- January - 5 entries, 29 links, 8 quotes
- 9th: Datasette 0.64, with a warning about SpatiaLite
- 13th: How to implement Q&A against your documentation with GPT3, embeddings and Datasette
- 15th: Weeknotes: AI hacking and a SpatiaLite tutorial
- 27th: Exploring MusicCaps, the evaluation data released to accompany Google's MusicLM text-to-music model
- 30th: datasette-scraper, Big Local News and other weeknotes
- February - 6 entries, 31 links, 13 quotes
- 9th: Weeknotes: A bunch of things I learned this week, plus datasette-explain
- 15th: Bing: "I will not harm you unless you harm me first"
- 17th: Analytics: Hacker News v.s. a tweet from Elon Musk
- 19th: I talked about Bing and tried to explain language models on live TV!
- 21st: In defense of prompt engineering
- 24th: Thoughts and impressions of AI-assisted search from Bing
- March - 10 entries, 46 links, 20 quotes
- 7th: Weeknotes: NICAR, and an appearance on KQED Forum
- 10th: ChatGPT couldn’t access the internet, even though it really looked like it could
- 11th: Large language models are having their Stable Diffusion moment
- 13th: Stanford Alpaca, and the acceleration of on-device large language model development
- 17th: Could you train a ChatGPT-beating model for $85,000 and run it in a browser?
- 18th: A conversation about prompt engineering with CBC Day 6
- 22nd: Don't trust AI to talk accurately about itself: Bard wasn't trained on Gmail
- 22nd: Weeknotes: AI won't slow down, a new newsletter and a huge Datasette refactor
- 24th: I built a ChatGPT plugin to answer questions about data hosted in Datasette
- 27th: AI-enhanced development makes me more ambitious with my projects
- April - 17 entries, 45 links, 19 quotes
- 2nd: What AI can do for you on the Theory of Change podcast
- 2nd: Think of language models like ChatGPT as a "calculator for words"
- 4th: Semi-automating a Substack newsletter with an Observable notebook
- 4th: Weeknotes: A new llm CLI tool, plus automating my weeknotes and newsletter
- 7th: We need to tell people ChatGPT will lie to them, not debate linguistics
- 8th: Working in public
- 8th: The Changelog podcast: LLMs break the internet
- 10th: Thoughts on AI safety in this era of increasingly powerful open source LLMs
- 12th: Running Python micro-benchmarks using the ChatGPT Code Interpreter alpha
- 14th: Prompt injection: What's the worst that can happen?
- 15th: sqlite-history: tracking changes to SQLite tables using triggers (also weeknotes)
- 16th: Web LLM runs the vicuna-7b Large Language Model entirely in your browser, and it's very impressive
- 17th: What's in the RedPajama-Data-1T LLM training set
- 20th: Data analysis with SQLite and Python for PyCon 2023
- 23rd: Weeknotes: Citus Con, PyCon and three new niche museums
- 25th: The Dual LLM pattern for building AI assistants that can resist prompt injection
- 29th: Enriching data with GPT3.5 and SQLite SQL functions
- May - 12 entries, 44 links, 14 quotes
- 1st: Weeknotes: Miscellaneous research into Rye, ChatGPT Code Interpreter and openai-to-sqlite
- 1st: Let's be bear or bunny
- 2nd: download-esm: a tool for downloading ECMAScript modules
- 2nd: Prompt injection explained, with video, slides, and a transcript
- 4th: Midjourney 5.1
- 4th: Leaked Google document: "We Have No Moat, And Neither Does OpenAI"
- 8th: Big Opportunities in Small Data
- 10th: Weeknotes: sqlite-utils 3.31, download-esm, Python in a sandbox
- 11th: Delimiters won't save you from prompt injection
- 18th: llm, ttok and strip-tags - CLI tools for working with ChatGPT and other LLMs
- 27th: Lawyer cites fake cases invented by ChatGPT, judge is not amused
- 30th: ChatGPT should include inline tips
- June - 5 entries, 18 links, 6 quotes
- 4th: It's infuriatingly hard to understand how closed models train on their input
- 4th: Weeknotes: Parquet in Datasette Lite, various talks, more LLM hacking
- 8th: Understanding GPT tokenizers
- 18th: Symbex: search Python code for functions and classes, then pipe them into a LLM
- 27th: Weeknotes: symbex, LLM prompt templates, a bit of a break
- July - 4 entries, 20 links, 7 quotes
- 12th: My LLM CLI tool now supports self-hosted language models via plugins
- 16th: Weeknotes: Self-hosted language models with LLM plugins, a new Datasette tutorial, a dozen package releases, a dozen TILs
- 18th: Accessing Llama 2 from the command-line with the llm-replicate plugin
- 24th: sqlite-utils now supports plugins
- August - 7 entries, 28 links, 10 quotes
- 1st: Run Llama 2 on your own Mac using LLM and Homebrew
- 3rd: Catching up on the weird world of LLMs
- 5th: Weeknotes: Plugins for LLM, sqlite-utils and Datasette
- 6th: How I make annotated presentations
- 16th: Datasette Cloud, Datasette 1.0a3, llm-mlc and more
- 27th: Making Large Language Models work for you
- 30th: Datasette 1.0a4 and 1.0a5, plus weeknotes
- September - 5 entries, 35 links, 5 quotes
- 4th: LLM now provides tools for working with embeddings
- 12th: Build an image search engine with llm-clip, chat with models with llm chat
- 17th: Weeknotes: Embeddings, more embeddings and Datasette Cloud
- 29th: Talking Large Language Models with Rooftop Ruby
- 30th: Things I've learned about building CLI tools in Python
- October - 8 entries, 27 links, 10 quotes
- 1st: Weeknotes: the Datasette Cloud API, a podcast appearance and more
- 14th: Multi-modal prompt injection image attacks against GPT-4V
- 17th: Open questions for AI engineering
- 22nd: Weeknotes: PyBay, AI Engineer Summit, Datasette metadata and JavaScript plugins
- 23rd: Embeddings: What they are and why they matter
- 26th: Execute Jina embeddings with a CLI using llm-embed-jina
- 26th: Now add a walrus: Prompt engineering in DALL‑E 3
- 30th: DALL-E 3, GPT4All, PMTiles, sqlite-migrate, datasette-edit-schema
- November - 8 entries, 30 links, 15 quotes
- 7th: ospeak: a CLI tool for speaking text in the terminal via OpenAI
- 10th: Financial sustainability for open source projects at GitHub Universe
- 15th: Exploring GPTs: ChatGPT in a trench coat?
- 22nd: Deciphering clues in a news article to understand how it was reported
- 22nd: Weeknotes: DevDay, GitHub Universe, OpenAI chaos
- 25th: I'm on the Newsroom Robots podcast, with thoughts on the OpenAI board
- 27th: Prompt injection explained, November 2023 edition
- 29th: llamafile is the new best way to run an LLM on your own computer
- December - 7 entries, 26 links, 12 quotes
- 1st: Datasette Enrichments: a new plugin framework for augmenting your data
- 8th: Weeknotes: datasette-enrichments, datasette-comments, sqlite-chronicle
- 14th: The AI trust crisis
- 18th: Many options for running Mistral models in your terminal using LLM
- 20th: Recommendations to help mitigate prompt injection: limit the blast radius
- 31st: Last weeknotes of 2023
- 31st: Stuff we figured out about AI in 2023