Merge main into dev#3865

Open

ilml wants to merge 144 commits intoNVIDIA:devfrom

ilml:merge-main-into-dev

Commits on Feb 26, 2026

Commits on Feb 28, 2026

Update copy-pr-bot.yaml [skip ci]
github-actions[bot]
committed

Commits on Mar 2, 2026

Commits on Mar 4, 2026

Commits on Mar 5, 2026

Commits on Mar 6, 2026

Commits on Mar 7, 2026

Update copy-pr-bot.yaml [skip ci]
github-actions[bot]
committed

Commits on Mar 9, 2026

Commits on Mar 10, 2026

Commits on Mar 11, 2026

Update copy-pr-bot.yaml [skip ci]
github-actions[bot]
committed
feat(checkpoint): zero-copy storage sharing in CheckpointWithoutOutput (#3649)

Victarry
and
claude
authored
Fuse MLA DOWN projection GEMMs (#3039)

cjld
and
yaox12
authored
fix: skip FSDP DTensor boundary validation under fake process group (#3686)

Victarry
and
claude
authored
[main] fix(moe): fix the bug where gate was not sliced when kv_head < tp_size. (#3575)

yuzhongw-nvidia
and
xiaotaoliu
authored
fix(offload): reset activation offload manager after eval as well as … (#3739)

rapatel
and
lhb8125
authored
chore: rotate oncall schedule
github-actions[bot]
committed
Improve error logging when invalid number of tokens is requested. (#3680)
yobibyte
authored
Add NVIDIA-Nemotron-3-Super-120B-A12B-BF16 to ModelOpt examples (#3805)
jenchen13
authored
build: Bump TE2.13 (#3800)
ko3n1g
authored
Ensure dummy_forward does not attempt to run cudagraphs (#3789)

jalbericiola
and
mathemakitten
authored
Add speculative decoding support with MTP layers (#3594)

authored
Shanmugamr1992/megatron inference ultra (#3784)

authored

Commits on Mar 12, 2026

Commits on Mar 13, 2026