AI Paper Atlas

Research map / Video / Video-language understanding

TEMPLE: Incentivizing Temporal Understanding of Video Large Language Models via Progressive Pre-SFT Alignment

Shicheng Li, Lei Li, Kun Ouyang, Shuhuai Ren, Yuanxin Liu, Yuanxing Zhang, Fuzheng Zhang, Lingpeng Kong, Qi Liu, Xu Sun

AAAI 2026 · Video › Video-language understanding · approach: Video LLMs with fine-grained understanding · 3 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
apply Progressive Pre-SFT DPO alignment on self-generated temporal preference dataQwen2-VL-7B · VideoMME temporal perceptionaccuracy: 70.9 -> 74.5Improved
apply Progressive Pre-SFT DPO alignment on self-generated temporal preference dataQwen2-VL-7B · Vinoground (text)accuracy: 34.0 -> 37.2Improved
apply Progressive Pre-SFT DPO alignment on self-generated temporal preference dataQwen2-VL-7B · Average of four benchmarksaccuracy: 57.7 -> 58.9 vs SFTImproved
apply Progressive Pre-SFT DPO alignment on self-generated temporal preference dataQwen2-VL-2B · TempCompass (multi-choice)accuracy: 64.6 -> 64.6No clear effect
apply DPO before SFT rather than after (ablation)Qwen2-VL-7B · VideoMME, MLVU, VinogroundaccuracyImproved
use curriculum of progressively harder DPO data (ablation)Qwen2-VL-7B · Vinoground (text)accuracyImproved
train on harder DPO data only (single difficulty) (ablation)Qwen2-VL-7B · Vinoground (text)accuracyHurt
apply TEMPLE with DPO data generated by Qwen2-VL-7B (ablation)Qwen2.5-VL-7B · Vinoground video scoreaccuracy: 28.8 -> 25.8Hurt
apply TEMPLE with DPO data generated by Qwen2-VL-7B (ablation)InternVL2.5-8B · Vinoground textaccuracy: 42.8 -> 45.0Mixed
use DPO data from Qwen2-VL-7B to tune Qwen2-VL-2B (ablation)Qwen2-VL-2B · VideoMME, MLVU, Vinogroundaccuracy: VideoMME 54.6 -> 52.9, Vinoground text 27.4 -> 26.0Hurt
use DPO data from Qwen2-VL-2B to tune Qwen2-VL-7B (ablation)Qwen2-VL-7B · Vinoground text, MLVUaccuracy: Vinoground 34.6 -> 37.4, MLVU 64.6 -> 64.3Mixed

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