Why Guild AI?
Guild AI is the easiest way to track experiments
Just run your script.
guild run train.py
With one command, you automatically record:
- Experiment inputs
- Experiment results
- Snapshot of source code
If you need to customize anything, Guild lets you add external configuration without changing your code.
It's superpowered
Guild AI is a feature-complete ML engineering toolkit.
- Run analysis, visualization, and diffing
- Pipeline automation
- Hyperparameter tuning with AutoML
- Scheduling
- Parallel processing
- Remote training, backup, and restore
100% open source
Guild AI is available under the Apache 2 open source license. You have full rights to use Guild source code — review it, change it, share it with others.
Guild AI is not affiliated with a platform vendor. There’s no bait-and-switch or hidden allegiance. Guild AI is an independent, open source company committed to empowering ML engineers with world-class tools.
No code change required
Don’t change your code just to track experiments. Guild adapts to your code with smart configuration. Avoid pointless changes to your code and stay independent.
No databases, exotic file systems, or agents
Guild is free of complicated requirements.
-
No databases — If your experiment tracking system requires that you install, configure, and maintain systems like MongoDB, think twice.
-
No exotic file systems — Storing an experiment should be straight forward. It shouldn’t require exotic file systems. If a tool requires anything beyond a standard mounted drive, think twice.
-
No agents — Experiments should just run, every time, without talking to a back-end service to capture results. If your experiment tracker talks to an agent, think twice.
Incredible community of ML engineers
The community of Guild AI users share a common vision.
-
Measure — Memory and intuition only take you so far. Measure results to systematically improve models and avoid costly regressions.
-
Automate — Manual steps are tedious and error-prone. Automate tasks to go faster and avoid costly mistakes.
-
Repeat — Engineers work with data scientists in joint-ownership of code. Anyone can run operations at any time to recreate and confirm results.
-
Improve — Work fast, measure, and iterate. This is the essence of Test Driven Development. It’s the key to effective software delivery.