List your company

Rationale Support Quality and Alignment on OpenBookQA

97.58Rationale Quality (Q)SFT
90.09292.03693.9895.924Jun 12, 2026

Evaluation results

MethodLinks
SFTModel=Ministral 3 8B2026.0697.582.58
CoRAModel=Ministral 3 8B2026.0696.093.66
GRPOModel=Ministral 3 8B2026.0695.914.59
BaseModel=Ministral 3 8B2026.0695.345.28
CoRAModel=Qwen2.5 7B2026.0694.655.65
CoRAModel=Gemma 2 9B2026.0692.67.29
GRPOModel=Gemma 2 9B2026.0692.147.59
BaseModel=Qwen2.5 7B2026.0691.9410.44
SFTModel=Qwen2.5 7B2026.0691.928.48
GRPOModel=Qwen2.5 7B2026.0690.99.46
SFTModel=Gemma 2 9B2026.0690.549.06
BaseModel=Gemma 2 9B2026.0690.389.02