Mathematical Reasoning on MATH500 (Pass@16, Avg@16)
100Pass@16SENT
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| SENTModel=Qwen3-14B2025.12 | 100 | 93.75 | |
| GRPOModel=Qwen3-14B2025.12 | 95.6 | 90.39 | |
| SENTbase_model=Qwen2.5-Math-7B2025.12 | 92 | 73.83 | |
| GRPObase_model=Qwen2.5-Math-7B2025.12 | 91.8 | 74.76 | |
| GRPO w/ Enbase_model=Qwen2.5-Math-7B2025.12 | 91 | 75.07 | |
| GRPO w/ Maskbase_model=Qwen2.5-Math-7B2025.12 | 91 | 74.49 |