Multiple-choice Audio Question Answering on VocalSound (test)
93.92AccuracyQwen2-Audio
Evaluation Results
| Method | Links | |
|---|---|---|
| Qwen2-AudioSize=7B, Training Hours=520k2024.06 | 93.92 | |
| VideoLLaMA2-AVSize=7B, Training Hours=5k2024.06 | 93.19 | |
| Qwen-AudioSize=7B, Training Hours=137k2024.06 | 92.89 | |
| VideoLLaMA2.1-AVSize=7B, Training Hours=5k2024.06 | 92.4 |