evaluate.py generates images different from results of paper

[T5eng]

Thank you for your excellent work at first. However when I tried to test images on the provides ckpt and codes, I found several problems.

I simply ran python evaluate.py without tuning ratio or any other parameters. But the output seems different from results in your paper. Here are 2 examples.

I also tried to run python evaluate_LOLdataset.py. and it seems only generate the illumination map.

[xenbaloch]

Which tool you're using to run the whole Training and Evaluation process. I can't see any visual outputs upon running evaluate_LOLdataset.py. I'm using PyCharm!