Wisdom Source Research Institute Announces Open Source Ultra-Long Video Understanding Model Video-XL-2

On June 3, Wisdom Source Research Institute announced that it has joined hands with Shanghai Jiao Tong University and other organizations to release a new generation of ultra-long video comprehension model: Video-XL-2. According to the introduction, the new model significantly extends the length of the processable video, supporting efficient processing of up to 10,000 frames of video input on a single graphics card. The model weights for Video-XL-2 are now fully available to the community. (TechCrunch Daily)

Search