search

raoyongming / DynamicViT

[NeurIPS 2021] [T-PAMI] DynamicViT: Efficient Vision Transformers with Dynamic Token Sparsification
https://dynamicvit.ivg-research.xyz/
MIT License
551 stars 69 forks source link

Pretrain LV-ViT-S and LV-ViT-M #4

Closed kristenkn closed 2 years ago

kristenkn commented 2 years ago

Hi Yongming,

Thanks for sharing your excellent work.

I tried to downloaded the pretrain model, but it seems the provided link of LV-ViT-S and LV-ViT-M models inside the download_pretrain.sh file are not available, will you be able to share them in here?

Thank you

raoyongming commented 2 years ago

Hi @kristenkn, thanks for your interest in our work. I have uploaded the pre-trained models used in our experiments to Tsinghua Cloud and Google Drive, and updated the README.