Generative Question Answering on DriveLM (test)
53.09BLEU-4DriveLM-Agent
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| DriveLM-AgentBackbone=DriveLM-Agent, Input Modality=Vision-based2025.12 | 53.09 | 36.19 | 66.79 | 2.79 | |
| MiniDriveBackbone=MiniDrive, Input Modality=Vision-based2025.12 | 50.2 | 37.4 | 73.5 | 3.32 | |
| BeLLA w/ QwenBackbone=Qwen, Input Modality=BEV2025.12 | 45.62 | 33.9 | 72.12 | 3.17 | |
| EM-VLM4AD BaseBackbone=EM-VLM4AD, Scale=Base, Input Modality=Vision-based2025.12 | 45.36 | 34.49 | 71.98 | 3.2 | |
| EM-VLM4ADQ-LargeBackbone=EM-VLM4ADQ, Scale=Large, Input Modality=Vision-based2025.12 | 40.11 | 34.34 | 70.72 | 3.1 |