Long-video Question Answering on Video-MME (val)
75AccuracyGemini-1.5-Pro
Evaluation Results
| Method | Links | |
|---|---|---|
| Gemini-1.5-ProSubtitles=without subtitles2025.01 | 75 | |
| InternVL2.5-78BSubtitles=without subtitles2025.01 | 72.1 | |
| GPT-4oSubtitles=without subtitles2025.01 | 71.9 | |
| Qwen2-VL-72BSubtitles=without subtitles2025.01 | 71.2 | |
| LLaVA-Video-72BSubtitles=without subtitles2025.01 | 70.5 | |
| Tarsier2-7BSubtitles=without subtitles, Number of frames=128f2025.01 | 64.5 | |
| InternVL2.5-8BSubtitles=without subtitles2025.01 | 64.2 | |
| LLaVA-Video-7BSubtitles=without subtitles2025.01 | 63.3 | |
| Qwen2-VL-7BSubtitles=without subtitles2025.01 | 63.3 | |
| VILA-1.5-40BSubtitles=without subtitles2025.01 | 60.1 | |
| Tarsier-34BSubtitles=without subtitles2025.01 | 52.3 | |
| Tarsier-7BSubtitles=without subtitles2025.01 | 42.2 |