All my activities in one place–mostly dev stuff.

Tag: TensorFlow

Using Combined Adversarial Loss in TensorFlow GANEstimator

This will be a quick tip on how to use combine_adversarial_loss in tf.contrib.gan.estimator.GANEstimator. In my latest projects, I have been using TensorFlow estimators. Estimators allow you to focus more on creating models and wraps the whole training (including saving, exporting, and putting a model in production) into few lines. Recently, I experienced the limits of estimators when I wanted to train a generative adversarial network (GAN) with a combined adversarial loss. In this article, I will show you a little trick how to do that.

TF tricks #1 Using GANs

Importing Multiple TensorFlow Models (Graphs)

Updated 6th March 2018: importing operations by name.

There are a lot of great things about TensorFlow. BUT once I figure out how to import my trained model (graph), I wasn’t able to import second model and use it alongside with the first one. The importing is pretty slow and I don’t want to do it more than once. On the other hand, squashing everything into one model seems to me pretty impractical.

In this tutorial, I will show how to save and import TensorFlow model. Even more, how to import multiple models alongside.

Importing multiple TensorFlow Graphs

2022 © Břetislav Hájek