Thomas McGrath

This work finds that simple difference of means vectors coherently represent reward hacking in Kimi K3, GLM 5.2, and Qwen 3.8 Max, and provides evidence that simple, white-box methods can be used to scalably study and monitor reward hacking behaviors in frontier open source models.

Uncertainty in LLM reasoning is largely a sampling artifact, allowing for more efficient analyses that cut costs without sacrificing accuracy.

Transforming post-training from opaque reward optimization into a transparent process of auditing and sculpting the learning signal could revolutionize how we guide model behavior.

Steering neural networks through the intrinsic geometry of their activations unlocks more natural and controllable behaviors than traditional linear interventions.