List your company

Reverse Chain-of-Thought Generation on ArenaHard

72ScoreSSR-D
30.81641.50852.262.892Feb 16, 2026

Evaluation results

MethodLinks
SSR-DBackbone=Qwen3-32B-Think2026.0272
SSRBackbone=Qwen3-32B-Think2026.0270.1
AUG-SUPBackbone=Qwen3-32B-Think2026.0269.1
SUPBackbone=Qwen3-32B-Think2026.0268.8
NEUBackbone=Qwen3-32B-Think2026.0268.2
SSR-DBackbone=Qwen3-8B-Think2026.0259.5
SSRBackbone=Qwen3-8B-Think2026.0256.3
AUG-SUPBackbone=Qwen3-8B-Think2026.0252.8
SUPBackbone=Qwen3-8B-Think2026.0251.5
NEUBackbone=Qwen3-8B-Think2026.0250.8
SSR-DBackbone=Qwen3-4B-Think-25072026.0250.8
SSRBackbone=Qwen3-4B-Think-25072026.0246.1
AUG-SUPBackbone=Qwen3-4B-Think-25072026.0243.5
SUPBackbone=Qwen3-4B-Think-25072026.0242.3
NEUBackbone=Qwen3-4B-Think-25072026.0241.7
SSR-DBackbone=NBG4-3B-Base2026.0237.2
SSRBackbone=NBG4-3B-Base2026.0234.6
AUG-SUPBackbone=NBG4-3B-Base2026.0233.2
SUPBackbone=NBG4-3B-Base2026.0232.9
NEUBackbone=NBG4-3B-Base2026.0232.4