microsoft/DeepSpeed

Pull Requests Commits

Merge branch 'master' into olruwase/fast_persist

PKUWZP committed 8 hours ago

Verified 223f06fa

anchor transformers version (#7316)

stas00 committed 15 hours ago

Verified b66c8107

Update gaudi2 nightly,ci to latest 1.21.0 build (#7313)

raza-sikander committed 18 hours ago

Verified ec6b254d

`tests/conftest.py`: automatically add local deepspeed repo when running tests (#7317)

stas00 committed 22 hours ago

Verified e5afb887

CI: prefer bf16 over fp16 (#7304)

stas00 committed 1 day ago

Verified b4cc079e

Fix: Update grad norm calculation for CPU offload (#7302)

therealnaveenkamal committed 2 days ago

Verified b9af5d8d

Fix the GPU memory usage of ZeRO-Offload (only update stage_1_and_2.py) (#7309)

arminzhu committed 2 days ago

Verified 17c8be07

Fix AutoTP gathering replaced layer params when bias is not None (#7257)

HollowMan6 committed 4 days ago

Verified b666844f

Merge branch 'master' into olruwase/fast_persist

sfc-gh-truwase committed 5 days ago

Verified e0a9e5a0

Update COMMITTERS.md (#7305)

PKUWZP committed 6 days ago

Verified d4032ec7

Update next version in version.txt after 0.16.9 release. (#7306)

loadams committed 6 days ago

Verified e3bd16fd

[XPU] Support XCCL on deepspeed side (#7299)

ys950902 committed 7 days ago

Verified bdba8231

Fix extra_repr_str when weight is None / in zero-3 (#7254)

HollowMan6 committed 7 days ago

Verified 0e3209a1

disable license check until the new license situation has been sorted… (#7301)

stas00 committed 7 days ago

Verified e290bf58

Add qwen3moe meta loading for AutoTP (#7297)

ranzhejiang committed 9 days ago

Verified 41fceade

Enable ZeRO set/get APIs for NVMe offload (#7046)

tjruwase committed 9 days ago

Verified 0e741714

Modernize system executable detection across components (#7290)

emmanuel-ferdman committed 10 days ago

Verified b048cc2b

Avoid graph break by removing another redundant requires grad false (#7263)

deepcharm committed 10 days ago

Verified d0ef6501

Add qwen3 meta loading for AutoTP (#7293)

delock committed 10 days ago

Verified 80bc7b76

Update patch version after 0.16.8 release (#7296)

loadams committed 10 days ago

Verified 88a1b5c0

rollback #6726 (#7258)

delock committed 10 days ago

Verified f4595025

Temporarily skip AIO tests due to an issue with runners (#7288)

loadams committed 10 days ago

Verified d46947db

Fix issues XPU tests hit with extra-index-url (#7291)

loadams committed 12 days ago

Verified 930ab46e

[XPU] update xpu-max1100 CI workflow to torch 2.7 (#7284)

Liangliang-Ma committed 14 days ago

Verified 5a4e7a08

Merge branch 'master' into olruwase/fast_persist

loadams committed 14 days ago

Verified e55d1ce6

Fix fp8 gemm (#7265)

RezaYazdaniAminabadi committed 20 days ago

Verified 069ec31c

add `Makefile` to ease maintenance (#7267)

stas00 committed 22 days ago

Verified e1ba9e61

Fix Ubuntu version

sfc-gh-truwase committed 27 days ago

0a587af8

sfc-gh-truwase committed 29 days ago

3ddea097

Merge branch 'master' into olruwase/fast_persist

sfc-gh-truwase committed 30 days ago

Verified 00e78db6

Older