Skip to the content
Know your tech before you buy it. Makers: get listed from $49/mo

Guild AI

guildai.org

  • Windows
  • macOS
  • Linux
  • Web
7.1 Club score Visit guildai.org

What Guild AI is

Guild AI is an open-source experiment tracking tool for machine learning projects, available across Linux, macOS, web, and Windows. It can run unmodified training scripts and save each run as a distinct experiment, recording flags, results, metadata, and source-code snapshots. Built-in batch operations support grid search, random search, and sequential Bayesian optimization; random optimization creates 20 trials by default, with the maximum adjustable through an option. Guild View provides a browser interface for runs, files, metadata, and output, while Guild Compare presents runs in a spreadsheet-style view and exports data as CSV. TensorBoard integration includes automatic HParam summaries, with TensorFlow and Keras integrations described in the documentation. Guild supports remote execution and synchronization through SSH, EC2, S3, Azure VM, Azure Blob Storage, and GitHub Gists, with capabilities varying by type. It can operate without a database or external system. Installation methods include pip, conda, Docker, and source code. Artifact tracking, dataset versioning, API and SDK access, and self-hosted deployment are listed.

Who it suits

Guild AI suits machine learning teams that need experiment records, parameter searches, run comparison, and remote synchronization. It also fits users wanting open-source, self-hosted tooling without a database.

Good to know, and what to watch

  • Runs unmodified training scripts
  • Supports grid, random, and Bayesian search
  • Browser and spreadsheet run views
  • TensorBoard, TensorFlow, and Keras integration
  • No database or external system required
  • GitHub Gist runs are public by default
  • Random optimizer defaults to 20 trials
  • Remote capabilities vary by system
  • No model registry

The verdict

Guild AI offers open-source experiment tracking with search, comparison, web views, and remote operation. Review the public-by-default Gist behavior and system-specific remote limits before deployment.

What the maker publishes

Experiment records
It records run details such as flags, results, metadata, and source code snapshots. source
Experiment tracking
Guild can run unmodified training scripts and capture each run as a unique experiment. source
Hyperparameter search
Built-in batch operations include grid search, random search, and sequential Bayesian optimization. source
Installation
The getting-started guide says to install Guild AI with pip install guildai. source
Install methods
The installation guide documents pip, conda, Docker, and source-code installation. source
Integrations
The documentation describes TensorBoard support and integrations involving TensorFlow and Keras. source
Local operation
The getting-started guide says Guild does not require databases or external systems. source
Notable limit
The documentation says the random optimizer generates 20 trials by default, adjustable with the max-trials option. source
Optimization
Guild supports grid search, random search, and Bayesian optimization. source
Purpose
Guild AI is an experiment tracking tool for machine learning projects. source
Purpose
Guild AI is an open-source toolkit for automating and optimizing machine learning experiments. source
Remote execution
Guild includes commands for running operations remotely and synchronizing remote runs. source
Remote systems
Supported remote types include SSH, EC2, S3, Azure VM, Azure Blob Storage, and GitHub Gists, with different capabilities by type. source
Result viewing
Guild View is a browser-based application for browsing runs, viewing metadata and files, and comparing results. source
Run comparison
Guild Compare lets users review run results and compare experiment metrics. source
Run comparisons
Guild Compare displays runs in a spreadsheet format and can export run data as CSV. source
Run data
Guild stores run scalars in TensorBoard summary file format and does not show summary files written outside the run directory. source
Security limit
Runs published to GitHub Gists are public by default, and users must change the gist visibility themselves to secure them. source
Support
The Guild AI community site directs users to its Troubleshooting category for help with issues. source
Support
The Guild AI forum says its support systems are moving to GitHub and that the forum will be read-only. source
Supported systems
The system requirements page lists Linux, macOS 10.14 and 10.15, and Windows 10 for precompiled packages. source
TensorBoard
Guild integrates with TensorBoard and automatically generates HParam summaries from Guild runs. source
Web interface
Guild View is a web application for viewing runs, comparing results, and viewing run files and output. source

Checked against the maker's pages on 4 Oct 2026.