[2/n][rollout] feat: flowgrpo - add diffusion agent loop support

[AndyZhou952]

What does this PR do?

Add diffusion agent loop support

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, veomni, 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, fully_async, one_step_off
  • 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.

[zhtmike]

Possibly inherited from AgentLoopWorker, check repeated methods?

[AndyZhou952]

Feels like it makes more sense to keep separate classes, considering the amount of overwrites needed. (Trace removed, DiffusersModelConfig, sampling params etc.)

[zhtmike]

fix CI pls

[Copilot]

Pull request overview

This PR introduces diffusion-model support to the experimental agent loop stack, adding a diffusion-specific single-turn agent loop, a diffusion worker pipeline for batching/postprocessing, and an initial test targeting diffusion rollouts.

Changes:

• Add DiffusionSingleTurnAgentLoop and register it as diffusion_single_turn_agent.
• Extend the agent loop manager/worker codepath with DiffusionAgentLoopWorker and diffusion-specific output schemas.
• Add a new diffusion agent loop test under tests/experimental/agent_loop.

Reviewed changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated 9 comments.

File	Description
verl/experimental/agent_loop/single_turn_agent_loop.py	Adds a diffusion-oriented single-turn agent loop that passes negative prompts and expects image outputs.
verl/experimental/agent_loop/agent_loop.py	Adds diffusion output models, diffusion worker implementation, and routes manager to the diffusion worker based on model_type.
verl/experimental/agent_loop/__init__.py	Exposes DiffusionAgentLoopWorker in the package exports.
tests/experimental/agent_loop/test_diffusion_agent_loop.py	Adds an end-to-end-style diffusion rollout test.

---

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

You can also share your feedback on Copilot code review. Take the survey.

[zhtmike]

move to verl PR