List your company

Reverse Chain-of-Thought Generation on IFEval

85.1AccuracySSR-D
73.86876.78479.782.616Feb 16, 2026

Evaluation results

MethodLinks
SSR-DBackbone=Qwen3-32B-Think2026.0285.1
SSRBackbone=Qwen3-32B-Think2026.0284.8
SUPBackbone=Qwen3-32B-Think2026.0284.5
SSR-DBackbone=Qwen3-4B-Think-25072026.0283.8
SSR-DBackbone=Qwen3-8B-Think2026.0283.7
NEUBackbone=Qwen3-32B-Think2026.0283.4
AUG-SUPBackbone=Qwen3-32B-Think2026.0283.2
SSRBackbone=Qwen3-8B-Think2026.0282.9
SSRBackbone=Qwen3-4B-Think-25072026.0281.6
AUG-SUPBackbone=Qwen3-8B-Think2026.0281.5
SUPBackbone=Qwen3-8B-Think2026.0281.2
AUG-SUPBackbone=Qwen3-4B-Think-25072026.0280.1
NEUBackbone=Qwen3-8B-Think2026.0280
SUPBackbone=Qwen3-4B-Think-25072026.0278.5
SSR-DBackbone=NBG4-3B-Base2026.0278.2
SSRBackbone=NBG4-3B-Base2026.0277.6
AUG-SUPBackbone=NBG4-3B-Base2026.0277.2
NEUBackbone=Qwen3-4B-Think-25072026.0276.7
SUPBackbone=NBG4-3B-Base2026.0276.5
NEUBackbone=NBG4-3B-Base2026.0274.3