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
2301_77200941
mindspore
提交
5d77480c
M
mindspore
项目概览
2301_77200941
/
mindspore
与 Fork 源项目一致
Fork自
MindSpore / mindspore
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
M
mindspore
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 >>
提交
5d77480c
编写于
5月 25, 2020
作者:
C
chengxianbin
浏览文件
操作
浏览文件
下载
电子邮件补丁
差异文件
modify yolov3&ssd shell script
上级
89a86611
变更
5
隐藏空白更改
内联
并排
Showing
5 changed file
with
121 addition
and
41 deletion
+121
-41
example/ssd_coco2017/run_distribute_train.sh
example/ssd_coco2017/run_distribute_train.sh
+37
-11
example/ssd_coco2017/train.py
example/ssd_coco2017/train.py
+11
-7
example/yolov3_coco2017/run_distribute_train.sh
example/yolov3_coco2017/run_distribute_train.sh
+42
-13
example/yolov3_coco2017/run_standalone_train.sh
example/yolov3_coco2017/run_standalone_train.sh
+20
-5
example/yolov3_coco2017/train.py
example/yolov3_coco2017/train.py
+11
-5
未找到文件。
example/ssd_coco2017/run_distribute_train.sh
浏览文件 @
5d77480c
...
...
@@ -14,13 +14,20 @@
# limitations under the License.
# ============================================================================
echo
"=============================================================================================================="
echo
"==============================================================================================================
===
"
echo
"Please run the scipt as: "
echo
"sh run_distribute_train.sh DEVICE_NUM EPOCH_SIZE
MINDSPORE_HCCL_CONFIG_PATH
"
echo
"for example: sh run_distribute_train.sh 8
150 coco /data/hccl.json
"
echo
"sh run_distribute_train.sh DEVICE_NUM EPOCH_SIZE
DATASET MINDSPORE_HCCL_CONFIG_PATH PRE_TRAINED PRE_TRAINED_EPOCH_SIZE
"
echo
"for example: sh run_distribute_train.sh 8
350 coco /data/hccl.json /opt/ssd-300.ckpt(optional) 200(optional)
"
echo
"It is better to use absolute path."
echo
"The learning rate is 0.4 as default, if you want other lr, please change the value in this script."
echo
"=============================================================================================================="
echo
"================================================================================================================="
if
[
$#
!=
4
]
&&
[
$#
!=
6
]
then
echo
"Usage: sh run_distribute_train.sh [DEVICE_NUM] [EPOCH_SIZE] [DATASET]
\
[MINDSPORE_HCCL_CONFIG_PATH] [PRE_TRAINED](optional) [PRE_TRAINED_EPOCH_SIZE](optional)"
exit
1
fi
# Before start distribute train, first create mindrecord files.
python train.py
--only_create_dataset
=
1
...
...
@@ -30,6 +37,8 @@ echo "After running the scipt, the network runs in the background. The log will
export
RANK_SIZE
=
$1
EPOCH_SIZE
=
$2
DATASET
=
$3
PRE_TRAINED
=
$5
PRE_TRAINED_EPOCH_SIZE
=
$6
export
MINDSPORE_HCCL_CONFIG_PATH
=
$4
...
...
@@ -43,12 +52,29 @@ do
export
RANK_ID
=
$i
echo
"start training for rank
$i
, device
$DEVICE_ID
"
env
>
env.log
python ../train.py
\
--distribute
=
1
\
--lr
=
0.4
\
--dataset
=
$DATASET
\
--device_num
=
$RANK_SIZE
\
--device_id
=
$DEVICE_ID
\
--epoch_size
=
$EPOCH_SIZE
>
log.txt 2>&1 &
if
[
$#
==
4
]
then
python ../train.py
\
--distribute
=
1
\
--lr
=
0.4
\
--dataset
=
$DATASET
\
--device_num
=
$RANK_SIZE
\
--device_id
=
$DEVICE_ID
\
--epoch_size
=
$EPOCH_SIZE
>
log.txt 2>&1 &
fi
if
[
$#
==
6
]
then
python ../train.py
\
--distribute
=
1
\
--lr
=
0.4
\
--dataset
=
$DATASET
\
--device_num
=
$RANK_SIZE
\
--device_id
=
$DEVICE_ID
\
--pre_trained
=
$PRE_TRAINED
\
--pre_trained_epoch_size
=
$PRE_TRAINED_EPOCH_SIZE
\
--epoch_size
=
$EPOCH_SIZE
>
log.txt 2>&1 &
fi
cd
../
done
example/ssd_coco2017/train.py
浏览文件 @
5d77480c
...
...
@@ -88,6 +88,7 @@ def main():
parser
.
add_argument
(
"--epoch_size"
,
type
=
int
,
default
=
70
,
help
=
"Epoch size, default is 70."
)
parser
.
add_argument
(
"--batch_size"
,
type
=
int
,
default
=
32
,
help
=
"Batch size, default is 32."
)
parser
.
add_argument
(
"--pre_trained"
,
type
=
str
,
default
=
None
,
help
=
"Pretrained Checkpoint file path."
)
parser
.
add_argument
(
"--pre_trained_epoch_size"
,
type
=
int
,
default
=
0
,
help
=
"Pretrained epoch size."
)
parser
.
add_argument
(
"--save_checkpoint_epochs"
,
type
=
int
,
default
=
5
,
help
=
"Save checkpoint epochs, default is 5."
)
parser
.
add_argument
(
"--loss_scale"
,
type
=
int
,
default
=
1024
,
help
=
"Loss scale, default is 1024."
)
args_opt
=
parser
.
parse_args
()
...
...
@@ -150,17 +151,20 @@ def main():
ckpt_config
=
CheckpointConfig
(
save_checkpoint_steps
=
dataset_size
*
args_opt
.
save_checkpoint_epochs
)
ckpoint_cb
=
ModelCheckpoint
(
prefix
=
"ssd"
,
directory
=
None
,
config
=
ckpt_config
)
lr
=
Tensor
(
get_lr
(
global_step
=
0
,
lr_init
=
0
,
lr_end
=
0
,
lr_max
=
args_opt
.
lr
,
warmup_epochs
=
max
(
args_opt
.
epoch_size
//
20
,
1
),
total_epochs
=
args_opt
.
epoch_size
,
steps_per_epoch
=
dataset_size
))
opt
=
nn
.
Momentum
(
filter
(
lambda
x
:
x
.
requires_grad
,
net
.
get_parameters
()),
lr
,
0.9
,
0.0001
,
loss_scale
)
net
=
TrainingWrapper
(
net
,
opt
,
loss_scale
)
if
args_opt
.
pre_trained
:
if
args_opt
.
pre_trained_epoch_size
<=
0
:
raise
KeyError
(
"pre_trained_epoch_size must be greater than 0."
)
param_dict
=
load_checkpoint
(
args_opt
.
pre_trained
)
load_param_into_net
(
net
,
param_dict
)
lr
=
Tensor
(
get_lr
(
global_step
=
args_opt
.
pre_trained_epoch_size
*
dataset_size
,
lr_init
=
0
,
lr_end
=
0
,
lr_max
=
args_opt
.
lr
,
warmup_epochs
=
max
(
350
//
20
,
1
),
total_epochs
=
350
,
steps_per_epoch
=
dataset_size
))
opt
=
nn
.
Momentum
(
filter
(
lambda
x
:
x
.
requires_grad
,
net
.
get_parameters
()),
lr
,
0.9
,
0.0001
,
loss_scale
)
net
=
TrainingWrapper
(
net
,
opt
,
loss_scale
)
callback
=
[
TimeMonitor
(
data_size
=
dataset_size
),
LossMonitor
(),
ckpoint_cb
]
model
=
Model
(
net
)
...
...
example/yolov3_coco2017/run_distribute_train.sh
浏览文件 @
5d77480c
...
...
@@ -14,18 +14,27 @@
# limitations under the License.
# ============================================================================
echo
"=============================================================================================================="
echo
"==============================================================================================================
=========================================
"
echo
"Please run the scipt as: "
echo
"sh run_distribute_train.sh DEVICE_NUM EPOCH_SIZE MINDRECORD_DIR IMAGE_DIR ANNO_PATH MINDSPORE_HCCL_CONFIG_PATH"
echo
"
for example: sh run_distribute_train.sh 8 100 /data/Mindrecord_train /data /data/train.txt /data/hccl.json
"
echo
"sh run_distribute_train.sh DEVICE_NUM EPOCH_SIZE MINDRECORD_DIR IMAGE_DIR ANNO_PATH MINDSPORE_HCCL_CONFIG_PATH
PRE_TRAINED PRE_TRAINED_EPOCH_SIZE
"
echo
"
For example: sh run_distribute_train.sh 8 150 /data/Mindrecord_train /data /data/train.txt /data/hccl.json /opt/yolov3-150.ckpt(optional) 100(optional)
"
echo
"It is better to use absolute path."
echo
"The learning rate is 0.005 as default, if you want other lr, please change the value in this script."
echo
"=============================================================================================================="
echo
"======================================================================================================================================================="
if
[
$#
!=
6
]
&&
[
$#
!=
8
]
then
echo
"Usage: sh run_distribute_train.sh [DEVICE_NUM] [EPOCH_SIZE] [MINDRECORD_DIR] [IMAGE_DIR] [ANNO_PATH] [MINDSPORE_HCCL_CONFIG_PATH]
\
[PRE_TRAINED](optional) [PRE_TRAINED_EPOCH_SIZE](optional)"
exit
1
fi
EPOCH_SIZE
=
$2
MINDRECORD_DIR
=
$3
IMAGE_DIR
=
$4
ANNO_PATH
=
$5
PRE_TRAINED
=
$7
PRE_TRAINED_EPOCH_SIZE
=
$8
# Before start distribute train, first create mindrecord files.
python train.py
--only_create_dataset
=
1
--mindrecord_dir
=
$MINDRECORD_DIR
--image_dir
=
$IMAGE_DIR
\
...
...
@@ -51,14 +60,34 @@ do
export
RANK_ID
=
$i
echo
"start training for rank
$i
, device
$DEVICE_ID
"
env
>
env.log
taskset
-c
$cmdopt
python ../train.py
\
--distribute
=
1
\
--lr
=
0.005
\
--device_num
=
$RANK_SIZE
\
--device_id
=
$DEVICE_ID
\
--mindrecord_dir
=
$MINDRECORD_DIR
\
--image_dir
=
$IMAGE_DIR
\
--epoch_size
=
$EPOCH_SIZE
\
--anno_path
=
$ANNO_PATH
>
log.txt 2>&1 &
if
[
$#
==
6
]
then
taskset
-c
$cmdopt
python ../train.py
\
--distribute
=
1
\
--lr
=
0.005
\
--device_num
=
$RANK_SIZE
\
--device_id
=
$DEVICE_ID
\
--mindrecord_dir
=
$MINDRECORD_DIR
\
--image_dir
=
$IMAGE_DIR
\
--epoch_size
=
$EPOCH_SIZE
\
--anno_path
=
$ANNO_PATH
>
log.txt 2>&1 &
fi
if
[
$#
==
8
]
then
taskset
-c
$cmdopt
python ../train.py
\
--distribute
=
1
\
--lr
=
0.005
\
--device_num
=
$RANK_SIZE
\
--device_id
=
$DEVICE_ID
\
--mindrecord_dir
=
$MINDRECORD_DIR
\
--image_dir
=
$IMAGE_DIR
\
--epoch_size
=
$EPOCH_SIZE
\
--pre_trained
=
$PRE_TRAINED
\
--pre_trained_epoch_size
=
$PRE_TRAINED_EPOCH_SIZE
\
--anno_path
=
$ANNO_PATH
>
log.txt 2>&1 &
fi
cd
../
done
example/yolov3_coco2017/run_standalone_train.sh
浏览文件 @
5d77480c
...
...
@@ -14,10 +14,25 @@
# limitations under the License.
# ============================================================================
echo
"=============================================================================================================="
echo
"==============================================================================================================
===========================
"
echo
"Please run the scipt as: "
echo
"sh run_standalone_train.sh DEVICE_ID EPOCH_SIZE MINDRECORD_DIR IMAGE_DIR ANNO_PATH"
echo
"for example: sh run_standalone_train.sh 0 50 ./Mindrecord_train ./dataset ./dataset/train.txt"
echo
"=============================================================================================================="
echo
"sh run_standalone_train.sh DEVICE_ID EPOCH_SIZE MINDRECORD_DIR IMAGE_DIR ANNO_PATH
PRE_TRAINED PRE_TRAINED_EPOCH_SIZE
"
echo
"for example: sh run_standalone_train.sh 0 50 ./Mindrecord_train ./dataset ./dataset/train.txt
/opt/yolov3-50.ckpt(optional) 30(optional)
"
echo
"==============================================================================================================
===========================
"
python train.py
--device_id
=
$1
--epoch_size
=
$2
--mindrecord_dir
=
$3
--image_dir
=
$4
--anno_path
=
$5
if
[
$#
!=
5
]
&&
[
$#
!=
7
]
then
echo
"Usage: sh run_standalone_train.sh [DEVICE_ID] [EPOCH_SIZE] [MINDRECORD_DIR] [IMAGE_DIR] [ANNO_PATH]
\
[PRE_TRAINED](optional) [PRE_TRAINED_EPOCH_SIZE](optional)"
exit
1
fi
if
[
$#
==
5
]
then
python train.py
--device_id
=
$1
--epoch_size
=
$2
--mindrecord_dir
=
$3
--image_dir
=
$4
--anno_path
=
$5
fi
if
[
$#
==
7
]
then
python train.py
--device_id
=
$1
--epoch_size
=
$2
--mindrecord_dir
=
$3
--image_dir
=
$4
--anno_path
=
$5
--pre_trained
=
$6
--pre_trained_epoch_size
=
$7
fi
example/yolov3_coco2017/train.py
浏览文件 @
5d77480c
...
...
@@ -71,6 +71,7 @@ def main():
parser
.
add_argument
(
"--epoch_size"
,
type
=
int
,
default
=
10
,
help
=
"Epoch size, default is 10"
)
parser
.
add_argument
(
"--batch_size"
,
type
=
int
,
default
=
32
,
help
=
"Batch size, default is 32."
)
parser
.
add_argument
(
"--pre_trained"
,
type
=
str
,
default
=
None
,
help
=
"Pretrained checkpoint file path"
)
parser
.
add_argument
(
"--pre_trained_epoch_size"
,
type
=
int
,
default
=
0
,
help
=
"Pretrained epoch size"
)
parser
.
add_argument
(
"--save_checkpoint_epochs"
,
type
=
int
,
default
=
5
,
help
=
"Save checkpoint epochs, default is 5."
)
parser
.
add_argument
(
"--loss_scale"
,
type
=
int
,
default
=
1024
,
help
=
"Loss scale, default is 1024."
)
parser
.
add_argument
(
"--mindrecord_dir"
,
type
=
str
,
default
=
"./Mindrecord_train"
,
...
...
@@ -133,14 +134,19 @@ def main():
ckpt_config
=
CheckpointConfig
(
save_checkpoint_steps
=
dataset_size
*
args_opt
.
save_checkpoint_epochs
)
ckpoint_cb
=
ModelCheckpoint
(
prefix
=
"yolov3"
,
directory
=
None
,
config
=
ckpt_config
)
lr
=
Tensor
(
get_lr
(
learning_rate
=
args_opt
.
lr
,
start_step
=
0
,
global_step
=
args_opt
.
epoch_size
*
dataset_size
,
decay_step
=
1000
,
decay_rate
=
0.95
,
steps
=
True
))
opt
=
nn
.
Adam
(
filter
(
lambda
x
:
x
.
requires_grad
,
net
.
get_parameters
()),
lr
,
loss_scale
=
loss_scale
)
net
=
TrainingWrapper
(
net
,
opt
,
loss_scale
)
if
args_opt
.
pre_trained
:
if
args_opt
.
pre_trained_epoch_size
<=
0
:
raise
KeyError
(
"pre_trained_epoch_size must be greater than 0."
)
param_dict
=
load_checkpoint
(
args_opt
.
pre_trained
)
load_param_into_net
(
net
,
param_dict
)
total_epoch_size
=
60
if
args_opt
.
distribute
:
total_epoch_size
=
160
lr
=
Tensor
(
get_lr
(
learning_rate
=
args_opt
.
lr
,
start_step
=
args_opt
.
pre_trained_epoch_size
*
dataset_size
,
global_step
=
total_epoch_size
*
dataset_size
,
decay_step
=
1000
,
decay_rate
=
0.95
,
steps
=
True
))
opt
=
nn
.
Adam
(
filter
(
lambda
x
:
x
.
requires_grad
,
net
.
get_parameters
()),
lr
,
loss_scale
=
loss_scale
)
net
=
TrainingWrapper
(
net
,
opt
,
loss_scale
)
callback
=
[
TimeMonitor
(
data_size
=
dataset_size
),
LossMonitor
(),
ckpoint_cb
]
...
...
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