distilbart-cnn-12-6-finetuned-pubmed
This model is a fine-tuned version of sshleifer/distilbart-cnn-12-6 on the pub_med_summarization_dataset dataset. It achieves the following results on the evaluation set:
- Loss: 1.9895
- Rouge1: 40.0985
- Rouge2: 16.5016
- Rougel: 24.8319
- Rougelsum: 36.0775
- Gen Len: 141.884
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: 2
- eval_batch_size: 2
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 5
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Rouge1 | Rouge2 | Rougel | Rougelsum | Gen Len |
|---|---|---|---|---|---|---|---|---|
| 2.1709 | 1.0 | 4000 | 2.0257 | 38.1012 | 15.112 | 23.4064 | 33.9373 | 141.9195 |
| 1.9495 | 2.0 | 8000 | 1.9593 | 39.529 | 16.1693 | 24.487 | 35.5238 | 141.9785 |
| 1.756 | 3.0 | 12000 | 1.9488 | 39.9623 | 16.5799 | 24.949 | 35.9194 | 141.8855 |
| 1.6032 | 4.0 | 16000 | 1.9732 | 39.672 | 16.1994 | 24.5996 | 35.7021 | 141.921 |
| 1.4817 | 5.0 | 20000 | 1.9895 | 40.0985 | 16.5016 | 24.8319 | 36.0775 | 141.884 |
Framework versions
- Transformers 4.17.0
- Pytorch 1.10.0+cu111
- Datasets 1.18.3
- Tokenizers 0.11.6