All the wet weather this winter has indeed brought May flowers in Northern California.
Since the high of the marathon last month I have fully recovered and started a new training cycle for my first Dipsea on June 11. The Dipsea is an 8 mile elevation-heavy trail race from downtown Mill Valley to Stinson Beach and is the oldest organized trail race in the US. It’s limited to 1500 people and berths go first to previous participants before opening to new racers, so I feel very lucky to have been accepted on my first try. I’ll drop a race report when it’s done.
Meanwhile, the tabs collected. No individual-link hot takes this month, but the AI content is unrelenting and difficult to absorb and reflect on.
Ok, maybe just one 🔥. The AI Cannon post from a16z has tens of hours of reading, which may or may not be required (essential?) to understand the nature of the current boom. They last boom they hyped was quite hollow, but that shouldn’t discount the insights they are sharing about the AI canard.
Dev
- Advanced Character Physics
- The Early History Of Smalltalk
- A practical guide to WebAssembly memory
- cohost! - “I want to talk about WebGPU”
- The UX Research Reckoning is Here | by Judd Antin | One Big Thought | May, 2023 | Medium
- Web3’s fake version of Web history - by Dave Karpf
- Creating a shine effect on a card with Tailwind CSS
- Building a modern design system in layers
- Pixels vs. Ems: Users DO Change Font Size | by Evan Minto | Medium
- macOS Internals
- 2023 Apple Design Award Finalists - Apple Design Awards - Apple Developer
Blogs
- Early Computer Art in the 50’s & 60’s — Amy Goodchild
- Ken Kocienda
- Building vs. optimizing | Adam Stoddard
- Erin Kissane
- The Cult of the Founders
AI
Navel-gazing
- Leaked Google document: “We Have No Moat, And Neither Does OpenAI”
- AI Canon | Andreessen Horowitz
- thread from a16z GP Anish Acharya
- LLM University, Generative AI, AI Product Moats
- Superintelligence: The Idea That Eats Smart People
- Google’s AI Hype Circle. We have to do Bard because everyone… | by Cory Doctorow | May, 2023 | Medium
Papers / Models
- StarCoder: A State-of-the-Art LLM for Code
- Gorilla
- Voyager | An Open-Ended Embodied Agent with Large Language Models
Code
- ianarawjo/ChainForge: An open-source visual programming environment for battle-testing prompts to LLMs.
- ray-project/llm-numbers: Numbers every LLM developer should know
- Jsonformer: A Bulletproof Way to Generate Structured JSON from Language Models
- SamurAIGPT/privateGPT: An app to interact privately with your documents using the power of GPT, 100% privately, no data leaks
- brexhq/prompt-engineering: Tips and tricks for working with Large Language Models like OpenAI’s GPT-4.
- Okapi BM25 - Wikipedia