use StableDiffusionXLPipeline_HPU

[Jianhong-Zhang]

use StableDiffusionXLPipeline_HPU for SDXL

[imangohari1]

hi @Jianhong-Zhang
what's the status of this PR? Could you rebase/sync this with main OH, and share some comments/description on what does this PR do with some results?
Thanks.

[imangohari1]

This PR needs some work:

• I made some changes that are needed. Pls. Apply the patch via git am <0001*
  0001-fea-fixes.patch
• The StableDiffusionXLPipeline_HPU crashes with below error. Pls address.
  This PR can't be merged right now.

python text_to_image_generation.py     --model_name_or_path stabilityai/stable-diffusion-xl-base-1.0     --prompts "Sailing ship painting by Van Gogh"     --num_images_per_prompt 28     --batch_size 7     --image_save_dir /tmp/stable_diffusion_xl_images     --scheduler euler_discrete     --use_habana     --use_hpu_graphs     --gaudi_config Habana/stable-diffusion     --bf16 --sdxl_mlperf

Traceback (most recent call last):
  File "/devops/sgohari/tests/codes/pr-reviews/1300/optimum-habana/examples/stable-diffusion/text_to_image_generation.py", line 549, in <module>
    main()
  File "/devops/sgohari/tests/codes/pr-reviews/1300/optimum-habana/examples/stable-diffusion/text_to_image_generation.py", line 518, in main
    outputs = pipeline(prompt=args.prompts, **kwargs_call)
  File "/usr/local/lib/python3.10/dist-packages/torch/utils/_contextlib.py", line 115, in decorate_context
    return func(*args, **kwargs)
  File "/usr/local/lib/python3.10/dist-packages/optimum/habana/diffusers/pipelines/stable_diffusion_xl/pipeline_stable_diffusion_xl_mlperf.py", line 599, in __call__
    if self.quantized:
  File "/usr/local/lib/python3.10/dist-packages/diffusers/configuration_utils.py", line 143, in __getattr__
    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
AttributeError: 'StableDiffusionXLPipeline_HPU' object has no attribute 'quantized'

[imangohari1]

@Jianhong-Zhang
Any update on this PR please? could you address my comments here? Thanks.

This PR needs some work:

* I made some changes that are needed. Pls. Apply the patch via `git am <0001*`
  [0001-fea-fixes.patch](https://github.com/user-attachments/files/16984799/0001-fea-fixes.patch)

* The `StableDiffusionXLPipeline_HPU` crashes with below error. Pls address.
  This PR can't be merged right now.

python text_to_image_generation.py     --model_name_or_path stabilityai/stable-diffusion-xl-base-1.0     --prompts "Sailing ship painting by Van Gogh"     --num_images_per_prompt 28     --batch_size 7     --image_save_dir /tmp/stable_diffusion_xl_images     --scheduler euler_discrete     --use_habana     --use_hpu_graphs     --gaudi_config Habana/stable-diffusion     --bf16 --sdxl_mlperf

Traceback (most recent call last):
  File "/devops/sgohari/tests/codes/pr-reviews/1300/optimum-habana/examples/stable-diffusion/text_to_image_generation.py", line 549, in <module>
    main()
  File "/devops/sgohari/tests/codes/pr-reviews/1300/optimum-habana/examples/stable-diffusion/text_to_image_generation.py", line 518, in main
    outputs = pipeline(prompt=args.prompts, **kwargs_call)
  File "/usr/local/lib/python3.10/dist-packages/torch/utils/_contextlib.py", line 115, in decorate_context
    return func(*args, **kwargs)
  File "/usr/local/lib/python3.10/dist-packages/optimum/habana/diffusers/pipelines/stable_diffusion_xl/pipeline_stable_diffusion_xl_mlperf.py", line 599, in __call__
    if self.quantized:
  File "/usr/local/lib/python3.10/dist-packages/diffusers/configuration_utils.py", line 143, in __getattr__
    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
AttributeError: 'StableDiffusionXLPipeline_HPU' object has no attribute 'quantized'