microsoft/DeepSpeed

Pull Requests Commits

Merge branch 'master' into sfc-gh-truwase/fp32_grads

sfc-gh-truwase committed 31 minutes ago

Verified 54cf4f23

sfc-gh-truwase committed 32 minutes ago

ae20d9d2

Merge branch 'sfc-gh-truwase/fp32_grads' of https://github.com/deepspeedai/DeepSpeed into sfc-gh-truwase/fp32_grads

sfc-gh-truwase committed 2 hours ago

52b59b6c

sfc-gh-truwase committed 2 hours ago

07a84412

Fix DeepCompile ZeRO-1 grad target lifetime (#8036)

tohtana committed 2 hours ago

Verified 11eeb7cd

Merge branch 'master' into sfc-gh-truwase/fp32_grads

sfc-gh-truwase committed 2 hours ago

Verified b1e61c56

Consistent fp32 grads flow

sfc-gh-truwase committed 2 hours ago

9acf2392

Avoid HF Hub access in CPU unit test setup (#8053)

tohtana committed 1 day ago

Verified 92cfacb9

Add test coverage for Muon muon_lr/adam_lr overrides (#8047)

sowndappan5 committed 3 days ago

Verified 3dc98deb

Fix ZenFlow ZeRO-3 selective optimizer crash with parameter offload on nvme (#8042)

Antlera committed 5 days ago

Verified 28a196f7

Fix DeepCompile ZeRO-3 release parameter lifetime (#8032)

tohtana committed 7 days ago

Verified 3e486feb

Bump version after 0.19.1 release (#8040)

tohtana committed 10 days ago

Verified 334230ff

Version fix to unblock pypi (#8039)

sfc-gh-truwase committed 10 days ago

Verified d5d118c2

Fix DeepCompile all-gather scheduler candidate selection (#8033)

tohtana committed 10 days ago

Verified 819af0e5

Simplify module_inject.transpose (#8028)

xbcReal committed 11 days ago

Verified 1afa3a53

feat(zero): enable torch.func transforms on engine for ZeRO 0/1/2 (#8026)

roycho96 committed 12 days ago

Verified 4370aa50

Add engine.coalesce_grad_reduction() for ZeRO 1/2/3 multi-backward (#7992)

roycho96 committed 12 days ago

Verified 60b242af

Bump version (#8030)

sfc-gh-truwase committed 12 days ago

Verified a0b98d34

Keep required CI checks visible for ignored paths (#8019)

tohtana committed 13 days ago

Verified 77a249f1

Fix full CI test isolation for ZeRO chmod and NVMe quantization tests (#8008)

tohtana committed 13 days ago

Verified 4570c508

Fix PR-target workflow concurrency groups (#8017)

tohtana committed 13 days ago

Verified b7aef4dc

Run AutoSP compile tests sequentially (#8020)

tohtana committed 13 days ago

Verified 2c8a007b

fix(fp16): filter requires_grad in FP16 optimizer flat buffer init (#8029)

avicooper1 committed 13 days ago

Verified 01e3a28b

Fix DeepCompile AOT kwargs patching for PyTorch >= v2.11 (#8024)

tohtana committed 15 days ago

Verified 510ebe58

Support flash-attn 2.7.0 in FPDT attention (#8022)

xbcReal committed 16 days ago

Verified 45429221

fix(zero): enable vmap on LinearFunctionForZeroStage3 (#8023)

roycho96 committed 16 days ago

Verified ae576f83

Sort and dedupe -gencode flags emitted by op_builder.builder (#8021)

adityasingh2400 committed 17 days ago

Verified 43166463

fix: add setup_context for torch.func compatibility (#7916)

roycho96 committed 19 days ago

Verified 67b9e212

fix gemma4 num attention head bugs (from #7975) (#7990)

delock committed 19 days ago

Verified bd71eed6

[Blog] Muon Optimizer Support in DeepSpeed (#7962)

delock committed 19 days ago

Verified de473091

Older