About the image range of metrics

[zhiyuanyou]

Hi～

Thanks for your nice work.

I have run your training code of auto-encoder.

I checked the code relating metrics, and I found that the metrics are calculated with the image range as [0, 1] (See: https://github.com/phizaz/diffae/blob/master/metrics.py#L121-L122).

For these metrics, the image range play a quite important role. For example, if you set the image range as [0, 255], the MSE error would be $255^2 = 65025$ times larger than image range [0, 1].

• My question is whether the metrics in the paper are reported with the image range [0, 1] or [0, 255].

Thanks.