Closed Huiyu-Li closed 2 years ago
Hello, Since the range of image is 0-255, why the code use data = data.to(device, dtype=torch.float) not data = data.to(device, dtype=torch.uint8)?
For image compression, unint8 is more efficient than float, right? Thank you.
Hello, Since the range of image is 0-255, why the code use data = data.to(device, dtype=torch.float) not data = data.to(device, dtype=torch.uint8)?
For image compression, unint8 is more efficient than float, right? Thank you.