According to your README, the requirements for TensorFlow are >= v1.3.
TF is not only used for TensorBoard (as you state) but also for HParams. The module contrib (which includes HParams) is no longer available in the newer versions (2.x). Thus, I simply created a dotted dictionary since you only use the hparams object to read the values of the hyper params.
This is of course compatible with any versions as long as it runs in Python 3.x (to work on Python 2.x, substitute .items() by .iteritems())
According to your README, the requirements for TensorFlow are >= v1.3.
TF is not only used for TensorBoard (as you state) but also for HParams. The module contrib (which includes HParams) is no longer available in the newer versions (2.x). Thus, I simply created a dotted dictionary since you only use the hparams object to read the values of the hyper params.
This is of course compatible with any versions as long as it runs in Python 3.x (to work on Python 2.x, substitute .items() by .iteritems())
For further uses in hyperaparameter tuning see the new alternative: https://www.tensorflow.org/tensorboard/hyperparameter_tuning_with_hparams.