bart_baseline_512
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: 1.0396
- Rouge1: 0.6863
- Rouge2: 0.4232
- Rougel: 0.6228
- Rougelsum: 0.6228
- Wer: 0.4678
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: 2e-05
- train_batch_size: 6
- eval_batch_size: 6
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 2
- mixed_precision_training: Native AMP
Training results
Training Loss | Epoch | Step | Validation Loss | Rouge1 | Rouge2 | Rougel | Rougelsum | Wer |
---|---|---|---|---|---|---|---|---|
No log | 0.13 | 250 | 1.2878 | 0.6496 | 0.3667 | 0.5801 | 0.5801 | 0.5226 |
2.1925 | 0.27 | 500 | 1.1932 | 0.6633 | 0.3849 | 0.5939 | 0.5939 | 0.5078 |
2.1925 | 0.4 | 750 | 1.1643 | 0.6635 | 0.3912 | 0.5958 | 0.5958 | 0.4988 |
1.2677 | 0.53 | 1000 | 1.1346 | 0.6741 | 0.3999 | 0.6056 | 0.6057 | 0.4946 |
1.2677 | 0.66 | 1250 | 1.1284 | 0.6685 | 0.3986 | 0.6024 | 0.6025 | 0.4905 |
1.2068 | 0.8 | 1500 | 1.1042 | 0.6783 | 0.4077 | 0.6131 | 0.613 | 0.4864 |
1.2068 | 0.93 | 1750 | 1.0896 | 0.677 | 0.4102 | 0.6127 | 0.6127 | 0.4816 |
1.1659 | 1.06 | 2000 | 1.0812 | 0.6803 | 0.4142 | 0.6167 | 0.6166 | 0.4788 |
1.1659 | 1.2 | 2250 | 1.0670 | 0.6862 | 0.4194 | 0.6214 | 0.6214 | 0.4753 |
1.0634 | 1.33 | 2500 | 1.0665 | 0.68 | 0.4162 | 0.6165 | 0.6165 | 0.4743 |
1.0634 | 1.46 | 2750 | 1.0542 | 0.6866 | 0.4223 | 0.6225 | 0.6226 | 0.4721 |
1.0708 | 1.6 | 3000 | 1.0495 | 0.6872 | 0.4232 | 0.6234 | 0.6233 | 0.4706 |
1.0708 | 1.73 | 3250 | 1.0459 | 0.6838 | 0.4209 | 0.6212 | 0.6211 | 0.47 |
1.0442 | 1.86 | 3500 | 1.0418 | 0.6868 | 0.4237 | 0.6235 | 0.6234 | 0.468 |
1.0442 | 1.99 | 3750 | 1.0396 | 0.6863 | 0.4232 | 0.6228 | 0.6228 | 0.4678 |
Framework versions
- Transformers 4.38.2
- Pytorch 2.2.1+cu121
- Datasets 2.18.0
- Tokenizers 0.15.2
- Downloads last month
- 8