bart-large-asqa-cb
This model is a fine-tuned version of facebook/bart-large on the None dataset. It achieves the following results on the evaluation set:
- Loss: 2.4791
- Rougelsum: 38.2862
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-06
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 20
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Rougelsum |
|---|---|---|---|---|
| 3.347 | 1.0 | 545 | 2.5353 | 37.3812 |
| 2.7829 | 2.0 | 1090 | 2.5087 | 37.6431 |
| 2.6973 | 3.0 | 1635 | 2.4906 | 37.9194 |
| 2.6125 | 4.0 | 2180 | 2.4812 | 38.1180 |
| 2.5697 | 5.0 | 2725 | 2.4762 | 38.1616 |
| 2.5086 | 6.0 | 3270 | 2.4773 | 38.1370 |
| 2.4678 | 7.0 | 3815 | 2.4831 | 37.9346 |
| 2.4404 | 8.0 | 4360 | 2.4896 | 38.1150 |
| 2.3866 | 9.0 | 4905 | 2.4775 | 38.2222 |
| 2.3791 | 10.0 | 5450 | 2.4791 | 38.2862 |
Framework versions
- Transformers 4.23.0.dev0
- Pytorch 1.12.1+cu102
- Datasets 2.4.0
- Tokenizers 0.12.1