Skip to content

Pull requests: NVIDIA/Megatron-LM

Author
Filter by author
Loading
Label
Filter by label
Loading
Use alt + click/return to exclude labels
or + click/return for logical OR
Projects
Filter by project
Loading
Milestones
Filter by milestone
Loading
Reviews
Assignee
Filter by who’s assigned
Assigned to nobody Loading
Sort

Pull requests list

[codex] Separate mFSDP v2 unit tests Run tests
#5640 opened Jul 3, 2026 by wujingyue Contributor Draft
[experimental] Add experimental/agent_compose placeholder with preview pointer docs-only documentation only (docs or docstrings)
#5639 opened Jul 3, 2026 by ISEEKYAN Contributor Loading…
[Dev] Fix FSDP backward hooks for TE-fused experts
#5636 opened Jul 3, 2026 by lhb8125 Contributor Loading…
Fix cu_seqlens broadcast shape for mbs > 1 with TP > 1 complexity: low Final Review PR is in the "final review" stage
#5635 opened Jul 3, 2026 by deepakn94 Contributor Loading…
4 tasks
Inference: Add super_v3 and ultra_v3 reasoning parsers
#5634 opened Jul 2, 2026 by sidsingh-nvidia Contributor Draft
6 tasks
chore: nightly sync main into dev (02_07_2026) Run functional tests Run MBridge tests Attach this for testing this PR against MBridge main
#5627 opened Jul 2, 2026 by svcnvidia-nemo-ci Draft
Ignore contributor DCO failures in Claude fix Approved All necessary approvals have been made complexity: low
#5625 opened Jul 2, 2026 by Phlip79 Member Loading…
1 task done
[dev] partial cuda graph support for dynamic cp
#5618 opened Jul 2, 2026 by HaochenYuan Contributor Loading…
6 tasks
Add sequence packing planning API
#5612 opened Jul 2, 2026 by ilml Contributor Draft
Callback system complexity: medium
#5610 opened Jul 1, 2026 by maanug-nv Contributor Loading…
6 tasks
return prefix cache hits data from the chat completions api
#5609 opened Jul 1, 2026 by sidsingh-nvidia Contributor Draft
1 of 6 tasks
Triton kernels - avoid recompilation and autotuning in prod complexity: low Final Review PR is in the "final review" stage
#5608 opened Jul 1, 2026 by sidsingh-nvidia Contributor Loading…
1 of 6 tasks
Inference: Add load aware routing to prefix caching.
#5607 opened Jul 1, 2026 by sidsingh-nvidia Contributor Draft
1 of 6 tasks
Add VLM support to DynamicInferenceEngine
#5606 opened Jul 1, 2026 by RPrenger Contributor Draft
6 tasks
ProTip! Type g i on any issue or pull request to go back to the issue listing page.