[trainer] fix: resolve dataset config in agent loop (verl-project#5034)

### What does this PR do?

line 516: `dataset_config=self.config.data`


https://github.com/volcengine/verl/blob/772c224935d136b05f5f2d10ddaa6d3b7a41403c/verl/experimental/agent_loop/agent_loop.py#L509-L517

self.config.data may have unresolved arguments like `'tool_config_path':
'${oc.select:actor_rollout_ref.rollout.multi_turn.tool_config_path,
null}'`, which cannot be instantiated. This will cause error

```
*** omegaconf.errors.InterpolationResolutionError: TypeError raised while resolving interpolation: Index 'actor_rollout_ref' (str) is not an int
```

### Solution:

Wrap it in DictConfigWrap, so that it will not be initialized in hydra
resolve config.

### 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`
- 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.

- [ ] Read the [Contribute
Guide](https://github.com/volcengine/verl/blob/main/CONTRIBUTING.md).
- [ ] 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: yyDing1

verl/experimental/agent_loop/agent_loop.py

@@ -200,7 +200,7 @@ def __init__(
         tokenizer: AutoTokenizer,
         processor: AutoProcessor,
         dataset_cls: type[RLHFDataset],
-        dataset_config: DictConfig,
+        dataset_config: DictConfigWrap,
         **kwargs,
     ):
         """Initialize agent loop, each sample will have its own loop instance.
@@ -211,15 +211,15 @@ def __init__(
             tokenizer (AutoTokenizer): Tokenizer for tokenize messages.
             processor (AutoProcessor): Processor for process messages.
             dataset_cls (type[Dataset]): Dataset class for creating dataset, Defaults to RLHFDataset.
-            dataset_config (DictConfig): Dataset config.
+            dataset_config (DictConfigWrap): Dataset config.
         """
         self.config = trainer_config.config
         self.server_manager = server_manager
         self.tokenizer = tokenizer
         self.processor = processor
         self.dataset_cls = dataset_cls
-        self.dataset_config = dataset_config
-        self.apply_chat_template_kwargs = dataset_config.get("apply_chat_template_kwargs", {})
+        self.dataset_config = dataset_config.config
+        self.apply_chat_template_kwargs = self.dataset_config.get("apply_chat_template_kwargs", {})
         self.system_prompt = initialize_system_prompt(self.tokenizer, **self.apply_chat_template_kwargs)
         self.loop = get_event_loop()
 
@@ -513,7 +513,7 @@ async def _run_agent_loop(
                 tokenizer=self.tokenizer,
                 processor=self.processor,
                 dataset_cls=self.dataset_cls,
-                dataset_config=self.config.data,
+                dataset_config=DictConfigWrap(self.config.data),
             )
             output: AgentLoopOutput = await agent_loop.run(sampling_params, **kwargs)
             return await self._agent_loop_postprocess(output, **kwargs)