Samuel Marks

Steering Evaluation-Aware Language Models To Act Like They Are Deployed

Liars' Bench: Evaluating Deception Detectors for AI Assistants

Steering Out-of-Distribution Generalization with Concept Ablation Fine-Tuning

Unsupervised Elicitation of Language Models

Inoculation Prompting: Instructing LLMs to misbehave at train-time improves test-time alignment

Eliciting Secret Knowledge from Language Models

Robustly Improving LLM Fairness in Realistic Settings via Interpretability

Sparse Feature Circuits: Discovering and Editing Interpretable Causal Graphs in Language Models

NNsight and NDIF: Democratizing Access to Open-Weight Foundation Model Internals

Erasing Conceptual Knowledge from Language Models

SAEBench: A Comprehensive Benchmark for Sparse Autoencoders in Language Model Interpretability

Erasing Conceptual Knowledge from Language Models