| { | |
| "best_metric": 0.1021631509065628, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-50", | |
| "epoch": 0.024468573176326655, | |
| "eval_steps": 25, | |
| "global_step": 50, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0004893714635265331, | |
| "grad_norm": 258.2685241699219, | |
| "learning_rate": 5e-05, | |
| "loss": 36.2057, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0004893714635265331, | |
| "eval_loss": 0.9609287977218628, | |
| "eval_runtime": 4.441, | |
| "eval_samples_per_second": 11.259, | |
| "eval_steps_per_second": 2.927, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0009787429270530662, | |
| "grad_norm": 263.50335693359375, | |
| "learning_rate": 0.0001, | |
| "loss": 28.2897, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0014681143905795992, | |
| "grad_norm": 136.62240600585938, | |
| "learning_rate": 9.990365154573717e-05, | |
| "loss": 17.9301, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.0019574858541061323, | |
| "grad_norm": 87.69647216796875, | |
| "learning_rate": 9.961501876182148e-05, | |
| "loss": 11.9713, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.0024468573176326656, | |
| "grad_norm": 67.17732238769531, | |
| "learning_rate": 9.913533761814537e-05, | |
| "loss": 9.9737, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.0029362287811591985, | |
| "grad_norm": 48.90814208984375, | |
| "learning_rate": 9.846666218300807e-05, | |
| "loss": 6.991, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.0034256002446857317, | |
| "grad_norm": 41.13584518432617, | |
| "learning_rate": 9.761185582727977e-05, | |
| "loss": 6.1007, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.003914971708212265, | |
| "grad_norm": 33.86528015136719, | |
| "learning_rate": 9.657457896300791e-05, | |
| "loss": 4.6095, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.004404343171738798, | |
| "grad_norm": 31.326583862304688, | |
| "learning_rate": 9.535927336897098e-05, | |
| "loss": 4.5279, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.004893714635265331, | |
| "grad_norm": 27.528942108154297, | |
| "learning_rate": 9.397114317029975e-05, | |
| "loss": 4.5319, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.005383086098791864, | |
| "grad_norm": 24.476354598999023, | |
| "learning_rate": 9.241613255361455e-05, | |
| "loss": 3.7506, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.005872457562318397, | |
| "grad_norm": 26.610166549682617, | |
| "learning_rate": 9.070090031310558e-05, | |
| "loss": 4.0902, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.006361829025844931, | |
| "grad_norm": 88.37236785888672, | |
| "learning_rate": 8.883279133655399e-05, | |
| "loss": 20.2102, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.0068512004893714635, | |
| "grad_norm": 98.64079284667969, | |
| "learning_rate": 8.681980515339464e-05, | |
| "loss": 18.9077, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.007340571952897996, | |
| "grad_norm": 52.15375900268555, | |
| "learning_rate": 8.467056167950311e-05, | |
| "loss": 9.0637, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.00782994341642453, | |
| "grad_norm": 38.73460006713867, | |
| "learning_rate": 8.239426430539243e-05, | |
| "loss": 6.5561, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.008319314879951063, | |
| "grad_norm": 43.262699127197266, | |
| "learning_rate": 8.000066048588211e-05, | |
| "loss": 7.039, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.008808686343477597, | |
| "grad_norm": 45.789329528808594, | |
| "learning_rate": 7.75e-05, | |
| "loss": 5.3116, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.009298057807004129, | |
| "grad_norm": 26.616962432861328, | |
| "learning_rate": 7.490299105985507e-05, | |
| "loss": 3.8364, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.009787429270530662, | |
| "grad_norm": 22.72823715209961, | |
| "learning_rate": 7.222075445642904e-05, | |
| "loss": 3.9739, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.010276800734057196, | |
| "grad_norm": 31.525386810302734, | |
| "learning_rate": 6.946477593864228e-05, | |
| "loss": 4.1897, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.010766172197583728, | |
| "grad_norm": 33.8955078125, | |
| "learning_rate": 6.664685702961344e-05, | |
| "loss": 3.9558, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.011255543661110262, | |
| "grad_norm": 38.87385177612305, | |
| "learning_rate": 6.377906449072578e-05, | |
| "loss": 3.0537, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.011744915124636794, | |
| "grad_norm": 20.20720100402832, | |
| "learning_rate": 6.087367864990233e-05, | |
| "loss": 3.2239, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.012234286588163328, | |
| "grad_norm": 21.866695404052734, | |
| "learning_rate": 5.794314081535644e-05, | |
| "loss": 2.2741, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.012234286588163328, | |
| "eval_loss": 0.1825588047504425, | |
| "eval_runtime": 4.5208, | |
| "eval_samples_per_second": 11.06, | |
| "eval_steps_per_second": 2.876, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.012723658051689861, | |
| "grad_norm": 79.81653594970703, | |
| "learning_rate": 5.500000000000001e-05, | |
| "loss": 23.5249, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.013213029515216393, | |
| "grad_norm": 67.56028747558594, | |
| "learning_rate": 5.205685918464356e-05, | |
| "loss": 11.0355, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.013702400978742927, | |
| "grad_norm": 39.948997497558594, | |
| "learning_rate": 4.912632135009769e-05, | |
| "loss": 5.2478, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.01419177244226946, | |
| "grad_norm": 26.83481788635254, | |
| "learning_rate": 4.6220935509274235e-05, | |
| "loss": 4.9199, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.014681143905795993, | |
| "grad_norm": 45.867279052734375, | |
| "learning_rate": 4.3353142970386564e-05, | |
| "loss": 4.4869, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.015170515369322526, | |
| "grad_norm": 46.26245880126953, | |
| "learning_rate": 4.053522406135775e-05, | |
| "loss": 4.9944, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.01565988683284906, | |
| "grad_norm": 29.81333351135254, | |
| "learning_rate": 3.777924554357096e-05, | |
| "loss": 3.8116, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.016149258296375594, | |
| "grad_norm": 23.598108291625977, | |
| "learning_rate": 3.509700894014496e-05, | |
| "loss": 2.9508, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.016638629759902126, | |
| "grad_norm": 17.001190185546875, | |
| "learning_rate": 3.250000000000001e-05, | |
| "loss": 3.2026, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.017128001223428658, | |
| "grad_norm": 14.1507568359375, | |
| "learning_rate": 2.9999339514117912e-05, | |
| "loss": 2.616, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.017617372686955193, | |
| "grad_norm": 15.094809532165527, | |
| "learning_rate": 2.760573569460757e-05, | |
| "loss": 2.635, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.018106744150481725, | |
| "grad_norm": 15.910400390625, | |
| "learning_rate": 2.53294383204969e-05, | |
| "loss": 2.3422, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.018596115614008257, | |
| "grad_norm": 30.014652252197266, | |
| "learning_rate": 2.3180194846605367e-05, | |
| "loss": 13.5659, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.019085487077534793, | |
| "grad_norm": 55.55118942260742, | |
| "learning_rate": 2.1167208663446025e-05, | |
| "loss": 9.5593, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.019574858541061325, | |
| "grad_norm": 25.363740921020508, | |
| "learning_rate": 1.9299099686894423e-05, | |
| "loss": 4.5981, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.020064230004587857, | |
| "grad_norm": 32.401058197021484, | |
| "learning_rate": 1.758386744638546e-05, | |
| "loss": 3.7437, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.020553601468114392, | |
| "grad_norm": 18.27694320678711, | |
| "learning_rate": 1.602885682970026e-05, | |
| "loss": 3.1648, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.021042972931640924, | |
| "grad_norm": 27.42363739013672, | |
| "learning_rate": 1.464072663102903e-05, | |
| "loss": 2.999, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.021532344395167456, | |
| "grad_norm": 24.510116577148438, | |
| "learning_rate": 1.3425421036992098e-05, | |
| "loss": 2.7106, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.022021715858693988, | |
| "grad_norm": 17.725000381469727, | |
| "learning_rate": 1.2388144172720251e-05, | |
| "loss": 2.29, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.022511087322220524, | |
| "grad_norm": 14.812396049499512, | |
| "learning_rate": 1.1533337816991932e-05, | |
| "loss": 2.5546, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.023000458785747056, | |
| "grad_norm": 15.039092063903809, | |
| "learning_rate": 1.0864662381854632e-05, | |
| "loss": 1.9962, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.023489830249273588, | |
| "grad_norm": 14.523675918579102, | |
| "learning_rate": 1.0384981238178534e-05, | |
| "loss": 1.8707, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.023979201712800123, | |
| "grad_norm": 12.050509452819824, | |
| "learning_rate": 1.0096348454262845e-05, | |
| "loss": 1.9962, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.024468573176326655, | |
| "grad_norm": 13.82956600189209, | |
| "learning_rate": 1e-05, | |
| "loss": 1.8693, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.024468573176326655, | |
| "eval_loss": 0.1021631509065628, | |
| "eval_runtime": 4.4908, | |
| "eval_samples_per_second": 11.134, | |
| "eval_steps_per_second": 2.895, | |
| "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": 1, | |
| "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": 5.6659811474866176e+17, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |