Re:Train a linear layer to increase your RAG system [Technic...
by PLBjt on Friday October 18, 2024 @07:59AM
(#41878573)
Key highlights:
- Uses a simple linear transformation on existing embeddings
- Boosts hit rate from 89% to 95% on real-world examples
- Minor increase on latency (less than 10ms)
- Works on top of blackbox embedding models (Mistral AI, OpenAI, Cohere,...)
- No dataset needed (just your documents)
- Train easily on CPU
[ Parent
| on HN ]