Using Neptune

Getting started

 In a hurry? Try the quickstart

 Missed the setup part? Go back to installation and setup

 Looking for a fully guided setup and orientation experience? Take the Neptune tutorial

---

Assumptions

For the topics in this section, we generally assume that you have:

1. Registered with Neptune. neptune.ai/register

2. Installed the latest version of the Neptune client library: pip install -U neptune (or see Installing Neptune)

   Note: We've just released a major version of the library. If you already have neptune-client installed, uninstall that before installing neptune. For details, see Upgrading.

3. Saved your API token and project name as environment variables.

   If you have not done this, you can pass your info when initializing Neptune, if needed:

   run = neptune.init_run(
       project="your-workspace-name/your-project-name",  # (1)!
       api_token="YourNeptuneApiToken",  # (2)!
   )
   

   1. The full project name. For example, "ml-team/classification". To copy it, navigate to the project settings → Properties.
   2. To copy your token, click your avatar in the Neptune app and select Get your API token. When you're done testing, save your API token as an environment variable instead of putting it here in the code!

For more, see Best practices.

Docs by category

RunsModelsProjectsAPIAppDevOpsImages and plots

• Creating a run
• Essential logging methods
• What you can log and display
• Tracking artifacts
• Logging to an existing run (Resuming a run)
• Re-running a failed training
• Logging from multiple places
• Setting a custom run ID
• Run comparison overview
• Stopping or aborting a run remotely

• Model registry overview
• Registering a model
• Creating a model version
• Managing the model version stage
• Querying metadata from the model registry

• About workspaces and projects
• Logging project-level metadata
• Tracking artifacts
• Collaboration in Neptune
• Trashing and deleting data
• Creating and deleting projects via API

• Python API overview
• Adding Neptune to your code
• What you can log and display
• Logging source code
• Creating and deleting projects via API
• Querying metadata from Neptune
• Using Neptune in different modes
• Field types reference
• Neptune CLI

• Neptune app overview
• Collaboration in Neptune
• Runs table overview
• Exploring a run
• Comparing runs
• Showing multiple graphs in one chart
• Creating a custom dashboard
• Downloading metadata from Neptune

• Setting up a service account
• Using Neptune with Docker
• Using Neptune with GitHub Actions
• Working with Amazon SageMaker
• Integrations ≫ Working with Kedro

• Logging images
• Logging charts and plots
• Logging interactive visualizations
• Comparing images between runs
• Viewing series of images in the gallery

Examples on GitHub

In the neptune-ai/examples repo on GitHub, you can browse our full library of example scripts, notebooks, and Neptune projects.

Visit examples repo 

Notebooks in Colab

All of our Jupyter Notebook examples are listed below.

TutorialsIntegrationsVisualization and plottingNeptune basics

• Tracking and organizing model-training runs
• Monitoring model training live
• Versioning datasets in runs
• Comparing datasets between runs
• Sharing dataset versions on project-level
• Tracking and visualizing cross-validation results
• Re-running a failed training
• Tracking HPO jobs

Notebooks used for our integration guides:

• Catalyst
• fastai
• 🤗 Transformers
• Keras
• LightGBM
• Lightning
• Optuna
• Prophet
• PyTorch
• Sacred
• scikit-learn
• skorch
• TensorFlow
• XGBoost

Notebooks that show how to log charts and images generated with common visualization libraries.

• Altair
• Bokeh
• Matplotlib
• pandas
• Plotly

• Quickstart (Hello Neptune)
• Using Neptune in Colab

Not what you're looking for?

Want to follow more comprehensive, practical examples? Tutorials

Wondering how to use Neptune with a specific library or tool? Integrations