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
机器未来
Paddle
提交
e26411ce
P
Paddle
项目概览
机器未来
/
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
1
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
1
Issue
1
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
未验证
提交
e26411ce
编写于
8月 28, 2019
作者:
C
chengduo
提交者:
GitHub
8月 28, 2019
浏览文件
操作
浏览文件
下载
电子邮件补丁
差异文件
Open test_parallel_dygraph_se_resnext (#19342)
* enabel test_parallel_dygraph_se_resnext test=develop
上级
caf59d0f
变更
2
隐藏空白更改
内联
并排
Showing
2 changed file
with
16 addition
and
11 deletion
+16
-11
python/paddle/fluid/tests/unittests/parallel_dygraph_se_resnext.py
...ddle/fluid/tests/unittests/parallel_dygraph_se_resnext.py
+10
-3
python/paddle/fluid/tests/unittests/test_parallel_dygraph_se_resnext.py
...fluid/tests/unittests/test_parallel_dygraph_se_resnext.py
+6
-8
未找到文件。
python/paddle/fluid/tests/unittests/parallel_dygraph_se_resnext.py
浏览文件 @
e26411ce
...
...
@@ -57,10 +57,15 @@ class ConvBNLayer(fluid.dygraph.Layer):
self
.
_batch_norm
=
BatchNorm
(
self
.
full_name
(),
num_filters
,
act
=
act
,
momentum
=
0.1
)
self
.
_layer_norm
=
fluid
.
dygraph
.
nn
.
LayerNorm
(
self
.
full_name
(),
begin_norm_axis
=
1
)
def
forward
(
self
,
inputs
):
y
=
self
.
_conv
(
inputs
)
y
=
self
.
_batch_norm
(
y
)
# FIXME(zcd): when compare the result of multi-card and single-card,
# we should replace batch_norm with layer_norm.
y
=
self
.
_layer_norm
(
y
)
# y = self._batch_norm(y)
return
y
...
...
@@ -278,7 +283,9 @@ class SeResNeXt(fluid.dygraph.Layer):
for
bottleneck_block
in
self
.
bottleneck_block_list
:
y
=
bottleneck_block
(
y
)
y
=
self
.
pool2d_avg
(
y
)
y
=
fluid
.
layers
.
dropout
(
y
,
dropout_prob
=
0.2
,
seed
=
1
)
# FIXME(zcd): the dropout should be removed when compare the
# result of multi-card and single-card.
# y = fluid.layers.dropout(y, dropout_prob=0.2, seed=1)
cost
=
self
.
fc
(
y
)
loss
=
fluid
.
layers
.
cross_entropy
(
cost
,
label
)
avg_loss
=
fluid
.
layers
.
mean
(
loss
)
...
...
@@ -290,7 +297,7 @@ class TestSeResNeXt(TestParallelDyGraphRunnerBase):
model
=
SeResNeXt
(
"se-resnext"
)
train_reader
=
paddle
.
batch
(
paddle
.
dataset
.
flowers
.
test
(
use_xmap
=
False
),
batch_size
=
2
,
batch_size
=
4
,
drop_last
=
True
)
opt
=
fluid
.
optimizer
.
SGD
(
learning_rate
=
1e-3
)
...
...
python/paddle/fluid/tests/unittests/test_parallel_dygraph_se_resnext.py
浏览文件 @
e26411ce
...
...
@@ -13,10 +13,11 @@
# limitations under the License.
from
__future__
import
print_function
#
import unittest
import
unittest
from
test_dist_base
import
TestDistBase
import
paddle.fluid
as
fluid
'''
class
TestParallelDygraphSeResNeXt
(
TestDistBase
):
def
_setup_config
(
self
):
self
.
_sync_mode
=
False
...
...
@@ -24,12 +25,9 @@ class TestParallelDygraphSeResNeXt(TestDistBase):
self
.
_dygraph
=
True
def
test_se_resnext
(
self
):
# TODO(Yancey1989): BN and Dropout is related with batchsize, so the delta is the 1,
# try to remove the BN and Dropout in the network and using delta = 1e-5
if
fluid
.
core
.
is_compiled_with_cuda
():
self.check_with_place("parallel_dygraph_se_resnext.py", delta=1)
'''
self
.
check_with_place
(
"parallel_dygraph_se_resnext.py"
,
delta
=
0.0
1
)
if
__name__
==
"__main__"
:
pass
#unittest.main()
unittest
.
main
()
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