Logical Reasoning on Reasoning Gym Zebra Puzzles
39.33Accuracy (*)TokUR (TU)
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| TokUR (TU)Base Model=Qwen-2.5-3B-Instruct2025.05 | 39.33 | 71.38 | 41.42 | |
| TokUR (AU)Base Model=Qwen-2.5-3B-Instruct2025.05 | 39.33 | 71.66 | 41.71 | |
| TokUR (EU)Base Model=Qwen-2.5-3B-Instruct2025.05 | 37.33 | 71.28 | 41.2 | |
| CoT (Lower-Bound)Base Model=Qwen-2.5-3B-Instruct2025.05 | 33.67 | — | — | |
| Self-CertaintyBase Model=Qwen-2.5-3B-Instruct2025.05 | 30 | 47.77 | 22.95 | |
| DeepConfBase Model=Qwen-2.5-3B-Instruct2025.05 | 26 | 42.41 | 21.07 | |
| PEBase Model=Qwen-2.5-3B-Instruct2025.05 | 24 | 44.7 | 21.76 | |
| LLBase Model=Qwen-2.5-3B-Instruct2025.05 | 22 | 42.15 | 21.02 |