Hi! Thank you for generously open-sourcing your code. I appreciate your work and the detailed documentation provided. However, I encountered some issues when trying to reproduce the results from your paper. Specifically, when I applied the OpenReMix operation to the mic baseline, the model achieved 51 mIoU on the private class. But when I further incorporated the DECON loss, the model's accuracy decreased to 49 mIoU. I would like to ask for your insights on why this might be happening. Please note that I did not make any changes to the code apart from the ablation operations.
Hi! Thank you for generously open-sourcing your code. I appreciate your work and the detailed documentation provided. However, I encountered some issues when trying to reproduce the results from your paper. Specifically, when I applied the OpenReMix operation to the mic baseline, the model achieved 51 mIoU on the private class. But when I further incorporated the DECON loss, the model's accuracy decreased to 49 mIoU. I would like to ask for your insights on why this might be happening. Please note that I did not make any changes to the code apart from the ablation operations.