Some question about global_torch

[junyizeng]

Thank you very much for your open source. In the reproduction of optimization and mapper, I saw amazing results, but I will reproduce your global_ Torch encountered some difficulties. First of all, you provided a website in StyleCLIP to download the pre training model, but when I used torch. load to load it, the following error occurred. How can I resolve it? In addition, can you share the test code for an experiment that does not appear to have the FFHQ dataset given in your paper?

[Zongze Wu]

Dear Junyizeng，

Thank you for your interest in our work.

Did you download the checkpoint from https://drive.google.com/file/d/1FlAb1rYa0r_--Zj_ML8e6shmaF28hQb5/view ? You need to go there page and click download. Directly wget will not work. It seems the checkpoint you download is corrupt.

Thank you for your help.

Best Wishes,

Zongze

[junyizeng]

Hello, I have downloaded the corresponding pre-trained model according to the URL you provided, but the previous error still occurs.

[Zongze Wu]

I just try the codes in codelab, and it works well.

Please just run the 'Setup (may take a few minutes)' section. Then download the checkpoint from google drive, upload it to codelab (this will take a few mins, pls make sure the uploading is completed), change the model path in here.

then add a cell after 'setup', and run

cd /content/StyleCLIP/global_torch/
!python StyleCLIP.py

This works well for me. If you meet other problems, please show me the errors. Thx

[junyizeng]

Thank you very much for your reply, the problem has been solved.