List your company

Mathematical Reasoning on AMC (Accuracy, Length)

91.56AccuracyDDPO
87.992888.918989.84590.7711Mar 19, 2026

Evaluation results

MethodLinks
DDPOBackbone=Qwen3-4B2026.0391.564,074
DDPOBackbone=Qwen3-14B2026.0391.093,622
DDPOBackbone=Qwen3-8B2026.0390.474,445
L1Backbone=Qwen3-14B2026.0390.314,957
GRPOBackbone=Qwen3-8B2026.0390.316,013
GRPOBackbone=Qwen3-14B2026.0389.844,080
SHORTER BETTERBackbone=Qwen3-14B2026.0388.443,600
GRPO-LEADBackbone=Qwen3-14B2026.0388.444,823
A-DLPBackbone=Qwen3-14B2026.0388.133,998
GRPOBackbone=Qwen3-4B2026.0388.136,031