Chain-of-Thought Reasoning on UGPhysics AtomicPhysics
15.1AccuracyMCNIG
Evaluation results
| Method | Links | |
|---|---|---|
| MCNIGParameter scale=14B2026.03 | 15.1 | |
| MCNIGParameter scale=8B2026.03 | 14.7 | |
| IGParameter scale=8B2026.03 | 13.7 | |
| QwenPRMParameter scale=7B2026.03 | 13.4 | |
| Majority voting2026.03 | 11.5 | |
| Granite PRM v22026.03 | 11 | |
| ImplicitPRM2026.03 | 10.6 | |
| ORMParameter scale=8B2026.03 | 10.4 | |
| MathShepherd2026.03 | 10 | |
| OVM2026.03 | 9.2 | |
| Single sampling2026.03 | 8 |