ClearML Documentation

ClearML (formerly Trains) is a complete, open source ML / DL experimentation and MLOps solution. Data scientists and engineers can focus on data and training, while ClearmML eliminates the time consuming and error-prone tasks associated with development, version tracking, and the full ML lifecycle for automation and scaling.

ClearML automatically tracks everything you need to document your work, visualize results, reproduce, tune, and compare experiments, and implement automated workflows such as hyperparameter optimization and other pipelines. It provides a robust set of tools to improve experimentation. MPOps in ClearML is fire-and-forget: go from local development to distributed clusters in different locations without cumbersome version management and repeated hand-offs from engineers to DevOps.

The ClearML Python Package, ClearML Hosted Service (or your own self-hosted ClearML Server), and the MLOps ClearML Agent are together a unified solution. They empower you to work programmatically, use the one-page, intuitive Web UI, and perform MLOps, executing experiments and other workflow on the GPU machines of your choice.


The ClearML solution

The ClearML solution provides you with the following:

  • ClearML Python Package clearml.

    • Automatically captures tracking (Git repository codebase, command-line parameters, hyperparameters, models and other artifacts, log, debug samples) and results (metrics and other data)

    • Supports:

      • Frameworks (TensorFlow/TensorBoard, PyTorch, Keras, Fastai, scikit-learn)

      • libraries (Pandas, Plotly, AutoKeras)

      • Visualization tools (Matplotlib, Seaborn)

      • Storage (file systems, S3, Google Cloud Storage, and Azure Storage)

    • Improves development with classes for experiments, explicit reporting, workflow automation, optimization (Optuna, HpBandster, random, grid, and custom search strategies), models, and storage.

  • ClearML Server clearml-server - Use the hosted service or deploy your own self-hosted ClearML Server.
    • Stores data from scripts, created by automated workflows, and edited in the ClearML Web UI.

    • Uses the stored data tracking, reproducibility, cloning, tuning, automation, and execution on target machines.

    • Allows you to work in the ClearML Web UI and programmatically.

  • ClearML Agent clearml-agent
    • Recreates the state of code from the original machine to a local or remote machine

    • Executes the code on a local or remote machine, while monitoring it and logging to the ClearML Server.

    • Eliminates repeated handoffs to DevOps. Once ClearML Agent is running on a machine, keep reusing it.

    • Permits anyone to run ClearML Agent, which everyone uses and reuses it for orchestration.

    • Supports configuration options and multiple use cases.

How to use this documentation

  • First, get started.

  • Next, learn about ClearML:

    • Concepts and Fundamentals - ClearML essential concepts.

    • Examples - Example scripts which are ready run and pre-loaded in ClearML.

    • Tutorials - Step-by-step instructions for explicitly reporting, tuning experiments, and creating tracking leaderboards.

    • Web UI - The ClearML Web UI user guide section. Learn how the experiment manager features and functions.

    • Integrations - ClearML integrations with ML / DL libraries and development tools.

  • For a deep-dive into ClearML see the references:

If you have questions, want to make feature requests, report bugs, and find additional resources, see our FAQ and Community.


ClearML supports integrations with popular libraries and development tools, including:

Next step