Yu Huang

Transformers Provably Learn Chain-of-Thought Reasoning with Length Generalization

Multi-head Transformers Provably Learn Symbolic Multi-step Reasoning via Gradient Descent

A Theoretical Analysis of Self-Supervised Learning for Vision Transformers

In-context Convergence of Transformers

In-Context Learning with Representations: Contextual Generalization of Trained Transformers

What Makes for Robust Multi-Modal Models in the Face of Missing Modalities?