Bring TensorBoard to MXNet.

TensorBoard is a suite of web applications for inspecting and understanding your TensorFlow runs and graphs. TensorBoard currently supports five visualizations: scalars, images, audio, histograms, and the graph.

This README gives an overview of key concepts in TensorBoard, as well as how to interpret the visualizations TensorBoard provides. For an in-depth example of using TensorBoard, see the tutorial: TensorBoard: Visualizing Learning. For in-depth information on the Graph Visualizer, see this tutorial: TensorBoard: Graph Visualization.

When installing from source you will build a pip wheel that you then install using pip. We provide a installer.sh and build_pip_package.sh for you to get that pip wheel.

We’re also working on providing a pre-built pip wheel for you, so you can install TensorBoard package more easily. We would let you know once we finish this feature but currently it has to be installed from source.

$ git clone https://github.com/dmlc/tensorboard.git

Note that this requires Protocol Buffers 3 compiler, so please install it.

Follow instructions here to install the dependencies for bazel.

# For Python 2.7:
$ sudo apt-get install python-numpy python-dev python-wheel python-mock python-protobuf
# For Python 3.x:
$ sudo apt-get install python3-numpy python3-dev python3-wheel python3-mock

Note that this requires Protocol Buffers 3 compiler, so please install it.

Follow instructions here to install the dependencies for bazel. You can then use homebrew to install bazel:

$ brew install bazel

You can install the python dependencies using easy_install or pip, or conda if you use Anaconda for virtual-env management. Using conda, run

$ conda install six, numpy, wheel, protobuf

After that, to build the first part, simply:

$ cd tensorboard
$ sh installer.sh
# In this process, it might need configuration or failed in bazel build, just retry the specific step.

For example(just type ’N’ for all case as we don’t need them):

$ ./configure
Please specify the location of python. [Default is /usr/bin/python]:
Do you wish to build TensorFlow with Google Cloud Platform support? [y/N] N
No Google Cloud Platform support will be enabled for TensorFlow
Do you wish to build TensorFlow with GPU support? [y/N] N
Do you wish to build TensorFlow with OpenCL support? [y/N] N

For logging, you can use the python package in your MXNet for logging some training/validation information, see README in Python package.

For rendering,

$ tensorboard --logdir=path/to/logs

Feel free to contribute your work and don't hesitate to discuss in issue with your ideas.