List your company

Mathematical Reasoning on AIME 24 (AUCOAA)

81.8AUCOAAFormat-Adaptive-Answer
19.50435.67751.8568.023Jan 6, 2026

Evaluation results

MethodLinks
Format-Adaptive-AnswerBackbone=Qwen3-8B2026.0181.8
Normalized-LengthBackbone=Qwen3-8B2026.0177.4
Adaptive-AnswerBackbone=Qwen3-8B2026.0175.8
Hard-Length 8k → 4kBackbone=Qwen3-8B2026.0175.1
TWYNBackbone=Qwen3-8B2026.0174.5
SFTBackbone=Qwen3-8B2026.0173.7
Hard-Length 8kBackbone=Qwen3-8B2026.0173.3
Soft-LengthBackbone=Qwen3-8B2026.0172.4
Hard-Length 16kBackbone=Qwen3-8B2026.0170.7
Base modelBackbone=Qwen3-8B2026.0168.6
No-ThinkingBackbone=Qwen3-8B2026.0121.9