Hey 👋🏽, my name is Will. I work as a Machine Learning Scientist at Tempus, where I focus on developing medically-focused AI with language models. See our recent paper “Large Language Models with Retrieval-Augmented Generation for Zero-Shot Disease Phenotyping” ( DGM4H NeurIPS 2023) to get a sense of the type of problems my group is trying to solve.
In a previous life, I worked in quantitative research/engineering in trading of the high frequency variety (i.e. “HFT”). Prior to that, I was a PhD student at Northwestern University in Applied Math.
Writing
These are my notes on Large Language Model/Transformer research. I started an earlier set of notes here. Inspired by Lilian Weng’s blog. I am writing these for the sake of writing.
→ What We Know About LLMs (A Primer) (July 23 2023) Hacker News Top Post 💫
→ Emergent Abilities of LLMs 🪴 (August 22 2023)
Note: I don’t have a background in academia, so writing papers hasn’t been a focus.
→ Will Thompson, David Vidmar, Jessica De Freitas, Gabriel Altay, Kabir Manghnani, Andrew Nelsen, Kellie Morland, John Pfeifer, Brandon Fornwalt, RuiJun Chen, Martin Stumpe, Riccardo Miotto “Large Language Models with Retrieval-Augmented Generation for Zero-Shot Disease Phenotyping.” DGM4H NeurIPS 2023.
→ David Vidmar, Will Thompson, Ruijun Chen, Dustin Hartzel, Daniel Rocha, Joseph Leader, Brandon Fornwalt, and Christopher M Haggerty. 2022. “Abstract 11934: Natural Language Processing Models Can Be Trained to Accurately Recognize the Presence of Disease Within Clinical Notes.” Circulation 146 (Suppl_1): A11934–A11934.
Miscellanea
- microstructure-plotter (2023) - visualize market microstructure data.
- tldr-transformers (2021)- earlier set of notes comparing transformer research threads.
- deeplearning-nlp-models (2020)- built a few deep learning models from scratch in Pytorch (w/ tests). Run on GPUs.