AI Paper Atlas

Research map / Video / Video-language understanding

TimeViper: A Hybrid Mamba-Transformer Vision-Language Model for Efficient Long Video Understanding

Boshen Xu, Zihan Xiao, Jiaze Li, Jianzhong Ju, Zhenbo Luo, Jian Luan, Qin Jin

CVPR 2026 · Video › Video-language understanding · approach: Long video understanding · 5 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use TransV token transfer instead of plain uniform token dropping at layer 7 (ablation)TimeViper · Charades-STAmIoU: 26.1 -> 38.1Improved
add TransV in the LLM (50% at layer 7, 90% attention-guided at layer 39) (ablation)TimeViper · VideoMME, Charades-STA, LVBenchaccuracy / mIoU: VideoMME 58.8 -> 56.9; Charades 40.5 -> 37.9; LVBench 35.5 -> 35.6Mixed
add TransV to a Transformer Qwen2.5-7B model (ablation)Qwen2.5-7B · LVBench and VDCaccuracy: LVBench 36.6 -> 37.0; VDC 42.0 -> 40.7Mixed
raise TransV compression at layer 7 from 50% to 90% (ablation)TimeViper · VideoMMEaccuracy: 56.7 -> 53.4Hurt
use attention-guided instead of uniform TransV at layer 39 (ablation)TimeViper · VideoMMEaccuracy: 56.2 -> 56.6Improved
place TransV at layer 7 instead of layer 2 (ablation)TimeViper · VideoMME and VDCaccuracy: VideoMME 56.1 -> 56.7; VDC 39.7 -> 38.9Mixed
add deep-layer uniform TransV (90% at layer 39) on top of shallow TransV (ablation)TimeViper · VideoMMEaccuracy: 56.7 -> 56.2Mixed
add TransV on top of token merging (ToMe) in the projector (ablation)TimeViper · GPU memory at 4,096 framesmemory: 54.8% less memoryImproved
add TransV on top of ToMe (ablation)TimeViper · prefilling time at 4,096 frameslatency: 15.7% less prefilling timeImproved
add token merging (ToMe) in the projector (ablation)TimeViper · frame capacitymax frames: out of memory at 128 frames -> about 5K framesImproved
use hybrid Nanov2-9B backbone instead of Transformer Qwen2.5-7B under identical recipeTimeViper · VideoMMEaccuracy: 56.6 -> 58.8No clear effect
increase test-time input frames from 256 to 512TimeViper · MLVU and LVBenchaccuracy: MLVU 65.64 -> 69.00; LVBench 35.53 -> 37.0Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy