// the find
hwchase17/chroma-langchain
Two Jupyter notebooks showing how to wire Chroma (vector store) with LangChain for document Q&A — one in-memory, one persistent. This is a demo repo from LangChain's creator, not a library or maintained project.
The notebooks are short and runnable, good for a first contact with retrieval-augmented generation. Using state_of_the_union.txt as demo data makes the examples concrete rather than toy-random. The persistent variant covers the part most tutorials skip: actually keeping the index between runs.
Last touched September 2023 — both LangChain and Chroma have broken their APIs multiple times since then, so there's a real chance the notebooks won't run as-is. Two notebooks is not enough to cover anything beyond the happy path. No error handling, no chunking strategy discussion, no evaluation of retrieval quality. Forks (82) and stars (376) suggest people cloned it to learn then never came back.