vespa icon indicating copy to clipboard operation
vespa copied to clipboard

Published 20 hours ago • verified publisher icon vespa-engine

ColPali `colbert-embedder`

Open rawsh-rubrik opened this issue 1 year ago • 7 comments
trafficstars

Are there plans on supporting multimodal colbert models for built in inference?

rawsh-rubrik avatar Sep 11 '24 03:09 rawsh-rubrik

Yes, that is on our roadmap for this autumn.

bratseth avatar Sep 11 '24 05:09 bratseth

Hey Jon, any news on this issue?

AmirEstiri avatar Nov 04 '24 17:11 AmirEstiri

We'll get to it this year!

bratseth avatar Nov 05 '24 23:11 bratseth

hey, any updates here?

simjak avatar Jan 01 '25 16:01 simjak

... 1st of January 😅

We're still working on it. Should be Real Soon Now.

bratseth avatar Jan 01 '25 17:01 bratseth

Greetings Jon, wanted to check for the ETA regarding the multi-modal embeddings support. Also, apart from Colpali, will there be support for other multi-modal embedding models such as ColQwen, Qwen-DSE ?

amanalok avatar Jan 16 '25 06:01 amanalok

Yes, we'll support some additional ones, maybe not right away. This will still be some weeks - I can't be more specific sorry.

bratseth avatar Jan 21 '25 09:01 bratseth

Hi @bratseth, thank you for your work. Do you happen to have any updates on this?

vitalijssilins avatar Sep 14 '25 16:09 vitalijssilins

Sorry, not yet!

bratseth avatar Sep 18 '25 12:09 bratseth