<!-- 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_5e-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: 2.8332
- Rouge1: 36.0185
- Rouge2: 12.6783
- Rougel: 23.3148
- Bleu1: 30.2418
- Bleu2: 17.381
- Bleu3: 10.3059
- Bleu4: 5.9599
- Gen Len: 50.9767
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: 5e-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 |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 2.5229 | 0.19 | 1000 | 2.9931 | 31.4246 | 10.0302 | 20.7531 | 25.3618 | 13.7797 | 7.3585 | 3.689 | 46.8019 |
| 2.3763 | 0.38 | 2000 | 2.8644 | 33.6125 | 11.6317 | 21.9202 | 27.7709 | 15.9381 | 8.996 | 4.8041 | 50.1562 |
| 2.3371 | 0.57 | 3000 | 2.7958 | 34.253 | 11.8488 | 22.4988 | 28.501 | 16.2829 | 9.1703 | 4.9873 | 48.2751 |
| 2.3018 | 0.76 | 4000 | 2.7559 | 34.3508 | 11.7971 | 22.5994 | 28.1757 | 15.9896 | 9.12 | 5.0712 | 42.9767 |
| 2.214 | 0.94 | 5000 | 2.7131 | 34.5451 | 12.4437 | 22.9456 | 28.3871 | 16.5087 | 9.9256 | 5.5757 | 46.0653 |
| 2.0007 | 1.13 | 6000 | 2.7207 | 35.0462 | 12.0128 | 22.3508 | 29.3657 | 16.7098 | 9.4792 | 5.0235 | 49.5152 |
| 1.9633 | 1.32 | 7000 | 2.7195 | 34.3249 | 11.9224 | 22.9618 | 28.2812 | 16.0876 | 9.3298 | 5.3695 | 46.7879 |
| 2.0002 | 1.51 | 8000 | 2.6799 | 35.783 | 12.7607 | 23.8872 | 29.6408 | 17.2382 | 10.1776 | 5.9003 | 46.5967 |
| 1.9783 | 1.7 | 9000 | 2.6615 | 34.7877 | 12.2492 | 23.0451 | 28.8199 | 16.6404 | 9.6347 | 5.2901 | 47.2681 |
| 1.955 | 1.89 | 10000 | 2.6337 | 35.3022 | 12.7166 | 23.4134 | 29.218 | 17.0785 | 9.925 | 5.6807 | 50.0559 |
| 1.671 | 2.08 | 11000 | 2.6997 | 35.3595 | 12.305 | 23.3744 | 29.525 | 16.937 | 9.6249 | 5.2743 | 48.4219 |
| 1.6756 | 2.27 | 12000 | 2.6986 | 34.8911 | 12.2688 | 23.1722 | 29.1454 | 16.7564 | 9.7788 | 5.5929 | 46.8648 |
| 1.663 | 2.45 | 13000 | 2.6974 | 35.4625 | 12.5317 | 23.3959 | 29.3184 | 17.0218 | 9.7629 | 5.4506 | 48.662 |
| 1.6896 | 2.64 | 14000 | 2.6792 | 34.6078 | 12.3596 | 23.1353 | 28.6652 | 16.697 | 9.9738 | 5.6329 | 45.1608 |
| 1.7114 | 2.83 | 15000 | 2.6765 | 35.3731 | 12.669 | 23.4203 | 29.6602 | 17.1914 | 10.0183 | 5.745 | 47.9557 |
| 1.4059 | 3.02 | 16000 | 2.7574 | 35.249 | 12.3037 | 23.0811 | 29.4765 | 16.9417 | 9.563 | 5.4593 | 50.3939 |
| 1.4559 | 3.21 | 17000 | 2.7695 | 35.3686 | 12.2559 | 23.1602 | 29.3155 | 16.7156 | 9.6546 | 5.4363 | 47.7226 |
| 1.4475 | 3.4 | 18000 | 2.7638 | 35.3241 | 12.5225 | 23.3305 | 29.5401 | 17.0816 | 9.7474 | 5.4129 | 48.6993 |
| 1.4459 | 3.59 | 19000 | 2.7679 | 35.64 | 12.6542 | 23.1888 | 30.0146 | 17.4051 | 10.2219 | 5.7042 | 51.8438 |
| 1.4678 | 3.78 | 20000 | 2.7604 | 35.1451 | 12.2282 | 23.1746 | 29.4539 | 16.8357 | 9.7948 | 5.321 | 49.1935 |
| 1.4478 | 3.97 | 21000 | 2.7555 | 36.2922 | 13.2416 | 24.0108 | 30.5121 | 17.9087 | 10.6678 | 6.2204 | 49.9417 |
| 1.2405 | 4.15 | 22000 | 2.8381 | 36.0049 | 12.868 | 23.5304 | 30.1701 | 17.6082 | 10.4209 | 5.7566 | 53.3916 |
| 1.2203 | 4.34 | 23000 | 2.8370 | 35.6913 | 12.5497 | 23.6024 | 29.8742 | 17.1319 | 9.9978 | 5.6913 | 49.7646 |
| 1.2756 | 4.53 | 24000 | 2.8360 | 35.3826 | 12.3329 | 22.8257 | 29.5363 | 16.8789 | 9.7444 | 5.4338 | 51.972 |
| 1.2452 | 4.72 | 25000 | 2.8362 | 35.7976 | 12.5759 | 23.2084 | 30.1391 | 17.3059 | 10.1375 | 5.6696 | 50.1888 |
| 1.241 | 4.91 | 26000 | 2.8332 | 36.0185 | 12.6783 | 23.3148 | 30.2418 | 17.381 | 10.3059 | 5.9599 | 50.9767 |
Framework versions
- Transformers 4.25.1
- Pytorch 1.13.0+cu117
- Datasets 2.7.1
- Tokenizers 0.13.2
