kh12043

Thank you for your prompt response. Upon reviewing the uploaded dataset_dict, I confirmed that the `novel_targets_shuffle` in my experimental results matches the one you provided. (The previous experiments I conducted...

Thank you for your response! I have already conducted several experiments and consistently observed a significant gap in the unseen class accuracy (based on Soft ACC). Since both the all...

Can you share your checkpoint from each stage of continual training on each dataset? It would help with reproducing the results.