-
Thanks for your wonderful work and detailed explaining ! I already successfully run `analyze/attnmap.py`, but the attention map is different from that obtained by deit. I wonder to know how I can gain…
-
Thank you for releasing your codes and pre-trained model!
As mentioned in the paper, the model is pre-trained on a subset of ImageNet-1k which is generated by removing similar or identical categor…
-
- [ ] Confirm structure and web pages by Tuesday, May 18
- [ ] Write content for the pages by Tuesday, June 1
-
Thank you for publishing your code. I saw the DeiT ablation in your paper. Is there a chance you could also provide code to reproduce that? If you'd prefer to contact me in private, my email is yusun@…
-
Can you provide the details of the model is fine-tuned for 1000 epochs with DeiT-style knowledge distillation? Thanks!
-
## Reference
- 2020-12 **[DeiT]** Training data-efficient image transformers & distillation through attention [[Paper](https://arxiv.org/abs/2012.12877)] [[Code](https://github.com/facebookresearch/d…
-
Thank you for your great work on this project and for sharing the code with it. I used the code given to Train and tried to recreate the work. But there was a problem. As you can see the best numbers …
-
Hello, again bother, I would like to confirm that. line 10 in fold.sh file: **train_weight_path = "backbones/PASCAL/FPTrans/one_shot_DeiT/fold0.pth**"; As far as I know, fold0.pth should be a file, w…
-
Unknown model (deit_small_MCTformerV2_patch16_224)
-
Hi all! I just wanna confirm one confusion.
1. Is it true that we could not get the training accuracy because we have a teacher to monitor that?
2. Is it possible to fine-tune the DeiT with distil…