Cannot repeat the results reported in your paper

[hathawayxxh]

Hi Chunmei,

Thanks for your excellent work and for releasing the source codes. I have tried to reimplement your codes, but unfortunately cannot repeat the results in your paper. I would like to know the parameter setting of the reconstruction task. Is the batch size set as 8 and learning rate set as 1e-4? Since the description in your paper is different from the settings in the file "config/reconstruction_multi_cross.py", I want to make sure what is the correct setting to produce the experimental results in your paper.

Looking forward to your reply. Thanks. Xiaohan