Re: Noise There's some probablistic programs that we run that not only don't need determinism, but are actively harmed by it. For example deep learning traini

Re: Noise There's some probablistic programs that we run that not only don't need determinism, but are actively harmed by it. For example deep learning training would probably work fine if there was a 1% destructive noise, as long as there were a massive increase in compute.