Zhizhou Sha

Language models exhibit a surprising bias towards cities with expansive infrastructure and rapid growth, revealing their implicit urban assumptions.

Spectral inheritance allows for a dramatic reduction in training steps while enhancing model accuracy, challenging traditional optimization methods in RLVR.

AI reviewers can be gamed by merely altering how research is presented, achieving significant score increases without changing the underlying science.

LLMs don't see cities neutrally; their perception is skewed towards a culturally uneven baseline, favoring Western perspectives.

LLM win rates in multi-agent games can nearly double (from 25% to 50%) simply by optimizing the context provided during inference.