Fetching the paper…

M2-CLIP: A Multimodal, Multi-task Adapting Framework for Video Action Recognition · Around