Ознакомьтесь с документацией PyTorch
Как они упоминали:
print(x.dtype) # Prints "torch.int64", currently 64-bit integer type
x = x.type(torch.FloatTensor)
print(x.dtype) # Prints "torch.float32", now 32-bit float
print(x.float()) # Still "torch.float32"
print(x.type(torch.DoubleTensor)) # Prints "tensor([0., 1., 2., 3.], dtype=torch.float64)"
print(x.type(torch.LongTensor)) # Cast back to int-64, prints "tensor([0, 1, 2, 3])"