Commonsense Reasoning on CSQA (Pass Rate)
80.6Pass RateSRPO
Evaluation results
| Method | Links | |
|---|---|---|
| SRPOModel=Qwen2.5-14B-Instruct2026.05 | 80.6 | |
| SCoReModel=Qwen2.5-14B-Instruct2026.05 | 78 | |
| RRPOModel=Qwen2.5-14B-Instruct2026.05 | 77.7 | |
| SRPOModel=OLMo-3-7B-Instruct2026.05 | 74.8 | |
| SPO-TreeModel=OLMo-3-7B-Instruct2026.05 | 73.8 | |
| GRPOModel=OLMo-3-7B-Instruct2026.05 | 72.3 | |
| BaseModel=OLMo-3-7B-Instruct2026.05 | 72.2 | |
| Cr-GRPOModel=OLMo-3-7B-Instruct2026.05 | 72.2 | |
| SCoReModel=OLMo-3-7B-Instruct2026.05 | 71.5 | |
| Cr-GRPOModel=Qwen2.5-14B-Instruct2026.05 | 66.8 | |
| SPO-TreeModel=Qwen2.5-14B-Instruct2026.05 | 66.7 | |
| GRPOModel=Qwen2.5-14B-Instruct2026.05 | 66.1 | |
| BaseModel=Qwen2.5-14B-Instruct2026.05 | 65.8 | |
| RRPOModel=OLMo-3-7B-Instruct2026.05 | 56.9 |