chaoer

@zouchuhang Hi, I run the `panoOptimization.m` and find many variables after each iteration. As you said, this script provide the 3D vertices for the layout. But, I was wondering whether...

> The original split of MillionAID is used for recognition. Our study is about pretraining, so we resplit the training and testing sets. The obtained training set is relatively large...

maybe you are looking for: https://github.com/sail-sg/EditAnything

> I'm not sure if I need to read this URL: https://mmpretrain.readthedocs.io/en/latest/papers/convnext_v2.html Thanks, that's it

same issue. I noticed that in [MedSAM](https://github.com/bowang-lab/MedSAM/blob/7e86549203bda3233afc50b8c8cc41c521d88c9b/segment_anything/modeling/mask_decoder.py#L126), they do the same modification as @bach05. compared to repeat all the embeddings, this seems to be more reliable? any suggestions?

@chaene Thanks for your kindly reply. Actually, I want to use some new samples (create by myself using SFM) to do the training work. So, I need to figure out...

maybe you can see: [def resize_pos_embed(](https://github.com/huggingface/pytorch-image-models/blob/2d0dbd17e388953ab81a5c56f80074eff962ea6b/timm/models/vision_transformer.py#L689)

> @chaoer can it work without finetune the model at other image size? yeap. I successfully finetune SAM (ViT-H) with image size 256 or any size else. But you should...

> > > @chaoer can it work without finetune the model at other image size? > > > > > > yeap. I successfully finetune SAM (ViT-H) with image size...

same problem. have you figured out what happened?