Researcher at Stanford University
An undergraduate student in Computer Science at the California Institute of Technology who conducts research.
How media typically covers Song Peiyang
Referenced in coverage
Stanford researchers introduce a comprehensive taxonomy framework categorizing LLM reasoning failures into embodied/non-embodied types and fundamental, application-specific, and robustness failure classes to systematically understand persistent reasoning gaps.
“Co-author of Stanford paper on Large Language Model Reasoning Failures survey.”