Oltking/adaption_agronomy_problem_solving

1{2  "best_global_step": null,3  "best_metric": null,4  "best_model_checkpoint": null,5  "epoch": 1.0,6  "eval_steps": 6,7  "global_step": 33,8  "is_hyper_param_search": false,9  "is_local_process_zero": true,10  "is_world_process_zero": true,11  "log_history": [12    {13      "epoch": 0.030303030303030304,14      "grad_norm": 1.4287192821502686,15      "learning_rate": 0.0,16      "loss": 2.0361328125,17      "step": 118    },19    {20      "epoch": 0.06060606060606061,21      "grad_norm": 1.437935471534729,22      "learning_rate": 2.5e-06,23      "loss": 2.1640625,24      "step": 225    },26    {27      "epoch": 0.09090909090909091,28      "grad_norm": 1.3686531782150269,29      "learning_rate": 5e-06,30      "loss": 2.01171875,31      "step": 332    },33    {34      "epoch": 0.12121212121212122,35      "grad_norm": 1.286812424659729,36      "learning_rate": 7.500000000000001e-06,37      "loss": 1.9638671875,38      "step": 439    },40    {41      "epoch": 0.15151515151515152,42      "grad_norm": 1.0939735174179077,43      "learning_rate": 1e-05,44      "loss": 1.8427734375,45      "step": 546    },47    {48      "epoch": 0.18181818181818182,49      "grad_norm": 1.7438583374023438,50      "learning_rate": 9.97362080719462e-06,51      "loss": 2.0869140625,52      "step": 653    },54    {55      "epoch": 0.21212121212121213,56      "grad_norm": 0.9894726872444153,57      "learning_rate": 9.89479250069539e-06,58      "loss": 1.76953125,59      "step": 760    },61    {62      "epoch": 0.24242424242424243,63      "grad_norm": 0.8804801106452942,64      "learning_rate": 9.764439270322612e-06,65      "loss": 1.7158203125,66      "step": 867    },68    {69      "epoch": 0.2727272727272727,70      "grad_norm": 0.6704576015472412,71      "learning_rate": 9.584089388519307e-06,72      "loss": 1.7607421875,73      "step": 974    },75    {76      "epoch": 0.2727272727272727,77      "eval_loss": 2.0078125,78      "eval_runtime": 2.5535,79      "eval_samples_per_second": 0.783,80      "eval_steps_per_second": 0.392,81      "step": 982    },83    {84      "epoch": 0.30303030303030304,85      "grad_norm": 1.9209635257720947,86      "learning_rate": 9.355857292754152e-06,87      "loss": 2.0673828125,88      "step": 1089    },90    {91      "epoch": 0.3333333333333333,92      "grad_norm": 0.8329216837882996,93      "learning_rate": 9.082418795675397e-06,94      "loss": 1.677734375,95      "step": 1196    },97    {98      "epoch": 0.36363636363636365,99      "grad_norm": 0.9212274551391602,100      "learning_rate": 8.76697971365409e-06,101      "loss": 1.9130859375,102      "step": 12103    },104    {105      "epoch": 0.3939393939393939,106      "grad_norm": 1.0851582288742065,107      "learning_rate": 8.413238281518225e-06,108      "loss": 1.873046875,109      "step": 13110    },111    {112      "epoch": 0.42424242424242425,113      "grad_norm": 0.9320775270462036,114      "learning_rate": 8.025341794130722e-06,115      "loss": 1.7841796875,116      "step": 14117    },118    {119      "epoch": 0.45454545454545453,120      "grad_norm": 0.6170089244842529,121      "learning_rate": 7.607837983149057e-06,122      "loss": 1.5927734375,123      "step": 15124    },125    {126      "epoch": 0.45454545454545453,127      "eval_loss": 1.8125,128      "eval_runtime": 2.5501,129      "eval_samples_per_second": 0.784,130      "eval_steps_per_second": 0.392,131      "step": 15132    },133    {134      "epoch": 0.48484848484848486,135      "grad_norm": 0.7249671816825867,136      "learning_rate": 7.165621699029615e-06,137      "loss": 1.7607421875,138      "step": 16139    },140    {141      "epoch": 0.5151515151515151,142      "grad_norm": 0.6273460984230042,143      "learning_rate": 6.703877523381495e-06,144      "loss": 1.666015625,145      "step": 17146    },147    {148      "epoch": 0.5454545454545454,149      "grad_norm": 0.6081604957580566,150      "learning_rate": 6.228018984487443e-06,151      "loss": 1.486328125,152      "step": 18153    },154    {155      "epoch": 0.5757575757575758,156      "grad_norm": 0.8183671832084656,157      "learning_rate": 5.74362508863438e-06,158      "loss": 1.7958984375,159      "step": 19160    },161    {162      "epoch": 0.6060606060606061,163      "grad_norm": 0.8645656704902649,164      "learning_rate": 5.256374911365621e-06,165      "loss": 1.6435546875,166      "step": 20167    },168    {169      "epoch": 0.6363636363636364,170      "grad_norm": 0.7197942137718201,171      "learning_rate": 4.771981015512559e-06,172      "loss": 1.6796875,173      "step": 21174    },175    {176      "epoch": 0.6363636363636364,177      "eval_loss": 1.703125,178      "eval_runtime": 2.5572,179      "eval_samples_per_second": 0.782,180      "eval_steps_per_second": 0.391,181      "step": 21182    },183    {184      "epoch": 0.6666666666666666,185      "grad_norm": 0.6852272152900696,186      "learning_rate": 4.296122476618507e-06,187      "loss": 1.71484375,188      "step": 22189    },190    {191      "epoch": 0.696969696969697,192      "grad_norm": 0.5617450475692749,193      "learning_rate": 3.834378300970385e-06,194      "loss": 1.509765625,195      "step": 23196    },197    {198      "epoch": 0.7272727272727273,199      "grad_norm": 0.7030121684074402,200      "learning_rate": 3.392162016850945e-06,201      "loss": 1.587890625,202      "step": 24203    },204    {205      "epoch": 0.7575757575757576,206      "grad_norm": 0.5215224623680115,207      "learning_rate": 2.9746582058692803e-06,208      "loss": 1.5263671875,209      "step": 25210    },211    {212      "epoch": 0.7878787878787878,213      "grad_norm": 0.5175567269325256,214      "learning_rate": 2.586761718481776e-06,215      "loss": 1.5361328125,216      "step": 26217    },218    {219      "epoch": 0.8181818181818182,220      "grad_norm": 0.505078136920929,221      "learning_rate": 2.2330202863459123e-06,222      "loss": 1.5234375,223      "step": 27224    },225    {226      "epoch": 0.8181818181818182,227      "eval_loss": 1.6484375,228      "eval_runtime": 2.5566,229      "eval_samples_per_second": 0.782,230      "eval_steps_per_second": 0.391,231      "step": 27232    },233    {234      "epoch": 0.8484848484848485,235      "grad_norm": 0.4583703279495239,236      "learning_rate": 1.9175812043246034e-06,237      "loss": 1.517578125,238      "step": 28239    },240    {241      "epoch": 0.8787878787878788,242      "grad_norm": 0.47751930356025696,243      "learning_rate": 1.6441427072458493e-06,244      "loss": 1.5185546875,245      "step": 29246    },247    {248      "epoch": 0.9090909090909091,249      "grad_norm": 0.5964409708976746,250      "learning_rate": 1.4159106114806943e-06,251      "loss": 1.4677734375,252      "step": 30253    },254    {255      "epoch": 0.9393939393939394,256      "grad_norm": 0.4563347399234772,257      "learning_rate": 1.2355607296773896e-06,258      "loss": 1.54833984375,259      "step": 31260    },261    {262      "epoch": 0.9696969696969697,263      "grad_norm": 0.48352736234664917,264      "learning_rate": 1.1052074993046102e-06,265      "loss": 1.44287109375,266      "step": 32267    },268    {269      "epoch": 1.0,270      "grad_norm": 0.5558485984802246,271      "learning_rate": 1.026379192805382e-06,272      "loss": 1.48291015625,273      "step": 33274    },275    {276      "epoch": 1.0,277      "eval_loss": 1.62109375,278      "eval_runtime": 2.5478,279      "eval_samples_per_second": 0.785,280      "eval_steps_per_second": 0.393,281      "step": 33282    }283  ],284  "logging_steps": 1.0,285  "max_steps": 33,286  "num_input_tokens_seen": 0,287  "num_train_epochs": 1,288  "save_steps": 0,289  "stateful_callbacks": {290    "TrainerControl": {291      "args": {292        "should_epoch_stop": false,293        "should_evaluate": false,294        "should_log": false,295        "should_save": true,296        "should_training_stop": true297      },298      "attributes": {}299    }300  },301  "total_flos": 6.715828683292017e+17,302  "train_batch_size": 1,303  "trial_name": null,304  "trial_params": null305}306