Learn by building
These tutorials teach you how BuddhistAI works by doing it. Each one is short. Do them in order, or pick the one you need.
| # | Tutorial | You will | You need | Cost |
|---|---|---|---|---|
| 1 | Run the scorer on your computer | Score stored results and see how a score is made | Python 3, git | Free |
| 2 | Evaluate your first model on a Buddhist task | Run an AI model on a real benchmark and score it | Tutorial 1, an OpenRouter API key | About $0.01 to $1 |
| 3 | Write your first skill | Write a SKILL.md, try it, share it, and have it measured | A text editor, a GitHub account | Free (or a few cents to test) |
| 4 | Add a benchmark | Turn material your team already checks by hand into a test set | Real examples with correct answers | Free |
| 5 | Improve your skill with GEPA | Let an optimizer rewrite your skill to raise its score | Tutorials 1 and 3, an Anthropic API key | A few dollars; always set a limit |
The tutorials use the Tibetan table-of-contents task, because it has a benchmark and a scorer today. The same steps work for every task with a benchmark.
Words you will meet
A task is one job, such as OCR or translation review. A benchmark is test material with correct answers for one task and language. A scorer compares answers with the correct ones and gives a number. A skill is a set of written instructions (and sometimes scripts) that make a model better at a task. See the Glossary for all terms.
Stuck?
Ask in GitHub Discussions. Found a mistake in a tutorial? Click “Edit this page” and fix it, or say so in Discussions. The FAQ answers common questions.

BuddhistAI Knowledge base