Update diffusers requirement from <=0.36.0,>=0.22.0 to >=0.22.0,<=0.37.0 in /tools/who_what_benchmark (openvinotoolkit#3460)

dependabot[bot] · web-flow · commit 1e3fce5f7314 · 2026-03-10T07:39:04.000Z
Updates the requirements on [diffusers](https://github.com/huggingface/diffusers) to permit the latest version. <details> <summary>Release notes</summary> <p><em>Sourced from <a href="https://github.com/huggingface/diffusers/releases">diffusers's releases</a>.</em></p> <blockquote> <h2>Diffusers 0.37.0: Modular Diffusers, New image and video pipelines, multiple core library improvements, and more 🔥</h2> <h2>Modular Diffusers</h2> <p>Modular Diffusers introduces a new way to build diffusion pipelines by composing reusable blocks. Instead of writing entire pipelines from scratch, you can now mix and match building blocks to create custom workflows tailored to your specific needs! This complements the existing <code>DiffusionPipeline</code> class, providing a more flexible way to create custom diffusion pipelines.</p> <p>Find more details on how to get started with Modular Diffusers <a href="https://huggingface.co/docs/diffusers/en/modular_diffusers/quickstart">here</a>, and also check out the <a href="https://huggingface.co/blog/modular-diffusers">announcement post</a>.</p> <h2>New Pipelines and Models</h2> <h3>Image 🌆</h3> <ul> <li><a href="https://huggingface.co/docs/diffusers/en/api/pipelines/z_image">Z Image Omni Base</a>: Z-Image is the foundation model of the Z-Image family, engineered for good quality, robust generative diversity, broad stylistic coverage, and precise prompt adherence. While Z-Image-Turbo is built for speed, Z-Image is a full-capacity, undistilled transformer designed to be the backbone for creators, researchers, and developers who require the highest level of creative freedom. Thanks to <a href="https://github.com/RuoyiDufor"><code>@​RuoyiDufor</code></a> for contributing this in <a href="https://redirect.github.com/huggingface/diffusers/issues/12857">#12857</a>.</li> <li><a href="https://huggingface.co/docs/diffusers/main/en/api/pipelines/flux2#diffusers.Flux2KleinPipeline">Flux2 Klein</a>:FLUX.2 [Klein] unifies generation and editing in a single compact architecture, delivering state-of-the-art quality with end-to-end inference in as low as under a second. Built for applications that require real-time image generation without sacrificing quality, and runs on consumer hardware, with as little as 13GB VRAM.</li> <li><a href="https://huggingface.co/Qwen/Qwen-Image-Layered">Qwen Image Layered</a>: Qwen-Image-Layered is a model capable of decomposing an image into multiple RGBA layers. This layered representation unlocks inherent editability: each layer can be independently manipulated without affecting other content. Thanks to <a href="https://github.com/naykun"><code>@​naykun</code></a> for contributing this in <a href="https://redirect.github.com/huggingface/diffusers/issues/12853">#12853</a>.</li> <li><a href="https://huggingface.co/docs/diffusers/main/en/api/pipelines/bria_fibo_edit">FIBO Edit</a>: Fibo Edit is an 8B parameter image-to-image model that introduces a new paradigm of structured control, operating on JSON inputs paired with source images to enable deterministic and repeatable editing workflows. Featuring native masking for granular precision, it moves beyond simple prompt-based diffusion to offer explicit, interpretable control optimized for production environments. Its lightweight architecture is designed for deep customization, empowering researchers to build specialized “Edit” models for domain-specific tasks while delivering top-tier aesthetic quality. Thanks galbria for contributing it in <a href="https://redirect.github.com/huggingface/diffusers/pull/12930">huggingface/diffusers#12930</a>.</li> <li><a href="https://huggingface.co/docs/diffusers/main/en/api/pipelines/cosmos">Cosmos Predict2.5</a>: Cosmos-Predict2.5, the latest version of the Cosmos World Foundation Models (WFMs) family, specialized for simulating and predicting the future state of the world. Thanks to <a href="https://github.com/miguelmartin75"><code>@​miguelmartin75</code></a> for contributing it in <a href="https://redirect.github.com/huggingface/diffusers/issues/12852">#12852</a>.</li> <li><a href="https://huggingface.co/docs/diffusers/main/en/api/pipelines/cosmos">Cosmos Transfer2.5</a>: Cosmos-Transfer2.5 is a conditional world generation model with adaptive multimodal control, that produces high-quality world simulations conditioned on multiple control inputs. These inputs can take different modalities—including edges, blurred video, segmentation maps, and depth maps. Thanks to <a href="https://github.com/miguelmartin75"><code>@​miguelmartin75</code></a> for contributing it in <a href="https://redirect.github.com/huggingface/diffusers/issues/13066">#13066</a>.</li> <li><a href="https://huggingface.co/docs/diffusers/main/en/api/pipelines/glm_image">GLM-Image</a>: GLM-Image is an image generation model adopts a hybrid autoregressive + diffusion decoder architecture, effectively pushing the upper bound of visual fidelity and fine-grained details. In general image generation quality, it aligns with industry-standard LDM-based approaches, while demonstrating significant advantages in knowledge-intensive image generation scenarios. Thanks to <a href="https://github.com/zRzRzRzRzRzRzR"><code>@​zRzRzRzRzRzRzR</code></a> for contributing it in <a href="https://redirect.github.com/huggingface/diffusers/pull/12973">huggingface/diffusers#12973</a>.</li> <li><a href="https://huggingface.co/docs/diffusers/main/api/models/autoencoder_rae">RAE</a>: Representation Autoencoders (aka RAE) are an exciting alternative to traditional VAEs, typically used in the area of latent-space diffusion models of image generation. RAEs leverage pre-trained vision encoders and train lightweight decoders for the task of reconstruction.</li> </ul> <h3>Video + audio 🎥 🎼</h3> <ul> <li><a href="https://huggingface.co/docs/diffusers/main/en/api/pipelines/ltx2">LTX-2</a>: LTX-2 is an audio-conditioned text-to-video generation model that can generate videos with synced audio. Full and distilled model inference, as well as two-stage inference with spatial sampling, is supported. We also support a conditioning pipeline that allows for passing different conditions (such as images, series of images, etc.). Check out the docs to learn more!</li> <li><a href="https://huggingface.co/docs/diffusers/main/api/pipelines/helios">Helios</a>: Helios is a 14B video generation model that runs at 17 FPS on a single NVIDIA H100 GPU and supports minute-scale generation while matching a strong baseline in quality. Thanks to <a href="https://github.com/SHYuanBest"><code>@​SHYuanBest</code></a> for contributing this in <a href="https://redirect.github.com/huggingface/diffusers/pull/13208">huggingface/diffusers#13208</a>.</li> </ul> <h2>Improvements to Core Library</h2> <h3>New caching methods</h3> <ul> <li><a href="https://redirect.github.com/huggingface/diffusers/pull/12744">MagCache</a> — thanks to <a href="https://github.com/AlanPonnachan"><code>@​AlanPonnachan</code></a>!</li> <li><a href="https://redirect.github.com/huggingface/diffusers/pull/12648/">TaylorSeer</a> — thanks to <a href="https://github.com/toilaluan"><code>@​toilaluan</code></a>!</li> </ul> <h3>New context-parallelism (CP) backends</h3> <ul> <li><a href="https://redirect.github.com/huggingface/diffusers/pull/12693">Unified Sequence Parallel attention</a> — thanks to <a href="https://github.com/Bissmella"><code>@​Bissmella</code></a>!</li> <li><a href="https://redirect.github.com/huggingface/diffusers/pull/12996">Ulysses Anything Attention</a> — thanks to <a href="https://github.com/DefTruth"><code>@​DefTruth</code></a>!</li> </ul> <h3>Misc</h3> <ul> <li>Mambo-G Guidance: New guider implementation (<a href="https://redirect.github.com/huggingface/diffusers/issues/12862">#12862</a>)</li> <li>Laplace Scheduler for DDPM (<a href="https://redirect.github.com/huggingface/diffusers/issues/11320">#11320</a>)</li> <li>Custom Sigmas in UniPCMultistepScheduler (<a href="https://redirect.github.com/huggingface/diffusers/issues/12109">#12109</a>)</li> <li>MultiControlNet support for SD3 Inpainting (<a href="https://redirect.github.com/huggingface/diffusers/issues/11251">#11251</a>)</li> <li>Context parallel in native flash attention (<a href="https://redirect.github.com/huggingface/diffusers/issues/12829">#12829</a>)</li> <li>NPU Ulysses Attention Support (<a href="https://redirect.github.com/huggingface/diffusers/issues/12919">#12919</a>)</li> <li>Fix Wan 2.1 I2V Context Parallel Inference (<a href="https://redirect.github.com/huggingface/diffusers/issues/12909">#12909</a>)</li> <li>Fix Qwen-Image Context Parallel Inference (<a href="https://redirect.github.com/huggingface/diffusers/issues/12970">#12970</a>)</li> <li>Introduction to <code>@apply_lora_scale</code> decorator for simplifying model definitions (<a href="https://redirect.github.com/huggingface/diffusers/issues/12994">#12994</a>)</li> <li>Introduction of pipeline-level “cpu” <code>device_map</code> (<a href="https://redirect.github.com/huggingface/diffusers/issues/12811">#12811</a>)</li> <li>Enable CP for kernels-based attention backends (<a href="https://redirect.github.com/huggingface/diffusers/issues/12812">#12812</a>)</li> </ul>  </blockquote> <p>... (truncated)</p> </details> <details> <summary>Commits</summary> <ul> <li><a href="https://github.com/huggingface/diffusers/commit/d791c5c024014784df4a8dac2601e19fb4d300fc"><code>d791c5c</code></a> Release: v0.37.0-release</li> <li><a href="https://github.com/huggingface/diffusers/commit/8ec0a5ccad96957c10388d2d2acc7fdd8e0fab84"><code>8ec0a5c</code></a> feat: implement rae autoencoder. (<a href="https://redirect.github.com/huggingface/diffusers/issues/13046">#13046</a>)</li> <li><a href="https://github.com/huggingface/diffusers/commit/29b91098f68847f4e50d099735532c7d5735b17e"><code>29b9109</code></a> [attention backends] change to updated repo and version. (<a href="https://redirect.github.com/huggingface/diffusers/issues/13161">#13161</a>)</li> <li><a href="https://github.com/huggingface/diffusers/commit/ae5881ba77fc26df801f0f76a9955cd66ddf68f0"><code>ae5881b</code></a> Fix Helios paper link in documentation (<a href="https://redirect.github.com/huggingface/diffusers/issues/13213">#13213</a>)</li> <li><a href="https://github.com/huggingface/diffusers/commit/ab6040ab2d84b1c6ce26a1f401dabad52cde4df5"><code>ab6040a</code></a> Add LTX2 Condition Pipeline (<a href="https://redirect.github.com/huggingface/diffusers/issues/13058">#13058</a>)</li> <li><a href="https://github.com/huggingface/diffusers/commit/20364fe5a2afbddbf2ec5097ec5d483ce0cc689a"><code>20364fe</code></a> adding lora support to z-image controlnet pipelines (<a href="https://redirect.github.com/huggingface/diffusers/issues/13200">#13200</a>)</li> <li><a href="https://github.com/huggingface/diffusers/commit/3902145b38619d00c8b486fc13ca27f74dbb4a08"><code>3902145</code></a> [lora] fix zimage lora conversion to support for more lora. (<a href="https://redirect.github.com/huggingface/diffusers/issues/13209">#13209</a>)</li> <li><a href="https://github.com/huggingface/diffusers/commit/5570f817da44520d25c053d7da562bd5b2f46989"><code>5570f81</code></a> [Z-Image] Fix more <code>do_classifier_free_guidance</code> thresholds (<a href="https://redirect.github.com/huggingface/diffusers/issues/13212">#13212</a>)</li> <li><a href="https://github.com/huggingface/diffusers/commit/33f785b4447f74eb9be5a798595e5b501b692d3b"><code>33f785b</code></a> Add Helios-14B Video Generation Pipelines (<a href="https://redirect.github.com/huggingface/diffusers/issues/13208">#13208</a>)</li> <li><a href="https://github.com/huggingface/diffusers/commit/06ccde949013eef3b2c57170bdc7873e9f58787e"><code>06ccde9</code></a> Fix group-offloading bug (<a href="https://redirect.github.com/huggingface/diffusers/issues/13211">#13211</a>)</li> <li>Additional commits viewable in <a href="https://github.com/huggingface/diffusers/compare/v0.22.0...v0.37.0">compare view</a></li> </ul> </details> <br /> Dependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting `@dependabot rebase`. [//]: # (dependabot-automerge-start) [//]: # (dependabot-automerge-end) --- <details> <summary>Dependabot commands and options</summary> <br /> You can trigger Dependabot actions by commenting on this PR: - `@dependabot rebase` will rebase this PR - `@dependabot recreate` will recreate this PR, overwriting any edits that have been made to it - `@dependabot show <dependency name> ignore conditions` will show all of the ignore conditions of the specified dependency - `@dependabot ignore this major version` will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself) - `@dependabot ignore this minor version` will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself) - `@dependabot ignore this dependency` will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself) </details> Signed-off-by: dependabot[bot] <support@github.com> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
diff --git a/tools/who_what_benchmark/requirements.txt b/tools/who_what_benchmark/requirements.txt
@@ -7,7 +7,7 @@ pandas>=2.0.3,<=2.3.1
 numpy==1.26.4; platform_system == "Darwin" and platform_machine == "x86_64"
 numpy>=1.16.6,<=2.4.2; platform_system != "Darwin" or platform_machine != "x86_64"
 tqdm>=4.66.1,<=4.67.1
-diffusers>=0.22.0,<=0.36.0
+diffusers>=0.22.0,<=0.37.0
 datasets>=3.6.0,<=4.6.0
 jinja2>=3.1.0,<=3.1.6
 scipy>=1.3.2,<=1.17.1
