Anthropic: "Applicants should not use AI assistants"

"A computer can never be held accountable"

Notes on OpenAI o3-mini

Qwen2.5-7B-Instruct-1M and Qwen2.5-14B-Instruct-1M

My AI/LLM predictions for the next 1, 3 and 6 years

Promising results from DeepSeek R1 for code

My approach to running a link blog

I still don't think companies serve you ads based on your microphone

Things we learned about LLMs in 2024

Trying out QvQ – Qwen's new visual reasoning model

Building Python tools with a one-shot prompt using uv run and Claude Projects

I can now run a GPT-4 class model on my laptop

First impressions of the new Amazon Nova LLMs (via a new llm-bedrock plugin)

Pelicans on a bicycle

Ask questions of SQLite databases and CSV/JSON files in your terminal

/usr/bin/env -S uv run

Storing Times for Human Events

Qwen2.5-Coder-32B is an LLM that can code well that runs on my Mac

Private School Labeler on Bluesky

Visualizing local election results with Datasette, Observable and MapLibre GL

Foursquare Open Source Places: A new foundational dataset

MDN Browser Support Timelines

SmolLM2

New OpenAI Feature: Predicted Outputs

Compiling and running sqlite3-rsync from a branch

You can now run prompts against images, audio and video in your terminal using LLM

LLM inference library written in Rust with mistral.rs

Video scraping: extracting JSON from a 35s screen capture for 1/10th of a cent

Notes on the new Claude analysis JavaScript code execution tool

The 3 AI Use Cases: Gods, Interns, and Cogs

More →