Color Analysis Plates from "Color problems; a practical manual for the lay student of color," Emily Noyes Vanderpoel, 1903.

Going through some Erdős problems that Mark Sellke and Mehtaab Sawhney found had already been solved using GPT5 pro for search. More precisely, these were listed as open, but solutions to them already existed in the literature, but not recognized as such. Quite remarkable. One example: [+]

The Entire City https://www.wikiart.org/en/max-ernst/the-entire-city-1935

First up in my series, El Maestro, Francisco Toledo. Not a strictly abstract artist, he used geometric shapes, textures, fantastical forms, ancient morphologies & symbols, native creatures in his art painting, drawing, sculpture, collage, ceramics, textiles. 🧵 1/? #MexicanArtist #AbstractArt

‘It almost feels like a big church – like I’m worshipping.’ Portia Zvavahera talks to Apollo about working in her Harare studio, which sits between a mountain and a dried-up riverbed

Portia Zvavahera

The painter maintains a flexible routine at her church-like studio in Harare, which sits between a mountain and a dried-up riverbed

buff.ly

Vision is gen-NI ! ('The shape of things unseen', Adam Zeman)

‘no sunset here’ #geometric #AbstractAug #abstract #eastcoastkin #abstractphotography #mobilephotography

bye webflow! today I moved my blog over to www.leandra.dev built with t3 + tiptap rich text editor + supabase for storage #buildinpublic

🌌🛰️🔭Want to explore universal visual features? Check out our interactive demo of concepts learned from our #ICML2025 paper "Universal Sparse Autoencoders: Interpretable Cross-Model Concept Alignment". Come see our poster at 4pm on Tuesday in East Exhibition hall A-B, E-1208!

Harry Thasarathan@hthasarathan.bsky.social · last yr.

🌌🛰️🔭Wanna know which features are universal vs unique in your models and how to find them? Excited to share our preprint: "Universal Sparse Autoencoders: Interpretable Cross-Model Concept Alignment"! arxiv.org/abs/2502.03714 (1/9)

new stream: replicating cool relief effect from amazing immersive-g.com done by Immersive Garden ninjas, with #threejs and TSL shaders, video: youtube.com/live/t64VFYm...

Has a perfectly good .alt, but I want to add that the beads are translucent, lit from the right so they glow like pieces of the sky against a black background.

Harmony merges vision-language training and self-supervised learning, achieving strong results on dense tasks with web-scraped data. It outperforms CLIP and SLIP, showing better performance in zero-shot classification and segmentation across various benchmarks. https://arxiv.org/abs/2405.14239

Harmony: A Joint Self-Supervised and Weakly-Supervised Framework for Learning General Purpose Visual Representations

ArXiv link for Harmony: A Joint Self-Supervised and Weakly-Supervised Framework for Learning General Purpose Visual Representations

arxiv.org

snapshots from the #CVPR2025 art gallery. thanks @elluba.bsky.social for organizing a great exhibit, so excited to have been included. explore the full gallery @ thecvf-art.com

a thread of the most exciting things I saw on #CVPR2025 day 1! (1) TerraMesh from IBM + ESA It’s a massive, multimodal dataset for Earth foundation models. Think ImageNet, but for the planet. 9M+ globally aligned image patches combining radar, optical, elevation, vegetation, and land use. 👇

I U N S E E N S I L E N C E I A face emerges in thick white strokes, unrecognizable yet familiar. A ghost of identity blurred by noise, time, and emotion. - The Asylum Art - Art and wordsby Yves #AbstractExpressionism #MonochromeArt #EmotionalArt #art #abstract #blueskyart #abstractart

Researchers present an analytic theory ex- https://arxiv.org/abs/2412.20292

An analytic theory of creativity in convolutional diffusion models

ArXiv link for An analytic theory of creativity in convolutional diffusion models

arxiv.org

Not Gaussian or NeRF! SuperDec rocks 3D scenes with superquadrics—compact, clever, and oh-so-smooth! This method uses transformer-based nets to predict superquadric params and soft segmentation, then refines them with Levenberg–Marquardt optimization.

Girl on a space hopper, 1971, Byker series, Newwcastle, UK , by Finnish photographer Sirkka-Liisa Konttinen #womensart #FridayFeeling

We propose Neurosymbolic Diffusion Models! We find diffusion is especially compelling for neurosymbolic approaches, combining powerful multimodal understanding with symbolic reasoning 🚀 Read more 👇

this workshop combined my loves: an abandoned wework + open source AI we built AI agents with open-source tools (smolagents, Qwen), and then used them (and our in-house robot arm) to generate and assemble mocktails! 🤖🍹 Huge thanks to 🤗 @hf.co for sponsoring & for creating smolagents 🦾

Happy birthday Arne Wolf. The under-the-radar artist, calligrapher, & educator, remembered as an influential teacher, experimental typographer, & bookmaker, was born today in 1929. “The transition from craft to art is equal to the movement from making letters to making content."

#SomethingBeautiful

Canadian Paintings@canadianpaintings.bsky.social · last yr.

Magnolia Kim Dorland 2023

Cryptic XV https://www.artic.edu/artworks/46139/

Happy bats in the margin Bodl. Library MS. Ashmole 304; 13th century; England, St. Albans; f.47v @bodleianlibraries.bsky.social

Title: Murmuration over Lanhydrock Park II Date: 2020 Medium: Oil on canvas Dimensions: 61 × 91 cm (24 × 36 in) Housing: likely private collection or with the artist #art #painting #painters #BlueSkyArt

New Paper: Continuous Thought Machines pub.sakana.ai/ctm/ Neurons in brains use timing and synchronization in the way that they compute, but this is largely ignored in modern neural nets. We believe neural timing is key for the flexibility and adaptability of biological intelligence. Thread ↓

Sakana AI@sakanaai.bsky.social · last yr.

“Continuous Thought Machines” Blog → sakana.ai/ctm Modern AI is powerful, but it's still distinct from human-like flexible intelligence. We believe neural timing is key. Our Continuous Thought Machine is built from the ground up to use neural dynamics as a powerful representation for intelligence.

[1/2] We've released the code for LegoGPT. Our autoregressive model generates physically stable and buildable designs from text prompts by integrating physics laws and assembly constraints into LLM training and inference. Code: github.com/AvaLovelace1... Website: avalovelace1.github.io/LegoGPT/

C5: Web-Scale Creative Commons Web Data - Effort to collect CC-licensed web data in one place - 147M documents explicitly CC-licensed - License detection from HTML links & metadata - Preserves exact license type & version - Maps to original URLS & crawl dates huggingface.co/datasets/Bra...

BramVanroy/CommonCrawl-CreativeCommons · Datasets at Hugging Face

We’re on a journey to advance and democratize artificial intelligence through open source and open science.

huggingface.co