Mathematical Reasoning on GSM8K (test) (AVG@8, PASS@8)
96.36AVG@8IAC
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| IACModel=Qwen3-14B2026.01 | 96.36 | 98.11 | |
| RLOOModel=Qwen3-14B2026.01 | 96.04 | 97.43 | |
| GRPOModel=Qwen3-14B2026.01 | 95.96 | 97.53 | |
| DrGRPOModel=Qwen3-14B2026.01 | 95.96 | 97.53 | |
| Reinforce++Model=Qwen3-14B2026.01 | 95.82 | 97.79 | |
| IACModel=Qwen3-8B2026.01 | 95.72 | 97.56 | |
| GRPOModel=Qwen3-8B2026.01 | 95.59 | 96.51 | |
| RLOOModel=Qwen3-8B2026.01 | 95.56 | 97.25 | |
| Reinforce++Model=Qwen3-8B2026.01 | 95.56 | 96.96 | |
| DrGRPOModel=Qwen3-8B2026.01 | 95.35 | 97.11 | |
| IACModel=Qwen3-4B2026.01 | 94.63 | 97.38 | |
| DrGRPOModel=Qwen3-4B2026.01 | 94.1 | 96.43 | |
| Reinforce++Model=Qwen3-4B2026.01 | 94.04 | 96.45 | |
| GRPOModel=Qwen3-4B2026.01 | 93.72 | 96.74 | |
| RLOOModel=Qwen3-4B2026.01 | 93.69 | 96.98 |