microsoft/DeepSpeed

Pull Requests Commits

Get full partition size

tjruwase committed 4 years ago

8e89a829

Fix UnboundLocalError

tjruwase committed 4 years ago

8d552017

update lr scheduler doc for doing per step or epoch update (#913)

cli99 committed 4 years ago

Verified c83e49f9

[zero] faster flatten/unflatten (cpp version) (#910)

stas00 committed 4 years ago

Verified 8b8ed2a7

document how to override ~/.cache/torch_extensions (#959)

stas00 committed 4 years ago

Verified 7003d447

[config] turn exponential notation back on for config dump (#955)

stas00 committed 4 years ago

Verified c87118b0

fix double linear override; spelling (#954)

stas00 committed 4 years ago

Verified adac058a

Delete check of pdsh (#941)

tma15 committed 4 years ago

Verified e6999ebd

jeffra committed 4 years ago

Verified 112ebff6

jeffra committed 4 years ago

0cd9abf2

Supporting different hidden dimensions for transformer kernels-v2 (#934)

RezaYazdaniAminabadi committed 4 years ago

Verified e721cb69

stas00 committed 4 years ago

Verified 31699291

Samyamr/stage 3 skip modules without parameters (#867)

samyam committed 4 years ago

Verified 7b46d11f

[zero doc] fix misspelled param (#878)

stas00 committed 4 years ago

Verified f19cf671

improved readability + typos (#895)

stas00 committed 4 years ago

Verified 5ca86ae4

[benchmarks] flatten/unflatten benchmarks (#919)

stas00 committed 4 years ago

Verified a128f34e

fix cpu_adam memory leak on deepspeed re-use in the same process (#896)

stas00 committed 4 years ago

Verified c79184eb

[zero3] GatheredParameters can now handle a list of params (#884)

stas00 committed 4 years ago

Verified 6d94afb5

Fix for fragmented linear inputs in ZeRO 3 Linear layers where reshap… (#881)

samyam committed 4 years ago

Verified b5f56b2c

Add space in help string (#926)

tma15 committed 4 years ago

Verified ce14cf1a

Add link to AML examples. (#916)

awan-10 committed 4 years ago

Verified ab5534fc

disable pipe test (#915)

jeffra committed 4 years ago

Verified 8db4fdf8

zero.Init() clarification (#880)

stas00 committed 4 years ago

Verified 5d721e09

[website] we're hiring!

jeffra committed 4 years ago

Verified c814abda

[website] We're hiring! + integration posts

jeffra committed 4 years ago

Verified c6b497df

We're hiring! + integration posts

jeffra committed 4 years ago

Verified 8c9e16eb

Bump kramdown from 2.3.0 to 2.3.1 in /docs (#905)

dependabot[bot] committed 4 years ago

Verified c0422642

update backward api doc (#903)

jeffra committed 4 years ago

Verified 23ff6cb7

update kramdown (#901)

jeffra committed 4 years ago

Verified af2d8fc5

Fix zero stage2 cpu_offload when some model trainable parameters skipped in training (#861)

ghosthamlet committed 4 years ago

Verified 7fcc8911

Older