Skip to content

Latest commit

 

History

History

dcgan_tf_keras

Folders and files

NameName
Last commit message
Last commit date

parent directory

..
images
images
 
 
README.md
README.md
 
 
const.yaml
const.yaml
 
 
data.py
data.py
 
 
dc_gan.py
dc_gan.py
 
 
distributed.yaml
distributed.yaml
 
 
export.py
export.py
 
 
model_def.py
model_def.py
 
 

README.md

DCGAN TensorFlow Keras GAN Example

This example demonstrates how to build a simple GAN on the MNIST dataset using Determined's TensorFlow Keras API. This example is adapted from this TensorFlow Tutorial. The DCGAN Keras model featured in this example subclasses tf.keras.Model and defines a custom train_step() and test_step(). This functionality was first added in TensorFlow 2.2.

Files

  • dc_gan.py: The code code defining the model.
  • data.py: The data loading and preparation code for the model.
  • model_def.py: Organizes the model into Determined's TensorFlow Keras API.
  • export.py: Exports a trained checkpoint and uses it to generate images.

Configuration Files

  • const.yaml: Train the model with constant hyperparameter values.
  • distributed.yaml: Same as const.yaml, but instead uses multiple GPUs (distributed training).

To Run

Installation instructions can be found under docs/install-admin.html or at Determined installation page. After configuring the settings in const.yaml, run the following command: det -m <master host:port> experiment create -f const.yaml .

To Export

Once the model has been trained, its top checkpoint can be exported and used to generate images by running:

python export.py --experiment-id <experimend_id> --master-url <master:port>

Generate Images