This PR incorporates TensorBoard into the HABs repo code. TensorBoard was added as a requirement in requirements.txt. In train.py, an import for SummaryWriter was added. Scalar write statements were included for the train loss and val loss in the training loop.
Now when ! pip install -r /content/habs/requirements.txt is called in Colab, TensorBoard will be installed. Locally, TensorBoard was already previously installed in my conda habs-env because my environment.yml file includes tensorflow, which automatically installs tensorboard. It is still important/necessary to include TensorBoard in requirements.txt because in the future we might remove the Tensorflow requirement because I think that was necessary for my original implementation that used Keras and I don't think we use it anymore. Also, in Colab, we install all of our dependencies with the requirements file. Therefore, we don't want to rely on only installing TensorBoard through TensorFlow in the environment file.
This PR incorporates
TensorBoard
into the HABs repo code. TensorBoard was added as a requirement inrequirements.txt
. Intrain.py
, an import forSummaryWriter
was added. Scalar write statements were included for the train loss and val loss in the training loop.Now when
! pip install -r /content/habs/requirements.txt
is called in Colab, TensorBoard will be installed. Locally, TensorBoard was already previously installed in my condahabs-env
because myenvironment.yml
file includes tensorflow, which automatically installs tensorboard. It is still important/necessary to includeTensorBoard
inrequirements.txt
because in the future we might remove the Tensorflow requirement because I think that was necessary for my original implementation that used Keras and I don't think we use it anymore. Also, in Colab, we install all of our dependencies with the requirements file. Therefore, we don't want to rely on only installing TensorBoard through TensorFlow in the environment file.Related Issue: #49