Skip to content

Pull requests: NVIDIA/Megatron-LM

New pull request New
802 Open 3,893 Closed
802 Open 3,893 Closed
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
Filter by reviews
No reviews Review required Approved review Changes requested
Assignee
Filter by who’s assigned
Assigned to nobody Loading
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Pull requests list

Remove leading NaN iteration times from golden values complexity: low
#6539 opened Aug 14, 2026 by balasaajay Contributor Loading…
1 of 6 tasks
2
Add flash-linear-attention GDP kernels complexity: high
#6538 opened Aug 14, 2026 by santhnm2 Contributor Loading…
6 tasks
2
docs: Update docs version for 0.19.0
#6537 opened Aug 14, 2026 by chtruong814 Contributor Loading…
6 tasks
docs: Add 0.19.0 to version picker docs-only documentation only (docs or docstrings)
#6536 opened Aug 14, 2026 by chtruong814 Contributor Loading…
6 tasks
2
[GTP] Symmetric Memory Registration for the GTP/EGTP ReduceScatter path
#6535 opened Aug 14, 2026 by prajwal1210 Contributor • Draft
6 tasks done
14
debug: do not merge pr-4547
#6534 opened Aug 14, 2026 by chtruong814 Contributor • Draft
6 tasks
3
debug: do not merge pr-3720
#6533 opened Aug 14, 2026 by chtruong814 Contributor • Draft
6 tasks
3
debug: do not merge pr-4087
#6532 opened Aug 14, 2026 by chtruong814 Contributor • Draft
6 tasks
3
debug: do not merge pr-5442
#6531 opened Aug 14, 2026 by chtruong814 Contributor • Draft
6 tasks
3
debug: do not merge pr-5726
#6530 opened Aug 14, 2026 by chtruong814 Contributor • Draft
6 tasks
3
debug: do not merge pr-5972
#6529 opened Aug 14, 2026 by chtruong814 Contributor • Draft
6 tasks
4
Fix missing golden values updates complexity: medium Run functional tests
#6528 opened Aug 14, 2026 by janEbert Contributor Loading…
1 task done
4
Fix swiglu test H100 golden values
#6527 opened Aug 13, 2026 by santhnm2 Contributor • Draft
6 tasks
1
test: add GDP hybrid dynamic-inference functional test (GB200)
#6526 opened Aug 13, 2026 by shanmugamr1992 Contributor • Draft
2 tasks
1
fix: ensure GTP params are ready before CUDA graph replay complexity: low nemotron
#6525 opened Aug 13, 2026 by JF-D Contributor Loading…
4 of 6 tasks
1
@JF-D
3
[NCCL EP] Add MXFP8 dispatch support complexity: low
#6524 opened Aug 13, 2026 by YangFei1990 Contributor Loading…
1 of 6 tasks
2
chore: nightly sync main into dev (13_08_2026) Run functional tests Run MBridge tests Attach this for testing this PR against MBridge main
#6522 opened Aug 13, 2026 by svcnvidia-nemo-ci Contributor • Draft
4
Batch-invariance generation with vLLM fused-MoE backend at improved performance community-request
#6521 opened Aug 13, 2026 by utkarsh530 Contributor • Draft
6 tasks done
23
fix(transformer): AUT-1497 support dual RoPE full recompute Run functional tests
#6519 opened Aug 13, 2026 by svcnemo-autobot Collaborator Loading…
2
fix: AUT-1503 abort when SSO list download fails
#6518 opened Aug 13, 2026 by svcnemo-autobot Collaborator • Draft
3
fix: AUT-1504 abort when SSO list download fails on dev
#6517 opened Aug 13, 2026 by svcnemo-autobot Collaborator Loading…
2
fix: AUT-1501 abort when SSO list download fails Approved All necessary approvals have been made complexity: low
#6516 opened Aug 13, 2026 by svcnemo-autobot Collaborator Loading…
19
[main] moe(perf): Refactor CP layout organization for Qwen3.5-style hybrid attention model complexity: high
#6515 opened Aug 13, 2026 by yuzhongw-nvidia Contributor Loading…
6 tasks
6
add gigatoken support for hf tokenizers
#6514 opened Aug 13, 2026 by dimapihtar Contributor • Draft
1 of 6 tasks
1
Ygalron/mtp hsm rank aware
#6512 opened Aug 13, 2026 by YanivDorGalron • Draft
6 tasks
2
ProTip! Add no:assignee to see everything that’s not assigned.