microsoft/DeepSpeed

Pull Requests Commits

Merge branch 'master' into tingfeng/zenflow_fix_backward

Antlera committed 10 days ago

Verified c2291bb3

Fix Muon optimizer module path (#7802)

tohtana committed 11 days ago

Verified 3bc882fa

Add full test suite workflow (#7795)

tohtana committed 11 days ago

Verified 5aa2d17d

Fix BF16_Optimizer being used without ZeRO (#7790)

tohtana committed 11 days ago

Verified 09885efe

Fix issue with BF16 optimizer selection (#7788)

tohtana committed 11 days ago

Verified 1393f75d

Skip empty parameters in gradient reduction (#7789)

tohtana committed 11 days ago

Verified 114f971c

Fix loss scaling logic in DeepSpeedEngine.

Antlera committed 11 days ago

0bf92ccd

Fix backward for pipeline engine (#7787)

tohtana committed 12 days ago

Verified 3d896d7e

Fix AutoTP test numerical tolerance with rtol (#7794)

tohtana committed 12 days ago

Verified 374f6d09

Add sequential allgather optimization for ZeRO-3 (#7661)

aeeeeeep committed 12 days ago

Verified 60876090

fix checkpointing/loading of z0+bf16 (#7786)

tohtana committed 12 days ago

Verified 991ebd72

Fix Evoformer compilation (#7760)

sdvillal committed 14 days ago

Verified 43125a75

Fix Ulysses PEFT test (#7784)

tohtana committed 14 days ago

Verified 491a38ce

Fix gradient checkpointing with use_reentrant=True / PyTorch-style backward / ZeRO-3 (#7780)

tohtana committed 15 days ago

Verified 311674ff

docs: update README.md (#7781)

eltociear committed 15 days ago

Verified 5f2f9bf3

fix: skip compressed allreduce for empty tensors (#7769)

T1mn committed 16 days ago

Verified 97e74301

Fix multiprocessing testcase (#7743)

k-artem committed 17 days ago

Verified 7bbb6c77

Update CLA to refer to DCO (#7778)

loadams committed 17 days ago

Verified e31daa8c

[Docs] Fix `docs/_pages/config-json.md` format (#7779)

ooooo-create committed 17 days ago

Verified 79af89be

Remove cron/PR triggers for outdated V100 tests (#7777)

loadams committed 19 days ago

Verified 53c87ce8

Add timeout to test workflows (#7774)

tohtana committed 19 days ago

Verified 52361686

Update README with newer status badges for CI

loadams committed 19 days ago

b80f453f

Merge branch 'tingfeng/zenflow_fix_backward' of github.com:deepspeedai/DeepSpeed into tingfeng/zenflow_fix_backward

Antlera committed 20 days ago

7f25ec2a

Refine logic and error handling based on feedback

Antlera committed 20 days ago

3583d35e

Merge branch 'master' into tingfeng/zenflow_fix_backward

tohtana committed 21 days ago

Verified 3e49ca6f

Antlera committed 21 days ago

af917331

fix: update Megatron-DeepSpeed tutorial to match current repo structure (#7761)

nathon-lee committed 21 days ago

Verified 8a9369d0

Refactor ZenFlow integration in AdamW optimizers.

Antlera committed 21 days ago

f7d9960a

Add ZenFlowAdamBuilder.

Antlera committed 21 days ago

e40eb3e7

fix: delete using namespace std. (#7766)

nathon-lee committed 22 days ago

Verified 816e4aef

Older