Add WFGY 1.0 (LLM self-healing / RAG debugging framework) to the LLM / Evaluation section? #304
Description
Hi, thanks for maintaining this awesome list – it has been a great map of the generative AI ecosystem.
I’d like to propose adding WFGY 1.0 as a resource in the LLM-related section (it sits at the intersection of LLM evaluation, RAG, and robustness):
- GitHub (MIT): https://github.com/onestardao/WFGY
- LLM ProblemMap (16 failure modes): https://github.com/onestardao/WFGY/blob/main/ProblemMap/README.md
- Tech report PDF with DOI: https://github.com/onestardao/WFGY/blob/main/I_am_not_lizardman/WFGY_All_Principles_Return_to_One_v1.0_PSBigBig_Public.pdf
Very short description:
- provides a 16-problem map of LLM / RAG failure modes (hallucination & chunk drift, retrieval traceability, entropy collapse, multi-agent chaos, deployment deadlock, pre-deploy collapse, etc.)
- combines this with a self-healing framework for large-scale LLM systems
- released under MIT license with a DOI-backed technical report for reproducibility
If you think it fits, an entry like this could work:
- WFGY 1.0 · Self-healing LLM Systems Framework – MIT-licensed framework for diagnosing and repairing LLM / RAG pipelines, built around a 16-problem failure taxonomy and a DOI-backed tech report.
Of course, happy to adjust the wording or open a PR instead if you prefer that workflow. And if you feel it’s out of scope for this list, feel free to close.