Trajectory Prediction on WOD-E2E (val)
7.292RFSMotionLM
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| MotionLMstatus=Baseline2026.05 | 7.292 | 7.083 | 7.227 | 3.195 | |
| IL+VL-HLA (Loss)HLA Strategy=Auxiliary Loss2026.05 | 7.841 | 7.799 | 7.779 | 2.825 | |
| Imitation Learningmode=Finetuned2026.05 | 7.844 | 7.861 | 7.855 | 3.075 | |
| IL+VL-HLA (Input)HLA Strategy=Model Input2026.05 | 8.06 | 8.032 | 8.096 | 2.715 | |
| IL+VL-DPOTraining Strategy=Preference Tuning2026.05 | 8.163 | 8.069 | 8.138 | 2.875 |