Sequence embeddings

[sudeepsahadevan]

Hello,
This project is extremely interesting and could be extremely useful to certain projects of us! What would be the recommended way to extract sequence embeddings from the model ? I have a couple of RNA sequences of interest and the aim is to cluster these sequences in terms of some kind of similarity. I guess, computing the cosine from the sequence embeddings would be the way to go. I found the function match in the module module but I couldn't understand why match_score is multiplied with 1.3 in line 148. Some comments on that would be great!

Thank you,

[tBai1994]

Hello, This project is extremely interesting and could be extremely useful to certain projects of us! What would be the recommended way to extract sequence embeddings from the model ? I have a couple of RNA sequences of interest and the aim is to cluster these sequences in terms of some kind of similarity. I guess, computing the cosine from the sequence embeddings would be the way to go. I found the function match in the module module but I couldn't understand why match_score is multiplied with 1.3 in line 148. Some comments on that would be great!

Thank you,

Did you solve your problem?

[Horikitasaku]

I'm not sure why 1.3, but I guess it's for increasing the weight of s?