CoolFace
Modelpublic

YOZ1/RadAI4

sourceHugging Faceupdated 3y agoView on Hugging Face
0likes4downloads
trainer_state.json401 linesDownload Raw Back to root
1{2  "best_metric": null,3  "best_model_checkpoint": null,4  "epoch": 14.349775784753364,5  "global_step": 3200,6  "is_hyper_param_search": false,7  "is_local_process_zero": true,8  "is_world_process_zero": true,9  "log_history": [10    {11      "epoch": 0.22,12      "learning_rate": 0.0002,13      "loss": 0.6202,14      "step": 5015    },16    {17      "epoch": 0.45,18      "learning_rate": 0.0002,19      "loss": 0.4487,20      "step": 10021    },22    {23      "epoch": 0.67,24      "learning_rate": 0.0002,25      "loss": 0.419,26      "step": 15027    },28    {29      "epoch": 0.9,30      "learning_rate": 0.0002,31      "loss": 0.4151,32      "step": 20033    },34    {35      "epoch": 1.12,36      "learning_rate": 0.0002,37      "loss": 0.3933,38      "step": 25039    },40    {41      "epoch": 1.35,42      "learning_rate": 0.0002,43      "loss": 0.3797,44      "step": 30045    },46    {47      "epoch": 1.57,48      "learning_rate": 0.0002,49      "loss": 0.3688,50      "step": 35051    },52    {53      "epoch": 1.79,54      "learning_rate": 0.0002,55      "loss": 0.3749,56      "step": 40057    },58    {59      "epoch": 2.02,60      "learning_rate": 0.0002,61      "loss": 0.3624,62      "step": 45063    },64    {65      "epoch": 2.24,66      "learning_rate": 0.0002,67      "loss": 0.3331,68      "step": 50069    },70    {71      "epoch": 2.47,72      "learning_rate": 0.0002,73      "loss": 0.3347,74      "step": 55075    },76    {77      "epoch": 2.69,78      "learning_rate": 0.0002,79      "loss": 0.3533,80      "step": 60081    },82    {83      "epoch": 2.91,84      "learning_rate": 0.0002,85      "loss": 0.3413,86      "step": 65087    },88    {89      "epoch": 3.14,90      "learning_rate": 0.0002,91      "loss": 0.3242,92      "step": 70093    },94    {95      "epoch": 3.36,96      "learning_rate": 0.0002,97      "loss": 0.3091,98      "step": 75099    },100    {101      "epoch": 3.59,102      "learning_rate": 0.0002,103      "loss": 0.3166,104      "step": 800105    },106    {107      "epoch": 3.81,108      "learning_rate": 0.0002,109      "loss": 0.3121,110      "step": 850111    },112    {113      "epoch": 4.04,114      "learning_rate": 0.0002,115      "loss": 0.304,116      "step": 900117    },118    {119      "epoch": 4.26,120      "learning_rate": 0.0002,121      "loss": 0.2776,122      "step": 950123    },124    {125      "epoch": 4.48,126      "learning_rate": 0.0002,127      "loss": 0.2783,128      "step": 1000129    },130    {131      "epoch": 4.71,132      "learning_rate": 0.0002,133      "loss": 0.2836,134      "step": 1050135    },136    {137      "epoch": 4.93,138      "learning_rate": 0.0002,139      "loss": 0.2931,140      "step": 1100141    },142    {143      "epoch": 5.16,144      "learning_rate": 0.0002,145      "loss": 0.2589,146      "step": 1150147    },148    {149      "epoch": 5.38,150      "learning_rate": 0.0002,151      "loss": 0.2525,152      "step": 1200153    },154    {155      "epoch": 5.61,156      "learning_rate": 0.0002,157      "loss": 0.255,158      "step": 1250159    },160    {161      "epoch": 5.83,162      "learning_rate": 0.0002,163      "loss": 0.2564,164      "step": 1300165    },166    {167      "epoch": 6.05,168      "learning_rate": 0.0002,169      "loss": 0.242,170      "step": 1350171    },172    {173      "epoch": 6.28,174      "learning_rate": 0.0002,175      "loss": 0.2092,176      "step": 1400177    },178    {179      "epoch": 6.5,180      "learning_rate": 0.0002,181      "loss": 0.2172,182      "step": 1450183    },184    {185      "epoch": 6.73,186      "learning_rate": 0.0002,187      "loss": 0.2233,188      "step": 1500189    },190    {191      "epoch": 6.95,192      "learning_rate": 0.0002,193      "loss": 0.2261,194      "step": 1550195    },196    {197      "epoch": 7.17,198      "learning_rate": 0.0002,199      "loss": 0.1944,200      "step": 1600201    },202    {203      "epoch": 7.4,204      "learning_rate": 0.0002,205      "loss": 0.1803,206      "step": 1650207    },208    {209      "epoch": 7.62,210      "learning_rate": 0.0002,211      "loss": 0.1852,212      "step": 1700213    },214    {215      "epoch": 7.85,216      "learning_rate": 0.0002,217      "loss": 0.1931,218      "step": 1750219    },220    {221      "epoch": 8.07,222      "learning_rate": 0.0002,223      "loss": 0.1755,224      "step": 1800225    },226    {227      "epoch": 8.3,228      "learning_rate": 0.0002,229      "loss": 0.1465,230      "step": 1850231    },232    {233      "epoch": 8.52,234      "learning_rate": 0.0002,235      "loss": 0.1535,236      "step": 1900237    },238    {239      "epoch": 8.74,240      "learning_rate": 0.0002,241      "loss": 0.1542,242      "step": 1950243    },244    {245      "epoch": 8.97,246      "learning_rate": 0.0002,247      "loss": 0.161,248      "step": 2000249    },250    {251      "epoch": 9.19,252      "learning_rate": 0.0002,253      "loss": 0.1205,254      "step": 2050255    },256    {257      "epoch": 9.42,258      "learning_rate": 0.0002,259      "loss": 0.1197,260      "step": 2100261    },262    {263      "epoch": 9.64,264      "learning_rate": 0.0002,265      "loss": 0.1242,266      "step": 2150267    },268    {269      "epoch": 9.87,270      "learning_rate": 0.0002,271      "loss": 0.1277,272      "step": 2200273    },274    {275      "epoch": 10.09,276      "learning_rate": 0.0002,277      "loss": 0.1151,278      "step": 2250279    },280    {281      "epoch": 10.31,282      "learning_rate": 0.0002,283      "loss": 0.0925,284      "step": 2300285    },286    {287      "epoch": 10.54,288      "learning_rate": 0.0002,289      "loss": 0.0969,290      "step": 2350291    },292    {293      "epoch": 10.76,294      "learning_rate": 0.0002,295      "loss": 0.1023,296      "step": 2400297    },298    {299      "epoch": 10.99,300      "learning_rate": 0.0002,301      "loss": 0.1016,302      "step": 2450303    },304    {305      "epoch": 11.21,306      "learning_rate": 0.0002,307      "loss": 0.0708,308      "step": 2500309    },310    {311      "epoch": 11.43,312      "learning_rate": 0.0002,313      "loss": 0.0735,314      "step": 2550315    },316    {317      "epoch": 11.66,318      "learning_rate": 0.0002,319      "loss": 0.0804,320      "step": 2600321    },322    {323      "epoch": 11.88,324      "learning_rate": 0.0002,325      "loss": 0.0833,326      "step": 2650327    },328    {329      "epoch": 12.11,330      "learning_rate": 0.0002,331      "loss": 0.072,332      "step": 2700333    },334    {335      "epoch": 12.33,336      "learning_rate": 0.0002,337      "loss": 0.0592,338      "step": 2750339    },340    {341      "epoch": 12.56,342      "learning_rate": 0.0002,343      "loss": 0.0585,344      "step": 2800345    },346    {347      "epoch": 12.78,348      "learning_rate": 0.0002,349      "loss": 0.0647,350      "step": 2850351    },352    {353      "epoch": 13.0,354      "learning_rate": 0.0002,355      "loss": 0.0656,356      "step": 2900357    },358    {359      "epoch": 13.23,360      "learning_rate": 0.0002,361      "loss": 0.0473,362      "step": 2950363    },364    {365      "epoch": 13.45,366      "learning_rate": 0.0002,367      "loss": 0.0509,368      "step": 3000369    },370    {371      "epoch": 13.68,372      "learning_rate": 0.0002,373      "loss": 0.0523,374      "step": 3050375    },376    {377      "epoch": 13.9,378      "learning_rate": 0.0002,379      "loss": 0.0531,380      "step": 3100381    },382    {383      "epoch": 14.13,384      "learning_rate": 0.0002,385      "loss": 0.0455,386      "step": 3150387    },388    {389      "epoch": 14.35,390      "learning_rate": 0.0002,391      "loss": 0.0407,392      "step": 3200393    }394  ],395  "max_steps": 3345,396  "num_train_epochs": 15,397  "total_flos": 5.0781774924831744e+17,398  "trial_name": null,399  "trial_params": null400}401