| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.0026274073619954284, | |
| "eval_steps": 25, | |
| "global_step": 75, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 3.5032098159939045e-05, | |
| "grad_norm": 8.220430374145508, | |
| "learning_rate": 3.3333333333333335e-05, | |
| "loss": 16.7863, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 3.5032098159939045e-05, | |
| "eval_loss": 2.912421226501465, | |
| "eval_runtime": 3850.8422, | |
| "eval_samples_per_second": 6.243, | |
| "eval_steps_per_second": 3.121, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 7.006419631987809e-05, | |
| "grad_norm": 10.870444297790527, | |
| "learning_rate": 6.666666666666667e-05, | |
| "loss": 16.9888, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.00010509629447981713, | |
| "grad_norm": 11.990473747253418, | |
| "learning_rate": 0.0001, | |
| "loss": 19.7956, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.00014012839263975618, | |
| "grad_norm": 16.78184700012207, | |
| "learning_rate": 9.99524110790929e-05, | |
| "loss": 19.0291, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.00017516049079969523, | |
| "grad_norm": 14.768333435058594, | |
| "learning_rate": 9.980973490458728e-05, | |
| "loss": 18.0519, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.00021019258895963426, | |
| "grad_norm": 14.792781829833984, | |
| "learning_rate": 9.957224306869053e-05, | |
| "loss": 15.9296, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.0002452246871195733, | |
| "grad_norm": 13.751137733459473, | |
| "learning_rate": 9.924038765061042e-05, | |
| "loss": 21.8163, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.00028025678527951236, | |
| "grad_norm": 11.666454315185547, | |
| "learning_rate": 9.881480035599667e-05, | |
| "loss": 17.2661, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.0003152888834394514, | |
| "grad_norm": 13.513507843017578, | |
| "learning_rate": 9.829629131445342e-05, | |
| "loss": 17.5924, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.00035032098159939047, | |
| "grad_norm": 13.285558700561523, | |
| "learning_rate": 9.768584753741134e-05, | |
| "loss": 16.776, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.00038535307975932947, | |
| "grad_norm": 14.108022689819336, | |
| "learning_rate": 9.698463103929542e-05, | |
| "loss": 16.9609, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.0004203851779192685, | |
| "grad_norm": 12.603458404541016, | |
| "learning_rate": 9.619397662556435e-05, | |
| "loss": 14.9006, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.00045541727607920757, | |
| "grad_norm": 14.595187187194824, | |
| "learning_rate": 9.53153893518325e-05, | |
| "loss": 17.0331, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.0004904493742391466, | |
| "grad_norm": 13.04394817352295, | |
| "learning_rate": 9.435054165891109e-05, | |
| "loss": 15.4006, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.0005254814723990856, | |
| "grad_norm": 15.335989952087402, | |
| "learning_rate": 9.330127018922194e-05, | |
| "loss": 14.6069, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.0005605135705590247, | |
| "grad_norm": 17.090463638305664, | |
| "learning_rate": 9.21695722906443e-05, | |
| "loss": 17.1589, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.0005955456687189637, | |
| "grad_norm": 14.205001831054688, | |
| "learning_rate": 9.09576022144496e-05, | |
| "loss": 15.1409, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.0006305777668789028, | |
| "grad_norm": 14.655081748962402, | |
| "learning_rate": 8.966766701456177e-05, | |
| "loss": 15.1379, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.0006656098650388418, | |
| "grad_norm": 12.469857215881348, | |
| "learning_rate": 8.83022221559489e-05, | |
| "loss": 13.5586, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.0007006419631987809, | |
| "grad_norm": 25.291624069213867, | |
| "learning_rate": 8.68638668405062e-05, | |
| "loss": 14.093, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.0007356740613587199, | |
| "grad_norm": 16.53830337524414, | |
| "learning_rate": 8.535533905932738e-05, | |
| "loss": 16.5948, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.0007707061595186589, | |
| "grad_norm": 14.183712005615234, | |
| "learning_rate": 8.377951038078302e-05, | |
| "loss": 14.1989, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.000805738257678598, | |
| "grad_norm": 20.805938720703125, | |
| "learning_rate": 8.213938048432697e-05, | |
| "loss": 15.1565, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.000840770355838537, | |
| "grad_norm": 13.752044677734375, | |
| "learning_rate": 8.043807145043604e-05, | |
| "loss": 14.4959, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.0008758024539984761, | |
| "grad_norm": 15.55850601196289, | |
| "learning_rate": 7.86788218175523e-05, | |
| "loss": 15.3995, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.0008758024539984761, | |
| "eval_loss": 1.7509880065917969, | |
| "eval_runtime": 3845.137, | |
| "eval_samples_per_second": 6.252, | |
| "eval_steps_per_second": 3.126, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.0009108345521584151, | |
| "grad_norm": 12.968778610229492, | |
| "learning_rate": 7.68649804173412e-05, | |
| "loss": 13.9149, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.0009458666503183541, | |
| "grad_norm": 18.30549430847168, | |
| "learning_rate": 7.500000000000001e-05, | |
| "loss": 14.3664, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.0009808987484782932, | |
| "grad_norm": 10.476753234863281, | |
| "learning_rate": 7.308743066175172e-05, | |
| "loss": 13.5992, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.0010159308466382324, | |
| "grad_norm": 13.235085487365723, | |
| "learning_rate": 7.113091308703498e-05, | |
| "loss": 14.0461, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.0010509629447981712, | |
| "grad_norm": 12.34658432006836, | |
| "learning_rate": 6.91341716182545e-05, | |
| "loss": 12.8937, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.0010859950429581103, | |
| "grad_norm": 10.956889152526855, | |
| "learning_rate": 6.710100716628344e-05, | |
| "loss": 12.5274, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.0011210271411180495, | |
| "grad_norm": 11.746081352233887, | |
| "learning_rate": 6.503528997521366e-05, | |
| "loss": 12.4693, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.0011560592392779886, | |
| "grad_norm": 15.51623821258545, | |
| "learning_rate": 6.294095225512603e-05, | |
| "loss": 12.3992, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.0011910913374379275, | |
| "grad_norm": 12.37278938293457, | |
| "learning_rate": 6.0821980696905146e-05, | |
| "loss": 12.1944, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.0012261234355978666, | |
| "grad_norm": 10.765904426574707, | |
| "learning_rate": 5.868240888334653e-05, | |
| "loss": 12.5313, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.0012611555337578057, | |
| "grad_norm": 14.782493591308594, | |
| "learning_rate": 5.6526309611002594e-05, | |
| "loss": 15.8179, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.0012961876319177446, | |
| "grad_norm": 13.46235466003418, | |
| "learning_rate": 5.435778713738292e-05, | |
| "loss": 14.1264, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.0013312197300776837, | |
| "grad_norm": 11.297664642333984, | |
| "learning_rate": 5.218096936826681e-05, | |
| "loss": 13.6005, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.0013662518282376228, | |
| "grad_norm": 11.96649169921875, | |
| "learning_rate": 5e-05, | |
| "loss": 10.5872, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.0014012839263975619, | |
| "grad_norm": 13.083653450012207, | |
| "learning_rate": 4.781903063173321e-05, | |
| "loss": 12.7705, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.0014363160245575008, | |
| "grad_norm": 12.904142379760742, | |
| "learning_rate": 4.564221286261709e-05, | |
| "loss": 15.0502, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.0014713481227174399, | |
| "grad_norm": 13.312074661254883, | |
| "learning_rate": 4.347369038899744e-05, | |
| "loss": 11.7917, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.001506380220877379, | |
| "grad_norm": 15.562873840332031, | |
| "learning_rate": 4.131759111665349e-05, | |
| "loss": 13.647, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.0015414123190373179, | |
| "grad_norm": 15.31185531616211, | |
| "learning_rate": 3.917801930309486e-05, | |
| "loss": 15.8205, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.001576444417197257, | |
| "grad_norm": 17.315444946289062, | |
| "learning_rate": 3.705904774487396e-05, | |
| "loss": 17.6123, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.001611476515357196, | |
| "grad_norm": 13.117803573608398, | |
| "learning_rate": 3.4964710024786354e-05, | |
| "loss": 12.6254, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.001646508613517135, | |
| "grad_norm": 15.708817481994629, | |
| "learning_rate": 3.289899283371657e-05, | |
| "loss": 14.3401, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.001681540711677074, | |
| "grad_norm": 22.316585540771484, | |
| "learning_rate": 3.086582838174551e-05, | |
| "loss": 12.2662, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.0017165728098370132, | |
| "grad_norm": 24.079978942871094, | |
| "learning_rate": 2.886908691296504e-05, | |
| "loss": 15.551, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.0017516049079969523, | |
| "grad_norm": 44.766876220703125, | |
| "learning_rate": 2.6912569338248315e-05, | |
| "loss": 16.2846, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.0017516049079969523, | |
| "eval_loss": 1.6892892122268677, | |
| "eval_runtime": 3848.953, | |
| "eval_samples_per_second": 6.246, | |
| "eval_steps_per_second": 3.123, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.0017866370061568912, | |
| "grad_norm": 11.697985649108887, | |
| "learning_rate": 2.500000000000001e-05, | |
| "loss": 10.0276, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.0018216691043168303, | |
| "grad_norm": 13.977499961853027, | |
| "learning_rate": 2.3135019582658802e-05, | |
| "loss": 12.4535, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.0018567012024767694, | |
| "grad_norm": 14.096190452575684, | |
| "learning_rate": 2.132117818244771e-05, | |
| "loss": 12.3367, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.0018917333006367083, | |
| "grad_norm": 13.248291969299316, | |
| "learning_rate": 1.9561928549563968e-05, | |
| "loss": 10.8346, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.0019267653987966474, | |
| "grad_norm": 16.24111557006836, | |
| "learning_rate": 1.7860619515673033e-05, | |
| "loss": 12.1668, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.0019617974969565865, | |
| "grad_norm": 16.580856323242188, | |
| "learning_rate": 1.622048961921699e-05, | |
| "loss": 11.8119, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.0019968295951165254, | |
| "grad_norm": 11.197458267211914, | |
| "learning_rate": 1.4644660940672627e-05, | |
| "loss": 11.7317, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.0020318616932764647, | |
| "grad_norm": 12.0034761428833, | |
| "learning_rate": 1.3136133159493802e-05, | |
| "loss": 15.88, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.0020668937914364036, | |
| "grad_norm": 11.464460372924805, | |
| "learning_rate": 1.1697777844051105e-05, | |
| "loss": 14.5566, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.0021019258895963425, | |
| "grad_norm": 9.426189422607422, | |
| "learning_rate": 1.0332332985438248e-05, | |
| "loss": 11.9118, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.002136957987756282, | |
| "grad_norm": 11.631290435791016, | |
| "learning_rate": 9.042397785550405e-06, | |
| "loss": 13.0088, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.0021719900859162207, | |
| "grad_norm": 9.637537956237793, | |
| "learning_rate": 7.830427709355725e-06, | |
| "loss": 12.759, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.0022070221840761596, | |
| "grad_norm": 11.821331024169922, | |
| "learning_rate": 6.698729810778065e-06, | |
| "loss": 13.3715, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.002242054282236099, | |
| "grad_norm": 12.932907104492188, | |
| "learning_rate": 5.649458341088915e-06, | |
| "loss": 16.1093, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.002277086380396038, | |
| "grad_norm": 9.686279296875, | |
| "learning_rate": 4.684610648167503e-06, | |
| "loss": 12.3969, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.002312118478555977, | |
| "grad_norm": 11.873120307922363, | |
| "learning_rate": 3.8060233744356633e-06, | |
| "loss": 11.2338, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.002347150576715916, | |
| "grad_norm": 10.407732009887695, | |
| "learning_rate": 3.0153689607045845e-06, | |
| "loss": 14.9716, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.002382182674875855, | |
| "grad_norm": 12.169331550598145, | |
| "learning_rate": 2.314152462588659e-06, | |
| "loss": 14.9951, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.0024172147730357942, | |
| "grad_norm": 14.04606819152832, | |
| "learning_rate": 1.70370868554659e-06, | |
| "loss": 15.8657, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.002452246871195733, | |
| "grad_norm": 9.512072563171387, | |
| "learning_rate": 1.1851996440033319e-06, | |
| "loss": 13.8743, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.002487278969355672, | |
| "grad_norm": 10.393471717834473, | |
| "learning_rate": 7.596123493895991e-07, | |
| "loss": 12.4094, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.0025223110675156113, | |
| "grad_norm": 10.671348571777344, | |
| "learning_rate": 4.277569313094809e-07, | |
| "loss": 12.0238, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.0025573431656755502, | |
| "grad_norm": 11.587846755981445, | |
| "learning_rate": 1.9026509541272275e-07, | |
| "loss": 13.5365, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.002592375263835489, | |
| "grad_norm": 9.723920822143555, | |
| "learning_rate": 4.7588920907110094e-08, | |
| "loss": 12.6632, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.0026274073619954284, | |
| "grad_norm": 10.614508628845215, | |
| "learning_rate": 0.0, | |
| "loss": 11.4377, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.0026274073619954284, | |
| "eval_loss": 1.660986304283142, | |
| "eval_runtime": 3847.2283, | |
| "eval_samples_per_second": 6.248, | |
| "eval_steps_per_second": 3.124, | |
| "step": 75 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 75, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 25, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 1.91356572008448e+17, | |
| "train_batch_size": 2, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |