[underdeserver]

So, how does this compare to AlphaFold?

[mentalgear]

seems like they use the normal transformer architecture versus deep fold's more specialised machine-learning approaches.