Hidenori Tanaka

Provable Low-Frequency Bias of In-Context Learning of Representations

Belief Dynamics Unify In-Context Learning and Activation Steering

Emergence of Hierarchical Emotion Organization in Large Language Models

New News: System-2 Fine-tuning for Robust Integration of New Knowledge

Competition Dynamics Shape Algorithmic Phases of In-Context Learning

A Percolation Model of Emergence: Analyzing Transformers Trained on a Formal Language

Swing-by Dynamics in Concept Learning and Compositional Generalization

ICLR: In-Context Learning of Representations

In-Context Learning Strategies Emerge Rationally

Forking Paths in Neural Text Generation

Representation Shattering in Transformers: A Synthetic Study with Knowledge Editing

Dynamical phases of short-term memory mechanisms in RNNs

Emergence of Hierarchical Emotion Representations in Large Language Models

Representation Shattering in Transformers: A Synthetic Study with Knowledge Editing

Continuous-Time Analysis of Adaptive Optimization and Normalization