Integrate AutoRound into Diffusers

[xin3he]

What does this PR do?

This pull request introduces support for the AutoRound quantization algorithm in the Diffusers library. AutoRound is a weight-only quantization method that enables efficient inference by optimizing weight rounding and min-max ranges, primarily targeting the W4A16 configuration (4-bit weights, 16-bit activations). The changes add a new quantization config, quantizer class, backend integration, and comprehensive documentation, while ensuring proper handling of optional dependencies.

Key changes:

AutoRound quantization support

• Added a new AutoRoundConfig class to quantization_config.py for configuring AutoRound quantization parameters, including bits, group size, symmetry, backend, and modules to exclude from quantization.
• Introduced the AutoRoundQuantizer class in quantizers/autoround/autoround_quantizer.py, implementing the logic for loading pre-quantized AutoRound models and integrating with the auto-round library.
• Registered AutoRoundConfig and AutoRoundQuantizer in the quantization auto-mapping logic, enabling selection via the "auto-round" key

Dependency management and import handling

• Added is_auto_round_available utility and integrated it into the main import structure and conditional imports, ensuring that AutoRound features are only available if the dependency is installed. Dummy objects are provided otherwise.
• Implemented a test utility require_auto_round_version_greater_or_equal for version-gated testing of AutoRound features.

Documentation

• Added a comprehensive user guide at docs/source/en/quantization/autoround.md, including usage examples, backend options, configuration details, and resource links.

These changes collectively enable seamless integration of AutoRound quantization into Diffusers, with robust configuration, backend selection, and user guidance.

Existed Model

• https://huggingface.co/INCModel/Z-Image-W4A16-AutoRound
• https://huggingface.co/INCModel/Z-Image-Turbo-W4A16-AutoRound

cc @wenhuach21 @thuang6 @hshen14

Before submitting

• This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
• Did you read the contributor guideline?
• Did you read our philosophy doc (important for complex PRs)?
• Was this discussed/approved via a GitHub issue or the forum? Please add a link to it if that's the case.
• Did you make sure to update the documentation with your changes? Here are the
  documentation guidelines, and
  here are tips on formatting docstrings.
• Did you write any new necessary tests?

Who can review?

@yiyixuxu @asomoza @stevhliu @sayakpaul

Anyone in the community is free to review the PR once the tests have passed. Feel free to tag
members/contributors who may be interested in your PR.

[xin3he]

• [2025/05] AutoRound has been integrated into vLLM: Usage, Medium blog, 小红书.

• [2025/05] AutoRound has been integrated into Transformers: Blog.

I would like to integrate AutoRound into Diffusers to support diffusion models. Although the performance improvement is not significant, memory usage is notably reduced.

[sayakpaul]

Thanks for this PR! Could you provide some example code using AutoRound and also some example outputs? Feel free to also report latency and memory consumption so that there's some signal into its effectiveness.

Cc: @SunMarc

[stevhliu]

thanks for the integration!

[thuang6]

LGTM

[xin3he]

Thank you all for your valuable feedback; I have updated the document and fixed the errors.

[xin3he]

Thanks for this PR! Could you provide some example code using AutoRound and also some example outputs? Feel free to also report latency and memory consumption so that there's some signal into its effectiveness.

Cc: @SunMarc

Apologies; due to policy restrictions, I am unable to provide specific latency data. Based on local testing, the time consumed by the W4A16 Marlin backend is comparable to that of BF16.

For the memory consumption, I observe that the original z-image takes about 25GB and the quantized one takes about 16GB

You are welcome to use the code provided in the model card to attempt to reproduce these results. the generated picture looks good.

[wenhuach21]

in case there is a better backend in the future, we'd better not to explicitly code like this. Besides, If users have install gptqmodel, we will use marlin. Otherwise, we will remind user to install it.

[wenhuach21]

how about just setting the device_map to "auto"

[wenhuach21]

device->auto

[wenhuach21]

0.10 ?

[sayakpaul]

Thanks for this PR! Could you provide some example code using AutoRound and also some example outputs? Feel free to also report latency and memory consumption so that there's some signal into its effectiveness.
Cc: @SunMarc

Apologies; due to policy restrictions, I am unable to provide specific latency data. Based on local testing, the time consumed by the W4A16 Marlin backend is comparable to that of BF16.

For the memory consumption, I observe that the original z-image takes about 25GB and the quantized one takes about 16GB

You are welcome to use the code provided in the model card to attempt to reproduce these results. the generated picture looks good.

I don't understand. Won't it run on a CUDA GPU? You cannot expect maintainers to run the code to address the bare minimums, I am afraid. Having the output samples and having comparisons to other methods gives us confidence in the quantization method itself.