clearml
ClearML - Auto-Magical CI/CD to streamline your AI workload. Experiment Management, Data Management, Pipeline, Orchestration, Scheduling & Serving in one MLOps/
What is clearml?
ClearML - Auto-Magical CI/CD to streamline your AI workload. Experiment Management, Data Management, Pipeline, Orchestration, Scheduling & Serving in one MLOps/LLMOps solution
Overview of clearml
What it is
ClearML is an open-source ML/DL development and production suite. It includes five main modules: Experiment Manager, MLOps/LLMOps, Data-Management, Model-Serving, and Reports. Additional components include an Orchestration Dashboard and Fractional GPU support. The suite is designed to track and control the process of training production-grade deep learning models, with minimal code integration.
Key features
- Automagical experiment tracking with just two lines of code added to existing scripts
- Full experiment setup log including source control, execution environment, hyper-parameters, and initial model weights
- Automatic capture of stdout, stderr, resource monitoring (CPU/GPU), model snapshots, artifacts, and Tensorboard scalars, metrics, histograms, images, audio, and video
- MLOps orchestration, automation, and pipelines for ML/DL/GenAI jobs on Kubernetes, cloud, or bare-metal
- Data management and version control on top of object storage (S3, GS, Azure, NAS)
- Model serving with GPU support backed by Nvidia-Triton and out-of-the-box model monitoring
- Hyper-parameter optimization using Bayesian algorithms and automation pipeline building
Use cases
ClearML is used for training production-grade deep learning models, boosting collaboration and visibility within teams, creating remote jobs from experiments with a single click, automating processes and building pipelines to collect experimentation logs, outputs, and data, and cataloging data on object storage for transparency.
Getting started
- Sign up for the free ClearML Hosted Service or set up your own server.2. Install the clearml Python package:pip install clearml3. Connect the SDK to the server by running:clearml-init4. Add two lines to your code:from clearml import Tasktask = Task.init(project_name='examples', task_name='hello world')
License and pricing
ClearML is licensed under the Apache License, Version 2.0. It is open-source and no pricing information is stated beyond that.
Based on the project's public README and repository metadata.
