[trainer, recipe] feat: Fully Async Policy add Rollout Importance Sampling

[ArronHZG]

What does this PR do?

In Fully Async Policy, support for recomputing rollout log probabilities is enabled, and Rollout Importance Sampling is activated.

future work for #2981

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).)

[gemini-code-assist]

Code Review

This pull request introduces Rollout Importance Sampling for the Fully Async Policy. This involves adding logic to recompute old_log_prob using the trainer's model, which requires saving and restoring model states to/from CPU. The changes include new utility functions for FSDP2 model state manipulation, modifications to the trainer logic to handle model state caching and recomputation, and updates to documentation and experiment scripts. My review has identified two high-severity issues. One is a maintainability risk in fstp_utils.py due to reliance on private PyTorch FSDP APIs. The other is a correctness issue in a new experiment script, where a necessary flag to enable the new feature is missing.

[szrlee]

related to Rollout Importance Sampling #3694
Rollout Correction #3984
and the original blog

@online{liu-li-2025-rl-collapse,
  title = {When Speed Kills Stability: Demystifying {RL} Collapse from the Training-Inference Mismatch},
  author = {Liu, Jiacai and Li, Yingru and Fu, Yuqian and Wang, Jiawei and Liu, Qian and Shen, Yu},
  year = {2025},
  month = sep,
  url = {https://richardli.xyz/rl-collapse}
}