1: Transformers.js . Run HuggingFace Transformers directly in your browser, with no need for a server. (Lots of AI applications use Python because lots of the research used Python because certain libraries worked well in that language. But to deploy an application with Python you need to host something like a Flask server somewhere. Whereas Javascript can run in browser!)
2: Stripe’s Operating Principles. An example of company values that actually seem useful. The section specific to leaders seems particularly helpful.
3: Shoelace formula - Wikipedia. An algorithm to determine the area of a polygon with lots of sides. Used in Advent of Code 2023
4: Jess’s Workshop via X. Design trends come and go. Are skeuomorphisms back?
5: Why you can’t afford to NOT work in sustainability. “Society’s hardest problems only get solved with capital and talent. And good talent is needed to advocate for the capital, so it’s more like talent^2. That’s why YOU need to choose.”
6: Mixtral-8x7B. Run 85B parameters with the same compute as 14B, and 6X faster than Llama 2 70B while outperforming. For reasons I don’t understand. How much better can open source models get? Time to rewrite When local LLMs will be useful
7: Funnel. Glad to see someone developing privacy first quick capture. I’d like to build something similar with plaintext that uses some LLM sorting and vector search.
8: Snipd . AI powered podcast player that lets you easily save a transcribed snippet to your notes. Incumbents have been slow to move — I guess the content ecosystem is going to get more complicated before it gets simpler? The number of apps I’m using to manage my content ecosystem keeps increasing.
9: Newborn King and Nativity Scene from Matt Chinworth. For Christmas, some beautiful renditions of Joseph with baby Jesus.
10: FunSearch: Making new discoveries in mathematical sciences using Large Language Models Google Deep Mind runs LLMs in a loop (prompt + evaluate) and that’s enough to discover some cutting edge programming discoveries?…
11: The Little Book of Deep Learning. A short, free introduction to deep learning released in September 2023. Also, designed to be read on a phone screen.
12: Rick Rubin production discography - Wikipedia. Who else has such a breadth of work over so many years in a given field?
13: Open Components Ecosystem. Open source “reusable building blocks for Bible Technology.”
14: Patrick Collison on Pollution A reminder to open the windows when I can.
15: Federal minimum wage effects less than 1.5% of hourly workers. Note the chart is in nominal terms (not inflation adjusted).
16: Charlottesville VA ambitious upzoning. In the works for a while but now passed by city council. Long small towns?
17: New US nuclear power capacity. Important because the NGFS scenarios core to most climate transaction models (perhaps understandably) don’t consider a possible nuclear comeback.
18: The Great Heirloom Transfer. I think cultivating taste will become more and more important.
19: Re NYT vs OpenAI Lawsuit - NYT added a clause against using its content to train AI systems to the TOU on August 3rd 2023. Feels a bit late on the draw?
20: Bento. Link-in-bio, but pretty. Not a skeuomorphism but again is taste becoming relatively more important?
21: Resolutions for human flourishing. “We need a self-help that shows us how to flourish as a whole person, not one that merely offers advice on improving your work habits and anxiously avoiding anxiety.”
Disclaimer: Something something sharing a link is a recommendation but not necessarily an endorsement or a sign of agreement something something.