Quite a candy for my optimization appetite :) As majority of inference are still on CPUs on mobile/edge accelerators, Mojo will be interesting to watch closely. I wonder how well it will support Triton or CUDA

https://youtu.be/6GvB5lZJqcE

Comments