Does a.matmul(b) where a and b are tensors implicitly call .cuda on the tensors if it is available?
no you will have to send the tensor to the device before the matmul
https://pytorch.org/docs/stable/notes/cuda.html
Does a.matmul(b) where a and b are tensors implicitly call .cuda on the tensors if it is available?
no you will have to send the tensor to the device before the matmul
https://pytorch.org/docs/stable/notes/cuda.html