MLOps backend for experiment/data management, pipelines, orchestration, scheduling & serving
Top 70.8% on sourcepulse
ClearML Server provides a self-hosted backend for ClearML, an MLOps platform for streamlining AI workloads. It targets ML engineers and researchers needing to manage experiments, data, pipelines, and serving, offering a centralized solution for collaboration and workflow automation.
How It Works
The ClearML Server comprises a web application for UI-based experiment management, a RESTful API for logging and querying experiment data, and a file server for storing artifacts like images and models. It supports single-domain or sub-domain configurations, allowing flexible deployment via Docker, cloud AMIs, or Kubernetes. The system leverages MongoDB and Elasticsearch for data storage.
Quick Start & Requirements
clearml-init
or manually edit ~/clearml.conf
to point the client to the server's api_server
, web_server
, and files_server
.Highlighted Details
Maintenance & Community
clearml
) and GitHub Issues for bug reports/feature requests.clearml@allegro.ai
Licensing & Compatibility
Limitations & Caveats
The ClearML Server relies on MongoDB and Elasticsearch, whose own licensing changes (SSPL, Elastic License) may impact downstream compatibility and commercial use. ClearML-Agent Services configuration is not supported on Windows installations.
1 week ago
1 day