[data] feat: TransferQueue - Unify the return of reward

[walterchenchn]

What does this PR do?

Unify the return values of the reward function to make the logic clearer

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.

[CLAassistant]

All committers have signed the CLA.

[gemini-code-assist]

Code Review

This pull request aims to unify the return values of the reward function, which is a good refactoring goal. However, the current implementation introduces a critical bug related to the REMAX baseline calculation and also makes the function's type hint and docstring incorrect and misleading. I've provided two review comments with high and critical severity to address these issues. Please review them carefully.

[Copilot]

Pull request overview

This PR refactors the _compute_or_extract_reward method in the PPO ray trainer to simplify and unify its return value logic. The parameter return_dict is renamed to reward_for_val to better reflect its purpose, and the return value handling is streamlined.

Changes:

• Renamed parameter return_dict to reward_for_val to clarify that it controls validation vs training reward computation paths
• Simplified return logic: now returns either a tensor (for REMAX with sum_reward) or a tuple (all other cases), removing the dictionary return type
• Updated all call sites to match the new return signature (tuple unpacking instead of dictionary access)

---

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

[0oshowero0]

Add descriptions for reward_extra_infos

[walterchenchn]

fixed

[0oshowero0]

unify with reward_extra_info_dict?

[walterchenchn]

fixed

[ji-huazhong]

please run pre-commit run first

[walterchenchn]

please run pre-commit run first

done