Hi, I’m Elliot, a researcher from Boston, MA. My primary research interest is integrating ideas from causal inference and philosophy of science into mechanistic interpretability. Another research interest involves LLM evaluation, reasoning, self-evaluation, and hallucination detection.

Background:

Machine Learning Engineer - Eluve Inc (2024-2026)

Machine Learning Engineer - Swarm Labs (2023-2024)

M.S. Computer Science - University of Massachusetts Amherst (2020-2022)

B.S. Mathematics, Philosophy - University of Massachusetts Amherst (2016-2020)

Selected Publications

• Rajarshi Das, Ameya Godbole, Ankita Naik, Elliot Tower, Manzil Zaheer, Hannaneh Hajishirzi, Robin Jia, Andrew McCallum Knowledge Base Question Answering by Case-based Reasoning over Subgraphs (ICML 2022)

Current Research

1. Factorized circuits

Factorizing transformer weights in order to analyze information flow with the residual stream

2. Weight-space circuits

Identifying specific functional heads and edges through analyzing weights (graph-based and spectral)

3. Mechanistic validity

A framework for evaluating mechanistic claims about neural networks, importing validation methodology from philosophy of science, neuroscience, pharmacology, and measurement theory — https://mechanistic-validity.github.io/mechanistic-validity/