1 weeks ago
Agents and Tool Use
How an LLM goes from a brain in a jar to an agent that acts: tool use and function calling (the model asks, your code runs it), th…
All AI/LLM Security articles
AI/LLM Security → All AI/LLM Security articles
1 weeks ago
RAG: Open-Book LLMs
Part 17 of the AI/LLM mastery series — the biggest practical fix for hallucination and stale knowledge. Retrieval-Augmented Genera…
1 weeks ago
Prompt Engineering
Part 16 of the AI/LLM mastery series — the cheapest, highest-leverage skill: no code, no GPU, just words. Why prompting works (it …
1 weeks ago
Why LLMs Make Things Up
Part 15 of the AI/LLM mastery series — the limit no setting can fix. What hallucination really is, why it happens (the model predi…
1 weeks ago
How an LLM Picks Words
Part 14 of the AI/LLM mastery series — decoding, the step where the model turns its probability distribution into an actual word. …
1 weeks ago
Alignment & RLHF
Part 13 of the AI/LLM mastery series — how a model learns not just to answer, but to answer the way humans prefer. RLHF explained:…
1 weeks ago
Fine-Tuning
Part 12 of the AI/LLM mastery series — turning a knowledgeable base model into a helpful assistant by changing its behaviour, not …
1 weeks ago
The Data Pipeline
Part 11 of the AI/LLM mastery series — the unglamorous machinery that decides model quality. How a raw web scrape (mostly junk) be…
1 weeks ago
Scaling Laws
Part 10 of the AI/LLM mastery series — the maths of "bigger, more data, more compute". The three levers (parameters, tok…
1 weeks ago
Pretraining
Part 9 of the AI/LLM mastery series — how a randomly-initialised GPT becomes one that knows the world. Pretraining: the trillions …