move folders & make for two-forward pass in training loop

[zhtmike]

What does this PR do?

Add concise overview of what this PR aims to achieve or accomplish. Reference related GitHub issues and PRs that help with the review.

Checklist Before Starting

• Search for similar PRs. Paste at least one query link here: ...
• Format the PR title as [{modules}] {type}: {description} (This will be checked by the CI)
  • {modules} include fsdp, megatron, sglang, vllm, rollout, trainer, ci, training_utils, recipe, hardware, deployment, ray, worker, single_controller, misc, perf, model, algo, env, tool, ckpt, doc, data, cfg, reward
  • If this PR involves multiple modules, separate them with , like [megatron, fsdp, doc]
  • {type} is in feat, fix, refactor, chore, test
  • If this PR breaks any API (CLI arguments, config, function signature, etc.), add [BREAKING] to the beginning of the title.
  • Example: [BREAKING][fsdp, megatron] feat: dynamic batching

Test

For changes that can not be tested by CI (e.g., algorithm implementation, new model support), validate by experiment(s) and show results like training curve plots, evaluation results, etc.

API and Usage Example

Demonstrate how the API changes if any, and provide usage example(s) if possible.

# Add code snippet or script demonstrating how to use this

Design & Code Changes

Demonstrate the high-level design if this PR is complex, and list the specific changes.

Checklist Before Submitting

Important

Please check all the following items before requesting a review, otherwise the reviewer might deprioritize this PR for review.

• Read the Contribute Guide.
• Apply pre-commit checks: pre-commit install && pre-commit run --all-files --show-diff-on-failure --color=always
• Add / Update the documentation.
• Add unit or end-to-end test(s) to the CI workflow to cover all the code. If not feasible, explain why: ...
• Once your PR is ready for CI, send a message in the ci-request channel in the verl Slack workspace. (If not accessible, please try the Feishu group (飞书群).)
• If your PR is related to the recipe submodule, please also update the reference to the submodule commit via git submodule update --remote or cd recipe && git pull origin main.

[Copilot]

Pull request overview

This PR reorganizes diffusers-related code by moving it to a new verl/workers/utils/diffusers_patch module and refactors the training loop in the FSDP implementation to use two separate forward passes instead of concatenating inputs for classifier-free guidance.

Key Changes

• Created new diffusers_patch module with scheduler, pipeline, and utility functions for Qwen Image diffusion models
• Modified FSDP diffuser implementation to execute conditional and unconditional forward passes separately instead of batching them
• Updated import paths to reflect the new module structure

Reviewed changes

Copilot reviewed 1 out of 8 changed files in this pull request and generated no comments.

Show a summary per file

File	Description
verl/workers/utils/diffusers_patch/utils.py	New utility functions for preparing and loading diffusion pipelines to devices/dtypes
verl/workers/utils/diffusers_patch/schedulers/scheduling_flow_match_sde_discrete.py	New scheduler implementation with SDE support and log probability computation
verl/workers/utils/diffusers_patch/schedulers/init.py	Exports FlowMatchSDEDiscreteScheduler
verl/workers/utils/diffusers_patch/pipelines/pipeline_qwenimage.py	Extended QwenImage pipeline with log probability tracking and SDE window support
verl/workers/utils/diffusers_patch/pipelines/init.py	Exports QwenImagePipelineWithLogProb
verl/workers/utils/diffusers_patch/patch.py	Function to inject SDE scheduler into existing pipelines
verl/workers/utils/diffusers_patch/init.py	Module-level exports for diffusers_patch functionality
verl/workers/engine/fsdp/diffuser_impl.py	Refactored to use two separate forward passes for CFG instead of input concatenation

---

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