[megatron] Add TIS support to megatron backend

[sharonyu-115]

What does this PR do?

Add the TIS support from #2953 to megatron actor

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 adds support for Truncated Importance Sampling (TIS) to the Megatron backend. The changes correctly plumb the rollout_log_probs data required for TIS from the data loader to the PPO loss calculation. My review found a critical issue where the code could crash if data.batch is None when checking for the presence of rollout_log_probs. I've provided a suggestion to fix this by adding a None check.

[gemini-code-assist]

The code data.batch.keys() could raise an AttributeError if data.batch is None, as the DataProto class definition allows batch to be None. To prevent a potential crash, you should first check if data.batch is not None before accessing its keys. Using the in operator directly on the TensorDict is also more idiomatic.

Suggested change

	assert "rollout_log_probs" in data.batch.keys(), (
	"Truncated Importance Sampling (TIS) requires to configure "
	"`actor_rollout_ref.rollout.calculate_log_probs=True` "
	"and is not currently supported in Server mode (agent loop)."
	)
	assert data.batch is not None and "rollout_log_probs" in data.batch, (
	"Truncated Importance Sampling (TIS) requires to configure "
	"`actor_rollout_ref.rollout.calculate_log_probs=True` "
	"and is not currently supported in Server mode (agent loop)."
	)