Neural Search Talks β€” Zeta Alpha cover image

ColBERT + ColBERTv2: late interaction at a reasonable inference cost

Neural Search Talks β€” Zeta Alpha

00:00

Colbert V3: A Deep Learning Perspective

I think this even goes back to your question about why people prefer single vector. It's like from a deep learning perspective. This is a lot of messing with the architecture that I think people don't want to do right now. Um, maybe they find a way to improve the, the centroid approach fat more, you know, maybe fewer centroids. Some other optimization in there. So if there were new dense retrieval techniques that could be applied, I'm sure they would be. Right now they're at the state of the art as far as what to apply.

Transcript
Play full episode

The AI-powered Podcast Player

Save insights by tapping your headphones, chat with episodes, discover the best highlights - and more!
App store bannerPlay store banner
Get the app