[megatron, worker] fix: use extract_multi_modal_inputs method for handling multi_modal_inputs

[HollowMan6]

Follow up for #3553

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.

Without those changes in #3315, the error when we train the mixture modal dataset will remain unresolved, so it would be a good idea to add them back.

File "verl/workers/actor/megatron_actor.py", line 639, in update_policy
  metric_micro_batch = self.forward_backward_batch(
                       ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "verl/workers/actor/megatron_actor.py", line 587, in forward_backward_batch
  losses_reduced = forward_backward_func(
                   ^^^^^^^^^^^^^^^^^^^^^^
File "megatron/core/pipeline_parallel/schedules.py", line 595, in forward_backward_no_pipelining
  output_tensor, num_tokens = forward_step(
                              ^^^^^^^^^^^^^
File "megatron/core/pipeline_parallel/schedules.py", line 402, in forward_step
  output_tensor, loss_func = forward_step_func(data_iterator, model)
                             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "verl/workers/actor/megatron_actor.py", line 497, in forward_step
  multi_modal_inputs[key] = torch.cat(
                            ^^^^^^^^^^
RuntimeError: torch.cat(): expected a non-empty list of Tensors

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 refactors the handling of multi-modal inputs in megatron_actor.py by introducing and using the extract_multi_modal_inputs utility function. This change effectively fixes a RuntimeError that occurred with mixed-modality datasets. The new approach is cleaner, more robust against edge cases like empty input lists, and centralizes the data processing logic. While the fix is correct, I've suggested moving the local import to the top of the file to adhere to best practices and improve code maintainability, especially in a performance-sensitive area.