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
毕竟曾有刹那
Mace
提交
d7533c48
Mace
项目概览
毕竟曾有刹那
/
Mace
与 Fork 源项目一致
Fork自
Xiaomi / Mace
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
DevOps
流水线
流水线任务
计划
Wiki
0
Wiki
分析
仓库
DevOps
项目成员
Pages
Mace
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
DevOps
DevOps
流水线
流水线任务
计划
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
流水线任务
提交
Issue看板
体验新版 GitCode,发现更多精彩内容 >>
提交
d7533c48
编写于
11月 28, 2019
作者:
叶
叶剑武
浏览文件
操作
浏览文件
下载
差异文件
Merge branch 'half' into 'master'
Fix half storage See merge request !1233
上级
86878fbc
23934855
变更
3
隐藏空白更改
内联
并排
Showing
3 changed file
with
15 addition
and
7 deletion
+15
-7
docs/user_guide/advanced_usage.rst
docs/user_guide/advanced_usage.rst
+7
-3
docs/user_guide/advanced_usage_cmake.rst
docs/user_guide/advanced_usage_cmake.rst
+7
-3
tools/python/utils/config_parser.py
tools/python/utils/config_parser.py
+1
-1
未找到文件。
docs/user_guide/advanced_usage.rst
浏览文件 @
d7533c48
...
...
@@ -577,12 +577,16 @@ so MACE provides several ways to reduce the model size with no or little perform
**
1.
Save
model
weights
in
half
-
precision
floating
point
format
**
The
d
efault
d
ata
type
of
a
regular
model
is
float
(
32
bit
).
To
reduce
the
model
weights
size
,
The
data
type
of
a
regular
model
is
float
(
32
bit
).
To
reduce
the
model
weights
size
,
half
(
16
bit
)
can
be
used
to
reduce
it
by
half
with
negligible
accuracy
degradation
.
Therefore
,
the
default
storage
type
for
a
regular
model
in
MACE
is
half
.
However
,
if
the
model
is
very
sensitive
to
accuracy
,
storage
type
can
be
changed
to
float
.
For
CPU
,
``
data_type
``
can
be
specified
as
``
fp16_fp32
``
in
the
deployment
file
to
save
the
weights
in
half
and
actual
inference
in
float
.
In
the
deployment
file
,
``
data_type
``
is
``
fp16_fp32
``
by
default
and
can
be
changed
to
``
fp32_fp32
``
.
For
GPU
,
``
fp16_fp32
``
is
default
.
The
ops
in
GPU
take
half
as
inputs
and
outputs
while
kernel
execution
in
float
.
For
CPU
,
``
fp16_fp32
``
means
that
the
weights
are
saved
in
half
and
actual
inference
is
in
float
.
For
GPU
,
``
fp16_fp32
``
means
that
the
ops
in
GPU
take
half
as
inputs
and
outputs
while
kernel
execution
in
float
.
**
2.
Save
model
weights
in
quantized
fixed
point
format
**
...
...
docs/user_guide/advanced_usage_cmake.rst
浏览文件 @
d7533c48
...
...
@@ -406,12 +406,16 @@ so MACE provides several ways to reduce the model size with no or little perform
**1. Save model weights in half-precision floating point format**
The d
efault d
ata type of a regular model is float (32bit). To reduce the model weights size,
The data type of a regular model is float (32bit). To reduce the model weights size,
half (16bit) can be used to reduce it by half with negligible accuracy degradation.
Therefore, the default storage type for a regular model in MACE is half. However,
if the model is very sensitive to accuracy, storage type can be changed to float.
For CPU, ``data_type`` can be specified as ``fp16_fp32`` in the deployment file to save the weights in half and actual inference in float
.
In the deployment file, ``data_type`` is ``fp16_fp32`` by default and can be changed to ``fp32_fp32``
.
For GPU, ``fp16_fp32`` is default. The ops in GPU take half as inputs and outputs while kernel execution in float.
For CPU, ``fp16_fp32`` means that the weights are saved in half and actual inference is in float.
For GPU, ``fp16_fp32`` means that the ops in GPU take half as inputs and outputs while kernel execution in float.
**2. Save model weights in quantized fixed point format**
...
...
tools/python/utils/config_parser.py
浏览文件 @
d7533c48
...
...
@@ -204,7 +204,7 @@ def normalize_model_config(conf):
conf
[
ModelKeys
.
platform
]
=
parse_platform
(
conf
[
ModelKeys
.
platform
])
conf
[
ModelKeys
.
runtime
]
=
parse_device_type
(
conf
[
ModelKeys
.
runtime
])
if
ModelKeys
.
quantize
in
conf
:
if
ModelKeys
.
quantize
in
conf
and
conf
[
ModelKeys
.
quantize
]
==
1
:
conf
[
ModelKeys
.
data_type
]
=
mace_pb2
.
DT_FLOAT
else
:
if
ModelKeys
.
data_type
in
conf
:
...
...
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