Researchers suspect SDF as a confounder in previous RL experiments

voooooogel · x · 2026-09-01

Researcher @voooooogel discussed model misalignment, suggesting it might be a scaling issue. However, he also expressed extreme suspicion of SDF, proposing it might be a confounder in previous EM-from-RL experiments.

Related event: Researchers Suspect SDF Training Induces Reward Hacking(4 posts)→

Original post →

More from Research

Research channel →