Research map / Video / Video-language understanding
TIME: Temporal-Sensitive Multi-Dimensional Instruction Tuning and Robust Benchmarking for Video-LLMs
Yunxiao Wang, Meng Liu, Wenqi Liu, Xuemeng Song, Bin Wen, Fan Yang, Tingting Gao, Di Zhang, Guorui Zhou, Liqiang Nie
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 |
|---|---|---|---|
| fine-tune with TIME dataset and MTP | Video-LLaVA · TIMEBench (AVG) | accuracy: 27.3 -> 31.4 | Improved |
| fine-tune with TIME dataset and MTP | VideoLLaMA 2 · TIMEBench (AVG) | accuracy: 27.4 -> 32.8 | Improved |
| fine-tune with TIME dataset and MTP | ShareGPT4Video · TIMEBench (AVG) | accuracy: 33.7 -> 35.6 | Improved |
| fine-tune with TIME dataset and MTP | InternVL 2.5 · TIMEBench (AVG) | accuracy: 34.4 -> 35.7 | Improved |
| fine-tune with TIME dataset and MTP | InternVL 2.5 · MVBench | accuracy: 72.1 -> 71.9 | No clear effect |
| fine-tune with TIME dataset and MTP | InternVL 2.5 · Video-MME | accuracy: 59.6 -> 59.0 | Hurt |
| fine-tune with TIME dataset and MTP | VideoLLaMA 2 · TempCompass | accuracy: 44.2 -> 58.2 | Improved |
| add TIME dataset alone (ablation) | VideoLLaMA 2 · TIMEBench (AVG) | accuracy: 27.4 -> 32.6 | Improved |
| add MTP alone (ablation) | VideoLLaMA 2 · TIMEBench (AVG) | accuracy: 27.4 -> 29.5 | Improved |
| add MTP on top of TIME (ablation) | VideoLLaMA 2 · TIMEBench (AVG) | accuracy: 32.6 -> 32.8 | Improved |
| mix TIME with original data instead of adding it after (ablation) | VideoLLaMA 2 · TIMEBench (AVG) | accuracy: 26.6 -> 32.6 | Improved |
| mix TIME with original data instead of adding it before (ablation) | VideoLLaMA 2 · TIMEBench (AVG) | accuracy: 28.7 -> 32.6 | Improved |
Closest papers by meaning
- On the Consistency of Video Large Language Models in Temporal Comprehension (CVPR 2025 · similarity 0.92)
- Enhancing Temporal Understanding in Video-LLMs through Stacked Temporal Attention in Vision Encoders (NeurIPS 2025 · similarity 0.92)
- Momentor: Advancing Video Large Language Model with Fine-Grained Temporal Reasoning (ICML 2024 · similarity 0.92)
- SlowFocus: Enhancing Fine-grained Temporal Understanding in Video LLM (NeurIPS 2024 · similarity 0.91)
- TimeChat: A Time-sensitive Multimodal Large Language Model for Long Video Understanding (CVPR 2024 · similarity 0.91)
- Video-STaR: Self-Training Enables Video Instruction Tuning with Any Supervision (ICLR 2025 · 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.
