[fsdp, megatron] refactor: Refactor Fully Async Implementation via Engine Workers (verl-project#5269)

### What does this PR do?
Refactor the fully async worker into an engine worker based on the
separate ray trainer.
> 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

- [x] Search for similar PRs. Paste at least one query link here: ...
- [x] 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`
- 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.

```python
# 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.

- [x] Read the [Contribute
Guide](https://github.com/volcengine/verl/blob/main/CONTRIBUTING.md).
- [x] Apply [pre-commit
checks](https://github.com/volcengine/verl/blob/main/CONTRIBUTING.md#code-linting-and-formatting):
`pre-commit install && pre-commit run --all-files --show-diff-on-failure
--color=always`
- [ ] Add / Update [the
documentation](https://github.com/volcengine/verl/tree/main/docs).
- [ ] Add unit or end-to-end test(s) to [the CI
workflow](https://github.com/volcengine/verl/tree/main/.github/workflows)
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](https://verl-project.slack.com/archives/C091TCESWB1) in [the
`verl` Slack
workspace](https://join.slack.com/t/verl-project/shared_invite/zt-3855yhg8g-CTkqXu~hKojPCmo7k_yXTQ).
(If not accessible, please try [the Feishu group
(飞书群)](https://applink.larkoffice.com/client/chat/chatter/add_by_link?link_token=772jd4f1-cd91-441e-a820-498c6614126a).)
- [ ] 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`.

Author: ZLiao097

verl/experimental/fully_async_policy/fully_async_main.py

@@ -80,29 +80,42 @@ def create_role_worker_mapping(config):
         dict: Mapping from roles to worker classes
     """
     # Select worker class based on strategy
-    if config.actor_rollout_ref.actor.strategy in ["fsdp", "fsdp2"]:
-        assert config.actor_rollout_ref.actor.strategy == config.critic.strategy
-        from verl.experimental.fully_async_policy.fsdp_workers import (
-            CriticWorker,
+    use_legacy_worker_impl = config.trainer.get("use_legacy_worker_impl", "auto")
+    if use_legacy_worker_impl == "disable":
+        from verl.experimental.separation.engine_workers import (
             DetachActorWorker,
             DetachAsyncRolloutWorker,
+            TrainingWorker,
         )
         from verl.single_controller.ray import RayWorkerGroup
 
         ray_worker_group_cls = RayWorkerGroup
 
-    elif config.actor_rollout_ref.actor.strategy == "megatron":
-        assert config.critic.strategy == "megatron"
-        from verl.experimental.fully_async_policy.megatron_worker import (
-            CriticWorker,
-            DetachActorWorker,
-            DetachAsyncRolloutWorker,
-        )
-        from verl.single_controller.ray import RayWorkerGroup
-
-        ray_worker_group_cls = RayWorkerGroup
+        CriticWorker = TrainingWorker
     else:
-        raise NotImplementedError(f"Unsupported strategy: {config.actor_rollout_ref.actor.strategy}")
+        if config.actor_rollout_ref.actor.strategy in ["fsdp", "fsdp2"]:
+            assert config.actor_rollout_ref.actor.strategy == config.critic.strategy
+            from verl.experimental.fully_async_policy.fsdp_workers import (
+                CriticWorker,
+                DetachActorWorker,
+                DetachAsyncRolloutWorker,
+            )
+            from verl.single_controller.ray import RayWorkerGroup
+
+            ray_worker_group_cls = RayWorkerGroup
+
+        elif config.actor_rollout_ref.actor.strategy == "megatron":
+            assert config.critic.strategy == "megatron"
+            from verl.experimental.fully_async_policy.megatron_worker import (
+                CriticWorker,
+                DetachActorWorker,
+                DetachAsyncRolloutWorker,
+            )
+            from verl.single_controller.ray import RayWorkerGroup
+
+            ray_worker_group_cls = RayWorkerGroup
+        else:
+            raise NotImplementedError(f"Unsupported strategy: {config.actor_rollout_ref.actor.strategy}")
 
     train_role = Role.ActorRollout if config.async_training.use_trainer_do_validate else Role.Actor
     role_worker_mapping = {

verl/experimental/one_step_off_policy/main_ppo.py

@@ -82,29 +82,42 @@ def create_role_worker_mapping(config):
         dict: Mapping from roles to worker classes
     """
     # Select worker class based on strategy
-    if config.actor_rollout_ref.actor.strategy in ["fsdp", "fsdp2"]:
-        assert config.actor_rollout_ref.actor.strategy == config.critic.strategy
-        from verl.experimental.one_step_off_policy.fsdp_workers import (
-            CriticWorker,
+    use_legacy_worker_impl = config.trainer.get("use_legacy_worker_impl", "auto")
+    if use_legacy_worker_impl == "disable":
+        from verl.experimental.separation.engine_workers import (
             DetachActorWorker,
             DetachAsyncRolloutWorker,
+            TrainingWorker,
         )
         from verl.single_controller.ray import RayWorkerGroup
 
         ray_worker_group_cls = RayWorkerGroup
 
-    elif config.actor_rollout_ref.actor.strategy == "megatron":
-        assert config.critic.strategy == "megatron"
-        from verl.experimental.one_step_off_policy.megatron_workers import (
-            CriticWorker,
-            DetachActorWorker,
-            DetachAsyncRolloutWorker,
-        )
-        from verl.single_controller.ray import RayWorkerGroup
-
-        ray_worker_group_cls = RayWorkerGroup
+        CriticWorker = TrainingWorker
     else:
-        raise NotImplementedError(f"Unsupported strategy: {config.actor_rollout_ref.actor.strategy}")
+        if config.actor_rollout_ref.actor.strategy in ["fsdp", "fsdp2"]:
+            assert config.actor_rollout_ref.actor.strategy == config.critic.strategy
+            from verl.experimental.one_step_off_policy.fsdp_workers import (
+                CriticWorker,
+                DetachActorWorker,
+                DetachAsyncRolloutWorker,
+            )
+            from verl.single_controller.ray import RayWorkerGroup
+
+            ray_worker_group_cls = RayWorkerGroup
+
+        elif config.actor_rollout_ref.actor.strategy == "megatron":
+            assert config.critic.strategy == "megatron"
+            from verl.experimental.one_step_off_policy.megatron_workers import (
+                CriticWorker,
+                DetachActorWorker,
+                DetachAsyncRolloutWorker,
+            )
+            from verl.single_controller.ray import RayWorkerGroup
+
+            ray_worker_group_cls = RayWorkerGroup
+        else:
+            raise NotImplementedError(f"Unsupported strategy: {config.actor_rollout_ref.actor.strategy}")
 
     role_worker_mapping = {
         Role.Actor: ray.remote(DetachActorWorker),