<!-- 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_3e-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: 2.7126
- Rouge1: 36.1419
- Rouge2: 13.0561
- Rougel: 23.9016
- Bleu1: 30.1069
- Bleu2: 17.658
- Bleu3: 10.4667
- Bleu4: 5.9043
- Gen Len: 51.1865
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: 3e-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 |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 2.6085 | 0.19 | 1000 | 3.0852 | 30.6379 | 9.5297 | 20.038 | 24.8006 | 13.3454 | 6.7991 | 3.2426 | 49.7669 |
| 2.437 | 0.38 | 2000 | 2.8738 | 33.0203 | 10.8205 | 21.5603 | 27.1297 | 14.9942 | 8.0371 | 4.2429 | 44.9184 |
| 2.3443 | 0.57 | 3000 | 2.7916 | 33.5593 | 11.3516 | 22.3152 | 27.5124 | 15.4457 | 8.6766 | 4.7093 | 47.8438 |
| 2.2552 | 0.76 | 4000 | 2.7385 | 33.9742 | 11.6764 | 22.1995 | 27.8055 | 15.7452 | 8.8451 | 4.8093 | 48.8998 |
| 2.1995 | 0.94 | 5000 | 2.6926 | 33.576 | 11.559 | 22.3242 | 27.3653 | 15.4487 | 8.7329 | 4.7069 | 44.9744 |
| 2.0435 | 1.13 | 6000 | 2.7022 | 33.7386 | 11.4953 | 22.4573 | 27.5194 | 15.5649 | 8.7561 | 4.7532 | 44.303 |
| 2.0249 | 1.32 | 7000 | 2.6774 | 35.7727 | 12.5817 | 23.0383 | 30.2068 | 17.3961 | 10.1085 | 5.5276 | 52.8228 |
| 2.0279 | 1.51 | 8000 | 2.6586 | 35.8848 | 13.1753 | 23.2714 | 29.9946 | 17.7741 | 10.5437 | 6.2265 | 54.1772 |
| 2.0246 | 1.7 | 9000 | 2.6345 | 34.6736 | 12.5099 | 22.9161 | 28.2366 | 16.4884 | 9.8084 | 5.5189 | 46.4499 |
| 2.0043 | 1.89 | 10000 | 2.6188 | 34.7284 | 12.4015 | 23.2798 | 28.4401 | 16.4485 | 9.591 | 5.4751 | 45.9184 |
| 1.7908 | 2.08 | 11000 | 2.6649 | 35.7231 | 12.9914 | 23.2967 | 29.9147 | 17.6362 | 10.3099 | 5.9712 | 51.9347 |
| 1.7773 | 2.27 | 12000 | 2.6449 | 35.1311 | 12.5855 | 23.0385 | 28.9588 | 16.8575 | 9.8356 | 5.4904 | 49.3007 |
| 1.7755 | 2.45 | 13000 | 2.6428 | 35.3833 | 12.7367 | 23.5528 | 29.6375 | 17.3555 | 10.2739 | 5.7493 | 50.8275 |
| 1.8226 | 2.64 | 14000 | 2.6308 | 35.5618 | 12.7614 | 23.4132 | 29.5545 | 17.1865 | 10.1054 | 5.5529 | 47.5897 |
| 1.8038 | 2.83 | 15000 | 2.6201 | 35.9127 | 13.1838 | 23.8028 | 29.7936 | 17.5444 | 10.3447 | 5.9587 | 49.8112 |
| 1.6303 | 3.02 | 16000 | 2.6723 | 35.6743 | 12.7891 | 23.4622 | 29.7027 | 17.3276 | 10.1388 | 5.6587 | 53.9487 |
| 1.6206 | 3.21 | 17000 | 2.6681 | 35.8034 | 12.7135 | 23.5982 | 29.7476 | 17.2447 | 9.9051 | 5.5885 | 47.8182 |
| 1.6431 | 3.4 | 18000 | 2.6802 | 35.9423 | 13.2482 | 23.813 | 29.9246 | 17.7619 | 10.5368 | 6.1582 | 49.4172 |
| 1.6123 | 3.59 | 19000 | 2.6747 | 36.0087 | 12.8361 | 23.6901 | 30.0722 | 17.4825 | 10.4133 | 5.8633 | 49.2587 |
| 1.5975 | 3.78 | 20000 | 2.6738 | 35.657 | 12.7534 | 23.4204 | 29.7909 | 17.2791 | 9.872 | 5.8953 | 48.5967 |
| 1.6147 | 3.97 | 21000 | 2.6757 | 36.2985 | 13.2254 | 23.5733 | 30.4337 | 17.9413 | 10.4105 | 6.1814 | 53.1119 |
| 1.4836 | 4.15 | 22000 | 2.7234 | 35.6085 | 12.7511 | 23.4656 | 29.607 | 17.2656 | 10.0315 | 5.671 | 51.3007 |
| 1.5084 | 4.34 | 23000 | 2.7132 | 36.4079 | 13.1961 | 23.9245 | 30.5969 | 17.9801 | 10.5547 | 6.0477 | 49.7529 |
| 1.485 | 4.53 | 24000 | 2.7115 | 36.4402 | 13.6026 | 24.0881 | 30.6095 | 18.2638 | 11.0541 | 6.5495 | 50.8508 |
| 1.5019 | 4.72 | 25000 | 2.7154 | 35.5796 | 12.7942 | 23.7473 | 29.5215 | 17.2522 | 10.2134 | 5.8349 | 49.331 |
| 1.4728 | 4.91 | 26000 | 2.7126 | 36.1419 | 13.0561 | 23.9016 | 30.1069 | 17.658 | 10.4667 | 5.9043 | 51.1865 |
Framework versions
- Transformers 4.25.1
- Pytorch 1.13.0+cu117
- Datasets 2.7.1
- Tokenizers 0.13.2
