Hyperparameter details for training the neural models for relation classification (for SpanBERT) and both components (Unsupervised Rationale and our approach). The numbers with * are the default values from the SpanBERT implementation available at: https://github.com/facebookresearch/SpanBERT.
| Approach . | SpanBERT . | Unsupervised Rationale . | Our Approach . |
|---|---|---|---|
| Number of epochs | 10* | 20 | 20 |
| Learning rate | 2e-5* | 1e-5 | 1e-5 |
| Dropout rate | 0.1* | 0.1 | 0.1 |
| Batch size | 32* | 32 | 32 |
| Max sequence length | 128* | 128 | 128 |
| Scheduler | Linear scheduler with warm up* | ||
| Approach . | SpanBERT . | Unsupervised Rationale . | Our Approach . |
|---|---|---|---|
| Number of epochs | 10* | 20 | 20 |
| Learning rate | 2e-5* | 1e-5 | 1e-5 |
| Dropout rate | 0.1* | 0.1 | 0.1 |
| Batch size | 32* | 32 | 32 |
| Max sequence length | 128* | 128 | 128 |
| Scheduler | Linear scheduler with warm up* | ||