-
Notifications
You must be signed in to change notification settings - Fork 3.9k
Pull requests: NVIDIA/Megatron-LM
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
[fix] Release MTP assertion when EP overlap with PP=1
complexity: low
#4796
opened May 14, 2026 by
Wohox
Contributor
Loading…
5 tasks
[draft] Add DeepEP v2 flex dispatcher backend
#4793
opened May 14, 2026 by
Autumn1998
Contributor
Loading…
5 tasks
[Dev] Fix full CUDA graph capture reverted by pull main
#4792
opened May 14, 2026 by
Victarry
Contributor
Loading…
chore: nightly sync main into dev (13_05_2026)
Run functional tests
Run MBridge tests
Attach this for testing this PR against MBridge main
#4788
opened May 13, 2026 by
svcnvidia-nemo-ci
•
Draft
ci: Update workflow to use same commit for building docker image and running tests
Approved
All necessary approvals have been made
complexity: low
#4787
opened May 13, 2026 by
balasaajay
Contributor
Loading…
5 tasks
Add TEFusedDenseMLP for Dense+Grouped GEMM fusion on SM100+ (#4318)
complexity: medium
#4786
opened May 13, 2026 by
sraman-rgb
Loading…
5 tasks done
Add --freeze-base-for-mtp to train MTP heads on frozen quantized base
#4785
opened May 13, 2026 by
yeyu-nvidia
Contributor
•
Draft
3 tasks
Add LLM PP>1 support for colocated MIMO training (NMFW-19)
#4784
opened May 13, 2026 by
yashaswikarnati
Contributor
•
Draft
3 of 4 tasks
chore: Restore golden values for functional tests
Run functional tests
#4783
opened May 13, 2026 by
balasaajay
Contributor
•
Draft
5 tasks
Thread custom process groups through MoE grad finalization
complexity: low
#4782
opened May 13, 2026 by
yashaswikarnati
Contributor
Loading…
5 tasks
Pass explicit process groups to hybrid logging
Approved
All necessary approvals have been made
complexity: low
#4781
opened May 13, 2026 by
yashaswikarnati
Contributor
Loading…
5 tasks
Tokenizers updates
complexity: low
Expert Review
[deprecated] Apply this label to indicate that your PR is ready for expert review.
#4780
opened May 13, 2026 by
dimapihtar
Contributor
Loading…
5 tasks
Add dev-feature preservation gate and change schedule
complexity: medium
#4773
opened May 13, 2026 by
Phlip79
Member
Loading…
chore: Bump TE to latest 2.14
Run functional tests
#4772
opened May 13, 2026 by
chtruong814
Contributor
Loading…
5 tasks
Route non-Muon params through DistributedOptimizer
complexity: medium
#4771
opened May 13, 2026 by
deepakn94
Contributor
Loading…
1 of 3 tasks
Previous Next
ProTip!
Mix and match filters to narrow down what you’re looking for.