Youngjung

Hi, You can use ‘—mode sample’ for that. Check [this part in README](https://github.com/clovaai/stargan-v2/blob/master/README.md#generating-interpolation-videos).

I am confused with your word "eval". Do you mean without "reference image"? If yes, you can do that with https://github.com/clovaai/stargan-v2/blob/e28bdee908fdf3cb787cc2dfa3310d586e4b60ed/core/utils.py#L78

We don't have an answer for Q1. We tested our code on linux only. Maybe someone else with windows experience could provide some help here. Q2 is resolved in #3...

You can use the script `download.sh` ``` bash download.sh pretrained-network-celeba-hq ``` Please refer to https://github.com/clovaai/stargan-v2#datasets-and-pre-trained-networks for more information.

Ah, I didn't notice we did not upload optims and net. We will discuss it and let you know.

Hi, `y` holds indices for the target domain which should be the same with the reference images. `idx` is an enumeration of indices over the minibatch, i.e., `[0, 1, ......

With your pre-trained model, I got 0.05750814338518269. With my-trained model (300K iters), I got 0.055037914931719. I uploaded my code at https://github.com/youngjung/layout2im with 3 more commits. Thanks, Youngjung

Oh, I did not mean to blame you but just wanted to help your repo and other users. Thanks for your response :)