Temporal Reasoning on RoboInter-VQA Temporal
81.9Visual TraceRoboInter-Qwen-7B
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| RoboInter-Qwen-7BParameters=7B, Backbone=Qwen2026.02 | 81.9 | 323 | 63.4 | 86.5 | 93 | |
| RoboInter-LlavaOV-7BParameters=7B, Backbone=LLaVA-OneVision2026.02 | 81.9 | 299 | 62.7 | 81.8 | 83.9 | |
| RoboInter-Qwen-3BParameters=3B, Backbone=Qwen2026.02 | 78.8 | 332 | 61.2 | 82.2 | 88.7 | |
| Gemini-2.5-flashSize=flash2026.02 | 49.4 | — | — | — | — | |
| QwenVL2.5-7BParameters=7B2026.02 | 39 | 1,702 | 22.4 | 64.5 | 60.5 | |
| Llava-OV-7BParameters=7B2026.02 | 37.7 | — | 11 | 44.9 | 63.5 | |
| QwenVL2.5-3BParameters=3B2026.02 | 37.5 | 2,712 | 20.3 | 60 | 59.7 | |
| InternVL3-2BParameters=2B2026.02 | 35.2 | — | 7.7 | 59.3 | 59.4 | |
| InternVL3-8BParameters=8B2026.02 | 34 | 1,035 | 8.1 | 71.5 | 60 | |
| Robobrain-2.0-3BParameters=3B2026.02 | 29.7 | 595 | 16 | 48.2 | 46.8 | |
| Robobrain-2.0-7BParameters=7B2026.02 | 29.5 | 541 | 15.3 | 57.8 | 46.4 | |
| InternVL3-1BParameters=1B2026.02 | 28.9 | — | 10.5 | 54.9 | 55.8 | |
| GPT4o-miniSize=mini2026.02 | 28.4 | 1,736 | 14.7 | 66.6 | 63.9 |