Reasoning Efficiency on Constraint Reasoning (CR)
123.3NtokensSEM-CTRL
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| SEM-CTRLBackbone=Llama 70B2025.03 | 123.3 | 15.9 | |
| o4-mini2025.03 | 1,190.5 | — | |
| DeepSeek-R12025.03 | 3,016.5 | — | |
| o1-preview2025.03 | 3,184.8 | — |
Loading the SOTA2 catalog…