Publications

Research papers and technical writing.

Filter:

CRL: Concept Bottleneck Sparse Autoencoders for Scalable Concept Representation Learning

CRL: Concept Bottleneck Sparse Autoencoders for Scalable Concept Representation Learning

Mechanistic Interpretability Sparse Autoencoders Steering Vectors LLM Safety

Confidence Manifold: Revealing LLM Internal States via Topological Inconsistency Analysis

Confidence Manifold: Revealing LLM Internal States via Topological Inconsistency Analysis

LLM Safety Mechanistic Interpretability

CorrSteer: Steering LLMs via Correlation-based Corrections

CorrSteer: Steering LLMs via Correlation-based Corrections

Steering Vectors Mechanistic Interpretability LLM Safety

FaithfulSAE: Evaluating Faithfulness of Sparse Autoencoders in Large Language Models

FaithfulSAE: Evaluating Faithfulness of Sparse Autoencoders in Large Language Models

Sparse Autoencoders Mechanistic Interpretability

LibVulnWatch: Automated Vulnerability Monitoring Using LLMs for Open-Source Libraries

LibVulnWatch: Automated Vulnerability Monitoring Using LLMs for Open-Source Libraries

RTSum: Relation Triple-based Interpretable Summarization with Multi-level Salience Visualization

RTSum: Relation Triple-based Interpretable Summarization with Multi-level Salience Visualization

NAACL 2024 Demo