Load trained model to produce an interactive interface.

[Chasexj]

Thanks for the great work. I followed your tutorial on "Train a ControlNet to Control SD" and reproduced the results, as well as a state dict saved under './models/control_sd15_ini.ckpt'. I am wondering if you can provide a brief explanation on how to save/load the generated model and yield an interactive interface from it by running a script such as the 'gradio_canny2image.py'. I tried replacing the line model.load_state_dict(load_state_dict('./models/control_sd15_depth.pth', location='cuda')) with model.load_state_dict(load_state_dict('./models/control_sd15_ini.ckpt', location='cuda')) but the outcome is not as expected.