| { |
| "best_global_step": null, |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 7.0, |
| "eval_steps": 500, |
| "global_step": 231, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.15151515151515152, |
| "grad_norm": 7.071829850114283, |
| "learning_rate": 6.666666666666667e-06, |
| "loss": 0.4885, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.15010309219360352, |
| "step": 5, |
| "valid_targets_mean": 8453.9, |
| "valid_targets_min": 1970 |
| }, |
| { |
| "epoch": 0.30303030303030304, |
| "grad_norm": 1.9804373529490895, |
| "learning_rate": 1.5000000000000002e-05, |
| "loss": 0.4116, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.11914314329624176, |
| "step": 10, |
| "valid_targets_mean": 8011.8, |
| "valid_targets_min": 1960 |
| }, |
| { |
| "epoch": 0.45454545454545453, |
| "grad_norm": 0.6412351971777132, |
| "learning_rate": 2.3333333333333336e-05, |
| "loss": 0.3555, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.10543923079967499, |
| "step": 15, |
| "valid_targets_mean": 8846.7, |
| "valid_targets_min": 4611 |
| }, |
| { |
| "epoch": 0.6060606060606061, |
| "grad_norm": 0.5392085970382771, |
| "learning_rate": 3.1666666666666666e-05, |
| "loss": 0.3169, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.10168177634477615, |
| "step": 20, |
| "valid_targets_mean": 9232.6, |
| "valid_targets_min": 3741 |
| }, |
| { |
| "epoch": 0.7575757575757576, |
| "grad_norm": 0.3582208500451154, |
| "learning_rate": 4e-05, |
| "loss": 0.2743, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.09165775775909424, |
| "step": 25, |
| "valid_targets_mean": 8923.8, |
| "valid_targets_min": 3026 |
| }, |
| { |
| "epoch": 0.9090909090909091, |
| "grad_norm": 0.26631192084809374, |
| "learning_rate": 3.994244399375679e-05, |
| "loss": 0.2516, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.08204009383916855, |
| "step": 30, |
| "valid_targets_mean": 9076.3, |
| "valid_targets_min": 4099 |
| }, |
| { |
| "epoch": 1.0606060606060606, |
| "grad_norm": 0.22624905586056185, |
| "learning_rate": 3.977010724441261e-05, |
| "loss": 0.2285, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.06856784224510193, |
| "step": 35, |
| "valid_targets_mean": 7689.8, |
| "valid_targets_min": 3519 |
| }, |
| { |
| "epoch": 1.2121212121212122, |
| "grad_norm": 0.19597958600025933, |
| "learning_rate": 3.9483981653469586e-05, |
| "loss": 0.2195, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.06643202900886536, |
| "step": 40, |
| "valid_targets_mean": 8430.5, |
| "valid_targets_min": 2426 |
| }, |
| { |
| "epoch": 1.3636363636363638, |
| "grad_norm": 0.18547010915072298, |
| "learning_rate": 3.908571404555758e-05, |
| "loss": 0.2076, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.06439946591854095, |
| "step": 45, |
| "valid_targets_mean": 9294.1, |
| "valid_targets_min": 1312 |
| }, |
| { |
| "epoch": 1.5151515151515151, |
| "grad_norm": 0.17502242360460948, |
| "learning_rate": 3.8577596689969346e-05, |
| "loss": 0.194, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.06616371870040894, |
| "step": 50, |
| "valid_targets_mean": 8604.2, |
| "valid_targets_min": 2696 |
| }, |
| { |
| "epoch": 1.6666666666666665, |
| "grad_norm": 0.15763655671676294, |
| "learning_rate": 3.7962554107273926e-05, |
| "loss": 0.1975, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.05964214354753494, |
| "step": 55, |
| "valid_targets_mean": 8201.2, |
| "valid_targets_min": 2157 |
| }, |
| { |
| "epoch": 1.8181818181818183, |
| "grad_norm": 0.15305317018363324, |
| "learning_rate": 3.724412623694427e-05, |
| "loss": 0.1887, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.05983173847198486, |
| "step": 60, |
| "valid_targets_mean": 8784.1, |
| "valid_targets_min": 2568 |
| }, |
| { |
| "epoch": 1.9696969696969697, |
| "grad_norm": 0.15685613419616962, |
| "learning_rate": 3.642644806287938e-05, |
| "loss": 0.1842, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.06401412934064865, |
| "step": 65, |
| "valid_targets_mean": 9732.0, |
| "valid_targets_min": 3691 |
| }, |
| { |
| "epoch": 2.121212121212121, |
| "grad_norm": 0.1710095040672745, |
| "learning_rate": 3.55142258140884e-05, |
| "loss": 0.1774, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.05440504476428032, |
| "step": 70, |
| "valid_targets_mean": 8249.6, |
| "valid_targets_min": 3584 |
| }, |
| { |
| "epoch": 2.2727272727272725, |
| "grad_norm": 0.1775001328477045, |
| "learning_rate": 3.451270987751598e-05, |
| "loss": 0.1751, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.05906001478433609, |
| "step": 75, |
| "valid_targets_mean": 8529.4, |
| "valid_targets_min": 3214 |
| }, |
| { |
| "epoch": 2.4242424242424243, |
| "grad_norm": 0.15374353835337343, |
| "learning_rate": 3.342766457891194e-05, |
| "loss": 0.1732, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.06226389855146408, |
| "step": 80, |
| "valid_targets_mean": 9319.1, |
| "valid_targets_min": 3163 |
| }, |
| { |
| "epoch": 2.5757575757575757, |
| "grad_norm": 0.16344244686978898, |
| "learning_rate": 3.226533500567433e-05, |
| "loss": 0.1768, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.06373937427997589, |
| "step": 85, |
| "valid_targets_mean": 8842.0, |
| "valid_targets_min": 4024 |
| }, |
| { |
| "epoch": 2.7272727272727275, |
| "grad_norm": 0.1590870643550257, |
| "learning_rate": 3.1032411062620544e-05, |
| "loss": 0.1675, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.05495002493262291, |
| "step": 90, |
| "valid_targets_mean": 9056.3, |
| "valid_targets_min": 2998 |
| }, |
| { |
| "epoch": 2.878787878787879, |
| "grad_norm": 0.1572481613873198, |
| "learning_rate": 2.973598896756697e-05, |
| "loss": 0.1658, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.05916336178779602, |
| "step": 95, |
| "valid_targets_mean": 9378.6, |
| "valid_targets_min": 4128 |
| }, |
| { |
| "epoch": 3.0303030303030303, |
| "grad_norm": 0.16522830538286623, |
| "learning_rate": 2.8383530408333285e-05, |
| "loss": 0.1705, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.06000653654336929, |
| "step": 100, |
| "valid_targets_mean": 9399.1, |
| "valid_targets_min": 4429 |
| }, |
| { |
| "epoch": 3.1818181818181817, |
| "grad_norm": 0.15664808473003858, |
| "learning_rate": 2.6982819596247373e-05, |
| "loss": 0.1662, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.051601044833660126, |
| "step": 105, |
| "valid_targets_mean": 8927.5, |
| "valid_targets_min": 3117 |
| }, |
| { |
| "epoch": 3.3333333333333335, |
| "grad_norm": 0.15996355844466018, |
| "learning_rate": 2.554191846333378e-05, |
| "loss": 0.1626, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.054302021861076355, |
| "step": 110, |
| "valid_targets_mean": 8787.9, |
| "valid_targets_min": 3691 |
| }, |
| { |
| "epoch": 3.484848484848485, |
| "grad_norm": 0.15935216780397668, |
| "learning_rate": 2.4069120261052682e-05, |
| "loss": 0.1571, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.047706879675388336, |
| "step": 115, |
| "valid_targets_mean": 7594.4, |
| "valid_targets_min": 1960 |
| }, |
| { |
| "epoch": 3.6363636363636362, |
| "grad_norm": 0.18279188836268426, |
| "learning_rate": 2.2572901827656626e-05, |
| "loss": 0.1568, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.05731729045510292, |
| "step": 120, |
| "valid_targets_mean": 8903.4, |
| "valid_targets_min": 1670 |
| }, |
| { |
| "epoch": 3.787878787878788, |
| "grad_norm": 0.16897393812957626, |
| "learning_rate": 2.1061874798894992e-05, |
| "loss": 0.1595, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.05671117454767227, |
| "step": 125, |
| "valid_targets_mean": 8410.8, |
| "valid_targets_min": 3052 |
| }, |
| { |
| "epoch": 3.9393939393939394, |
| "grad_norm": 0.16560200225340949, |
| "learning_rate": 1.9544736042877886e-05, |
| "loss": 0.1608, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.0500858873128891, |
| "step": 130, |
| "valid_targets_mean": 8883.9, |
| "valid_targets_min": 2364 |
| }, |
| { |
| "epoch": 4.090909090909091, |
| "grad_norm": 0.17765486125420513, |
| "learning_rate": 1.8030217604376628e-05, |
| "loss": 0.1568, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.05107717961072922, |
| "step": 135, |
| "valid_targets_mean": 7918.4, |
| "valid_targets_min": 2880 |
| }, |
| { |
| "epoch": 4.242424242424242, |
| "grad_norm": 0.16121021939384844, |
| "learning_rate": 1.6527036446661396e-05, |
| "loss": 0.1572, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.056909848004579544, |
| "step": 140, |
| "valid_targets_mean": 8202.9, |
| "valid_targets_min": 2989 |
| }, |
| { |
| "epoch": 4.393939393939394, |
| "grad_norm": 0.15780609040060248, |
| "learning_rate": 1.5043844280142005e-05, |
| "loss": 0.1516, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.05371875315904617, |
| "step": 145, |
| "valid_targets_mean": 9110.0, |
| "valid_targets_min": 2450 |
| }, |
| { |
| "epoch": 4.545454545454545, |
| "grad_norm": 0.16910637578622967, |
| "learning_rate": 1.358917776657806e-05, |
| "loss": 0.1537, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.0523504763841629, |
| "step": 150, |
| "valid_targets_mean": 8569.7, |
| "valid_targets_min": 3473 |
| }, |
| { |
| "epoch": 4.696969696969697, |
| "grad_norm": 0.15315713301670586, |
| "learning_rate": 1.2171409385463218e-05, |
| "loss": 0.1539, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.051369499415159225, |
| "step": 155, |
| "valid_targets_mean": 9342.3, |
| "valid_targets_min": 4669 |
| }, |
| { |
| "epoch": 4.848484848484849, |
| "grad_norm": 0.15519091165573462, |
| "learning_rate": 1.0798699245376959e-05, |
| "loss": 0.1537, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.05453380197286606, |
| "step": 160, |
| "valid_targets_mean": 8980.6, |
| "valid_targets_min": 2311 |
| }, |
| { |
| "epoch": 5.0, |
| "grad_norm": 0.15338125297734576, |
| "learning_rate": 9.478948117658577e-06, |
| "loss": 0.152, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.048132434487342834, |
| "step": 165, |
| "valid_targets_mean": 9222.9, |
| "valid_targets_min": 3021 |
| }, |
| { |
| "epoch": 5.151515151515151, |
| "grad_norm": 0.15963497291261625, |
| "learning_rate": 8.219751962722726e-06, |
| "loss": 0.1513, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.05217127129435539, |
| "step": 170, |
| "valid_targets_mean": 8096.7, |
| "valid_targets_min": 1663 |
| }, |
| { |
| "epoch": 5.303030303030303, |
| "grad_norm": 0.1701013399219855, |
| "learning_rate": 7.028358210744881e-06, |
| "loss": 0.1483, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.04974319785833359, |
| "step": 175, |
| "valid_targets_mean": 8601.6, |
| "valid_targets_min": 3238 |
| }, |
| { |
| "epoch": 5.454545454545454, |
| "grad_norm": 0.15824877992240952, |
| "learning_rate": 5.911624048347757e-06, |
| "loss": 0.1515, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.054281268268823624, |
| "step": 180, |
| "valid_targets_mean": 9436.0, |
| "valid_targets_min": 3369 |
| }, |
| { |
| "epoch": 5.606060606060606, |
| "grad_norm": 0.16388221222616883, |
| "learning_rate": 4.875976951373633e-06, |
| "loss": 0.1498, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.04890450835227966, |
| "step": 185, |
| "valid_targets_mean": 8094.2, |
| "valid_targets_min": 2998 |
| }, |
| { |
| "epoch": 5.757575757575758, |
| "grad_norm": 0.15237739649000462, |
| "learning_rate": 3.927377690900436e-06, |
| "loss": 0.1468, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.04883284121751785, |
| "step": 190, |
| "valid_targets_mean": 9055.8, |
| "valid_targets_min": 4367 |
| }, |
| { |
| "epoch": 5.909090909090909, |
| "grad_norm": 0.1678247264659432, |
| "learning_rate": 3.071286025423983e-06, |
| "loss": 0.1513, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.050182901322841644, |
| "step": 195, |
| "valid_targets_mean": 8657.7, |
| "valid_targets_min": 2063 |
| }, |
| { |
| "epoch": 6.0606060606060606, |
| "grad_norm": 0.1550783342570967, |
| "learning_rate": 2.312629276668554e-06, |
| "loss": 0.1532, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.047827623784542084, |
| "step": 200, |
| "valid_targets_mean": 8015.0, |
| "valid_targets_min": 2224 |
| }, |
| { |
| "epoch": 6.212121212121212, |
| "grad_norm": 0.1551034370223493, |
| "learning_rate": 1.6557739698909436e-06, |
| "loss": 0.1483, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.04756705462932587, |
| "step": 205, |
| "valid_targets_mean": 8738.5, |
| "valid_targets_min": 973 |
| }, |
| { |
| "epoch": 6.363636363636363, |
| "grad_norm": 0.16985454441654288, |
| "learning_rate": 1.1045007019049182e-06, |
| "loss": 0.1478, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.050542425364255905, |
| "step": 210, |
| "valid_targets_mean": 8397.1, |
| "valid_targets_min": 3736 |
| }, |
| { |
| "epoch": 6.515151515151516, |
| "grad_norm": 0.17287398062167753, |
| "learning_rate": 6.619823814758786e-07, |
| "loss": 0.1519, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.049958501011133194, |
| "step": 215, |
| "valid_targets_mean": 9084.4, |
| "valid_targets_min": 1312 |
| }, |
| { |
| "epoch": 6.666666666666667, |
| "grad_norm": 0.15268537058766385, |
| "learning_rate": 3.307659673251595e-07, |
| "loss": 0.1489, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.044708285480737686, |
| "step": 220, |
| "valid_targets_mean": 8133.6, |
| "valid_targets_min": 1325 |
| }, |
| { |
| "epoch": 6.818181818181818, |
| "grad_norm": 0.17158051378559555, |
| "learning_rate": 1.1275780885282806e-07, |
| "loss": 0.145, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.047260358929634094, |
| "step": 225, |
| "valid_targets_mean": 9015.6, |
| "valid_targets_min": 3238 |
| }, |
| { |
| "epoch": 6.96969696969697, |
| "grad_norm": 0.1521451879785229, |
| "learning_rate": 9.212673951897177e-09, |
| "loss": 0.1523, |
| "loss_nan_ranks": 0, |
| "loss_rank_avg": 0.04904649779200554, |
| "step": 230, |
| "valid_targets_mean": 8038.4, |
| "valid_targets_min": 1670 |
| }, |
| { |
| "epoch": 7.0, |
| "step": 231, |
| "total_flos": 1.7248182061550797e+18, |
| "train_loss": 0.0, |
| "train_runtime": 6.073, |
| "train_samples_per_second": 3642.329, |
| "train_steps_per_second": 38.037 |
| } |
| ], |
| "logging_steps": 5, |
| "max_steps": 231, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 7, |
| "save_steps": 300, |
| "stateful_callbacks": { |
| "TrainerControl": { |
| "args": { |
| "should_epoch_stop": false, |
| "should_evaluate": false, |
| "should_log": false, |
| "should_save": true, |
| "should_training_stop": true |
| }, |
| "attributes": {} |
| } |
| }, |
| "total_flos": 1.7248182061550797e+18, |
| "train_batch_size": 1, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
