List your company

Mathematical Reasoning on AIME 25 (Last score, Best score)

0.5261Last ScoreRPG-URKL
0.3277720.3792610.430750.482239May 23, 2025

Evaluation results

MethodLinks
RPG-URKLContext length=8k2025.050.52610.4938
RPG-REINFORCE-URKLContext length=8k2025.050.50730.5208
RPG-REINFORCE-UFKLContext length=8k2025.050.48330.5031
RPG-UFKLContext length=8k2025.050.46980.4865
DAPOContext length=8k2025.050.44690.474
GRPOContext length=8k2025.050.33540.4063