search

Jingkang50 / OpenOOD

Benchmarking Generalized Out-of-Distribution Detection
MIT License
831 stars 106 forks source link

Submit results to the leaderboard. #219

Closed kai422 closed 7 months ago

kai422 commented 7 months ago

Hi,

Great benchmark! I wonder if I could submit my results to the leaderboard?

Regards, Kai

zjysteven commented 7 months ago

Yes of course! You can update your results in this thread and we will add them to the leaderboard. Meanwhile, if you want you can also contribute your method to OpenOOD by opening a pull request.

kai422 commented 7 months ago

Thanks! I will submit a pull request shortly. By the way, this is a conflict between the environment.yml and setup.py, which is confusing for the installation.

kai422 commented 7 months ago

Hi Jingyang,

I have uploaded the results file to SCALE; you can view the results in this folder. The pretrained model for ISH is available on Google Drive. Additionally, I have created a pull request for our postprocessor, SCALE, and will be uploading our training code shortly.

Regards, Kai

zjysteven commented 7 months ago

Sure. We will update to the leaderboard asap. And for the environment setup, yeah it's a bit confusing. Please just use setup.py for pip install. We will remove environment.yml later.

zjysteven commented 7 months ago

@kai422 Congratulations to your ICLR paper and hitting a new SOTA on ImageNet leaderboard with SCALE and ISH! We have updated the leaderboard to include the results. For the pull request, let us know when the training code is ready just so we can merge once for both SCALE and ISH.