Usage

#1
by Mousey - opened

Hi Kaikai, great work, however, I am not really able to run it or I'm not sure I'm doing that right.

pipe = DiffusionPipeline.from_pretrained(
pretrained_model_name_or_path,
torch_dtype=torch.float16
).to("cuda")

gives an error, so I had to add low_cpu_mem_usage=False to it. Then it runs but the results are not equal to the ones you are getting here. Any ideas ?

Hi Mousey,
thanks for asking. From the infomation given, I can't really tell what's going wrong. Usually, the problem has to do with memory. But, if you can run the model, I suspect the problem has to do with the prompt. Did you use the template keywords before and after in the prompt?

Sign up or log in to comment