CantTalkAboutThis: Aligning Language Models to Stay on Topic in Dialogues
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_ | 1866911928365875200 |
|---|---|
| author | Sreedhar, Makesh Narsimhan Rebedea, Traian Ghosh, Shaona Zeng, Jiaqi Parisien, Christopher |
| author_facet | Sreedhar, Makesh Narsimhan Rebedea, Traian Ghosh, Shaona Zeng, Jiaqi Parisien, Christopher |
| contents | Recent advancements in instruction-tuning datasets have predominantly focused on specific tasks like mathematical or logical reasoning. There has been a notable gap in data designed for aligning language models to maintain topic relevance in conversations - a critical aspect for deploying chatbots to production. We introduce the CantTalkAboutThis dataset to help language models remain focused on the subject at hand during task-oriented interactions. It consists of synthetic dialogues on a wide range of conversation topics from different domains. These dialogues are interspersed with distractor turns that intentionally divert the chatbot from the predefined topic. Fine-tuning language models on this dataset helps make them resilient to deviating from the role assigned and improves their ability to maintain topical coherence compared to general-purpose instruction-tuned LLMs like GPT-4-turbo and Mixtral-Instruct. Additionally, preliminary observations suggest that training models on this dataset also enhance their performance on fine-grained instruction following tasks, including safety alignment. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2404_03820 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | CantTalkAboutThis: Aligning Language Models to Stay on Topic in Dialogues Sreedhar, Makesh Narsimhan Rebedea, Traian Ghosh, Shaona Zeng, Jiaqi Parisien, Christopher Computation and Language Recent advancements in instruction-tuning datasets have predominantly focused on specific tasks like mathematical or logical reasoning. There has been a notable gap in data designed for aligning language models to maintain topic relevance in conversations - a critical aspect for deploying chatbots to production. We introduce the CantTalkAboutThis dataset to help language models remain focused on the subject at hand during task-oriented interactions. It consists of synthetic dialogues on a wide range of conversation topics from different domains. These dialogues are interspersed with distractor turns that intentionally divert the chatbot from the predefined topic. Fine-tuning language models on this dataset helps make them resilient to deviating from the role assigned and improves their ability to maintain topical coherence compared to general-purpose instruction-tuned LLMs like GPT-4-turbo and Mixtral-Instruct. Additionally, preliminary observations suggest that training models on this dataset also enhance their performance on fine-grained instruction following tasks, including safety alignment. |
| title | CantTalkAboutThis: Aligning Language Models to Stay on Topic in Dialogues |
| topic | Computation and Language |
| url | https://arxiv.org/abs/2404.03820 |