search

OpenGVLab / unmasked_teacher

[ICCV2023 Oral] Unmasked Teacher: Towards Training-Efficient Video Foundation Models
https://arxiv.org/abs/2303.16058
MIT License
267 stars 13 forks source link

UMT-VIT module #36

Closed JackeyZhang1001 closed 3 months ago

JackeyZhang1001 commented 4 months ago

I want to load vit module only, how can I load it from pertained?

Andy1621 commented 4 months ago

You can simply download the model weights from here and use ViT module.

JackeyZhang1001 commented 4 months ago

I download the model weights, but I don't know how to init the model that I can load the state_dict, could you please give me a demo?

Andy1621 commented 4 months ago

You can simply use the model code here and load the model. For a demo, you can follow those use in VideoMAE, since we share the same model code.