How to I get the embeddings for a list of sequences?

[ajrferrari]

Hi, sorry if this is obvious from what is available on the code or at the jupyter notebook, but I could work out through it.

How do I get the embeddings from thousands of protein sequences efficiently? I saw there is a function ".get_rep(self, seq)" that returns the vector of interest for one sequence, but it's very inefficient to iterate over that for thousands of sequences.

Thanks,
Állan

[RaphaelBouvet]

You can use the jax reimplementation of unirep, it is faster than the original implementation
https://github.com/ElArkk/jax-unirep

[jlotthammer]

@RaphaelBouvet are the JAX sequence embeddings equivalent to the original model?