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
xiaopch
PaddleRec
提交
f8d55c9f
P
PaddleRec
项目概览
xiaopch
/
PaddleRec
与 Fork 源项目一致
Fork自
PaddlePaddle / PaddleRec
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
PaddleRec
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
f8d55c9f
编写于
4月 14, 2020
作者:
T
tangwei
浏览文件
操作
浏览文件
下载
电子邮件补丁
差异文件
code clean
上级
357f0da7
变更
10
隐藏空白更改
内联
并排
Showing
10 changed file
with
45 addition
and
252 deletion
+45
-252
fleetrec/models/base.py
fleetrec/models/base.py
+5
-5
fleetrec/models/ctr_dnn/dataloader.py
fleetrec/models/ctr_dnn/dataloader.py
+0
-83
fleetrec/models/ctr_dnn/dataset.py
fleetrec/models/ctr_dnn/dataset.py
+0
-69
fleetrec/models/ctr_dnn/model.py
fleetrec/models/ctr_dnn/model.py
+10
-25
fleetrec/trainer/cluster_trainer.py
fleetrec/trainer/cluster_trainer.py
+2
-6
fleetrec/trainer/factory.py
fleetrec/trainer/factory.py
+6
-45
fleetrec/trainer/single_trainer.py
fleetrec/trainer/single_trainer.py
+3
-10
fleetrec/trainer/trainer.py
fleetrec/trainer/trainer.py
+0
-3
fleetrec/trainer/transpiler_trainer.py
fleetrec/trainer/transpiler_trainer.py
+7
-5
fleetrec/utils/util.py
fleetrec/utils/util.py
+12
-1
未找到文件。
fleetrec/models/base.py
浏览文件 @
f8d55c9f
...
...
@@ -63,7 +63,7 @@ def create(config):
model
=
None
if
config
[
'mode'
]
==
'fluid'
:
model
=
YamlModel
(
config
)
model
.
build_model
()
model
.
net
()
return
model
...
...
@@ -94,13 +94,13 @@ class Model(object):
return
self
.
_fetch_interval
@
abc
.
abstractmethod
def
shrink
(
self
,
params
):
def
net
(
self
):
"""R
"""
pass
@
abc
.
abstractmethod
def
build_model
(
self
):
def
shrink
(
self
,
params
):
"""R
"""
pass
...
...
@@ -140,7 +140,7 @@ class YamlModel(Model):
self
.
_build_param
=
{
'layer'
:
{},
'inner_layer'
:
{},
'layer_extend'
:
{},
'model'
:
{}}
self
.
_inference_meta
=
{
'dependency'
:
{},
'params'
:
{}}
def
build_model
(
self
):
def
net
(
self
):
"""R
build a fluid model with config
Return:
...
...
@@ -287,4 +287,4 @@ class YamlModel(Model):
dependency_list
=
copy
.
deepcopy
(
dependencys
)
for
dependency
in
dependencys
:
dependency_list
=
dependency_list
+
self
.
get_dependency
(
layer_graph
,
dependency
)
return
list
(
set
(
dependency_list
))
\ No newline at end of file
return
list
(
set
(
dependency_list
))
fleetrec/models/ctr_dnn/dataloader.py
已删除
100644 → 0
浏览文件 @
357f0da7
# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from
...utils
import
envs
# There are 13 integer features and 26 categorical features
continous_features
=
range
(
1
,
14
)
categorial_features
=
range
(
14
,
40
)
continous_clip
=
[
20
,
600
,
100
,
50
,
64000
,
500
,
100
,
50
,
500
,
10
,
10
,
10
,
50
]
class
CriteoDataset
(
object
):
def
__init__
(
self
,
sparse_feature_dim
):
self
.
cont_min_
=
[
0
,
-
3
,
0
,
0
,
0
,
0
,
0
,
0
,
0
,
0
,
0
,
0
,
0
]
self
.
cont_max_
=
[
20
,
600
,
100
,
50
,
64000
,
500
,
100
,
50
,
500
,
10
,
10
,
10
,
50
]
self
.
cont_diff_
=
[
20
,
603
,
100
,
50
,
64000
,
500
,
100
,
50
,
500
,
10
,
10
,
10
,
50
]
self
.
hash_dim_
=
sparse_feature_dim
# here, training data are lines with line_index < train_idx_
self
.
train_idx_
=
41256555
self
.
continuous_range_
=
range
(
1
,
14
)
self
.
categorical_range_
=
range
(
14
,
40
)
def
_reader_creator
(
self
,
file_list
,
is_train
,
trainer_num
,
trainer_id
):
def
reader
():
for
file
in
file_list
:
with
open
(
file
,
'r'
)
as
f
:
line_idx
=
0
for
line
in
f
:
line_idx
+=
1
features
=
line
.
rstrip
(
'
\n
'
).
split
(
'
\t
'
)
dense_feature
=
[]
sparse_feature
=
[]
for
idx
in
self
.
continuous_range_
:
if
features
[
idx
]
==
''
:
dense_feature
.
append
(
0.0
)
else
:
dense_feature
.
append
(
(
float
(
features
[
idx
])
-
self
.
cont_min_
[
idx
-
1
])
/
self
.
cont_diff_
[
idx
-
1
])
for
idx
in
self
.
categorical_range_
:
sparse_feature
.
append
([
hash
(
str
(
idx
)
+
features
[
idx
])
%
self
.
hash_dim_
])
label
=
[
int
(
features
[
0
])]
yield
[
dense_feature
]
+
sparse_feature
+
[
label
]
return
reader
def
train
(
self
,
file_list
,
trainer_num
,
trainer_id
):
return
self
.
_reader_creator
(
file_list
,
True
,
trainer_num
,
trainer_id
)
def
test
(
self
,
file_list
):
return
self
.
_reader_creator
(
file_list
,
False
,
1
,
0
)
def
Train
():
sparse_feature_number
=
envs
.
get_global_env
(
"sparse_feature_number"
)
train_generator
=
CriteoDataset
(
sparse_feature_number
)
return
train_generator
.
train
def
Evaluate
():
sparse_feature_number
=
envs
.
get_global_env
(
"sparse_feature_number"
)
train_generator
=
CriteoDataset
(
sparse_feature_number
)
return
train_generator
.
test
fleetrec/models/ctr_dnn/dataset.py
已删除
100644 → 0
浏览文件 @
357f0da7
# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from
__future__
import
print_function
import
sys
import
paddle.fluid.incubate.data_generator
as
dg
cont_min_
=
[
0
,
-
3
,
0
,
0
,
0
,
0
,
0
,
0
,
0
,
0
,
0
,
0
,
0
]
cont_max_
=
[
20
,
600
,
100
,
50
,
64000
,
500
,
100
,
50
,
500
,
10
,
10
,
10
,
50
]
cont_diff_
=
[
20
,
603
,
100
,
50
,
64000
,
500
,
100
,
50
,
500
,
10
,
10
,
10
,
50
]
hash_dim_
=
1000001
continuous_range_
=
range
(
1
,
14
)
categorical_range_
=
range
(
14
,
40
)
class
CriteoDataset
(
dg
.
MultiSlotDataGenerator
):
"""
DacDataset: inheritance MultiSlotDataGeneratior, Implement data reading
Help document: http://wiki.baidu.com/pages/viewpage.action?pageId=728820675
"""
def
generate_sample
(
self
,
line
):
"""
Read the data line by line and process it as a dictionary
"""
def
reader
():
"""
This function needs to be implemented by the user, based on data format
"""
features
=
line
.
rstrip
(
'
\n
'
).
split
(
'
\t
'
)
dense_feature
=
[]
sparse_feature
=
[]
for
idx
in
continuous_range_
:
if
features
[
idx
]
==
""
:
dense_feature
.
append
(
0.0
)
else
:
dense_feature
.
append
(
(
float
(
features
[
idx
])
-
cont_min_
[
idx
-
1
])
/
cont_diff_
[
idx
-
1
])
for
idx
in
categorical_range_
:
sparse_feature
.
append
(
[
hash
(
str
(
idx
)
+
features
[
idx
])
%
hash_dim_
])
label
=
[
int
(
features
[
0
])]
process_line
=
dense_feature
,
sparse_feature
,
label
feature_name
=
[
"dense_input"
]
for
idx
in
categorical_range_
:
feature_name
.
append
(
"C"
+
str
(
idx
-
13
))
feature_name
.
append
(
"label"
)
yield
zip
(
feature_name
,
[
dense_feature
]
+
sparse_feature
+
[
label
])
return
reader
d
=
CriteoDataset
()
d
.
run_from_stdin
()
fleetrec/models/ctr_dnn/model.py
浏览文件 @
f8d55c9f
...
...
@@ -19,7 +19,7 @@ from fleetrec.utils import envs
from
fleetrec.models.base
import
Model
class
Train
(
Model
):
class
Train
Model
(
Model
):
def
__init__
(
self
,
config
):
Model
.
__init__
(
self
,
config
)
self
.
namespace
=
"train.model"
...
...
@@ -34,7 +34,7 @@ class Train(Model):
lod_level
=
1
,
dtype
=
"int64"
)
for
i
in
range
(
1
,
ids
)
]
return
sparse_input_ids
,
[
var
.
name
for
var
in
sparse_input_ids
]
return
sparse_input_ids
def
dense_input
():
dim
=
envs
.
get_global_env
(
"hyper_parameters.dense_input_dim"
,
None
,
self
.
namespace
)
...
...
@@ -42,23 +42,20 @@ class Train(Model):
dense_input_var
=
fluid
.
layers
.
data
(
name
=
"dense_input"
,
shape
=
[
dim
],
dtype
=
"float32"
)
return
dense_input_var
,
dense_input_var
.
name
return
dense_input_var
def
label_input
():
label
=
fluid
.
layers
.
data
(
name
=
"label"
,
shape
=
[
1
],
dtype
=
"int64"
)
return
label
,
label
.
name
return
label
self
.
sparse_inputs
,
self
.
sparse_input_varnames
=
sparse_inputs
()
self
.
dense_input
,
self
.
dense_input_varname
=
dense_input
()
self
.
label_input
,
self
.
label_input_varname
=
label_input
()
self
.
sparse_inputs
=
sparse_inputs
()
self
.
dense_input
=
dense_input
()
self
.
label_input
=
label_input
()
def
input
_var
s
(
self
):
def
inputs
(
self
):
return
[
self
.
dense_input
]
+
self
.
sparse_inputs
+
[
self
.
label_input
]
def
input_varnames
(
self
):
return
[
input
.
name
for
input
in
self
.
input_vars
()]
def
build_model
(
self
):
def
net
(
self
):
def
embedding_layer
(
input
):
sparse_feature_number
=
envs
.
get_global_env
(
"hyper_parameters.sparse_feature_number"
,
None
,
self
.
namespace
)
sparse_feature_dim
=
envs
.
get_global_env
(
"hyper_parameters.sparse_feature_dim"
,
None
,
self
.
namespace
)
...
...
@@ -120,20 +117,8 @@ class Train(Model):
optimizer
=
fluid
.
optimizer
.
Adam
(
learning_rate
,
lazy_mode
=
True
)
return
optimizer
def
dump_model_program
(
self
,
path
):
pass
def
dump_inference_param
(
self
,
params
):
pass
def
dump_inference_program
(
self
,
inference_layer
,
path
):
pass
def
shrink
(
self
,
params
):
pass
class
Evaluate
(
object
):
class
Evaluate
Model
(
object
):
def
input
(
self
):
pass
...
...
fleetrec/trainer/cluster_trainer.py
浏览文件 @
f8d55c9f
...
...
@@ -32,11 +32,7 @@ logger = logging.getLogger("fluid")
logger
.
setLevel
(
logging
.
INFO
)
class
ClusterTrainerWithDataloader
(
TranspileTrainer
):
pass
class
ClusterTrainerWithDataset
(
TranspileTrainer
):
class
ClusterTrainer
(
TranspileTrainer
):
def
processor_register
(
self
):
role
=
PaddleCloudRoleMaker
()
fleet
.
init
(
role
)
...
...
@@ -71,7 +67,7 @@ class ClusterTrainerWithDataset(TranspileTrainer):
def
init
(
self
,
context
):
self
.
model
.
input
()
self
.
model
.
build_model
()
self
.
model
.
net
()
self
.
model
.
metrics
()
self
.
model
.
avg_loss
()
optimizer
=
self
.
model
.
optimizer
()
...
...
fleetrec/trainer/factory.py
浏览文件 @
f8d55c9f
...
...
@@ -10,46 +10,19 @@
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.# Copyright (c) 2019 PaddlePaddle Authors. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import
os
import
sys
import
yaml
from
fleetrec.trainer.single_trainer
import
SingleTrainerWithDataloader
from
fleetrec.trainer.single_trainer
import
SingleTrainerWithDataset
from
fleetrec.trainer.cluster_trainer
import
ClusterTrainerWithDataloader
from
fleetrec.trainer.cluster_trainer
import
ClusterTrainerWithDataset
from
fleetrec.trainer.local_engine
import
Launch
from
fleetrec.trainer.single_trainer
import
SingleTrainer
from
fleetrec.trainer.cluster_trainer
import
ClusterTrainer
from
fleetrec.trainer.ctr_trainer
import
CtrPaddleTrainer
from
fleetrec.utils
import
envs
def
str2bool
(
v
):
if
isinstance
(
v
,
bool
):
return
v
if
v
.
lower
()
in
(
'yes'
,
'true'
,
't'
,
'y'
,
'1'
):
return
True
elif
v
.
lower
()
in
(
'no'
,
'false'
,
'f'
,
'n'
,
'0'
):
return
False
else
:
raise
ValueError
(
'Boolean value expected.'
)
from
fleetrec.utils
import
util
class
TrainerFactory
(
object
):
...
...
@@ -61,21 +34,10 @@ class TrainerFactory(object):
print
(
envs
.
pretty_print_envs
(
envs
.
get_global_envs
()))
train_mode
=
envs
.
get_global_env
(
"train.trainer"
)
reader_mode
=
envs
.
get_global_env
(
"train.reader.mode"
)
if
train_mode
==
"SingleTraining"
:
if
reader_mode
==
"dataset"
:
trainer
=
SingleTrainerWithDataset
()
elif
reader_mode
==
"dataloader"
:
trainer
=
SingleTrainerWithDataloader
()
else
:
raise
ValueError
(
"reader only support dataset/dataloader"
)
trainer
=
SingleTrainer
()
elif
train_mode
==
"ClusterTraining"
:
if
reader_mode
==
"dataset"
:
trainer
=
ClusterTrainerWithDataset
()
elif
reader_mode
==
"dataloader"
:
trainer
=
ClusterTrainerWithDataloader
()
else
:
raise
ValueError
(
"reader only support dataset/dataloader"
)
trainer
=
ClusterTrainer
()
elif
train_mode
==
"CtrTrainer"
:
trainer
=
CtrPaddleTrainer
(
config
)
else
:
...
...
@@ -108,7 +70,7 @@ class TrainerFactory(object):
envs
.
set_global_envs
(
_config
)
mode
=
envs
.
get_global_env
(
"train.trainer"
)
container
=
envs
.
get_global_env
(
"train.container"
)
instance
=
str2bool
(
os
.
getenv
(
"CLUSTER_INSTANCE"
,
"0"
))
instance
=
util
.
str2bool
(
os
.
getenv
(
"CLUSTER_INSTANCE"
,
"0"
))
if
mode
==
"ClusterTraining"
and
container
==
"local"
and
not
instance
:
trainer
=
TrainerFactory
.
_build_engine
(
config
)
...
...
@@ -124,4 +86,3 @@ if __name__ == "__main__":
raise
ValueError
(
"need a yaml file path argv"
)
trainer
=
TrainerFactory
.
create
(
sys
.
argv
[
1
])
trainer
.
run
()
fleetrec/trainer/single_trainer.py
浏览文件 @
f8d55c9f
...
...
@@ -17,25 +17,18 @@ Training use fluid with one node only.
"""
from
__future__
import
print_function
import
os
import
time
import
numpy
as
np
import
logging
import
paddle.fluid
as
fluid
from
.transpiler_trainer
import
TranspileTrainer
from
.
.utils
import
envs
from
fleetrec.trainer
.transpiler_trainer
import
TranspileTrainer
from
fleetrec
.utils
import
envs
logging
.
basicConfig
(
format
=
"%(asctime)s - %(levelname)s - %(message)s"
)
logger
=
logging
.
getLogger
(
"fluid"
)
logger
.
setLevel
(
logging
.
INFO
)
class
SingleTrainerWithDataloader
(
TranspileTrainer
):
pass
class
SingleTrainerWithDataset
(
TranspileTrainer
):
class
SingleTrainer
(
TranspileTrainer
):
def
processor_register
(
self
):
self
.
regist_context_processor
(
'uninit'
,
self
.
instance
)
self
.
regist_context_processor
(
'init_pass'
,
self
.
init
)
...
...
fleetrec/trainer/trainer.py
浏览文件 @
f8d55c9f
...
...
@@ -14,11 +14,8 @@
import
abc
import
time
import
yaml
from
paddle
import
fluid
from
..utils
import
envs
class
Trainer
(
object
):
"""R
...
...
fleetrec/trainer/transpiler_trainer.py
浏览文件 @
f8d55c9f
...
...
@@ -18,10 +18,9 @@ Training use fluid with DistributeTranspiler
import
os
import
paddle.fluid
as
fluid
from
paddle.fluid.incubate.fleet.parameter_server.distribute_transpiler
import
fleet
from
fleetrec.trainer
import
Trainer
from
fleetrec.trainer
.trainer
import
Trainer
from
fleetrec.utils
import
envs
...
...
@@ -39,15 +38,18 @@ class TranspileTrainer(Trainer):
def
_get_dataset
(
self
):
namespace
=
"train.reader"
inputs
=
self
.
model
.
input
_var
s
()
inputs
=
self
.
model
.
inputs
()
threads
=
envs
.
get_global_env
(
"train.threads"
,
None
)
batch_size
=
envs
.
get_global_env
(
"batch_size"
,
None
,
namespace
)
pipe_command
=
envs
.
get_global_env
(
"pipe_command"
,
None
,
namespace
)
reader_class
=
envs
.
get_global_env
(
"class"
,
None
,
namespace
)
abs_dir
=
os
.
path
.
dirname
(
os
.
path
.
abspath
(
__file__
))
reader
=
os
.
path
.
join
(
abs_dir
,
'..'
,
'reader_implement.py'
)
pipe_cmd
=
"python {} {} {}"
.
format
(
reader
,
reader_class
,
"TRAIN"
)
train_data_path
=
envs
.
get_global_env
(
"train_data_path"
,
None
,
namespace
)
dataset
=
fluid
.
DatasetFactory
().
create_dataset
()
dataset
.
set_use_var
(
inputs
)
dataset
.
set_pipe_command
(
pipe_c
omman
d
)
dataset
.
set_pipe_command
(
pipe_c
m
d
)
dataset
.
set_batch_size
(
batch_size
)
dataset
.
set_thread
(
threads
)
file_list
=
[
...
...
fleetrec/utils/util.py
浏览文件 @
f8d55c9f
...
...
@@ -15,7 +15,18 @@
import
os
import
time
import
datetime
from
..
utils
import
fs
as
fs
from
..utils
import
fs
as
fs
def
str2bool
(
v
):
if
isinstance
(
v
,
bool
):
return
v
if
v
.
lower
()
in
(
'yes'
,
'true'
,
't'
,
'y'
,
'1'
):
return
True
elif
v
.
lower
()
in
(
'no'
,
'false'
,
'f'
,
'n'
,
'0'
):
return
False
else
:
raise
ValueError
(
'Boolean value expected.'
)
def
get_env_value
(
env_name
):
...
...
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