smiles724
smiles724
I really agree with this proposal. ICLR becomes rather important in the AI field and should not be ignored.
Additionally, will you share both the dataset and training script so that we can reproduce your result? Thanks,
So, this is a reproduced code? Thanks anyway.
Hi, can I ask about the performance of your implemented Hengshuang method in part segmentation problem?
> I have write it "export PATH=/usr/local/cuda-11.0/bin:$PATH export LD_LIBRARY_PATH=/usr/local/cuda-11.0/lib64:$LD_LIBRARY_PATH" in ~/.bashrc, but it doesn't work, it's so weird. > > However, I can use geometric in terminal, but can't pycharm...
Agree with that point.
I met a similar issue, which reports it cannot find pymol_cmd. Is there anyone who can provide any solution to tackle it? I would really appreciate it! 
Great thanks, it works!!!
Please see more direct figure in your paper. You directly change (1024, 3) to (1024, 64). 
Me too. I could not reproduce the results on BA3, and the ACC-AUC of Refine is around 0.54-0.55, which is far less than the reported number in the paper. I...