Skip to content

Pull requests: NVIDIA/Megatron-LM

New pull request New
734 Open 3,760 Closed
734 Open 3,760 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

chore(ci): bump community workflow to v1.8.8 Approved All necessary approvals have been made complexity: low
#6322 opened Aug 6, 2026 by ko3n1g Contributor Loading…
1
Exit non-zero when --exit-on-missing-checkpoint triggers
#6321 opened Aug 6, 2026 by shanmugamr1992 Contributor • Draft
2 tasks
1
[Dev] Skip broadcast for empty cu_seqlens tensors
#6320 opened Aug 6, 2026 by lhb8125 Contributor • Draft
1
[Dev] Preserve explicit NCCL NVLS configuration
#6319 opened Aug 6, 2026 by lhb8125 Contributor • Draft
1
[Dev] Scope FSDP symmetric UBR to dense inner gathers
#6318 opened Aug 6, 2026 by lhb8125 Contributor • Draft
1
[Dev] Make FSDP manual UBR registration deterministic
#6317 opened Aug 6, 2026 by lhb8125 Contributor • Draft
1
[Dev] Generalize FSDP persistent communication buffers
#6316 opened Aug 6, 2026 by lhb8125 Contributor • Draft
1
feat(multimodal): migrate Qwen3.5-VL to HybridModel
#6315 opened Aug 6, 2026 by BestJuly Contributor • Draft
1
Refactor hybrid layers to use per-layer configs
#6313 opened Aug 6, 2026 by Phlip79 Member • Draft
6 tasks
2
Add Nemotron 3.5 Lightning GB200 nightly test complexity: low
#6312 opened Aug 6, 2026 by Phlip79 Member Loading…
1 task done
1
[Dev] Recompute EP A2A overlap activations per layer instead of per model chunk
#6311 opened Aug 6, 2026 by Wohox Contributor Loading…
1
Add HybridModel Python recipe API and launcher support
#6310 opened Aug 6, 2026 by Phlip79 Member • Draft
3 of 5 tasks
1
Support reentrant MFSDP activation recomputation
#6309 opened Aug 6, 2026 by wujingyue Contributor • Draft
1
[DEV] Support energon dataloader in Qwen35 community-request
#6308 opened Aug 6, 2026 by VVsssssk Loading…
6 tasks
1
Propagate rope_scaling_factor from args to GPT model construction community-request
#6306 opened Aug 5, 2026 by Alvorecer721 • Draft
3 of 6 tasks
1
1
tests: Update cerno.yml with linear issue status Approved All necessary approvals have been made complexity: low
#6304 opened Aug 5, 2026 by balasaajay Contributor Loading…
1 of 6 tasks
2
Require an explicit process-group collection in LanguageModule
#6303 opened Aug 5, 2026 by Connor-XY Contributor • Draft
5 of 6 tasks
2
Reject unsupported heterogeneous hybrid inference
#6301 opened Aug 5, 2026 by Phlip79 Member • Draft
3
Add Nemotron direct hybrid architecture examples
#6300 opened Aug 5, 2026 by Phlip79 Member • Draft
4
Integrate direct hybrid architectures with training
#6299 opened Aug 5, 2026 by Phlip79 Member • Draft
5
Support direct hybrid MTP layer specs
#6298 opened Aug 5, 2026 by Phlip79 Member • Draft
5
Execute direct hybrid layers across PP and VPP
#6297 opened Aug 5, 2026 by Phlip79 Member • Draft
5
Add direct hybrid architecture descriptors
#6295 opened Aug 5, 2026 by Phlip79 Member • Draft
33
Remove global process-group reads from megatron/core (207 -> 155)
#6293 opened Aug 5, 2026 by Connor-XY Contributor • Draft
5 of 6 tasks
2
[DEV] fix(multimodal_dev): derive mock samples from the sample index complexity: low
#6287 opened Aug 5, 2026 by xuwchen Contributor Loading…
6 tasks
1
ProTip! Add no:assignee to see everything that’s not assigned.