ROGUEQUANT_

// QUANTITATIVE RESEARCH · AI · MARKETS

Signal from noise.

Notes on AI, quantitative trading and financial technology — written from the trenches, not the sidelines.

FEATURED

2026.09.05 // LLM FOR DUMMIES // 8 MIN

LLM for Dummies #5 — From Autocomplete to Assistant: The Persona Is Learned

Pretraining buys a brilliant alien autocomplete with no manners. Part 5 is about the surprisingly small dose of extra training that turns it into something that answers questions, follows instructions and says no — instruction tuning, RLHF, the chat template, and why 'the assistant' is a character the model was taught to play.

2026.08.05 // LLM FOR DUMMIES // 7 MIN

LLM for Dummies #4 — Training: How Guessing the Next Word Buys You a Mind

We have the machine. Now, why are its trillion numbers right? Because of one absurdly simple game played ten trillion times: hide the next token, guess it, measure the surprise, adjust. This part explains why that game forces a model to learn facts, grammar and a bit of reasoning — and what a loss curve really measures.

LATEST ENTRIES

All entries