Since TensorFlow 2.0 is about to be released, I spend a few days on the new official tutorial and documentations and find it would be better to be prepared.
The current branch master is using relatively low level API to construct the network. This makes the code redundant and the training process complicated. The Keras API is recommended in the tutorial and I tried to update this project by using Keras API.
Please checkout the new branch keras if you are interested.
Since TensorFlow 2.0 is about to be released, I spend a few days on the new official tutorial and documentations and find it would be better to be prepared.
The current branch
master
is using relatively low level API to construct the network. This makes the code redundant and the training process complicated. The Keras API is recommended in the tutorial and I tried to update this project by using Keras API.Please checkout the new branch
keras
if you are interested.