Mathematical Reasoning on MATH500 (Avg@3 Accuracy)
91.2Avg@3 AccuracyQwen3-8B-Base + PASC-GRPO
Evaluation Results
| Method | Links | |
|---|---|---|
| Qwen3-8B-Base + PASC-GRPOmax_completion_tokens=8K2026.01 | 91.2 | |
| Gemma-3-27B-ITmax_completion_tokens=8K2026.01 | 90 | |
| Gemma-3-12B-ITmax_completion_tokens=8K2026.01 | 85.6 | |
| Qwen3-4B-Base + PASC-GRPOmax_completion_tokens=8K2026.01 | 85.32 | |
| Qwen3-8B-Base + GRP-SFTmax_completion_tokens=8K2026.01 | 81.6 | |
| Qwen3-4B-Base + GRP-SFTmax_completion_tokens=8K2026.01 | 72.12 | |
| Qwen3-8B-Basemax_completion_tokens=8K2026.01 | 60.08 | |
| Qwen3-4B-Basemax_completion_tokens=8K2026.01 | 55.43 | |
| LLaMA-3.1-8B-Instructmax_completion_tokens=8K2026.01 | 54.8 |