Closed normanheckscher closed 7 years ago
Cool. Thanks!
No worries.
Do you think you could create a branch from late November 2016 and call it tf-0.12.1 so that a point is saved for the old library?
Nice job getting compatibility with 1.0. This is awesome! Such a big release :+1:
After watching most of the TensorFlow Dev Summit 2017 videos, I think there is a need to use named arguments with all tf python functions. If my understanding of what was said is correct, this will allow for backward compatibility as we move forward with 2.0 & 3.0 & etc releases.
tf.mul() -> tf.multiply()