DeepSpeed
A faster and more memory-efficient implementation of `zero_to_fp32`
#6658

Merged

A faster and more memory-efficient implementation of `zero_to_fp32` #6658

loadams merged 12 commits into deepspeedai:master from xu-song:patch-4

xu-song requested a review from

tjruwase 1 year ago

xu-song requested a review from

awan-10 1 year ago

tjruwase commented on 2024-10-23

tjruwase removed review request from

awan-10 1 year ago

tjruwase requested a review from

tohtana 1 year ago

xu-song closed this 1 year ago

xu-song reopened this 1 year ago

Faster and more memory-efficient impl of zero_to_fp32

ec734974

xu-song force pushed from e1d12bb6 to ec734974 1 year ago

tjruwase commented on 2024-10-28

Merge branch 'master' into patch-4

27e8ae27

add unit test

e0f1d53f

xu-song requested a review from

loadams 1 year ago

add comments

4d005b8a

fix yapf formatting issue

5684be04

Merge branch 'master' into patch-4

f3f8b271

tjruwase approved these changes on 2024-11-06

Merge branch 'master' into patch-4

b8c9a5e4

rename lazy_mege to lazy_mode; fix Trailing Whitespace

d2b98b03

Merge branch 'master' into patch-4

f36925e9

Merge branch 'master' into patch-4

5a753b97

Merge branch 'master' into patch-4

af82bae8

Merge branch 'master' into patch-4

517210cd

loadams enabled auto-merge 1 year ago

loadams merged dd402694 into master 1 year ago

Reviewers

tjruwase

tohtana

loadams

Assignees

No one assigned

Labels

None yet

Milestone

No milestone

DeepSpeed A faster and more memory-efficient implementation of `zero_to_fp32` #6658 Merged

A faster and more memory-efficient implementation of `zero_to_fp32` #6658

DeepSpeed
A faster and more memory-efficient implementation of `zero_to_fp32`
#6658

Merged