Research map / Video / Video-language understanding
VAST: Video Ability-Stratified Taxonomy for Data-Efficient Video Reasoning
Zhongan Wang, Xiaoyu Wen, Lingxiao Du, Kun Li, Zhiliang Wu, Xingcheng Xu, Qiaosheng Zhang, Chaochao Lu, Hehe Fan
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 changed | Where | Measured | Effect |
|---|---|---|---|
| add consistency reward (no SFT) (ablation) | VideoVAST · Overall avg (16 frames) | accuracy: 55.5 -> 55.8 | Improved |
| add consistency reward (no SFT) (ablation) | VideoVAST · Overall avg (32 frames) | accuracy: 55.8 -> 56.3 | Improved |
| add consistency reward (no SFT) (ablation) | VideoVAST · Overall avg (64 frames) | accuracy: 56.7 -> 56.7 | No clear effect |
| add consistency reward on SFT-initialized model (ablation) | VideoVAST · Overall avg (32 frames) | accuracy: 55.0 -> 56.2 | Improved |
| add consistency reward (ablation) | VideoVAST · Video-Holmes (32 frames) | accuracy: 37.7 -> 41.6 | Improved |
| add SFT initialization before RL (with CR) (ablation) | VideoVAST · Overall avg (16 frames) | accuracy: 55.8 -> 55.2 | Hurt |
| add SFT initialization before RL (with CR) (ablation) | VideoVAST · Overall avg (32 frames) | accuracy: 56.3 -> 56.2 | Hurt |
| add SFT initialization before RL (no CR) (ablation) | VideoVAST · Overall avg (16 frames) | accuracy: 55.5 -> 54.3 | Hurt |
| scale ability-stratified data from 1.5K to 15K (ablation) | VideoVAST · Overall avg (16 frames) | accuracy: 53.5 -> 55.8 | Improved |
| use ability-stratified VAST-15K instead of random 15K samples (ablation) | VideoVAST · Overall avg (16 frames) | accuracy: 47.6 -> 55.8 | Improved |
| use ability-stratified data instead of random on MMVU (ablation) | VideoVAST · MMVU (mc) | accuracy: 34.4 -> 66.1 | Improved |
| increase inference frames from 16 to 64 (RL with CR) (ablation) | VideoVAST · Overall avg | accuracy: 55.8 -> 56.7 | Improved |
Closest papers by meaning
- Video-R1: Reinforcing Video Reasoning in MLLMs (NeurIPS 2025 · similarity 0.93)
- Incentivizing Versatile Video Reasoning in MLLMs via Data-Efficient Reinforcement Learning (CVPR 2026 · similarity 0.91)
- VideoRFT: Incentivizing Video Reasoning Capability in MLLMs via Reinforced Fine-Tuning (NeurIPS 2025 · similarity 0.91)
- Find, Fix, Reason: Context Repair for Video Reasoning (ICML 2026 · similarity 0.90)
- Scaling RL to Long Videos (NeurIPS 2025 · similarity 0.90)
- OneThinker: All-in-one Reasoning Model for Image and Video (CVPR 2026 · similarity 0.90)
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.
