codechrl/bert-micro-cybersecurity

1{2  "best_global_step": null,3  "best_metric": null,4  "best_model_checkpoint": null,5  "epoch": 3.0,6  "eval_steps": 500,7  "global_step": 3000,8  "is_hyper_param_search": false,9  "is_local_process_zero": true,10  "is_world_process_zero": true,11  "log_history": [12    {13      "epoch": 0.1,14      "grad_norm": 18.73765754699707,15      "learning_rate": 2.7500000000000004e-05,16      "loss": 2.1984,17      "step": 10018    },19    {20      "epoch": 0.2,21      "grad_norm": 51.54933166503906,22      "learning_rate": 4.966312056737589e-05,23      "loss": 2.1209,24      "step": 20025    },26    {27      "epoch": 0.3,28      "grad_norm": 32.52346420288086,29      "learning_rate": 4.789007092198582e-05,30      "loss": 2.1384,31      "step": 30032    },33    {34      "epoch": 0.4,35      "grad_norm": 35.90546417236328,36      "learning_rate": 4.6117021276595746e-05,37      "loss": 2.095,38      "step": 40039    },40    {41      "epoch": 0.5,42      "grad_norm": 29.83684730529785,43      "learning_rate": 4.4343971631205674e-05,44      "loss": 1.9609,45      "step": 50046    },47    {48      "epoch": 0.6,49      "grad_norm": 25.847257614135742,50      "learning_rate": 4.25709219858156e-05,51      "loss": 1.8644,52      "step": 60053    },54    {55      "epoch": 0.7,56      "grad_norm": 44.531307220458984,57      "learning_rate": 4.079787234042554e-05,58      "loss": 1.8201,59      "step": 70060    },61    {62      "epoch": 0.8,63      "grad_norm": 36.09579849243164,64      "learning_rate": 3.902482269503546e-05,65      "loss": 1.6903,66      "step": 80067    },68    {69      "epoch": 0.9,70      "grad_norm": 25.78478240966797,71      "learning_rate": 3.7251773049645395e-05,72      "loss": 1.5981,73      "step": 90074    },75    {76      "epoch": 1.0,77      "grad_norm": 26.695831298828125,78      "learning_rate": 3.547872340425532e-05,79      "loss": 1.6652,80      "step": 100081    },82    {83      "epoch": 1.1,84      "grad_norm": 31.150781631469727,85      "learning_rate": 3.3705673758865245e-05,86      "loss": 1.5633,87      "step": 110088    },89    {90      "epoch": 1.2,91      "grad_norm": 37.66817092895508,92      "learning_rate": 3.193262411347518e-05,93      "loss": 1.5541,94      "step": 120095    },96    {97      "epoch": 1.3,98      "grad_norm": 26.20791244506836,99      "learning_rate": 3.0159574468085105e-05,100      "loss": 1.516,101      "step": 1300102    },103    {104      "epoch": 1.4,105      "grad_norm": 22.03032684326172,106      "learning_rate": 2.8386524822695037e-05,107      "loss": 1.6995,108      "step": 1400109    },110    {111      "epoch": 1.5,112      "grad_norm": 19.463829040527344,113      "learning_rate": 2.6613475177304965e-05,114      "loss": 1.4198,115      "step": 1500116    },117    {118      "epoch": 1.6,119      "grad_norm": 12.715448379516602,120      "learning_rate": 2.4840425531914897e-05,121      "loss": 1.2346,122      "step": 1600123    },124    {125      "epoch": 1.7,126      "grad_norm": 23.795001983642578,127      "learning_rate": 2.3067375886524825e-05,128      "loss": 1.4306,129      "step": 1700130    },131    {132      "epoch": 1.8,133      "grad_norm": 35.95954132080078,134      "learning_rate": 2.129432624113475e-05,135      "loss": 1.5738,136      "step": 1800137    },138    {139      "epoch": 1.9,140      "grad_norm": 35.67465591430664,141      "learning_rate": 1.9521276595744682e-05,142      "loss": 1.4514,143      "step": 1900144    },145    {146      "epoch": 2.0,147      "grad_norm": 19.942232131958008,148      "learning_rate": 1.774822695035461e-05,149      "loss": 1.4458,150      "step": 2000151    },152    {153      "epoch": 2.1,154      "grad_norm": 18.1514949798584,155      "learning_rate": 1.597517730496454e-05,156      "loss": 1.5136,157      "step": 2100158    },159    {160      "epoch": 2.2,161      "grad_norm": 28.748600006103516,162      "learning_rate": 1.420212765957447e-05,163      "loss": 1.5163,164      "step": 2200165    },166    {167      "epoch": 2.3,168      "grad_norm": 16.781105041503906,169      "learning_rate": 1.2429078014184398e-05,170      "loss": 1.3713,171      "step": 2300172    },173    {174      "epoch": 2.4,175      "grad_norm": 13.730096817016602,176      "learning_rate": 1.0656028368794328e-05,177      "loss": 1.5527,178      "step": 2400179    },180    {181      "epoch": 2.5,182      "grad_norm": 40.0937614440918,183      "learning_rate": 8.882978723404256e-06,184      "loss": 1.3158,185      "step": 2500186    },187    {188      "epoch": 2.6,189      "grad_norm": 58.328670501708984,190      "learning_rate": 7.109929078014185e-06,191      "loss": 1.4293,192      "step": 2600193    },194    {195      "epoch": 2.7,196      "grad_norm": 22.44051742553711,197      "learning_rate": 5.336879432624114e-06,198      "loss": 1.3321,199      "step": 2700200    },201    {202      "epoch": 2.8,203      "grad_norm": 38.144752502441406,204      "learning_rate": 3.563829787234043e-06,205      "loss": 1.2445,206      "step": 2800207    },208    {209      "epoch": 2.9,210      "grad_norm": 21.987308502197266,211      "learning_rate": 1.790780141843972e-06,212      "loss": 1.5435,213      "step": 2900214    },215    {216      "epoch": 3.0,217      "grad_norm": 32.320709228515625,218      "learning_rate": 1.773049645390071e-08,219      "loss": 1.3325,220      "step": 3000221    }222  ],223  "logging_steps": 100,224  "max_steps": 3000,225  "num_input_tokens_seen": 0,226  "num_train_epochs": 3,227  "save_steps": 500,228  "stateful_callbacks": {229    "TrainerControl": {230      "args": {231        "should_epoch_stop": false,232        "should_evaluate": false,233        "should_log": false,234        "should_save": true,235        "should_training_stop": true236      },237      "attributes": {}238    }239  },240  "total_flos": 4092733440000.0,241  "train_batch_size": 1,242  "trial_name": null,243  "trial_params": null244}245