Welcome to the Comet Docs!

Data science and machine learning teams use Comet’s ML platform to track, compare, explain, and optimize their models across the complete ML lifecycle – from managing experiments to monitoring models in production.

Get Started for Free Watch Demo

Get started with two lines of code¶

# For Comet to start tracking a training run,
# just add these two lines at the top of
# your training script:
import comet_ml

experiment = comet_ml.Experiment(
    api_key="<Your API Key>",
    project_name="<Your Project Name>"
)

# Metrics from this training run will now be
# available in the Comet UI

Build Better Models Faster

Comet’s machine learning platform tracks the lifecycle of a model in one user interface, so you can build, collaborate, and iterate faster. For enterprise, bring Comet to what you build; we treat virtual private cloud (VPC) and on-premises environments as first-class citizens.

Try Comet free

Discover Comet

Track models and training runs

Start managing experiments using the tools, libraries, and frameworks you use today.

Tracking a training run Integrations with machine learning frameworks Training models in notebooks

Custom visualizations for faster iteration

Iterate, debug, and evaluate models faster with custom visualizations you can build yourself or choose from our library of templates.

Using Comet's built-in visualizations Creating custom visualizations Comparing training runs

Reproduce experiments

Create dataset versions and track hyperparameters for faster and easier reproducibility and collaboration.

Getting started with artifacts Creating a dataset artifact Using remote and local artifacts

Model Registry

Save model versions and deploy registered models using your computing environment.

Tracking model versions Registering an experiment model Integrating with CI/CD pipelines

New at doc

LLMOps Tools

Comet's LLMOps suite of tools allows you to explore LLMs through our plaground as well as track and visualize LLM chains.

Configure webhooks

Configure webhooks when a model is deployed to development, staging or production in order to trigger a CI/CD pipeline.

Kubeflow

The Kubeflow integration allows you to track the status of the DAG and individual steps straight from the Comet UI.