search

hustvl / Vim

[ICML 2024] Vision Mamba: Efficient Visual Representation Learning with Bidirectional State Space Model
Apache License 2.0
2.56k stars 160 forks source link

Issue with Loading Models from Huggingface Repo #27

Open cskyl opened 4 months ago

cskyl commented 4 months ago

Hi, Thank you for your fantastic project.

I've encountered a problem when trying to evaluate the model downloaded from the Huggingface repo given in your project. I'm facing an issue with both the small and tiny vim models. When I loaded the models, I receive the following error message: "size mismatch for pos_embed: copying a param with shape torch.Size([1, 730, 384]) from checkpoint, the shape in the current model is torch.Size([1, 197, 384])."

Could you please offer some guidance on how to address this problem? Your assistance would be greatly appreciated.

Thank you so much for this matter!

a912289748 commented 3 months ago

metoo ,what is file

a912289748 commented 3 months ago

yes i got

cskyl commented 3 months ago

yes i got

Did you make it work with any of the models? I tried both the Tiny and Small versions but none of them works for me..

jiabingxiing commented 1 month ago

me too, anyone solved this issue?