AI Paper Atlas

Research map / Large language models / Agents, code and math

FactGuard: Agentic Video Misinformation Detection via Reinforcement Learning

Zehao Li, Hongwei Yu, Hao Jiang, Qiang Sheng, Yilong Xu, Baolong Bi, Yang Li, Zhenlong Yuan, Yujun Cai, Zhaoqi Wang

ICML 2026 · Large language models › Agents, code and math · approach: LLM planning with symbolic tools · 0 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 reinforcement learning (decision-aware GRPO) after SFT (ablation)FactGuard · FakeSVaccuracy: 62.0 -> 79.3Improved
add agentic SFT before RL (ablation)FactGuard · FakeSVaccuracy: 73.1 -> 79.3Improved
add tool-usage reward Rtool (ablation)FactGuard · FakeSVaccuracy: 77.7 -> 79.3Improved
add risk-aware reward Rrisk (ablation)FactGuard · FakeSVaccuracy: 78.5 -> 79.3Improved
add decision-aware RL to regulate tool use (ablation)FactGuard · average FactProbe calls per sampletool calls: 4-5 -> 1-2Improved
add FactProbe and ClipScout tools without training (ablation)Qwen2.5-VL-7B · FakeSVaccuracy: 55.6 -> 57.6No clear effect
add FactProbe to FactGuard (ablation)FactGuard · FakeSVaccuracy: 76.1 -> 79.3Improved
add ClipScout to FactGuard (ablation)FactGuard · FakeSVaccuracy: 78.2 -> 79.3Improved
increase GRPO rollouts from 6 to 8 (ablation)FactGuard · FakeSVaccuracy: 77.3 -> 79.3Improved
increase GRPO rollouts from 8 to 12 (ablation)FactGuard · FakeSVaccuracy: 79.3 -> 79.5No clear effect
change KL coefficient from 0.04 to 0.02 or 0.06 (ablation)FactGuard · FakeSVaccuracy: 79.3 -> 78.2 (0.02), 78.8 (0.06)Hurt
swap teacher model from Qwen2.5-VL-72B to InternVL2.5-78B-MPO (ablation)FactGuard SFT data · FakeSVaccuracy: 79.3 -> 78.7 after SFT+RL (62.0 -> 60.8 SFT only)No clear effect

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