List your company

Mathematical Reasoning on GSM8K (pass@1/mean@32)

86.2Accuracy (pass@1/mean@32)No sampling
85.78485.8928686.108Apr 30, 2026

Evaluation results

MethodLinks
No samplingSetting=GRPO2026.0486.2
No samplingSetting=ZVF2026.0486.1
p*Setting=GRPO2026.0486
p* smooth(alpha = 0.01)Setting=GRPO2026.0486
p*Setting=ZVF2026.0485.9
p* smooth(alpha = 0.01)Setting=ZVF2026.0485.8