microsoft/DeepSpeed

Pull Requests Commits

Merge branch 'tohtana/deepcompile' into tohtana/deepcompile_fix_scheduling

Masahiro Tanaka committed 277 days ago

fb83c8d9

Masahiro Tanaka committed 277 days ago

8b993904

Merge branch 'tohtana/deepcompile' of github.com:deepspeedai/DeepSpeed into tohtana/deepcompile

Masahiro Tanaka committed 277 days ago

7c78179e

remove debugging code

Masahiro Tanaka committed 277 days ago

37134a3d

Masahiro Tanaka committed 282 days ago

23c7d2d7

Merge branch 'tohtana/deepcompile' of github.com:deepspeedai/DeepSpeed into tohtana/deepcompile

Masahiro Tanaka committed 285 days ago

9feaefc1

add deepcompile dependency

Masahiro Tanaka committed 285 days ago

61395e77

Merge branch 'master' into tohtana/deepcompile

tjruwase committed 285 days ago

Verified e639c85f

Update to fix pydantic warning (#7193)

loadams committed 285 days ago

Verified 027ee21a

free real input to forward

Masahiro Tanaka committed 286 days ago

22b7bc29

Reland perf fix for nan inf check (#7184)

nelyahu committed 286 days ago

Verified 3c1817f3

Update to new torch grad hook API: BF16Optimizer and Stage2 (#7189)

deepcharm committed 287 days ago

Verified 79ff1627

Add qwen3 autotp support (#7187)

Yejing-Lai committed 287 days ago

Verified 11aca297

fix return value

Masahiro Tanaka committed 288 days ago

9b23d6fc

Fix issue #5242 grad_norm and loss is nan (#7171)

Glaceon-Hyy committed 291 days ago

Verified 1f706621

async tp allreduce (#7115)

inkcherry committed 291 days ago

Verified b8cc1eb0

Cross layer overlapping for domino (#7178)

hwchen2017 committed 291 days ago

Verified f355b9ea

Update version.txt after 0.16.5 release (#7180)

loadams committed 292 days ago

Verified 3054b934

Merge branch 'tohtana/deepcompile' of github.com:deepspeedai/DeepSpeed into tohtana/deepcompile

Masahiro Tanaka committed 292 days ago

d15eb6ca

add scipy to requirement

Masahiro Tanaka committed 292 days ago

8af3f964

surpress building deepcompile ops for windows

Masahiro Tanaka committed 292 days ago

721bc58f

Merge branch 'master' into tohtana/deepcompile

loadams committed 292 days ago

Verified dd7dbfe4

Variable batch size and LR scheduler (#7104)

bm-synth committed 292 days ago

Verified 20f988ea

Merge branch 'master' into tohtana/deepcompile

tohtana committed 293 days ago

Verified 429a4231

Masahiro Tanaka committed 293 days ago

832a147f

Fix typos in GDS blog (#7177)

loadams committed 293 days ago

Verified ac295aa0

fix `seq_parallel_communication_data_type` constant. (#7175)

stas00 committed 293 days ago

Verified 2dbb5d94

Masahiro Tanaka committed 294 days ago

4a26c44d

add sync offload func

Masahiro Tanaka committed 294 days ago

f66102af

Merge branch 'master' into tohtana/deepcompile

tohtana committed 295 days ago

Verified af396e71

Older