Fix llama meta tensor loading in AutoTP and kernel injected inference #3608
Adapt to Llama when using meta tensor to load
d66ae982
Fix gated mlp parameter mp
5d1fcf42
zeyugao
force pushed
from
b29ef812
to
5d1fcf42
2 years ago
zeyugao
changed the title Adapte to Llama when using meta tensor to load Fix llama meta tensor loading, model tensor parallelism inference 2 years ago
Re-enable meta tensor for kernel injection
f8ce148d
Merge branch 'master' into master
fe0512fc
Merge remote-tracking branch 'origin/master' into pr-master
779bbc3e
zeyugao
changed the title Fix llama meta tensor loading, model tensor parallelism inference Fix llama meta tensor loading in AutoTP and kernel injected inference 2 years ago
Revert mlp_inter_mp for gated mlp as it is fixed
eb695312
Merge remote-tracking branch 'origin/master' into pr-master
02309b50
Monkey patch for fixing llama output
3f684cb5
zeyugao
force pushed
from
eb695312
to
9d79cfd1
2 years ago
Merge branch 'master' of https://github.com/zeyugao/DeepSpeed into ze…
9fbd189a
t push origin masterMerge branch 'zeyugao-master'
166469cb
Merge branch 'master' of https://github.com/microsoft/DeepSpeed
e35f460b
Fix formatting
f2f92fe9
Merge branch 'master' into master
07ac3c7d
Merge branch 'master' into master
a98e6463
Add comment
f51feaee
Merge branch 'master' of https://github.com/zeyugao/DeepSpeed
0a5dd861
lekurile
approved these changes
on 2023-09-20
lekurile
merged
4fc2c8e7
into master 2 years ago
Assignees
No one assigned
Login to write a write a comment.
Login via GitHub