AI Paper Atlas

Research map / Multi-agent systems and games / Cooperative multi-agent RL

VideoChat-M1: Collaborative Policy Planning for Video Understanding via Multi-Agent Reinforcement Learning

Boyu Chen, Zikang Wang, Zhengrong Yue, Kainan Yan, Chenyun Yu, Yi Huang, Zijun Liu, Yafei Wen, Xiaoxin Chen, Yang Liu, Peng Li, Yali Wang

CVPR 2026 · Multi-agent systems and games › Cooperative multi-agent RL · approach: MARL benchmarks and grouping · 6 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
add collaborative multi-agent policy planning with 4 agents instead of 1 (Qwen3-8B) (ablation)VideoChat-M1 · LongVideoBenchaccuracy: 61.9 -> 82.3Improved
add result reward R_res (ablation)VideoChat-M1 MARL · Video-Holmesaccuracy: 32.4 -> 60.5Improved
add collaboration (process) reward R_col (ablation)VideoChat-M1 MARL · LongVideoBenchaccuracy: 81.1 -> 82.3Improved
add format reward R_format (ablation)VideoChat-M1 MARL · LongVideoBenchaccuracy: 82.0 -> 82.3Improved
add agent dropout (random DAG communication topology) (ablation)VideoChat-M1 MARL · LongVideoBenchaccuracy: 79.9 -> 82.3Improved
add MARL after SFT (ablation)VideoChat-M1 · LongVideoBenchaccuracy: 75.9 -> 82.3Improved
add SFT warm-up before MARL (ablation)VideoChat-M1 · LongVideoBenchaccuracy: 80.2 -> 82.3Improved
use heterogeneous agent architectures instead of 4 identical Qwen3-8B agents (ablation)4-agent CPP group · LongVideoBenchaccuracy: 79.2 -> 82.3Improved
MARL-trained small agent group instead of untrained GPT-4o/DeepSeek-R1 groups under CPP (ablation)CPP agent group · LongVideoBenchaccuracy: 75.9 -> 82.3Improved
full-parameter finetuning instead of LoRA (ablation)VideoChat-M1 · LongVideoBenchaccuracy: 81.2 -> 82.3No clear effect
majority vote instead of a single designated agent deciding (ablation)VideoChat-M1 · LongVideoBenchaccuracy: 81.6 -> 82.3Improved
increase homogeneous agents beyond four (ablation)CPP agent group · Qwen3-8B homogeneous groupsaccuracyNo clear effect

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