Some details about the training parameters.

I trained the network from scratch but got poor results. Below are my training parameters(same as the training guide you provide).

--program_name=twingan
--dataset_name="image_only"
--dataset_dir="dir to celeba tfrecord files"
--unpaired_target_dataset_name="anime_faces"
--unpaired_target_dataset_dir="dir to the anime tfrecord you provided"
--train_dir="dir to save results"
--dataset_split_name=train
--preprocessing_name="danbooru"
--resize_mode=RANDOM_CROP
--do_random_cropping=True
--learning_rate=0.0001
--learning_rate_decay_type=fixed
--is_training=True
--generator_network="pggan"
--loss_architecture=dragan
--pggan_max_num_channels=256
--generator_norm_type=batch_renorm
--use_ttur=True
--num_images_per_resolution=50000

Compared with official PGGAN repo, I found some differences.

1. num_image_per_resolution of TwinGAN is 50000 while PGGAN is 600000.
2. TwinGAN uses RANDOM_CROP while PGGAN uses RESIZE directly.

Could you please help me out here?

[jerryli27]

Hi @lionel3 I updated the training documentation. There was indeed a bug in my default parameters. After fixing that I am able to reproduce my previous results.

Please sync to the latest version and see https://github.com/jerryli27/TwinGAN/blob/master/docs/training.md .

The parameters I added are:

--do_pixel_norm=True
--l_content_weight=0.1
--l_cycle_weight=1.0

Please reopen this issue if you cannot reproduce. Thanks!

[jerryli27]

I added two lines to the training script. It should work now. --gradient_penalty_lambda=0.25 --use_unet=True

The whole script now looks like:

python pggan_runner.py
--program_name=twingan
--dataset_name="image_only"
# Assume you have data like 
# ./data/celeba/train-00000-of-00100.tfrecord,  
# ./data/celeba/train-00001-of-00100.tfrecord ...
--dataset_dir="./data/celeba/"
--unpaired_target_dataset_name="anime_faces"
--unpaired_target_dataset_dir="./data/anime_faces/"
--train_dir="./checkpoints/twingan_faces/"
--dataset_split_name=train
--preprocessing_name="danbooru"
--resize_mode=RESHAPE
--do_random_cropping=True
--learning_rate=0.0001
--learning_rate_decay_type=fixed
--is_training=True
--generator_network="pggan"
--use_unet=True
--num_images_per_resolution=300000
--loss_architecture=dragan
--gradient_penalty_lambda=0.25
--pggan_max_num_channels=256
--generator_norm_type=batch_renorm
--hw_to_batch_size="{4: 8, 8: 8, 16: 8, 32: 8, 64: 8, 128: 4, 256: 3, 512: 2}"

I haven’t tested with the multi-gpu setting thoroughly yet due to limits in hardware, so yes there may be some bug, but you can try to add the following flags.

--sync_replicas=False
--replicas_to_aggregate=1
--num_clones=2
--worker_replicas=1

I updated the training readme with the comments above.