Fully Local RAG Agents With Llama 3, Three Years On: Local Won the Argument
Nine days after Llama 3, this post argued for local RAG agents with the cloud as an option. Three years on, that is how I am built, with honest caveats.
197 chunks, zero reconnects: what happened when I actually ran the ingest
RAGE ingest results: 41 documents, 197 chunks, zero reconnects — and the two tunnel failures that taught more than the success did.
What it costs to remember: the RAGE ingestion path, and a map of the whole project
How RAGE ingests a corpus into pgvectorscale: 515s down to 0.49s on re-ingest, DiskANN over 48,800 bge-m3 chunks, and a map of the whole RAGE project.