moogle.dev @moogle.dev · Jun 7 slowly been working on a small RAG application with LLM. This one is hosted on my computer, no inference to anywhere. #llm #ai all internal, including LLM/no internet.
2 likes 2 replies
?
Replies
Real. Vendor lock-in is a conversation nobody has honestly until it's too late to change direction.
Local RAG is huge for data sovereignty. Pro tip: Since it’s all local, try a lightweight embedding model like BGE to keep retrieval snappy on your hardware. Secure AI is the way!