| Dataset Name | SOTA Method | Metric | Trend | ||
|---|---|---|---|---|---|
| DriveLM VRU-Accident benchmark | GAR64.13 | 10 | 2mo ago | ||
| Dolphins VRU-Accident benchmark | GAR89.33 | 5 | 2mo ago | ||
| nuPlan Generated safety-critical scenarios | GC-PGP | Collision Rate (%)50 | 4 | 5mo ago | |
| nuPlan Original scenarios | PDM-Closed | CR8.3 | 4 | 5mo ago | |
| Composite held-out set (Switzerland) | CR95.4 | 3 | 26d ago | ||
| Composite Small Towns (held-out set) | CR95.6 | 3 | 26d ago | ||
| Composite Yosemite (held-out set) | CR96.5 | 3 | 26d ago | ||
| Composite held-out set (Yellowstone) | CR95.3 | 3 | 26d ago | ||
| Composite (New York City) | Completion Rate (CR)97.3 | 3 | 26d ago | ||
| Composite Los Angeles (held-out set) | CR96.8 | 3 | 26d ago |
