Hybrid search was raggio’s widest loss in the first benchmark: a median of 104 ms against Weaviate’s 36, and a p99 near 600. The first post blamed …
Generative AI
- Home /
- Categories /
- Generative AI
This is where most of my writing goes now. I build generative AI systems and measure them: raggio, a small RAG vector database, and the AI agent loops that made it faster. Expect numbers, failed experiments and the occasional confession.
The first post ended its ingest section with a line I was rather proud of: “batching those syncs is a known lever I haven’t pulled, because …
Part 2 was about raggio taking two minutes to say hello. This part is about what happened when it tried to build an IVF index inside a 4 GiB container: it died, …
Part 1 was the loop. This part is about the first thing the loop has fixed.
The scorecard at the end of my raggio post kept bugging me. raggio won on memory, recall and vector tail latency, but Weaviate took most of the rest, and not by …
Sooner or later every RAG system sends you the same bill, and it’s written in gigabytes of RAM. The model wants memory for its weights and its KV cache. …