Audio-Visual Question Answering on VALOR (test) using Model-as-Judge
44.67M.J. ScoreM3KG-RAG
Evaluation Results
| Method | Links | |
|---|---|---|
| M3KG-RAGMLLM=Qwen2.5-Omni2025.12 | 44.67 | |
| VAT-KGMLLM=Qwen2.5-Omni2025.12 | 35.44 | |
| VTKGMLLM=Qwen2.5-Omni2025.12 | 32.7 | |
| NoneMLLM=Qwen2.5-Omni2025.12 | 32.42 | |
| M2ConceptBaseMLLM=Qwen2.5-Omni2025.12 | 32.31 | |
| WikidataMLLM=Qwen2.5-Omni2025.12 | 30.28 | |
| M3KG-RAGMLLM=VideoLLaMA22025.12 | 29.25 | |
| VAT-KGMLLM=VideoLLaMA22025.12 | 28.3 | |
| WikidataMLLM=VideoLLaMA22025.12 | 26.43 | |
| M2ConceptBaseMLLM=VideoLLaMA22025.12 | 25.93 | |
| VTKGMLLM=VideoLLaMA22025.12 | 25.92 | |
| NoneMLLM=VideoLLaMA22025.12 | 25.66 |