LensVLM: Compressing long context as images, expanding only relevant pages

28 points | by victormustar 4 hours ago

2 comments