Stay updated daily with trending AI research
7 days free trialPick your own topicsAutomated AI summaries

Long Context Transfer from Language to Vision

LMMs
long context transfer
Large Multimodal Models
vision understanding
benchmark
state-of-the-art performance
arXiv:2406.16852 - [arXivPDF]
30
52
267
Long Context Transfer from Language to Vision
Abstract
Video sequences offer valuable temporal information, but existing large multimodal models (LMMs) fall short in understanding extremely long videos. Many works address this by reducing the number of visual tokens using visual resamplers. Alternatively, in this paper, we approach this problem from the perspective of the language model. By simply extrapolating the context length of the language backbone, we enable LMMs to comprehend orders of magnitude more visual tokens without any video training. We call this phenomenon long context transfer and carefully ablate its properties. To effectively measure LMMs' ability to generalize to long contexts in the vision modality, we develop V-NIAH (Visual Needle-In-A-Haystack), a purely synthetic long vision benchmark inspired by the language model's NIAH test. Our proposed Long Video Assistant (LongVA) can process 2000 frames or over 200K visual tokens without additional complexities. With its extended context length, LongVA achieves state-of-the-art performance on Video-MME among 7B-scale models by densely sampling more input frames. Our work is open-sourced at https://github.com/EvolvingLMMs-Lab/LongVA.
30
52
267
Sign up to continue reading AI summary
Stay updated on the latest trending research with our newsletter. Never miss a release date!
Sign Up
© 2025 Adaptive Plus Inc.1216 Broadway, Suite 213,575 Market Str, San Francisco, CA