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

Support GEMM + Swiglu fused MLP complexity: high
#3971 opened Mar 20, 2026 by ksivaman Loading…
5 tasks
Enable cpu_offloading with Full iteration CUDA graph Final Review PR is in the "final review" stage
#3969 opened Mar 20, 2026 by vasunvidia Loading…
5 tasks
ci: Split out inference tests
#3966 opened Mar 20, 2026 by ko3n1g Draft
5 tasks
Core 0.16
Fix bug with non-partial rollouts complexity: low
#3964 opened Mar 20, 2026 by tdene Loading…
5 tasks
Core 0.16
Fix: qk-clip causes a tensor shape error when tensor-model-parallel-size > 1 community-request Final Review PR is in the "final review" stage
#3963 opened Mar 20, 2026 by xzy-xzy Loading…
5 tasks
Small quality-of-life improvements in megatron/training Approved All necessary approvals have been made complexity: low
#3957 opened Mar 19, 2026 by deepakn94 Loading… Core 0.16
Add moe loss normalization for RL SFT complexity: low
#3956 opened Mar 19, 2026 by pthombre Loading…
5 tasks
Fix completions endpoint Approved All necessary approvals have been made complexity: low Run functional tests
#3940 opened Mar 19, 2026 by santhnm2 Loading…
5 tasks
Core 0.16
Fix key error in layer_wise_optimizer.sharded_state_dict community-request Final Review PR is in the "final review" stage
#3939 opened Mar 19, 2026 by chenchun Loading…
1 of 5 tasks
Fix FSDP checkpoint conversion and loading for Qwen3.5-VL community-request Final Review PR is in the "final review" stage
#3936 opened Mar 19, 2026 by DAISY-gh Loading…
5 tasks
Make text generation server hostname configurable complexity: low Final Review PR is in the "final review" stage
#3935 opened Mar 18, 2026 by santhnm2 Loading…
5 tasks
Refit optimization Approved All necessary approvals have been made complexity: medium
#3933 opened Mar 18, 2026 by wdykas Loading…
5 tasks
Core 0.16
Improve load balancing behavior for prefix cache-aware routing complexity: low Final Review PR is in the "final review" stage
#3930 opened Mar 18, 2026 by santhnm2 Loading…
5 tasks
Core 0.16
Add --muon-coefficient-type argument for Muon optimizer complexity: low Final Review PR is in the "final review" stage
#3927 opened Mar 18, 2026 by mchrzanowski Loading…
ProTip! What’s not been updated in a month: updated:<2026-02-20.