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
TonyTonyFun
Paddle
提交
a2b39320
P
Paddle
项目概览
TonyTonyFun
/
Paddle
与 Fork 源项目一致
Fork自
PaddlePaddle / Paddle
通知
1
Star
0
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
前往新版Gitcode,体验更适合开发者的 AI 搜索 >>
未验证
提交
a2b39320
编写于
7月 22, 2022
作者:
X
xiaoxiaohehe001
提交者:
GitHub
7月 22, 2022
浏览文件
操作
浏览文件
下载
电子邮件补丁
差异文件
shufflechannelfix (#44516)
上级
4f86092b
变更
1
隐藏空白更改
内联
并排
Showing
1 changed file
with
6 addition
and
9 deletion
+6
-9
paddle/fluid/inference/tensorrt/convert/shuffle_channel_op.cc
...le/fluid/inference/tensorrt/convert/shuffle_channel_op.cc
+6
-9
未找到文件。
paddle/fluid/inference/tensorrt/convert/shuffle_channel_op.cc
浏览文件 @
a2b39320
...
...
@@ -46,26 +46,23 @@ class ShuffleChannelOpConverter : public OpConverter {
#if IS_TRT_VERSION_GE(8000)
if
(
engine_
->
with_dynamic_shape
())
{
auto
*
input_shape_tensor
=
Shape
(
input
);
auto
*
batch_shape_tensor
=
GetEleTensorOfShape
(
input_shape_tensor
,
0
);
auto
*
channel_shape_tensor
=
GetEleTensorOfShape
(
input_shape_tensor
,
1
);
auto
*
group_tensor
=
Add1DConstantLayer
(
group
,
output_name
+
"_group_tensor_"
);
auto
*
new_channel_shape_tensor
=
Div
(
channel_shape_tensor
,
group_tensor
);
std
::
vector
<
int32_t
>
shape_dim
3
{
0
,
2
,
3
};
auto
*
shape_dim
3_tensor
=
Gather
(
input_shape_tensor
,
shape_dim3
);
std
::
vector
<
int32_t
>
shape_dim
2
{
2
,
3
};
auto
*
shape_dim
2_tensor
=
Gather
(
input_shape_tensor
,
shape_dim2
);
std
::
vector
<
nvinfer1
::
ITensor
*>
itensors
;
itensors
.
push_back
(
shape_dim3
_tensor
);
itensors
.
push_back
(
batch_shape
_tensor
);
itensors
.
push_back
(
group_tensor
);
itensors
.
push_back
(
new_channel_shape_tensor
);
itensors
.
push_back
(
shape_dim2_tensor
);
auto
*
reshape_tensor
=
Concat
(
itensors
);
auto
*
reshape_layer
=
TRT_ENGINE_ADD_LAYER
(
engine_
,
Shuffle
,
*
reshape_tensor
);
nvinfer1
::
Permutation
transpose_new_input
{
0
,
3
,
4
,
1
,
2
};
reshape_layer
->
setSecondTranspose
(
transpose_new_input
);
auto
*
layer
=
TRT_ENGINE_ADD_LAYER
(
engine_
,
Shuffle
,
*
input
);
layer
->
setInput
(
1
,
*
(
reshape_
layer
->
getOutput
(
0
)
));
layer
->
setInput
(
1
,
*
(
reshape_
tensor
));
nvinfer1
::
Permutation
transpose_embed
{
0
,
2
,
1
,
3
,
4
};
layer
->
setSecondTranspose
(
transpose_embed
);
auto
*
output
=
layer
->
getOutput
(
0
);
...
...
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