CoolFace
Modelpublic

Azfarhashmi/adaption_market_analysis_reasoning

sourceHugging Faceotherupdated 2mo agoView on Hugging Face
0likes9downloads
trainer_state.json516 linesDownload Raw Back to root
1{2  "best_global_step": null,3  "best_metric": null,4  "best_model_checkpoint": null,5  "epoch": 3.0,6  "eval_steps": 12,7  "global_step": 63,8  "is_hyper_param_search": false,9  "is_local_process_zero": true,10  "is_world_process_zero": true,11  "log_history": [12    {13      "epoch": 0.047619047619047616,14      "grad_norm": 0.7285156846046448,15      "learning_rate": 0.0,16      "loss": 2.265625,17      "step": 118    },19    {20      "epoch": 0.09523809523809523,21      "grad_norm": 0.8744295835494995,22      "learning_rate": 2.5e-05,23      "loss": 2.2958984375,24      "step": 225    },26    {27      "epoch": 0.14285714285714285,28      "grad_norm": 0.707972526550293,29      "learning_rate": 5e-05,30      "loss": 2.197509765625,31      "step": 332    },33    {34      "epoch": 0.19047619047619047,35      "grad_norm": 0.47509288787841797,36      "learning_rate": 7.500000000000001e-05,37      "loss": 1.9765625,38      "step": 439    },40    {41      "epoch": 0.23809523809523808,42      "grad_norm": 0.15270085632801056,43      "learning_rate": 0.0001,44      "loss": 1.84521484375,45      "step": 546    },47    {48      "epoch": 0.2857142857142857,49      "grad_norm": 0.10096657276153564,50      "learning_rate": 9.99362213054543e-05,51      "loss": 1.78857421875,52      "step": 653    },54    {55      "epoch": 0.3333333333333333,56      "grad_norm": 0.11879778653383255,57      "learning_rate": 9.974506600945618e-05,58      "loss": 1.83447265625,59      "step": 760    },61    {62      "epoch": 0.38095238095238093,63      "grad_norm": 0.1394820213317871,64      "learning_rate": 9.942707596246052e-05,65      "loss": 1.79443359375,66      "step": 867    },68    {69      "epoch": 0.42857142857142855,70      "grad_norm": 0.12889999151229858,71      "learning_rate": 9.8983152541804e-05,72      "loss": 1.72900390625,73      "step": 974    },75    {76      "epoch": 0.47619047619047616,77      "grad_norm": 0.11798744648694992,78      "learning_rate": 9.841455409665323e-05,79      "loss": 1.6611328125,80      "step": 1081    },82    {83      "epoch": 0.5238095238095238,84      "grad_norm": 0.10778991132974625,85      "learning_rate": 9.772289238107715e-05,86      "loss": 1.657958984375,87      "step": 1188    },89    {90      "epoch": 0.5714285714285714,91      "grad_norm": 0.11877302080392838,92      "learning_rate": 9.691012798535524e-05,93      "loss": 1.617919921875,94      "step": 1295    },96    {97      "epoch": 0.6190476190476191,98      "grad_norm": 0.11993581056594849,99      "learning_rate": 9.59785647784711e-05,100      "loss": 1.664794921875,101      "step": 13102    },103    {104      "epoch": 0.6666666666666666,105      "grad_norm": 0.10702720284461975,106      "learning_rate": 9.493084337754573e-05,107      "loss": 1.51708984375,108      "step": 14109    },110    {111      "epoch": 0.7142857142857143,112      "grad_norm": 0.0843004509806633,113      "learning_rate": 9.376993366272128e-05,114      "loss": 1.591796875,115      "step": 15116    },117    {118      "epoch": 0.7142857142857143,119      "eval_loss": 1.55517578125,120      "eval_runtime": 5.081,121      "eval_samples_per_second": 0.787,122      "eval_steps_per_second": 0.197,123      "step": 15124    },125    {126      "epoch": 0.7619047619047619,127      "grad_norm": 0.07435919344425201,128      "learning_rate": 9.249912635871318e-05,129      "loss": 1.574951171875,130      "step": 16131    },132    {133      "epoch": 0.8095238095238095,134      "grad_norm": 0.0768747329711914,135      "learning_rate": 9.112202370689337e-05,136      "loss": 1.6044921875,137      "step": 17138    },139    {140      "epoch": 0.8571428571428571,141      "grad_norm": 0.09450701624155045,142      "learning_rate": 8.964252925434579e-05,143      "loss": 1.5732421875,144      "step": 18145    },146    {147      "epoch": 0.9047619047619048,148      "grad_norm": 0.08037487417459488,149      "learning_rate": 8.806483678883803e-05,150      "loss": 1.53759765625,151      "step": 19152    },153    {154      "epoch": 0.9523809523809523,155      "grad_norm": 0.08016908168792725,156      "learning_rate": 8.639341845107433e-05,157      "loss": 1.48974609375,158      "step": 20159    },160    {161      "epoch": 1.0,162      "grad_norm": 0.08489606529474258,163      "learning_rate": 8.463301205792675e-05,164      "loss": 1.484375,165      "step": 21166    },167    {168      "epoch": 1.0476190476190477,169      "grad_norm": 0.08712036162614822,170      "learning_rate": 8.278860767257865e-05,171      "loss": 1.51904296875,172      "step": 22173    },174    {175      "epoch": 1.0952380952380953,176      "grad_norm": 0.07777861505746841,177      "learning_rate": 8.086543345964832e-05,178      "loss": 1.52294921875,179      "step": 23180    },181    {182      "epoch": 1.1428571428571428,183      "grad_norm": 0.06942926347255707,184      "learning_rate": 7.886894086538841e-05,185      "loss": 1.4566650390625,186      "step": 24187    },188    {189      "epoch": 1.1904761904761905,190      "grad_norm": 0.06922606378793716,191      "learning_rate": 7.680478916496926e-05,192      "loss": 1.4736328125,193      "step": 25194    },195    {196      "epoch": 1.2380952380952381,197      "grad_norm": 0.10756262391805649,198      "learning_rate": 7.46788294206485e-05,199      "loss": 1.421630859375,200      "step": 26201    },202    {203      "epoch": 1.2857142857142856,204      "grad_norm": 0.0812968835234642,205      "learning_rate": 7.249708789629944e-05,206      "loss": 1.4189453125,207      "step": 27208    },209    {210      "epoch": 1.2857142857142856,211      "eval_loss": 1.43017578125,212      "eval_runtime": 5.061,213      "eval_samples_per_second": 0.79,214      "eval_steps_per_second": 0.198,215      "step": 27216    },217    {218      "epoch": 1.3333333333333333,219      "grad_norm": 0.06487224251031876,220      "learning_rate": 7.026574897531137e-05,221      "loss": 1.484619140625,222      "step": 28223    },224    {225      "epoch": 1.380952380952381,226      "grad_norm": 0.11446287482976913,227      "learning_rate": 6.799113763028295e-05,228      "loss": 1.4599609375,229      "step": 29230    },231    {232      "epoch": 1.4285714285714286,233      "grad_norm": 0.061298832297325134,234      "learning_rate": 6.567970149420018e-05,235      "loss": 1.4345703125,236      "step": 30237    },238    {239      "epoch": 1.4761904761904763,240      "grad_norm": 0.06884041428565979,241      "learning_rate": 6.333799258392015e-05,242      "loss": 1.37255859375,243      "step": 31244    },245    {246      "epoch": 1.5238095238095237,247      "grad_norm": 0.06530533730983734,248      "learning_rate": 6.097264872776749e-05,249      "loss": 1.3865966796875,250      "step": 32251    },252    {253      "epoch": 1.5714285714285714,254      "grad_norm": 0.07631056010723114,255      "learning_rate": 5.859037474988874e-05,256      "loss": 1.355712890625,257      "step": 33258    },259    {260      "epoch": 1.619047619047619,261      "grad_norm": 0.07293955236673355,262      "learning_rate": 5.6197923464699875e-05,263      "loss": 1.431884765625,264      "step": 34265    },266    {267      "epoch": 1.6666666666666665,268      "grad_norm": 0.08120140433311462,269      "learning_rate": 5.3802076535300135e-05,270      "loss": 1.27734375,271      "step": 35272    },273    {274      "epoch": 1.7142857142857144,275      "grad_norm": 0.06899567693471909,276      "learning_rate": 5.140962525011126e-05,277      "loss": 1.392578125,278      "step": 36279    },280    {281      "epoch": 1.7619047619047619,282      "grad_norm": 0.06586313992738724,283      "learning_rate": 4.902735127223252e-05,284      "loss": 1.38916015625,285      "step": 37286    },287    {288      "epoch": 1.8095238095238095,289      "grad_norm": 0.0632728636264801,290      "learning_rate": 4.6662007416079864e-05,291      "loss": 1.43505859375,292      "step": 38293    },294    {295      "epoch": 1.8571428571428572,296      "grad_norm": 0.06368006765842438,297      "learning_rate": 4.432029850579983e-05,298      "loss": 1.4033203125,299      "step": 39300    },301    {302      "epoch": 1.8571428571428572,303      "eval_loss": 1.36083984375,304      "eval_runtime": 5.0652,305      "eval_samples_per_second": 0.79,306      "eval_steps_per_second": 0.197,307      "step": 39308    },309    {310      "epoch": 1.9047619047619047,311      "grad_norm": 0.06019105762243271,312      "learning_rate": 4.2008862369717074e-05,313      "loss": 1.37841796875,314      "step": 40315    },316    {317      "epoch": 1.9523809523809523,318      "grad_norm": 0.06852934509515762,319      "learning_rate": 3.973425102468864e-05,320      "loss": 1.324951171875,321      "step": 41322    },323    {324      "epoch": 2.0,325      "grad_norm": 0.07731995731592178,326      "learning_rate": 3.750291210370057e-05,327      "loss": 1.33251953125,328      "step": 42329    },330    {331      "epoch": 2.0476190476190474,332      "grad_norm": 0.06647676974534988,333      "learning_rate": 3.5321170579351516e-05,334      "loss": 1.37451171875,335      "step": 43336    },337    {338      "epoch": 2.0952380952380953,339      "grad_norm": 0.06389626860618591,340      "learning_rate": 3.319521083503075e-05,341      "loss": 1.386962890625,342      "step": 44343    },344    {345      "epoch": 2.142857142857143,346      "grad_norm": 0.06924697011709213,347      "learning_rate": 3.113105913461159e-05,348      "loss": 1.33056640625,349      "step": 45350    },351    {352      "epoch": 2.1904761904761907,353      "grad_norm": 0.06576833128929138,354      "learning_rate": 2.9134566540351693e-05,355      "loss": 1.358642578125,356      "step": 46357    },358    {359      "epoch": 2.238095238095238,360      "grad_norm": 0.06895660609006882,361      "learning_rate": 2.721139232742137e-05,362      "loss": 1.3031005859375,363      "step": 47364    },365    {366      "epoch": 2.2857142857142856,367      "grad_norm": 0.09544253349304199,368      "learning_rate": 2.5366987942073267e-05,369      "loss": 1.299560546875,370      "step": 48371    },372    {373      "epoch": 2.3333333333333335,374      "grad_norm": 0.06581106781959534,375      "learning_rate": 2.3606581548925695e-05,376      "loss": 1.381591796875,377      "step": 49378    },379    {380      "epoch": 2.380952380952381,381      "grad_norm": 0.06757261604070663,382      "learning_rate": 2.1935163211161985e-05,383      "loss": 1.36279296875,384      "step": 50385    },386    {387      "epoch": 2.4285714285714284,388      "grad_norm": 0.06766699999570847,389      "learning_rate": 2.0357470745654214e-05,390      "loss": 1.34130859375,391      "step": 51392    },393    {394      "epoch": 2.4285714285714284,395      "eval_loss": 1.328125,396      "eval_runtime": 5.0567,397      "eval_samples_per_second": 0.791,398      "eval_steps_per_second": 0.198,399      "step": 51400    },401    {402      "epoch": 2.4761904761904763,403      "grad_norm": 0.07995021343231201,404      "learning_rate": 1.8877976293106646e-05,405      "loss": 1.28076171875,406      "step": 52407    },408    {409      "epoch": 2.5238095238095237,410      "grad_norm": 0.06969735026359558,411      "learning_rate": 1.7500873641286825e-05,412      "loss": 1.30712890625,413      "step": 53414    },415    {416      "epoch": 2.571428571428571,417      "grad_norm": 0.06776302307844162,418      "learning_rate": 1.6230066337278724e-05,419      "loss": 1.2781982421875,420      "step": 54421    },422    {423      "epoch": 2.619047619047619,424      "grad_norm": 0.07185522466897964,425      "learning_rate": 1.5069156622454287e-05,426      "loss": 1.357177734375,427      "step": 55428    },429    {430      "epoch": 2.6666666666666665,431      "grad_norm": 0.07692070305347443,432      "learning_rate": 1.4021435221528906e-05,433      "loss": 1.1956787109375,434      "step": 56435    },436    {437      "epoch": 2.7142857142857144,438      "grad_norm": 0.06928940117359161,439      "learning_rate": 1.3089872014644771e-05,440      "loss": 1.32763671875,441      "step": 57442    },443    {444      "epoch": 2.761904761904762,445      "grad_norm": 0.06275974214076996,446      "learning_rate": 1.2277107618922843e-05,447      "loss": 1.3272705078125,448      "step": 58449    },450    {451      "epoch": 2.8095238095238093,452      "grad_norm": 0.07041693478822708,453      "learning_rate": 1.1585445903346783e-05,454      "loss": 1.37841796875,455      "step": 59456    },457    {458      "epoch": 2.857142857142857,459      "grad_norm": 0.06966356188058853,460      "learning_rate": 1.1016847458196001e-05,461      "loss": 1.3466796875,462      "step": 60463    },464    {465      "epoch": 2.9047619047619047,466      "grad_norm": 0.0679149404168129,467      "learning_rate": 1.0572924037539496e-05,468      "loss": 1.3291015625,469      "step": 61470    },471    {472      "epoch": 2.9523809523809526,473      "grad_norm": 0.06559661775827408,474      "learning_rate": 1.0254933990543831e-05,475      "loss": 1.273193359375,476      "step": 62477    },478    {479      "epoch": 3.0,480      "grad_norm": 0.0778159499168396,481      "learning_rate": 1.0063778694545714e-05,482      "loss": 1.28759765625,483      "step": 63484    },485    {486      "epoch": 3.0,487      "eval_loss": 1.3125,488      "eval_runtime": 5.0513,489      "eval_samples_per_second": 0.792,490      "eval_steps_per_second": 0.198,491      "step": 63492    }493  ],494  "logging_steps": 1.0,495  "max_steps": 63,496  "num_input_tokens_seen": 0,497  "num_train_epochs": 3,498  "save_steps": 0,499  "stateful_callbacks": {500    "TrainerControl": {501      "args": {502        "should_epoch_stop": false,503        "should_evaluate": false,504        "should_log": false,505        "should_save": true,506        "should_training_stop": true507      },508      "attributes": {}509    }510  },511  "total_flos": 2.53785675170159e+18,512  "train_batch_size": 1,513  "trial_name": null,514  "trial_params": null515}516