Home

Welcome to the tensorboard-pytorch wiki!

Before debugging

You might want to update the package if there is any unexpected behavior.

pip uninstall tensorboard-pytorch tensorboardX

pip install tensorboardX --upgrade

Install the newest version from source:

git clone https://github.com/lanpa/tensorboardX && cd tensorboardX && python setup.py install

FAQ

why called tensorboardX?

At first, the package was named tensorboard, and soon there are issues about name confliction. The first alternative name came to my mind is tensorboard-pytorch, but in order to make it more general, I chose tensorboardX which stands for tensorboard for X.

tensorboard vs. tensorboardX vs. tensorflow-tensorboard

When you pip install tensorflow[-gpu], you also get an official tensorboard called tensorflow-tensorboard. But if you later pip install tensorboard you end up with an unofficial tensorboard that overwrites the official one if you already have Tensorflow installed.

The unofficial tensorboard aims to provide a standalone package that is independent of tensorflow. And they registered tensorboard in PyPI before google's tensorboard team. (pip install tensorboard now installs official tensorboard) That's how tensorflow-tensorboard comes from. tensorboardX is a pure package aiming at the logging part. The resulting log file should work with either tensorboard mentioned above. Of course, you will need the official one for newer features such as the precision-recall curve.

read more:

https://github.com/tensorflow/tensorboard/issues/684

https://github.com/dmlc/tensorboard/issues/50

https://github.com/conda-forge/staged-recipes/pull/4778

https://github.com/lanpa/tensorboard-pytorch/issues/72

add_histogram() shows strange results?

Should be fixed in 85a3c8, please pip install git+https://github.com/lanpa/tensorboard-pytorch and try again.

I can not import!

from tensorboard import ... (X)

from tensorboardX import ... (O)