Logical Reasoning on KK hard
73.5AccuracyRL w/ SIE
Evaluation Results
| Method | Links | |
|---|---|---|
| RL w/ SIEModel=Qwen3-8B2025.09 | 73.5 | |
| CoTModel=Qwen3-8B2025.09 | 59.5 | |
| RL w/ SIEModel=Qwen2.5-7B-Instruct2025.09 | 29 | |
| RL w/ SIEModel=Qwen2.5-7B2025.09 | 27.5 | |
| CoTModel=Qwen2.5-7B-Instruct2025.09 | 19.5 | |
| RL w/ SIEModel=Llama3.1-8B-Instruct2025.09 | 15.5 | |
| CoTModel=Qwen2.5-7B2025.09 | 15.5 | |
| CoTModel=Llama3.1-8B-Instruct2025.09 | 6 |