Research map / Video / Video-language understanding
CHAMPAGNE: Learning Real-world Conversation from Large-Scale Web Videos
Seungju Han, Jack Hessel, Nouha Dziri, Yejin Choi, Youngjae Yu
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 |
|---|---|---|---|
| train on 18M instead of 2M dialogues (ablation) | Champagne-Large · Image Chat First Turn | PPL: 28.3 -> 21.6 | Improved |
| train on 18M instead of 2M dialogues (ablation) | Champagne-Large · CMU-MOSEI | accuracy (%): 83.2 -> 83.9 | Improved |
| add three video frames instead of a single frame (ablation) | Champagne-Large on YTD-2M · CMU-MOSEI | accuracy (%): 78.9 -> 83.2 | Improved |
| add video frames to training (ablation) | Champagne-Large on YTD-2M · CMU-MOSEI | accuracy (%): 75.6 -> 83.2 | Improved |
| add video frames to training (ablation) | Champagne-Large on YTD-2M · Image Chat First Turn | Dist-3: 0.073 -> 0.190 | Improved |
| convert noisy transcripts into dialogue format with speaker information (ablation) | Champagne-Large on YTD-2M · CMU-MOSEI | accuracy (%): 74.9 -> 83.2 | Improved |
| convert transcripts into dialogue format (ablation) | Champagne-Large on YTD-2M · Image Chat First Turn | Dist-3: 0.143 -> 0.190 | Improved |
| pretrain on YTD-18M before fine-tuning instead of Unified-IO PT initialization | Champagne-XL · CMU-MOSEI | accuracy (%): 81.5 -> 86.1 | Improved |
| pretrain on YTD-18M before fine-tuning instead of Unified-IO PT initialization | Champagne-XL · Visual Dialog fine-tuned | NDCG x100: 65.6 -> 68.2 | Improved |
| pretrain on YTD-18M versus text-only Reddit pretraining (Blender 2.7B) | Champagne-XL · BST perplexity | PPL: 14.6 -> 15.1 | Hurt |
| scale model from Base to XL (ablation) | Champagne · CMU-MOSEI | accuracy (%): 82.3 -> 86.1 | Improved |
Closest papers by meaning
- Chat-UniVi: Unified Visual Representation Empowers Large Language Models with Image and Video Understanding (CVPR 2024 · similarity 0.88)
- V^2Dial: Unification of Video and Visual Dialog via Multimodal Experts (CVPR 2025 · similarity 0.88)
- SAMA: Towards Multi-Turn Referential Grounded Video Chat with Large Language Models (NeurIPS 2025 · similarity 0.88)
- Structure-Aware Multimodal Sequential Learning for Visual Dialog (AAAI 2024 · similarity 0.87)
- HourVideo: 1-Hour Video-Language Understanding (NeurIPS 2024 · similarity 0.87)
- Distilling Vision-Language Models on Millions of Videos (CVPR 2024 · similarity 0.87)
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.
