CoolFace
Modelpublic

anjandash/JavaBERT-small

sourceHugging Facemitupdated 4y agoView on Hugging Face
0likes27downloads
trainer_state.json143 linesDownload Raw Back to root
1{2  "best_metric": 6.430952072143555,3  "best_model_checkpoint": "finetuned-bert__java-8m-methods__jemma-java-bert-tokenizer__XXXM/checkpoint-45000",4  "epoch": 2.626185404569461,5  "global_step": 45000,6  "is_hyper_param_search": false,7  "is_local_process_zero": true,8  "is_world_process_zero": true,9  "log_history": [10    {11      "epoch": 0.29,12      "learning_rate": 0.0004998919902242132,13      "loss": 5.8968,14      "step": 500015    },16    {17      "epoch": 0.29,18      "eval_loss": 6.462705612182617,19      "eval_runtime": 1805.2005,20      "eval_samples_per_second": 27.514,21      "eval_steps_per_second": 6.879,22      "step": 500023    },24    {25      "epoch": 0.58,26      "learning_rate": 0.0004994534515508689,27      "loss": 5.9119,28      "step": 1000029    },30    {31      "epoch": 0.58,32      "eval_loss": 6.476224422454834,33      "eval_runtime": 1804.4274,34      "eval_samples_per_second": 27.526,35      "eval_steps_per_second": 6.882,36      "step": 1000037    },38    {39      "epoch": 0.88,40      "learning_rate": 0.0004986784236353977,41      "loss": 5.9089,42      "step": 1500043    },44    {45      "epoch": 0.88,46      "eval_loss": 6.472481727600098,47      "eval_runtime": 1804.1115,48      "eval_samples_per_second": 27.531,49      "eval_steps_per_second": 6.883,50      "step": 1500051    },52    {53      "epoch": 1.17,54      "learning_rate": 0.0004975676423415004,55      "loss": 5.9054,56      "step": 2000057    },58    {59      "epoch": 1.17,60      "eval_loss": 6.4574785232543945,61      "eval_runtime": 1804.4293,62      "eval_samples_per_second": 27.526,63      "eval_steps_per_second": 6.882,64      "step": 2000065    },66    {67      "epoch": 1.46,68      "learning_rate": 0.0004961227282936757,69      "loss": 5.9018,70      "step": 2500071    },72    {73      "epoch": 1.46,74      "eval_loss": 6.441328525543213,75      "eval_runtime": 1805.0751,76      "eval_samples_per_second": 27.516,77      "eval_steps_per_second": 6.879,78      "step": 2500079    },80    {81      "epoch": 1.75,82      "learning_rate": 0.0004943460202721656,83      "loss": 5.9057,84      "step": 3000085    },86    {87      "epoch": 1.75,88      "eval_loss": 6.43818998336792,89      "eval_runtime": 1791.0913,90      "eval_samples_per_second": 27.731,91      "eval_steps_per_second": 6.933,92      "step": 3000093    },94    {95      "epoch": 2.04,96      "learning_rate": 0.0004922392052032267,97      "loss": 5.9076,98      "step": 3500099    },100    {101      "epoch": 2.04,102      "eval_loss": 6.455899238586426,103      "eval_runtime": 1805.3008,104      "eval_samples_per_second": 27.513,105      "eval_steps_per_second": 6.879,106      "step": 35000107    },108    {109      "epoch": 2.33,110      "learning_rate": 0.0004898059684069413,111      "loss": 5.8996,112      "step": 40000113    },114    {115      "epoch": 2.33,116      "eval_loss": 6.460456848144531,117      "eval_runtime": 1805.5197,118      "eval_samples_per_second": 27.51,119      "eval_steps_per_second": 6.878,120      "step": 40000121    },122    {123      "epoch": 2.63,124      "learning_rate": 0.00048704920398382135,125      "loss": 5.9083,126      "step": 45000127    },128    {129      "epoch": 2.63,130      "eval_loss": 6.430952072143555,131      "eval_runtime": 1807.4508,132      "eval_samples_per_second": 27.48,133      "eval_steps_per_second": 6.87,134      "step": 45000135    }136  ],137  "max_steps": 428375,138  "num_train_epochs": 25,139  "total_flos": 7.5543354473762e+17,140  "trial_name": null,141  "trial_params": null142}143