A while back I wrote about how we needed a word for citing statements that had been sourced from an LLM but which are low-stakes enough not to have been fact-checked.

I like the word "sprently," for this. It's already an English word that seems more or less interchangeable with …

I found myself writing "s/apparently/according to Claude 3.7 Sonnet/".

There needs to be a word.

(That string is programmer code for "whenever I write 'apparently' switch it to 'according to a possibly-hallucinating LLM.')

The essay in which I found myself writing that substitution doesn't use "apparently," except …

OpenAI claims ChatGPT 04-mini-high is its best model for coding. Screw that. The thing lies about its ability to correct its hallucinating behavior after I gave it every chance to clarify its shortcoming. If I were a junior, much less "vibe" coder, I would likely believe its assurances about doing …

ML Architecture and papers

Scaling Monosemanticity: Extracting Interpretable Features from Claude 3 Sonnet Really important article, seemingly a breakthough in peering inside LLMs. A nice adjunct to this article is the less dense An Intuitive Explanation of Sparse Autoencoders for Mechanistic Interpretability of LLMs. My big worry about this is …

“Hundreds of Beavers” restored my faith in the comedic art of cinema.

It's a silent b&w movie that combines a few live-action actors with Terry Gilliam-style animation to tell the story of Our Hero learning to survive and become a competent-enough trapper to win the hand of The Fur-Trader's …

LLMs are bland. Two new papers, Art or Artifice? Large Language Models and the False Promise of Creativity and A Robot Walks into a Bar: Can Language Models Serve as Creativity Support Tools for Comedy? An Evaluation of LLMs’ Humour Alignment with Comedians ] reinforces the findings of a more rigorous …

• There’s a new version of YOLO: https://github.com/THU-MIG/yolov10

• I’ve studied this proposal for hours and I still can’t figure out if it would be a help or a hindrance. It’s a diagramming language for tensor math, which is a good idea, but I …

• Interesting article about a classic pen-and-paper game: Ace of Aces: or, why you should Do Maths as a game designer

• Activism is not a Social Club

• “AI” Presets for Underwater Photography: Adaptive AI Preset Pack This is a good idea, especially if they really are doing the feasible thing, which …

BirdCLEF is a bird-call recognition contest that runs yearly on Kaggle. (CLEF stands for Cross Language Education and Function.)

Last year, I participated in BirdCLEF because the domain was native Hawaiian birds, so even though I didn't know much of anything about audio ML, I had a leg up in …