Runtime error when building Space with Stable Diffusion backend

I’ve created a (yet private) space which produces Beatles-like lyrics, a song title and an album cover. The text is generated using fine-tuned NLP models, the cover is created with SD.

A few weeks ago everything was running fine, but now I’m stuck on a runtime error without any details. I’m calling SD using the following Gradio block:

stable_diffusion = gr.Blocks.load(name=“spaces/stabilityai/stable-diffusion-1”)

In a Colab environment (without GPU) this is working just fine, it generates images in about 20sec. However, in a HF Space the build ends with an error. Any ideas of what might might be?

Hi @wvangils !

Very cool space idea and thank you for posting.

My guess is that the upstream stable diffusion space and your space and on different gradio versions and your gradio version was released before the batch parameter was introduced.

Can you try upgrading your space version to match the upstream space?

Thx a million @freddyaboulton ! This exactly what was going on. I’ve made a new space, copied the code and it is flying.