text-multilingual-embedding-002 - deprecation policy

[geco1235]

hi

we are using multi-lingual text embedding for similarity search and clustering.

I see that once in a while,  embedding models are being deprecated.

my questions is: when moving to a newer embedding model, does it mean I need to re-embed my entire text database? (so that new vectors will be in the same space as the old ones?)

Or does new models preserve backward compatibility of the vector space?

If 

[caryna · Accepted solution]

Hi @geco1235,

Welcome to Google Cloud Community!

Yes, when switching to a new model version you must re-embed your entire text database to ensure compatibility and accuracy to your use cases.

Currently, there is no specific discontinuation date for text-multilingual-embedding-002. In fact text-multilingual-embedding-002 is the recommended latest version for multilingual text. You may refer to this documentation for more details.

Was this helpful? If so, please accept this answer as “Solution”. If you need additional assistance, reply here within 2 business days and I’ll be happy to help.

[geco1235]

thanks for your repy.

But eventually it will be deprecated right? so we need to create a migration mechanism in advance or use an open source under our control? or do you see any other option?