News for hackers. Stuff that matters.
Re:AlphaFold2 at CASP14: “It feels like one’s child has left... by rsfern on Saturday December 12, 2020 @03:02PM (#25401131)
While interesting, I think this model is only viable for a certain subset of computational workloads.

AF2 took like two weeks of computation on 200 TPUs to train. I’m not sure if they used a data parallel or model parallel distributed training approach, but I don’t imagine either would scale particularly well to a globally distributed network of worker machines in a folding@home analogue...

There are no comments on this comment yet.