siahuat0727

@WEIFZH 咳。。我后来用了这个 https://github.com/Qv2ray/Qv2ray

Oh! You are right, I made a terrible mistake here! The result without abs() can't represent its orthogonality among others. I will fix it and re-run the experiments then. Thanks...

Sorry for the late reply. My results are the same as yours. Maybe there are more bugs in my implementation. I also hope you can let me know when you...

No problem. Keep in touch!

@BigFishMaster Hi, I will re-run it this weekend and update the results on [my blog](https://siahuat0727.github.io/2019/03/17/repr/).

@BigFishMaster Thanks! I've just updated my re-run results.

@philokey That's my mistake, thanks a lot! For the channels initialization part, I have tried to initialize it randomly or with all zeros, and the former looks better. I also...

@tiandunx Hi, thanks for checking. 1. I think I didn't get it. I also wonder what's the difference between my experiment settings and the paper's. 2. If I understand correctly,...

@niuzchinanick I just noticed the email. Did you solve it? Which line is wrong? Can you copy the entire error message? Maybe PyTorch version not compatible. I used PyTorch 1.0.1...

Hi, I mask the weights that connected to the pruned filters with zeros. [code](https://github.com/siahuat0727/RePr/blob/b0f46f496b0f98d082a386e1bdf71ec0a5d6f40b/main.py#L205) Note that this operation doesn't mentioned in the paper.