[megatron] fix: Bump Megatron-Bridge commit for PEFT recompute

[HollowMan6]

What does this PR do?

Resolves:

• [megatron] feat: Integrate Megatron-Bridge and support LoRA/PEFT #4063 (comment)
• grad_norm always equals 0 when training with megatron + lora #4514

Related to: #4566

NVIDIA-NeMo/Megatron-Bridge@953aabf

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, 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.

# 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 updates the recommended Megatron-Bridge commit hash in the documentation and example scripts to incorporate a fix for PEFT recompute. The changes are consistent across all modified files and correctly point to the new commit. The update is straightforward and appears correct.

[Copilot]

Pull request overview

This PR updates the Megatron-Bridge dependency to a newer commit (953aabf) that includes fixes for PEFT (Parameter-Efficient Fine-Tuning) recompute functionality. The update resolves issues related to LoRA training with gradient checkpointing.

Key Changes:

• Updated Megatron-Bridge commit reference from a489bed to 953aabf across all documentation, example scripts, and CI workflows

Reviewed changes

Copilot reviewed 5 out of 5 changed files in this pull request and generated no comments.

Show a summary per file

File	Description
examples/grpo_trainer/run_qwen3moe-30b_megatron_lora.sh	Updated recommended Megatron-Bridge commit hash in comment
examples/grpo_trainer/run_qwen2-7b_math_megatron_lora.sh	Updated recommended Megatron-Bridge commit hash in comment
docs/advance/ppo_lora.rst	Updated Megatron-Bridge commit reference link in documentation
.github/workflows/e2e_ppo_trainer_megatron_vllm_2.yml	Updated Megatron-Bridge installation command to use new commit hash
.github/workflows/e2e_ppo_trainer_megatron_vllm.yml	Updated Megatron-Bridge installation command to use new commit hash

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.