Publications

• Reconstructing the Mind's Eye: fMRI-to-Image with Contrastive Learning and Diffusion Priors - NeurIPS 2023, Spotlight Poster
• Challenging Error Correction in Recognized Byzantine Greek - ML4AL 2024, Best Paper
• Competition Report: Finding Universal Jailbreak Backdoors in Aligned LLMs - IEEE SaTML 2024
• Scaling Sparse Feature Circuits For Studying In-Context Learning - ICML 2025, poster
• Interpreting Large Text-to-Image Diffusion Models with Dictionary Learning - CVPR 2025 Mechanistic Interpretability for Vision Workshop, poster

Preprints

• Transcoders Beat Sparse Autoencoders for Interpretability
  • Work done at EleutherAI.

Blog posts

• EleutherAI
  • Minetester: A fully open RL environment built on Minetest
  • Product Key Memory Sparse Coders
• Lesswrong
  • Self-explaining SAE features
  • Extracting SAE task features for in-context learning
  • SAE features for refusal and sycophancy steering vectors
  • Evolutionary prompt optimization for SAE feature visualization