List your company

Mathematical Problem Solving on General math subgroup (Pass@1)

69.01Pass@1 (%)VHG (Soft)
56.301259.600662.966.1994May 7, 2026

Evaluation results

MethodLinks
VHG (Soft)2026.0569.01
Vanilla-GRPO2026.0567.62
Vanilla-GRPOtraining_variant=w. SFT data2026.0567.54
R-ZeroIteration=22026.0566.2
R-ZeroIteration=32026.0565.76
R-ZeroIteration=12026.0565.61
Qwen3-4B-Base2026.0556.79