List your company

Reward Prediction on S2 10-task benchmark unconventional

0.031Demo L MSESARM2
0.027440.051470.07550.09953Jun 9, 2026

Evaluation results

MethodLinks
SARM2standard_full=true2026.060.031
SARM2Multi-Goal=excluded2026.060.038
SARM2Spectral Embedding=excluded2026.060.04
RW2026.060.044
RobometerFine-tuning=true2026.060.052
RobometerFine-tuning=false2026.060.113
TR2026.060.12