Question about PSNR implementation

[xuefei1]

In utility.py, the calc_psnr(sr, hr, scale, rgb_range, benchmark=False) function, I see that you commented out the if/else check on benchmark, and always execute the block of code that performs the conversion. Just wondering why you did this, because in the original EDSR code benchmark is set to False by default I believe, so this conversion code would never execute?

[xuefei1]

I figured out why, benchmark is set to True when perform validation on Set5, Set14, etc, and during training we never compute the PSNR on the train set, so this would not affect anything.