PR #5336 FP [6,8,12] quantizer op

FP [6,8,12] quantizer op #5336

loadams merged 11 commits into deepspeedai:master from Snowflake-Labs:fp-quantizer

test

010fa309

fp[6,8,12] quantizer op

99367962

op builder

cc3096f3

cleanup

483d8d93

jeffra requested a review from

mrwyattii 1 year ago

jeffra requested a review from

awan-10 1 year ago

jeffra requested a review from

arashb 1 year ago

jeffra requested a review from

tjruwase 1 year ago

jeffra requested a review from

loadams 1 year ago

fp quantizer assumes ampere and above arch, also disable ninja for pr…

a53cd0b1

ifdef bf16 in reduction utils

4092e467

skip on cpu

90c13db7

cannot run fp quant on v100

bf390f4d

fix missing import

ee4aa3f9

move qtorch import to be after pytest skip

23c06bf5

Merge branch 'master' into fp-quantizer

fd564a72

mrwyattii approved these changes on 2024-04-03

JamesTheZ commented on 2024-04-03

JamesTheZ approved these changes on 2024-04-04

loadams merged 3fbd01cc into master 1 year ago

Reviewers

JamesTheZ

mrwyattii

sfc-gh-reyazda

awan-10

arashb

tjruwase

loadams

Assignees

No one assigned

Labels

None yet

Milestone

No milestone

DeepSpeed FP [6,8,12] quantizer op #5336 Merged

FP [6,8,12] quantizer op #5336

DeepSpeed
FP [6,8,12] quantizer op
#5336

Merged