Audio-Visual Question Answering on Music-AVQA 30 (test)
84.3Overall AccuracyCAT-7B-FT
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| CAT-7B-FTEvaluation Protocol=Task-specific fine-tuning2025.03 | 84.3 | 84.9 | 86.1 | 83.2 | |
| PAVE-7B (w/ audio)Evaluation Protocol=Task-specific fine-tuning, TFLOPS=98.63, Total/Trainable Params=8.2B / 170.5M2025.03 | 82.3 | 79.7 | 93 | 78 | |
| VASTEvaluation Protocol=Task-specific fine-tuning2025.03 | 80.7 | — | — | — | |
| VALOREvaluation Protocol=Task-specific fine-tuning2025.03 | 78.9 | — | — | — | |
| LLaVA-OV-7B-FTEvaluation Protocol=Task-specific fine-tuning, TFLOPS=98.53, Total/Trainable Params=8.2B / 161.5M2025.03 | 77.4 | 75.4 | 89.3 | 72.3 | |
| LLaVA-OV-7BEvaluation Protocol=Zero-shot, TFLOPS=98.53, Total/Trainable Params=8.2B / -2025.03 | 60.4 | 68.8 | 70.6 | 52.8 | |
| CAT-7BEvaluation Protocol=Zero-shot2025.03 | 48.6 | — | — | — |