Neptune-HiPlot Integration

This integration lets you analyze multiple experiments in Neptune using HiPlot visualization. HiPlot is a lightweight interactive visualization tool published by the Facebook AI group.

parallel plot header

Parallel coordinates plot is a powerful tool that allows AI researchers to analyze correlations and patterns between experiment metrics, parameters and properties.

Parallel plots are especially useful when inspecting hyperparameter optimization jobs that usually consists of hundreds of experiments. Neptune allows you to very easily generate such plots in a Jupyter Notebook or Python script.


This feature is implemented as a part of neptune-contrib. Make sure that you have all dependencies installed:

  • neptune-client

  • neptune-contrib[viz]

  • hiplot

Use this command to install them:

pip install neptune-client neptune-contrib[viz] hiplot

Generate parallel coordinates plot


Make sure you have your project set: neptune.init('USERNAME/example-project')

import neptune

from neptunecontrib.viz.parallel_coordinates_plot import make_parallel_coordinates_plot


                               metrics= ['epoch_accuracy', 'epoch_loss', 'eval_accuracy', 'eval_loss'],
                               params = ['activation', 'batch_size', 'dense_units', 'dropout', 'learning_rate', 'optimizer'],
parallel plot overview

Customize visualization to your needs

Perform the following steps:

  1. Set axes order.

  2. Drop the unused axes.

  3. Apply coloring to the axis.

  4. Sort by clicking on the axis.

  5. Select range in the axis and slide.

parallel plot customization options

Inspect experiments lineage

Perform the following steps:

  1. Right-click on the axis name.

  2. Use options ‘Set as X axis’ and ‘Set as Y axis’ (in the menu XY group at the bottom).

When both are selected, you will see the lineage plot below the parallel coordinates plot.

experiments lineage

Check example notebooks in Neptune

  1. credit-default-prediction

  2. example-project

These notebooks are tracked in Neptune public projects. Feel free to play with the plots - they are interactive.

Learn more

Check integration documentation for more details.