Refusal Tokens: A Simple Way to Calibrate Refusals in Large Language Models
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2024
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866909758789779456 |
|---|---|
| author | Jain, Neel Shrivastava, Aditya Zhu, Chenyang Liu, Daben Samuel, Alfy Panda, Ashwinee Kumar, Anoop Goldblum, Micah Goldstein, Tom |
| author_facet | Jain, Neel Shrivastava, Aditya Zhu, Chenyang Liu, Daben Samuel, Alfy Panda, Ashwinee Kumar, Anoop Goldblum, Micah Goldstein, Tom |
| contents | A key component of building safe and reliable language models is enabling the models to appropriately refuse to follow certain instructions or answer certain questions. We may want models to output refusal messages for various categories of user queries, for example, ill-posed questions, instructions for committing illegal acts, or queries which require information past the model's knowledge horizon. Engineering models that refuse to answer such questions is complicated by the fact that an individual may want their model to exhibit varying levels of sensitivity for refusing queries of various categories, and different users may want different refusal rates. The current default approach involves training multiple models with varying proportions of refusal messages from each category to achieve the desired refusal rates, which is computationally expensive and may require training a new model to accommodate each user's desired preference over refusal rates. To address these challenges, we propose refusal tokens, one such token for each refusal category or a single refusal token, which are prepended to the model's responses during training. We then show how to increase or decrease the probability of generating the refusal token for each category during inference to steer the model's refusal behavior. Refusal tokens enable controlling a single model's refusal rates without the need of any further fine-tuning, but only by selectively intervening during generation. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2412_06748 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Refusal Tokens: A Simple Way to Calibrate Refusals in Large Language Models Jain, Neel Shrivastava, Aditya Zhu, Chenyang Liu, Daben Samuel, Alfy Panda, Ashwinee Kumar, Anoop Goldblum, Micah Goldstein, Tom Machine Learning Computation and Language A key component of building safe and reliable language models is enabling the models to appropriately refuse to follow certain instructions or answer certain questions. We may want models to output refusal messages for various categories of user queries, for example, ill-posed questions, instructions for committing illegal acts, or queries which require information past the model's knowledge horizon. Engineering models that refuse to answer such questions is complicated by the fact that an individual may want their model to exhibit varying levels of sensitivity for refusing queries of various categories, and different users may want different refusal rates. The current default approach involves training multiple models with varying proportions of refusal messages from each category to achieve the desired refusal rates, which is computationally expensive and may require training a new model to accommodate each user's desired preference over refusal rates. To address these challenges, we propose refusal tokens, one such token for each refusal category or a single refusal token, which are prepended to the model's responses during training. We then show how to increase or decrease the probability of generating the refusal token for each category during inference to steer the model's refusal behavior. Refusal tokens enable controlling a single model's refusal rates without the need of any further fine-tuning, but only by selectively intervening during generation. |
| title | Refusal Tokens: A Simple Way to Calibrate Refusals in Large Language Models |
| topic | Machine Learning Computation and Language |
| url | https://arxiv.org/abs/2412.06748 |