Is RAP algorithm applicable to coding tasks?

[Zeyuan-Liu]

I think coding task is as important as math and reasoning tasks, and extending llm-reasoners on coding tasks will further improve its influence. If I want to apply RAP algorithm to coding tasks (e.g. humaneval, mbpp datasets), how should I modify the algorithm? Thanks for your help!

[Ber666]

Thanks for your interests! Maybe it's not intuitive to define the world model for coding task, but LLM reasoners could still be helpful if you want to build a search framework for coding (e.g., this paper).