[crucio]

how well would this work for 762 dimensional vectors? At 3072, they're starting with such a high number of dimensions that the accuracy loss may not be representative of what others would see

[esafak]

You'd have to look at the precision-recall curves for your data set and make the trade-off. There are studies on this topic.

[mediaman]

Generally, it seems that people are starting to see more problems when making vectors of fewer than 1,000 dimensions binary.

[blublubdub]

seems to work pretty well check this out: https://huggingface.co/blog/embedding-quantization