Neptune is a metadata store for MLOps, built for teams that run a lot of experiments.

It's used for:

• Experiment tracking: Log, display, organize, and compare ML experiments in a single place.
• Model registry: Version, store, manage, and query trained models, and model building metadata.
• Monitoring ML runs live: Record and monitor model training, evaluation, or production runs live

In this repo, you'll find examples of using Neptune to log and retrieve your ML metadata.

You can run every example with zero setup as an "ANONYMOUS" Neptune user (no registration needed).

Note: This readme is best viewed in the GitHub Light theme.

	Docs	Neptune	GitHub	Colab
Quickstart

	Docs	Neptune	GitHub	Colab
Organize ML experiments
DDP training experiments
Re-run failed training
Use Neptune in HPO training job

	Docs	Neptune	GitHub	Colab
Log model building metadata

	Docs	Neptune	GitHub	Colab
Monitor model training runs live

	Docs	Neptune	GitHub	Colab
Version datasets in model training runs
Compare datasets between runs

	Docs	Neptune	GitHub	Colab
Resume run
Pass run object between files
Use Neptune in distributed computing
Use Neptune in parallel computing
Use Neptune in Pipelines
Log to multiple runs in one script
Create and delete projects

	Docs	Neptune	GitHub	Colab
Do GroupBy on runs
Do sorting

	Docs	Neptune	GitHub	Colab
Python
R

	Docs	Neptune	GitHub	Colab
Catalyst
fastai
Keras
lightGBM
Prophet
PyTorch
PyTorch Ignite
PyTorch Lightning
scikit-learn
skorch
🤗 Transformers
TensorFlow
XGBoost

	Docs	Neptune	GitHub	Colab
Keras Tuner
Optuna
Scikit-Optimize

	Docs	Neptune	GitHub	Colab
Altair
Bokeh
Dalex
HiPlot
HTML
Matplotlib
Pandas
Plotly

	Docs	Neptune	GitHub	Colab
Kedro

	Docs	Neptune	GitHub	Colab
MLflow
Sacred
TensorBoard

	Docs	Neptune	GitHub	Colab
Any IDE
Amazon SageMaker notebooks
Deepnote
Google Colab
Jupyter Notebook and JupyterLab

	Docs	Neptune	GitHub	Colab
AWS S3
Data Version Control (DVC)
Local filesystem

	Docs	Neptune	GitHub	Colab
GitHub Actions

	Docs	Neptune	GitHub
Using Neptune in training jobs with custom Docker containers
Using Neptune in training jobs with PyTorch Estimator