microsoft/DeepSpeed

Pull Requests Commits

test(zero): add training step coverage for accelerator flatten path

delock committed 44 days ago

7f8aff51

remove unnecessary detach comment

delock committed 44 days ago

0bfe45e3

fix(zero): detach flat buffer to prevent autograd inplace error on CPU accelerator

delock committed 44 days ago

b70e8629

AGENTS.md: Add pre-commit command to existing CI requirements line (#7930)

delock committed 48 days ago

Verified abb88ceb

fix(superoffload) preserve multi-group updates with shared cpu buffers (#7905) (#7906)

xylian86 committed 50 days ago

Verified 729df6ca

Add news item for ASPLOS 2026 Best Paper Award (#7923)

PKUWZP committed 52 days ago

Verified 62c3e6d8

Extending Muon Optimizer Support for ZeRO Stage 3 (#7919)

PKUWZP committed 52 days ago

Verified 956ec6fb

Fix backward compatibility of torch.amp.custom_fwd for PyTorch < 2.4 (#7920)

tohtana committed 52 days ago

Verified 138f20d9

fix: handle non-existent path in is_nfs_path for Triton autotune cache (#7921)

Krishnachaitanyakc committed 53 days ago

Verified f887b982

Add HuggingFace tp_plan support for AutoTP (#7901)

delock committed 53 days ago

Verified a240c4da

Don't detect local GPU if `$DS_IGNORE_CUDA_DETECTION` is set (#7896)

Flamefire committed 54 days ago

Verified 26c954ff

fix: remove unnecessary shell=True in ROCm GPU architecture detection (#7915)

instantraaamen committed 54 days ago

Verified 5ce3abbf

Add Feature Universal Checkpoint for AutoTP (#7908)

nathon-lee committed 54 days ago

Verified f2bb1ec6

Respect `$TRITON_HOME` (#7907)

Flamefire committed 55 days ago

Verified 38bd11a7

Update version (#7903)

loadams committed 65 days ago

Verified 5f7b6870

fix(zero3): use current_stream() instead of default_stream() for grad… (#7898)

michaelroyzen committed 65 days ago

Verified be60451f

Add AGENTS.md and CLAUDE.md with project rules for AI coding agents (#7902)

delock committed 65 days ago

Verified 49a82a0d

fix: Validate fp16.loss_scale is finite and non-negative (#7889)

nathon-lee committed 65 days ago

Verified 63eeb114

Fix Evoformer's multi-arch dispatch root cause (#7881)

tohtana committed 65 days ago

Verified 784cc26e

Fix async_io ops building error on Huawei Ascend NPU (#7894)

huangyifan0610 committed 66 days ago

Verified f88d0f8d

double reduction user-friendly error (#7895)

stas00 committed 68 days ago

Verified b6346bf0

[Bloom] Fix hangs of bloom test (#7890)

k-artem committed 68 days ago

Verified d9a4aadb

Suppress see_memory_usage logs (#7891)

sfc-gh-truwase committed 70 days ago

Verified 285cae30

Fix hook count performance regression from v0.18.5 (#7886)

tohtana committed 73 days ago

Verified 6c59d544

Add document section explaining autocast nesting (#7883)

tohtana committed 74 days ago

Verified 4dba1e20

Remove amp() from abstract accelerator (#7879)

delock committed 76 days ago

Verified a41a96b1

XPU use stock pytorch instead of Intel Extension for PyTorch (#7877)

delock committed 76 days ago

Verified d8e15da4

Fix import deepspeed crash on PyTorch v2.3 + Python 3.12 (#7875)

tohtana committed 76 days ago

Verified 04d69cc6

Fix no-grad grad-fn lookup in ZeRO hook counting on PyTorch 2.3 (#7830) (#7841)

tohtana committed 76 days ago

Verified bffaf457

Fix: only add parameter with grads to parameter group (#7869)

delock committed 76 days ago

Verified 116dbe28

Older