List your company

Mathematical Reasoning on AMC, AIME, HMMT, and OlympiadBench Aggregate

54.11AccuracyDAPO w/ EDAS
11.282822.401433.5244.6386May 17, 2026

Evaluation results

MethodLinks
DAPO w/ EDASBackbone=Qwen3-8B2026.0554.11
DAPO w/ EDASBackbone=Qwen3-4B2026.0553.73
DAPOBackbone=Qwen3-4B2026.0551.84
DAPOBackbone=Qwen3-8B2026.0547.82
DAPO w/ En AdvBackbone=Qwen3-8B2026.0544.78
DAPO w/ En AdvBackbone=Qwen3-4B2026.0541.87
GRPO w/ EDASBackbone=Qwen3-8B2026.0538.35
DAPO w/ EDASBackbone=Qwen3-4B-Base2026.0537.02
GRPOBackbone=Qwen3-8B2026.0536.75
PKPOBackbone=Qwen3-8B2026.0536.57
BaselineBackbone=Qwen3-4B2026.0533.68
BaselineBackbone=Qwen3-8B2026.0533.63
DAPOBackbone=Qwen3-4B-Base2026.0527.14
DAPO w/ En AdvBackbone=Qwen3-4B-Base2026.0526.9
BaselineBackbone=Qwen3-4B-Base2026.0512.93