metadata
library_name: transformers
license: apache-2.0
base_model: facebook/bart-large
tags:
- generated_from_trainer
metrics:
- bleu
model-index:
- name: bart-large-wiki-doc-full
results: []
bart-large-wiki-doc-full
This model is a fine-tuned version of facebook/bart-large on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 4.9843
- Sari: 51.0933
- Sari Add: 12.516
- Sari Keep: 45.354
- Sari Del: 95.4099
- Fkgl: 6.642
- Bleu: 19.5688
- D Sari: 0.4555
- D Sari Keep: 0.3857
- D Sari Del: 0.8046
- D Sari Add: 0.1763
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: 2
- eval_batch_size: 2
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 8
- optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.06
- num_epochs: 15
- label_smoothing_factor: 0.3
Training results
| Training Loss | Epoch | Step | Validation Loss | Sari | Sari Add | Sari Keep | Sari Del | Fkgl | Bleu | D Sari | D Sari Keep | D Sari Del | D Sari Add |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 5.6237 | 0.2862 | 500 | 5.1875 | 47.043 | 5.9566 | 40.0258 | 95.1466 | 7.2471 | 12.9924 | 0.4057 | 0.3169 | 0.7965 | 0.1038 |
| 5.0682 | 0.5725 | 1000 | 5.1042 | 48.192 | 9.0538 | 40.5192 | 95.0031 | 7.2139 | 15.7017 | 0.4288 | 0.3477 | 0.8055 | 0.1332 |
| 5.0363 | 0.8587 | 1500 | 5.0920 | 48.4536 | 9.2349 | 41.6306 | 94.4954 | 7.6174 | 20.3138 | 0.4245 | 0.3505 | 0.7871 | 0.1358 |
| 4.9284 | 1.1448 | 2000 | 5.0709 | 49.1161 | 9.9088 | 42.5012 | 94.9385 | 7.3117 | 18.5046 | 0.4339 | 0.3599 | 0.7961 | 0.1455 |
| 4.8644 | 1.4311 | 2500 | 5.0343 | 49.604 | 10.9617 | 42.8337 | 95.0164 | 7.1277 | 18.8397 | 0.4403 | 0.3681 | 0.7991 | 0.1536 |
| 4.8532 | 1.7173 | 3000 | 5.0192 | 49.3028 | 10.3908 | 42.1003 | 95.4172 | 6.9579 | 14.4955 | 0.4389 | 0.3588 | 0.8116 | 0.1462 |
| 4.8343 | 2.0034 | 3500 | 5.0058 | 49.2487 | 11.0341 | 42.1651 | 94.5467 | 7.2777 | 21.4311 | 0.4401 | 0.3693 | 0.7926 | 0.1583 |
| 4.719 | 2.2897 | 4000 | 5.0043 | 49.8632 | 11.2301 | 42.9018 | 95.4578 | 6.5783 | 15.364 | 0.4509 | 0.3735 | 0.8138 | 0.1654 |
| 4.6898 | 2.5759 | 4500 | 4.9904 | 49.9624 | 11.3392 | 43.3503 | 95.1978 | 6.9214 | 18.1433 | 0.449 | 0.3764 | 0.8068 | 0.1637 |
| 4.7061 | 2.8622 | 5000 | 4.9840 | 49.9367 | 11.8606 | 42.6481 | 95.3013 | 6.9934 | 16.9996 | 0.4516 | 0.3748 | 0.8158 | 0.1643 |
| 4.6392 | 3.1483 | 5500 | 4.9898 | 50.5502 | 11.928 | 44.3611 | 95.3614 | 6.9017 | 18.7169 | 0.4523 | 0.3813 | 0.8091 | 0.1665 |
| 4.578 | 3.4345 | 6000 | 4.9893 | 50.39 | 12.0634 | 43.6891 | 95.4175 | 6.8384 | 17.0137 | 0.4567 | 0.3837 | 0.8154 | 0.171 |
| 4.5981 | 3.7208 | 6500 | 4.9843 | 51.0933 | 12.516 | 45.354 | 95.4099 | 6.642 | 19.5688 | 0.4555 | 0.3857 | 0.8046 | 0.1763 |
| 4.5826 | 4.0069 | 7000 | 4.9818 | 50.788 | 12.8273 | 44.4249 | 95.1118 | 6.6555 | 21.2107 | 0.4594 | 0.3942 | 0.8049 | 0.1791 |
| 4.4927 | 4.2931 | 7500 | 5.0009 | 50.4098 | 12.2269 | 43.9055 | 95.0968 | 6.754 | 20.3637 | 0.4602 | 0.3948 | 0.8096 | 0.1763 |
| 4.5009 | 4.5794 | 8000 | 4.9908 | 50.7179 | 12.8769 | 43.783 | 95.4938 | 6.5702 | 17.565 | 0.4601 | 0.3863 | 0.8152 | 0.1787 |
| 4.5128 | 4.8656 | 8500 | 4.9830 | 50.7341 | 12.5935 | 44.2741 | 95.3346 | 6.7056 | 18.9024 | 0.4588 | 0.3905 | 0.8076 | 0.1783 |
| 4.457 | 5.1517 | 9000 | 5.0095 | 50.8947 | 12.7442 | 44.829 | 95.111 | 6.8278 | 21.5163 | 0.462 | 0.3987 | 0.8062 | 0.181 |
Framework versions
- Transformers 4.57.3
- Pytorch 2.9.1+cu128
- Datasets 3.6.0
- Tokenizers 0.22.1