Surbhi Goel

Emergent Alignment Via Competition

In Good GRACES: Principled Teacher Selection for Knowledge Distillation

Weight Clipping for Robust Conformal Inference under Unbounded Covariate Shifts

Why Do Transformers Fail to Forecast Time Series In-Context?

Probabilistic Stability Guarantees for Feature Attributions

Progressive distillation induces an implicit curriculum

Logicbreaks: A Framework for Understanding Subversion of Rule-based Inference

Conformal Language Model Reasoning with Coherent Factuality