23 posts found
A model does not look anything up and does not decide what is true. It estimates which token comes next. Almost everythi...
Models do not see characters or words. They see tokens — and once you know how text becomes tokens, several odd behaviou...
Turning temperature down does not make a model more accurate. It makes it more repeatable — and confusing the two is how...
Teams reach for fine-tuning when they need context, and for prompting when they need behaviour. Knowing which problem ea...
Bigger context windows did not give models memory. They gave you a larger envelope to fill on every single request — and...
Hallucination is not a glitch that a better model will one day remove. It is what generation does when it has nothing to...
A system prompt sets the rules for everything that follows. Written well it is the cheapest quality improvement availabl...
If your code needs a value from a model, ask for JSON and validate it. Regexing an answer out of a paragraph is a bug wa...
Most production traffic does not need the largest model available. Routing by task rather than defaulting to the top of...
Strip away the branding and an agent is a small loop: call the model, run the tool it asked for, feed the result back, r...
A tool definition is documentation with a schema attached. Write it for a capable colleague who has never seen your syst...
An agent that handles five steps beautifully can fall apart at twenty. The reason is rarely reasoning — it is that the c...