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
路上走着呢
Paddle
提交
e43614d7
P
Paddle
项目概览
路上走着呢
/
Paddle
与 Fork 源项目一致
Fork自
PaddlePaddle / Paddle
通知
1
Star
1
Fork
0
代码
文件
提交
分支
Tags
贡献者
分支图
Diff
Issue
0
列表
看板
标记
里程碑
合并请求
0
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
0
Issue
0
列表
看板
标记
里程碑
合并请求
0
合并请求
0
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
体验新版 GitCode,发现更多精彩内容 >>
未验证
提交
e43614d7
编写于
6月 20, 2018
作者:
Y
Yu Yang
提交者:
GitHub
6月 20, 2018
浏览文件
操作
浏览文件
下载
差异文件
Merge pull request #11554 from chengduoZH/fix_doc_parallel_execturo
Refine ParallelExe Doc
上级
3587d921
73f224d0
变更
1
隐藏空白更改
内联
并排
Showing
1 changed file
with
67 addition
and
39 deletion
+67
-39
python/paddle/fluid/parallel_executor.py
python/paddle/fluid/parallel_executor.py
+67
-39
未找到文件。
python/paddle/fluid/parallel_executor.py
浏览文件 @
e43614d7
...
...
@@ -27,6 +27,40 @@ BuildStrategy = core.ParallelExecutor.BuildStrategy
class
ParallelExecutor
(
object
):
"""
ParallelExecutor can run program in parallel.
Args:
use_cuda (bool): Whether to use CUDA or not.
loss_name (str): The loss name must set in training. Default None.
main_program (Program): The program that need to run, if not provided,
then default_main_program will be used. Default None.
share_vars_from(ParallelExecutor): If provied, it will share variables
from the specified ParallelExecutor. Default None.
num_trainers(int): If greater than 1, NCCL will be initialized with
multiple rank of nodes, each node should have same number of GPUs.
Distributed training will be enabled then. Default 1.
trainer_id(int: Must use together with num_trainers. trainer_id is the
"rank" of current node starts from 0. Default 0.
Returns:
ParallelExecutor: The initialized ParallelExecutor object.
Raises:
TypeError: If share_vars_from is provided, but not ParallelExecutor object.
Examples:
.. code-block:: python
train_exe = fluid.ParallelExecutor(use_cuda=True, loss_name=loss.name)
test_exe = fluid.ParallelExecutor(use_cuda=True,
main_program=test_program,
share_vars_from=train_exe)
train_loss, = train_exe.run([loss.name], feed=feed_dict)
test_loss, = test_exe.run([loss.name], feed=feed_dict)
"""
def
__init__
(
self
,
use_cuda
,
loss_name
=
None
,
...
...
@@ -37,42 +71,7 @@ class ParallelExecutor(object):
num_trainers
=
1
,
trainer_id
=
0
,
**
kwargs
):
"""
ParallelExecutor can run program in parallel.
Args:
use_cuda(bool): Whether to use CUDA or not.
loss_name(str, default None): The loss name must set in training.
main_program(Program, default None): The program that need to run,
if not provided, then default_main_program will be used.
share_vars_from(ParallelExecutor, default None): If provied,
it will share variables from the specified ParallelExecutor.
num_trainers(int, default 1): If greater than 1, NCCL will be
initialized with multpile rank of nodes, each node should have
same number of GPUs. Distributed training will be enabled then.
trainer_id(int, default 0): Must use together with num_trainers.
trainer_id is the "rank" of current node starts from 0.
Returns:
A ParallelExecutor object.
Raises:
TypeError: If share_vars_from is provided, but not ParallelExecutor
object.
Examples:
.. code-block:: python
train_exe = fluid.ParallelExecutor(
use_cuda=True, loss_name=loss.name)
test_exe = fluid.ParallelExecutor(
use_cuda=True,
main_program=test_program,
share_vars_from=train_exe)
train_loss, = train_exe.run([loss.name], feed=feed_dict)
test_loss, = test_exe.run([loss.name], feed=feed_dict)
"""
if
len
(
kwargs
)
!=
0
:
err_msg
=
""
for
key
in
kwargs
:
...
...
@@ -135,6 +134,7 @@ class ParallelExecutor(object):
if
share_vars_from
and
not
isinstance
(
share_vars_from
,
ParallelExecutor
):
raise
TypeError
(
"share_vars_from must be ParallelExecutor."
)
local_scopes
=
share_vars_from
.
executor
.
local_scopes
(
)
if
share_vars_from
else
[]
...
...
@@ -166,12 +166,14 @@ class ParallelExecutor(object):
element in the list will be copied to each device directly.
For example, if the feed is a dict:
>>> exe = ParallelExecutor()
>>> # the image will be splitted into devices. If there is two devices
>>> # each device will process an image with shape (24, 1, 28, 28)
>>> exe.run(feed={'image': numpy.random.random(size=(48, 1, 28, 28))})
For example, if the feed is a list:
>>> exe = ParallelExecutor()
>>> # each device will process each element in the list.
>>> # the 1st device will process an image with shape (48, 1, 28, 28)
...
...
@@ -182,18 +184,40 @@ class ParallelExecutor(object):
>>> {"image": numpy.random.random(size=(32, 1, 28, 28))},
>>> ])
Args:
fetch_list(list): The fetched variable names
feed(list|dict|None): The feed variables. If the feed is a dict,
tensors in that dict will be splitted into each devices. If
the feed is a list, each element of the list will be copied
to each device.
to each device.
Default None.
feed_dict: Alias for feed parameter, for backward compatibility.
This parameter is deprecated.
This parameter has been deprecated. Default None.
Returns:
List: The fetched result list.
Returns: fetched result list.
Raises:
ValueError: If the feed is a list, but its length is not equal the
length of active places, or its element's is not dict.
NOTES:
1. If the feed's type is dict, the number of data that feeds to
ParallelExecutor must be bigger than active places. Otherwise,
it will throw exception from C++ side. Special attention should be
paid to check whether the last batch of the dataset is bigger
than active places.
2. If active places are more than one, the fetch results for each
variable is a list, and each element of this list is the variable of
respective active place.
Examples:
.. code-block:: python
pe = fluid.ParallelExecutor(use_cuda=use_cuda,
loss_name=avg_cost.name,
main_program=fluid.default_main_program())
loss = pe.run(feed=feeder.feed(cur_batch),
fetch_list=[avg_cost.name]))
"""
if
feed
is
None
and
feed_dict
is
not
None
:
feed
=
feed_dict
...
...
@@ -241,6 +265,10 @@ class ParallelExecutor(object):
return
[
arr
[
i
]
for
i
in
range
(
len
(
arr
))]
def
bcast_params
(
self
):
"""
Broadcast the parameters to other devices. It is used during
distributed training.
"""
self
.
executor
.
bcast_params
(
set
(
self
.
persistable_vars
))
@
property
...
...
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