YAML
Last Updated: July 29, 2026 | By Mihail Sebastian | AI Dictionary
A data format that uses indentation instead of brackets, designed for humans to read and edit – the standard for ML training configs and CI pipelines.
What is YAML?
YAML (YAML Ain’t Markup Language) is a data format that uses indentation instead of brackets to represent structure, designed to be easy for people to read and edit. It is the standard format for configuration files.
Most ML infrastructure runs on YAML somewhere: training configs, Kubernetes manifests, Docker Compose files, and CI workflows in GitHub Actions or GitLab are all written in it.
How YAML Works
A YAML document expresses nesting through indentation, lists with leading dashes, and key-value pairs with a colon. Comments start with #, which is one reason it suits files that humans maintain.
learning_rate: 0.001
batch_size: 32
epochs: 10
Three lines define a training run. Change a value, commit the file, and the experiment is reproducible: the config records exactly which hyperparameters produced which model.
YAML vs JSON
YAML is built for humans editing configuration files; JSON is built for programs exchanging data. YAML reads cleanly and allows comments, but its reliance on indentation makes it easy to break with a stray space. JSON’s stricter syntax is harder to hand-edit and simpler to parse.
The two are related. YAML 1.2 is designed as a superset of JSON, so any valid JSON document is also valid YAML.
| Criterion | YAML | JSON |
|---|---|---|
| Primary use | Configuration files edited by hand | Data exchange between programs (APIs, logs, annotations) |
| Syntax | Indentation, minimal punctuation | Braces, brackets, quoted keys |
| Comments | Supported with # | Not supported |
Example of YAML
A team training an image classifier keeps a config.yaml in the repository next to the code: dataset path, model architecture, learning rate, augmentation settings. The training script loads it at startup, so nobody edits source code to change an experiment.
The same repository’s CI file is also YAML. A GitHub Actions workflow lists the steps (install dependencies, run the tests, train on a sample) that execute on every commit. The pipeline itself lives under version control, the same as the code.
Related AI terms: JSON · Dockerfile · Hyperparameter · Pipeline
Did you like the YAML gist?
Learn about 250+ need-to-know artificial intelligence terms in the AI Dictionary.
Mihail Sebastian — Writes about AI governance, regulation, and the technology behind them. Placeholder bio — replace with a real credential line. About