Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

V-LynX: Token Interface Alignment for Video+X LLMs

About

This study introduces an intriguing phenomenon in Video LLMs: rather than merely translating frames into textual embeddings, Video LLMs establish a continuous manifold, token interface, allowing visual tokens to operate as standalone entities within the architecture. Exploiting this discovery, we propose V-LynX, a scalable framework that integrates novel modalities into Video LLMs by repurposing the internalized interface. Departing from conventional paradigms that necessitate heavy modality-specific encoders or paired supervision, V-LynX employs a lightweight auxiliary pathway in parallel with the frozen vision encoder. Our method integrates new sensory inputs with intrinsic video priors by aligning both attention responses and statistical distributions using unpaired unimodal data sets. This ensures manifold compatibility while preserving the integrity of the Video LLMs. Extensive benchmarks demonstrate that V-LynX achieves SOTA and efficiency across audio-visual QA, 3D reasoning, high-frame-rate, and multi-view video understanding. The code is available at https://github.com/park-jungin/lynx.

Jungin Park, Jiyoung Lee, Kwanghoon Sohn• 2026

Related benchmarks

TaskDatasetResultRank
Video Question AnsweringMLVU
Accuracy68.4
194
Audio-Visual Question AnsweringAVQA
Accuracy94.2
85
3D Question AnsweringSQA3D
Exact Match (EM)60.5
34
Video Question AnsweringMVBench
Average Score61.2
20
Proficiency estimationEgo-Exo4D--
16
3D Question AnsweringScanQA (evaluation)
CIDEr107.4
13
Audio-visual understandingAVUT AV-Human
Accuracy0.4691
12
Audio-Visual QAMUSIC-AVQA
Overall Accuracy83
11
Audio-Visual QAAVSD
CIDEr163
10
Video Question AnsweringVideoMME
Short VQA Accuracy73
6
Showing 10 of 10 rows

Other info

Follow for update