HealthBench is a new evaluation benchmark for AI in healthcare which evaluates models in realistic scenarios. Built with input from 250+ physicians, it aims to provide a shared standard for model performance and safety in health.

Discover how Nix can revolutionize your software supply chain security, enabling verifiable integrity and offline rebuilds from source.

We introduce the Byte Latent Transformer (BLT), a new byte-level LLM architecture that, for the first time, matches tokenization-based LLM performance at scale with significant improvements in...

A security research blog.

choose a classifier

No more rigging your own fixes, download the official part.

A Conversation with an AI Expert on Technology’s Role in Science and National Security

That doesn’t spy on you

When GADTs (Generalized Algebraic DataTypes) landed inOCaml, I wasn’t particularly happy about it. I assumed that it was the kind ofnonsense you get when you...

Compliance-as-a-service platform for tech startups to achieve SOC 2 readiness efficiently and affordably.

Midi IN and OUT. Standard midi file parser and player. Midi Softsynth implementation. - meshula/LabMidi

has anyone coined the term "fast tech" yet? used in the same sense as "fast fashion" - stuff that's designed to be cheap and disposable, chasing short term fads and tends, built in wasteful quantities in full knowledge that many units will go to landfill…

It's Not a Hypothetical, I've Already Lost My Job to AI For The Last Year

Air traffic control has been in the news lately, on account of my country's declining ability to do it. Well, that's a long-term trend, resulting from decades of under-investment, severe capture by our increasingly incompetent defense-industrial complex,…

What if humanity forgot how to make CPUs?

Imagine Zero Tape-out Day (Z-Day), the moment where no further silicon designs ever get manufactured. Advanced core designs fare out very badly.

Assuming we keep our existing supply, here’s how it would play out:

Background

This repository contains the official implementation of "FastVLM: Efficient Vision Encoding for Vision Language Models" - CVPR 2025 - apple/ml-fastvlm