Open-Ended Audio-Video QA on MUSIC-QA
80.9AccuracyVideoLLaMA2.1-AV
Evaluation Results
| Method | Links | |
|---|---|---|
| VideoLLaMA2.1-AVNumber of Parameters=7B, # Pairs=2.0M2024.06 | 80.9 | |
| VideoLLaMA2-AVNumber of Parameters=7B, # Pairs=1.9M2024.06 | 79.2 | |
| CREMANumber of Parameters=4B, Evaluation Protocol=Specialized model fine-tuning2024.06 | 75.6 | |
| CREMANumber of Parameters=4B, Evaluation Protocol=Zero-shot2024.06 | 52.6 | |
| AVicunaNumber of Parameters=7B, # Pairs=1.1M, Evaluation Protocol=Zero-shot2024.06 | 49.6 | |
| OneLLMNumber of Parameters=7B, # Pairs=1007M, Evaluation Protocol=Zero-shot2024.06 | 47.6 | |
| AV-LLMNumber of Parameters=13B, # Pairs=1.6M, Evaluation Protocol=Zero-shot2024.06 | 45.2 | |
| X-InstructBLIPNumber of Parameters=13B, # Pairs=32M, Evaluation Protocol=Zero-shot2024.06 | 44.5 | |
| VideoLLaMANumber of Parameters=7B, # Pairs=2.8M, Evaluation Protocol=Zero-shot2024.06 | 36.6 | |
| PandaGPTNumber of Parameters=13B, # Pairs=128M, Evaluation Protocol=Zero-shot2024.06 | 33.7 | |
| Macaw-LLMNumber of Parameters=7B, # Pairs=0.3M, Evaluation Protocol=Zero-shot2024.06 | 31.8 |