DeepSpeed is a deep learning optimization library that makes distributed training and inference easy, efficient, and effective.
COMMITS
/ op_builder/builder.py March 16, 2022
S
[build] support cuda-11.6 (#1836)
Stas Bekman committed
March 7, 2022
J
Invoke hipify from op builder for JIT extension builds (#1807)
Jithun Nair committed
March 3, 2022
J
AMD support (#1430)
Jeff Rasley committed
February 11, 2022
J
unset torch arch list for JIT mode (#1765)
Jeff Rasley committed
January 19, 2022
J
MoE inference + PR-MoE model support (#1705)
Jeff Rasley committed
January 10, 2022
J
add -lcurand to fix torch-nightly issue w. JIT (#1688)
Jeff Rasley committed
January 5, 2022
V
support CPU Adam and Adagrad on Windows with SDK 10.0.22000 (#1634)
Victor committed
December 21, 2021
J
follow-up to #1652, resolved a100-80gb issue (#1655)
Jeff Rasley committed
December 20, 2021
J
add backup cpu-arch detection if py-cpuinfo fails (#1652)
Jeff Rasley committed
December 11, 2021
V
December 9, 2021
O
Control ds_report output (#1622)
Olatunji Ruwase committed
November 23, 2021
W
Enable AVX256 on AMD CPU (#1360)
Wenhao Hu committed
November 15, 2021
S
[build] support cuda-11.5 (#1558)
Stas Bekman committed
November 2, 2021
A
Prevent creation of local temp directory (#1494)
Alex Hedges committed
October 2, 2021
A
Fix many typos (#1423)
Alex Hedges committed
August 30, 2021
S
update for cuda-11.4 (#1329)
Stas Bekman committed
August 27, 2021
R
use scalar cpu-adam in case of exception in builder (#1259)
Reza Yazdani committed
July 29, 2021
A
query for libaio package using known package managers (#1250)
Adam Moody committed
July 24, 2021
A
aio: test link against libaio using distutils (#1247)
Adam Moody committed
July 15, 2021
A
enable async io op on powerpc architectures (#1224)
Adam Moody committed
June 16, 2021
J
Remove hard torch dependency at install (#1166)
Jeff Rasley committed
June 10, 2021
June 7, 2021
S
[builder] allow minor cuda version differences (#1139)
Stas Bekman committed
May 24, 2021
R
Quantization + inference release (#1091)
Reza Yazdani committed
March 8, 2021
S
ZeRO 3 Offload (#834)
Samyam Rajbhandari committed
January 7, 2021
X
Update builder.py (#642)
Xingjian Shi committed
December 23, 2020
J
Elastic training support (#602)
Jeff Rasley committed
December 11, 2020
S
[build] fix computer capability arch flags, add PTX, handle PTX (#591)
Stas Bekman committed
December 7, 2020
S
[build] add compute_86 (#577)
Stas Bekman committed
S
[build] make builder smarter and configurable wrt compute capabilities + docs (#578)
Stas Bekman committed
December 3, 2020
S
[build] build against installed cuda-11.1 while torch built w/ cuda-11.0 (#570)
Stas Bekman committed
J
Add compute capability 8.0 if on cuda 11+ (#572)
Jeff Rasley committed
November 12, 2020
J
DeepSpeed JIT op + PyPI support (#496)
Jeff Rasley committed