microsoft/DeepSpeed

Pull Requests Commits

bump DSE commit

jeffra committed 4 years ago

5986dbca

jeffra committed 4 years ago

5fbff461

Fix the workspace allocation for the transformer kernel (#1397)

RezaYazdaniAminabadi committed 4 years ago

Verified bc7778ea

Make the replace module more configurable (#1366)

kamalkraj committed 4 years ago

Verified c6d1418d

[CI] HF tests `image-classification` requirements (#1425)

stas00 committed 4 years ago

Verified 86457c2d

Revise param_shapes to be a list of ordered dict (#1424)

eelxpeng committed 4 years ago

Verified dd0c8fa7

rm max_seq_length arg:DeepSpeedTransformerConfig (#1362)

kamalkraj committed 4 years ago

Verified 466b0e63

[zero_to_fp32] fix to handle world_size (#1422)

stas00 committed 4 years ago

Verified dd6bf4d0

Fix non-fp tensor bugs of contiguous activation checkpointing (#1376)

hyunwoongko committed 4 years ago

Verified d1e72f29

Fix many typos (#1423)

aphedges committed 4 years ago

Verified be789b16

Add flexibility of pipeline parallel module and engine (#1399)

hyunwoongko committed 4 years ago

Verified 30965ea7

Add assert to ensure we don't skip unsupported grad dtypes (#1418)

jeffra committed 4 years ago

Verified 0457bb1c

Improve inference documentation (#1421)

aphedges committed 4 years ago

Verified aa8d97e6

Save the model parallel group at inference engine statically (#1411)

RezaYazdaniAminabadi committed 4 years ago

Verified 9f17087f

Fix typo in assert message (#1420)

manuelciosici committed 4 years ago

Verified 56124621

[CI] make build names consistent

jeffra committed 4 years ago

Verified a4ffc831

[ci] hf dependencies (#1417)

stas00 committed 4 years ago

Verified 8c5ffcd9

Big science related changes (#1407)

jeffra committed 4 years ago

Verified e2fdd254

Bump nokogiri from 1.11.4 to 1.12.5 in /docs (#1409)

dependabot[bot] committed 4 years ago

Verified 39c77440

[CI] install fix for HF tests and use dict instead of json for some tests (#1405)

jeffra committed 4 years ago

Verified 9e5c0c5c

[CI] Add HF transformers tests (#958)

awan-10 committed 4 years ago

Verified c1829c45

Fix from Felipe and Young for loading checkpoints. (#1389)

awan-10 committed 4 years ago

Verified 86dd6a64

[zero_to_fp32] fix padding removal (#1380)

stas00 committed 4 years ago

Verified 364994ad

add tutorial on pytorch profiler usage (#1350)

cli99 committed 4 years ago

Verified 51a2e916

Sparse attn triton v1.0 support + torch1.8 test runner (#1374)

jeffra committed 4 years ago

Verified 6996bb01

fix: support three digit layer numbers (#1377)

adammoody committed 4 years ago

Verified 4ad8019c

jeffra committed 4 years ago

45a498d0

[zero_to_fp32] adapt to 4-bytes alignment in z2 (#1372)

stas00 committed 4 years ago

Verified 30537e71

[zero Init] fix regression (#1373)

stas00 committed 4 years ago

Verified cf22a691

Introduce a device rank when setting device (#1370)

Sean Naren committed 4 years ago

Verified 90398a70

Newer Older