About the pre-training model

[QHua-Lin]

Dear author,thank you very much for your open-source contribution.
I have tested the LOL_v1 and LOL_v2 datasets using the pre-trained weight "llie_real.pth", but the results appear to be unusual. I adhered strictly to the instructions provided in the operating documentation. Could there be any steps that I overlooked, or might there be an issue with the model?

LOL_v2_real
paper results: PSNR 22.97 SSIM 0.858
my results: PSNR 33.052 SSIM 0.933

This is really confusing to me, looking forward to your answer！

[ZhengkaiZhao]

I had also met this problem. I calculated the PSNR on LOLv2_real, and the result was 33.11 dB.

[HanZm521]

Dear author,thank you very much for your open-source contribution. I have tested the LOL_v1 and LOL_v2 datasets using the pre-trained weight "llie_real.pth", but the results appear to be unusual. I adhered strictly to the instructions provided in the operating documentation. Could there be any steps that I overlooked, or might there be an issue with the model?

LOL_v2_real paper results: PSNR 22.97 SSIM 0.858 my results: PSNR 33.052 SSIM 0.933

This is really confusing to me, looking forward to your answer！

The authors say the real weights they give are actually LLOL -V1 trained, not LLOL -V2-real.