List your company

Mathematical Reasoning on GSM-Hard (Accuracy and Cost)

63.28AccuracyMACA
46.723251.021655.3259.6184May 25, 2026

Evaluation results

MethodLinks
MACABackbone=Qwen2.5-14B-Instruct2026.0563.282,966.4
MaASBackbone=Qwen2.5-14B-Instruct2026.0561.373,784.3
AgentPruneBackbone=Qwen2.5-14B-Instruct2026.0555.924,236.5
DyLANBackbone=Qwen2.5-14B-Instruct2026.0552.6413,926.8
CoTBackbone=Qwen2.5-14B-Instruct2026.0551.28391.4
VanillaBackbone=Qwen2.5-14B-Instruct2026.0547.36266.8