Cannot reproduce the results reported in the paper

[SoundingSilence]

Hello, thanks for the sharing of such an awesome research work.

About this work, I have one question. When I run the baseline experiments by using the default configurations in the released code, the obtained result is much lower than the results reported in the paper. To reproduce the same result of the paper, could the author provide the training detail configurations of all models?

Thanks! Wait for your reply.