Audio-to-Text temporal grounding on ChronusAV
90.5Recall@IoU=0.5ChronusOmni
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| ChronusOmniModel Size=7B2025.12 | 90.5 | 79.85 | |
| Qwen3-OmniModel Size=30B-A3B2025.12 | 46.7 | 33.1 | |
| ARC-Hunyuan-VideoModel Size=7B2025.12 | 36.85 | 24.25 | |
| Qwen2.5-OmniModel Size=7B2025.12 | 10.1 | 3.7 | |
| AVicunaModel Size=7B2025.12 | 8.2 | 3.45 | |
| OlaModel Size=7B2025.12 | 5.5 | — | |
| LongVALE-LLMModel Size=7B2025.12 | 4.25 | 1.25 | |
| VideoLLaMAModel Size=7B2025.12 | 1.7 | 0.8 |