method · first seen 5 Oct 2026
training-free memory compression for video models
Compressing and retrieving historical visual context in pretrained video models without additional training.
Mentions per week
PapersCodeDiscussion
Earliest sources
- ReMem: Streaming Video Understanding With Long Context RetentionPapers · 5 Oct 2026 · Li Yiheng, He Xu, Wang Shaobo, Shao Ling, Lu Shijian
Latest sources
- ReMem: Streaming Video Understanding With Long Context RetentionPapers · 5 Oct 2026 · Li Yiheng, He Xu, Wang Shaobo, Shao Ling, Lu Shijian