MSLS pretrained model release

[Idanlau]

It would be great if you could release the MSLS Melbourne pre-trained model for reference. For some reason, I cannot replicate some of the experimental results for MSLS. I would love to double-check on my end to see where I would need to adjust my setup, thanks!

[oravus]

Hi @Idanlau, unfortunately, I don't have a handy access to that specific model. I could retrain it but it will hae to wait for a few weeks. Instead, it might be better to directly help you replicate. Can you please elaborate?

Also, have you been able to replicate results corresponding to the other released models (here)?

[Idanlau]

Sorry for the late response it was my mistake that caused the issue, thanks for the great work on seqnet!