Audio-Visual Question Answering on MusicAVQA (Audio-centric target domain)
40.2Average ScoreVISCOP
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| VISCOPArchitecture=Universal Encoder (UE)2025.10 | 40.2 | 11.9 | |
| UE + LLM (LoRA)Architecture=Universal Encoder (UE), Tuning=LoRA2025.10 | 39.8 | 11.5 | |
| Base VLM2025.10 | 28.3 | — |