- For confidence estimation of LLMs, there's excellent work in the computer vision community.
- Learning from feedback in LLMs? Look into program repair and code generation.

- If you're researching function calling of LLMs, consider exploring literature on semantic parsing.
- Working on task planning with LLMs? Check out techniques from dialogue state tracking.

Finished reviewing for ACL Rolling Review. A common theme in submissions was the lack of connection between similar lines of research. Spending some extra time on targeted literature searches outside the immediate field can go a long way.

Some specific examples I'm familiar with: 👇