#machine-learning
11 posts
-
Neural Networks, Transformers, and How LLMs Learn
The machinery the deep dive skipped: how a neural network and the Transformer work, what loss really means, and how reinforcement learning trains a model.
machine-learningllmsneural-networksfundamentals -
A Deep Dive into How LLMs Are Built
How a model like ChatGPT is built stage by stage — pretraining, fine-tuning, reinforcement learning — and why it hallucinates and needs tokens to think.
machine-learningllmsfundamentals -
Intro to Large Language Models
A plain-English tour of how LLMs work: what they physically are, how they're trained, the OS-like ecosystem around them, and their security issues.
machine-learningllmsfundamentals -
Vectors and Matrices: The Language Neural Networks Speak
A ground-up introduction to vectors, matrices, dot products, and matrix multiplication — the operations every neural network is built from.
mathematicsmachine-learningfundamentals -
Derivatives and Gradients: Teaching Machines to Improve
Derivatives, the chain rule, partial derivatives, the gradient, and gradient descent — the calculus that drives every step of neural network learning.
mathematicsmachine-learningfundamentals -
Probability and the Gaussian: How Neural Networks Express Uncertainty
Probability distributions, the Gaussian, softmax, and cross-entropy loss — the tools neural networks use to express uncertainty and produce predictions.
mathematicsmachine-learningfundamentals -
A Neural Network from Scratch: Perceptrons, Layers, and Forward Pass
Build a neural network layer by layer — the perceptron, activation functions (ReLU, sigmoid, tanh), and a complete worked forward pass.
mathematicsmachine-learningfundamentals -
Backpropagation: How Neural Networks Learn from Mistakes
A complete walkthrough of backpropagation — the chain rule applied to computation graphs. Includes a worked numerical example through a 2-layer network.
mathematicsmachine-learningfundamentals -
Embeddings and Similarity: Turning Words into Vectors
How neural networks represent words as dense vectors, why dot products measure similarity, and how cosine similarity finds related concepts.
mathematicsmachine-learningfundamentals -
The Attention Mechanism: How Transformers Focus
A detailed walkthrough of scaled dot-product attention — Query, Key, and Value matrices, the softmax operation, and a complete numerical example.
mathematicsmachine-learningfundamentals -
The Transformer and GPT: Putting It All Together
Multi-head attention, positional encoding, layer normalisation, and the feed-forward sublayer. A complete step-by-step forward pass through GPT.
mathematicsmachine-learningfundamentals