microsoft/DeepSpeed

Pull Requests Commits

Merge branch 'master' into stas00/diff-driven-test-selection

stas00 committed 7 hours ago

Verified 468d68a4

Merge branch 'stas00/diff-driven-test-selection' of https://github.com:/deepspeedai/DeepSpeed into stas00/diff-driven-test-selection

stas00 committed 7 hours ago

cce08f30

address feedback

stas00 committed 7 hours ago

9afd433e

address feedback

stas00 committed 7 hours ago

0afd7a16

[CI] diff driven test selection

stas00 committed 7 hours ago

be3d8e6e

fix: close file descriptor in deepspeed_io_handle_t::wait() to prevent fd leak (#8075)

MarkCLChang committed 8 hours ago

Verified 3c3c68f7

address feedback

stas00 committed 9 hours ago

75500c42

Fix ZeRO-3: Use per-param dtype for output buffers in _allgather_params_coalesced (#8073)

albertvillanova committed 19 hours ago

Verified b5b3fded

Stop obsolete CI jobs on workflow cancellation (#8034)

tohtana committed 1 day ago

Verified 329f9372

Add merge queue DCO workflow (#8086)

tohtana committed 1 day ago

Verified 8bda651d

Validate fp16 dynamic loss scaling parameters are positive (#8050)

aryanputta committed 1 day ago

Verified ded23491

address feedback

stas00 committed 5 days ago

6577c03f

[CI] diff driven test selection

stas00 committed 5 days ago

f84aec4e

Update version post release (#8070)

loadams committed 7 days ago

Verified ad026a1f

Mixed-precision: per-policy param/buffer dtype cast (preserve fp32 buffers) (#8066)

sfc-gh-truwase committed 7 days ago

Verified b919284a

Add Biren SUPA accelerator support (#8054)

frozenleaves committed 7 days ago

Verified 7ad41089

Pass expected grad dtype to register_z3_param in ZeRO-3 release test (#8063)

tohtana committed 9 days ago

Verified 3a917de5

Fix: ZenFlow Adam integration for updated PyTorch backward flow (#7759) (#7771)

Antlera committed 11 days ago

Verified b2145896

Add AutoEP (#7938)

tohtana committed 12 days ago

Verified 7c3fbdc0

Consistent fp32 grads flow (#8056)

sfc-gh-truwase committed 13 days ago

Verified 56e7e0c5

zero3: invalidate coordinator trace on hook re-registration (#8043)

roycho96 committed 13 days ago

Verified 87f31d55

fix(transformer): use correct stride in Transpose_Kernel shared memory indexing to eliminate bank conflicts (#8055)

flutist committed 13 days ago

Verified 65e5f6fc

Remove AutoSP assertion against Transformers version (#8044)

tohtana committed 14 days ago

Verified a14e4e18

Normalize ZeRO-3 DeepCompile grad dtype before reduction (#8038)

tohtana committed 14 days ago

Verified 4f5fb834

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

tohtana committed 14 days ago

Verified 11eeb7cd

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

tohtana committed 15 days ago

Verified 92cfacb9

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

sowndappan5 committed 17 days ago

Verified 3dc98deb

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

Antlera committed 19 days ago

Verified 28a196f7

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

tohtana committed 21 days ago

Verified 3e486feb

Bump version after 0.19.1 release (#8040)

tohtana committed 24 days ago

Verified 334230ff

Older