Add tensor parallel inference unit tests #2232
jeffra
merged
34 commits into
deepspeedai:master
from
mrwyattii:add-mp-inf-tests
added mp test
770c59a8
Merge branch 'master' into add-mp-inf-tests
706617fd
make text-gen compare DS output to baseline
c0925a14
Merge branch 'master' into add-mp-inf-tests
36bb2c5e
Merge branch 'master' into add-mp-inf-tests
f942b5a5
remove non-gpt models from multigpu test
adaabe0d
add more models and fixes for OOM
f9a7eae5
added job ID to environment printout
606a669e
made multi-GPU tests sequential
aef76031
Merge branch 'master' into add-mp-inf-tests
27a9825b
update inference workflow with sequential tests
7cb4136f
jeffra
commented
on 2022-08-26
jeffra
approved these changes
on 2022-08-26
fix for multi-gpu
409af10b
fix for selecting inference sequential tests
28c81024
change default to not include inference tests
185d4b91
restrict gptneox to half precision
5c84dcab
increas world_size to 4 for multigpu
6edb13b5
Merge branch 'master' into add-mp-inf-tests
a75d5fea
Merge branch 'master' into add-mp-inf-tests
a79bb153
Merge branch 'master' into add-mp-inf-tests
2ab79f07
use fp16 revision of gpt-j
19839cb2
revert non-gptj
776be031
use gpt-j fp16 revision
429c88dd
Merge branch 'master' into add-mp-inf-tests
b6d2b3be
Merge branch 'master' into add-mp-inf-tests
af8d94d3
fix for broken tests
335406a7
fix invalid model/task check
d0bb608f
correct gpt-neo test that should have been skipped
4678f05b
add exception for bloom models failing to match text-gen output
5b5e65a2
restrict number of tokens generated
19f2bf09
fix for opt models
7feac82d
Update test_inference.py
cfdf629d
Merge branch 'master' into add-mp-inf-tests
3e87d74e
formatting
42ccea56
Merge branch 'master' into add-mp-inf-tests
db48d24a
jeffra
commented
on 2022-09-09
jeffra
approved these changes
on 2022-09-09
Reviewers
Assignees
No one assigned
Labels
None yet
Milestone
No milestone
Login to write a write a comment.
Login via GitHub