AI Paper Atlas

Research map / Vision-language and multimodal / Cross-modal learning

PRIM:Cooperative Dynamic Token Compression for Efficient Large Multimodal Models

Song Li, yongping xiong

ICML 2026 · Vision-language and multimodal › Cross-modal learning · approach: Audio-visual speech separation

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
apply PRIM compression at 50% token retention instead of full tokensQwen2.5-Omni-7B · Video benchmarks averageaverage score: 61.4 -> 60.4No clear effect
use PRIM instead of FastVID at 50% retentionQwen2.5-Omni-7B · AVUT averageaccuracy: 60.2 -> 62.3Improved
use PRIM instead of FastVID at 35% retentionQwen2.5-Omni-7B · Video benchmarks averageaverage score: 55.7 -> 56.7Improved
use PRIM instead of FastVID on long videos at 50% retentionQwen2.5-Omni-7B · VideoMME longaccuracy: 49.4 -> 48.9Hurt
use PRIM instead of FastVID on LLaVA-Video at 25% retentionLLaVA-Video · VideoMME overallaccuracy: 59.3 -> 61.8Improved
use PRIM instead of FastVID on LLaVA-OV-7B at 50% retentionLLaVA-OV-7B · WorldSense averageaccuracy: 50.4 -> 51.0Improved
use PRIM instead of FastVID at 50% retention on Qwen2.5-Omni-7BQwen2.5-Omni-7B · WorldSense latency per examplelatency (s): 3.64s -> 3.29sImproved
use PRIM instead of FastVID on Qwen2.5-Omni-3B AVUT event localizationQwen2.5-Omni-3B · AVUT ELaccuracy: 42.1 -> 43.7Improved
use PRIM instead of FastVID on counting subtask of AVUT at 35%Qwen2.5-Omni-3B · AVUT CCaccuracy: 39.6 -> 38.7Hurt

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 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