Practical explanations and project-based series on coding, AI, and connected devices.
Start here · 10 parts · Executed Python notebooks
Start with what fine-tuning means, then learn full tuning, adapters, LoRA, QLoRA, supervised instruction tuning, and preference tuning through ten clear articles, numerical examples, diagrams, and Python notebooks.
Project-based Python series
Eight project-based articles for developers who know basic Python. Build routing, retrieval, persistent human approval, retries, an optional local LLM, and a tested final application. Every part includes a downloadable code ZIP.
Read the LangGraph series Download the complete codeSame prompt, three sentence completions. Language models sample the next token from a probability distribution — they do not look up a stored answer. Here is why, and how to lock the output down when you need the same reply twice.
A shop-desk agent: look up an order, read the return policy, check today's date, and draft a customer email. Same loop, two tickets, two different replies.
Move from teaching models to an optional LoRA training script, then connect each notebook to its website article.
Compare five adaptation methods, inspect a confusion matrix, and verify a saved model reloads correctly.
Calculate the DPO objective and train a tiny response-choice policy against a frozen reference.
Pretrain a tiny language model, mask prompt tokens, and teach a simple instruction-to-label format.
Quantize a small matrix, measure reconstruction error, and train an update around a frozen quantized base.
Calculate a LoRA update by hand, train it, and prove that merging preserves predictions.
Learn the PEFT family by training a bottleneck adapter and inspecting bias-only updates.
Adapt a pretrained miniature router and inspect exactly which parameters changed.
Create a labeled workshop dataset, keep paraphrases together, and learn what a baseline reveals.
Understand weights, training, and useful applications with a one-weight numerical experiment.
Start with a small working program, learn what a workflow means, and identify the problems a graph will help us organize.
Build and run a StateGraph, then trace how nodes return changes to a shared dictionary-shaped state.
Give technical questions and human-only requests different paths, and keep a readable log of the nodes that ran.
Find relevant local notes, attach source references to a reply, and stop a failed revision loop after a deliberate limit.
Store a review checkpoint in SQLite, end the process, and continue the same ticket with an explicit approval or rejection.
Recover from a simulated transient lookup failure, show the nodes that complete, and prevent a repeated queue operation from creating duplicate rows.
Replace one drafting function with an optional Ollama adapter, validate its JSON response, and keep source checks and human review around generated text.
Run the assembled application across separate commands, inspect the local outbox, and evaluate a fixed set of tickets before publishing the repository.
Breaking down the complex concepts of neural networks into simple terms
A comprehensive guide to creating effective data visualizations
Exploring the latest trends and developments in AI technology
