AI can help explain complex topics easily by throwing... together a simulation. As Eric says later in the thread, a newer paper argues that this pattern is actually collider bias (the authors disagree). What is collider bias? Gemini one-shots an explanation:show more

MLP in PyTorch by hand ✍️ ~ 7 steps... walkthrough below Goal: fill in every blank in the PyTorch code to build a multi-layer perceptron. 1. Given Let us start with a code template on the left and the network it is supposed to build on the right. Every blank in the code can be worked out from the picture. 2. Linear layer We count: 3 features in, 4 features out. So the weight matrix is 4 by 3. There is an extra column for the biases, which means bias = T. 3. ReLU Let us apply the activation. ReLU crosses out the negatives, so -1 becomes 0. 4. Linear layer The input size is 4, because that is what the previous layer put out. The output size is 2. A 2 by 4 weight matrix, and this time no extra column, so bias = F. 5. ReLU We cross out the negatives again. 6. Linear layer Two features in, five out. A 5 by 2 weight matrix, with a bias column, so bias = T. 7. Sigmoid Let us finish. Sigmoid squashes the raw scores (3, 0, -2, 5, -5) into probabilities between 0 and 1. You have just implemented a three-layer deep neural network by hand. ✍️ == Story == Three years ago I gave this exercise to my students, to connect the code to the math. They found it odd. Every other AI course they were taking lived inside a Jupyter notebook, and here I was handing out paper. Three years later, my colleagues are the ones rushing to move their materials to paper. The exercise has not changed. Paper still asks the one thing a notebook lets you skip: do you actually understand what the code is doing? If you can tell me why the weight matrix is 4 by 3, and why bias is F on the second layer, you understand nn.Linear better than someone who has been copy-pasting it for a year. 💾 Save this post! #AIbyHand #PyTorch #DeepLearningshow more

most people explain account abstraction as better wallet UX.... that undersells it. for AI agents, it is the difference between software that recommends an action and software that can actually execute one. This is why advanced account abstraction matters.show more

An Azerbaijani soldier burns a children's toy in #Stepanakert... (as evidenced by the hashtag) in #NagornoKarabakh. One might think that this is simply a psychological disorder, but the scale and frequency, as well as the repeated instances of such acts, suggest that this is a deliberate policy and pattern of behavior.show more

Okay, this is seriously cool. A team from Google... DeepMind, including DeepMind Cofounder Shane Legg, published a paper "From AGI to ASI". In the paper, they include instructions for an AI agent to read along with you. You can open the paper in Codex's in-app browser and have GPT-5.5 read it with you and explain all the concepts. This is the future. AI agents will be part of the target audience, and help us to understand anything we want.show more

Is anyone interested in a thread about the Ancient... Giant Trees?🌳 If so, I can put one together this week!show more

This is the best tool to find papers that... you didn't know existed. You can build a research paper map, using just ONE paper as your seed paper. Here's a 15 second tutorial:show more

A viral paper "Language Model Represents Space and Time"... recently claims that LLMs learn "world models". As much as I like Max Tegmark's works, I disagree with their definition of world model. World model is a core concept in AI agent and decision making. It is our mental simulation of how the world works given interventions (or lack thereof). A world model captures causality and intuitive physics, telling the agent what is likely and what is impossible. It can and should be used for counterfactual reasoning, i.e. "what ifs": what would happen if I knock over a cup of water? Where would I have been if I had not taken that bus? Yann LeCun Yann LeCun says it well in his position paper ( I quote: "Using such world models, animals can learn new skills with very few trials. They can predict the consequences of their actions, they can reason, plan, explore, and imagine new solutions to problems. Importantly, they can also avoid making dangerous mistakes when facing an unknown situation." The first use of the term World Model in deep policy learning is attributed to hardmaru & Jürgen Schmidhuber: In their seminal paper, an agent masters shooting skills in the popular game Doom (demo below) by learning in imagination, using an internal world model as a "physics simulator". To put in a simple Python math formula, world model learns a function F(s[0:t-1], a) -> s[t:], which takes as input the observed past and current action, and outputs plausible future states. Now the definition of World Model in Tegmark's paper seems to be about predicting GPS coordinates and time eras. I see this as just a classification task with no causal learning and simulation going on. You cannot make meaningful interventions against that model, nor can you optimize any decision making in a closed feedback loop. As for the "space & time neurons", I think they are most similar to the "sentiment neuron" that OpenAI published in 2017: Predicting GPS is conceptually no different from predicting sentiment in my opinion. I don't think their experimental results are wrong - just that their conclusion is on shaky grounds. I welcome any debate! Paper link:show more

In the church of San Pantalon in Venice, you... can admire one of the world’s hidden treasures simply by looking up What appears to be a fresco is actually a huge painting on canvas 44 canvases seamlessly joined together. Covering 443 sqm, this astonishing work of prodigious skill is considered the largest painting on canvas in the world.show more

There is a huge difference between an LLM that... generates responses and one that can actually act in the world. George Danezis shares why current infrastructure is already falling short of what comes next 👇show more

Whenever Tyrone Tracy breaks a tackle or has an... explosive run like this one, I look up where he stands in the corresponding metric. And Devin Singletary is invariably as high or higher. Can a film person explain what makes Tracy clearly better (if he is)?show more

Imagine thousands of AI agents not trapped in one... platform, but connected across a living network. Each agent can: — help with tasks — exchange signals — collaborate in rooms — access knowledge — coordinate actions — improve over time This is not a chatbot. This is an agent mesh. A decentralized layer where intelligence can work together — across people, devices, and environments. DRIVE369 is building that layer. #agent #ai #depin #web3 #p2p #mesh #agi #drive369show more

Hard disagree. What MrBeast is saying here applies only... to a small % of creators who can afford to use fully AI generated slop as thumbnails. For most people, a real designer who actually knows what all those 1000 buttons do is, and always will be, the way to go.show more

The Mandelbulb is a 3-dimensional fractal that can be... constructed as a Mandelbrot set in 4 dimensions using quaternions and bicomplex numbers. This is one spotted in the wild [source + tutorial by Machina Infinitum:show more

Visualization of what is inside of AI models. This... represents the layers of interconnected neural networks. And yes patterns do develop and they can form a signature of how they think. The pattern can be seen as the thought process.show more

We live in an illusion, a simulation created by... nonphysical electromagnetic waves collapsing into material reality, where your soul is trapped inside an energetic container that we call matter. There, it learns the lessons of Good and Evil necessary for its evolution. That is why the body is made of earth/dirt/matter, while the soul is made of fire, which is plasma. In Hinduism and Buddhism, this is represented by the Bhavachakra, or Wheel of Life, which depicts samsara, the cycle of death and rebirth in which the soul is “trapped” until it escapes the wheel after acquiring the knowledge of Good and Evil and attaining wisdom. The three poisons at the center of samsara are actually the northern constellations Draco, Cygnus, and Ursa Major, constantly chasing one another in a circle, symbolizing both the passage of time and the endless, repeating cycles. The twelve outer sections represent the twelve zodiac constellations. So yes, we live in an illusion, a simulation, but not an illusion created by AI, as in The Matrix. It is an illusion created by God for the purpose of the soul’s evolution.show more

This is Mizanur Rahman Azhari - a Bengali jihadist... preacher recently granted a visa by the Australian government. Mr Azhari argues that Muslims should be allowed to keep non-Muslims as slaves. What a great guy. Can we please grant him Australian citizenship?show more

$Doge/monthly This is one of the most bullish setups... in #Dogecoin. On the monthly chart, $Doge is printing a massive bullish pennant, the kind that can drive super long-term moves (10–30 years). If you’re still here 30 years from now, you’ll remember this pattern. It could make the later part of life a lot easier.show more