Joe Barrow
Overview
About Hi, I’m Joe Barrow.I’m an AI research scientist with a PhD from the University of Maryland.Most of my work is on documents (commonforms, pdftriage, segmentation), document collections (scaffolding), and building datasets (locus, commonforms).If you want to see what I’ve actually built, the software page lists tools and projects.If you’re interested in my research, check the publications page.
News
Don't Diffuse, Speculate
Diffusion and speculative decoding both trade breadth for depth. In most circumstances, speculative decoding is the better way to make that trade!
Read more
Report
A Visual Guide to the Roofline Model
My attempt at a simple/visual explanation of arithmetic intensity and the roofline model, and why they matter for making LLMs go fast.
Read more
Report
Book Review, Inference Engineering
Inference Engineering by Philip Kiely is an excellent resource and if you think you need to read it, you probably need to read it.
Read more
Report
Report
Building a 1-Outlet, 4-GPU Workstation
As a grad student, the one thing I desperately wanted was a GPU workstation. As a gainfully employed adult I can finally make that happen.
Read more
Report
Agents have (Information) Needs
Information retrieval is about satisfying an information need, but a query is a poor stand-in. Your agent is capable of expressing one, so you should probably use it.
Read more
Report
OCR's Cambrian Explosion 1 - Introduction
Strong, open VLMs enabled an explosion of open OCR model releases, with little sign of things letting up. In this survey, I detail the models, their evaluation, research trends, and open questions.
Read more
Report
Google Gemini 102 - Advanced Structured Outputs
Probing the supported output types of Gemini.
Read more
Report
Google Gemini 101 - Object Detection
Navigating Gemini's API for object detection with vision and Structured Outputs.
Read more
Report
Be Careful Interpreting Averaged Benchmarks
Thoughts on averaged benchmarks and hidden correlations.
Read more
Report
Show more news