AI Paper Atlas

Research map / Video / Video-language understanding

Learning Transferable Temporal Primitives for Video Reasoning via Synthetic Videos

Songtao Jiang, Sibo Song, Chenyi Zhou, Yuan Wang, Ruizhe Chen, Tongkun Guan, Ruilin Luo, Yan Zhang, Zhihang Tang, Yuchong Sun, Hang Zhang, Zhibo Yang and 3 more

CVPR 2026 · Video › Video-language understanding · approach: Long video understanding · 2 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
post-train with SynRL (synthetic SFT + GRPO) on Qwen3-VL-4BQwen3-VL-4B · TOMATOaccuracy: 32.1 -> 36.7Improved
post-train with SynRL on Qwen3-VL-8BQwen3-VL-8B · TOMATOaccuracy: 33.2 -> 38.1Improved
post-train with SynRLQwen3-VL-4B · RexTimeR@0.3: 26.2 -> 38.8Improved
post-train with SynRLQwen3-VL-4B · Charades-STAmIoU: 41.9 -> 47.0Improved
prompt Qwen3-VL-4B with CoT only, no SynRL trainingQwen3-VL-4B · Charades-STAmIoU: 41.9 -> 34.5Hurt
add cold-start SFT before RL (ablation)SynRL on Qwen3-VL-4B · TOMATOaccuracy: 32.8 -> 36.9Improved
add RL (GRPO) after SFT (ablation)SynRL on Qwen3-VL-4B · TOMATOaccuracy: 34.2 -> 36.9Improved
use 7.7K synthetic CoT cold-start instead of Video-R1 165K CoT (ablation)Qwen3-VL-4B · TOMATOaccuracy: 28.2 -> 34.2Improved
cold-start with Video-R1 CoT-165K (ablation)Qwen3-VL-4B · vinogroundaccuracy: 40.8 -> 36.6Hurt
add second RL round with 3K longer, harder long-term videos (ablation)SynRL on Qwen3-VL-4B · TOMATOaccuracy: 36.7 -> 37.0Improved
add second RL round (ablation)SynRL on Qwen3-VL-4B · Video-MMEaccuracy: 62.0 -> 62.4Improved
post-train with SynRLQwen3-VL-4B · Minervaaccuracy: 32.5 -> 32.8No 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