Problem on esp branch

[cskyl]

I am trying to run the experiments for the esp branch (inference contrast), however, it seems that the pretrain weight of the contrast model download in the repo doesn't match the Network setup in the code. Also, it seems that there are multiple network type selection, I was wondering which one should we choose for training the esp network.

Thank you and hope to get your reply!

[huiqing-su]

@cskyl Hi，I have the same question , have you solve this problem？