Skip to content
体验新版
项目
组织
正在加载...
登录
切换导航
打开侧边栏
PaddlePaddle
Paddle
提交
d3105dbf
P
Paddle
项目概览
PaddlePaddle
/
Paddle
大约 2 年 前同步成功
通知
2325
Star
20933
Fork
5424
代码
文件
提交
分支
Tags
贡献者
分支图
Diff
Issue
1423
列表
看板
标记
里程碑
合并请求
543
Wiki
0
Wiki
分析
仓库
DevOps
项目成员
Pages
P
Paddle
项目概览
项目概览
详情
发布
仓库
仓库
文件
提交
分支
标签
贡献者
分支图
比较
Issue
1,423
Issue
1,423
列表
看板
标记
里程碑
合并请求
543
合并请求
543
Pages
分析
分析
仓库分析
DevOps
Wiki
0
Wiki
成员
成员
收起侧边栏
关闭侧边栏
动态
分支图
创建新Issue
提交
Issue看板
提交
d3105dbf
编写于
2月 07, 2021
作者:
S
sandyhouse
浏览文件
操作
浏览文件
下载
电子邮件补丁
差异文件
update
上级
7aa0cc3c
变更
2
显示空白变更内容
内联
并排
Showing
2 changed file
with
39 addition
and
28 deletion
+39
-28
python/paddle/distributed/fleet/meta_optimizers/sharding/gradient_clip_helper.py
...ed/fleet/meta_optimizers/sharding/gradient_clip_helper.py
+36
-17
python/paddle/distributed/fleet/meta_optimizers/sharding/utils.py
...addle/distributed/fleet/meta_optimizers/sharding/utils.py
+3
-11
未找到文件。
python/paddle/distributed/fleet/meta_optimizers/sharding/gradient_clip_helper.py
浏览文件 @
d3105dbf
...
@@ -16,8 +16,8 @@ from paddle.distributed.fleet.meta_optimizers.common import OP_ROLE_KEY, OpRole
...
@@ -16,8 +16,8 @@ from paddle.distributed.fleet.meta_optimizers.common import OP_ROLE_KEY, OpRole
class
GradientClipHelper
(
object
):
class
GradientClipHelper
(
object
):
def
__init__
(
self
,
sharding
_ring_id
):
def
__init__
(
self
,
mp
_ring_id
):
self
.
sharding_ring_id
=
sharding
_ring_id
self
.
mp_ring_id
=
mp
_ring_id
def
_is_gradient_clip_op
(
self
,
op
):
def
_is_gradient_clip_op
(
self
,
op
):
return
op
.
desc
.
has_attr
(
"op_namescope"
)
\
return
op
.
desc
.
has_attr
(
"op_namescope"
)
\
...
@@ -31,6 +31,7 @@ class GradientClipHelper(object):
...
@@ -31,6 +31,7 @@ class GradientClipHelper(object):
"""
"""
deperated_vars
=
set
()
deperated_vars
=
set
()
deperate_op_idx
=
set
()
deperate_op_idx
=
set
()
reversed_x_paramname
=
[]
for
idx
,
op
in
enumerate
(
block
.
ops
):
for
idx
,
op
in
enumerate
(
block
.
ops
):
if
not
self
.
_is_gradient_clip_op
(
op
):
if
not
self
.
_is_gradient_clip_op
(
op
):
continue
continue
...
@@ -44,6 +45,8 @@ class GradientClipHelper(object):
...
@@ -44,6 +45,8 @@ class GradientClipHelper(object):
if
shard
.
is_param
(
param_name
)
and
\
if
shard
.
is_param
(
param_name
)
and
\
not
shard
.
has_param
(
param_name
):
not
shard
.
has_param
(
param_name
):
deperate_op
=
True
deperate_op
=
True
elif
shard
.
is_param
(
param_name
):
reversed_x_paramname
.
append
(
param_name
)
if
deperate_op
:
if
deperate_op
:
deperate_op_idx
.
add
(
idx
)
deperate_op_idx
.
add
(
idx
)
...
@@ -65,31 +68,47 @@ class GradientClipHelper(object):
...
@@ -65,31 +68,47 @@ class GradientClipHelper(object):
for
input_name
in
op
.
desc
.
input_arg_names
():
for
input_name
in
op
.
desc
.
input_arg_names
():
if
input_name
not
in
deperated_vars
:
if
input_name
not
in
deperated_vars
:
reversed_inputs
.
append
(
input_name
)
reversed_inputs
.
append
(
input_name
)
op
.
desc
.
set_input
(
"X"
,
reversed_inputs
)
op
.
desc
.
set_input
(
"X"
,
reversed_inputs
)
assert
(
len
(
op
.
desc
.
output_arg_names
())
==
1
)
assert
(
len
(
op
.
desc
.
output_arg_names
())
==
1
)
sum_res
=
op
.
desc
.
output_arg_names
()[
0
]
sum_res
=
op
.
desc
.
output_arg_names
()[
0
]
block
.
_insert_op_without_sync
(
idx
+
1
,
# this allreduce should not overlap with calc and should be scheduled in calc stream
type
=
'c_sync_comm_stream'
,
# block._insert_op_without_sync(
inputs
=
{
'X'
:
sum_res
},
# idx + 1,
outputs
=
{
'Out'
:
sum_res
},
# type='c_sync_comm_stream',
attrs
=
{
'ring_id'
:
0
,
# inputs={'X': sum_res},
OP_ROLE_KEY
:
OpRole
.
Optimize
})
# outputs={'Out': sum_res},
# attrs={'ring_id': 0,
# OP_ROLE_KEY: OpRole.Optimize})
block
.
_insert_op_without_sync
(
block
.
_insert_op_without_sync
(
idx
+
1
,
idx
+
1
,
type
=
'c_allreduce_sum'
,
type
=
'c_allreduce_sum'
,
inputs
=
{
'X'
:
sum_res
},
inputs
=
{
'X'
:
sum_res
},
outputs
=
{
'Out'
:
sum_res
},
outputs
=
{
'Out'
:
sum_res
},
attrs
=
{
attrs
=
{
'ring_id'
:
self
.
sharding_ring_id
,
'ring_id'
:
self
.
mp_ring_id
,
OP_ROLE_KEY
:
OpRole
.
Optimize
'op_namescope'
:
"/gradient_clip_model_parallelism"
,
'use_calc_stream'
:
True
,
OP_ROLE_KEY
:
OpRole
.
Optimize
,
})
})
block
.
_insert_op_without_sync
(
# block._insert_op_without_sync(
idx
+
1
,
# idx + 1,
type
=
'c_sync_calc_stream'
,
# type='c_sync_calc_stream',
inputs
=
{
'X'
:
sum_res
},
# inputs={'X': sum_res},
outputs
=
{
'Out'
:
sum_res
},
# outputs={'Out': sum_res},
attrs
=
{
OP_ROLE_KEY
:
OpRole
.
Optimize
})
# attrs={OP_ROLE_KEY: OpRole.Optimize})
# the grad sum here should take the all and only param in the current shard
to_check_param
=
set
(
reversed_x_paramname
)
should_check_param
=
set
(
shard
.
global_params
).
intersection
(
set
([
param
for
param
,
worker_idx
in
shard
.
global_param2device
.
items
()
if
worker_idx
==
shard
.
worker_idx
]))
assert
to_check_param
==
should_check_param
,
"amp check_finite_and_unscale checking miss [{}] and got unexpected [{}]"
.
format
(
should_check_param
-
to_check_param
,
to_check_param
-
should_check_param
)
for
var_name
in
deperated_vars
:
for
var_name
in
deperated_vars
:
block
.
_remove_var
(
var_name
,
sync
=
False
)
block
.
_remove_var
(
var_name
,
sync
=
False
)
...
...
python/paddle/distributed/fleet/meta_optimizers/sharding/utils.py
100755 → 100644
浏览文件 @
d3105dbf
...
@@ -227,14 +227,9 @@ def get_valid_op_role(block, insert_idx):
...
@@ -227,14 +227,9 @@ def get_valid_op_role(block, insert_idx):
return OpRole.Forward or OpRole.Backward
return OpRole.Forward or OpRole.Backward
"""
"""
op_role
=
block
.
ops
[
insert_idx
].
attr
(
'op_role'
)
op_role
=
block
.
ops
[
insert_idx
].
attr
(
'op_role'
)
#if (insert_idx >= len(block.ops)) or (
if
(
insert_idx
>=
len
(
block
.
ops
))
or
(
# op_role in [int(OpRole.Backward), int(OpRole.Optimize)]):
op_role
in
[
int
(
OpRole
.
Backward
),
int
(
OpRole
.
Optimize
)]):
# return OpRole.Backward
return
OpRole
.
Backward
#if op_role in [int(OpRole.Forward), int(OpRole.Loss)]:
# return OpRole.Forward
if
insert_idx
>=
len
(
block
.
ops
):
return
OpRole
.
Optimize
if
op_role
==
int
(
OpRole
.
Backward
):
return
OpRole
.
Backward
if
op_role
==
int
(
OpRole
.
Optimize
):
return
OpRole
.
Optimize
if
op_role
in
[
int
(
OpRole
.
Forward
),
int
(
OpRole
.
Loss
)]:
if
op_role
in
[
int
(
OpRole
.
Forward
),
int
(
OpRole
.
Loss
)]:
return
OpRole
.
Forward
return
OpRole
.
Forward
...
@@ -485,9 +480,6 @@ def save_persistables(exe, dirname, main_program, filename=None):
...
@@ -485,9 +480,6 @@ def save_persistables(exe, dirname, main_program, filename=None):
This function handles the model saving for sharding training.
This function handles the model saving for sharding training.
"""
"""
if
main_program
.
_pipeline_opt
:
main_program
=
main_program
.
_pipeline_opt
[
'section_program'
][
'program'
]
def
is_opt_vars
(
var
):
def
is_opt_vars
(
var
):
# NOTE(liangjianzhong): The checks should be updated when add new compatible optimizer
# NOTE(liangjianzhong): The checks should be updated when add new compatible optimizer
# now only Momentum and adam are compatible with sharding
# now only Momentum and adam are compatible with sharding
...
...
编辑
预览
Markdown
is supported
0%
请重试
或
添加新附件
.
添加附件
取消
You are about to add
0
people
to the discussion. Proceed with caution.
先完成此消息的编辑!
取消
想要评论请
注册
或
登录