Visual Question Answering on VQA v2 (Accuracy and Iteration Timing)
75.62AccuracyQ-Mamba
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| Q-MambaQuery Length=7292026.06 | 75.62 | 7.54 | 0.147 | |
| Q-MambaQuery Length=5122026.06 | 75.42 | 6.54 | 0.127 | |
| CobraQuery Length=-2026.06 | 75.38 | 7.62 | 0.129 | |
| Q-MambaQuery Length=2562026.06 | 75.01 | 5.94 | 0.099 | |
| Q-MambaQuery Length=1282026.06 | 74.51 | 5.52 | 0.095 | |
| VL-MambaQuery Length=-2026.06 | 74.38 | 7.26 | 0.152 | |
| VL-MambaQuery Length=-, Visual scanning order=forward scan only2026.06 | 72.34 | — | — | |
| VL-MambaQuery Length=-, Visual scanning order=backward scan only2026.06 | 72.06 | — | — |