Mathematical Reasoning on AMC 23 (Performance %)
47.66Performance (%)CARE
Evaluation Results
| Method | Links | |
|---|---|---|
| CAREBackbone=Phi4-mini-instruct2026.05 | 47.66 | |
| GTBackbone=Phi4-mini-instruct2026.05 | 47.26 | |
| OracleBackbone=Phi4-mini-instruct2026.05 | 42.97 | |
| TTRLBackbone=Phi4-mini-instruct2026.05 | 42.34 | |
| ProbBackbone=Phi4-mini-instruct2026.05 | 42.34 | |
| RandomBackbone=Phi4-mini-instruct2026.05 | 41.64 | |
| EntropyBackbone=Phi4-mini-instruct2026.05 | 40.63 | |
| VanillaBackbone=Phi4-mini-instruct2026.05 | 19.3 |