AI Paper Atlas

Research map / Video / Video-language understanding

VideoKR: Towards Knowledge- and Reasoning-Intensive Video Understanding

Lin Fu, Zheyuan Yang, Yang Wang, Tingyu Song, Arman Cohan, Yilun Zhao

ICML 2026 · Video › Video-language understanding · approach: Long video understanding · 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 the knowledge-intensive reasoning skill (KnowVidR) to VidR + KnowVid SFT data (ablation)Qwen2.5-VL-7B SFT · Knowledge-intensive averageaccuracy: 41.3 -> 42.4Improved
add KnowVid to VidR only SFT data (ablation)Qwen2.5-VL-7B SFT · Knowledge-intensive averageaccuracy: 41.4 -> 41.3No clear effect
add KnowVid to VidR only SFT data (ablation)Qwen2.5-VL-7B SFT · VideoKR-Evalaccuracy: 35.3 -> 35.9Improved
add KnowVidR to VidR + KnowVid (ablation)Qwen2.5-VL-7B SFT · VideoKR-Evalaccuracy: 35.9 -> 36.8Improved
add chain-of-thought rationales to SFT data instead of direct output (ablation)Qwen2.5-VL-7B SFT · Knowledge-intensive averageaccuracy: 39.4 -> 42.4Improved
add chain-of-thought rationales to SFT data instead of direct output (ablation)Qwen2.5-VL-7B SFT · General video reasoning averageaccuracy: 61.4 -> 58.3Hurt
train on VideoKR-SFT-201K instead of Video-R1-CoT-165kQwen2.5-VL-7B SFT · Knowledge-intensive averageaccuracy: 36.2 -> 42.4Improved
train on VideoKR-SFT-201K instead of VideoRFT-CoT-102KQwen2.5-VL-7B SFT · Knowledge-intensive averageaccuracy: 38.4 -> 42.4Improved
GRPO on VideoKR-RL-114K instead of VideoAuto-R1-83KQwen2.5-VL-7B zero-RL · Knowledge-intensive averageaccuracy: 42.7 -> 43.0Improved
add RL after SFT on VideoKR (ablation)Qwen3-VL-8B · Knowledge-intensive averageaccuracy: 49.2 -> 51.5Improved
zero RL only versus SFT only on VideoKR (ablation)Qwen3-VL-8B · Knowledge-intensive averageaccuracy: 49.2 -> 50.6Improved
post-train Qwen2.5-VL-7B with VideoKR SFT + RLQwen2.5-VL-7B · Knowledge-intensive averageaccuracy: 41.9 -> 46.6 (+4.7)Improved

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