Naomi Saphra (original) (raw)
Naomi Saphra
Research Fellow
About Me
I am a research fellow at the Kempner Institute at Harvard University and incoming Assistant Professor in Boston University’s faculty of Computing & Data Science. I am interested in NLP training dynamics: how models learn to encode linguistic patterns or other structure and how we can encode useful inductive biases into the training process. Recently, I have begun collaborating with natural and social scientists to use interpretability to understand the world around us. I have become particularly interested in fish. Previously, I earned a PhD from the University of Edinburgh on Training Dynamics of Neural Language Models; worked at NYU, Google and Facebook; and attended Johns Hopkins and Carnegie Mellon University. Outside of research, I play roller derby under the name Gaussian Retribution and perform standup comedy.
I am recruiting PhD students to begin in 2026 at Boston University. Do not email me before reading my contact notes if you want me to read your message.
Interests
- Language modeling
- Interpretability
- Training Dynamics
- Generalization
- AI for Scientific Understanding
Education
- PhD in Informatics
University of Edinburgh - MEng in Computer Science
Johns Hopkins University - BSc Artificial Intelligence
Carnegie Mellon University
My Research
I want to completely and comprehensively understand language model training. This objective combines linguistics, optimization, learning dynamics, science of deep learning, interpretability, and behavioral analysis. Recently, I have begun using similar approaches to study scientific discovery models and enhance broader scientific understanding.
My top three current research goals are:
- Leveraging training trajectories and variation between runs to identify what concepts are significant and distinct to a model and how they relate.
- Predicting model behavior by holistically understanding models on a computational and algorithmic level.
- Expanding human understanding of the world by studying how models learn to simulate it. To achieve this interdisciplinary objective, my current collaborations aim to deeply understand fish, stars, and weather as well as language.
My current publication list is available on my Google Scholar.
Sep 29, 2025
Apr 24, 2025
Sep 17, 2023
Jun 7, 2022
Contacting me
You do not need to email me to apply to my lab. If you want to cold email me anyway, please follow these steps to ensure I read it:
- Include the word “cuttlefish” in the subject line. (It doesn’t need context.)
- Open your email by discussing a detail from one of my papers and explaining why you found it interesting. The detail must not be described in the paper’s abstract.
If I receive a cold email asking about opportunities at my lab which does not follow the above directions, I won’t read it. I do still welcome any specific connections and questions about my research, though I may direct you towards my coauthors who did the real work related to your inquiry.
I welcome any messages from fellow disabled researchers looking to connect—I have direct personal experience in this arena.