[data] Filter Overlonged Prompts in tool_agent_loop

[HwCARI]

What does this PR do?

Fixes #2069.

filter_overlong_prompts was silently skipped in the multiturn setting in verl/verl/experimental/agent_loop/tool_agent_loop.py, causing prompts exceeding data.max_prompt_length to pass through unfiltered. This resulted in tensor size mismatches downstream when batching sequences:

RuntimeError: Sizes of tensors must match except in dimension 0.
Expected size 2048 but got size 2106 for tensor number 2 in the list.

This PR ensures the overlong prompt filter is applied consistently regardless of whether multiturn tool use is enabled.

Fix

Apply filter_overlong_prompts in verl/verl/experimental/agent_loop/tool_agent_loop.py.

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.

[gemini-code-assist]

Code Review

This pull request introduces a length check for prompt IDs in the agent loop to prevent runtime errors from overlong prompts by terminating the state early. Feedback suggests that this approach may result in blank samples being passed to the trainer, recommending the use of specific metric flags or default rewards for filtered samples and the inclusion of request IDs in logs for improved observability.

[gemini-code-assist]

While this check correctly prevents the RuntimeError during batching by terminating early for overlong prompts, returning AgentState.TERMINATED at this stage leaves agent_data.prompt_ids as an empty list. This results in a "blank" sample (all padding) being passed to the trainer.

To improve observability and ensure the trainer handles the filtered sample correctly, consider setting a default reward or a specific metric flag indicating the sample was filtered due to length constraints. Additionally, including the request_id in the warning log would significantly aid in debugging specific failed samples in large-scale runs.