3 ms·
How does this work under the hood? Does it build an in-memory vector database of the input sources and runs queries on top of that data to supplement the contex
by m4r71n 1y ago
How does this work under the hood? Does it build an in-memory vector database of the input sources and runs queries on top of that data to supplement the context window?
- brokegrammer 1y agoNo idea how it's implemented because it's proprietary. Details here: https://support.anthropic.com/en/articles/11473015-retrieval-augmented-generation-rag-for-projects https://support.anthropic.com/en/articles/11473015-retrieval...
- menaerus 1y agoRAG commonly implies some sort of vector database to be built and which will then be used for response augmentation. If it operates over the repo, I believe it will index your codebase using those vector embeddings.