In the paint() function, you are using the torch.autocast() function to cast the batch tensor to the cuda device. However, the batch tensor is already on the cuda device, so there is no need to cast it again. This can be fixed by removing the torch.autocast() call from the paint() function. Am I right?
In the
paint()
function, you are using thetorch.autocast()
function to cast the batch tensor to thecuda
device. However, thebatch
tensor is already on thecuda
device, so there is no need to cast it again. This can be fixed by removing thetorch.autocast()
call from thepaint()
function. Am I right?