mustehsannisarrao/pseudocode-to-python

1{2  "best_global_step": 800,3  "best_metric": 0.8735545873641968,4  "best_model_checkpoint": "./gpt2-pseudocode-to-python/checkpoint-800",5  "epoch": 8.421052631578947,6  "eval_steps": 100,7  "global_step": 800,8  "is_hyper_param_search": false,9  "is_local_process_zero": true,10  "is_world_process_zero": true,11  "log_history": [12    {13      "epoch": 0.5263157894736842,14      "grad_norm": 161976.90625,15      "learning_rate": 2.45e-05,16      "loss": 1.0892,17      "step": 5018    },19    {20      "epoch": 1.0526315789473684,21      "grad_norm": 273256.0625,22      "learning_rate": 4.9500000000000004e-05,23      "loss": 0.7226,24      "step": 10025    },26    {27      "epoch": 1.0526315789473684,28      "eval_loss": 1.2260727882385254,29      "eval_runtime": 7.4214,30      "eval_samples_per_second": 25.602,31      "eval_steps_per_second": 3.234,32      "step": 10033    },34    {35      "epoch": 1.5789473684210527,36      "grad_norm": 119407.9609375,37      "learning_rate": 4.711764705882353e-05,38      "loss": 0.6145,39      "step": 15040    },41    {42      "epoch": 2.1052631578947367,43      "grad_norm": 144676.25,44      "learning_rate": 4.4176470588235296e-05,45      "loss": 0.5795,46      "step": 20047    },48    {49      "epoch": 2.1052631578947367,50      "eval_loss": 1.0508249998092651,51      "eval_runtime": 7.0559,52      "eval_samples_per_second": 26.928,53      "eval_steps_per_second": 3.401,54      "step": 20055    },56    {57      "epoch": 2.6315789473684212,58      "grad_norm": 139209.640625,59      "learning_rate": 4.123529411764706e-05,60      "loss": 0.556,61      "step": 25062    },63    {64      "epoch": 3.1578947368421053,65      "grad_norm": 100604.8359375,66      "learning_rate": 3.8294117647058826e-05,67      "loss": 0.522,68      "step": 30069    },70    {71      "epoch": 3.1578947368421053,72      "eval_loss": 0.982334315776825,73      "eval_runtime": 7.0698,74      "eval_samples_per_second": 26.875,75      "eval_steps_per_second": 3.395,76      "step": 30077    },78    {79      "epoch": 3.6842105263157894,80      "grad_norm": 93355.3359375,81      "learning_rate": 3.535294117647059e-05,82      "loss": 0.4965,83      "step": 35084    },85    {86      "epoch": 4.2105263157894735,87      "grad_norm": 120237.671875,88      "learning_rate": 3.2411764705882356e-05,89      "loss": 0.491,90      "step": 40091    },92    {93      "epoch": 4.2105263157894735,94      "eval_loss": 0.9479882717132568,95      "eval_runtime": 7.0795,96      "eval_samples_per_second": 26.838,97      "eval_steps_per_second": 3.39,98      "step": 40099    },100    {101      "epoch": 4.7368421052631575,102      "grad_norm": 96738.1875,103      "learning_rate": 2.9470588235294117e-05,104      "loss": 0.4977,105      "step": 450106    },107    {108      "epoch": 5.2631578947368425,109      "grad_norm": 103758.7578125,110      "learning_rate": 2.6529411764705885e-05,111      "loss": 0.4549,112      "step": 500113    },114    {115      "epoch": 5.2631578947368425,116      "eval_loss": 0.917313277721405,117      "eval_runtime": 7.0732,118      "eval_samples_per_second": 26.862,119      "eval_steps_per_second": 3.393,120      "step": 500121    },122    {123      "epoch": 5.7894736842105265,124      "grad_norm": 97584.5078125,125      "learning_rate": 2.3588235294117647e-05,126      "loss": 0.4538,127      "step": 550128    },129    {130      "epoch": 6.315789473684211,131      "grad_norm": 114618.953125,132      "learning_rate": 2.0647058823529415e-05,133      "loss": 0.4523,134      "step": 600135    },136    {137      "epoch": 6.315789473684211,138      "eval_loss": 0.8995645642280579,139      "eval_runtime": 7.051,140      "eval_samples_per_second": 26.946,141      "eval_steps_per_second": 3.404,142      "step": 600143    },144    {145      "epoch": 6.842105263157895,146      "grad_norm": 102502.5703125,147      "learning_rate": 1.7705882352941177e-05,148      "loss": 0.4441,149      "step": 650150    },151    {152      "epoch": 7.368421052631579,153      "grad_norm": 103262.21875,154      "learning_rate": 1.4764705882352942e-05,155      "loss": 0.4493,156      "step": 700157    },158    {159      "epoch": 7.368421052631579,160      "eval_loss": 0.8839536905288696,161      "eval_runtime": 7.1126,162      "eval_samples_per_second": 26.713,163      "eval_steps_per_second": 3.374,164      "step": 700165    },166    {167      "epoch": 7.894736842105263,168      "grad_norm": 80773.8046875,169      "learning_rate": 1.1823529411764707e-05,170      "loss": 0.4266,171      "step": 750172    },173    {174      "epoch": 8.421052631578947,175      "grad_norm": 92954.6328125,176      "learning_rate": 8.88235294117647e-06,177      "loss": 0.446,178      "step": 800179    },180    {181      "epoch": 8.421052631578947,182      "eval_loss": 0.8735545873641968,183      "eval_runtime": 7.0958,184      "eval_samples_per_second": 26.776,185      "eval_steps_per_second": 3.382,186      "step": 800187    }188  ],189  "logging_steps": 50,190  "max_steps": 950,191  "num_input_tokens_seen": 0,192  "num_train_epochs": 10,193  "save_steps": 200,194  "stateful_callbacks": {195    "EarlyStoppingCallback": {196      "args": {197        "early_stopping_patience": 3,198        "early_stopping_threshold": 0.0199      },200      "attributes": {201        "early_stopping_patience_counter": 0202      }203    },204    "TrainerControl": {205      "args": {206        "should_epoch_stop": false,207        "should_evaluate": false,208        "should_log": false,209        "should_save": true,210        "should_training_stop": false211      },212      "attributes": {}213    }214  },215  "total_flos": 3344538009600000.0,216  "train_batch_size": 8,217  "trial_name": null,218  "trial_params": null219}220