Team Ai
Modelpublic

Isotr0py/phi-2-test-sql-lora

sourceHugging Faceupdated 2y agoView on Hugging Face
2likes37downloads
trainer_state.json424 linesDownload Raw Back to root
1{2  "best_metric": null,3  "best_model_checkpoint": null,4  "epoch": 0.1,5  "eval_steps": 50,6  "global_step": 250,7  "is_hyper_param_search": false,8  "is_local_process_zero": true,9  "is_world_process_zero": true,10  "log_history": [11    {12      "epoch": 0.002,13      "grad_norm": 0.2585286796092987,14      "learning_rate": 0.0001,15      "loss": 0.9729,16      "step": 517    },18    {19      "epoch": 0.004,20      "grad_norm": 0.3368561565876007,21      "learning_rate": 0.0002,22      "loss": 0.9899,23      "step": 1024    },25    {26      "epoch": 0.006,27      "grad_norm": 0.499980092048645,28      "learning_rate": 0.00019583333333333334,29      "loss": 0.7853,30      "step": 1531    },32    {33      "epoch": 0.008,34      "grad_norm": 0.3696683347225189,35      "learning_rate": 0.00019166666666666667,36      "loss": 0.6654,37      "step": 2038    },39    {40      "epoch": 0.01,41      "grad_norm": 0.40797653794288635,42      "learning_rate": 0.0001875,43      "loss": 0.6311,44      "step": 2545    },46    {47      "epoch": 0.012,48      "grad_norm": 0.33503052592277527,49      "learning_rate": 0.00018333333333333334,50      "loss": 0.6248,51      "step": 3052    },53    {54      "epoch": 0.014,55      "grad_norm": 0.3338172733783722,56      "learning_rate": 0.0001791666666666667,57      "loss": 0.6046,58      "step": 3559    },60    {61      "epoch": 0.016,62      "grad_norm": 0.3330083191394806,63      "learning_rate": 0.000175,64      "loss": 0.6173,65      "step": 4066    },67    {68      "epoch": 0.018,69      "grad_norm": 0.3610495328903198,70      "learning_rate": 0.00017083333333333333,71      "loss": 0.6598,72      "step": 4573    },74    {75      "epoch": 0.02,76      "grad_norm": 0.5117771625518799,77      "learning_rate": 0.0001666666666666667,78      "loss": 0.701,79      "step": 5080    },81    {82      "epoch": 0.02,83      "eval_loss": 0.5943231582641602,84      "eval_runtime": 1078.9543,85      "eval_samples_per_second": 5.423,86      "eval_steps_per_second": 0.678,87      "step": 5088    },89    {90      "epoch": 0.022,91      "grad_norm": 0.24727080762386322,92      "learning_rate": 0.00016250000000000002,93      "loss": 0.5523,94      "step": 5595    },96    {97      "epoch": 0.024,98      "grad_norm": 0.32829391956329346,99      "learning_rate": 0.00015833333333333332,100      "loss": 0.5601,101      "step": 60102    },103    {104      "epoch": 0.026,105      "grad_norm": 0.2365810126066208,106      "learning_rate": 0.00015416666666666668,107      "loss": 0.5529,108      "step": 65109    },110    {111      "epoch": 0.028,112      "grad_norm": 0.23763389885425568,113      "learning_rate": 0.00015000000000000001,114      "loss": 0.5683,115      "step": 70116    },117    {118      "epoch": 0.03,119      "grad_norm": 0.2370971143245697,120      "learning_rate": 0.00014583333333333335,121      "loss": 0.5734,122      "step": 75123    },124    {125      "epoch": 0.032,126      "grad_norm": 0.23187637329101562,127      "learning_rate": 0.00014166666666666668,128      "loss": 0.5589,129      "step": 80130    },131    {132      "epoch": 0.034,133      "grad_norm": 0.2783190906047821,134      "learning_rate": 0.0001375,135      "loss": 0.5833,136      "step": 85137    },138    {139      "epoch": 0.036,140      "grad_norm": 0.2846642732620239,141      "learning_rate": 0.00013333333333333334,142      "loss": 0.5855,143      "step": 90144    },145    {146      "epoch": 0.038,147      "grad_norm": 0.30239176750183105,148      "learning_rate": 0.00012916666666666667,149      "loss": 0.6083,150      "step": 95151    },152    {153      "epoch": 0.04,154      "grad_norm": 0.4522075057029724,155      "learning_rate": 0.000125,156      "loss": 0.6666,157      "step": 100158    },159    {160      "epoch": 0.04,161      "eval_loss": 0.5652860403060913,162      "eval_runtime": 1078.4557,163      "eval_samples_per_second": 5.425,164      "eval_steps_per_second": 0.679,165      "step": 100166    },167    {168      "epoch": 0.042,169      "grad_norm": 0.34593072533607483,170      "learning_rate": 0.00012083333333333333,171      "loss": 0.5283,172      "step": 105173    },174    {175      "epoch": 0.044,176      "grad_norm": 0.24053940176963806,177      "learning_rate": 0.00011666666666666668,178      "loss": 0.5376,179      "step": 110180    },181    {182      "epoch": 0.046,183      "grad_norm": 0.25374647974967957,184      "learning_rate": 0.00011250000000000001,185      "loss": 0.5268,186      "step": 115187    },188    {189      "epoch": 0.048,190      "grad_norm": 0.2413879930973053,191      "learning_rate": 0.00010833333333333333,192      "loss": 0.5466,193      "step": 120194    },195    {196      "epoch": 0.05,197      "grad_norm": 0.22842754423618317,198      "learning_rate": 0.00010416666666666667,199      "loss": 0.558,200      "step": 125201    },202    {203      "epoch": 0.052,204      "grad_norm": 0.23777280747890472,205      "learning_rate": 0.0001,206      "loss": 0.5304,207      "step": 130208    },209    {210      "epoch": 0.054,211      "grad_norm": 0.2853969633579254,212      "learning_rate": 9.583333333333334e-05,213      "loss": 0.5773,214      "step": 135215    },216    {217      "epoch": 0.056,218      "grad_norm": 0.28491005301475525,219      "learning_rate": 9.166666666666667e-05,220      "loss": 0.5898,221      "step": 140222    },223    {224      "epoch": 0.058,225      "grad_norm": 0.3216317892074585,226      "learning_rate": 8.75e-05,227      "loss": 0.6376,228      "step": 145229    },230    {231      "epoch": 0.06,232      "grad_norm": 0.47678303718566895,233      "learning_rate": 8.333333333333334e-05,234      "loss": 0.6448,235      "step": 150236    },237    {238      "epoch": 0.06,239      "eval_loss": 0.5443492531776428,240      "eval_runtime": 1078.9513,241      "eval_samples_per_second": 5.423,242      "eval_steps_per_second": 0.678,243      "step": 150244    },245    {246      "epoch": 0.062,247      "grad_norm": 0.3121783435344696,248      "learning_rate": 7.916666666666666e-05,249      "loss": 0.5081,250      "step": 155251    },252    {253      "epoch": 0.064,254      "grad_norm": 0.2449740618467331,255      "learning_rate": 7.500000000000001e-05,256      "loss": 0.511,257      "step": 160258    },259    {260      "epoch": 0.066,261      "grad_norm": 0.2097545564174652,262      "learning_rate": 7.083333333333334e-05,263      "loss": 0.509,264      "step": 165265    },266    {267      "epoch": 0.068,268      "grad_norm": 0.23882268369197845,269      "learning_rate": 6.666666666666667e-05,270      "loss": 0.5411,271      "step": 170272    },273    {274      "epoch": 0.07,275      "grad_norm": 0.26276838779449463,276      "learning_rate": 6.25e-05,277      "loss": 0.5539,278      "step": 175279    },280    {281      "epoch": 0.072,282      "grad_norm": 0.26204946637153625,283      "learning_rate": 5.833333333333334e-05,284      "loss": 0.5692,285      "step": 180286    },287    {288      "epoch": 0.074,289      "grad_norm": 0.27385151386260986,290      "learning_rate": 5.4166666666666664e-05,291      "loss": 0.5599,292      "step": 185293    },294    {295      "epoch": 0.076,296      "grad_norm": 0.3151029646396637,297      "learning_rate": 5e-05,298      "loss": 0.5622,299      "step": 190300    },301    {302      "epoch": 0.078,303      "grad_norm": 0.372055321931839,304      "learning_rate": 4.5833333333333334e-05,305      "loss": 0.6128,306      "step": 195307    },308    {309      "epoch": 0.08,310      "grad_norm": 0.4943559765815735,311      "learning_rate": 4.166666666666667e-05,312      "loss": 0.6352,313      "step": 200314    },315    {316      "epoch": 0.08,317      "eval_loss": 0.5273013114929199,318      "eval_runtime": 1079.0525,319      "eval_samples_per_second": 5.422,320      "eval_steps_per_second": 0.678,321      "step": 200322    },323    {324      "epoch": 0.082,325      "grad_norm": 0.28252649307250977,326      "learning_rate": 3.7500000000000003e-05,327      "loss": 0.4963,328      "step": 205329    },330    {331      "epoch": 0.084,332      "grad_norm": 0.2909074127674103,333      "learning_rate": 3.3333333333333335e-05,334      "loss": 0.5165,335      "step": 210336    },337    {338      "epoch": 0.086,339      "grad_norm": 0.2641913890838623,340      "learning_rate": 2.916666666666667e-05,341      "loss": 0.5264,342      "step": 215343    },344    {345      "epoch": 0.088,346      "grad_norm": 0.2438758909702301,347      "learning_rate": 2.5e-05,348      "loss": 0.525,349      "step": 220350    },351    {352      "epoch": 0.09,353      "grad_norm": 0.2509464621543884,354      "learning_rate": 2.0833333333333336e-05,355      "loss": 0.5451,356      "step": 225357    },358    {359      "epoch": 0.092,360      "grad_norm": 0.26488572359085083,361      "learning_rate": 1.6666666666666667e-05,362      "loss": 0.5512,363      "step": 230364    },365    {366      "epoch": 0.094,367      "grad_norm": 0.2820191979408264,368      "learning_rate": 1.25e-05,369      "loss": 0.548,370      "step": 235371    },372    {373      "epoch": 0.096,374      "grad_norm": 0.3010896146297455,375      "learning_rate": 8.333333333333334e-06,376      "loss": 0.578,377      "step": 240378    },379    {380      "epoch": 0.098,381      "grad_norm": 0.38558459281921387,382      "learning_rate": 4.166666666666667e-06,383      "loss": 0.5913,384      "step": 245385    },386    {387      "epoch": 0.1,388      "grad_norm": 0.6697375178337097,389      "learning_rate": 0.0,390      "loss": 0.6351,391      "step": 250392    },393    {394      "epoch": 0.1,395      "eval_loss": 0.5142252445220947,396      "eval_runtime": 1078.9757,397      "eval_samples_per_second": 5.423,398      "eval_steps_per_second": 0.678,399      "step": 250400    }401  ],402  "logging_steps": 5,403  "max_steps": 250,404  "num_input_tokens_seen": 0,405  "num_train_epochs": 1,406  "save_steps": 50,407  "stateful_callbacks": {408    "TrainerControl": {409      "args": {410        "should_epoch_stop": false,411        "should_evaluate": false,412        "should_log": false,413        "should_save": true,414        "should_training_stop": true415      },416      "attributes": {}417    }418  },419  "total_flos": 2.64628687263744e+16,420  "train_batch_size": 20,421  "trial_name": null,422  "trial_params": null423}424