DeepSpeed
[CPU] add fp16 support to shm inference_all_reduce
#5669

Merged

[CPU] add fp16 support to shm inference_all_reduce #5669

loadams merged 12 commits into deepspeedai:master from delock:gma/fp16_allreduce_support

add fp16 support to shm allreduce

0aa626fa

delock requested a review from

awan-10 2 years ago

delock requested a review from

mrwyattii 2 years ago

delock requested a review from

arashb 2 years ago

fix format

cd89810d

Merge branch 'master' into gma/fp16_allreduce_support

94094678

loadams requested a review from

tjruwase 2 years ago

loadams requested a review from

tohtana 2 years ago

add more data types for test inference_all_reduce

206e29a4

delock requested a review from

loadams 2 years ago

adk9 approved these changes on 2024-06-18

fix FP32+world_size=1 bug

44007e2d

remove unneeded code

52084b07

fix format

4719cba1

remove unnecessary comments

9e08a420

Merge branch 'master' into gma/fp16_allreduce_support

7b62634b

Merge branch 'master' into gma/fp16_allreduce_support

e077d572

Merge branch 'master' into gma/fp16_allreduce_support

812579f9

loadams enabled auto-merge 2 years ago

disabled auto-merge 2 years ago
Manually disabled by user

Merge branch 'master' into gma/fp16_allreduce_support

ae7497a9

loadams enabled auto-merge 2 years ago

loadams merged 19da95f7 into master 2 years ago

Reviewers

adk9

awan-10

mrwyattii

arashb

tjruwase

tohtana

loadams

Assignees

No one assigned

Labels

None yet

Milestone

No milestone

DeepSpeed [CPU] add fp16 support to shm inference_all_reduce #5669 Merged

[CPU] add fp16 support to shm inference_all_reduce #5669

DeepSpeed
[CPU] add fp16 support to shm inference_all_reduce
#5669

Merged