[todo] BERT loss on MSA

lucidrains / alphafold2

To eventually become an unofficial Pytorch implementation / replication of Alphafold2, as details of the architecture get released

MIT License

1.54k stars 256 forks source link

Open lucidrains opened 3 years ago

lucidrains commented 3 years ago

needs to be added

lucidrains commented 3 years ago

need to make noise accurate to the paper (contiguous spans?)

lucidrains commented 3 years ago

lucidrains commented 3 years ago

used at inference time.. 🤔

intersun commented 3 years ago

used at inference time.. 🤔

Yep, I also noticed this part, which seems really weird to me...