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

17 points

by victormustar

3 hours ago

1 comments