<!-- 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. -->
kobart_8_4e-5_datav2_min30_lp5.0_temperature1.0
This model is a fine-tuned version of gogamza/kobart-base-v2 on the None dataset. It achieves the following results on the evaluation set:
- Loss: 2.7690
- Rouge1: 35.7198
- Rouge2: 12.6777
- Rougel: 23.5157
- Bleu1: 29.7798
- Bleu2: 17.2442
- Bleu3: 10.1198
- Bleu4: 5.5845
- Gen Len: 50.2914
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: 4e-05
- train_batch_size: 8
- eval_batch_size: 128
- 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: 5.0
Training results
| Training Loss | Epoch | Step | Validation Loss | Rouge1 | Rouge2 | Rougel | Bleu1 | Bleu2 | Bleu3 | Bleu4 | Gen Len |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 2.5571 | 0.19 | 1000 | 3.0256 | 30.6752 | 9.655 | 20.3793 | 24.9545 | 13.4562 | 7.0852 | 3.6167 | 47.2378 |
| 2.3748 | 0.38 | 2000 | 2.8633 | 33.6862 | 11.3467 | 21.6442 | 27.6602 | 15.5034 | 8.564 | 4.7708 | 52.5921 |
| 2.3327 | 0.57 | 3000 | 2.7965 | 34.1286 | 11.5936 | 22.3078 | 28.2895 | 15.9539 | 9.0344 | 5.0261 | 46.4336 |
| 2.2987 | 0.76 | 4000 | 2.7423 | 33.7844 | 11.4184 | 22.2715 | 27.9016 | 15.7678 | 8.887 | 4.9817 | 44.1305 |
| 2.2137 | 0.94 | 5000 | 2.6925 | 34.4899 | 12.4798 | 23.0933 | 28.5676 | 16.7234 | 9.854 | 5.4929 | 46.5431 |
| 2.0205 | 1.13 | 6000 | 2.6899 | 35.1651 | 12.2364 | 22.6918 | 29.561 | 16.9967 | 9.5871 | 5.4011 | 51.4126 |
| 1.9818 | 1.32 | 7000 | 2.7037 | 34.1708 | 12.01 | 22.3273 | 28.597 | 16.3676 | 9.6473 | 5.2881 | 48.0979 |
| 2.0085 | 1.51 | 8000 | 2.6568 | 35.1423 | 12.6615 | 23.3564 | 29.0896 | 16.9543 | 10.0793 | 5.8229 | 47.014 |
| 1.9972 | 1.7 | 9000 | 2.6399 | 35.3604 | 12.6992 | 23.3829 | 29.2344 | 17.0287 | 9.9469 | 5.5226 | 46.4336 |
| 1.963 | 1.89 | 10000 | 2.6225 | 34.992 | 12.3573 | 23.0134 | 29.0142 | 16.8063 | 9.6906 | 5.5045 | 51.4452 |
| 1.718 | 2.08 | 11000 | 2.6629 | 34.8932 | 12.2868 | 23.2794 | 28.7742 | 16.5584 | 9.6199 | 5.4499 | 47.5804 |
| 1.7171 | 2.27 | 12000 | 2.6648 | 35.4343 | 12.7376 | 23.4355 | 29.4051 | 17.1878 | 10.2903 | 5.824 | 46.4359 |
| 1.695 | 2.45 | 13000 | 2.6578 | 35.0225 | 12.1733 | 22.9686 | 28.8901 | 16.5961 | 9.3781 | 5.2049 | 49.0443 |
| 1.7282 | 2.64 | 14000 | 2.6435 | 33.9569 | 11.9783 | 22.9137 | 27.9425 | 16.0888 | 9.3867 | 5.3915 | 46.0886 |
| 1.7541 | 2.83 | 15000 | 2.6469 | 34.6347 | 12.1309 | 22.7496 | 28.9934 | 16.6886 | 9.7165 | 5.2098 | 49.62 |
| 1.4855 | 3.02 | 16000 | 2.7137 | 35.3936 | 12.7873 | 23.3762 | 29.4388 | 17.1262 | 10.0549 | 5.9223 | 50.0256 |
| 1.5382 | 3.21 | 17000 | 2.7161 | 35.211 | 12.7758 | 23.8604 | 29.1727 | 17.007 | 10.1639 | 6.0141 | 46.8159 |
| 1.5243 | 3.4 | 18000 | 2.7222 | 35.6339 | 12.683 | 23.5104 | 29.8071 | 17.3418 | 10.178 | 5.5185 | 49.5944 |
| 1.5265 | 3.59 | 19000 | 2.7210 | 35.4469 | 12.5754 | 23.3784 | 29.5035 | 17.1414 | 9.8427 | 5.5385 | 50.7762 |
| 1.5394 | 3.78 | 20000 | 2.7193 | 35.9595 | 12.9418 | 23.5227 | 30.0655 | 17.5487 | 10.115 | 5.6725 | 50.3357 |
| 1.5364 | 3.97 | 21000 | 2.7000 | 35.6398 | 12.9591 | 23.8267 | 29.9125 | 17.587 | 10.4197 | 5.985 | 48.4476 |
| 1.343 | 4.15 | 22000 | 2.7756 | 35.8172 | 12.7519 | 23.5584 | 29.7877 | 17.2715 | 10.219 | 5.9187 | 49.2984 |
| 1.3182 | 4.34 | 23000 | 2.7813 | 35.2382 | 12.7271 | 23.3914 | 29.5501 | 17.3306 | 10.3873 | 6.1428 | 50.8228 |
| 1.3771 | 4.53 | 24000 | 2.7716 | 35.4267 | 12.6279 | 23.3564 | 29.6336 | 17.245 | 10.2511 | 5.9128 | 51.8695 |
| 1.3522 | 4.72 | 25000 | 2.7700 | 35.8057 | 12.9656 | 23.6143 | 29.8501 | 17.475 | 10.2721 | 5.7671 | 50.6946 |
| 1.3508 | 4.91 | 26000 | 2.7690 | 35.7198 | 12.6777 | 23.5157 | 29.7798 | 17.2442 | 10.1198 | 5.5845 | 50.2914 |
Framework versions
- Transformers 4.25.1
- Pytorch 1.13.0+cu117
- Datasets 2.7.1
- Tokenizers 0.13.2
