[trainer, worker] fix: setting old log probs equal to log probs for on policy training

[sahilpatelsp]

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.

Since training backend (FSDP/Megatron)'s recompute of log probs are not accurate, so given an exact batch forwarding twice, the old_log_probs vs log_probs are not the same even in on policy training. This PR quickly fix this issue.

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
  • 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).)

[sahilpatelsp]

@zhaochenyang20 will help to take this merged.

[gemini-code-assist]

Code Review

This pull request aims to fix an issue with old_log_probs during on-policy training by setting them equal to the newly computed log_probs. The logic for identifying on-policy training and updating old_log_probs is sound. However, there is a critical bug in the implementation where the old_log_prob variable is used before it is defined, which will cause a runtime error. This needs to be fixed by reordering the code.

[zhaochenyang20]

Also, vint find this out:

id: vint-1
email: [email protected]

will add him as the co-author.