Size mismatch error when loading pretrained/finetuned model

[fazliimam]

Hi, I get the following error when I try to load the finetune temporal checkpoint vit large model:
RuntimeError: Error(s) in loading state_dict for VisionTransformer: size mismatch for pos_embed: copying a param with shape torch.Size([1, 197, 640]) from checkpoint, the shape in current model is torch.Size([1, 197, 1024]).

I get similar shape mismatch error when I tried loading other saved checkpoints as well

[jayshrivastava0]

Hi, I was facing the similar problem.

RuntimeError: Error(s) in loading state_dict for GroupChannelsVisionTransformer:
	size mismatch for cls_token: copying a param with shape torch.Size([1, 1, 768]) from checkpoint, the shape in current model is torch.Size([1, 1, 1024]).
	size mismatch for patch_embed.0.proj.weight: copying a param with shape torch.Size([768, 4, 8, 8]) from checkpoint, the shape in current model is torch.Size([1024, 4, 8, 8]).

I looked into the documentation, this error was resolved using ViT Large, instead of ViT base, I believe they have changed the size for ViT Base since this repo used it.