(arxiv.org)

147 points fzliu | 3 comments | 29 Aug 25 20:25 UTC | HN request time: 0.56s | source

we used multi-vector models at Morphik, and I can confirm the real-world effectiveness, especially when compared with dense-vector retrieval.

2. codingjaguar ◴[30 Aug 25 00:00 UTC] No.45070748[source]▶

Curious is that colbert-like ones?

3. bjornsing ◴[30 Aug 25 09:03 UTC] No.45073127[source]▶

Can you share more about the multi-vector approach? Is it open source?

The Theoretical Limitations of Embedding-Based Retrieval