Mathematical Reasoning on Minerva (Pass@1, Pass@5, Pass@10)
46.1Pass@1LP
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| LPModel=Qwen3-8B2026.03 | 46.1 | 56.2 | 59.1 | |
| S-RLModel=Qwen3-8B2026.03 | 46 | 55.7 | 58.3 | |
| AvgEntModel=Qwen3-8B2026.03 | 44.2 | 54.4 | 57.9 | |
| CH2Model=Qwen3-8B2026.03 | 44 | 53.4 | 56.4 | |
| EntModel=Qwen3-8B2026.03 | 43.9 | 54.4 | 58.2 | |
| AvgEntModel=Qwen3-1.7B2026.03 | 36.1 | 50 | 54.5 | |
| S-RLModel=Qwen3-1.7B2026.03 | 35.4 | 49.2 | 53.9 | |
| CH2Model=Qwen3-1.7B2026.03 | 35.2 | 48.6 | 52.7 | |
| EntModel=Qwen3-1.7B2026.03 | 34.4 | 48.3 | 52.9 | |
| LPModel=Qwen3-1.7B2026.03 | 34.4 | 47.9 | 52.4 | |
| BaseModel=Qwen3-1.7B2026.03 | 30.6 | 42 | 46 | |
| BaseModel=Qwen3-8B2026.03 | 29.8 | 39.9 | 44.8 | |
| S-RLModel=Llama3.1-8B2026.03 | 25.7 | 45.5 | 52.3 | |
| S-RLModel=DeepSeek-Distill Llama-8B2026.03 | 24.4 | 42.7 | 49.9 | |
| AvgEnt†Model=DeepSeek-Distill Llama-8B2026.03 | 23.2 | 40.6 | 48 | |
| BaseModel=DeepSeek-Distill Llama-8B2026.03 | 22.7 | 40.2 | 47.8 | |
| LP†Model=DeepSeek-Distill Llama-8B2026.03 | 22.2 | 40.7 | 48.3 | |
| Ent†Model=DeepSeek-Distill Llama-8B2026.03 | 21.8 | 39.9 | 47.2 | |
| CH2†Model=DeepSeek-Distill Llama-8B2026.03 | 21.8 | 38.6 | 46.1 | |
| BaseModel=Llama3.1-8B2026.03 | 21.4 | 39.8 | 47.1 |