CoolFace
Modelpublic

anjandash/JavaBERT-mini

sourceHugging Facemitupdated 4y agoView on Hugging Face
0likes31downloads
trainer_state.json129 linesDownload Raw Back to root
1{2  "best_metric": 1.2563587427139282,3  "best_model_checkpoint": "finetuned-bert__java-8m-methods__jemma-java-bert-tokenizer/checkpoint-5000",4  "epoch": 4.669020454611771,5  "global_step": 40000,6  "is_hyper_param_search": false,7  "is_local_process_zero": true,8  "is_world_process_zero": true,9  "log_history": [10    {11      "epoch": 0.58,12      "learning_rate": 0.0004888116319438376,13      "loss": 3.0057,14      "step": 500015    },16    {17      "epoch": 0.58,18      "eval_loss": 1.2563587427139282,19      "eval_runtime": 836.9828,20      "eval_samples_per_second": 59.343,21      "eval_steps_per_second": 7.418,22      "step": 500023    },24    {25      "epoch": 1.17,26      "learning_rate": 0.00044517243817696517,27      "loss": 1.5788,28      "step": 1000029    },30    {31      "epoch": 1.17,32      "eval_loss": 6.417298793792725,33      "eval_runtime": 835.6884,34      "eval_samples_per_second": 59.435,35      "eval_steps_per_second": 7.43,36      "step": 1000037    },38    {39      "epoch": 1.75,40      "learning_rate": 0.00037428394451934934,41      "loss": 5.9109,42      "step": 1500043    },44    {45      "epoch": 1.75,46      "eval_loss": 6.413046836853027,47      "eval_runtime": 833.6127,48      "eval_samples_per_second": 59.583,49      "eval_steps_per_second": 7.448,50      "step": 1500051    },52    {53      "epoch": 2.33,54      "learning_rate": 0.00028607867802092244,55      "loss": 5.9036,56      "step": 2000057    },58    {59      "epoch": 2.33,60      "eval_loss": 6.433701038360596,61      "eval_runtime": 833.0303,62      "eval_samples_per_second": 59.624,63      "eval_steps_per_second": 7.454,64      "step": 2000065    },66    {67      "epoch": 2.92,68      "learning_rate": 0.00019284648488204127,69      "loss": 5.9003,70      "step": 2500071    },72    {73      "epoch": 2.92,74      "eval_loss": 6.432703971862793,75      "eval_runtime": 833.1905,76      "eval_samples_per_second": 59.613,77      "eval_steps_per_second": 7.452,78      "step": 2500079    },80    {81      "epoch": 3.5,82      "learning_rate": 0.00010757762475595697,83      "loss": 5.8971,84      "step": 3000085    },86    {87      "epoch": 3.5,88      "eval_loss": 6.4252238273620605,89      "eval_runtime": 846.7073,90      "eval_samples_per_second": 58.661,91      "eval_steps_per_second": 7.333,92      "step": 3000093    },94    {95      "epoch": 4.09,96      "learning_rate": 4.217367458660157e-05,97      "loss": 5.8972,98      "step": 3500099    },100    {101      "epoch": 4.09,102      "eval_loss": 6.429380416870117,103      "eval_runtime": 846.9722,104      "eval_samples_per_second": 58.643,105      "eval_steps_per_second": 7.331,106      "step": 35000107    },108    {109      "epoch": 4.67,110      "learning_rate": 5.6997776508484205e-06,111      "loss": 5.892,112      "step": 40000113    },114    {115      "epoch": 4.67,116      "eval_loss": 6.426671504974365,117      "eval_runtime": 847.0538,118      "eval_samples_per_second": 58.637,119      "eval_steps_per_second": 7.33,120      "step": 40000121    }122  ],123  "max_steps": 42835,124  "num_train_epochs": 5,125  "total_flos": 6.741181575431455e+17,126  "trial_name": null,127  "trial_params": null128}129