adammoss/ddpm-ema-dsprites-cond · Model running instructions

Hi @adammoss ,

I am interested in using this model. When I try to generate an image guided by a simple text prompt, I get an error. I understand that the DDPM is an unconditional model type, so I suspect the class is not designed to take a text prompt input.

I also tried to run the model with no text prompt input, I also get an error about a positional argument missing from a UNetConditionModel forward pass function call:
TypeError: UNet2DConditionModel.forward() missing 1 required positional argument: 'encoder_hidden_states'

How is the model intended to be used? I can't seem to find any documentation.

I have provided my code below

Thanks!
Ricardo

from diffusers import DDPMPipeline
from PIL import Image

pipeline: DDPMPipeline = DDPMPipeline.from_pretrained("adammoss/ddpm-ema-dsprites-cond").to("cuda")

image: Image = pipeline("a circle located in the bottom left corner of the image").images[0]
image.save("circle.png")
image.show()