reading

Simon Willison summarizes Andrej Karpathy's review of Meta's Llama 3, noting the increase in training tokens and tokenizer size but also the disappointingly small context length.

via @simonw

Wired's Steven Levy details how eight Google employees invented the "transformer" concept, leading to a recent tech breakthrough. via kottke.org

The future of non-alcoholic beverage retail is explored following the bankruptcy of Boisson, highlighting emerging trends. via Modern Retail.

Authors Maria Farrell and Robin Berjon draw parallels between ecosystem monoculture and the internet's consolidation under a few tech giants, advocating for antitrust laws and interoperability mandates to "rewild" it. via Noema Magazine

Ken Kantzer shares practical lessons learned from using OpenAI's models, including a simple token estimation method and challenges with structured data extraction.

via Simon Willison's Weblog

Simon Willison quotes Molly White's argument that large language models may only offer small productivity gains because they're only somewhat useful for mundane tasks. via @simonw