Neil Pagarkar Rathi

Hi! I am an undergraduate at Stanford advised by Dan Jurafsky.

My research focuses on modeling cognition and language using insights from machine learning, experimental psych, linguistic theory, and neuroscience.

In my free time, I run logistics for the Stanford Math Tournament, play Hindustani classical sitar, and go on drives down the California coast. I also take notes.

You can reach me via email at rathi at stanford dot edu.

also on: github | scholar | twitter

Publications

Neil Rathi*, Johannes Mehrer*, Badr Alkhamissi, Taha Huraib, Nicholas Blauch and Martin Schrimpf. 2024. TopoLM: brain-like spatio-functional organization in a topographic language model. arXiv preprint. [pdf]

Neil Rathi, Brandon Waldon, and Judith Degen. 2024. Informativity and accessibility in incremental production of the dative alternation. CogSci 2024. [pdf] [code]

Neil Rathi. 2023. The architecture gave the language model the constituent length preferences. BlackBoxNLP. [pdf]

Neil Rathi, Michael Hahn, Richard Futrell. 2022. Explaining patterns of fusion in morphological paradigms using the memory-surprisal tradeoff. CogSci 2022. Best Undergrad Paper. [pdf] [code]

Neil Rathi, Michael Hahn, Richard Futrell. 2021. An Information-Theoretic Characterization of Morphological Fusion. EMNLP 2021. SIGTYP Best Paper. [pdf] [code]

Neil Rathi. 2021. Dependency locality and neural surprisal as predictors of processing difficulty: Evidence from reading times. CMCL @ NAACL 2021. [pdf]

Other Stuff

A pipeline for running massive online spoken production experiments with Empirica.
A script for automatic reading-time dataset annotation.
My CS 428a final paper, on learning function word semantics with/without a language of thought.
My CS 231n final paper, on activation patching for adversarial example circuits in GoogLeNet
My writing in the major paper for Math 120, on impossibility of construction.