Comment Comment by cschmid Can I also interpret this as: 'AMD's pytorch support is so abysmal that inference is 10x slower than it should be'? parent cschmid 3 years ago
