Train generative adversarial network (GAN) models for image synthesis. Supports DCGAN, WGAN-GP, LSGAN, GGAN, PGGAN, and StyleGAN architectures.
Supported computing resources
Inputs and outputs
Input ports
-
Connect the Read File Data component to specify the Object Storage Service (OSS) path of your training images.
-
Or set the oss path to train data parameter directly to the OSS path of your training images.
Output port
The trained model is saved to the OSS path set in oss path to output distributed.
The following table describes the component parameters.
Configure the component
Add the image generation component to your Machine Learning Designer pipeline and configure the following parameters.
|
Tab |
Parameter |
Required |
Description |
Default value |
|
|
Fields Setting |
oss path to train data |
No |
If no upstream data source is connected, specify the OSS path of your training images. |
None |
|
|
oss path to pretrained model |
No |
OSS path to a pre-trained model for transfer learning. If empty, training starts from scratch. Must be in the same OSS bucket as your training images. |
None |
||
|
oss path to output distributed |
Yes |
OSS path for the trained model output. Must be in the same OSS bucket as your training images. |
None |
||
|
Parameters Setting |
image generation network |
Yes |
GAN architecture for training. Options: DCGAN, WGAN-GP, LSGAN, GGAN, PGGAN, and StyleGAN. |
dcgan |
|
|
learning rate |
Yes |
Learning rate for model optimization. |
0.001 |
||
|
number of train iterations |
Yes |
Total training iterations. Each iteration processes one batch of images (set by batch size). |
10000 |
||
|
visualization iterations |
Yes |
Interval for generating sample images from random noise. Samples are saved to the training_samples folder under oss path to output distributed. |
1000 |
||
|
batch size |
Yes |
Number of images per training iteration. |
32 |
||
|
model save interval |
No |
Checkpoint save frequency during training, in epochs. |
1 |
||
|
Tuning |
Select Resource Group |
Public Resource Group |
No |
Select the instance type and VPC. This algorithm requires a GPU instance. |
None |
|
Dedicated resource group |
No |
Specify CPU cores, memory, shared memory, and GPU count. |
None |
||
|
Maximum Running Duration (seconds) |
No |
Maximum execution time. Training stops when this limit is reached. |
None |
||