# Public LB 15.81605 solution

Here is one of my final submitted kernels (I selected quite similar two kernels for final submission).
[[Sub] RaLS AC-BigGAN with MinibatchStddev](https://www.kaggle.com/yukia18/sub-rals-ac-biggan-with-minibatchstddev)
Version 3 got public LB 15.81605.
FID emphasizes sample diversity too much. My images generated from roose threshold (2~) truncated latents are not good as for quality but get good score. Once you download images.zip from my kernel, you can see that my generated images are crappy quality but diversed.
Key points:
Model: based on BigGAN
shared embedding of dog breed labels
hierarchical latent noise
projection
auxiliary classifier (ACGAN)
Loss: RaLS with weighted auxiliary classification loss
Batch size: 128
Exponential Moving Average of generator weights
Uniform input noise on real images