David McClure

real-time color naming in the browser with a lightweight model distilled from CLIP embeddings—

Lot of dystopian AI discussion on here as usual. Much is out of my control on that front, but I did use AI to build nice map of (imo) best 80,000 books of all time: bookscape.site

Select your cell: I'm prototyping an interactive sc RNA-seq portal where the UMAP is dynamically linked to morphing 3D artistic renders of each cell type. You can play with the initial build below.

Explore Wikipedia through a data map. Pages are grouped by semantic similarity, for topic clusters. Hover to see details, zoom to explore fine-grained topics, click to go to a page. Search page names to find interesting starting points for exploration.🧵 lmcinnes.github.io/datamapplot_ex…

mapping out the visual language of film using a multimodal llm: i fed frames of a short film to a vision-language model and mapped out its ratings of surrealism and presence of human figure in each moment along the timeline. the result is an interactive playback interface based on these 2 dimensions:

My new favourite thing to do is to just watch GPT2 think and attempt to answer the meaning of life

"Find visually similar images." Ok but which feature space?

Open Syllabus is proud to launch the Course Matcher, which will make it easier for the roughly 1 million US college students who transfer schools every year to receive credit for their classes. coursematcher.opensyllabus.org

Amazing graphical representation of a neural net, never seen anything like it.

Ellsworth Kelly Spectrum Colors Arranged by Chance III, 1951

@christopher @lichess Really cool - how do you encode the positions?

Some more fun with chess position embeddings using @lichess puzzles. Screenshot from an interactive version using only a subset of 1M positions for now. Clusters are just k-means clusters for debugging purposes.

Got some time to tidy up a bit this exploration and publish a live link! SPH is a little monster that needs to be harnessed :) Link: fluid.felixmartinez.dev (double click to play with the presets) #threejs #realtime #troikatext

@DavidSHolz @nickfloats

@nickfloats my personalization version is quite a bit different!

The most broadly applicable prompting technique: 1. Collect a random subset of failing examples from a training set 2. Add the examples and a correct response to your prompt 3. Repeat Doing the above 5 times has solved 90% of prompting challenges I’ve seen

Mixing colors [📹 safespacestudios]

Ask ChatGPT to pick a number between 1 and 100 - which does it pick? (by @Leniolabs_)

ask an LLM to generate a list of random words, and each one gives you its own distribution

hello world in python... using a genetic algorithm

Dot pull

Orbits. Made with #python, #matplotlib and #numpy.