<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->
8_koelectra-base-v3-finetuned-korquad_augment_korquad-1_2_aihub-final
This model is a fine-tuned version of monologg/koelectra-base-v3-finetuned-korquad on the None dataset. It achieves the following results on the evaluation set:
- Exact Match: 68.4291
- F1: 83.2037
- Loss: 1.0459
Model description
More information needed
Intended uses & limitations
More information needed
Training and evaluation data
More information needed
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 1e-05
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 15
- mixed_precision_training: Native AMP
- label_smoothing_factor: 0.1
Training results
| Training Loss | Epoch | Step | Exact Match | F1 | Validation Loss |
|---|---|---|---|---|---|
| 1.7364 | 0.16 | 6000 | 61.8627 | 78.1626 | 1.2363 |
| 1.129 | 0.33 | 12000 | 64.0978 | 79.8857 | 1.1445 |
| 1.0497 | 0.49 | 18000 | 64.9327 | 80.4719 | 1.1084 |
| 1.0207 | 0.65 | 24000 | 65.2334 | 80.6901 | 1.0947 |
| 1.0206 | 0.82 | 30000 | 65.4443 | 80.9532 | 1.0732 |
| 1.0106 | 0.98 | 36000 | 65.7271 | 81.1378 | 1.0879 |
| 0.9736 | 1.15 | 42000 | 66.0189 | 81.3277 | 1.0820 |
| 0.9575 | 1.31 | 48000 | 66.1804 | 81.5178 | 1.0547 |
| 0.9448 | 1.47 | 54000 | 66.3330 | 81.6589 | 1.0430 |
| 0.9513 | 1.64 | 60000 | 66.5485 | 81.5404 | 1.0752 |
| 0.9463 | 1.8 | 66000 | 66.6607 | 81.7489 | 1.0430 |
| 0.9334 | 1.96 | 72000 | 66.9255 | 81.9937 | 1.0244 |
| 0.906 | 2.13 | 78000 | 66.9120 | 81.9072 | 1.0225 |
| 0.9177 | 2.29 | 84000 | 66.8671 | 82.0070 | 1.0342 |
| 0.9157 | 2.45 | 90000 | 67.0422 | 82.0179 | 1.0234 |
| 0.8941 | 2.62 | 96000 | 67.3250 | 82.2431 | 1.0176 |
| 0.896 | 2.78 | 102000 | 67.0781 | 82.0828 | 1.0195 |
| 0.8949 | 2.95 | 108000 | 67.3205 | 82.3027 | 1.0039 |
| 0.878 | 3.11 | 114000 | 67.3160 | 82.3084 | 1.0088 |
| 0.8676 | 3.27 | 120000 | 67.5224 | 82.3786 | 1.0137 |
| 0.8743 | 3.44 | 126000 | 67.5583 | 82.3468 | 1.0049 |
| 0.8779 | 3.6 | 132000 | 67.3968 | 82.3132 | 1.0146 |
| 0.872 | 3.76 | 138000 | 67.6212 | 82.4477 | 1.0098 |
| 0.868 | 3.93 | 144000 | 67.7244 | 82.5110 | 1.0068 |
| 0.8539 | 4.09 | 150000 | 67.7154 | 82.4795 | 1.0195 |
| 0.8434 | 4.25 | 156000 | 67.5718 | 82.4538 | 1.0166 |
| 0.8552 | 4.42 | 162000 | 67.9039 | 82.6227 | 1.0293 |
| 0.8443 | 4.58 | 168000 | 67.7513 | 82.6146 | 1.0039 |
| 0.8452 | 4.75 | 174000 | 67.8411 | 82.6710 | 1.0156 |
| 0.856 | 4.91 | 180000 | 67.8366 | 82.7559 | 0.9932 |
| 0.8402 | 5.07 | 186000 | 67.7558 | 82.5685 | 1.0146 |
| 0.8191 | 5.24 | 192000 | 67.9174 | 82.7676 | 1.0059 |
| 0.8216 | 5.4 | 198000 | 67.7917 | 82.6217 | 1.0186 |
| 0.8366 | 5.56 | 204000 | 67.7154 | 82.6135 | 1.0059 |
| 0.8341 | 5.73 | 210000 | 68.0341 | 82.7362 | 0.9985 |
| 0.8191 | 5.89 | 216000 | 67.9847 | 82.7773 | 1.0059 |
| 0.8114 | 6.05 | 222000 | 67.9578 | 82.7099 | 1.0 |
| 0.8117 | 6.22 | 228000 | 67.7469 | 82.6885 | 1.0020 |
| 0.8174 | 6.38 | 234000 | 68.0745 | 82.9214 | 1.0146 |
| 0.8094 | 6.55 | 240000 | 68.0296 | 82.9424 | 1.0107 |
| 0.7991 | 6.71 | 246000 | 67.9578 | 82.8998 | 1.0010 |
| 0.8193 | 6.87 | 252000 | 67.9803 | 82.9003 | 0.9951 |
| 0.8075 | 7.04 | 258000 | 68.0566 | 82.8525 | 0.9961 |
| 0.7876 | 7.2 | 264000 | 68.0835 | 82.8723 | 1.0098 |
| 0.7874 | 7.36 | 270000 | 68.1822 | 82.8987 | 1.0195 |
| 0.7941 | 7.53 | 276000 | 68.1912 | 82.9071 | 1.0107 |
| 0.8024 | 7.69 | 282000 | 68.2989 | 83.0104 | 1.0137 |
| 0.7872 | 7.85 | 288000 | 68.3124 | 83.0462 | 1.0 |
| 0.7894 | 8.02 | 294000 | 68.2720 | 83.0188 | 1.0176 |
| 0.7754 | 8.18 | 300000 | 68.0969 | 82.8837 | 1.0176 |
| 0.7771 | 8.35 | 306000 | 68.2136 | 83.0676 | 1.0137 |
| 0.7835 | 8.51 | 312000 | 68.4066 | 83.2256 | 1.0039 |
| 0.7716 | 8.67 | 318000 | 68.4964 | 83.1383 | 1.0332 |
| 0.7874 | 8.84 | 324000 | 68.2899 | 83.0772 | 1.0273 |
| 0.7851 | 9.0 | 330000 | 68.5054 | 83.1645 | 1.0244 |
| 0.7641 | 9.16 | 336000 | 68.4560 | 83.1627 | 1.0195 |
| 0.7609 | 9.33 | 342000 | 68.4111 | 83.1645 | 1.0283 |
| 0.7712 | 9.49 | 348000 | 68.3438 | 83.0170 | 1.0166 |
| 0.7621 | 9.65 | 354000 | 68.1912 | 83.0456 | 1.0078 |
| 0.7596 | 9.82 | 360000 | 68.4695 | 83.2600 | 1.0127 |
| 0.7593 | 9.98 | 366000 | 68.4560 | 83.1607 | 1.0156 |
| 0.7484 | 10.15 | 372000 | 68.4381 | 83.2253 | 1.0303 |
| 0.7495 | 10.31 | 378000 | 68.2765 | 83.0905 | 1.0420 |
| 0.7568 | 10.47 | 384000 | 68.5054 | 83.2846 | 1.0156 |
| 0.7464 | 10.64 | 390000 | 68.5278 | 83.1653 | 1.0469 |
| 0.7521 | 10.8 | 396000 | 68.2944 | 83.0742 | 1.0225 |
| 0.7663 | 10.96 | 402000 | 68.3707 | 83.1869 | 1.0234 |
| 0.7358 | 11.13 | 408000 | 68.4785 | 83.2487 | 1.0371 |
| 0.7415 | 11.29 | 414000 | 68.3528 | 83.1453 | 1.0322 |
| 0.7339 | 11.45 | 420000 | 68.4022 | 83.1251 | 1.0469 |
| 0.7437 | 11.62 | 426000 | 68.3438 | 83.0958 | 1.0479 |
| 0.7375 | 11.78 | 432000 | 68.3932 | 83.2728 | 1.0264 |
| 0.7404 | 11.95 | 438000 | 68.4874 | 83.2539 | 1.0225 |
| 0.7283 | 12.11 | 444000 | 68.3303 | 83.1379 | 1.0293 |
| 0.728 | 12.27 | 450000 | 68.4022 | 83.2441 | 1.0342 |
| 0.7268 | 12.44 | 456000 | 68.5144 | 83.2234 | 1.0361 |
| 0.7259 | 12.6 | 462000 | 68.3573 | 83.2379 | 1.0293 |
| 0.7409 | 12.76 | 468000 | 68.2271 | 83.1034 | 1.0361 |
| 0.7305 | 12.93 | 474000 | 68.3438 | 83.1846 | 1.0605 |
| 0.7109 | 13.09 | 480000 | 68.4156 | 83.2237 | 1.0498 |
| 0.7124 | 13.25 | 486000 | 68.5233 | 83.2077 | 1.0479 |
| 0.7167 | 13.42 | 492000 | 68.4066 | 83.1147 | 1.0498 |
| 0.7248 | 13.58 | 498000 | 68.3662 | 83.0887 | 1.0391 |
| 0.711 | 13.75 | 504000 | 68.5323 | 83.2364 | 1.0479 |
| 0.7101 | 13.91 | 510000 | 68.3752 | 83.2171 | 1.0498 |
| 0.7126 | 14.07 | 516000 | 68.2899 | 83.1185 | 1.0781 |
| 0.7139 | 14.24 | 522000 | 68.4919 | 83.2476 | 1.0557 |
| 0.707 | 14.4 | 528000 | 68.5278 | 83.2351 | 1.0508 |
| 0.6963 | 14.56 | 534000 | 68.4381 | 83.2389 | 1.0615 |
| 0.7129 | 14.73 | 540000 | 68.5009 | 83.2322 | 1.0654 |
| 0.7101 | 14.89 | 546000 | 68.4291 | 83.2037 | 1.0459 |
Framework versions
- Transformers 4.25.1
- Pytorch 1.13.0+cu117
- Datasets 2.7.1
- Tokenizers 0.13.2
