| video-question-answering-on-videommmu | ResAdapt (Qwen3-VL-8B, 128 frames) | #8 | Accuracy: 59.6 |
| video-question-answering-on-video-mmmu | Qwen3-VL-8B + ResAdapt (128 frames, 22.9% retention) | #23 | Accuracy: 59.6 |
| video-question-answering-on-mmvu | Qwen3-VL-8B + ResAdapt (128 frames, 22.9% retention) | #28 | Accuracy: 56.3 |
| chart-question-answering-on-chartqa | Qwen2.5-VL-7B + ResAdapt-RL (zero-shot, 105% retention) | #29 | 1:1 Accuracy: 86.6 |
| video-question-answering-on-mlvu | Qwen3-VL-8B + ResAdapt (128 frames, 22.9% retention) | #31 | Accuracy: 70.8 |
| moment-retrieval-on-charades-sta | ResAdapt (Qwen2.5-VL-7B, 128 frames) | #76 | mIoU: 42.0 |