About training problem with stage 2

[nldhuyen0047]

Hi, I finished training model with stage 1, but with stage 2, I see the content_image_file in load_data function, my json file does not have this variable. So the content_prompt with the content_image_file are similar? Should I need to add this variable in my json file before training in stage 2?

Thank you so much, I look forward to hearing from you soon.

[Jeoyal]

content_image_file is the content input image that has been processed by contouring. Please follow our tutorial to prepare it (preferably resized to 512x512) and add this variable in json file.

[nldhuyen0047]

Thank you so muchhh.

[Jeoyal]

:)

[nldhuyen0047]

Hi, T_T I have a question that, I do not know how to inference the code after training with the new checkpoint, could you please explain to me?

Thank you so much.

[Jeoyal]

Please change ip_ckpt and style_aware_encoder_path in each inference script into your checkpoints.

It might follow this path structure: output/checkpoint-xxxxx/pytorch_model.bin (ip_ckpt) output/checkpoint-xxxxx/pytorch_model_1.bin (style_aware_encoder_path)

[nldhuyen0047]

I could run.

Thank you so much.

[Jeoyal]

Hope you achieve the results you desire. :)

[nldhuyen0047]

Yeah, thank you so much. I hope so ^^.