks": { |
| "TrainerControl": { |
| "args": { |
| "should_epoch_stop": false, |
| "should_evaluate": false, |
| "should_log": false, |
| "should_save": true, |
| "should_training_stop": false |
| }, |
| "attributes": {} |
| } |
| }, |
| "total_flos": 1.8923148064220774e+17, |
| "train_batch_size": 1, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
