←
Blake Richards23,087 posts
Blake Richards avatar

Blake Richards

@tyrell_turing

Researcher at @mcgillu combining AI and neuroscience. Also on Bluesky (@tyrellturing.bsky.social) and Mastodon: @tyrell_turing@fediscience.org.

◷ Joined April 2013▧ 549 Photos and videos
Blake Richards avatar
Blake Richards@tyrell_turing·Oct 30

1/4) We're hiring in my lab, **please share**! 🚀 There are two postdoc opportunities in cog neuro and AI. This is for a new project supported by a Schmidt Polymath Award. This project aims to build an engaging video game to collect data for neurofoundation models. 🧠🤖

Blaise Agüera (@blaiseaguera.bsky.social) avatar
⟳ Blake Richards retweeted
Blaise Agüera (@blaiseaguera.bsky.social)@blaiseaguera·21h

Investment in AI infrastructure now dwarfs historic spending on railroads and the Internet combined, yet investment in our civic and scientific institutions hasn't kept pace. In @ScienceMagazine, @profjamesevans, visiting faculty in our Paradigms of Intelligence team, and @glenweyl explore how to close that gap: science.org/doi/10.1126/scie…

Ravid Shwartz Ziv avatar
⟳ Blake Richards retweeted
Ravid Shwartz Ziv@ziv_ravid·Nov 9

1/ Two years ago I scrolled to the bottom of the ICLR rankings and found art. This year I downloaded 125,680 ICLR 2027 reviews the day after they went public. There was more art 🧵

clareLab avatar
⟳ Blake Richards retweeted
clareLab@c1areLab·Nov 6

ICLR 2027 Rating Distribution

Post attachment
alphaXiv avatar
⟳ Blake Richards retweeted
alphaXiv@askalphaxiv·Nov 1

Yann LeCun and Randall Balestriero just Proved the Platonic Representation Hypothesis, and found a better way to achieve it. The Platonic Representation Hypothesis suggests that AI models trained on different modalities, like images and text, eventually learn similar representations of the world. But this paper shows that how models are trained matters more than simply scaling them up. So they mathematically show that, under certain conditions, two JEPAs trained completely separately can learn essentially the same underlying representation of the world. One learns from different images of the same scene, while the other learns from different descriptions of that scene, without ever being trained on image-text pairs. Their approach achieves the same image-text alignment with 42x fewer training samples than next-token prediction. A 109M-parameter text JEPA even outperforms LLMs up to 72B parameters on cross-modal alignment benchmarks. alphaxiv.org/abs/2610.39302

Post attachment
Anton Arkhipov avatar
⟳ Blake Richards retweeted
Anton Arkhipov@AntonSArkhipov·Nov 1

Paper submission for our 2nd AAAI Workshop on 𝗡𝗲𝘂𝗿𝗼 𝗳𝗼𝗿 𝗔𝗜 & 𝗔𝗜 𝗳𝗼𝗿 𝗡𝗲𝘂𝗿𝗼 is up and running, due 11/20! We'll have archival full and non-archival short papers. All details and the OpenReview submission link are at neuroai-multimodal-workshop.…

Post attachment
Blake Richards avatar
Blake Richards@tyrell_turing·Oct 30

4/4) Additional notes: 📆 Application deadline is Nov 30, 2026 🌍 International applicants welcome 🕐 Initial contract for one year with possibility of extension 💰 Salary will be based on experience and qualifications (benefits will be included) Please spread the word!

Blake Richards avatar
Blake Richards@tyrell_turing·Oct 30

2/4) Position 1: 🧠 Postdoctoral Fellow – Cognitive Assessment and Psychometrics Focus: Adapting cognitive assessments into engaging video game mechanics to measure working memory, response inhibition, cognitive flexibility, time perception, etc. 🔗 mcgill.wd3.myworkdayjobs.com…

Randall Balestriero avatar
⟳ Blake Richards retweeted
Randall Balestriero@randall_balestr·Oct 30

The Platonic Representation Hypothesis is now a Theorem ◼️ Train JEPAs separately, on any modalities: they provably learn the same representation up to one rotation. The result? A 109M JEPA-Text model that aligns with vision models better than 72B LLMs. w/ @ylecun 🧵

1:45
Lacuna avatar
⟳ Blake Richards retweeted
Lacuna@Lacuna_Research·Oct 30

SCAMPER is quite cool. Made a little demo: drag the slider and watch the virtual rat go from walk to trot to gallop, gaits it never saw in its training data. By Aidan Sirbu (@aidan_sirbu), Talmo Pereira (@talmop), Blake Richards (@tyrell_turing) and collaborators. "Simulated Control via Augmented Motor Priors for Embodied Reinforcement learning": lacuna.tiptreesystems.com/x/…

0:15
Eric Elmoznino avatar
⟳ Blake Richards retweeted
Eric Elmoznino@EricElmoznino·Oct 29

Couldn't be happier to announce my new paper on defining compositional structure — and generating it! I see this project as the culmination of my PhD, as it unifies all my interests: compositionality, compression, structure, and open-endedness. More below! arxiv.org/abs/2610.20922

Post attachment
Doris Tsao avatar
⟳ Blake Richards retweeted
Doris Tsao@doristsao·Oct 17

1/ We discovered something totally unexpected about the neural structure of consciousness that overturns decades of thinking on its biological basis and brings us significantly closer to the recipe for generating conscious awareness. New preprint: biorxiv.org/content/10.64898… (led by @HesseJanis, with @FrankLanfranchi, @_conor_mcgrory, @Yuelin_Shi, and @EngelTatiana)