Cognitive Load Theory gets plenty of attention on EduTwitter (which is great!), but Ausubel’s Assimilation Theory rarely gets mentioned. I think it’s just as important. 🧵

What a BRILLIANT read! And a pleasure to make a summary of Zach Groshell's fantastic new book on explaining. So many useful ideas for the classroom and a must read for anyone wanting to find out about explicit instruction or refine their practice. Pdfs here: tinyurl.com/ykh5rrvn

Hey Grok 3 can you analyze Elon Musk’s gorklon rust account from the point of view of accuracy for someone interested in political science and media sources? Answer: x.com/i/grok/share/8… Grok holds nothing back.

Andrej Karpathy calls large language models the new computing paradigm, called LLM OS: CPU -> LLM bytes -> tokens RAM -> context window

LLM Post-Training: A Deep Dive into Reasoning Large Language Models

This is a fascinating study in Science!

Time Complexity of Machine Learning Algorithms:

About that claim he made about “$8 million in funding for research to turn mice transgender” which had the entire Republican side of the room bursting with laughter: it turns out the studies were on TRANSGENIC mice. Not transgender mice. Trump is just a transphobe who can’t read.

Popper on definitions youtu.be/oUbJqb08Nfs

Congratulations to Richard Sutton and Andrew Barto on receiving the Turing Award in recognition of their significant contributions to ML. I also stand with them: Releasing models to the public without the right technical and societal safeguards is irresponsible.

Been trying to get X under some sort of prompting direction towards what I want to find out about, but it is becoming increasingly wild and out of kilter, so have also joined bluesky to moderate a little

This is how 𝗔𝗜 𝗔𝗴𝗲𝗻𝘁 𝗠𝗲𝗺𝗼𝗿𝘆 works. In general, the memory for an agent is something that we provide via context in the prompt passed to LLM that helps the agent to better plan and react given past interactions or data not immediately available. It is useful to

# A new type of information theory this paper is not super well-known but has changed my opinion of how deep learning works more than almost anything else it says that we should measure the amount of information available in some representation based on how *extractable* it is,

the best researchers from Meta, Yale, Stanford, Google DeepMind, and Microsoft laid out all we know about Agents in a 264-page paper [book], here are some of their key findings:

RL is hard to explain to people because it doesn't really make any sense without internalizing strong intuitions about everything going on here

Info theory offers powerful measures for capturing complexity & interaction among elements of a complex system, like the brain! 🧠 Here's our new unified reference for key info-theoretic time series measures ft. 📊 visuals, ➗equations, & 💬descriptions: arxiv.org/abs/2505.13080

founders waking up to realise their startup is now a bullet point in Google’s keynote.

I’m stoked to share our new paper: “Harnessing the Universal Geometry of Embeddings” with jack morris, Collin Zhang, and Vitaly Shmatikov. We present the first method to translate text embeddings across different spaces without any paired data or encoders. Here's why we're excited: 🧵👇🏾

Prompt Theory (Made with Veo 3) What if AI-generated characters refused to believe they were AI-generated?