Support CP (no sequence packing)#4086
Draft
ankisinha-nvidia wants to merge 2 commits intoNVIDIA:mainfrom
Draft
Conversation
Align the RL CP>1 path with the working branch and add launcher/inference fixes so interactive smoke runs can collect rollouts, surface inference progress, and complete short training iterations reliably.
Remove the temporary inference request logging and restore the original host behavior while keeping the core RL context-parallel changes intact.
Contributor
|
This PR has been automatically converted to draft because all PRs must start as drafts. When you are ready for review, click Ready for Review to begin the review process. This will:
See the contribution guide for more details. |
wdykas
reviewed
Apr 1, 2026
|
|
||
| def cp_gather_logprobs(local_logprobs): | ||
| """All-gather and un-zigzag CP-local logprobs back to [B, S-1].""" | ||
| cp_size = mpu.get_context_parallel_world_size() |
Contributor
There was a problem hiding this comment.
please dont grab from global groups. you can see grabbing groups here https://github.com/ankisinha-nvidia/Megatron-LM/blob/24d79d247a683ef4648039d39c657c7257d19871/megatron/rl/rl_utils.py#L527 how it can be done
wdykas
reviewed
Apr 1, 2026
|
|
||
| gathered = [torch.empty_like(local_logprobs) for _ in range(cp_size)] | ||
| torch.distributed.all_gather( | ||
| gathered, local_logprobs.contiguous(), group=mpu.get_context_parallel_group() |
wdykas
reviewed
Apr 1, 2026
| use_cp_nonpacked_path = ( | ||
| labels is not None | ||
| and not sequence_packing | ||
| and mpu.get_context_parallel_world_size() > 1 |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What does this PR do ?
Align the RL CP>1 path with the working branch and add launcher/inference fixes so interactive smoke runs can collect rollouts, surface inference progress, and complete short training iterations reliably.
Contribution process
Pre-checks
Code review
Feel free to message or comment the @mcore-oncall to help accelerate your merge into main. The less complex your PR is, the faster it will be approved and merged!
All PRs start as draft. If you open a non-draft PR, it will be automatically converted to draft.
Step 1: Mark PR as "Ready for Review"
.github/CODEOWNERS.Final Review might get declined if these requirements are not fulfilled.
Step 2: Final Review
For PRs that change
megatron/core, once all expert reviewers have approved, theFinal Reviewlabel is applied automatically and final reviewers are assigned.For PRs outside
megatron/core, this step is skipped.Step 3: Approved
Once all required reviewers have approved, the
Approvedlabel is applied automatically.Merge
Any member of mcore-engineers will be able to merge your PR.
For MRs into `dev` branch
The proposed review process for `dev` branch is under active discussion.MRs are mergable after one approval by either
eharper@nvidia.comorzijiey@nvidia.com.