COSP and USP: New methods to advance reasoning in LLMs

[TheIronYuppie]

https://pub.towardsai.net/inside-cosp-and-usp-google-research-new-methods-to-advance-reasoning-in-llms-07338b323dfd

[sgt101]

Self adaptive prompting (prompt generation based on some sort of feedback from the model I think).

USP: https://arxiv.org/abs/2305.14926

COSP: https://aclanthology.org/2023.findings-acl.216/

A hall of mirrors or a wonderful telescope? Time will tell.

[TwentyPosts]

This appears to be for "paid subscribers".

[sgt101 → TwentyPosts]

Google can be your friend brother (still).

https://pub.towardsai.net/inside-cosp-and-usp-google-researc...

[dang → sgt101]

Ok, we'll use that instead of https://thesequence.substack.com/p/about-cosp-and-usp-two-ne..., which looks hardwalled. Thanks!