Action Explanation Generation on Autonomous Driving Scenarios
73Goal AccuracyAXIS
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| AXISBackbone=GPT-4.1, Mode=AXIS2025.05 | 73 | 2.5 | 3.23 | 73 | |
| AXISBackbone=DSV3, Mode=AXIS2025.05 | 73 | 3.49 | 3.18 | 45 | |
| AXISBackbone=o4-mini, Mode=AXIS2025.05 | 68 | 3.15 | 4.23 | 55 | |
| AXISBackbone=Llama 70B, Mode=AXIS2025.05 | 64 | 2.36 | 3.59 | 64 | |
| GPT-4.1Backbone=GPT-4.1, Mode=Model-only baseline2025.05 | 59 | 2.86 | 3 | 59 | |
| NoExpExplanation=None2025.05 | 56 | — | — | 53 | |
| o4-miniBackbone=o4-mini, Mode=Model-only baseline2025.05 | 55 | 3.19 | 3.59 | 59 | |
| DSV3Backbone=DSV3, Mode=Model-only baseline2025.05 | 50 | 2.69 | 2.64 | 59 | |
| DSR1Backbone=DSR1, Mode=Model-only baseline2025.05 | 50 | 3.33 | 3.73 | 59 | |
| AXISBackbone=DSR1, Mode=AXIS2025.05 | 50 | 3.37 | 4 | 50 | |
| Llama 70BBackbone=Llama 70B, Mode=Model-only baseline2025.05 | 45 | 2.81 | 3.14 | 27 |