Hi, and thanks for maintaining this guide. It has become a de facto reference for many people learning how to work with LLMs, prompts, RAG, and evaluation.
I am the author of WFGY 16 Problem Map, a focused list of sixteen common failure modes in RAG and LLM infra, each with a short explanation and minimal structural fix. The idea is to give practitioners a shared vocabulary for “what went wrong” when long context, retrieval or infra starts to behave in strange ways.
Repo: https://github.com/onestardao/WFGY
Problem Map: https://github.com/onestardao/WFGY/tree/main/ProblemMap/README.md
The map includes patterns such as retrieval hallucination on top of valid context, vector store ingestion mistakes, bootstrap ordering and infra races, secrets and config drift at first deploy, and other real world issues that appear once systems go beyond toy level.
Suggestion: in the parts of the guide where you discuss RAG, evaluation, or failure analysis, would it be in scope to add a short “Further reading” bullet that links to WFGY 16 Problem Map as an external resource? It might be useful for readers who want to go deeper into concrete failure pattern taxonomy for production RAG systems.
If this looks reasonable, I can draft a small PR that only adds a short reference line in the relevant section.
Thank you again for the high quality guide.
Hi, and thanks for maintaining this guide. It has become a de facto reference for many people learning how to work with LLMs, prompts, RAG, and evaluation.
I am the author of WFGY 16 Problem Map, a focused list of sixteen common failure modes in RAG and LLM infra, each with a short explanation and minimal structural fix. The idea is to give practitioners a shared vocabulary for “what went wrong” when long context, retrieval or infra starts to behave in strange ways.
Repo: https://github.com/onestardao/WFGY
Problem Map: https://github.com/onestardao/WFGY/tree/main/ProblemMap/README.md
The map includes patterns such as retrieval hallucination on top of valid context, vector store ingestion mistakes, bootstrap ordering and infra races, secrets and config drift at first deploy, and other real world issues that appear once systems go beyond toy level.
Suggestion: in the parts of the guide where you discuss RAG, evaluation, or failure analysis, would it be in scope to add a short “Further reading” bullet that links to WFGY 16 Problem Map as an external resource? It might be useful for readers who want to go deeper into concrete failure pattern taxonomy for production RAG systems.
If this looks reasonable, I can draft a small PR that only adds a short reference line in the relevant section.
Thank you again for the high quality guide.