List your company

Logical Reasoning on ProofWriter CWA 3-hop

73.8AccuracyLSS-CoT
45.51252.85660.267.544Apr 21, 2026

Evaluation results

MethodLinks
LSS-CoTModel=Gemma-2-9B It2026.0473.82.4
3-shot-CoTModel=Gemma-2-9B It2026.0472.8—
SC-3Model=Gemma-2-9B It2026.0472.8—
Greedy-CoTModel=Gemma-2-9B It2026.0471.4—
3-shot-CoTModel=Phi-3-mini-4k Instruct2026.0469.2—
3-shot-CoTModel=Llama-3.1-8B Instruct2026.0466.4—
SC-3Model=Phi-3-mini-4k Instruct2026.0466.4—
LSS-CoTModel=Phi-3-mini-4k Instruct2026.0463.43.8
Greedy-CoTModel=Phi-3-mini-4k Instruct2026.0459.6—
3-shot-CoTModel=LLaMA-2-13B Chat2026.0458—
3-shot-CoTModel=LLaMA-3.2-3B Instruct2026.0456.8—
LSS-CoTModel=LLaMA-2-13B Chat2026.0456.62
LSS-CoTModel=Llama-3.1-8B Instruct2026.0455.64.2
Greedy-CoTModel=LLaMA-2-13B Chat2026.0454.6—
SC-3Model=LLaMA-2-13B Chat2026.0452.6—
Greedy-CoTModel=Llama-3.1-8B Instruct2026.0451.4—
LSS-CoTModel=LLaMA-3.2-3B Instruct2026.0449.83.2
SC-3Model=Llama-3.1-8B Instruct2026.0447.2—
SC-3Model=LLaMA-3.2-3B Instruct2026.0446.8—
Greedy-CoTModel=LLaMA-3.2-3B Instruct2026.0446.6—