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
PaddlePaddle
Paddle
提交
16146088
P
Paddle
项目概览
PaddlePaddle
/
Paddle
1 年多 前同步成功
通知
2302
Star
20931
Fork
5422
代码
文件
提交
分支
Tags
贡献者
分支图
Diff
Issue
1423
列表
看板
标记
里程碑
合并请求
543
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
1,423
Issue
1,423
列表
看板
标记
里程碑
合并请求
543
合并请求
543
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
16146088
编写于
8月 17, 2021
作者:
K
Kaipeng Deng
提交者:
GitHub
8月 17, 2021
浏览文件
操作
浏览文件
下载
电子邮件补丁
差异文件
fix drop_last not work on IterableDataset (#34801)
* fix drop_last not work in IterableDataset. test=develop
上级
181f7cec
变更
5
隐藏空白更改
内联
并排
Showing
5 changed file
with
36 addition
and
8 deletion
+36
-8
python/paddle/fluid/dataloader/dataloader_iter.py
python/paddle/fluid/dataloader/dataloader_iter.py
+4
-3
python/paddle/fluid/dataloader/worker.py
python/paddle/fluid/dataloader/worker.py
+4
-3
python/paddle/fluid/reader.py
python/paddle/fluid/reader.py
+1
-0
python/paddle/fluid/tests/unittests/test_multiprocess_dataloader_dataset.py
...d/tests/unittests/test_multiprocess_dataloader_dataset.py
+25
-0
python/paddle/fluid/tests/unittests/test_multiprocess_dataloader_exception.py
...tests/unittests/test_multiprocess_dataloader_exception.py
+2
-2
未找到文件。
python/paddle/fluid/dataloader/dataloader_iter.py
浏览文件 @
16146088
...
...
@@ -59,6 +59,7 @@ class _DataLoaderIterBase(object):
self
.
_places
=
loader
.
places
self
.
_return_list
=
loader
.
return_list
self
.
_batch_sampler
=
loader
.
batch_sampler
self
.
_drop_last
=
loader
.
drop_last
self
.
_auto_collate_batch
=
loader
.
auto_collate_batch
self
.
_num_workers
=
loader
.
num_workers
self
.
_use_buffer_reader
=
loader
.
use_buffer_reader
...
...
@@ -111,7 +112,7 @@ class _DataLoaderIterSingleProcess(_DataLoaderIterBase):
self
.
_dataset_fetcher
=
_DatasetKind
.
create_fetcher
(
self
.
_dataset_kind
,
self
.
_dataset
,
self
.
_auto_collate_batch
,
self
.
_collate_fn
,
True
)
self
.
_collate_fn
,
self
.
_drop_last
)
# NOTE: _structrue_infos used to record the data structure of
# batch to restore batch structure after reading Tensor
...
...
@@ -309,8 +310,8 @@ class _DataLoaderIterMultiProcess(_DataLoaderIterBase):
args
=
(
self
.
_dataset
,
self
.
_dataset_kind
,
indices_queue
,
self
.
_data_queue
,
self
.
_workers_done_event
,
self
.
_auto_collate_batch
,
self
.
_collate_fn
,
self
.
_
worker_init_fn
,
i
,
self
.
_num_workers
,
self
.
_use_shared_memory
))
self
.
_
drop_last
,
self
.
_worker_init_fn
,
i
,
self
.
_
num_workers
,
self
.
_
use_shared_memory
))
worker
.
daemon
=
True
worker
.
start
()
self
.
_workers
.
append
(
worker
)
...
...
python/paddle/fluid/dataloader/worker.py
浏览文件 @
16146088
...
...
@@ -253,7 +253,7 @@ def _generate_states(base_seed=0, worker_id=0):
def
_worker_loop
(
dataset
,
dataset_kind
,
indices_queue
,
out_queue
,
done_event
,
auto_collate_batch
,
collate_fn
,
init_fn
,
worker_id
,
auto_collate_batch
,
collate_fn
,
drop_last
,
init_fn
,
worker_id
,
num_workers
,
use_shared_memory
):
try
:
# NOTE: [ mmap files clear ] When the child process exits unexpectedly,
...
...
@@ -282,8 +282,9 @@ def _worker_loop(dataset, dataset_kind, indices_queue, out_queue, done_event,
try
:
if
init_fn
is
not
None
:
init_fn
(
worker_id
)
fetcher
=
_DatasetKind
.
create_fetcher
(
dataset_kind
,
dataset
,
auto_collate_batch
,
collate_fn
,
True
)
fetcher
=
_DatasetKind
.
create_fetcher
(
dataset_kind
,
dataset
,
auto_collate_batch
,
collate_fn
,
drop_last
)
except
:
init_exception
=
_WorkerException
(
worker_id
)
...
...
python/paddle/fluid/reader.py
浏览文件 @
16146088
...
...
@@ -401,6 +401,7 @@ class DataLoader(object):
shuffle
=
shuffle
,
drop_last
=
drop_last
)
self
.
drop_last
=
drop_last
self
.
auto_collate_batch
=
self
.
batch_sampler
is
not
None
self
.
pin_memory
=
False
...
...
python/paddle/fluid/tests/unittests/test_multiprocess_dataloader_dataset.py
浏览文件 @
16146088
...
...
@@ -397,5 +397,30 @@ class TestDataLoaderGenerateStates(unittest.TestCase):
assert
out
==
outp
class
TestDatasetWithDropLast
(
unittest
.
TestCase
):
def
run_main
(
self
,
dataset
,
num_samples
,
batch_size
):
for
num_workers
in
[
0
,
1
]:
for
drop_last
in
[
True
,
False
]:
steps
=
(
num_samples
+
(
1
-
int
(
drop_last
))
*
\
(
batch_size
-
1
))
//
batch_size
dataloader
=
DataLoader
(
dataset
,
batch_size
=
batch_size
,
drop_last
=
drop_last
,
num_workers
=
num_workers
)
datas
=
[]
for
data
in
dataloader
:
datas
.
append
(
data
)
assert
len
(
datas
)
==
steps
def
test_map_dataset
(
self
):
dataset
=
RandomDataset
(
10
)
self
.
run_main
(
dataset
,
10
,
3
)
def
test_iterable_dataset
(
self
):
dataset
=
RandomIterableDataset
(
10
)
self
.
run_main
(
dataset
,
10
,
3
)
if
__name__
==
'__main__'
:
unittest
.
main
()
python/paddle/fluid/tests/unittests/test_multiprocess_dataloader_exception.py
浏览文件 @
16146088
...
...
@@ -180,7 +180,7 @@ class TestDataLoaderWorkerLoop(unittest.TestCase):
indices_queue
.
put
(
None
)
_worker_loop
(
loader
.
_dataset
,
0
,
indices_queue
,
loader
.
_data_queue
,
loader
.
_workers_done_event
,
True
,
_collate_fn
,
_init_fn
,
0
,
1
,
True
,
_collate_fn
,
True
,
_init_fn
,
0
,
1
,
loader
.
_use_shared_memory
)
self
.
assertTrue
(
False
)
except
AssertionError
:
...
...
@@ -224,7 +224,7 @@ class TestDataLoaderWorkerLoop(unittest.TestCase):
loader
.
_workers_done_event
.
set
()
_worker_loop
(
loader
.
_dataset
,
0
,
indices_queue
,
loader
.
_data_queue
,
loader
.
_workers_done_event
,
True
,
_collate_fn
,
_init_fn
,
0
,
1
,
True
,
_collate_fn
,
True
,
_init_fn
,
0
,
1
,
loader
.
_use_shared_memory
)
self
.
assertTrue
(
True
)
except
AssertionError
:
...
...
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