microsoft/DeepSpeed

Pull Requests Commits

fix import error and add lazy import

jeffra committed 2 years ago

5193b268

jeffra committed 2 years ago

9c756cf8

Requires grad checking. (#3789)

jomayeri committed 2 years ago

Verified e5b1eadb

Missing strided copy for gated MLP (#3788)

cmikeh2 committed 2 years ago

Verified b7e1010b

Bug Fixes for autotuner and flops profiler (#1880)

cli99 committed 2 years ago

Verified b4a2c0af

Add ZeRO++ Japanese blog (#3797)

tohtana committed 2 years ago

Verified 319b64ed

adding zero++ to navigation panel of deepspeed.ai (#3796)

HeyangQin committed 2 years ago

Verified 01b843aa

HeyangQin committed 2 years ago

Verified e1119d8b

DeepSpeed-Triton for Inference (#3748)

stephen-youn committed 2 years ago

Verified 4dc65f7b

remove staging trigger (#3792)

jeffra committed 2 years ago

Verified 2c62cb4c

ZeRO++ chinese blog (#3793)

HeyangQin committed 2 years ago

Verified 49a0a1bb

jeffra committed 2 years ago

40045dc7

revert PR #3611 (#3786)

jeffra committed 2 years ago

Verified fc9e1ee0

use `Flops Profiler` to test `model.generate()` (#2515)

CaffreyR committed 2 years ago

Verified c4c442f0

fix interpolate flops compute (#3782)

cli99 committed 2 years ago

Verified a8c182a4

[Fix] _conv_flops_compute when padding is a str and stride=1 (#3169)

zhiruiluo committed 2 years ago

Verified d81a6ad6

zero++ tutorial PR (#3783)

HeyangQin committed 2 years ago

Verified df1859d6

Add H100 workflow and status badge. (#3754)

loadams committed 2 years ago

Verified dd593410

Update deepspeed-chat/japanese/README.md (#3765)

eltociear committed 2 years ago

Verified 1491e14e

Fix url in getting-started guide (docs) (#3768)

acforvs committed 2 years ago

Verified 044dd0e2

fix ccl_backend and residual_add problems (#3642)

Dino Chen committed 2 years ago

Verified 3f5e4931

Account for expert parameters when calculating the total number of parameters in the model (#3720)

alito committed 2 years ago

Verified 9a8f6a1d

add Chinese Zhihu social account (#3755)

conglongli committed 2 years ago

Verified b692d236

remove UtilsBuilder load, use torch (un)flatten ops (#3728)

inkcherry committed 2 years ago

Verified 5a5340d0

Fix output transpose dimension bugs (#3747)

loadams committed 2 years ago

Verified cd911f9a

fix hybrid engine mlp module (#3736)

tensor-tang committed 2 years ago

Verified 45466afa

Include cublas error details when getting cublas handle fails (#3695)

jli committed 2 years ago

Verified 46bb08c2

Fix autotuner get_gas_from_user_config (#3664)

straywarrior committed 2 years ago

Verified 09332dbf

Fix apex install bugs (#3741)

loadams committed 2 years ago

Verified 1b401823

FP8 unittest for H100 (#3731)

jomayeri committed 2 years ago

Verified 6f4fc30b

Older