A review of recent research on concerning characteristics of LLMs — CONTENT WARNING: This post contains examples of biased, toxic text generated by LLMs. This post provides a deep dive into recent research on bias, toxicity, and jailbreaking of large language models (LLMs), especially ChatGPT and GPT-4. I’ll discuss the ethical guidelines companies are currently using in LLM development and the…

Highlights from 95 Pages — I recently finished reading the intriguing paper Sparks of Artificial General Intelligence: Early experiments with GPT-4 published by researchers at Microsoft in spring 2023. In this post, I provide a condensed overview of the paper’s content. The central claim of the paper is that the large language model GPT-4 demonstrates…

Becoming a Physician Scientist — In May 2022 I became the first person to graduate from Duke University with an MD and a PhD in Computer Science. The process took eight years: three years of medical school and five years of graduate school. One of the most common questions I heard from other students and…

Creating an LLM that isn’t racist or sexist — Large language models (LLMs) like ChatGPT are racist, sexist, homophobic, and in general packed full of all of the worst of society’s biases, because they are trained on biased data. As Brown et al. state, “internet-trained models have internet-scale biases.” …

Manipulate dataframes like a pro for quicker data cleaning — Back when I spent a lot of my time cleaning datasets, I had a bright yellow sticky note on my desk with lines of numpy and pandas syntax that I used all the time. …

Bring a Chair on Rounds and Other Tips — In my last year of medical school, I completed most of my clinical clerkships while pregnant, including an internal medicine sub-internship during my third trimester. …

I am the CEO of an early-stage health tech startup, Cydoc. Over the past four years I’ve interviewed hundreds of people for software engineering positions. …

I am thrilled to announce that as of today, 3,630 whole CT scans from the RAD-ChestCT dataset are publicly available on Zenodo, along with abnormality and location labels! You can access the dataset here. After a multi-year saga of bureaucracy and technical barriers, the data is finally up and you…

SARLE is a customizable technique for extracting structured abnormality and location labels automatically from free text radiology reports — Classification labels specifying the presence and absence of abnormalities are necessary to train computer vision models on radiology images. However, obtaining these classification labels by hand is time-consuming and limits the size of the final dataset as well as the number of abnormalities that can be considered. …