mit-han-lab / gan-compression

[CVPR 2020] GAN Compression: Efficient Architectures for Interactive Conditional GANs
Other
1.1k stars 148 forks source link

Training compressed model in the first place #7

Closed s1ddok closed 4 years ago

s1ddok commented 4 years ago

Is it necessary to first train regular pix2pix and then compress it with the code in this repository? Is it possible to train compressed model straight away?

lmxyy commented 4 years ago

In order to compress the GAN generator with our repo, you need a pre-trained big model. We have released the pre-trained models at https://hanlab.mit.edu/files/gan_compression/pretrained. You could directly download the pre-trained models and compress the models with them. A more detailed tutorial will be released later.