Open-Ended Audio-Video QA on VGGSound
0.714AccuracyVideoLLaMA2.1-AV
Evaluation Results
| Method | Links | |
|---|---|---|
| VideoLLaMA2.1-AVNumber of Parameters=7B, # Pairs=2.0M, Evaluation Protocol=In-domain2024.06 | 0.714 | |
| VideoLLaMA2-AVNumber of Parameters=7B, # Pairs=1.9M, Evaluation Protocol=In-domain2024.06 | 0.709 | |
| AV-LLMNumber of Parameters=13B, # Pairs=1.6M, Evaluation Protocol=In-domain2024.06 | 0.476 | |
| VideoLLaMANumber of Parameters=7B, # Pairs=2.8M, Evaluation Protocol=Zero-shot2024.06 | 0.408 | |
| Macaw-LLMNumber of Parameters=7B, # Pairs=0.3M, Evaluation Protocol=Zero-shot2024.06 | 0.361 | |
| PandaGPTNumber of Parameters=13B, # Pairs=128M, Evaluation Protocol=Zero-shot2024.06 | 0.327 |