CoolFace
Modelpublic

Lazyhope/python-clone-detection

sourceHugging Facemitupdated 4y agoView on Hugging Face
3likes74downloads
trainer_state.json248 linesDownload Raw Back to root
1{2  "best_metric": 0.9843817398562538,3  "best_model_checkpoint": "./exps/1_fold_RobertaRBERT/checkpoint-22000",4  "epoch": 1.1827271994194017,5  "global_step": 22000,6  "is_hyper_param_search": false,7  "is_local_process_zero": true,8  "is_world_process_zero": true,9  "log_history": [10    {11      "epoch": 0.05,12      "learning_rate": 1.0693175711982806e-05,13      "loss": 0.3694,14      "step": 100015    },16    {17      "epoch": 0.11,18      "learning_rate": 1.9924167397640137e-05,19      "loss": 0.1227,20      "step": 200021    },22    {23      "epoch": 0.11,24      "eval_accuracy": 0.9525065963060686,25      "eval_loss": 0.12871456146240234,26      "eval_runtime": 1262.9688,27      "eval_samples_per_second": 141.341,28      "eval_steps_per_second": 4.417,29      "step": 200030    },31    {32      "epoch": 0.16,33      "learning_rate": 1.9358252454656067e-05,34      "loss": 0.077,35      "step": 300036    },37    {38      "epoch": 0.22,39      "learning_rate": 1.8792337511671998e-05,40      "loss": 0.0597,41      "step": 400042    },43    {44      "epoch": 0.22,45      "eval_accuracy": 0.977625777971979,46      "eval_loss": 0.07220922410488129,47      "eval_runtime": 1261.8214,48      "eval_samples_per_second": 141.469,49      "eval_steps_per_second": 4.421,50      "step": 400051    },52    {53      "epoch": 0.27,54      "learning_rate": 1.8226422568687928e-05,55      "loss": 0.0492,56      "step": 500057    },58    {59      "epoch": 0.32,60      "learning_rate": 1.766050762570386e-05,61      "loss": 0.0429,62      "step": 600063    },64    {65      "epoch": 0.32,66      "eval_accuracy": 0.9803035141085323,67      "eval_loss": 0.07511359453201294,68      "eval_runtime": 1262.1321,69      "eval_samples_per_second": 141.434,70      "eval_steps_per_second": 4.42,71      "step": 600072    },73    {74      "epoch": 0.38,75      "learning_rate": 1.709459268271979e-05,76      "loss": 0.0385,77      "step": 700078    },79    {80      "epoch": 0.43,81      "learning_rate": 1.652867773973572e-05,82      "loss": 0.0356,83      "step": 800084    },85    {86      "epoch": 0.43,87      "eval_accuracy": 0.9810037589141164,88      "eval_loss": 0.07426761090755463,89      "eval_runtime": 1264.8606,90      "eval_samples_per_second": 141.129,91      "eval_steps_per_second": 4.411,92      "step": 800093    },94    {95      "epoch": 0.48,96      "learning_rate": 1.5963328711694633e-05,97      "loss": 0.0326,98      "step": 900099    },100    {101      "epoch": 0.54,102      "learning_rate": 1.5397413768710567e-05,103      "loss": 0.0288,104      "step": 10000105    },106    {107      "epoch": 0.54,108      "eval_accuracy": 0.9811886235427906,109      "eval_loss": 0.0728357657790184,110      "eval_runtime": 1260.8301,111      "eval_samples_per_second": 141.581,112      "eval_steps_per_second": 4.425,113      "step": 10000114    },115    {116      "epoch": 0.59,117      "learning_rate": 1.4831498825726494e-05,118      "loss": 0.0281,119      "step": 11000120    },121    {122      "epoch": 0.65,123      "learning_rate": 1.4265583882742424e-05,124      "loss": 0.0266,125      "step": 12000126    },127    {128      "epoch": 0.65,129      "eval_accuracy": 0.9783652364866757,130      "eval_loss": 0.08940344303846359,131      "eval_runtime": 1265.0154,132      "eval_samples_per_second": 141.112,133      "eval_steps_per_second": 4.41,134      "step": 12000135    },136    {137      "epoch": 0.7,138      "learning_rate": 1.370023485470134e-05,139      "loss": 0.0244,140      "step": 13000141    },142    {143      "epoch": 0.75,144      "learning_rate": 1.313431991171727e-05,145      "loss": 0.0225,146      "step": 14000147    },148    {149      "epoch": 0.75,150      "eval_accuracy": 0.9823538308992824,151      "eval_loss": 0.06773301213979721,152      "eval_runtime": 1260.1554,153      "eval_samples_per_second": 141.656,154      "eval_steps_per_second": 4.427,155      "step": 14000156    },157    {158      "epoch": 0.81,159      "learning_rate": 1.2568970883676185e-05,160      "loss": 0.0206,161      "step": 15000162    },163    {164      "epoch": 0.86,165      "learning_rate": 1.2003055940692115e-05,166      "loss": 0.0208,167      "step": 16000168    },169    {170      "epoch": 0.86,171      "eval_accuracy": 0.9828524052008583,172      "eval_loss": 0.08000749349594116,173      "eval_runtime": 1262.4987,174      "eval_samples_per_second": 141.393,175      "eval_steps_per_second": 4.419,176      "step": 16000177    },178    {179      "epoch": 0.91,180      "learning_rate": 1.1437140997708046e-05,181      "loss": 0.0199,182      "step": 17000183    },184    {185      "epoch": 0.97,186      "learning_rate": 1.087179196966696e-05,187      "loss": 0.0182,188      "step": 18000189    },190    {191      "epoch": 0.97,192      "eval_accuracy": 0.9813734881714647,193      "eval_loss": 0.08186674118041992,194      "eval_runtime": 1260.8606,195      "eval_samples_per_second": 141.577,196      "eval_steps_per_second": 4.425,197      "step": 18000198    },199    {200      "epoch": 1.02,201      "learning_rate": 1.030587702668289e-05,202      "loss": 0.0159,203      "step": 19000204    },205    {206      "epoch": 1.08,207      "learning_rate": 9.739962083698821e-06,208      "loss": 0.0137,209      "step": 20000210    },211    {212      "epoch": 1.08,213      "eval_accuracy": 0.9839727968897927,214      "eval_loss": 0.062463145703077316,215      "eval_runtime": 1261.7007,216      "eval_samples_per_second": 141.483,217      "eval_steps_per_second": 4.422,218      "step": 20000219    },220    {221      "epoch": 1.13,222      "learning_rate": 9.174047140714752e-06,223      "loss": 0.0123,224      "step": 21000225    },226    {227      "epoch": 1.18,228      "learning_rate": 8.60926402761665e-06,229      "loss": 0.0138,230      "step": 22000231    },232    {233      "epoch": 1.18,234      "eval_accuracy": 0.9843817398562538,235      "eval_loss": 0.07732487469911575,236      "eval_runtime": 1262.988,237      "eval_samples_per_second": 141.339,238      "eval_steps_per_second": 4.417,239      "step": 22000240    }241  ],242  "max_steps": 37202,243  "num_train_epochs": 2,244  "total_flos": 2.7232075889551824e+17,245  "trial_name": null,246  "trial_params": null247}248