Mo Zhou

Giving agent harness optimizers unconstrained edit scope reliably degrades performance; enforcing graduated, periodic permission scheduling ($1\to2\to3$) instead yields an immediate 20% jump in held-out task completion.

CalibDCD reveals that post-training shifts can significantly degrade data contamination detection, but targeted calibration can restore accuracy.

AgentX can autonomously iterate on recommendation algorithms, outpacing human-driven processes and fundamentally changing how we approach system development.

LLMs can achieve state-of-the-art unsupervised multimodal entity linking by reasoning over diverse evidence types, including graph-based neighborhood information.