Missing number of seen classes

[bhargaviparanjape]

Thanks for the great work! I am interested in reproducing some experimental results. When running CoCoOp/CPL for Oxford Pets dataset in the few-shot setting, I get an error in Line 184/186 in CPL/datasets/oxford_pets.py since the number of seen classes is currently not being passed as a 4th argument to the subsample_classes function. I wasn't able to also figure out how many classes were seen/unseen in the results reported in Table 1 in the paper. Where can I find this information?

[jkooy]

Hi, thanks for interests. Seen is passed as base. See here https://github.com/eric-ai-lab/CPL/tree/master/CPL/dataset_processing/classification