Mathematical Reasoning on AIME 2025 (pass@16, avg@16)
73.3Pass@16Base
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| BaseModel Family=Qwen3-8B2026.05 | 73.3 | 47.7 | |
| Self-ReSETModel Family=Qwen3-8B2026.05 | 73.3 | 46.9 | |
| BaseModel Family=DS-Qwen-7B2026.05 | 70 | 37.5 | |
| RECAPModel Family=Qwen3-8B2026.05 | 70 | 46.5 | |
| DAPOModel Family=Qwen3-8B2026.05 | 66.7 | 46.3 | |
| Self-ReSETModel Family=DS-Qwen-7B2026.05 | 63.3 | 37.3 | |
| RECAPModel Family=DS-Qwen-7B2026.05 | 60 | 35 | |
| DAPOModel Family=DS-Qwen-7B2026.05 | 60 | 34.2 | |
| DAPOModel Family=DS-Llama-8B2026.05 | 56.7 | 25 | |
| Self-ReSETModel Family=DS-Llama-8B2026.05 | 56.7 | 25 | |
| RECAPModel Family=DS-Llama-8B2026.05 | 53.3 | 24.6 | |
| BaseModel Family=DS-Llama-8B2026.05 | 50 | 24.6 |