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
a96d54ac
P
Paddle
项目概览
PaddlePaddle
/
Paddle
大约 1 年 前同步成功
通知
2299
Star
20931
Fork
5422
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
未验证
提交
a96d54ac
编写于
7月 29, 2020
作者:
D
Dong Daxiang
提交者:
GitHub
7月 29, 2020
浏览文件
操作
浏览文件
下载
电子邮件补丁
差异文件
Generate final strategy (#25782)
* refine strategy compiler and meta optimizers make async as a_sync
上级
0ca1bb56
变更
9
隐藏空白更改
内联
并排
Showing
9 changed file
with
131 addition
and
46 deletion
+131
-46
paddle/fluid/framework/distributed_strategy.proto
paddle/fluid/framework/distributed_strategy.proto
+2
-2
python/paddle/fleet/base/distributed_strategy.py
python/paddle/fleet/base/distributed_strategy.py
+21
-17
python/paddle/fleet/base/fleet_base.py
python/paddle/fleet/base/fleet_base.py
+14
-7
python/paddle/fleet/base/strategy_compiler.py
python/paddle/fleet/base/strategy_compiler.py
+41
-5
python/paddle/fleet/meta_optimizers/graph_execution_optimizer.py
...paddle/fleet/meta_optimizers/graph_execution_optimizer.py
+9
-0
python/paddle/fleet/meta_optimizers/meta_optimizer_base.py
python/paddle/fleet/meta_optimizers/meta_optimizer_base.py
+3
-0
python/paddle/fleet/meta_optimizers/recompute_optimizer.py
python/paddle/fleet/meta_optimizers/recompute_optimizer.py
+6
-1
python/paddle/fluid/tests/unittests/test_fleet_distributed_strategy.py
.../fluid/tests/unittests/test_fleet_distributed_strategy.py
+12
-12
python/paddle/fluid/tests/unittests/test_fleet_meta_optimizer.py
...paddle/fluid/tests/unittests/test_fleet_meta_optimizer.py
+23
-2
未找到文件。
paddle/fluid/framework/distributed_strategy.proto
浏览文件 @
a96d54ac
...
...
@@ -85,7 +85,7 @@ message DistributedStrategy {
optional
bool
pipeline
=
9
[
default
=
false
];
optional
bool
elastic
=
10
[
default
=
false
];
optional
bool
auto
=
11
[
default
=
false
];
optional
bool
async
=
12
[
default
=
true
];
optional
bool
a
_
sync
=
12
[
default
=
true
];
optional
bool
sync_nccl_allreduce
=
13
[
default
=
true
];
optional
int32
nccl_comm_num
=
14
[
default
=
1
];
optional
bool
use_hierarchical_allreduce
=
15
[
default
=
false
];
...
...
@@ -99,7 +99,7 @@ message DistributedStrategy {
optional
LocalSGDConfig
localsgd_configs
=
103
;
optional
GradientMergeConfig
gradient_merge_configs
=
104
;
optional
PipelineConfig
pipeline_configs
=
106
;
optional
AsyncConfig
async_configs
=
107
;
optional
AsyncConfig
a
_
sync_configs
=
107
;
optional
BuildStrategy
build_strategy
=
201
;
optional
ExecutionStrategy
execution_strategy
=
202
;
...
...
python/paddle/fleet/base/distributed_strategy.py
浏览文件 @
a96d54ac
...
...
@@ -201,7 +201,7 @@ class DistributedStrategy(object):
f
.
name
).
extend
(
getattr
(
strategy
,
f
.
name
))
@
property
def
a
sync_update
(
self
):
def
a
_sync
(
self
):
"""
Indicating whether we are using asynchronous stocastic gradient descent updates
for training. This property is valid when we are using parameter server training,
...
...
@@ -216,29 +216,29 @@ class DistributedStrategy(object):
fleet.init(role_maker)
strategy = fleet.DistributedStrategy()
strategy.a
sync_update
= True # by default this is True
strategy.a
_sync
= True # by default this is True
# code block for defining loss and local optimizer
# sgd = fleet.distributed_optimizer(optimizer, strategy)
"""
return
self
.
strategy
.
async
return
self
.
strategy
.
a
_
sync
@
a
sync_update
.
setter
def
a
sync_update
(
self
,
flag
):
@
a
_sync
.
setter
def
a
_sync
(
self
,
flag
):
if
isinstance
(
flag
,
bool
):
self
.
strategy
.
async
=
flag
self
.
strategy
.
a
_
sync
=
flag
else
:
print
(
"WARNING: a
sync_update
should have value of bool type"
)
print
(
"WARNING: a
_sync
should have value of bool type"
)
@
property
def
a
sync_update
_configs
(
self
):
def
a
_sync
_configs
(
self
):
"""
Set async update configurations. In general, asynchronous parameter server
Set a
_
sync update configurations. In general, asynchronous parameter server
training has serveral configurable settings that can be configured through
a dict.
**Notes**:
**Detailed arguments for a
sync_update
_configs**
**Detailed arguments for a
_sync
_configs**
**k_step**: number of local optimization updates before communication
**max_merge_var_num**: maximum number of merged gradients before communication
**send_queue_size**: a buffer size of worker communication
...
...
@@ -255,19 +255,20 @@ class DistributedStrategy(object):
fleet.init(role_maker)
strategy = fleet.DistributedStrategy()
strategy.a
sync_update
= True # by default this is True
strategy.a
_sync
= True # by default this is True
configs = {"k_step": 10000, "send_queue_size": 32}
strategy.a
sync_update
_configs = configs
strategy.a
_sync
_configs = configs
# code block for defining loss and local optimizer
# sgd = fleet.distributed_optimizer(optimizer, strategy)
"""
return
get_msg_dict
(
self
.
strategy
.
async_configs
)
return
get_msg_dict
(
self
.
strategy
.
a
_
sync_configs
)
@
async_update_configs
.
setter
def
async_update_configs
(
self
,
configs
):
check_configs_key
(
self
.
strategy
.
async_configs
,
configs
,
"async_configs"
)
assign_configs_value
(
self
.
strategy
.
async_configs
,
configs
)
@
a_sync_configs
.
setter
def
a_sync_configs
(
self
,
configs
):
check_configs_key
(
self
.
strategy
.
a_sync_configs
,
configs
,
"a_sync_configs"
)
assign_configs_value
(
self
.
strategy
.
a_sync_configs
,
configs
)
@
property
def
amp
(
self
):
...
...
@@ -584,4 +585,7 @@ class DistributedStrategy(object):
print
(
"WARNING: auto should have value of bool type"
)
def
__repr__
(
self
):
fields
=
self
.
strategy
.
DESCRIPTOR
.
fields
for
f
in
fields
:
print
(
"{}: {}"
.
format
(
f
.
name
,
f
.
default_value
))
return
str
(
self
.
strategy
)
python/paddle/fleet/base/fleet_base.py
浏览文件 @
a96d54ac
...
...
@@ -228,6 +228,7 @@ class Fleet(object):
"""
self
.
user_defined_optimizer
=
optimizer
self
.
user_defined_strategy
=
strategy
self
.
valid_strategy
=
None
return
self
def
minimize
(
self
,
...
...
@@ -292,8 +293,10 @@ class Fleet(object):
distributed_optimizer_list
=
\
MetaOptimizerFactory
().
_get_valid_meta_optimizers
(
self
.
user_defined_optimizer
)
valid_optimizer_list
=
[]
valid_graph_optimizer_list
=
[]
can_not_apply_optimizer_list
=
[]
# recall meta optimizers for ranking
for
opt
in
distributed_optimizer_list
:
opt
.
_set_basic_info
(
loss
,
self
.
_role_maker
,
...
...
@@ -301,15 +304,21 @@ class Fleet(object):
self
.
user_defined_strategy
)
if
opt
.
_can_apply
()
and
not
opt
.
_is_graph_out
():
valid_optimizer_list
.
append
(
opt
)
if
opt
.
_can_apply
()
and
opt
.
_is_graph_out
():
el
if
opt
.
_can_apply
()
and
opt
.
_is_graph_out
():
valid_graph_optimizer_list
.
append
(
opt
)
else
:
can_not_apply_optimizer_list
.
append
(
opt
)
# combine recalled meta optimizers to be a valid meta optimizer
meta_optimizer
,
graph_optimizer
,
final_dist_strategy
=
\
meta_optimizer
,
graph_optimizer
=
\
self
.
strategy_compiler
.
generate_optimizer
(
loss
,
self
.
_role_maker
,
self
.
user_defined_optimizer
,
self
.
user_defined_strategy
,
valid_optimizer_list
,
valid_graph_optimizer_list
)
valid_strategy
=
self
.
strategy_compiler
.
_get_valid_strategy
(
self
.
user_defined_strategy
,
can_not_apply_optimizer_list
)
self
.
valid_strategy
=
valid_strategy
optimize_ops
=
[]
params_grads
=
[]
if
meta_optimizer
:
...
...
@@ -332,12 +341,10 @@ class Fleet(object):
if
self
.
_runtime_handle
is
None
:
self
.
_runtime_handle
=
RuntimeFactory
().
_create_runtime
(
final_dist_strategy
,
self
.
_role_maker
,
optimize_ops
,
params_grads
)
valid_strategy
,
self
.
_role_maker
,
optimize_ops
,
params_grads
)
if
self
.
_util
is
None
:
self
.
_util
=
UtilFactory
().
_create_util
(
final_dist_strategy
,
self
.
_role_maker
,
optimize_ops
,
params_grads
)
self
.
_util
=
UtilFactory
().
_create_util
(
valid_strategy
,
self
.
_role_maker
,
optimize_ops
,
params_grads
)
return
optimize_ops
,
params_grads
python/paddle/fleet/base/strategy_compiler.py
浏览文件 @
a96d54ac
...
...
@@ -30,7 +30,7 @@ def maximum_path_len_algo(optimizer_list):
return
None
for
idx
,
opt
in
enumerate
(
candidates
[
max_idx
][:
-
1
]):
opt
.
_update_inner_optimizer
(
candidates
[
max_idx
][
idx
+
1
])
return
candidates
[
max_idx
]
[
0
]
return
candidates
[
max_idx
]
class
StrategyCompilerBase
(
object
):
...
...
@@ -51,19 +51,55 @@ class StrategyCompiler(StrategyCompilerBase):
def
__init__
(
self
):
super
(
StrategyCompiler
,
self
).
__init__
()
self
.
_meta_optimizer
=
None
self
.
_graph_optimizer
=
None
self
.
_valid_optimizer_list
=
None
self
.
_user_defined_strategy
=
None
self
.
_meta_optimizer_candidates
=
[]
self
.
_graph_optimizer_candidates
=
[]
def
_get_valid_strategy
(
self
,
dist_strategy
,
can_not_apply_optimizer_list
):
import
copy
valid_strategy
=
copy
.
copy
(
dist_strategy
)
invalid_optimizers
=
[]
for
candidate
in
self
.
_meta_optimizer_candidates
:
is_valid
=
False
for
valid
in
self
.
_meta_optimizers
:
if
candidate
.
__class__
.
__name__
==
valid
.
__class__
.
__name__
:
is_valid
=
True
break
if
not
is_valid
:
invalid_optimizers
.
append
(
candidate
)
for
opt
in
invalid_optimizers
:
opt
.
_disable_strategy
(
valid_strategy
)
for
opt
in
can_not_apply_optimizer_list
:
opt
.
_disable_strategy
(
valid_strategy
)
return
valid_strategy
def
generate_optimizer
(
self
,
loss
,
role_maker
,
optimizer
,
user
d
_defined_strategy
,
meta_optimizer_list
,
user_defined_strategy
,
meta_optimizer_list
,
graph_optimizer_list
):
self
.
_user_defined_strategy
=
user_defined_strategy
self
.
_meta_optimizer_candidates
=
meta_optimizer_list
self
.
_graph_optimizer_candidates
=
graph_optimizer_list
if
len
(
meta_optimizer_list
)
==
0
and
len
(
graph_optimizer_list
)
==
0
:
return
optimizer
,
None
else
:
# currently, we use heuristic algorithm to select
# meta optimizers combinations
meta_optimizer
=
maximum_path_len_algo
(
meta_optimizer_list
)
graph_optimizer
=
maximum_path_len_algo
(
graph_optimizer_list
)
meta_optimizer
s
=
maximum_path_len_algo
(
meta_optimizer_list
)
graph_optimizer
s
=
maximum_path_len_algo
(
graph_optimizer_list
)
# should design a distributed strategy update interface
# when we have finally decided the combination of meta_optimizer
# and graph_optimizer, the corresponding distributed strategy
# should be updated.
return
meta_optimizer
,
graph_optimizer
,
None
self
.
_meta_optimizers
=
meta_optimizers
self
.
_graph_optimizers
=
graph_optimizers
return_meta
=
None
if
meta_optimizers
==
None
else
meta_optimizers
[
0
]
return_graph
=
None
if
graph_optimizers
==
None
else
graph_optimizers
[
0
]
return
return_meta
,
return_graph
python/paddle/fleet/meta_optimizers/graph_execution_optimizer.py
浏览文件 @
a96d54ac
...
...
@@ -16,6 +16,7 @@ from paddle.fluid.framework import core
from
paddle.fluid
import
compiler
from
.meta_optimizer_base
import
MetaOptimizerBase
from
..base.private_helper_function
import
wait_server_ready
import
logging
class
GraphExecutionOptimizer
(
MetaOptimizerBase
):
...
...
@@ -32,6 +33,10 @@ class GraphExecutionOptimizer(MetaOptimizerBase):
"""
Basically, this is PE, and almost all programs can be executed here
"""
if
not
self
.
role_maker
.
_is_collective
:
# update me. currently, if parameter server is used
# graph execution optimizer can not be applied
return
False
return
True
def
backward
(
self
,
...
...
@@ -178,6 +183,10 @@ class GraphExecutionOptimizer(MetaOptimizerBase):
return
self
.
_compiled_program
def
_disable_strategy
(
self
,
dist_strategy
):
# TODO(guru4elephant): should close all PE related flags here
pass
def
minimize
(
self
,
loss
,
startup_program
=
None
,
...
...
python/paddle/fleet/meta_optimizers/meta_optimizer_base.py
浏览文件 @
a96d54ac
...
...
@@ -39,6 +39,9 @@ class MetaOptimizerBase(object):
if
str
(
optimizer
.
__class__
.
__name__
)
in
self
.
meta_optimizers_white_list
:
return
True
def
_disable_strategy
(
self
,
dist_strategy
):
raise
NotImplementedError
(
"you should implement disable strategy"
)
def
minimize_impl
(
self
,
loss
,
startup_program
=
None
,
...
...
python/paddle/fleet/meta_optimizers/recompute_optimizer.py
浏览文件 @
a96d54ac
...
...
@@ -34,11 +34,16 @@ class RecomputeOptimizer(MetaOptimizerBase):
def
_can_apply
(
self
):
if
self
.
user_defined_strategy
.
recompute
==
True
:
if
len
(
self
.
user_defined_strategy
.
recompute_checkpoints
)
==
0
:
if
len
(
self
.
user_defined_strategy
.
recompute_configs
[
"checkpoints"
])
==
0
:
return
False
else
:
return
True
def
_disable_strategy
(
self
,
dist_strategy
):
dist_strategy
.
recompute
=
False
dist_strategy
.
recompute_configs
=
{
"checkpoints"
:
[]}
def
backward
(
self
,
loss
,
startup_program
=
None
,
...
...
python/paddle/fluid/tests/unittests/test_fleet_distributed_strategy.py
浏览文件 @
a96d54ac
...
...
@@ -178,20 +178,20 @@ class TestStrategyConfig(unittest.TestCase):
strategy
.
lamb
=
"True"
self
.
assertEqual
(
strategy
.
lamb
,
False
)
def
test_a
sync_update
(
self
):
def
test_a
_sync
(
self
):
strategy
=
paddle
.
fleet
.
DistributedStrategy
()
strategy
.
a
sync_update
=
True
self
.
assertEqual
(
strategy
.
a
sync_update
,
True
)
strategy
.
a
sync_update
=
False
self
.
assertEqual
(
strategy
.
a
sync_update
,
False
)
strategy
.
a
sync_update
=
"True"
self
.
assertEqual
(
strategy
.
a
sync_update
,
False
)
def
test_async_configs
(
self
):
strategy
.
a
_sync
=
True
self
.
assertEqual
(
strategy
.
a
_sync
,
True
)
strategy
.
a
_sync
=
False
self
.
assertEqual
(
strategy
.
a
_sync
,
False
)
strategy
.
a
_sync
=
"True"
self
.
assertEqual
(
strategy
.
a
_sync
,
False
)
def
test_a
_
sync_configs
(
self
):
strategy
=
paddle
.
fleet
.
DistributedStrategy
()
configs
=
{
"k_steps"
:
1000
}
strategy
.
a
sync_update
_configs
=
configs
self
.
assertEqual
(
strategy
.
a
sync_update
_configs
[
"k_steps"
],
1000
)
strategy
.
a
_sync
_configs
=
configs
self
.
assertEqual
(
strategy
.
a
_sync
_configs
[
"k_steps"
],
1000
)
def
test_elastic
(
self
):
strategy
=
paddle
.
fleet
.
DistributedStrategy
()
...
...
@@ -213,7 +213,7 @@ class TestStrategyConfig(unittest.TestCase):
def
test_strategy_prototxt
(
self
):
strategy
=
paddle
.
fleet
.
DistributedStrategy
()
strategy
.
a
sync_update
=
True
strategy
.
a
_sync
=
True
strategy
.
localsgd
=
True
strategy
.
dgc
=
True
localsgd_configs
=
{
"k_steps"
:
5
}
...
...
python/paddle/fluid/tests/unittests/test_fleet_meta_optimizer.py
浏览文件 @
a96d54ac
...
...
@@ -25,6 +25,27 @@ class TestFleetMetaOptimizer(unittest.TestCase):
os
.
environ
[
"PADDLE_PSERVERS_IP_PORT_LIST"
]
=
\
"127.0.0.1:36001,127.0.0.2:36001"
def
test_graph_execution_optimizer_not_apply
(
self
):
import
paddle.fleet
as
fleet
import
paddle.fluid.incubate.fleet.base.role_maker
as
role_maker
role
=
role_maker
.
PaddleCloudRoleMaker
()
fleet
.
init
(
role
)
input_x
=
paddle
.
fluid
.
layers
.
data
(
name
=
"x"
,
shape
=
[
32
],
dtype
=
'float32'
)
input_y
=
paddle
.
fluid
.
layers
.
data
(
name
=
"y"
,
shape
=
[
1
],
dtype
=
'int64'
)
fc_1
=
paddle
.
fluid
.
layers
.
fc
(
input
=
input_x
,
size
=
64
,
act
=
'tanh'
)
fc_2
=
paddle
.
fluid
.
layers
.
fc
(
input
=
fc_1
,
size
=
64
,
act
=
'tanh'
)
prediction
=
paddle
.
fluid
.
layers
.
fc
(
input
=
[
fc_2
],
size
=
2
,
act
=
'softmax'
)
cost
=
paddle
.
fluid
.
layers
.
cross_entropy
(
input
=
prediction
,
label
=
input_y
)
avg_cost
=
paddle
.
fluid
.
layers
.
mean
(
x
=
cost
)
strategy
=
paddle
.
fleet
.
DistributedStrategy
()
optimizer
=
paddle
.
optimizer
.
SGD
(
learning_rate
=
0.01
)
optimizer
=
fleet
.
distributed_optimizer
(
optimizer
,
strategy
=
strategy
)
optimizer
.
minimize
(
avg_cost
)
def
test_graph_execution_optimizer
(
self
):
import
paddle.fleet
as
fleet
import
paddle.fluid.incubate.fleet.base.role_maker
as
role_maker
...
...
@@ -42,7 +63,7 @@ class TestFleetMetaOptimizer(unittest.TestCase):
avg_cost
=
paddle
.
fluid
.
layers
.
mean
(
x
=
cost
)
strategy
=
paddle
.
fleet
.
DistributedStrategy
()
strategy
.
nccl_comm_num
=
2
optimizer
=
paddle
.
optimizer
.
SGD
(
learning_rate
=
0.01
)
optimizer
=
fleet
.
distributed_optimizer
(
optimizer
,
strategy
=
strategy
)
optimizer
.
minimize
(
avg_cost
)
...
...
@@ -65,7 +86,7 @@ class TestFleetMetaOptimizer(unittest.TestCase):
strategy
=
paddle
.
fleet
.
DistributedStrategy
()
strategy
.
recompute
=
True
strategy
.
recompute_c
heckpoints
=
[
fc_2
]
strategy
.
recompute_c
onfigs
=
{
"checkpoints"
:
[
"fc2"
]}
optimizer
=
paddle
.
optimizer
.
SGD
(
learning_rate
=
0.01
)
optimizer
=
fleet
.
distributed_optimizer
(
optimizer
,
strategy
=
strategy
)
...
...
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