The Stable Diffusion notebook is broken.

[zhanyong-wan]

📚 Documentation

The README points to a Stable Diffusion notebook to help a user get started. However, this notebook cannot be run successfully:

1. The import torch_xla step results in an error:

oduleNotFoundError                       Traceback (most recent call last)
/tmp/ipykernel_27/3499457412.py in <module>
----> 1 import torch_xla
      2 torch_xla.__version__

ModuleNotFoundError: No module named 'torch_xla'

This can be fixed by

!pip install torch~=2.6.0 'torch_xla[tpu]~=2.6.0' \
  -f https://storage.googleapis.com/libtpu-releases/index.html \
  -f [https://storage.googleapis.com/libtpu-wheels/index.html](https://storage.googleapis.com/libtpu-wheels/index.html%60)

2. Later, the image = pipeline(prompt, callback=lambda *args: xm.mark_step(), generator=generator).images[0] step failed with

/usr/local/lib/python3.11/dist-packages/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion.py:894: FutureWarning: `callback` is deprecated and will be removed in version 1.0.0. Passing `callback` as an input argument to `__call__` is deprecated, consider using `callback_on_step_end`
  deprecate(
  2%
 1/50 [01:16<1:02:27, 76.48s/it]
---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
[<ipython-input-10-049c86b52afd>](https://localhost:8080/#) in <cell line: 0>()
      2 # xm.mark_step compiles and executes the graph after each iteration.
      3 # The first few steps will be much slower than the rest.
----> 4 image = pipeline(prompt, callback=lambda *args: xm.mark_step(), generator=generator).images[0]
      5 image

1 frames
[/usr/local/lib/python3.11/dist-packages/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion.py](https://localhost:8080/#) in __call__(self, prompt, height, width, num_inference_steps, timesteps, sigmas, guidance_scale, negative_prompt, num_images_per_prompt, eta, generator, latents, prompt_embeds, negative_prompt_embeds, ip_adapter_image, ip_adapter_image_embeds, output_type, return_dict, cross_attention_kwargs, guidance_rescale, clip_skip, callback_on_step_end, callback_on_step_end_tensor_inputs, **kwargs)
   1068                 if i == len(timesteps) - 1 or ((i + 1) > num_warmup_steps and (i + 1) % self.scheduler.order == 0):
   1069                     progress_bar.update()
-> 1070                     if callback is not None and i % callback_steps == 0:
   1071                         step_idx = i // getattr(self.scheduler, "order", 1)
   1072                         callback(step_idx, t, latents)

TypeError: unsupported operand type(s) for %: 'int' and 'NoneType'