How to use the model trained with dreambooth?

[wangqiang9]

I obtained the model and weight files by finetuning dreambooth stable diffusion and placed them in the folder: work_dirs/dreambooth/
Then I used the inference code as follows:

from mmagic.apis import MMagicInferencer
model_ckpt = './work_dirs/dreambooth/iter_1000.pth'
model_config = './work_dirs/dreambooth/dreambooth.py'

sd_inferencer = MMagicInferencer(model_name='stable_diffusion',
                                 model_config=model_config,
                                 model_ckpt=model_ckpt)
text_prompts = 'a sks dog on the desk'
result_out_dir = 'output/result.png'
sd_inferencer.infer(text=text_prompts, result_out_dir=result_out_dir)

The error is as follows:

TypeError: argument of type 'NoneType' is not iterable.

How should the correct inference code be written?

[Z-Fran]

Use sd config to infer:

from mmagic.apis import MMagicInferencer
model_ckpt = './work_dirs/dreambooth/iter_100.pth'
model_config = './configs/stable_diffusion/stable-diffusion_ddim_denoisingunet.py'

sd_inferencer = MMagicInferencer(model_name='stable_diffusion',
                                 model_config=model_config,
                                 model_ckpt=model_ckpt)
text_prompts = 'a sks dog on the desk'
result_out_dir = 'output/result.png'
sd_inferencer.infer(text=text_prompts, result_out_dir=result_out_dir)