You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
First of all, we would like to thank the authors for their excellent work. However, we would like to raise the following questions:
1 How does aligning the denoised student model's features with the teacher's latent space features benefit the optimization of the student model?
2 During inference, the student model's features are not denoised. How does this approach improve the student model's performance?
3 How can this improvement be demonstrated through implementation?
4 How can the effectiveness of this approach be substantiated with data?
The text was updated successfully, but these errors were encountered:
First of all, we would like to thank the authors for their excellent work. However, we would like to raise the following questions:
1 How does aligning the denoised student model's features with the teacher's latent space features benefit the optimization of the student model?
2 During inference, the student model's features are not denoised. How does this approach improve the student model's performance?
3 How can this improvement be demonstrated through implementation?
4 How can the effectiveness of this approach be substantiated with data?
The text was updated successfully, but these errors were encountered: