List your company

Commonsense Reasoning on CommonQA (Accuracy %)

85.18AccuracySC
74.873677.549380.22582.9007Apr 3, 2026

Evaluation results

MethodLinks
SCBase Model=Qwen2.5-32b-instruct2026.0485.18
HCP-MADBase Model=Qwen2.5-32b-instruct + Llama-3.1-70b-instruct2026.0484.85
SRBase Model=Qwen2.5-32b-instruct2026.0484.6
CoTBase Model=Qwen2.5-32b-instruct2026.0484.44
MADBase Model=Qwen2.5-32b-instruct2026.0484.36
DOWNBase Model=Qwen2.5-32b-instruct + Llama-3.1-70b-instruct2026.0484.11
Heter-MADBase Model=Qwen2.5-32b-instruct + Llama-3.1-70b-instruct2026.0483.52
MARSBase Model=Qwen2.5-32b-instruct + Llama-3.1-70b-instruct2026.0479.85
MADBase Model=Llama-3.1-70b-instruct2026.0478.3
SCBase Model=Llama-3.1-70b-instruct2026.0476.49
SRBase Model=Llama-3.1-70b-instruct2026.0476
CoTBase Model=Llama-3.1-70b-instruct2026.0475.27