Add transformer cache context for SkyReels-V2 pipelines & Update docs

[tolgacangoz]

This PR adds transformer cache context for SkyReels-V2 pipelines and & updates docs.

• Pipelines and pipeline callbacks: @yiyixuxu and @asomoza

[tolgacangoz]

Similar to https://huggingface.co/docs/transformers/main/en/model_doc/speech_to_text_2#:~:text=This%20model%20was%20contributed%20by%20Patrick%20von%20Platen.

[tolgacangoz]

As the contributor of this model spending almost 2 months on its integration, I request that the mention be added.

Because SkyReels-V2/I wasn't mentioned in the 0.35.x releases at the new pipelines section.

I cannot understand why such a small PR has been waiting for 3 weeks!

@yiyixuxu @sayakpaul @DN6 @asomoza @dg845

[yiyixuxu]

Your contribution is included in the release notes under the "Significant community contributions" section. The highlights section at the top is reserved for a small number of the most impactful additions — not every merged PR is featured there, regardless of the effort involved.

[tolgacangoz]

I disagree with you that you evaluated the integration of SkyReels-V2 as not impactful enough. This model was the first open-source video generative model employing AutoRegressive Diffusion-Forcing architecture. And, Aryan's mentioning.

[yiyixuxu]

The decision on what goes in the highlights section is at the maintainers' discretion and isn't something we debate on a per-PR basis. Your contribution is credited appropriately in the release notes.
I'm going to consider this discussion closed.

[tolgacangoz]

FLF2V was producing bad results at the 1.3B model; I didn't even create its HF repo, but put it here as a placeholder at that time.

[Copilot]

Pull request overview

This PR adds transformer cache context support to SkyReels-V2 pipelines to enable caching optimizations during inference. The cache contexts separate conditional and unconditional forward passes with "cond" and "uncond" labels respectively.

• Wraps transformer calls with cache_context() for both conditional and unconditional predictions
• Applies the pattern consistently across all five SkyReels-V2 pipeline variants
• Updates documentation with contributor credit (unrelated change)

Reviewed changes

Copilot reviewed 6 out of 6 changed files in this pull request and generated 1 comment.

Show a summary per file

File	Description
src/diffusers/pipelines/skyreels_v2/pipeline_skyreels_v2_i2v.py	Wraps conditional and unconditional transformer calls with cache_context("cond") and cache_context("uncond")
src/diffusers/pipelines/skyreels_v2/pipeline_skyreels_v2_diffusion_forcing_v2v.py	Wraps conditional and unconditional transformer calls with cache_context("cond") and cache_context("uncond")
src/diffusers/pipelines/skyreels_v2/pipeline_skyreels_v2_diffusion_forcing_i2v.py	Wraps conditional and unconditional transformer calls with cache_context("cond") and cache_context("uncond")
src/diffusers/pipelines/skyreels_v2/pipeline_skyreels_v2_diffusion_forcing.py	Wraps conditional and unconditional transformer calls with cache_context("cond") and cache_context("uncond")
src/diffusers/pipelines/skyreels_v2/pipeline_skyreels_v2.py	Wraps conditional and unconditional transformer calls with cache_context("cond") and cache_context("uncond")
docs/source/en/api/pipelines/skyreels_v2.md	Removes FLF2V model listing and adds contributor credit (unrelated to cache context changes)

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

Pull request overview

Copilot reviewed 6 out of 6 changed files in this pull request and generated 1 comment.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[yiyixuxu]

thanks!

[HuggingFaceDocBuilderDev]

The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update.