Audio-Visual Dense Narration on EgoAVU-Bench
2.66S ScoreEgoAVU-Instruct (Full)
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| EgoAVU-Instruct (Full)Size=7B, Tuning=Full2026.02 | 2.66 | 12.5 | 17.32 | |
| EgoAVU-Instruct (LoRA)Size=7B, Tuning=LoRA2026.02 | 2.6 | 12.2 | 17.19 | |
| Qwen2.5-OmniSize=7B2026.02 | 2.37 | 10.69 | 14.74 | |
| MiniCPM-oSize=8B2026.02 | 2.27 | 10.84 | 14.77 | |
| Qwen2.5-OmniSize=3B2026.02 | 2.17 | 8.55 | 13.45 | |
| Intern-OmniSize=8.7B2026.02 | 1.95 | 5.2 | 10.11 | |
| Baichuan-OmniSize=7B2026.02 | 1.92 | 4.82 | 9.75 | |
| VideoLLaMA2Size=7B2026.02 | 1.88 | 3.65 | 8.32 | |
| Phi4-mmSize=8B2026.02 | 1.59 | 8.79 | 13.13 |