Reasoning on LightEval (test)
0.7452Average Success Rate @8IAC
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| IACModel=Qwen3-14B2026.01 | 0.7452 | 0.7728 | |
| Reinforce++Model=Qwen3-14B2026.01 | 0.7396 | 0.7674 | |
| IACModel=Qwen3-8B2026.01 | 0.7385 | 0.7869 | |
| DrGRPOModel=Qwen3-14B2026.01 | 0.7385 | 0.7657 | |
| GRPOModel=Qwen3-14B2026.01 | 0.7383 | 0.7654 | |
| RLOOModel=Qwen3-14B2026.01 | 0.7364 | 0.7636 | |
| RLOOModel=Qwen3-8B2026.01 | 0.723 | 0.7617 | |
| IACModel=Qwen3-4B2026.01 | 0.7215 | 0.773 | |
| Reinforce++Model=Qwen3-8B2026.01 | 0.7214 | 0.7596 | |
| DrGRPOModel=Qwen3-8B2026.01 | 0.7156 | 0.7536 | |
| GRPOModel=Qwen3-8B2026.01 | 0.7112 | 0.7519 | |
| GRPOModel=Qwen3-4B2026.01 | 0.7061 | 0.7543 | |
| RLOOModel=Qwen3-4B2026.01 | 0.7024 | 0.7539 | |
| Reinforce++Model=Qwen3-4B2026.01 | 0.7021 | 0.7492 | |
| DrGRPOModel=Qwen3-4B2026.01 | 0.6954 | 0.7387 |