dzungpham/graphcodebert-code-classification

1{2  "best_global_step": null,3  "best_metric": null,4  "best_model_checkpoint": null,5  "epoch": 0.0256,6  "eval_steps": 1000,7  "global_step": 400,8  "is_hyper_param_search": false,9  "is_local_process_zero": true,10  "is_world_process_zero": true,11  "log_history": [12    {13      "epoch": 0.00064,14      "grad_norm": 1.6144306659698486,15      "learning_rate": 1.1520000000000002e-08,16      "loss": 0.729,17      "step": 1018    },19    {20      "epoch": 0.00128,21      "grad_norm": 2.0952296257019043,22      "learning_rate": 2.4320000000000002e-08,23      "loss": 0.7295,24      "step": 2025    },26    {27      "epoch": 0.00192,28      "grad_norm": 1.3587689399719238,29      "learning_rate": 3.7120000000000004e-08,30      "loss": 0.73,31      "step": 3032    },33    {34      "epoch": 0.00256,35      "grad_norm": 1.2531732320785522,36      "learning_rate": 4.9920000000000006e-08,37      "loss": 0.7221,38      "step": 4039    },40    {41      "epoch": 0.0032,42      "grad_norm": 1.437932014465332,43      "learning_rate": 6.272000000000001e-08,44      "loss": 0.7209,45      "step": 5046    },47    {48      "epoch": 0.00384,49      "grad_norm": 1.418426752090454,50      "learning_rate": 7.552e-08,51      "loss": 0.729,52      "step": 6053    },54    {55      "epoch": 0.00448,56      "grad_norm": 1.9476298093795776,57      "learning_rate": 8.832e-08,58      "loss": 0.7242,59      "step": 7060    },61    {62      "epoch": 0.00512,63      "grad_norm": 1.7948051691055298,64      "learning_rate": 1.0112000000000001e-07,65      "loss": 0.7227,66      "step": 8067    },68    {69      "epoch": 0.00576,70      "grad_norm": 1.6534360647201538,71      "learning_rate": 1.1392e-07,72      "loss": 0.7234,73      "step": 9074    },75    {76      "epoch": 0.0064,77      "grad_norm": 1.0920158624649048,78      "learning_rate": 1.2672e-07,79      "loss": 0.7328,80      "step": 10081    },82    {83      "epoch": 0.00704,84      "grad_norm": 1.977837085723877,85      "learning_rate": 1.3952000000000002e-07,86      "loss": 0.7263,87      "step": 11088    },89    {90      "epoch": 0.00768,91      "grad_norm": 1.388983130455017,92      "learning_rate": 1.5232000000000003e-07,93      "loss": 0.7286,94      "step": 12095    },96    {97      "epoch": 0.00832,98      "grad_norm": 1.2956682443618774,99      "learning_rate": 1.6512e-07,100      "loss": 0.7251,101      "step": 130102    },103    {104      "epoch": 0.00896,105      "grad_norm": 1.8125052452087402,106      "learning_rate": 1.7792e-07,107      "loss": 0.7251,108      "step": 140109    },110    {111      "epoch": 0.0096,112      "grad_norm": 1.626846194267273,113      "learning_rate": 1.9072e-07,114      "loss": 0.727,115      "step": 150116    },117    {118      "epoch": 0.01024,119      "grad_norm": 2.3243086338043213,120      "learning_rate": 2.0352e-07,121      "loss": 0.726,122      "step": 160123    },124    {125      "epoch": 0.01088,126      "grad_norm": 1.4734737873077393,127      "learning_rate": 2.1632e-07,128      "loss": 0.7252,129      "step": 170130    },131    {132      "epoch": 0.01152,133      "grad_norm": 2.090498685836792,134      "learning_rate": 2.2912e-07,135      "loss": 0.7273,136      "step": 180137    },138    {139      "epoch": 0.01216,140      "grad_norm": 1.7563093900680542,141      "learning_rate": 2.4192000000000004e-07,142      "loss": 0.719,143      "step": 190144    },145    {146      "epoch": 0.0128,147      "grad_norm": 1.449843168258667,148      "learning_rate": 2.5472000000000005e-07,149      "loss": 0.7237,150      "step": 200151    },152    {153      "epoch": 0.01344,154      "grad_norm": 144219.625,155      "learning_rate": 5.350742447516642e-07,156      "loss": 0.7218,157      "step": 210158    },159    {160      "epoch": 0.01408,161      "grad_norm": 105046.0234375,162      "learning_rate": 5.606758832565284e-07,163      "loss": 0.718,164      "step": 220165    },166    {167      "epoch": 0.01472,168      "grad_norm": 126142.4296875,169      "learning_rate": 5.862775217613928e-07,170      "loss": 0.7107,171      "step": 230172    },173    {174      "epoch": 0.01536,175      "grad_norm": 92423.2265625,176      "learning_rate": 6.118791602662571e-07,177      "loss": 0.7271,178      "step": 240179    },180    {181      "epoch": 0.016,182      "grad_norm": 98091.828125,183      "learning_rate": 6.374807987711214e-07,184      "loss": 0.7123,185      "step": 250186    },187    {188      "epoch": 0.01664,189      "grad_norm": 131949.578125,190      "learning_rate": 6.630824372759858e-07,191      "loss": 0.7204,192      "step": 260193    },194    {195      "epoch": 0.01728,196      "grad_norm": 112228.5625,197      "learning_rate": 6.8868407578085e-07,198      "loss": 0.722,199      "step": 270200    },201    {202      "epoch": 0.01792,203      "grad_norm": 64587.734375,204      "learning_rate": 7.142857142857143e-07,205      "loss": 0.7263,206      "step": 280207    },208    {209      "epoch": 0.01856,210      "grad_norm": 99893.203125,211      "learning_rate": 7.398873527905787e-07,212      "loss": 0.7169,213      "step": 290214    },215    {216      "epoch": 0.0192,217      "grad_norm": 135749.875,218      "learning_rate": 7.65488991295443e-07,219      "loss": 0.7122,220      "step": 300221    },222    {223      "epoch": 0.01984,224      "grad_norm": 103292.5703125,225      "learning_rate": 7.910906298003073e-07,226      "loss": 0.7183,227      "step": 310228    },229    {230      "epoch": 0.02048,231      "grad_norm": 86927.28125,232      "learning_rate": 8.166922683051716e-07,233      "loss": 0.7192,234      "step": 320235    },236    {237      "epoch": 0.02112,238      "grad_norm": 153738.390625,239      "learning_rate": 8.422939068100359e-07,240      "loss": 0.711,241      "step": 330242    },243    {244      "epoch": 0.02176,245      "grad_norm": 69994.7734375,246      "learning_rate": 8.678955453149002e-07,247      "loss": 0.7176,248      "step": 340249    },250    {251      "epoch": 0.0224,252      "grad_norm": 141370.6875,253      "learning_rate": 8.934971838197646e-07,254      "loss": 0.7105,255      "step": 350256    },257    {258      "epoch": 0.02304,259      "grad_norm": 71139.453125,260      "learning_rate": 9.190988223246289e-07,261      "loss": 0.7126,262      "step": 360263    },264    {265      "epoch": 0.02368,266      "grad_norm": 82039.1953125,267      "learning_rate": 9.447004608294931e-07,268      "loss": 0.7078,269      "step": 370270    },271    {272      "epoch": 0.02432,273      "grad_norm": 71275.7890625,274      "learning_rate": 9.703020993343575e-07,275      "loss": 0.7145,276      "step": 380277    },278    {279      "epoch": 0.02496,280      "grad_norm": 145801.21875,281      "learning_rate": 9.959037378392218e-07,282      "loss": 0.7102,283      "step": 390284    },285    {286      "epoch": 0.0256,287      "grad_norm": 171507.0,288      "learning_rate": 1.021505376344086e-06,289      "loss": 0.7123,290      "step": 400291    }292  ],293  "logging_steps": 10,294  "max_steps": 78125,295  "num_input_tokens_seen": 0,296  "num_train_epochs": 5,297  "save_steps": 200,298  "stateful_callbacks": {299    "EarlyStoppingCallback": {300      "args": {301        "early_stopping_patience": 3,302        "early_stopping_threshold": 0.0303      },304      "attributes": {305        "early_stopping_patience_counter": 0306      }307    },308    "TrainerControl": {309      "args": {310        "should_epoch_stop": false,311        "should_evaluate": false,312        "should_log": false,313        "should_save": true,314        "should_training_stop": false315      },316      "attributes": {}317    }318  },319  "total_flos": 3367295286497280.0,320  "train_batch_size": 32,321  "trial_name": null,322  "trial_params": null323}324