Rylan Schaeffer ❤️‍🔥
Rylan Schaeffer

AI/ML CS PhD Student

Stanford Trustworthy AI Research (STAIR)

About Me

I’m currently an AI/ML Computer Science PhD student at Stanford in Professor Sanmi Koyejo’s Stanford Trustworthy AI Research (STAIR) lab researching the science, engineering and mathematics of intelligence. I am supported by a Stanford Data Science PhD Scholarship and an OpenAI Superalignment Fast Grant.

This new website is currently a work-in-progress!

Education

  • PhD Artificial Intelligence

    Stanford University

  • MS Computational Science & Engineering

    Harvard University

  • MRes Cognitive Neuroscience

    University College London

  • BS Computer Science Engineering & BS Computational Statistics

    University of California, Davis

Recent Publications
Collapse or Thrive? Perils and Promises of Synthetic Data in a Self-Generating World
ZIP-FIT: Embedding-Free Data Selection via Compression-Based Alignment
Position: Maximizing Neural Regression Scores May Not Identify Good Models of the Brain
Failures to Find Transferable Image Jailbreaks Between Vision-Language Models
Uncovering Latent Memories: Assessing Data Leakage and Memorization Patterns in Large Language Models
In-Context Learning of Energy Functions
Quantifying Variance in Evaluation Benchmarks
Towards an Improved Understanding and Utilization of Maximum Manifold Capacity Representations
Why Has Predicting Downstream Capabilities of Frontier AI Models with Scale Remained Elusive?
Is Model Collapse Inevitable? Breaking the Curse of Recursion by Accumulating Real and Synthetic Data
Many-shot Jailbreaking
Bridging Associative Memory and Probabilistic Modeling
Investigating Data Contamination for Pre-training Language Models
Disentangling Fact from Grid Cell Fiction in Trained Deep Path Integrators
Testing Assumptions Underlying a Unified Theory for the Origin of Grid Cells
Self-Supervised Learning of Representations for Space Generates Multi-Modular Grid Cells
Pretraining on the Test Set Is All You Need

Rylan’s Approach to Research TLDR Research is an emotionally tough but rewarding journey. It requires struggling, learning, exploring, communicating. The key logistical details are: Each research project is a coordinate ascent problem: iterate between (a) trying to find a better goal and (b) improving your approach to achieve your current goal Especially when we’re starting to work together, let’s overcommunicate to make sure we’re on the same page Send me personal feedback early, bluntly and often Let’s meet once every week or once every two weeks (unless you have a strong personal reason to prefer not to) Maintain at least two documents: (a) a daily personal research journal and (b) a slidedeck of results with key results and the minimal information necessary to understand them Why Do Research?

Recent & Upcoming Talks
Example Talk
Example Talk

An example talk using Hugo Blox Builder's Markdown slides feature.

Jun 1, 2030