zero3 performance optimizations #3622
Remove dead code
e1e4b753
Prevent evaluation of debug strings
e3dbb7a3
BacharL
force pushed
from
3b588087
to
f8c7ad3a
2 years ago
BacharL
force pushed
from
f8c7ad3a
to
a942dfa1
2 years ago
BacharL
force pushed
from
a942dfa1
to
3a87dfa2
2 years ago
Use contiguous gradients tensor reduce scatter between ranks
bd4d724a
BacharL
force pushed
from
3a87dfa2
to
d6a8711a
2 years ago
BacharL
force pushed
from
d6a8711a
to
bd4d724a
2 years ago
move overflow tracker to optimizer.step
9cf826df
Merge branch 'master' into perf1
6847b306
Merge branch 'master' into perf1
9b80d605
Merge branch 'master' into perf1
a813c774
Merge branch 'master' into perf1
b0bfdc1a
Merge branch 'master' into perf1
7b832241
Merge branch 'master' into perf1
89add18d
Merge branch 'master' into perf1
16e6f1cf
tjruwase
approved these changes
on 2023-06-07
Merge branch 'master' into perf1
290bdbd9
Merge branch 'master' into perf1
df88b940
tjruwase
merged
0977106a
into master 2 years ago
Assignees
No one assigned
Login to write a write a comment.
Login via GitHub