madaan/self-refine

This project helps developers improve the quality of text and code generated by large language models (LLMs). It allows an LLM to generate an initial output, then critique its own work by providing feedback, and finally use that feedback to refine the output. This iterative self-correction process is useful for anyone working with LLMs to produce better, more accurate results for tasks like generating acronyms, improving code readability, or creating dialogue responses.

785 stars. No commits in the last 6 months.