3D-Uniformer Ablation Weights

[omubarek]

Hello, In Uniformer: Unified Transformer for Efficient Spatiotemporal Representation Learning paper Table 4a you have done an ablation study on the local and global affinity at different stages.

Can you please share the pretrained weights on 2D-Uniformer for the following configurations (such that I can load them for the 3D-Uniformer):

• G-G-G-G
• L-L-L-L
• L-G-G-G

Thank you,

[Andy1621]

Sorry, I forgot to reply to your email. These pre-trained weights for ablation studies are lost, since I have started a new internship in another company.

I think it's simple to retrain these models, you can simply adopt the settings for Image Classification. To maintain similar FLOPs, you should adjust the depth of each stage. Besides, for GGGG, I adopt the SA in MViT, wherein the K and V are downsampled by depth-wise convolution.

[omubarek]

Thank you.