cls_KD
cls_KD copied to clipboard
'NKD and USKD' (ICCV 2023) and 'ViTKD' (CVPRW 2024)
Hi Zhendong, In ViTKD, we only distill the knowledge from unmasked area, while full area in MGD. My questions are: 1) Why ViTKD only distill the knowledge only from unmasked...
训练
KeyError: 'ClsVisualizer is not in the mmengine::visualizer registry. Please check whether the value of `ClsVisualizer` is correct or it was registered as expected. More details can be found at https://mmengine.readthedocs.io/en/latest/advanced_tutorials/config.html#import-the-custom-module'...
Hello, Thank you for your great work! I am very impressed by your research. According to your ViT KD paper, the reported baseline performance of DeiT-tiny and DeiT-small in your...
请问是用uskd_loss()损失来代替任务中的nn.CrossEntropyLoss()交叉熵损失吗? 比如: loss_id = criterion_id(logits , labels) 换为 loss_id = self.uskd_loss(fea_mid, logits , labels) 还是直接加上uskd_loss() 就可以了?