Visual Question Answering on TextVQA (Accuracy, Training time, Inference time)
51.2AccuracyQ-Mamba
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Q-MambaQuery Length=7292026.06 | 51.2 | 7.54 | 0.147 | |
| Q-MambaQuery Length=5122026.06 | 50.2 | 6.54 | 0.127 | |
| Q-MambaQuery Length=2562026.06 | 48.8 | 5.94 | 0.099 | |
| VL-MambaQuery Length=-2026.06 | 48.7 | 7.26 | 0.152 | |
| Q-MambaQuery Length=1282026.06 | 47.1 | 5.52 | 0.095 | |
| VL-MambaQuery Length=-, Visual scanning order=forward scan only2026.06 | 45.6 | — | — | |
| VL-MambaQuery Length=-, Visual scanning order=backward scan only2026.06 | 45.1 | — | — | |
| CobraQuery Length=-2026.06 | 44.9 | 7.62 | 0.129 |