Mathematical Reasoning on AMC23 (Pass@16, Avg@16)
100Pass@16GRPO
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| GRPOModel=Qwen3-14B2025.12 | 100 | 90.16 | |
| SENTModel=Qwen3-14B2025.12 | 100 | 93.75 | |
| SENTbase_model=Qwen2.5-Math-7B2025.12 | 95 | 69.22 | |
| GRPO w/ Enbase_model=Qwen2.5-Math-7B2025.12 | 90 | 66.56 | |
| GRPO w/ Maskbase_model=Qwen2.5-Math-7B2025.12 | 87.5 | 64.69 | |
| GRPObase_model=Qwen2.5-Math-7B2025.12 | 85 | 62.5 |