Skip to content
GitLab
Menu
Projects
Groups
Snippets
Loading...
Help
Help
Support
Community forum
Keyboard shortcuts
?
Submit feedback
Contribute to GitLab
Sign in / Register
Toggle navigation
Menu
Open sidebar
wuxk1
Megatron-LM
Commits
aaa5715a
"git@developer.sourcefind.cn:modelzoo/resnet50_tensorflow.git" did not exist on "79b885f091319c37cb47c9bb91ec2c55186afd31"
Commit
aaa5715a
authored
Aug 16, 2022
by
Mohammad Shoeybi
Committed by
Jared Casper
Aug 16, 2022
Browse files
fixed grad scalar warning for bf16
parent
91384a5a
Changes
2
Hide whitespace changes
Inline
Side-by-side
Showing
2 changed files
with
7 additions
and
5 deletions
+7
-5
megatron/optimizer/distrib_optimizer.py
megatron/optimizer/distrib_optimizer.py
+3
-2
megatron/optimizer/optimizer.py
megatron/optimizer/optimizer.py
+4
-3
No files found.
megatron/optimizer/distrib_optimizer.py
View file @
aaa5715a
...
@@ -449,8 +449,9 @@ class DistributedOptimizer(MixedPrecisionOptimizer):
...
@@ -449,8 +449,9 @@ class DistributedOptimizer(MixedPrecisionOptimizer):
# Grad scaler.
# Grad scaler.
if
'grad_scaler'
not
in
state_dict
:
if
'grad_scaler'
not
in
state_dict
:
print_rank_0
(
'***WARNING*** found an old checkpoint, will not '
if
self
.
fp16
:
'load grad scaler ...'
)
print_rank_0
(
'***WARNING*** found an old checkpoint, will not '
'load grad scaler ...'
)
else
:
else
:
if
self
.
grad_scaler
:
if
self
.
grad_scaler
:
self
.
grad_scaler
.
load_state_dict
(
state_dict
[
'grad_scaler'
])
self
.
grad_scaler
.
load_state_dict
(
state_dict
[
'grad_scaler'
])
...
...
megatron/optimizer/optimizer.py
View file @
aaa5715a
...
@@ -679,9 +679,10 @@ class Float16OptimizerWithFloat16Params(MixedPrecisionOptimizer):
...
@@ -679,9 +679,10 @@ class Float16OptimizerWithFloat16Params(MixedPrecisionOptimizer):
self
.
optimizer
.
load_state_dict
(
state_dict
[
optimizer_key
])
self
.
optimizer
.
load_state_dict
(
state_dict
[
optimizer_key
])
# Grad scaler.
# Grad scaler.
if
self
.
fp16
and
'grad_scaler'
not
in
state_dict
:
if
'grad_scaler'
not
in
state_dict
:
print_rank_0
(
'***WARNING*** found an old checkpoint, will not '
if
self
.
fp16
:
'load grad scaler ...'
)
print_rank_0
(
'***WARNING*** found an old checkpoint, will not '
'load grad scaler ...'
)
else
:
else
:
if
self
.
grad_scaler
:
if
self
.
grad_scaler
:
self
.
grad_scaler
.
load_state_dict
(
state_dict
[
'grad_scaler'
])
self
.
grad_scaler
.
load_state_dict
(
state_dict
[
'grad_scaler'
])
...
...
Write
Preview
Markdown
is supported
0%
Try again
or
attach a new file
.
Attach a file
Cancel
You are about to add
0
people
to the discussion. Proceed with caution.
Finish editing this message first!
Cancel
Please
register
or
sign in
to comment