|
--- |
|
license: mit |
|
base_model: facebook/bart-large-cnn |
|
tags: |
|
- generated_from_trainer |
|
model-index: |
|
- name: bart_samsum_v2 |
|
results: [] |
|
--- |
|
|
|
<!-- 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. --> |
|
|
|
# bart_samsum_v2 |
|
|
|
This model is a fine-tuned version of [facebook/bart-large-cnn](https://huggingface.co./facebook/bart-large-cnn) on the None dataset. |
|
It achieves the following results on the evaluation set: |
|
- Loss: 0.0236 |
|
|
|
## 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: 4 |
|
- eval_batch_size: 4 |
|
- seed: 42 |
|
- gradient_accumulation_steps: 16 |
|
- total_train_batch_size: 64 |
|
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 |
|
- lr_scheduler_type: linear |
|
- lr_scheduler_warmup_steps: 8 |
|
- num_epochs: 15 |
|
|
|
### Training results |
|
|
|
| Training Loss | Epoch | Step | Validation Loss | |
|
|:-------------:|:-----:|:----:|:---------------:| |
|
| 9.4233 | 0.17 | 1 | 9.1990 | |
|
| 9.5213 | 0.34 | 2 | 8.5394 | |
|
| 8.7467 | 0.52 | 3 | 8.1115 | |
|
| 8.4697 | 0.69 | 4 | 7.5747 | |
|
| 7.752 | 0.86 | 5 | 6.8712 | |
|
| 7.0515 | 1.03 | 6 | 5.8670 | |
|
| 6.0874 | 1.2 | 7 | 4.6814 | |
|
| 5.0408 | 1.38 | 8 | 3.8055 | |
|
| 4.14 | 1.55 | 9 | 2.6678 | |
|
| 2.9893 | 1.72 | 10 | 1.9701 | |
|
| 2.4337 | 1.89 | 11 | 1.5191 | |
|
| 1.9451 | 2.06 | 12 | 1.2105 | |
|
| 1.53 | 2.24 | 13 | 0.9714 | |
|
| 1.2369 | 2.41 | 14 | 0.7905 | |
|
| 1.0014 | 2.58 | 15 | 0.6478 | |
|
| 0.8419 | 2.75 | 16 | 0.5493 | |
|
| 0.7338 | 2.92 | 17 | 0.4770 | |
|
| 0.6393 | 3.1 | 18 | 0.4151 | |
|
| 0.5747 | 3.27 | 19 | 0.3691 | |
|
| 0.4962 | 3.44 | 20 | 0.3293 | |
|
| 0.4516 | 3.61 | 21 | 0.2935 | |
|
| 0.3995 | 3.78 | 22 | 0.2614 | |
|
| 0.3618 | 3.96 | 23 | 0.2346 | |
|
| 0.3246 | 4.13 | 24 | 0.2129 | |
|
| 0.2929 | 4.3 | 25 | 0.1938 | |
|
| 0.278 | 4.47 | 26 | 0.1770 | |
|
| 0.2493 | 4.65 | 27 | 0.1627 | |
|
| 0.2273 | 4.82 | 28 | 0.1500 | |
|
| 0.2067 | 4.99 | 29 | 0.1381 | |
|
| 0.1917 | 5.16 | 30 | 0.1274 | |
|
| 0.1805 | 5.33 | 31 | 0.1174 | |
|
| 0.1557 | 5.51 | 32 | 0.1081 | |
|
| 0.1495 | 5.68 | 33 | 0.1002 | |
|
| 0.1394 | 5.85 | 34 | 0.0933 | |
|
| 0.1261 | 6.02 | 35 | 0.0868 | |
|
| 0.1155 | 6.19 | 36 | 0.0809 | |
|
| 0.1114 | 6.37 | 37 | 0.0755 | |
|
| 0.1041 | 6.54 | 38 | 0.0705 | |
|
| 0.0952 | 6.71 | 39 | 0.0657 | |
|
| 0.0881 | 6.88 | 40 | 0.0615 | |
|
| 0.0823 | 7.05 | 41 | 0.0577 | |
|
| 0.0778 | 7.23 | 42 | 0.0545 | |
|
| 0.071 | 7.4 | 43 | 0.0515 | |
|
| 0.07 | 7.57 | 44 | 0.0487 | |
|
| 0.0625 | 7.74 | 45 | 0.0463 | |
|
| 0.0589 | 7.91 | 46 | 0.0440 | |
|
| 0.0567 | 8.09 | 47 | 0.0422 | |
|
| 0.0537 | 8.26 | 48 | 0.0411 | |
|
| 0.05 | 8.43 | 49 | 0.0398 | |
|
| 0.0472 | 8.6 | 50 | 0.0384 | |
|
| 0.0458 | 8.77 | 51 | 0.0363 | |
|
| 0.0455 | 8.95 | 52 | 0.0347 | |
|
| 0.0412 | 9.12 | 53 | 0.0340 | |
|
| 0.0414 | 9.29 | 54 | 0.0326 | |
|
| 0.0403 | 9.46 | 55 | 0.0333 | |
|
| 0.0384 | 9.63 | 56 | 0.0303 | |
|
| 0.0353 | 9.81 | 57 | 0.0298 | |
|
| 0.0348 | 9.98 | 58 | 0.0293 | |
|
| 0.0342 | 10.15 | 59 | 0.0275 | |
|
| 0.0311 | 10.32 | 60 | 0.0272 | |
|
| 0.0317 | 10.49 | 61 | 0.0270 | |
|
| 0.0315 | 10.67 | 62 | 0.0261 | |
|
| 0.0289 | 10.84 | 63 | 0.0253 | |
|
| 0.0285 | 11.01 | 64 | 0.0247 | |
|
| 0.0273 | 11.18 | 65 | 0.0244 | |
|
| 0.0277 | 11.35 | 66 | 0.0240 | |
|
| 0.0267 | 11.53 | 67 | 0.0237 | |
|
| 0.0263 | 11.7 | 68 | 0.0237 | |
|
| 0.0258 | 11.87 | 69 | 0.0237 | |
|
| 0.0254 | 12.04 | 70 | 0.0238 | |
|
| 0.0248 | 12.22 | 71 | 0.0239 | |
|
| 0.0246 | 12.39 | 72 | 0.0239 | |
|
| 0.0249 | 12.56 | 73 | 0.0237 | |
|
| 0.0239 | 12.73 | 74 | 0.0236 | |
|
| 0.0247 | 12.9 | 75 | 0.0236 | |
|
|
|
|
|
### Framework versions |
|
|
|
- Transformers 4.38.1 |
|
- Pytorch 2.1.0+cu121 |
|
- Datasets 2.17.1 |
|
- Tokenizers 0.15.2 |
|
|
