CoolFace
Modelpublic

marid/sst2

sourceHugging Faceapache-2.0updated 2y agoView on Hugging Face
0likes2downloads
trainer_state.json347 linesDownload Raw Back to root
1{2  "best_metric": 0.4493387111903199,3  "best_model_checkpoint": "/workspace/cv-corpus-8.0-2022-01-19/output/checkpoint-1000",4  "epoch": 10.0,5  "global_step": 15630,6  "is_hyper_param_search": false,7  "is_local_process_zero": true,8  "is_world_process_zero": true,9  "log_history": [10    {11      "epoch": 0.32,12      "learning_rate": 9.92e-05,13      "loss": 6.7916,14      "step": 50015    },16    {17      "epoch": 0.64,18      "learning_rate": 0.00019920000000000002,19      "loss": 1.354,20      "step": 100021    },22    {23      "epoch": 0.64,24      "eval_loss": 0.4108898937702179,25      "eval_runtime": 71.0896,26      "eval_samples_per_second": 146.125,27      "eval_steps_per_second": 1.829,28      "eval_wer": 0.4493387111903199,29      "step": 100030    },31    {32      "epoch": 0.96,33      "learning_rate": 0.00019321941216678059,34      "loss": 0.7084,35      "step": 150036    },37    {38      "epoch": 1.28,39      "learning_rate": 0.00018638414217361587,40      "loss": 0.5886,41      "step": 200042    },43    {44      "epoch": 1.28,45      "eval_loss": 0.2797781527042389,46      "eval_runtime": 71.0704,47      "eval_samples_per_second": 146.165,48      "eval_steps_per_second": 1.829,49      "eval_wer": 0.3099334021198762,50      "step": 200051    },52    {53      "epoch": 1.6,54      "learning_rate": 0.00017954887218045112,55      "loss": 0.5386,56      "step": 250057    },58    {59      "epoch": 1.92,60      "learning_rate": 0.0001727136021872864,61      "loss": 0.4977,62      "step": 300063    },64    {65      "epoch": 1.92,66      "eval_loss": 0.23867885768413544,67      "eval_runtime": 71.0833,68      "eval_samples_per_second": 146.138,69      "eval_steps_per_second": 1.829,70      "eval_wer": 0.2673295188068662,71      "step": 300072    },73    {74      "epoch": 2.24,75      "learning_rate": 0.00016587833219412168,76      "loss": 0.4531,77      "step": 350078    },79    {80      "epoch": 2.56,81      "learning_rate": 0.00015905673274094328,82      "loss": 0.4253,83      "step": 400084    },85    {86      "epoch": 2.56,87      "eval_loss": 0.22657370567321777,88      "eval_runtime": 71.5875,89      "eval_samples_per_second": 145.109,90      "eval_steps_per_second": 1.816,91      "eval_wer": 0.2523215458212175,92      "step": 400093    },94    {95      "epoch": 2.88,96      "learning_rate": 0.00015222146274777856,97      "loss": 0.413,98      "step": 450099    },100    {101      "epoch": 3.2,102      "learning_rate": 0.0001453861927546138,103      "loss": 0.3942,104      "step": 5000105    },106    {107      "epoch": 3.2,108      "eval_loss": 0.21706202626228333,109      "eval_runtime": 72.0993,110      "eval_samples_per_second": 144.079,111      "eval_steps_per_second": 1.803,112      "eval_wer": 0.2437294812869337,113      "step": 5000114    },115    {116      "epoch": 3.52,117      "learning_rate": 0.00013855092276144907,118      "loss": 0.3741,119      "step": 5500120    },121    {122      "epoch": 3.84,123      "learning_rate": 0.00013171565276828435,124      "loss": 0.3619,125      "step": 6000126    },127    {128      "epoch": 3.84,129      "eval_loss": 0.20762862265110016,130      "eval_runtime": 72.0546,131      "eval_samples_per_second": 144.168,132      "eval_steps_per_second": 1.804,133      "eval_wer": 0.22530719444705,134      "step": 6000135    },136    {137      "epoch": 4.16,138      "learning_rate": 0.00012489405331510595,139      "loss": 0.3435,140      "step": 6500141    },142    {143      "epoch": 4.48,144      "learning_rate": 0.00011805878332194122,145      "loss": 0.3245,146      "step": 7000147    },148    {149      "epoch": 4.48,150      "eval_loss": 0.2087564468383789,151      "eval_runtime": 71.9965,152      "eval_samples_per_second": 144.285,153      "eval_steps_per_second": 1.806,154      "eval_wer": 0.21862864646843636,155      "step": 7000156    },157    {158      "epoch": 4.8,159      "learning_rate": 0.0001112235133287765,160      "loss": 0.3135,161      "step": 7500162    },163    {164      "epoch": 5.12,165      "learning_rate": 0.0001044019138755981,166      "loss": 0.308,167      "step": 8000168    },169    {170      "epoch": 5.12,171      "eval_loss": 0.2086208015680313,172      "eval_runtime": 68.9232,173      "eval_samples_per_second": 150.718,174      "eval_steps_per_second": 1.886,175      "eval_wer": 0.22063596285526685,176      "step": 8000177    },178    {179      "epoch": 5.44,180      "learning_rate": 9.756664388243337e-05,181      "loss": 0.292,182      "step": 8500183    },184    {185      "epoch": 5.76,186      "learning_rate": 9.073137388926864e-05,187      "loss": 0.2881,188      "step": 9000189    },190    {191      "epoch": 5.76,192      "eval_loss": 0.20888157188892365,193      "eval_runtime": 70.2391,194      "eval_samples_per_second": 147.895,195      "eval_steps_per_second": 1.851,196      "eval_wer": 0.21048682112372197,197      "step": 9000198    },199    {200      "epoch": 6.08,201      "learning_rate": 8.389610389610389e-05,202      "loss": 0.2717,203      "step": 9500204    },205    {206      "epoch": 6.4,207      "learning_rate": 7.706083390293917e-05,208      "loss": 0.2557,209      "step": 10000210    },211    {212      "epoch": 6.4,213      "eval_loss": 0.20148096978664398,214      "eval_runtime": 70.1726,215      "eval_samples_per_second": 148.035,216      "eval_steps_per_second": 1.853,217      "eval_wer": 0.20035643935840916,218      "step": 10000219    },220    {221      "epoch": 6.72,222      "learning_rate": 7.022556390977444e-05,223      "loss": 0.2536,224      "step": 10500225    },226    {227      "epoch": 7.04,228      "learning_rate": 6.33902939166097e-05,229      "loss": 0.248,230      "step": 11000231    },232    {233      "epoch": 7.04,234      "eval_loss": 0.2043762356042862,235      "eval_runtime": 70.2184,236      "eval_samples_per_second": 147.938,237      "eval_steps_per_second": 1.851,238      "eval_wer": 0.19529124847575274,239      "step": 11000240    },241    {242      "epoch": 7.36,243      "learning_rate": 5.655502392344498e-05,244      "loss": 0.2308,245      "step": 11500246    },247    {248      "epoch": 7.68,249      "learning_rate": 4.971975393028025e-05,250      "loss": 0.2251,251      "step": 12000252    },253    {254      "epoch": 7.68,255      "eval_loss": 0.20575200021266937,256      "eval_runtime": 70.8946,257      "eval_samples_per_second": 146.527,258      "eval_steps_per_second": 1.834,259      "eval_wer": 0.19315261232529782,260      "step": 12000261    },262    {263      "epoch": 8.0,264      "learning_rate": 4.291182501708818e-05,265      "loss": 0.2207,266      "step": 12500267    },268    {269      "epoch": 8.32,270      "learning_rate": 3.6076555023923446e-05,271      "loss": 0.2052,272      "step": 13000273    },274    {275      "epoch": 8.32,276      "eval_loss": 0.21170856058597565,277      "eval_runtime": 70.7957,278      "eval_samples_per_second": 146.732,279      "eval_steps_per_second": 1.836,280      "eval_wer": 0.18776850201669637,281      "step": 13000282    },283    {284      "epoch": 8.64,285      "learning_rate": 2.9241285030758714e-05,286      "loss": 0.2026,287      "step": 13500288    },289    {290      "epoch": 8.96,291      "learning_rate": 2.2406015037593985e-05,292      "loss": 0.1976,293      "step": 14000294    },295    {296      "epoch": 8.96,297      "eval_loss": 0.21043309569358826,298      "eval_runtime": 71.1895,299      "eval_samples_per_second": 145.92,300      "eval_steps_per_second": 1.826,301      "eval_wer": 0.18249695150548728,302      "step": 14000303    },304    {305      "epoch": 9.28,306      "learning_rate": 1.5570745044429256e-05,307      "loss": 0.1875,308      "step": 14500309    },310    {311      "epoch": 9.6,312      "learning_rate": 8.735475051264526e-06,313      "loss": 0.1845,314      "step": 15000315    },316    {317      "epoch": 9.6,318      "eval_loss": 0.21563756465911865,319      "eval_runtime": 71.0722,320      "eval_samples_per_second": 146.161,321      "eval_steps_per_second": 1.829,322      "eval_wer": 0.18212175218084609,323      "step": 15000324    },325    {326      "epoch": 9.92,327      "learning_rate": 1.9138755980861244e-06,328      "loss": 0.1837,329      "step": 15500330    },331    {332      "epoch": 10.0,333      "step": 15630,334      "total_flos": 9.942412569719006e+20,335      "train_loss": 0.1316310991176183,336      "train_runtime": 23113.6031,337      "train_samples_per_second": 173.054,338      "train_steps_per_second": 0.676339    }340  ],341  "max_steps": 15630,342  "num_train_epochs": 10,343  "total_flos": 9.942412569719006e+20,344  "trial_name": null,345  "trial_params": null346}347