mlepori1

NeuroSurgeon. NeuroSurgeon is a package that enables researchers to uncover and manipulate subnetworks within models in Huggingface Transformers

Compositional_Subnetworks. Code for the paper Break It Down: Evidence for Structural Compositionality in Neural Networks, presented at NeurIPS 2023

Circuit_Probing. Codebase for Uncovering Causal Variables in Transformers Using Circuit Probing

Representations_Of_Syntax. Code and Resources for the paper 'Representations of Syntax [MASK] Useful: Effects of Constituency and Dependency Structure in Recursive LSTMs' by Lepori, Linzen, and McCoy

Picking_BERTs_Brain. Code and Data for Picking BERT's Brain: Probing for Linguistic Dependencies in Contextualized Embeddings Using Representational Similarity Analysis by Michael Lepori and R. Thomas McCoy, presented at COLING 2020

Racing_Thoughts. Code and data to accompany Racing Thoughts by Lepori et al. 2025

Interpretable_Encoding_Models. Jupyter Notebook

TreeTransformers. Modified transformer architecture, replacing linear positional encodings with tree-based positional encodings.