Why does torch.FloatTensor change an image's appearance?

Hello I have attached an image that explains my query ,can anyone please explain this behavior?

Thank you!

matplotlib raises a warning, which explains the different output:

Clipping input data to the valid range for imshow with RGB data ([0..1] for floats of [0..255] for integers).

Based on this warning it seems the FloatTensor image contains values outside of [0, 1], which will then be clipped.