May 12 · 7 min read · The whole thing with AI models like ChatGPT is kind of weird when you think about it. You ask a question and it spits out an answer that sounds completely sure of itself — but nobody really knows why
Join discussion
May 11 · 6 min read · Part 87 of Forgotten Theories, a series re-testing old scientific claims with modern tools. Find the rest under the #forgotten-theories tag. Donald Hebb's 1949 book "The Organization of Behavior" is one of those texts that almost everyone in neurosc...
Join discussionMay 11 · 7 min read · Part 86 of Forgotten Theories, a series re-testing old scientific claims with modern tools. Find the rest under the #forgotten-theories tag. In 1943 Warren McCulloch, a neuropsychiatrist with a taste for logic, and Walter Pitts, a teenage runaway wh...
Join discussionMay 10 · 6 min read · The Inversion at ML's Core Most developers understand traditional programming intuitively: Input + Rules (code) → Output You write a function. It takes input. It applies your logic. It returns an out
Join discussionMay 3 · 21 min read · TLDR: Softmax converts a vector of raw scores (logits) into a valid probability distribution by exponentiating each value and dividing by the total. Subtracting the max before exponentiating prevents floating-point overflow. Temperature scaling contr...
Join discussionMay 3 · 30 min read · The goal of this blog post is to demonstrate so far the the neural network is a relatively simple algorithm to understand and show using mathematics and machine learning that it is a universal functio
TIyanuloluwa commentedMay 3 · 3 min read · Let's be honest: the current meta of AI development is just brute-forcing parameters until we run out of VRAM. It’s expensive, it’s inefficient, and it’s definitely not how a biological brain actually
Join discussionMay 3 · 21 min read · TLDR: The dot product multiplies corresponding elements of two vectors and sums the results. In machine learning it does three critical jobs: it scores semantic similarity between embeddings, computes every activation in a fully connected layer, and ...
Join discussion
