[ETOgaosion]

What does this PR do?

1. add profiler post hook, current torch profiler API works with:

    actor_rollout_ref.actor.profiler.enable=True
    actor_rollout_ref.actor.profiler.ranks=${profile_ranks}
    actor_rollout_ref.actor.profiler.all_ranks=${profile_ranks_all}
    actor_rollout_ref.actor.profiler.tool_config.torch.discrete=${profile_discrete}
    actor_rollout_ref.actor.profiler.tool_config.torch.contents=${profile_contents}
    actor_rollout_ref.actor.profiler.tool_config.torch.schedule.skip_first=${profile_schedule_skip_first}
    actor_rollout_ref.actor.profiler.tool_config.torch.schedule.wait=${profile_schedule_wait}
    actor_rollout_ref.actor.profiler.tool_config.torch.schedule.warmup=${profile_schedule_warmup}
    actor_rollout_ref.actor.profiler.tool_config.torch.schedule.active=${profile_schedule_active}
    actor_rollout_ref.actor.profiler.tool_config.torch.schedule.repeat=${profile_schedule_repeat}
    actor_rollout_ref.actor.profiler.finish_hook_cmd="hdfs upload ${profile_save_path}"
    actor_rollout_ref.actor.profiler.finish_hook_ranks=${profile_ranks}

    // same with rollout/ref/critic
    global_profiler.tool=torch
    global_profiler.steps=${profile_steps}
    global_profiler.save_path=${profile_save_path}
    global_profiler.relocate_results=True    // for rollout and nsys profiler, relocate results to profile_save_path

2. fix behaviors and allow comprehensive torch profiler.

Now torch profiler discrete mode creates:

When discrete=False, the actors' profiling results fused to actor-rollout-ref_train_step1_rank0-of-16_pid10256_20260814122731089.json

For compute log probs:

For Rollout:

For trainer:

For discrete=False:

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.

[wuxibin89]

Do we really need to profile rollout(vllm/sglang) in RL framework? Maybe prometheus metrics is a better option.