A plugin for TensorBoard to customize and export Scalar plots

Home Page: https://pypi.org/project/tensorboard-plugin-customizable-plots/

License: Apache License 2.0

Python 0.37% CSS 0.31% JavaScript 99.28% HTML 0.04%

tensorboard tensorboard-visualizations tensorboardx

tensorboard_plugin_customizable_plots's Introduction

Customizable Plots - Tensorboard Plugin

This plugin is a simple Tensorboard dashboard to visualize, customize and export ready-to-use scalar plots for publication.

Built with plotly.js, react and material-ui.

Installation

pip install tensorboard-plugin-customizable-plots

After running Tensorboard, a new tab labeled CUSTMOZIABLE PLOTS will be added to the Tensorboard dashboard.

Features

The plugin has almost the same features as the TIME SERIES or the SCALARS dashboards, plus plotly.js features:

The ability to customize the plot title and the axis labels.
The ability to customize the colors.
Legends are attached with each plot.
X-axis and Y-axis both support log scale.
The customized plots can be exported to many image formats including svg and png.

Check plotly.js documentation for the full list of features and options.

Limitations

Not as good and stylish as the TIME SERIES plugin.
The settings with a text field are not applied until Enter is pressed (I tried to implement onChange but it makes the plugin slower)

So until the Tensorboard authors implement the above features, one might use the TIME SERIES dashboard to track the progress of the plots (as it is more responsive) and use this plugin to customize and export them.

License

The plugin is licensed under the Apache License, Version 2.0. See LICENSE for the full license text.

tensorboard_plugin_customizable_plots's People

Contributors

Stargazers

Watchers

Forkers

gurpreet-web qitaozhao

tensorboard_plugin_customizable_plots's Issues

Suggestion: support regular expression in the filter input widget

Hello, thank you for making this plugin, it is great, and saves me a lot of time.

Currently, the filter input widget seems not support regular expressions to filter runs, which is a little bit inconvenient.
For example, the default filter widget in the TIME SERIES tab:

Below is the result in the CUSTOMIZABLE PLOT:

how to change the order of legend items

Hello, for the generated figure below, I want to show the items NBC2-small + 7️⃣, NBC2-small + 7️⃣x2, NBC2-small + 7️⃣x3 close. How could I achieve that? And would it possible to provide a button to quickly change the colors of the filtered runs? Because often, I find that the filtered runs have close colors hard to distinguish, e.g. the NBC2-small and the NBC2-small + 7️⃣x2.

Saving the plot

Hi,
Thanks for this plugin..
I had a problem while using it.. when trying to save the plot (after naming the file in "ToImageButtonOptions") the file did not saved anywhere and the new tab where should I see it did not appear. any advice!!

thanks in advance.

Startup error

Following error appears when clicked on the "customizable plots tab" after successful server initialization.

E0913 11:26:42.467386 139851626170112 _internal.py:113] Error on request:
Traceback (most recent call last):
  File "/usr/local/lib/python3.6/dist-packages/werkzeug/serving.py", line 323, in run_wsgi
    execute(self.server.app)
  File "/usr/local/lib/python3.6/dist-packages/werkzeug/serving.py", line 312, in execute
    application_iter = app(environ, start_response)
  File "/usr/local/lib/python3.6/dist-packages/tensorboard/backend/application.py", line 563, in __call__
    return self._app(environ, start_response)
  File "/usr/local/lib/python3.6/dist-packages/tensorboard/backend/application.py", line 155, in wrapper
    return wsgi_app(*args)
  File "/usr/local/lib/python3.6/dist-packages/tensorboard/backend/security_validator.py", line 82, in __call__
    return self._application(environ, start_response_proxy)
  File "/usr/local/lib/python3.6/dist-packages/tensorboard/backend/path_prefix.py", line 71, in __call__
    return self._application(environ, start_response)
  File "/usr/local/lib/python3.6/dist-packages/tensorboard/backend/experiment_id.py", line 76, in __call__
    return self._application(environ, start_response)
  File "/usr/local/lib/python3.6/dist-packages/tensorboard/backend/empty_path_redirect.py", line 47, in __call__
    return self._application(environ, start_response)
  File "/usr/local/lib/python3.6/dist-packages/tensorboard/backend/application.py", line 586, in _route_request
    return self.exact_routes[clean_path](environ, start_response)
  File "/usr/local/lib/python3.6/dist-packages/werkzeug/wrappers/base_request.py", line 238, in application
    resp = f(*args[:-2] + (request,))
  File "/usr/local/lib/python3.6/dist-packages/tensorboard_plugin_customizable_plots/plugin.py", line 144, in _serve_runs
    ctx = plugin_util.context(request.environ)
AttributeError: module 'tensorboard.plugin_util' has no attribute 'context'

abdeladim-s / tensorboard_plugin_customizable_plots Goto Github PK