search

OpenGVLab / InternVideo

[ECCV2024] Video Foundation Models & Data for Multimodal Understanding
Apache License 2.0
1.31k stars 84 forks source link

Downstream classification task checkpoint link not working #179

Open mustafahalimeh opened 5 days ago

mustafahalimeh commented 5 days ago

I am trying to download a downstream classification task model (VideoMAE-L K700) but got this instead:

This XML file does not appear to have any style information associated with it. The document tree is shown below.
<Error>
<Code>NoSuchBucket</Code>
<Message>The specified bucket does not exist.</Message>
<RequestId>66E40608641DD935366F31E7</RequestId>
<HostId>pjlab-gvm-data.oss-cn-shanghai.aliyuncs.com</HostId>
<BucketName>pjlab-gvm-data</BucketName>
<EC>0015-00000101</EC>
<RecommendDoc>https://api.aliyun.com/troubleshoot?q=0015-00000101</RecommendDoc>
</Error

Is there any code snippet or notebook showing how I can create a model from the .pth file and then use it on video clips to get the last features before the K400 classification layer? Thank you

yinanhe commented 5 days ago

For easier access, we have uploaded the model to huggingface. The model can be found at hf. Thanks for the reminder, we have corrected the model link in the table.

mustafahalimeh commented 2 days ago

Thank you so much for the quick reply. Would you kindly provide an example how to load an .pth file and use it on a video to get the predicions probabilities (e.g. [vit_l_hybrid_pt_800e_k700_ft.pth]).