| { |
| "best_metric": 3.0149381160736084, |
| "best_model_checkpoint": "miner_id_24/checkpoint-50", |
| "epoch": 0.009938382031405287, |
| "eval_steps": 25, |
| "global_step": 50, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.00019876764062810574, |
| "grad_norm": 0.08420488983392715, |
| "learning_rate": 2e-05, |
| "loss": 1.6777, |
| "step": 1 |
| }, |
| { |
| "epoch": 0.00019876764062810574, |
| "eval_loss": 3.6612236499786377, |
| "eval_runtime": 67.3581, |
| "eval_samples_per_second": 125.805, |
| "eval_steps_per_second": 15.737, |
| "step": 1 |
| }, |
| { |
| "epoch": 0.0003975352812562115, |
| "grad_norm": 0.17416247725486755, |
| "learning_rate": 4e-05, |
| "loss": 2.4276, |
| "step": 2 |
| }, |
| { |
| "epoch": 0.0005963029218843172, |
| "grad_norm": 0.21623319387435913, |
| "learning_rate": 6e-05, |
| "loss": 2.6007, |
| "step": 3 |
| }, |
| { |
| "epoch": 0.000795070562512423, |
| "grad_norm": 0.15903286635875702, |
| "learning_rate": 8e-05, |
| "loss": 2.646, |
| "step": 4 |
| }, |
| { |
| "epoch": 0.0009938382031405286, |
| "grad_norm": 0.1743364781141281, |
| "learning_rate": 0.0001, |
| "loss": 2.9006, |
| "step": 5 |
| }, |
| { |
| "epoch": 0.0011926058437686344, |
| "grad_norm": 0.18579961359500885, |
| "learning_rate": 9.987820251299122e-05, |
| "loss": 2.7491, |
| "step": 6 |
| }, |
| { |
| "epoch": 0.0013913734843967402, |
| "grad_norm": 0.19385971128940582, |
| "learning_rate": 9.951340343707852e-05, |
| "loss": 2.7176, |
| "step": 7 |
| }, |
| { |
| "epoch": 0.001590141125024846, |
| "grad_norm": 0.20547613501548767, |
| "learning_rate": 9.890738003669029e-05, |
| "loss": 2.9435, |
| "step": 8 |
| }, |
| { |
| "epoch": 0.0017889087656529517, |
| "grad_norm": 0.21542897820472717, |
| "learning_rate": 9.806308479691595e-05, |
| "loss": 2.9636, |
| "step": 9 |
| }, |
| { |
| "epoch": 0.0019876764062810573, |
| "grad_norm": 0.20447668433189392, |
| "learning_rate": 9.698463103929542e-05, |
| "loss": 2.7969, |
| "step": 10 |
| }, |
| { |
| "epoch": 0.0021864440469091632, |
| "grad_norm": 0.22560608386993408, |
| "learning_rate": 9.567727288213005e-05, |
| "loss": 3.3065, |
| "step": 11 |
| }, |
| { |
| "epoch": 0.002385211687537269, |
| "grad_norm": 0.23869244754314423, |
| "learning_rate": 9.414737964294636e-05, |
| "loss": 3.1795, |
| "step": 12 |
| }, |
| { |
| "epoch": 0.002583979328165375, |
| "grad_norm": 0.21942728757858276, |
| "learning_rate": 9.24024048078213e-05, |
| "loss": 2.9018, |
| "step": 13 |
| }, |
| { |
| "epoch": 0.0027827469687934803, |
| "grad_norm": 0.24622339010238647, |
| "learning_rate": 9.045084971874738e-05, |
| "loss": 3.3347, |
| "step": 14 |
| }, |
| { |
| "epoch": 0.0029815146094215863, |
| "grad_norm": 0.2358148694038391, |
| "learning_rate": 8.83022221559489e-05, |
| "loss": 3.2586, |
| "step": 15 |
| }, |
| { |
| "epoch": 0.003180282250049692, |
| "grad_norm": 0.2474563717842102, |
| "learning_rate": 8.596699001693255e-05, |
| "loss": 3.2522, |
| "step": 16 |
| }, |
| { |
| "epoch": 0.003379049890677798, |
| "grad_norm": 0.26249605417251587, |
| "learning_rate": 8.345653031794292e-05, |
| "loss": 3.0814, |
| "step": 17 |
| }, |
| { |
| "epoch": 0.0035778175313059034, |
| "grad_norm": 0.27410539984703064, |
| "learning_rate": 8.07830737662829e-05, |
| "loss": 3.3012, |
| "step": 18 |
| }, |
| { |
| "epoch": 0.003776585171934009, |
| "grad_norm": 0.25694385170936584, |
| "learning_rate": 7.795964517353735e-05, |
| "loss": 3.1564, |
| "step": 19 |
| }, |
| { |
| "epoch": 0.0039753528125621145, |
| "grad_norm": 0.26552635431289673, |
| "learning_rate": 7.500000000000001e-05, |
| "loss": 3.2579, |
| "step": 20 |
| }, |
| { |
| "epoch": 0.0041741204531902205, |
| "grad_norm": 0.25721579790115356, |
| "learning_rate": 7.191855733945387e-05, |
| "loss": 3.2505, |
| "step": 21 |
| }, |
| { |
| "epoch": 0.0043728880938183265, |
| "grad_norm": 0.2668185234069824, |
| "learning_rate": 6.873032967079561e-05, |
| "loss": 3.0924, |
| "step": 22 |
| }, |
| { |
| "epoch": 0.0045716557344464325, |
| "grad_norm": 0.26348620653152466, |
| "learning_rate": 6.545084971874738e-05, |
| "loss": 3.2819, |
| "step": 23 |
| }, |
| { |
| "epoch": 0.004770423375074538, |
| "grad_norm": 0.26392704248428345, |
| "learning_rate": 6.209609477998338e-05, |
| "loss": 3.102, |
| "step": 24 |
| }, |
| { |
| "epoch": 0.004969191015702644, |
| "grad_norm": 0.26735320687294006, |
| "learning_rate": 5.868240888334653e-05, |
| "loss": 3.1296, |
| "step": 25 |
| }, |
| { |
| "epoch": 0.004969191015702644, |
| "eval_loss": 3.111262798309326, |
| "eval_runtime": 67.118, |
| "eval_samples_per_second": 126.255, |
| "eval_steps_per_second": 15.793, |
| "step": 25 |
| }, |
| { |
| "epoch": 0.00516795865633075, |
| "grad_norm": 0.3099224269390106, |
| "learning_rate": 5.522642316338268e-05, |
| "loss": 3.1568, |
| "step": 26 |
| }, |
| { |
| "epoch": 0.005366726296958855, |
| "grad_norm": 0.2683943808078766, |
| "learning_rate": 5.174497483512506e-05, |
| "loss": 3.3075, |
| "step": 27 |
| }, |
| { |
| "epoch": 0.005565493937586961, |
| "grad_norm": 0.290770560503006, |
| "learning_rate": 4.825502516487497e-05, |
| "loss": 3.1469, |
| "step": 28 |
| }, |
| { |
| "epoch": 0.005764261578215067, |
| "grad_norm": 0.2689608931541443, |
| "learning_rate": 4.477357683661734e-05, |
| "loss": 3.1745, |
| "step": 29 |
| }, |
| { |
| "epoch": 0.005963029218843173, |
| "grad_norm": 0.27947020530700684, |
| "learning_rate": 4.131759111665349e-05, |
| "loss": 3.1196, |
| "step": 30 |
| }, |
| { |
| "epoch": 0.006161796859471278, |
| "grad_norm": 0.3059743344783783, |
| "learning_rate": 3.790390522001662e-05, |
| "loss": 3.2627, |
| "step": 31 |
| }, |
| { |
| "epoch": 0.006360564500099384, |
| "grad_norm": 0.2834652066230774, |
| "learning_rate": 3.4549150281252636e-05, |
| "loss": 3.0715, |
| "step": 32 |
| }, |
| { |
| "epoch": 0.00655933214072749, |
| "grad_norm": 0.31416839361190796, |
| "learning_rate": 3.12696703292044e-05, |
| "loss": 3.1854, |
| "step": 33 |
| }, |
| { |
| "epoch": 0.006758099781355596, |
| "grad_norm": 0.3347357213497162, |
| "learning_rate": 2.8081442660546125e-05, |
| "loss": 3.1765, |
| "step": 34 |
| }, |
| { |
| "epoch": 0.006956867421983701, |
| "grad_norm": 0.30826708674430847, |
| "learning_rate": 2.500000000000001e-05, |
| "loss": 3.223, |
| "step": 35 |
| }, |
| { |
| "epoch": 0.007155635062611807, |
| "grad_norm": 0.32336822152137756, |
| "learning_rate": 2.2040354826462668e-05, |
| "loss": 3.3838, |
| "step": 36 |
| }, |
| { |
| "epoch": 0.007354402703239913, |
| "grad_norm": 0.3186890482902527, |
| "learning_rate": 1.9216926233717085e-05, |
| "loss": 3.1607, |
| "step": 37 |
| }, |
| { |
| "epoch": 0.007553170343868018, |
| "grad_norm": 0.330779105424881, |
| "learning_rate": 1.6543469682057106e-05, |
| "loss": 3.2356, |
| "step": 38 |
| }, |
| { |
| "epoch": 0.007751937984496124, |
| "grad_norm": 0.3563855290412903, |
| "learning_rate": 1.4033009983067452e-05, |
| "loss": 3.1177, |
| "step": 39 |
| }, |
| { |
| "epoch": 0.007950705625124229, |
| "grad_norm": 0.36722344160079956, |
| "learning_rate": 1.1697777844051105e-05, |
| "loss": 3.0539, |
| "step": 40 |
| }, |
| { |
| "epoch": 0.008149473265752336, |
| "grad_norm": 0.3383886516094208, |
| "learning_rate": 9.549150281252633e-06, |
| "loss": 3.2582, |
| "step": 41 |
| }, |
| { |
| "epoch": 0.008348240906380441, |
| "grad_norm": 0.35494890809059143, |
| "learning_rate": 7.597595192178702e-06, |
| "loss": 3.0954, |
| "step": 42 |
| }, |
| { |
| "epoch": 0.008547008547008548, |
| "grad_norm": 0.3515838086605072, |
| "learning_rate": 5.852620357053651e-06, |
| "loss": 3.3631, |
| "step": 43 |
| }, |
| { |
| "epoch": 0.008745776187636653, |
| "grad_norm": 0.37979280948638916, |
| "learning_rate": 4.322727117869951e-06, |
| "loss": 3.3569, |
| "step": 44 |
| }, |
| { |
| "epoch": 0.008944543828264758, |
| "grad_norm": 0.3924817144870758, |
| "learning_rate": 3.0153689607045845e-06, |
| "loss": 3.2494, |
| "step": 45 |
| }, |
| { |
| "epoch": 0.009143311468892865, |
| "grad_norm": 0.39909616112709045, |
| "learning_rate": 1.9369152030840556e-06, |
| "loss": 3.3279, |
| "step": 46 |
| }, |
| { |
| "epoch": 0.00934207910952097, |
| "grad_norm": 0.46893322467803955, |
| "learning_rate": 1.0926199633097157e-06, |
| "loss": 3.1981, |
| "step": 47 |
| }, |
| { |
| "epoch": 0.009540846750149075, |
| "grad_norm": 0.4818413555622101, |
| "learning_rate": 4.865965629214819e-07, |
| "loss": 3.4108, |
| "step": 48 |
| }, |
| { |
| "epoch": 0.009739614390777182, |
| "grad_norm": 0.6217837333679199, |
| "learning_rate": 1.2179748700879012e-07, |
| "loss": 3.674, |
| "step": 49 |
| }, |
| { |
| "epoch": 0.009938382031405287, |
| "grad_norm": 0.9694095253944397, |
| "learning_rate": 0.0, |
| "loss": 4.1192, |
| "step": 50 |
| }, |
| { |
| "epoch": 0.009938382031405287, |
| "eval_loss": 3.0149381160736084, |
| "eval_runtime": 67.076, |
| "eval_samples_per_second": 126.334, |
| "eval_steps_per_second": 15.803, |
| "step": 50 |
| } |
| ], |
| "logging_steps": 1, |
| "max_steps": 50, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 1, |
| "save_steps": 25, |
| "stateful_callbacks": { |
| "EarlyStoppingCallback": { |
| "args": { |
| "early_stopping_patience": 5, |
| "early_stopping_threshold": 0.0 |
| }, |
| "attributes": { |
| "early_stopping_patience_counter": 0 |
| } |
| }, |
| "TrainerControl": { |
| "args": { |
| "should_epoch_stop": false, |
| "should_evaluate": false, |
| "should_log": false, |
| "should_save": true, |
| "should_training_stop": true |
| }, |
| "attributes": {} |
| } |
| }, |
| "total_flos": 1725805984481280.0, |
| "train_batch_size": 8, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
