SG161222--RealVisXL_V4.0项目的vae中缺少diffusion_pytorch_model.bin文件该如何解决

[kangxi1818]

报错如下 但是该项目中本来就没有diffusion_pytorch_model.bin文件

[Z-YuPeng]

HI, thanks for you feedback, whether you use the latest code and run python gradio_app_sdxl_specific_id_low_vram.py ?

[TonySCruze]

try adding use_safetensors=True to your pipeline StableDiffusionXLPipeline.from_pretrained(sd_model_path, variant="fp16", torch_dtype=torch.float16, use_safetensors=True)

I'm not sure this is the correct answer but I was able to bypass that error you're receiving. The issue I'm having now is that it's producing all black images. But this may be unrelated to safetensors or directly related. idk.

[xiehust]

@TonySCruze You are right~ Fixed it

[Z-YuPeng]

Currently, the latest code has been tested on all four models and will not report errors. Please download the latest code and run "python gradio_app_sdxl_specific_id_low_vram.py".