Chapter 1Free
Your first model, measured.
Run a small model on your laptop, call it over HTTP, and measure its speed.
About 2 h 45 min
What you will do
- 01Guess firstBefore a new idea, write down what you think. It is never graded.
- 02Learn4 short lessons, one small step at a time.
- 03Run itFill your own speed table on your laptop, then break it 3 ways.
- 04Check10 questions on a recorded log. Pass with 8 right, including time to first token and tokens per second.
In this chapter
- What inference is, and your first modelAfter this you can run Qwen3-0.6B on your laptop with Ollama, and get an answer over HTTP from curl, JavaScript and Python.About 30 min
- Measure its speedAfter this you can say how fast your model is, with numbers you trust.About 30 min
- A peek at thinking modeAfter this you can turn thinking on and off on purpose, read both fields in a stream, and time the first answer word.About 25 min
- Where this fits in real jobsAfter this you can say which jobs use TTFT and tokens/s, and check one tool against a speed budget.About 20 min
- Practice: your speed tableFill a table of TTFT and tokens/s for 3 prompt lengths and 2 model sizes, thinking off.About 25 min
- Break itChange one thing, guess first, look, and explain.About 25 min
- Chapter check10 questions on one recorded log. Pass with 8 right, including time to first token and tokens per second.About 10 min
Stuck? Known errors for this chapter
Next chapter: Chapter 2, What just happened: tokens, the loop, the two phases, and "will it fit?"