Embodied Question Answering on RoboVQA
77.1BLEU-1Tarsier2
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| Tarsier22025.01 | 77.1 | 67.4 | 61.5 | 56.8 | — | |
| MLCD2025.01 | 73.2 | 66.4 | 60.6 | 56.6 | — | |
| Fast-ThinkAct-7BModel Size=7B2026.01 | 70.4 | 63.3 | 57.3 | 53.2 | 61.1 | |
| ThinkAct-7BModel Size=7B2026.01 | 69.1 | 61.8 | 56 | 52.4 | 59.8 | |
| RoboMamba2025.01 | 54.9 | 44.2 | 39.5 | 36.3 | — | |
| Qwen2.5-VL-7BModel Size=7B2026.01 | 47.8 | 41.2 | 36.2 | 33.7 | 39.7 | |
| RoboBrain2.0-7BModel Size=7B2026.01 | 44.9 | 38.2 | 34.7 | 33.5 | 37.8 | |
| InternVL3-8BModel Size=8B2026.01 | 44.3 | 36.5 | 31.6 | 28.9 | 35.3 | |
| NVILA-8BModel Size=8B2026.01 | 42.7 | 39.7 | 37.6 | 36.1 | 39 | |
| InternVL2.5-8BModel Size=8B2026.01 | 40.5 | 33.3 | 29.6 | 27.5 | 32.7 | |
| Magma-8BModel Size=8B2026.01 | 38.6 | 31.5 | 28.1 | 26.7 | 31.2 | |
| LLaVA-OV-7B2025.01 | 38.1 | 33.6 | 31.8 | 31 | — | |
| LLaMA-AdapterV22025.01 | 27.8 | 16 | 10.9 | 8.1 | — |