Andrew Lee

Into the Rabbit Hull: From Task-Relevant Concepts in DINO to Minkowski Geometry

Agentic reinforcement learning for search is unsafe

Why Can't Transformers Learn Multiplication? Reverse-Engineering Implicit Chain-of-Thought Reveals Challenges of Learning Long-Range Dependencies

The Geometry of Self-Verification in a Task-Specific Reasoning Model

From Isolation to Entanglement: When Do Interpretability Methods Identify and Disentangle Known Concepts?

Shared Global and Local Geometry of Language Model Embeddings

ICLR: In-Context Learning of Representations