Skip to content

P
Paddle

PaddlePaddle / Paddle
大约 1 年 前同步成功

通知 2298
Star 20931
Fork 5422
P
Paddle
提交 559d3632 编写于 作者: M minqiyang
浏览文件
操作

Apply 2to3 to current paddle main python code

上级 3ade95d0
隐藏空白更改
内联 并排
Showing with 805 addition and 745 deletion
paddle/fluid/pybind/pybind.cc
浏览文件 @ 559d3632
......@@ -394,8 +394,10 @@ All parameter, weight, gradient are variables in Paddle.
InferenceOptimize(*(origin.Proto()), &pruned_desc);
return new ProgramDesc(pruned_desc);
});
m.def("empty_var_name", []() { return framework::kEmptyVarName; });
m.def("grad_var_suffix", []() { return framework::kGradVarSuffix; });
m.def("empty_var_name",
[]() { return std::string(framework::kEmptyVarName); });
m.def("grad_var_suffix",
[]() { return std::string(framework::kGradVarSuffix); });
m.def_submodule(
"var_names",
"The module will return special predefined variable name in Paddle")
......
python/paddle/dataset/cifar.py
浏览文件 @ 559d3632
......@@ -28,7 +28,7 @@ images per class.
"""
import cPickle
import pickle
import itertools
import numpy
import paddle.dataset.common
......@@ -48,7 +48,7 @@ def reader_creator(filename, sub_name, cycle=False):
data = batch['data']
labels = batch.get('labels', batch.get('fine_labels', None))
assert labels is not None
for sample, label in itertools.izip(data, labels):
for sample, label in zip(data, labels):
yield (sample / 255.0).astype(numpy.float32), int(label)
def reader():
......@@ -58,7 +58,7 @@ def reader_creator(filename, sub_name, cycle=False):
while True:
for name in names:
batch = cPickle.load(f.extractfile(name))
batch = pickle.load(f.extractfile(name))
for item in read_batch(batch):
yield item
if not cycle:
......
python/paddle/dataset/common.py
浏览文件 @ 559d3632
......@@ -20,9 +20,9 @@ import shutil
import sys
import importlib
import paddle.dataset
import cPickle
import pickle
import glob
import cPickle as pickle
import pickle as pickle
__all__ = [
'DATA_HOME',
......@@ -75,13 +75,13 @@ def download(url, module_name, md5sum, save_name=None):
retry_limit = 3
while not (os.path.exists(filename) and md5file(filename) == md5sum):
if os.path.exists(filename):
print "file md5", md5file(filename), md5sum
print(("file md5", md5file(filename), md5sum))
if retry < retry_limit:
retry += 1
else:
raise RuntimeError("Cannot download {0} within retry limit {1}".
format(url, retry_limit))
print "Cache file %s not found, downloading %s" % (filename, url)
print(("Cache file %s not found, downloading %s" % (filename, url)))
r = requests.get(url, stream=True)
total_length = r.headers.get('content-length')
......@@ -104,8 +104,9 @@ def download(url, module_name, md5sum, save_name=None):
def fetch_all():
for module_name in filter(lambda x: not x.startswith("__"),
dir(paddle.dataset)):
for module_name in [
x for x in dir(paddle.dataset) if not x.startswith("__")
]:
if "fetch" in dir(
importlib.import_module("paddle.dataset.%s" % module_name)):
getattr(
......@@ -114,8 +115,9 @@ def fetch_all():
def fetch_all_recordio(path):
for module_name in filter(lambda x: not x.startswith("__"),
dir(paddle.dataset)):
for module_name in [
x for x in dir(paddle.dataset) if not x.startswith("__")
]:
if "convert" in dir(
importlib.import_module("paddle.dataset.%s" % module_name)) and \
not module_name == "common":
......@@ -126,7 +128,7 @@ def fetch_all_recordio(path):
"convert")(ds_path)
def split(reader, line_count, suffix="%05d.pickle", dumper=cPickle.dump):
def split(reader, line_count, suffix="%05d.pickle", dumper=pickle.dump):
"""
you can call the function as:
......@@ -167,7 +169,7 @@ def split(reader, line_count, suffix="%05d.pickle", dumper=cPickle.dump):
def cluster_files_reader(files_pattern,
trainer_count,
trainer_id,
loader=cPickle.load):
loader=pickle.load):
"""
Create a reader that yield element from the given files, select
a file set according trainer count and trainer_id
......@@ -188,7 +190,7 @@ def cluster_files_reader(files_pattern,
my_file_list = []
for idx, fn in enumerate(file_list):
if idx % trainer_count == trainer_id:
print "append file: %s" % fn
print(("append file: %s" % fn))
my_file_list.append(fn)
for fn in my_file_list:
with open(fn, "r") as f:
......@@ -221,7 +223,7 @@ def convert(output_path, reader, line_count, name_prefix):
for l in lines:
# FIXME(Yancey1989):
# dumps with protocol: pickle.HIGHEST_PROTOCOL
writer.write(cPickle.dumps(l))
writer.write(pickle.dumps(l))
writer.close()
lines = []
......
python/paddle/dataset/conll05.py
浏览文件 @ 559d3632
......@@ -87,12 +87,12 @@ def corpus_reader(data_path, words_name, props_name):
sentences = []
labels = []
one_seg = []
for word, label in itertools.izip(words_file, props_file):
for word, label in zip(words_file, props_file):
word = word.strip()
label = label.strip().split()
if len(label) == 0: # end of sentence
for i in xrange(len(one_seg[0])):
for i in range(len(one_seg[0])):
a_kind_lable = [x[i] for x in one_seg]
labels.append(a_kind_lable)
......
python/paddle/dataset/flowers.py
浏览文件 @ 559d3632
......@@ -28,10 +28,10 @@ Graphics and Image Processing (2008)
http://www.robots.ox.ac.uk/~vgg/publications/papers/nilsback08.{pdf,ps.gz}.
"""
import cPickle
import pickle
import itertools
import functools
from common import download
from .common import download
import tarfile
import scipy.io as scio
from paddle.dataset.image import *
......@@ -116,10 +116,10 @@ def reader_creator(data_file,
file = file.strip()
batch = None
with open(file, 'r') as f:
batch = cPickle.load(f)
batch = pickle.load(f)
data = batch['data']
labels = batch['label']
for sample, label in itertools.izip(data, batch['label']):
for sample, label in zip(data, batch['label']):
yield sample, int(label) - 1
if not cycle:
break
......
python/paddle/dataset/image.py
浏览文件 @ 559d3632
......@@ -36,7 +36,7 @@ except ImportError:
cv2 = None
import os
import tarfile
import cPickle
import pickle
__all__ = [
"load_image_bytes", "load_image", "resize_short", "to_chw", "center_crop",
......@@ -86,10 +86,10 @@ def batch_images_from_tar(data_file,
output = {}
output['label'] = labels
output['data'] = data
cPickle.dump(
pickle.dump(
output,
open('%s/batch_%d' % (out_path, file_id), 'w'),
protocol=cPickle.HIGHEST_PROTOCOL)
protocol=pickle.HIGHEST_PROTOCOL)
file_id += 1
data = []
labels = []
......@@ -97,10 +97,10 @@ def batch_images_from_tar(data_file,
output = {}
output['label'] = labels
output['data'] = data
cPickle.dump(
pickle.dump(
output,
open('%s/batch_%d' % (out_path, file_id), 'w'),
protocol=cPickle.HIGHEST_PROTOCOL)
protocol=pickle.HIGHEST_PROTOCOL)
with open(meta_file, 'a') as meta:
for file in os.listdir(out_path):
......
python/paddle/dataset/imdb.py
浏览文件 @ 559d3632
......@@ -42,13 +42,13 @@ def tokenize(pattern):
# sequential access of member files, other than
# tarfile.extractfile, which does random access and might
# destroy hard disks.
tf = tarf.next()
tf = next(tarf)
while tf != None:
if bool(pattern.match(tf.name)):
# newline and punctuations removal and ad-hoc tokenization.
yield tarf.extractfile(tf).read().rstrip("\n\r").translate(
None, string.punctuation).lower().split()
tf = tarf.next()
tf = next(tarf)
def build_dict(pattern, cutoff):
......@@ -62,11 +62,11 @@ def build_dict(pattern, cutoff):
word_freq[word] += 1
# Not sure if we should prune less-frequent words here.
word_freq = filter(lambda x: x[1] > cutoff, word_freq.items())
word_freq = [x for x in list(word_freq.items()) if x[1] > cutoff]
dictionary = sorted(word_freq, key=lambda x: (-x[1], x[0]))
words, _ = list(zip(*dictionary))
word_idx = dict(zip(words, xrange(len(words))))
word_idx = dict(list(zip(words, list(range(len(words))))))
word_idx['<unk>'] = len(words)
return word_idx
......
python/paddle/dataset/imikolov.py
浏览文件 @ 559d3632
......@@ -64,11 +64,11 @@ def build_dict(min_word_freq=50):
# remove <unk> for now, since we will set it as last index
del word_freq['<unk>']
word_freq = filter(lambda x: x[1] > min_word_freq, word_freq.items())
word_freq = [x for x in list(word_freq.items()) if x[1] > min_word_freq]
word_freq_sorted = sorted(word_freq, key=lambda x: (-x[1], x[0]))
words, _ = list(zip(*word_freq_sorted))
word_idx = dict(zip(words, xrange(len(words))))
word_idx = dict(list(zip(words, list(range(len(words))))))
word_idx['<unk>'] = len(words)
return word_idx
......
python/paddle/dataset/mnist.py
浏览文件 @ 559d3632
......@@ -65,7 +65,7 @@ def reader_creator(image_filename, label_filename, buffer_size):
images = images / 255.0 * 2.0 - 1.0
for i in xrange(buffer_size):
for i in range(buffer_size):
yield images[i, :], int(labels[i])
finally:
try:
......
python/paddle/dataset/movielens.py
浏览文件 @ 559d3632
......@@ -187,7 +187,7 @@ def max_movie_id():
Get the maximum value of movie id.
"""
__initialize_meta_info__()
return reduce(__max_index_info__, MOVIE_INFO.viewvalues()).index
return reduce(__max_index_info__, list(MOVIE_INFO.values())).index
def max_user_id():
......@@ -195,7 +195,7 @@ def max_user_id():
Get the maximum value of user id.
"""
__initialize_meta_info__()
return reduce(__max_index_info__, USER_INFO.viewvalues()).index
return reduce(__max_index_info__, list(USER_INFO.values())).index
def __max_job_id_impl__(a, b):
......@@ -210,7 +210,7 @@ def max_job_id():
Get the maximum value of job id.
"""
__initialize_meta_info__()
return reduce(__max_job_id_impl__, USER_INFO.viewvalues()).job_id
return reduce(__max_job_id_impl__, list(USER_INFO.values())).job_id
def movie_categories():
......@@ -243,7 +243,7 @@ def unittest():
for test_count, _ in enumerate(test()()):
pass
print train_count, test_count
print((train_count, test_count))
def fetch():
......
python/paddle/dataset/mq2007.py
浏览文件 @ 559d3632
......@@ -26,7 +26,7 @@ http://research.microsoft.com/en-us/um/beijing/projects/letor/LETOR4.0/Data/MQ20
import os
import functools
import rarfile
from common import download
from .common import download
import numpy as np
# URL = "http://research.microsoft.com/en-us/um/beijing/projects/letor/LETOR4.0/Data/MQ2007.rar"
......@@ -330,4 +330,4 @@ if __name__ == "__main__":
mytest = functools.partial(
__reader__, filepath="MQ2007/MQ2007/Fold1/sample", format="listwise")
for label, query in mytest():
print label, query
print((label, query))
python/paddle/dataset/sentiment.py
浏览文件 @ 559d3632
......@@ -43,11 +43,11 @@ def download_data_if_not_yet():
nltk.data.path.append(paddle.dataset.common.DATA_HOME)
movie_reviews.categories()
except LookupError:
print "Downloading movie_reviews data set, please wait....."
print("Downloading movie_reviews data set, please wait.....")
nltk.download(
'movie_reviews', download_dir=paddle.dataset.common.DATA_HOME)
print "Download data set success....."
print "Path is " + nltk.data.find('corpora/movie_reviews').path
print("Download data set success.....")
print(("Path is " + nltk.data.find('corpora/movie_reviews').path))
def get_word_dict():
......@@ -64,7 +64,7 @@ def get_word_dict():
for field in movie_reviews.fileids(category):
for words in movie_reviews.words(field):
word_freq_dict[words] += 1
words_sort_list = word_freq_dict.items()
words_sort_list = list(word_freq_dict.items())
words_sort_list.sort(cmp=lambda a, b: b[1] - a[1])
for index, word in enumerate(words_sort_list):
words_freq_sorted.append((word[0], index))
......@@ -80,7 +80,8 @@ def sort_files():
files_list = list()
neg_file_list = movie_reviews.fileids('neg')
pos_file_list = movie_reviews.fileids('pos')
files_list = list(chain.from_iterable(zip(neg_file_list, pos_file_list)))
files_list = list(
chain.from_iterable(list(zip(neg_file_list, pos_file_list))))
return files_list
......
python/paddle/dataset/tests/common_test.py
浏览文件 @ 559d3632
......@@ -36,7 +36,7 @@ class TestCommon(unittest.TestCase):
def test_split(self):
def test_reader():
def reader():
for x in xrange(10):
for x in range(10):
yield x
return reader
......@@ -49,7 +49,7 @@ class TestCommon(unittest.TestCase):
def test_cluster_file_reader(self):
_, temp_path = tempfile.mkstemp()
for x in xrange(5):
for x in range(5):
with open(temp_path + '/%05d.test' % x) as f:
f.write('%d\n' % x)
reader = paddle.dataset.common.cluster_files_reader(
......@@ -63,7 +63,7 @@ class TestCommon(unittest.TestCase):
def test_reader():
def reader():
for x in xrange(record_num):
for x in range(record_num):
yield x
return reader
......
python/paddle/dataset/tests/imikolov_test.py
浏览文件 @ 559d3632
......@@ -59,7 +59,7 @@ class TestMikolov(unittest.TestCase):
self.assertEqual(first_line, read_line)
def test_total(self):
_, idx = zip(*WORD_DICT.items())
_, idx = list(zip(*list(WORD_DICT.items())))
self.assertEqual(sorted(idx)[-1], len(WORD_DICT) - 1)
......
python/paddle/dataset/tests/test_sentiment.py
浏览文件 @ 559d3632
......@@ -24,9 +24,8 @@ from nltk.corpus import movie_reviews
class TestSentimentMethods(unittest.TestCase):
def test_get_word_dict(self):
word_dict = st.get_word_dict()[0:10]
test_word_list = [(u',', 0), (u'the', 1), (u'.', 2), (u'a', 3),
(u'and', 4), (u'of', 5), (u'to', 6), (u"'", 7),
(u'is', 8), (u'in', 9)]
test_word_list = [(',', 0), ('the', 1), ('.', 2), ('a', 3), ('and', 4),
('of', 5), ('to', 6), ("'", 7), ('is', 8), ('in', 9)]
for idx, each in enumerate(word_dict):
self.assertEqual(each, test_word_list[idx])
self.assertTrue("/root/.cache/paddle/dataset" in nltk.data.path)
......
python/paddle/dataset/uci_housing.py
浏览文件 @ 559d3632
......@@ -49,9 +49,12 @@ def feature_range(maximums, minimums):
import matplotlib.pyplot as plt
fig, ax = plt.subplots()
feature_num = len(maximums)
ax.bar(range(feature_num), maximums - minimums, color='r', align='center')
ax.bar(list(range(feature_num)),
maximums - minimums,
color='r',
align='center')
ax.set_title('feature scale')
plt.xticks(range(feature_num), feature_names)
plt.xticks(list(range(feature_num)), feature_names)
plt.xlim([-1, feature_num])
fig.set_figheight(6)
fig.set_figwidth(10)
......@@ -71,7 +74,7 @@ def load_data(filename, feature_num=14, ratio=0.8):
maximums, minimums, avgs = data.max(axis=0), data.min(axis=0), data.sum(
axis=0) / data.shape[0]
feature_range(maximums[:-1], minimums[:-1])
for i in xrange(feature_num - 1):
for i in range(feature_num - 1):
data[:, i] = (data[:, i] - avgs[i]) / (maximums[i] - minimums[i])
offset = int(data.shape[0] * ratio)
UCI_TRAIN_DATA = data[:offset]
......
python/paddle/dataset/wmt14.py
浏览文件 @ 559d3632
......@@ -154,8 +154,8 @@ def get_dict(dict_size, reverse=True):
tar_file = paddle.dataset.common.download(URL_TRAIN, 'wmt14', MD5_TRAIN)
src_dict, trg_dict = __read_to_dict(tar_file, dict_size)
if reverse:
src_dict = {v: k for k, v in src_dict.items()}
trg_dict = {v: k for k, v in trg_dict.items()}
src_dict = {v: k for k, v in list(src_dict.items())}
trg_dict = {v: k for k, v in list(trg_dict.items())}
return src_dict, trg_dict
......
python/paddle/dataset/wmt16.py
浏览文件 @ 559d3632
......@@ -70,7 +70,9 @@ def __build_dict(tar_file, dict_size, save_path, lang):
fout.write("%s\n%s\n%s\n" % (START_MARK, END_MARK, UNK_MARK))
for idx, word in enumerate(
sorted(
word_dict.iteritems(), key=lambda x: x[1], reverse=True)):
iter(list(word_dict.items())),
key=lambda x: x[1],
reverse=True)):
if idx + 3 == dict_size: break
fout.write("%s\n" % (word[0]))
......
python/paddle/fluid/__init__.py
浏览文件 @ 559d3632
......@@ -14,49 +14,49 @@
from __future__ import print_function
# import all class inside framework into fluid module
import framework
from framework import *
from . import framework
from .framework import *
# import all class inside executor into fluid module
import executor
from executor import *
import trainer
from trainer import Trainer
from trainer import BeginEpochEvent
from trainer import EndEpochEvent
from trainer import BeginStepEvent
from trainer import EndStepEvent
from trainer import CheckpointConfig
import inferencer
from inferencer import Inferencer
import io
import evaluator
import initializer
import layers
import contrib
import nets
import optimizer
import backward
import regularizer
import average
import metrics
import transpiler
from param_attr import ParamAttr, WeightNormParamAttr
from data_feeder import DataFeeder
from core import LoDTensor, LoDTensorArray, CPUPlace, CUDAPlace, CUDAPinnedPlace, Scope
from transpiler import DistributeTranspiler, InferenceTranspiler, \
from . import executor
from .executor import *
from . import trainer
from .trainer import Trainer
from .trainer import BeginEpochEvent
from .trainer import EndEpochEvent
from .trainer import BeginStepEvent
from .trainer import EndStepEvent
from .trainer import CheckpointConfig
from . import inferencer
from .inferencer import Inferencer
from . import io
from . import evaluator
from . import initializer
from . import layers
from . import contrib
from . import nets
from . import optimizer
from . import backward
from . import regularizer
from . import average
from . import metrics
from . import transpiler
from .param_attr import ParamAttr, WeightNormParamAttr
from .data_feeder import DataFeeder
from .core import LoDTensor, LoDTensorArray, CPUPlace, CUDAPlace, CUDAPinnedPlace, Scope
from .transpiler import DistributeTranspiler, InferenceTranspiler, \
memory_optimize, release_memory, DistributeTranspilerConfig
from concurrency import (Go, make_channel, channel_send, channel_recv,
channel_close, Select)
from lod_tensor import create_lod_tensor, create_random_int_lodtensor
import clip
import profiler
import unique_name
import recordio_writer
import parallel_executor
from parallel_executor import *
from .concurrency import (Go, make_channel, channel_send, channel_recv,
channel_close, Select)
from .lod_tensor import create_lod_tensor, create_random_int_lodtensor
from . import clip
from . import profiler
from . import unique_name
from . import recordio_writer
from . import parallel_executor
from .parallel_executor import *
from paddle.fluid.layers.math_op_patch import monkey_patch_variable
Tensor = LoDTensor
......@@ -99,8 +99,8 @@ def __bootstrap__():
None
"""
import sys
import core
import os
from . import core
in_test = 'unittest' in sys.modules
......
python/paddle/fluid/annotations.py
浏览文件 @ 559d3632
......@@ -12,6 +12,7 @@
# See the License for the specific language governing permissions and
# limitations under the License.
from __future__ import print_function
import functools
import sys
......@@ -28,7 +29,7 @@ def deprecated(since, instead, extra_message=""):
@functools.wraps(func)
def wrapper(*args, **kwargs):
print >> sys.stderr, err_msg
print(err_msg, file=sys.stderr)
return func(*args, **kwargs)
wrapper.__doc__ += "\n "
......
python/paddle/fluid/backward.py
浏览文件 @ 559d3632
......@@ -16,7 +16,8 @@ from paddle.fluid import framework as framework
from . import core
import collections
import copy
import unique_name
import six
from . import unique_name
__all__ = ['append_backward']
......@@ -44,17 +45,25 @@ def _create_op_desc_(op_type, inputs, outputs, attrs):
"""
op_desc = core.OpDesc()
op_desc.set_type(op_type)
for para, args in inputs.iteritems():
op_desc.set_input(para, args)
for para, args in outputs.iteritems():
op_desc.set_output(para, args)
for para, args in list(inputs.items()):
op_desc.set_input(
para,
list(
map(lambda arg: arg.decode() if isinstance(arg, six.binary_type) else arg,
args)))
for para, args in list(outputs.items()):
op_desc.set_output(
para,
list(
map(lambda arg: arg.decode() if isinstance(arg, six.binary_type) else arg,
args)))
op_role_attr_name = core.op_proto_and_checker_maker.kOpRoleAttrName()
if op_role_attr_name not in attrs:
attrs[
op_role_attr_name] = core.op_proto_and_checker_maker.OpRole.Backward
for name, val in attrs.iteritems():
for name, val in list(attrs.items()):
if isinstance(val, framework.Block):
op_desc.set_block_attr(name, val.desc)
else:
......@@ -105,7 +114,9 @@ def _strip_grad_suffix_(name):
e.g. x@GRAD ==> x
y@GRAD@RENAME@1 ==> y
"""
pos = name.find(core.grad_var_suffix())
if isinstance(name, six.text_type):
name = name.encode()
pos = name.find(six.b(core.grad_var_suffix()))
return name[:pos] if pos != -1 else name
......@@ -114,7 +125,9 @@ def _append_grad_suffix_(name):
Append grad suffix to the given variable name
e.g. x ==> x@GRAD
"""
return name + core.grad_var_suffix()
if isinstance(name, six.text_type):
name = name.encode()
return name + six.b(core.grad_var_suffix())
def _addup_repetitive_outputs_(op_descs):
......@@ -174,7 +187,7 @@ def _addup_repetitive_outputs_(op_descs):
op_desc.set_output(param_name, arg_names)
renamed_vars[var_name].append(new_name)
for var_name, inputs in renamed_vars.iteritems():
for var_name, inputs in list(renamed_vars.items()):
if len(inputs) > 1:
pending_sum_ops.append(
(_create_op_desc_("sum", {"X": inputs}, {"Out": [var_name]},
......@@ -198,16 +211,19 @@ def _remove_no_grad_branch_(op_descs, no_grad_set):
out_arg_names = op_desc.output_arg_names()
if len(out_arg_names) == 0 or _all_in_set_(out_arg_names, no_grad_set):
return True
if _all_in_set_(
filter(lambda name: name.find(core.grad_var_suffix()) != -1,
op_desc.input_arg_names()), no_grad_set):
if _all_in_set_([
name for name in op_desc.input_arg_names()
if name.find(core.grad_var_suffix()) != -1
], no_grad_set):
no_grad_set.update(out_arg_names)
return True
return False
# Remove ops whose outputs are all in no_grad_dict
op_descs = filter(
lambda op_desc: not _op_can_be_removed_(op_desc, no_grad_set), op_descs)
op_descs = [
op_desc for op_desc in op_descs
if not _op_can_be_removed_(op_desc, no_grad_set)
]
# Insert fill_zeros_like_op
to_insert = []
for idx, op_desc in enumerate(op_descs):
......@@ -217,12 +233,12 @@ def _remove_no_grad_branch_(op_descs, no_grad_set):
"X": [_strip_grad_suffix_(arg)]
}, {"Out": [arg]}, {}), idx))
map(lambda p: op_descs.insert(p[1], p[0]), reversed(to_insert))
list([op_descs.insert(p[1], p[0]) for p in reversed(to_insert)])
return op_descs
import proto.framework_pb2 as framework_pb2
from .proto import framework_pb2
def serialize_op_decs(op_desc):
......@@ -244,8 +260,10 @@ def _callback_lookup_(op):
if op.type == 'parallel_do' and op.attr('use_nccl'):
all_vars = op.block.vars
param_names = set(op.input('parameters'))
param_names = filter(lambda name: all_vars[name].stop_gradient is False,
param_names)
param_names = [
name for name in param_names
if all_vars[name].stop_gradient is False
]
param_grad_names = [n + "@GRAD" for n in param_names]
class ParallelDoCallBack(object):
......@@ -399,7 +417,7 @@ def _append_backward_vars_(block, start_op_idx, grad_to_var, grad_info_map):
continue
block.desc.var(grad_var_name)
new_vars.add(grad_var_name)
if not grad_to_var.has_key(grad_var_name):
if grad_var_name not in grad_to_var:
continue
grad_info_map[grad_to_var[grad_var_name]] = (grad_var_name, block)
# infer_shape and infer_type
......@@ -427,7 +445,7 @@ def _rename_grad_(block, start_op_idx, grad_to_var, target_grad_map):
op_desc.rename_output(name, new_name)
var_map[name] = new_name
for g, ng in var_map.iteritems():
for g, ng in list(var_map.items()):
if g in grad_to_var:
grad_to_var[ng] = grad_to_var[g]
grad_to_var.pop(g)
......@@ -439,7 +457,7 @@ def _get_stop_gradients_(program):
for block in program.blocks:
assert isinstance(block, framework.Block)
block_no_grad_set = set()
for var in block.vars.itervalues():
for var in list(block.vars.values()):
assert isinstance(var, framework.Variable)
if var.stop_gradient:
block_no_grad_set.add(_append_grad_suffix_(var.name))
......@@ -452,51 +470,51 @@ def append_backward(loss, parameter_list=None, no_grad_set=None,
"""
Append backward part to main_program.
A complete neural network training is made up of forward and backward
propagation. However, when we configure a network, we only need to
specify its forwrd part. The backward part is generated automatically
A complete neural network training is made up of forward and backward
propagation. However, when we configure a network, we only need to
specify its forwrd part. The backward part is generated automatically
according to the forward part by this function.
In most cases, users do not need to invoke this function manually. It
In most cases, users do not need to invoke this function manually. It
will be automatically invoked by the optimizer's `minimize` function.
Args:
loss(Variable): The loss variable of the network.
parameter_list(list[string]|None): Names of parameters that need
to be updated by optimizers.
If it is None, all parameters
parameter_list(list[string]|None): Names of parameters that need
to be updated by optimizers.
If it is None, all parameters
will be updated.
Default: None
no_grad_set(set|None): Variables in the Block 0 whose gradients
should be ignored. All variables with
`step_gradient=True` from all blocks will
no_grad_set(set|None): Variables in the Block 0 whose gradients
should be ignored. All variables with
`step_gradient=True` from all blocks will
be automatically added into this set.
Default: None
callbacks(list[callable object]|None): The callbacks are used for
doing some custom jobs during
backward part building. All
callable objects in it will
be invoked once each time a
new gradient operator is added
into the program. The callable
object must has two input
parameters: 'block' and 'context'.
The 'block' is the block which
the new gradient operator will
be added to. The 'context' is a
map, whose keys are gradient
variable names and values are
callbacks(list[callable object]|None): The callbacks are used for
doing some custom jobs during
backward part building. All
callable objects in it will
be invoked once each time a
new gradient operator is added
into the program. The callable
object must has two input
parameters: 'block' and 'context'.
The 'block' is the block which
the new gradient operator will
be added to. The 'context' is a
map, whose keys are gradient
variable names and values are
corresponding original variables.
In addition to this, the 'context'
has another special key-value pair:
the key is string '__current_op_desc__'
and the value is the op_desc of the
gradient operator who has just
triggered the callable object.
In addition to this, the 'context'
has another special key-value pair:
the key is string '__current_op_desc__'
and the value is the op_desc of the
gradient operator who has just
triggered the callable object.
Returns:
list[(Variable,Variable)]: Pairs of parameter and its
corresponding gradients. The key is the parameter and the
list[(Variable,Variable)]: Pairs of parameter and its
corresponding gradients. The key is the parameter and the
value is gradient variable.
Raises:
......@@ -535,7 +553,7 @@ def append_backward(loss, parameter_list=None, no_grad_set=None,
no_grad_set = set()
no_grad_set = copy.copy(no_grad_set)
no_grad_dict = _get_stop_gradients_(program)
no_grad_dict[0].update(map(_append_grad_suffix_, no_grad_set))
no_grad_dict[0].update(list(map(_append_grad_suffix_, no_grad_set)))
grad_info_map = dict()
root_block = program.block(0)
......@@ -558,7 +576,7 @@ def append_backward(loss, parameter_list=None, no_grad_set=None,
block_no_grad_set = set(map(_strip_grad_suffix_, no_grad_dict[0]))
op_path = _find_op_path_(root_block, [loss], [], block_no_grad_set)
no_grad_dict[0].update(map(_append_grad_suffix_, block_no_grad_set))
no_grad_dict[0].update(list(map(_append_grad_suffix_, block_no_grad_set)))
_append_backward_ops_(root_block, op_path, root_block, no_grad_dict,
grad_to_var, callbacks)
......@@ -699,7 +717,7 @@ def calc_gradient(targets, inputs, target_gradients=None, no_grad_set=None):
no_grad_set = set()
no_grad_set = copy.copy(no_grad_set)
no_grad_dict = _get_stop_gradients_(prog)
no_grad_dict[0].update(map(_append_grad_suffix_, no_grad_set))
no_grad_dict[0].update(list(map(_append_grad_suffix_, no_grad_set)))
fwd_op_num = block.desc.op_size()
......@@ -733,7 +751,7 @@ def calc_gradient(targets, inputs, target_gradients=None, no_grad_set=None):
block_no_grad_set = set(map(_strip_grad_suffix_, no_grad_dict[0]))
op_path = _find_op_path_(block, targets, inputs, block_no_grad_set)
no_grad_dict[0].update(map(_append_grad_suffix_, block_no_grad_set))
no_grad_dict[0].update(list(map(_append_grad_suffix_, block_no_grad_set)))
grad_to_var = dict()
grad_info_map = dict()
_append_backward_ops_(block, op_path, block, no_grad_dict, grad_to_var)
......
python/paddle/fluid/clip.py
浏览文件 @ 559d3632
......@@ -15,8 +15,8 @@
import copy
import functools
import layers
import framework
from . import layers
from . import framework
from . import core
__all__ = [
......@@ -80,8 +80,7 @@ def error_clip_callback(block, context):
# the context is a grad_to_var map
grad_to_var = context
op_desc = block.desc.op(block.desc.op_size() - 1)
for grad_n in filter(lambda n: grad_to_var.has_key(n),
op_desc.output_arg_names()):
for grad_n in [n for n in op_desc.output_arg_names() if n in grad_to_var]:
fwd_var = block._var_recursive(grad_to_var[grad_n])
error_clip = getattr(fwd_var, "error_clip", None)
if not (error_clip is None or isinstance(error_clip,
......@@ -247,7 +246,7 @@ class GradientClipByGlobalNorm(BaseGradientClipAttr):
"""
def __init__(self, clip_norm, group_name="default_group"):
if not isinstance(group_name, basestring):
if not isinstance(group_name, str):
raise TypeError("'group_name' must be a basestring.")
self.clip_norm = clip_norm
......@@ -284,7 +283,7 @@ class GradientClipByGlobalNorm(BaseGradientClipAttr):
x=clip_var,
y=layers.elementwise_max(
x=clip_var, y=group_norm_var))
assert group_scale_var.shape == (1L, )
assert group_scale_var.shape == (1, )
self.context[group_scale_name] = group_scale_var
new_grad = layers.elementwise_mul(
......@@ -313,7 +312,7 @@ def set_gradient_clip(clip, param_list=None, program=None):
program = framework.default_main_program()
if param_list is None:
param_list = program.block(0).all_parameters()
if all(isinstance(elem, basestring) for elem in param_list):
if all(isinstance(elem, str) for elem in param_list):
param_list = [program.block(0).var(elem) for elem in param_list]
if not all(isinstance(elem, framework.Parameter) for elem in param_list):
raise TypeError(
......
python/paddle/fluid/concurrency.py
浏览文件 @ 559d3632
......@@ -12,11 +12,11 @@
# See the License for the specific language governing permissions and
# limitations under the License.
from layers.control_flow import BlockGuard, equal
from .layers.control_flow import BlockGuard, equal
from .framework import Operator
from layer_helper import LayerHelper, unique_name
from layers import fill_constant
import core
from .layer_helper import LayerHelper, unique_name
from .layers import fill_constant
from . import core
__all__ = [
'Go', 'make_channel', 'channel_send', 'channel_recv', 'channel_close',
......
python/paddle/fluid/data_feeder.py
浏览文件 @ 559d3632
......@@ -12,14 +12,13 @@
# See the License for the specific language governing permissions and
# limitations under the License.
from __future__ import print_function
import core
from . import core
import numpy
import os
import six.moves as six
import multiprocessing
from framework import Variable, default_main_program
from .framework import Variable, default_main_program
__all__ = ['DataFeeder']
......@@ -142,7 +141,7 @@ class DataFeeder(object):
if program is None:
program = default_main_program()
for each_var in feed_list:
if isinstance(each_var, basestring):
if isinstance(each_var, str):
each_var = program.block(0).var(each_var)
if not isinstance(each_var, Variable):
raise TypeError("Feed list should contain a list of variable")
......
python/paddle/fluid/debugger.py
浏览文件 @ 559d3632
......@@ -14,8 +14,8 @@
import sys
import re
from graphviz import GraphPreviewGenerator
import proto.framework_pb2 as framework_pb2
from .graphviz import GraphPreviewGenerator
from .proto import framework_pb2
from google.protobuf import text_format
_vartype2str_ = [
......
python/paddle/fluid/evaluator.py
浏览文件 @ 559d3632
......@@ -15,11 +15,11 @@
import warnings
import numpy as np
import layers
from framework import Program, Variable, program_guard
import unique_name
from layer_helper import LayerHelper
from initializer import Constant
from . import layers
from .framework import Program, Variable, program_guard
from . import unique_name
from .layer_helper import LayerHelper
from .initializer import Constant
__all__ = [
'ChunkEvaluator',
......
python/paddle/fluid/executor.py
浏览文件 @ 559d3632
......@@ -14,7 +14,7 @@
import numpy as np
import contextlib
from framework import Program, default_main_program, Variable
from .framework import Program, default_main_program, Variable
from . import core
__all__ = [
......@@ -204,19 +204,19 @@ def fetch_var(name, scope=None, return_numpy=True):
def _get_program_cache_key(feed, fetch_list):
feed_var_names = feed.keys()
feed_var_names = list(feed.keys())
def to_name_str(var):
if isinstance(var, Variable):
return var.desc.name()
elif isinstance(var, str):
return var
elif isinstance(var, basestring):
elif isinstance(var, str):
return str(var)
else:
raise TypeError(str(var) + " should be Variable or str")
fetch_var_names = map(to_name_str, fetch_list)
fetch_var_names = list(map(to_name_str, fetch_list))
return str(feed_var_names + fetch_var_names)
......@@ -345,7 +345,7 @@ class Executor(object):
def _fetch_data(self, fetch_list, fetch_var_name, scope):
outs = [
core.get_fetch_variable(scope, fetch_var_name, i)
for i in xrange(len(fetch_list))
for i in range(len(fetch_list))
]
return outs
......
python/paddle/fluid/framework.py
浏览文件 @ 559d3632
......@@ -15,21 +15,22 @@
import collections
import contextlib
import re
import six
import numpy as np
import proto.framework_pb2 as framework_pb2
from .proto import framework_pb2
try:
from . import core
except ImportError, e:
except ImportError as e:
raise ImportError(
"""NOTE: You may need to run \"export LD_LIBRARY_PATH=/usr/local/lib:$LD_LIBRARY_PATH\"
if you encounters \"libmkldnn.so not found\" errors. If you have python
installed in other directory, replace \"/usr/local/lib\" with your own
directory. The original error is: \n""" + e.message)
except Exception, e:
except Exception as e:
raise e
import unique_name
from . import unique_name
__all__ = [
'Program',
......@@ -86,7 +87,7 @@ def convert_np_dtype_to_dtype_(np_dtype):
elif dtype == np.uint8:
return core.VarDesc.VarType.UINT8
else:
raise ValueError("Not supported numpy dtype " + str(dtype))
raise ValueError("Not supported numpy dtype " + six.binary_type(dtype))
def dtype_is_floating(dtype):
......@@ -129,15 +130,15 @@ def _debug_string_(proto, throw_on_error=True):
class Variable(object):
"""
In Fluid, every input and output of an operator is a variable. In most
cases, variables are used for holding different kinds of data or training
labels. A variable belongs to a block. All variable has its own name and
In Fluid, every input and output of an operator is a variable. In most
cases, variables are used for holding different kinds of data or training
labels. A variable belongs to a block. All variable has its own name and
two variables in different blocks could have the same name.
There are many kinds of variables. Each kind of them has its own attributes
and usages. Please reference the framework.proto for details.
There are many kinds of variables. Each kind of them has its own attributes
and usages. Please reference the framework.proto for details.
Most of a Variable's member variables can be setted to be None. It mean
Most of a Variable's member variables can be setted to be None. It mean
it is not available or will be specified later.
Args:
......@@ -197,6 +198,7 @@ class Variable(object):
if name is None:
name = unique_name.generate('_generated_var')
is_new_var = False
name = name if isinstance(name, six.binary_type) else name.encode()
self.desc = self.block.desc.find_var(name)
if self.desc is None:
......@@ -290,13 +292,13 @@ class Variable(object):
assert isinstance(throw_on_error, bool) and isinstance(with_details,
bool)
protostr = self.desc.serialize_to_string()
proto = framework_pb2.VarDesc.FromString(str(protostr))
proto = framework_pb2.VarDesc.FromString(six.binary_type(protostr))
res_str = _debug_string_(proto, throw_on_error)
if with_details:
additional_attr = ("error_clip", "stop_gradient")
for attr_name in additional_attr:
res_str += "%s: %s\n" % (attr_name,
str(getattr(self, attr_name)))
res_str += "%s: %s\n" % (
attr_name, six.binary_type(getattr(self, attr_name)))
return res_str
__repr__ = __str__
......@@ -369,7 +371,7 @@ def get_all_op_protos():
protostrs = core.get_all_op_protos()
ret_values = []
for pbstr in protostrs:
op_proto = framework_pb2.OpProto.FromString(str(pbstr))
op_proto = framework_pb2.OpProto.FromString(six.binary_type(pbstr))
ret_values.append(op_proto)
return ret_values
......@@ -472,7 +474,6 @@ class Operator(object):
inputs=None,
outputs=None,
attrs=None):
self.block = block
self.desc = desc
self.attrs = attrs
......@@ -523,10 +524,15 @@ class Operator(object):
% (in_proto.name, len(in_args)))
in_arg_names = []
for arg in in_args:
if isinstance(arg, basestring):
if issubclass(arg.__class__, six.string_types):
in_arg_names.append(arg)
elif isinstance(arg, six.binary_type):
in_arg_names.append(arg.decode())
else:
in_arg_names.append(arg.name)
if issubclass(arg.name.__class__, six.string_types):
in_arg_names.append(arg.name)
elif isinstance(arg.name, six.binary_type):
in_arg_names.append(arg.name.decode())
self.desc.set_input(in_proto.name, in_arg_names)
else:
self.desc.set_input(in_proto.name, [])
......@@ -541,8 +547,9 @@ class Operator(object):
if not given == need:
raise ValueError(("Incorrect setting for output(s) of "
"operator \"%s\". Need: [%s] Given: [%s]") %
(type, ", ".join(str(e) for e in need),
", ".join(str(e) for e in given)))
(type,
", ".join(six.binary_type(e) for e in need),
", ".join(six.binary_type(e) for e in given)))
for out_proto in proto.outputs:
out_args = outputs[out_proto.name]
......@@ -554,7 +561,12 @@ class Operator(object):
(out_proto.name, len(out_args)))
out_arg_names = []
for arg in out_args:
out_arg_names.append(arg.name)
if issubclass(arg.name.__class__, six.string_types):
out_arg_names.append(arg.name)
elif isinstance(arg.name, six.binary_type):
out_arg_names.append(arg.name.decode())
else:
out_arg_names.append(six.u(arg.name))
arg.op = self
self.desc.set_output(out_proto.name, out_arg_names)
......@@ -590,7 +602,7 @@ class Operator(object):
"""
protostr = self.desc.serialize_to_string()
proto = framework_pb2.OpDesc.FromString(str(protostr))
proto = framework_pb2.OpDesc.FromString(six.binary_type(protostr))
return _debug_string_(proto, throw_on_error)
def __str__(self):
......@@ -845,7 +857,7 @@ class Block(object):
re_add_indent = re.compile(r"\n(.)")
res_str = "blocks {\n idx: %d\n parent_idx: %d" % (
self.idx, self.parent_idx)
for var in self.vars.itervalues():
for var in list(self.vars.values()):
res_str += "\n vars {\n %s }" % re_add_indent.sub(
r"\n \1", var.to_string(throw_on_error, with_details))
for op in self.ops:
......@@ -854,7 +866,8 @@ class Block(object):
res_str += "\n}"
else:
protostr = self.desc.serialize_to_string()
proto = framework_pb2.BlockDesc.FromString(str(protostr))
proto = framework_pb2.BlockDesc.FromString(
six.binary_type(protostr))
res_str = _debug_string_(proto, throw_on_error)
return res_str
......@@ -898,10 +911,11 @@ class Block(object):
Returns:
Variable: the Variable with the giving name.
"""
if not isinstance(name, basestring):
raise TypeError(
"var require string as parameter, but get %s instead." %
(type(name)))
if not issubclass(name.__class__, six.string_types):
if not isinstance(name, six.binary_type):
raise TypeError(
"var require string as parameter, but get %s instead." %
(type(name)))
v = self.vars.get(name, None)
if v is None:
raise ValueError("var %s not in this block" % name)
......@@ -949,10 +963,10 @@ class Block(object):
raise ValueError("Var {0} is not found recursively".format(name))
def all_parameters(self):
return list(self._iter_parameters())
return list(self.iter_parameters())
def _iter_parameters(self):
return (item[1] for item in self.vars.iteritems()
def iter_parameters(self):
return (item[1] for item in list(self.vars.items())
if isinstance(item[1], Parameter))
def create_var(self, *args, **kwargs):
......@@ -1113,7 +1127,7 @@ class Block(object):
self.create_var(name=var.name(), desc=var, type=var.type())
# sync variables removed from c++ end
for var in self.vars.keys():
for var in list(self.vars.keys()):
if not self.desc.find_var(var):
self.vars.pop(var)
......@@ -1185,7 +1199,7 @@ class Block(object):
if not isinstance(other, Block):
raise TypeError(
"_copy_param_info_from should be invoked with Block")
for p in other._iter_parameters():
for p in other.iter_parameters():
assert isinstance(p, Parameter)
v = self.vars.get(p.name, None)
if v is None:
......@@ -1384,7 +1398,8 @@ class Program(object):
res_str += block.to_string(throw_on_error, with_details)
else:
protostr = self.desc.serialize_to_string()
proto = framework_pb2.ProgramDesc.FromString(str(protostr))
proto = framework_pb2.ProgramDesc.FromString(
six.binary_type(protostr))
res_str = _debug_string_(proto, throw_on_error)
return res_str
......@@ -1482,7 +1497,7 @@ class Program(object):
else:
p = Program()
p.desc = core.ProgramDesc(self.desc)
p.blocks = [Block(p, i) for i in xrange(self.desc.num_blocks())]
p.blocks = [Block(p, i) for i in range(self.desc.num_blocks())]
p._sync_with_cpp()
p._copy_param_info_from(self)
......@@ -1534,7 +1549,7 @@ class Program(object):
targets_idx.append([t.block.idx, t.idx])
res = Program()
res.desc = core.prune(self.desc, targets_idx)
res.blocks = [Block(res, i) for i in xrange(res.desc.num_blocks())]
res.blocks = [Block(res, i) for i in range(res.desc.num_blocks())]
res._sync_with_cpp()
return res
......@@ -1554,13 +1569,13 @@ class Program(object):
# core.inference_optimize being fixed.
res = Program()
res.desc = core.ProgramDesc(self.desc)
for i in xrange(res.desc.num_blocks()):
for i in range(res.desc.num_blocks()):
block = res.desc.block(i)
for j in xrange(block.op_size()):
for j in range(block.op_size()):
op = block.op(j)
if op.has_attr('is_test'):
op.set_attr('is_test', True)
res.blocks = [Block(res, i) for i in xrange(res.desc.num_blocks())]
res.blocks = [Block(res, i) for i in range(res.desc.num_blocks())]
res._sync_with_cpp()
return res
......@@ -1573,14 +1588,14 @@ class Program(object):
and deserialization.
Args:
binary_str(str): The binary prootbuf string.
binary_str_type(str): The binary prootbuf string.
Returns:
Program: A deserialized program desc.
"""
p = Program()
p.desc = core.ProgramDesc(binary_str)
p.blocks = [Block(p, i) for i in xrange(p.desc.num_blocks())]
p.blocks = [Block(p, i) for i in range(p.desc.num_blocks())]
p._sync_with_cpp()
return p
......@@ -1608,7 +1623,7 @@ class Program(object):
self._seed = seed
def __repr__(self):
return str(self)
return self.__str__()
def global_block(self):
"""
......@@ -1719,7 +1734,7 @@ class Program(object):
if len(self.blocks) != len(other.blocks):
raise ValueError("_copy_param_info_from should be invoked with two "
"program, with represent the same topology")
for var in other.global_block().vars.itervalues():
for var in list(other.global_block().vars.values()):
if var.is_data:
self.global_block().var(var.name).is_data = True
......@@ -1731,15 +1746,15 @@ class Program(object):
iterable: The generator will yield every variable in this program.
"""
for each_block in self.blocks:
for each_var in each_block.vars.itervalues():
for each_var in list(each_block.vars.values()):
yield each_var
class Parameter(Variable):
"""
Parameter is derived from Variable. A parameter is a persistable
Parameter is derived from Variable. A parameter is a persistable
Variable, and will be updated by optimizers after each iteration.
The training of a neural network is essentially the updating of
The training of a neural network is essentially the updating of
its parameters.
Relative to a general Variable, a Parameter has several its own
......@@ -1805,8 +1820,8 @@ class Parameter(Variable):
additional_attr = ("trainable", "optimize_attr", "regularizer",
"gradient_clip_attr", "do_model_average")
for attr_name in additional_attr:
res_str += "%s: %s\n" % (attr_name,
str(getattr(self, attr_name)))
res_str += "%s: %s\n" % (
attr_name, six.binary_type(getattr(self, attr_name)))
else:
res_str = Variable.to_string(self, throw_on_error, False)
return res_str
......
python/paddle/fluid/graphviz.py
浏览文件 @ 559d3632
......@@ -19,7 +19,7 @@ import logging
def crepr(v):
if type(v) is str or type(v) is unicode:
if type(v) is str or type(v) is str:
return '"%s"' % v
return str(v)
......@@ -104,7 +104,7 @@ class Graph(object):
def _rank_repr(self):
ranks = sorted(
self.rank_groups.items(),
list(self.rank_groups.items()),
cmp=lambda a, b: a[1].priority > b[1].priority)
repr = []
for x in ranks:
......@@ -148,7 +148,7 @@ class Node(object):
name=self.name,
label=self.label,
extra=',' + ','.join("%s=%s" % (key, crepr(value))
for key, value in self.attrs.items())
for key, value in list(self.attrs.items()))
if self.attrs else "")
return reprs
......@@ -172,7 +172,7 @@ class Edge(object):
target=self.target.name,
extra="" if not self.attrs else
"[" + ','.join("{}={}".format(attr[0], crepr(attr[1]))
for attr in self.attrs.items()) + "]")
for attr in list(self.attrs.items())) + "]")
return repr
......
python/paddle/fluid/inferencer.py
浏览文件 @ 559d3632
......@@ -14,14 +14,14 @@
import contextlib
import core
import executor
import framework
import io
import parallel_executor
import unique_name
from trainer import check_and_get_place
from . import core
from . import executor
from . import framework
from . import io
from . import parallel_executor
from . import unique_name
from .trainer import check_and_get_place
__all__ = ['Inferencer', ]
......
python/paddle/fluid/initializer.py
浏览文件 @ 559d3632
......@@ -12,11 +12,11 @@
# See the License for the specific language governing permissions and
# limitations under the License.
import framework
from . import framework
import numpy as np
import contextlib
from framework import convert_np_dtype_to_dtype_
from core import VarDesc
from .framework import convert_np_dtype_to_dtype_
from .core import VarDesc
__all__ = [
'Constant', 'Uniform', 'Normal', 'Xavier', 'Bilinear', 'MSRA',
......
python/paddle/fluid/io.py
浏览文件 @ 559d3632
......@@ -16,6 +16,7 @@ import os
import errno
import time
import shutil
import six
from paddle.fluid.evaluator import Evaluator
from paddle.fluid.framework import Program, Parameter, default_main_program, default_startup_program, Variable
......@@ -92,34 +93,34 @@ def save_vars(executor,
"""
Save variables to the given directory by executor.
There are two ways to specify variables to be saved: The first way, list
variables in a list and assign it to the `vars`. The second way, assign the
`main_program` with an existing program, then all variables in the program
will be saved. The first way has a higher priority. In other words, if `vars`
There are two ways to specify variables to be saved: The first way, list
variables in a list and assign it to the `vars`. The second way, assign the
`main_program` with an existing program, then all variables in the program
will be saved. The first way has a higher priority. In other words, if `vars`
are assigned, the `main_program` and the `predicate` will be ignored.
The `dirname` are used to specify the folder where to save variables.
If you prefer to save variables in separate files in the folder `dirname`,
set `filename` None; if you prefer to save all variables in a single file,
The `dirname` are used to specify the folder where to save variables.
If you prefer to save variables in separate files in the folder `dirname`,
set `filename` None; if you prefer to save all variables in a single file,
use `filename` to specify it.
Args:
executor(Executor): The executor to run for saving variables.
dirname(str): The directory path.
main_program(Program|None): The program whose variables will be saved.
If it is None, the default main program will
main_program(Program|None): The program whose variables will be saved.
If it is None, the default main program will
be used automatically.
Default: None
vars(list[Variable]|None): The list that contains all variables to save.
vars(list[Variable]|None): The list that contains all variables to save.
It has a higher priority than the `main_program`.
Default: None
predicate(function|None): If it is not None, only variables in the
`main_program` that makes predicate(variable)==True
will be saved. It only works when we are using the
`main_program` to specify variables (In other words
predicate(function|None): If it is not None, only variables in the
`main_program` that makes predicate(variable)==True
will be saved. It only works when we are using the
`main_program` to specify variables (In other words
`vars` is None).
Default: None
filename(str|None): The file which to save all variables. If you prefer to save
filename(str|None): The file which to save all variables. If you prefer to save
variables separately, set it to None.
Default: None
......@@ -149,7 +150,7 @@ def save_vars(executor,
# The second usage: using `vars` to specify variables
var_list = [var_a, var_b, var_c]
fluid.io.save_vars(executor=exe, dirname=path, vars=var_list,
fluid.io.save_vars(executor=exe, dirname=path, vars=var_list,
filename="vars_file")
# var_a, var_b and var_c will be saved. And they are going to be
# saved in the same file named 'var_file' in the path "./my_paddle_model".
......@@ -163,7 +164,7 @@ def save_vars(executor,
save_vars(
executor,
dirname=dirname,
vars=filter(predicate, main_program.list_vars()),
vars=list(filter(predicate, main_program.list_vars())),
filename=filename)
else:
save_program = Program()
......@@ -203,14 +204,14 @@ def save_params(executor, dirname, main_program=None, filename=None):
This function filters out all parameters from the give `main_program`
and then save them to the folder `dirname` or the file `filename`.
Use the `dirname` to specify the saving folder. If you would like to
save parameters in separate files, set `filename` None; if you would
like to save all parameters in a single file, use `filename` to specify
Use the `dirname` to specify the saving folder. If you would like to
save parameters in separate files, set `filename` None; if you would
like to save all parameters in a single file, use `filename` to specify
the file name.
NOTICE: Some variables are not Parameter while they are necessary for
training. So you can NOT save and continue your training just by
`save_params()` and `load_params()`. Please use `save_persistables()`
NOTICE: Some variables are not Parameter while they are necessary for
training. So you can NOT save and continue your training just by
`save_params()` and `load_params()`. Please use `save_persistables()`
and `load_persistables()` instead.
Args:
......@@ -220,8 +221,8 @@ def save_params(executor, dirname, main_program=None, filename=None):
saved. If it is None, the default
main program will be used automatically.
Default: None
filename(str|None): The file to save all parameters. If you prefer
to save parameters in differnet files, set it
filename(str|None): The file to save all parameters. If you prefer
to save parameters in differnet files, set it
to None.
Default: None
......@@ -234,7 +235,7 @@ def save_params(executor, dirname, main_program=None, filename=None):
exe = fluid.Executor(fluid.CPUPlace())
param_path = "./my_paddle_model"
prog = fluid.default_main_program()
fluid.io.save_params(executor=exe, dirname=param_path,
fluid.io.save_params(executor=exe, dirname=param_path,
main_program=None)
"""
save_vars(
......@@ -248,23 +249,23 @@ def save_params(executor, dirname, main_program=None, filename=None):
def save_persistables(executor, dirname, main_program=None, filename=None):
"""
This function filters out all variables with `persistable==True` from the
give `main_program` and then saves these variables to the folder `dirname`
This function filters out all variables with `persistable==True` from the
give `main_program` and then saves these variables to the folder `dirname`
or file `filename`.
The `dirname` is used to specify the folder where persistable variables
are going to be saved. If you would like to save variables in separate
files, set `filename` None; if you would like to save all variables in a
The `dirname` is used to specify the folder where persistable variables
are going to be saved. If you would like to save variables in separate
files, set `filename` None; if you would like to save all variables in a
single file, use `filename` to specify the file name.
Args:
executor(Executor): The executor to run for saving persistable variables.
dirname(str): The directory path.
main_program(Program|None): The program whose persistbale variables will
be saved. If it is None, the default main
main_program(Program|None): The program whose persistbale variables will
be saved. If it is None, the default main
program will be used automatically.
Default: None
filename(str|None): The file to saved all variables. If you prefer to
filename(str|None): The file to saved all variables. If you prefer to
save variables in differnet files, set it to None.
Default: None
......@@ -277,7 +278,7 @@ def save_persistables(executor, dirname, main_program=None, filename=None):
exe = fluid.Executor(fluid.CPUPlace())
param_path = "./my_paddle_model"
prog = fluid.default_main_program()
fluid.io.save_persistables(executor=exe, dirname=param_path,
fluid.io.save_persistables(executor=exe, dirname=param_path,
main_program=None)
"""
save_vars(
......@@ -298,34 +299,34 @@ def load_vars(executor,
"""
Load variables from the given directory by executor.
There are two ways to specify variables to be loaded: The first way, list
variables in a list and assign it to the `vars`. The second way, assign the
`main_program` with an existing program, then all variables in the program
will be loaded. The first way has a higher priority. In other words if `vars`
There are two ways to specify variables to be loaded: The first way, list
variables in a list and assign it to the `vars`. The second way, assign the
`main_program` with an existing program, then all variables in the program
will be loaded. The first way has a higher priority. In other words if `vars`
are assigned, the `main_program` and the `predicate` will be ignored.
The `dirname` are used to specify the folder where to load variables.
If variables were saved in separate files in the folder `dirname`,
set `filename` None; if all variables were saved in a single file,
The `dirname` are used to specify the folder where to load variables.
If variables were saved in separate files in the folder `dirname`,
set `filename` None; if all variables were saved in a single file,
use `filename` to specify it.
Args:
executor(Executor): The executor to run for loading variables.
dirname(str): The directory path.
main_program(Program|None): The program whose variables will be loaded.
If it is None, the default main program will
main_program(Program|None): The program whose variables will be loaded.
If it is None, the default main program will
be used automatically.
Default: None
vars(list[Variable]|None): The list that contains all variables to load.
vars(list[Variable]|None): The list that contains all variables to load.
It has a higher priority than the `main_program`.
Default: None
predicate(function|None): If it is not None, only variables in the
`main_program` that makes predicate(variable)==True
will be loaded. It only works when we are using the
`main_program` to specify variables (In other words
predicate(function|None): If it is not None, only variables in the
`main_program` that makes predicate(variable)==True
will be loaded. It only works when we are using the
`main_program` to specify variables (In other words
`vars` is None).
Default: None
filename(str|None): The file which saved all required variables. If variables
filename(str|None): The file which saved all required variables. If variables
were saved in differnet files, set it to None.
Default: None
......@@ -355,9 +356,9 @@ def load_vars(executor,
# The second usage: using `vars` to specify variables
var_list = [var_a, var_b, var_c]
fluid.io.load_vars(executor=exe, dirname=path, vars=var_list,
fluid.io.load_vars(executor=exe, dirname=path, vars=var_list,
filename="vars_file")
# var_a, var_b and var_c will be loaded. And they are supposed to haven
# var_a, var_b and var_c will be loaded. And they are supposed to haven
# been saved in the same file named 'var_file' in the path "./my_paddle_model".
"""
if vars is None:
......@@ -369,7 +370,7 @@ def load_vars(executor,
load_vars(
executor,
dirname=dirname,
vars=filter(predicate, main_program.list_vars()),
vars=list(filter(predicate, main_program.list_vars())),
filename=filename)
else:
load_prog = Program()
......@@ -410,15 +411,15 @@ def load_params(executor, dirname, main_program=None, filename=None):
and then trys to load these parameters from the folder `dirname` or
the file `filename`.
Use the `dirname` to specify the folder where parameters were saved. If
parameters were saved in separate files in the folder `dirname`, set
`filename` None; if all parameters were saved in a single file, use
Use the `dirname` to specify the folder where parameters were saved. If
parameters were saved in separate files in the folder `dirname`, set
`filename` None; if all parameters were saved in a single file, use
`filename` to specify the file name.
NOTICE: Some variables are not Parameter while they are necessary for
training. So you can NOT save and continue your training just by
`save_params()` and `load_params()`. Please use `save_persistables()`
and `load_persistables()` instead.
NOTICE: Some variables are not Parameter while they are necessary for
training. So you can NOT save and continue your training just by
`save_params()` and `load_params()`. Please use `save_persistables()`
and `load_persistables()` instead.
Args:
executor(Executor): The executor to run for loading parameters.
......@@ -427,7 +428,7 @@ def load_params(executor, dirname, main_program=None, filename=None):
loaded. If it is None, the default
main program will be used automatically.
Default: None
filename(str|None): The file which saved all parameters. If parameters
filename(str|None): The file which saved all parameters. If parameters
were saved in differnet files, set it to None.
Default: None
......@@ -440,7 +441,7 @@ def load_params(executor, dirname, main_program=None, filename=None):
exe = fluid.Executor(fluid.CPUPlace())
param_path = "./my_paddle_model"
prog = fluid.default_main_program()
fluid.io.load_params(executor=exe, dirname=param_path,
fluid.io.load_params(executor=exe, dirname=param_path,
main_program=None)
"""
load_vars(
......@@ -453,23 +454,23 @@ def load_params(executor, dirname, main_program=None, filename=None):
def load_persistables(executor, dirname, main_program=None, filename=None):
"""
This function filters out all variables with `persistable==True` from the
give `main_program` and then trys to load these variables from the folder
This function filters out all variables with `persistable==True` from the
give `main_program` and then trys to load these variables from the folder
`dirname` or the file `filename`.
Use the `dirname` to specify the folder where persistable variables were
saved. If variables were saved in separate files, set `filename` None;
if all variables were saved in a single file, use `filename` to specify
Use the `dirname` to specify the folder where persistable variables were
saved. If variables were saved in separate files, set `filename` None;
if all variables were saved in a single file, use `filename` to specify
the file name.
Args:
executor(Executor): The executor to run for loading persistable variables.
dirname(str): The directory path.
main_program(Program|None): The program whose persistbale variables will
be loaded. If it is None, the default main
main_program(Program|None): The program whose persistbale variables will
be loaded. If it is None, the default main
program will be used automatically.
Default: None
filename(str|None): The file which saved all variables. If variables were
filename(str|None): The file which saved all variables. If variables were
saved in differnet files, set it to None.
Default: None
......@@ -482,7 +483,7 @@ def load_persistables(executor, dirname, main_program=None, filename=None):
exe = fluid.Executor(fluid.CPUPlace())
param_path = "./my_paddle_model"
prog = fluid.default_main_program()
fluid.io.load_persistables(executor=exe, dirname=param_path,
fluid.io.load_persistables(executor=exe, dirname=param_path,
main_program=None)
"""
load_vars(
......@@ -561,20 +562,20 @@ def save_inference_model(dirname,
Args:
dirname(str): The directory path to save the inference model.
feeded_var_names(list[str]): Names of variables that need to be feeded data
feeded_var_names(list[str]): Names of variables that need to be feeded data
during inference.
target_vars(list[Variable]): Variables from which we can get inference
target_vars(list[Variable]): Variables from which we can get inference
results.
executor(Executor): The executor that saves the inference model.
main_program(Program|None): The original program, which will be pruned to
build the inference model. If is setted None,
main_program(Program|None): The original program, which will be pruned to
build the inference model. If is setted None,
the default main program will be used.
Default: None.
model_filename(str|None): The name of file to save the inference program
itself. If is setted None, a default filename
model_filename(str|None): The name of file to save the inference program
itself. If is setted None, a default filename
`__model__` will be used.
params_filename(str|None): The name of file to save all related parameters.
If it is setted None, parameters will be saved
params_filename(str|None): The name of file to save all related parameters.
If it is setted None, parameters will be saved
in separate files .
Returns:
......@@ -592,20 +593,34 @@ def save_inference_model(dirname,
fluid.io.save_inference_model(dirname=path, feeded_var_names=['img'],
target_vars=[predict_var], executor=exe)
# In this exsample, the function will prune the default main program
# to make it suitable for infering the `predict_var`. The pruned
# inference program is going to be saved in the "./infer_model/__model__"
# In this exsample, the function will prune the default main program
# to make it suitable for infering the `predict_var`. The pruned
# inference program is going to be saved in the "./infer_model/__model__"
# and parameters are going to be saved in separate files under folder
# "./infer_model".
# "./infer_model".
"""
if isinstance(feeded_var_names, basestring):
if isinstance(feeded_var_names, six.binary_type):
feeded_var_names = [feeded_var_names]
elif isinstance(feeded_var_names, six.text_type):
feeded_var_names = [feeded_var_names.encode()]
else:
if len(feeded_var_names) > 0:
if not (bool(feeded_var_names) and all(
isinstance(name, basestring) for name in feeded_var_names)):
raise ValueError("'feed_var_names' should be a list of str.")
isinstance(name, six.binary_type)
for name in feeded_var_names)):
if not (all(
isinstance(name, six.text_type)
for name in feeded_var_names)):
import sys
print([type(name) for name in feeded_var_names])
sys.stdout.flush()
raise ValueError(
"'feed_var_names' should be a list of str.")
else:
feeded_var_names = [
name.encode() for name in feeded_var_names
]
if isinstance(target_vars, Variable):
target_vars = [target_vars]
......@@ -662,22 +677,22 @@ def load_inference_model(dirname,
dirname(str): The directory path
executor(Executor): The executor to run for loading inference model.
model_filename(str|None): The name of file to load inference program.
If it is None, the default filename
If it is None, the default filename
'__model__' will be used.
Default: None
params_filename(str|None): The name of file to load all parameters.
It is only used for the case that all
parameters were saved in a single binary
file. If parameters were saved in separate
It is only used for the case that all
parameters were saved in a single binary
file. If parameters were saved in separate
files, set it as 'None'.
Returns:
tuple: The return of this function is a tuple with three elements:
(program, feed_target_names, fetch_targets). The `program` is a
Program, it's the program for inference. The `feed_target_names` is
a list of str, it contains Names of variables that need to feed
data in the inference program. The `fetch_targets` is a list of
Variable. It contains variables from which we can get inference
(program, feed_target_names, fetch_targets). The `program` is a
Program, it's the program for inference. The `feed_target_names` is
a list of str, it contains Names of variables that need to feed
data in the inference program. The `fetch_targets` is a list of
Variable. It contains variables from which we can get inference
results.
Raises:
......@@ -688,17 +703,17 @@ def load_inference_model(dirname,
exe = fluid.Executor(fluid.CPUPlace())
path = "./infer_model"
[inference_program, feed_target_names, fetch_targets] =
[inference_program, feed_target_names, fetch_targets] =
fluid.io.load_inference_model(dirname=path, executor=exe)
results = exe.run(inference_program,
feed={feed_target_names[0]: tensor_img},
fetch_list=fetch_targets)
# In this exsample, the inference program was saved in the
# "./infer_model/__model__" and parameters were saved in
# separate files in ""./infer_model".
# After getting inference program, feed target names and
# fetch targets, we can use an Executor to run the inference
# In this exsample, the inference program was saved in the
# "./infer_model/__model__" and parameters were saved in
# separate files in ""./infer_model".
# After getting inference program, feed target names and
# fetch targets, we can use an Executor to run the inference
# program to get the inference result.
"""
......
python/paddle/fluid/layer_helper.py
浏览文件 @ 559d3632
......@@ -15,11 +15,11 @@
import copy
import itertools
from framework import Variable, Parameter, default_main_program, default_startup_program, dtype_is_floating
import unique_name
from .framework import Variable, Parameter, default_main_program, default_startup_program, dtype_is_floating
from . import unique_name
from paddle.fluid.initializer import Constant, Xavier
from param_attr import ParamAttr, WeightNormParamAttr
import core
from .param_attr import ParamAttr, WeightNormParamAttr
from . import core
class LayerHelper(object):
......@@ -83,7 +83,7 @@ class LayerHelper(object):
raise ValueError("parameter number mismatch")
elif len(param_attr) == 1 and length != 1:
tmp = [None] * length
for i in xrange(length):
for i in range(length):
tmp[i] = copy.deepcopy(param_attr[0])
param_attr = tmp
return param_attr
......@@ -91,7 +91,7 @@ class LayerHelper(object):
def iter_inputs_and_params(self, input_param_name='input'):
inputs = self.multiple_input(input_param_name)
param_attrs = self.multiple_param_attr(len(inputs))
for ipt, param_attr in itertools.izip(inputs, param_attrs):
for ipt, param_attr in zip(inputs, param_attrs):
yield ipt, param_attr
def input_dtype(self, input_param_name='input'):
......@@ -218,7 +218,7 @@ class LayerHelper(object):
norm = __norm_op(reshape, dim=0, block=block)
__reshape_op(norm, out=out, shape=out_shape, block=block)
else:
perm = range(len(x.shape))
perm = list(range(len(x.shape)))
perm[0], perm[dim] = dim, 0
transpose = __transpose_op(x, perm, block=block)
norm = __norm_op(transpose, dim=0, block=block)
......@@ -397,7 +397,7 @@ class LayerHelper(object):
act = self.kwargs.get('act', None)
if act is None:
return input_var
if isinstance(act, basestring):
if isinstance(act, str):
act = {'type': act}
if 'use_cudnn' in self.kwargs and self.kwargs.get('use_cudnn'):
......
python/paddle/fluid/layers/__init__.py
浏览文件 @ 559d3632
......@@ -12,25 +12,25 @@
# See the License for the specific language governing permissions and
# limitations under the License.
import ops
from ops import *
import nn
from nn import *
import io
from io import *
import tensor
from tensor import *
import control_flow
from control_flow import *
import device
from device import *
import math_op_patch
from math_op_patch import *
import detection
from detection import *
import metric_op
from metric_op import *
from learning_rate_scheduler import *
from . import ops
from .ops import *
from . import nn
from .nn import *
from . import io
from .io import *
from . import tensor
from .tensor import *
from . import control_flow
from .control_flow import *
from . import device
from .device import *
from . import math_op_patch
from .math_op_patch import *
from . import detection
from .detection import *
from . import metric_op
from .metric_op import *
from .learning_rate_scheduler import *
__all__ = []
__all__ += nn.__all__
......
python/paddle/fluid/layers/control_flow.py
浏览文件 @ 559d3632
......@@ -13,14 +13,15 @@
# limitations under the License.
import contextlib
from layer_function_generator import autodoc, templatedoc
from tensor import assign, fill_constant
from .layer_function_generator import autodoc, templatedoc
from .tensor import assign, fill_constant
from .. import core
from ..framework import Program, Variable, Operator
from ..layer_helper import LayerHelper, unique_name
from ..initializer import force_init_on_cpu
from ops import logical_and, logical_not, logical_or
from .ops import logical_and, logical_not, logical_or
import numpy
from functools import reduce
__all__ = [
'While',
......@@ -597,7 +598,7 @@ class StaticRNN(object):
boot_memories = []
pre_memories = []
memories = []
for _, mem in self.memories.iteritems():
for _, mem in list(self.memories.items()):
boot_memories.append(mem.init)
pre_memories.append(mem.pre_mem.name)
mem_var = rnn_block.var(mem.mem.name)
......@@ -1508,7 +1509,7 @@ class IfElse(object):
def __call__(self):
if self.status != self.OUT_IF_ELSE_BLOCKS:
raise ValueError("IfElse::__call__ must be out of sub-block")
false_len, true_len = map(len, self.output_table)
false_len, true_len = list(map(len, self.output_table))
if false_len == 0 and true_len == 0:
raise ValueError("Must invoke true_block/false_block before "
"__call__")
......
python/paddle/fluid/layers/detection.py
浏览文件 @ 559d3632
......@@ -15,12 +15,13 @@
All layers just related to the detection neural network.
"""
from layer_function_generator import generate_layer_fn
from layer_function_generator import autodoc, templatedoc
from .layer_function_generator import generate_layer_fn
from .layer_function_generator import autodoc, templatedoc
from ..layer_helper import LayerHelper
import tensor
import nn
from . import tensor
from . import nn
import math
from functools import reduce
__all__ = [
'prior_box',
......@@ -1031,7 +1032,7 @@ def multi_box_head(inputs,
min_sizes = []
max_sizes = []
step = int(math.floor(((max_ratio - min_ratio)) / (num_layer - 2)))
for ratio in xrange(min_ratio, max_ratio + 1, step):
for ratio in range(min_ratio, max_ratio + 1, step):
min_sizes.append(base_size * ratio / 100.)
max_sizes.append(base_size * (ratio + step) / 100.)
min_sizes = [base_size * .10] + min_sizes
......
python/paddle/fluid/layers/device.py
浏览文件 @ 559d3632
......@@ -15,7 +15,7 @@
All util layers.
"""
from layer_function_generator import autodoc
from .layer_function_generator import autodoc
from ..framework import unique_name
from ..layer_helper import LayerHelper
from ..annotations import deprecated
......
python/paddle/fluid/layers/io.py
浏览文件 @ 559d3632
......@@ -16,8 +16,8 @@ import multiprocessing
import threading
from ..data_feeder import DataFeeder
from control_flow import BlockGuard
from layer_function_generator import templatedoc
from .control_flow import BlockGuard
from .layer_function_generator import templatedoc
from .. import core
from ..executor import global_scope
from ..framework import convert_np_dtype_to_dtype_, default_main_program, \
......@@ -69,7 +69,7 @@ def data(name,
"""
helper = LayerHelper('data', **locals())
shape = list(shape)
for i in xrange(len(shape)):
for i in range(len(shape)):
if shape[i] is None:
shape[i] = -1
append_batch_size = False
......@@ -387,9 +387,9 @@ def random_data_generator(low, high, shapes, lod_levels, for_parallel=True):
Create a uniform random data generator
This layer returns a Reader Variable.
Instead of opening a file and reading data from it, this
Reader Variable generates float uniform random data by itself.
It can be used as a dummy reader to test a network without
Instead of opening a file and reading data from it, this
Reader Variable generates float uniform random data by itself.
It can be used as a dummy reader to test a network without
opening a real file.
Args:
......@@ -710,9 +710,9 @@ def open_files(filenames,
"""
Open files
This layer takes a list of files to read from and returns a Reader Variable.
Via the Reader Variable, we can get data from given files. All files must
have name suffixs to indicate their formats, e.g., '*.recordio'.
This layer takes a list of files to read from and returns a Reader Variable.
Via the Reader Variable, we can get data from given files. All files must
have name suffixs to indicate their formats, e.g., '*.recordio'.
Args:
filenames(list): The list of file names.
......@@ -828,9 +828,9 @@ def shuffle(reader, buffer_size):
def batch(reader, batch_size):
"""
This layer is a reader decorator. It takes a reader and adds
'batching' decoration on it. When reading with the result
decorated reader, output data will be automatically organized
This layer is a reader decorator. It takes a reader and adds
'batching' decoration on it. When reading with the result
decorated reader, output data will be automatically organized
to the form of batches.
Args:
......@@ -855,11 +855,11 @@ def batch(reader, batch_size):
# If we read data with the raw_reader:
# data = fluid.layers.read_file(raw_reader)
# We can only get data instance by instance.
#
#
# However, if we read data with the batch_reader:
# data = fluid.layers.read_file(batch_reader)
# Each 5 adjacent instances will be automatically combined together
# to become a batch. So what we get('data') is a batch data instead
# Each 5 adjacent instances will be automatically combined together
# to become a batch. So what we get('data') is a batch data instead
# of an instance.
"""
return __create_unshared_decorated_reader__(
......@@ -906,8 +906,8 @@ def read_file(reader):
"""
Execute the given reader and get data via it.
A reader is also a Variable. It can be a raw reader generated by
`fluid.layers.open_files()` or a decorated one generated by
A reader is also a Variable. It can be a raw reader generated by
`fluid.layers.open_files()` or a decorated one generated by
`fluid.layers.double_buffer()` and so on.
Args:
......@@ -1008,7 +1008,7 @@ class Preprocessor(object):
source_lod_levels = self.underlying_reader.desc.lod_levels()
self.source_var_names = [
unique_name("preprocessor_source")
for _ in xrange(len(source_shapes))
for _ in range(len(source_shapes))
]
source_vars = []
for var_name, shape, dtype, lod_level in zip(
......
python/paddle/fluid/layers/layer_function_generator.py
浏览文件 @ 559d3632
......@@ -12,11 +12,11 @@
# See the License for the specific language governing permissions and
# limitations under the License.
import re
import cStringIO
import functools
import warnings
import string
from six.moves import cStringIO
from ..proto import framework_pb2
from ..framework import OpProtoHolder, Variable
from ..layer_helper import LayerHelper
......@@ -70,7 +70,7 @@ def _generate_doc_string_(op_proto):
if not isinstance(op_proto, framework_pb2.OpProto):
raise TypeError("OpProto should be `framework_pb2.OpProto`")
buf = cStringIO.StringIO()
buf = cStringIO()
buf.write(escape_math(op_proto.comment))
buf.write('\nArgs:\n')
for each_input in op_proto.inputs:
......@@ -119,9 +119,9 @@ def generate_layer_fn(op_type):
"""
op_proto = OpProtoHolder.instance().get_op_proto(op_type)
not_intermediate_outputs = \
filter(lambda output: not output.intermediate, op_proto.outputs)
[output for output in op_proto.outputs if not output.intermediate]
intermediate_outputs = \
filter(lambda output: output.intermediate, op_proto.outputs)
[output for output in op_proto.outputs if output.intermediate]
if len(not_intermediate_outputs) != 1:
raise ValueError("Only one non intermediate output operator can be",
......
python/paddle/fluid/layers/learning_rate_scheduler.py
浏览文件 @ 559d3632
......@@ -20,10 +20,10 @@ User can also implement their own learning_rate_decay
strategy according to this module.
"""
import control_flow
import nn
import ops
import tensor
from . import control_flow
from . import nn
from . import ops
from . import tensor
from ..initializer import init_on_cpu
from ..framework import default_main_program, Parameter
......
python/paddle/fluid/layers/math_op_patch.py
浏览文件 @ 559d3632
......@@ -13,7 +13,7 @@
# limitations under the License.
from ..framework import Variable, unique_name
from layer_function_generator import OpProtoHolder
from .layer_function_generator import OpProtoHolder
from ..initializer import force_init_on_cpu
......
python/paddle/fluid/layers/metric_op.py
浏览文件 @ 559d3632
......@@ -20,7 +20,7 @@ from ..layer_helper import LayerHelper
from ..initializer import Normal, Constant
from ..framework import Variable
from ..param_attr import ParamAttr
import nn
from . import nn
__all__ = ['accuracy', 'auc']
......
python/paddle/fluid/layers/nn.py
浏览文件 @ 559d3632
......@@ -33,11 +33,12 @@ from ..layer_helper import LayerHelper
from ..initializer import Normal, Constant
from ..framework import Variable
from ..param_attr import ParamAttr
from layer_function_generator import autodoc, templatedoc
from tensor import concat
import utils
from .layer_function_generator import autodoc, templatedoc
from .tensor import concat
from . import utils
import random
from .. import unique_name
from functools import reduce
__all__ = [
'fc',
......@@ -4843,7 +4844,7 @@ def dice_loss(input, label, epsilon=0.00001):
loss = fluid.layers.dice_loss(input=predictions, label=label, 2)
"""
label = one_hot(label, depth=input.shape[-1])
reduce_dim = range(1, len(input.shape))
reduce_dim = list(range(1, len(input.shape)))
inse = reduce_sum(input * label, dim=reduce_dim)
dice_denominator = reduce_sum(
input, dim=reduce_dim) + reduce_sum(
......
python/paddle/fluid/layers/ops.py
浏览文件 @ 559d3632
......@@ -11,7 +11,7 @@
# 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 layer_function_generator import generate_layer_fn
from .layer_function_generator import generate_layer_fn
__activations__ = [
'sigmoid',
......
python/paddle/fluid/layers/tensor.py
浏览文件 @ 559d3632
......@@ -18,7 +18,7 @@ from ..framework import convert_np_dtype_to_dtype_
from ..framework import Variable
from ..initializer import Constant, force_init_on_cpu
from ..core import VarDesc
from layer_function_generator import templatedoc
from .layer_function_generator import templatedoc
import numpy
__all__ = [
......
python/paddle/fluid/lod_tensor.py
浏览文件 @ 559d3632
......@@ -12,7 +12,7 @@
# See the License for the specific language governing permissions and
# limitations under the License.
import core
from . import core
import numpy as np
__all__ = ['create_lod_tensor', 'create_random_int_lodtensor']
......@@ -24,7 +24,7 @@ def create_lod_tensor(data, recursive_seq_lens, place):
Create a lod tensor by doing the following:
1. Check that the length-based level of detail (LoD) also known as
1. Check that the length-based level of detail (LoD) also known as
recursive_sequence_lengths of the input is valid.
2. Convert recursive_sequence_lengths to a offset-based LoD.
......@@ -33,7 +33,7 @@ def create_lod_tensor(data, recursive_seq_lens, place):
CPU or GPU device (based on input place).
4. Set the level of detail (LoD) using the offset-based LoD.
Examples:
Suppose we want LoDTensor to hold data for sequences of word, where each
......@@ -51,7 +51,7 @@ def create_lod_tensor(data, recursive_seq_lens, place):
Args:
data(numpy.ndarray|list|LoDTensor): a numpy array or a LoDTensor or a
list holding the data to be copied.
recursive_seq_lens(list): a list of lists indicating the length-based level of detail
recursive_seq_lens(list): a list of lists indicating the length-based level of detail
info specified by the user.
place(Place): CPU or GPU place indicating where the data in the new
LoDTensor will be stored.
......@@ -62,10 +62,10 @@ def create_lod_tensor(data, recursive_seq_lens, place):
if isinstance(data, core.LoDTensor):
return create_lod_tensor(np.array(data), recursive_seq_lens, place)
elif isinstance(data, list):
# When input data is a list, it only deal with the case where the base element
# is an index of shape [1] and dtype int64 (e.g., word id). Hence, the generated
# LoDTensor will be of shape [n, 1] and dtype int64, where `n` is the total number
# of words or other indexes in the sequence.
# When input data is a list, it only deal with the case where the base element
# is an index of shape [1] and dtype int64 (e.g., word id). Hence, the generated
# LoDTensor will be of shape [n, 1] and dtype int64, where `n` is the total number
# of words or other indexes in the sequence.
new_recursive_seq_lens = []
for seq in data:
new_recursive_seq_lens.append(len(seq))
......@@ -109,12 +109,12 @@ def create_random_int_lodtensor(recursive_seq_lens, base_shape, place, low,
Suppose we want LoDTensor to hold data for sequences of word, where each
word is represented by an integer. If we want to create a LoDTensor to
represent two sentences, one of 2 words, and one of 3 words. Then
'base_shape' is [1], input length-based 'recursive_seq_lens' is [[2, 3]].
Then the overall shape of the LoDTensor would be [5, 1], holding 5 words
'base_shape' is [1], input length-based 'recursive_seq_lens' is [[2, 3]].
Then the overall shape of the LoDTensor would be [5, 1], holding 5 words
for two sentences.
Args:
recursive_seq_lens(list): a list of lists indicating the length-based
recursive_seq_lens(list): a list of lists indicating the length-based
level of detail info specified by the user.
base_shape(list): the shape of the basic element to be held by the
LoDTensor.
......@@ -124,11 +124,11 @@ def create_random_int_lodtensor(recursive_seq_lens, base_shape, place, low,
high(int): the upper bound of the random integers.
Returns:
A fluid LoDTensor object with tensor data and recursive_seq_lens info.
A fluid LoDTensor object with tensor data and recursive_seq_lens info.
"""
assert isinstance(base_shape, list), "base_shape should be a list"
# append the total number of basic elements to the front of its shape
overall_shape = [sum(recursive_seq_lens[-1])] + base_shape
# the range of integer data elements is [low, high]
# the range of integer data elements is [low, high]
data = np.random.random_integers(low, high, overall_shape).astype("int64")
return create_lod_tensor(data, recursive_seq_lens, place)
python/paddle/fluid/metrics.py
浏览文件 @ 559d3632
......@@ -79,10 +79,10 @@ class MetricBase(object):
"""
states = {
attr: value
for attr, value in self.__dict__.iteritems()
for attr, value in list(self.__dict__.items())
if not attr.startswith("_")
}
for attr, value in states.iteritems():
for attr, value in list(states.items()):
if isinstance(value, int):
setattr(self, attr, 0)
elif isinstance(value, float):
......@@ -105,7 +105,7 @@ class MetricBase(object):
"""
states = {
attr: value
for attr, value in self.__dict__.iteritems()
for attr, value in list(self.__dict__.items())
if not attr.startswith("_")
}
config = {}
......
python/paddle/fluid/net_drawer.py
浏览文件 @ 559d3632
......@@ -24,7 +24,7 @@ logger = logging.getLogger(__name__)
logger.setLevel(logging.INFO)
try:
from graphviz import Digraph
from .graphviz import Digraph
except ImportError:
logger.info(
'Cannot import graphviz, which is required for drawing a network. This '
......@@ -77,7 +77,7 @@ def parse_graph(program, graph, var_dict, **kwargs):
# fill the known variables
for block in program.blocks:
for var in block.vars:
if not var_dict.has_key(var):
if var not in var_dict:
var_dict[var] = "Feed"
temp_id = 0
......@@ -93,17 +93,17 @@ def parse_graph(program, graph, var_dict, **kwargs):
var_dict[arg] = op.type
for e in op.inputs:
for arg in e.arguments:
if var_dict.has_key(arg):
if arg in var_dict:
graph.edge(**draw_edge(var_dict, op, e, arg))
break # only plot the first block
def draw_graph(startup_program, main_program, **kwargs):
if kwargs.has_key("graph_attr"):
if "graph_attr" in kwargs:
GRAPH_STYLE.update(kwargs[graph_attr])
if kwargs.has_key("node_attr"):
if "node_attr" in kwargs:
OP_STYLE.update(kwargs[node_attr])
if kwargs.has_key("edge_attr"):
if "edge_attr" in kwargs:
VAR_STYLE.update(kwargs[edge_attr])
graph_id = unique_id()
......
python/paddle/fluid/nets.py
浏览文件 @ 559d3632
......@@ -11,7 +11,7 @@
# 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 layers
from . import layers
__all__ = [
"simple_img_conv_pool",
......@@ -210,7 +210,7 @@ def img_conv_group(input,
conv_with_batchnorm = __extend_list__(conv_with_batchnorm)
conv_batchnorm_drop_rate = __extend_list__(conv_batchnorm_drop_rate)
for i in xrange(len(conv_num_filter)):
for i in range(len(conv_num_filter)):
local_conv_act = conv_act
if conv_with_batchnorm[i]:
local_conv_act = None
......@@ -488,10 +488,11 @@ def scaled_dot_product_attention(queries,
trans_x = layers.transpose(x, perm=[0, 2, 1, 3])
return layers.reshape(
x=trans_x,
shape=map(int, [
trans_x.shape[0], trans_x.shape[1],
trans_x.shape[2] * trans_x.shape[3]
]))
shape=list(
map(int, [
trans_x.shape[0], trans_x.shape[1], trans_x.shape[2] *
trans_x.shape[3]
])))
q, k, v = __compute_qkv(queries, keys, values, num_heads)
......
python/paddle/fluid/op.py
浏览文件 @ 559d3632
......@@ -12,6 +12,8 @@
# See the License for the specific language governing permissions and
# limitations under the License.
import six
import paddle.fluid.core as core
import paddle.fluid.proto.framework_pb2 as framework_pb2
......@@ -24,13 +26,13 @@ def get_all_op_protos():
protostrs = core.get_all_op_protos()
ret_values = []
for pbstr in protostrs:
op_proto = framework_pb2.OpProto.FromString(str(pbstr))
op_proto = framework_pb2.OpProto.FromString(six.binary_type(pbstr))
ret_values.append(op_proto)
return ret_values
def is_str(s):
return isinstance(s, str) or isinstance(s, unicode)
return isinstance(s, str) or isinstance(s, str)
class OpDescCreationMethod(object):
......@@ -189,7 +191,7 @@ class OperatorFactory(object):
return self.get_op_info(t).method(**kwargs)
def types(self):
return self.op_methods.keys()
return list(self.op_methods.keys())
def get_op_info(self, t):
if t not in self.op_methods:
......@@ -197,13 +199,13 @@ class OperatorFactory(object):
return self.op_methods.get(t)
def get_op_input_names(self, type):
return map(lambda x: x[0], self.get_op_info(type).inputs)
return [x[0] for x in self.get_op_info(type).inputs]
def get_op_inputs(self, type):
return self.get_op_info(type).inputs
def get_op_output_names(self, type):
return map(lambda x: x[0], self.get_op_info(type).outputs)
return [x[0] for x in self.get_op_info(type).outputs]
def get_op_outputs(self, type):
return self.get_op_info(type).outputs
......
python/paddle/fluid/optimizer.py
浏览文件 @ 559d3632
......@@ -14,15 +14,15 @@
import re
from collections import defaultdict
from paddle.fluid.framework import Program, Variable
import framework
import layers
from backward import append_backward
from framework import program_guard
import unique_name
from initializer import Constant
from layer_helper import LayerHelper
from regularizer import append_regularization_ops
from clip import append_gradient_clip_ops, error_clip_callback
from . import framework
from . import layers
from .backward import append_backward
from .framework import program_guard
from . import unique_name
from .initializer import Constant
from .layer_helper import LayerHelper
from .regularizer import append_regularization_ops
from .clip import append_gradient_clip_ops, error_clip_callback
from contextlib import contextmanager
__all__ = [
......@@ -106,7 +106,7 @@ class Optimizer(object):
param_lr = param.optimize_attr['learning_rate']
if type(param_lr) == Variable:
# param learning rate has been updated (LARS)
print("returns updated param lr ", param_lr)
print(("returns updated param lr ", param_lr))
return param_lr
else:
if param_lr == 1.0:
......
python/paddle/fluid/parallel_executor.py
浏览文件 @ 559d3632
......@@ -12,10 +12,11 @@
# See the License for the specific language governing permissions and
# limitations under the License.
import core
from __future__ import print_function
import multiprocessing
import framework
import executor
from . import core
from . import framework
from . import executor
import warnings
import sys
import os
......@@ -94,7 +95,7 @@ class ParallelExecutor(object):
self._places = []
self._act_places = []
if use_cuda:
for i in xrange(core.get_cuda_device_count()):
for i in range(core.get_cuda_device_count()):
p = core.Place()
self._act_places.append(core.CUDAPlace(i))
p.set_place(self._act_places[-1])
......@@ -102,7 +103,7 @@ class ParallelExecutor(object):
else:
cpu_num = int(
os.environ.get('CPU_NUM', multiprocessing.cpu_count()))
for i in xrange(cpu_num):
for i in range(cpu_num):
p = core.Place()
self._act_places.append(core.CPUPlace())
p.set_place(self._act_places[-1])
......@@ -143,16 +144,16 @@ class ParallelExecutor(object):
) if share_vars_from else []
self.persistable_vars = [
v.name
for v in filter(
lambda var: var.persistable and var.type != core.VarDesc.VarType.RAW,
main.list_vars())
v.name for v in [
var for var in main.list_vars()
if var.persistable and var.type != core.VarDesc.VarType.RAW
]
]
self.executor = core.ParallelExecutor(
self._places,
set([
p.name for p in main.global_block()._iter_parameters()
p.name for p in main.global_block().iter_parameters()
if not p.stop_gradient
]),
set(self.persistable_vars), main.desc, loss_name
......@@ -227,7 +228,9 @@ class ParallelExecutor(object):
"""
if feed is None and feed_dict is not None:
feed = feed_dict
print >> sys.stderr, "`feed_dict` is deprecated. Please use `feed=`"
print(
"`feed_dict` is deprecated. Please use `feed=`",
file=sys.stderr)
if isinstance(feed, dict):
feed_tensor_dict = dict()
......
python/paddle/fluid/param_attr.py
浏览文件 @ 559d3632
......@@ -12,8 +12,8 @@
# See the License for the specific language governing permissions and
# limitations under the License.
from initializer import Initializer, Xavier, Constant
from regularizer import WeightDecayRegularizer
from .initializer import Initializer, Xavier, Constant
from .regularizer import WeightDecayRegularizer
__all__ = [
'ParamAttr',
......@@ -134,7 +134,7 @@ class ParamAttr(object):
return [ParamAttr._to_attr(a) for a in arg]
elif isinstance(arg, ParamAttr):
return arg
elif isinstance(arg, str) or isinstance(arg, unicode):
elif isinstance(arg, str) or isinstance(arg, str):
return ParamAttr(name=arg)
elif isinstance(arg, Initializer):
return ParamAttr(initializer=arg)
......
python/paddle/fluid/profiler.py
浏览文件 @ 559d3632
......@@ -12,7 +12,7 @@
# See the License for the specific language governing permissions and
# limitations under the License.
import core
from . import core
from contextlib import contextmanager
import os
......@@ -224,7 +224,7 @@ def profiler(state, sorted_key=None, profile_path='/tmp/profile'):
If the state == 'All', a profile proto file will be written to
`profile_path`. This file records timeline information during the execution.
Then users can visualize this file to see the timeline, please refer
Then users can visualize this file to see the timeline, please refer
https://github.com/PaddlePaddle/Paddle/blob/develop/doc/fluid/howto/optimization/timeline.md
Args:
......
python/paddle/fluid/recordio_writer.py
浏览文件 @ 559d3632
......@@ -13,8 +13,8 @@
# limitations under the License.
import os
import core
import contextlib
from . import core
__all__ = [
'convert_reader_to_recordio_file', 'convert_reader_to_recordio_files'
]
......
python/paddle/fluid/regularizer.py
浏览文件 @ 559d3632
......@@ -12,7 +12,7 @@
# See the License for the specific language governing permissions and
# limitations under the License.
import framework
from . import framework
from . import core
__all__ = ['L1Decay', 'L2Decay', 'L1DecayRegularizer', 'L2DecayRegularizer']
......
python/paddle/fluid/tests/book/high-level-api/fit_a_line/test_fit_a_line.py
浏览文件 @ 559d3632
......@@ -63,7 +63,7 @@ def train(use_cuda, train_program, params_dirname):
if event.step == 10:
test_metrics = trainer.test(
reader=test_reader, feed_order=['x', 'y'])
print test_metrics
print(test_metrics)
'''
...
['25.768919467926025']
......@@ -94,7 +94,7 @@ def infer(use_cuda, inference_program, params_dirname=None):
tensor_x = numpy.random.uniform(0, 10, [batch_size, 13]).astype("float32")
results = inferencer.infer({'x': tensor_x})
print("infer results: ", results[0])
print(("infer results: ", results[0]))
def main(use_cuda):
......
python/paddle/fluid/tests/book/high-level-api/image_classification/cifar10_small_test_set.py
浏览文件 @ 559d3632
......@@ -28,7 +28,7 @@ images per class.
"""
import cPickle
import pickle
import itertools
import numpy
import paddle.v2.dataset.common
......@@ -46,7 +46,7 @@ def reader_creator(filename, sub_name, batch_size=None):
data = batch['data']
labels = batch.get('labels', batch.get('fine_labels', None))
assert labels is not None
for sample, label in itertools.izip(data, labels):
for sample, label in zip(data, labels):
yield (sample / 255.0).astype(numpy.float32), int(label)
def reader():
......@@ -56,7 +56,7 @@ def reader_creator(filename, sub_name, batch_size=None):
batch_count = 0
for name in names:
batch = cPickle.load(f.extractfile(name))
batch = pickle.load(f.extractfile(name))
for item in read_batch(batch):
if isinstance(batch_size, int) and batch_count > batch_size:
break
......
python/paddle/fluid/tests/book/high-level-api/image_classification/test_image_classification_resnet.py
浏览文件 @ 559d3632
......@@ -12,8 +12,6 @@
# See the License for the specific language governing permissions and
# limitations under the License.
from __future__ import print_function
import paddle
import paddle.fluid as fluid
import numpy
......@@ -107,7 +105,7 @@ def train(use_cuda, train_program, params_dirname):
avg_cost, accuracy = trainer.test(
reader=test_reader, feed_order=['pixel', 'label'])
print('Loss {0:2.2}, Acc {1:2.2}'.format(avg_cost, accuracy))
print(('Loss {0:2.2}, Acc {1:2.2}'.format(avg_cost, accuracy)))
if accuracy > 0.01: # Low threshold for speeding up CI
if params_dirname is not None:
......@@ -136,7 +134,7 @@ def infer(use_cuda, inference_program, params_dirname=None):
tensor_img = numpy.random.rand(1, 3, 32, 32).astype("float32")
results = inferencer.infer({'pixel': tensor_img})
print("infer results: ", results)
print(("infer results: ", results))
def main(use_cuda):
......
python/paddle/fluid/tests/book/high-level-api/image_classification/test_image_classification_vgg.py
浏览文件 @ 559d3632
......@@ -12,8 +12,6 @@
# See the License for the specific language governing permissions and
# limitations under the License.
from __future__ import print_function
import paddle
import paddle.fluid as fluid
import numpy
......@@ -84,7 +82,7 @@ def train(use_cuda, train_program, params_dirname):
avg_cost, accuracy = trainer.test(
reader=test_reader, feed_order=['pixel', 'label'])
print('Loss {0:2.2}, Acc {1:2.2}'.format(avg_cost, accuracy))
print(('Loss {0:2.2}, Acc {1:2.2}'.format(avg_cost, accuracy)))
if accuracy > 0.01: # Low threshold for speeding up CI
if params_dirname is not None:
......@@ -113,7 +111,7 @@ def infer(use_cuda, inference_program, params_dirname=None):
tensor_img = numpy.random.rand(1, 3, 32, 32).astype("float32")
results = inferencer.infer({'pixel': tensor_img})
print("infer results: ", results)
print(("infer results: ", results))
def main(use_cuda):
......
python/paddle/fluid/tests/book/high-level-api/label_semantic_roles/test_label_semantic_roles_newapi.py
浏览文件 @ 559d3632
......@@ -12,8 +12,6 @@
# See the License for the specific language governing permissions and
# limitations under the License.
from __future__ import print_function
import paddle
import paddle.fluid as fluid
import numpy as np
......@@ -173,19 +171,20 @@ def train(use_cuda, train_program, params_dirname):
# get avg cost
avg_cost = np.array(avg_cost_set).mean()
print("avg_cost: %s" % avg_cost)
print(("avg_cost: %s" % avg_cost))
if float(avg_cost) < 100.0: # Large value to increase CI speed
trainer.save_params(params_dirname)
else:
print('BatchID {0}, Test Loss {1:0.2}'.format(event.epoch + 1,
float(avg_cost)))
print(
('BatchID {0}, Test Loss {1:0.2}'.format(event.epoch + 1,
float(avg_cost))))
if math.isnan(float(avg_cost)):
sys.exit("got NaN loss, training failed.")
elif isinstance(event, fluid.EndStepEvent):
print("Step {0}, Epoch {1} Metrics {2}".format(
event.step, event.epoch, map(np.array, event.metrics)))
print(("Step {0}, Epoch {1} Metrics {2}".format(
event.step, event.epoch, list(map(np.array, event.metrics)))))
if event.step == 1: # Run 2 iterations to speed CI
trainer.save_params(params_dirname)
trainer.stop()
......@@ -249,7 +248,7 @@ def infer(use_cuda, inference_program, params_dirname):
},
return_numpy=False)
print("infer results: ", np.array(results[0]).shape)
print(("infer results: ", np.array(results[0]).shape))
def main(use_cuda):
......
python/paddle/fluid/tests/book/high-level-api/machine_translation/test_machine_translation.py
浏览文件 @ 559d3632
......@@ -197,7 +197,7 @@ def train(use_cuda, is_sparse, is_local=True):
def event_handler(event):
if isinstance(event, fluid.EndStepEvent):
print('pass_id=' + str(event.epoch) + ' batch=' + str(event.step))
print(('pass_id=' + str(event.epoch) + ' batch=' + str(event.step)))
if event.step == 10:
trainer.stop()
......@@ -250,7 +250,7 @@ def decode_main(use_cuda, is_sparse):
feeder = fluid.DataFeeder(feed_list, place)
for data in train_data():
feed_dict = feeder.feed(map(lambda x: [x[0]], data))
feed_dict = feeder.feed([[x[0]] for x in data])
feed_dict['init_ids'] = init_ids
feed_dict['init_scores'] = init_scores
......@@ -259,7 +259,7 @@ def decode_main(use_cuda, is_sparse):
feed=feed_dict,
fetch_list=[translation_ids, translation_scores],
return_numpy=False)
print result_ids.recursive_sequence_lengths()
print((result_ids.recursive_sequence_lengths()))
break
......
python/paddle/fluid/tests/book/high-level-api/recognize_digits/test_recognize_digits_conv.py
浏览文件 @ 559d3632
......@@ -11,7 +11,7 @@
# 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 argparse
import paddle.fluid as fluid
import paddle.fluid.core as core
......@@ -78,19 +78,21 @@ def train(use_cuda, train_program, params_dirname):
avg_cost, acc = trainer.test(
reader=test_reader, feed_order=['img', 'label'])
print("avg_cost: %s" % avg_cost)
print("acc : %s" % acc)
print(("avg_cost: %s" % avg_cost))
print(("acc : %s" % acc))
if acc > 0.2: # Smaller value to increase CI speed
trainer.save_params(params_dirname)
else:
print('BatchID {0}, Test Loss {1:0.2}, Acc {2:0.2}'.format(
event.epoch + 1, avg_cost, acc))
print(('BatchID {0}, Test Loss {1:0.2}, Acc {2:0.2}'.format(
event.epoch + 1, avg_cost, acc)))
if math.isnan(avg_cost):
sys.exit("got NaN loss, training failed.")
elif isinstance(event, fluid.EndStepEvent):
print("Step {0}, Epoch {1} Metrics {2}".format(
event.step, event.epoch, map(numpy.array, event.metrics)))
print(
("Step {0}, Epoch {1} Metrics {2}".format(
event.step, event.epoch,
list(map(numpy.array, event.metrics)))))
train_reader = paddle.batch(
paddle.reader.shuffle(
......@@ -116,7 +118,7 @@ def infer(use_cuda, inference_program, params_dirname=None):
results = inferencer.infer({'img': tensor_img})
print("infer results: ", results[0])
print(("infer results: ", results[0]))
def main(use_cuda):
......
python/paddle/fluid/tests/book/high-level-api/recognize_digits/test_recognize_digits_mlp.py
浏览文件 @ 559d3632
......@@ -11,7 +11,7 @@
# 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 argparse
import paddle.fluid as fluid
import paddle
......@@ -61,14 +61,14 @@ def train(use_cuda, train_program, params_dirname):
avg_cost, acc = trainer.test(
reader=test_reader, feed_order=['img', 'label'])
print("avg_cost: %s" % avg_cost)
print("acc : %s" % acc)
print(("avg_cost: %s" % avg_cost))
print(("acc : %s" % acc))
if acc > 0.2: # Smaller value to increase CI speed
trainer.save_params(params_dirname)
else:
print('BatchID {0}, Test Loss {1:0.2}, Acc {2:0.2}'.format(
event.epoch + 1, avg_cost, acc))
print(('BatchID {0}, Test Loss {1:0.2}, Acc {2:0.2}'.format(
event.epoch + 1, avg_cost, acc)))
if math.isnan(avg_cost):
sys.exit("got NaN loss, training failed.")
......@@ -96,7 +96,7 @@ def infer(use_cuda, inference_program, params_dirname=None):
results = inferencer.infer({'img': tensor_img})
print("infer results: ", results[0])
print(("infer results: ", results[0]))
def main(use_cuda):
......
python/paddle/fluid/tests/book/high-level-api/recommender_system/test_recommender_system_newapi.py
浏览文件 @ 559d3632
......@@ -180,14 +180,15 @@ def train(use_cuda, train_program, params_dirname):
# get avg cost
avg_cost = np.array(avg_cost_set).mean()
print("avg_cost: %s" % avg_cost)
print(("avg_cost: %s" % avg_cost))
if float(avg_cost) < 4: # Smaller value to increase CI speed
trainer.save_params(params_dirname)
trainer.stop()
else:
print('BatchID {0}, Test Loss {1:0.2}'.format(event.epoch + 1,
float(avg_cost)))
print(
('BatchID {0}, Test Loss {1:0.2}'.format(event.epoch + 1,
float(avg_cost))))
if math.isnan(float(avg_cost)):
sys.exit("got NaN loss, training failed.")
......@@ -239,7 +240,7 @@ def infer(use_cuda, inference_program, params_dirname):
},
return_numpy=False)
print("infer results: ", np.array(results[0]))
print(("infer results: ", np.array(results[0])))
def main(use_cuda):
......
python/paddle/fluid/tests/book/high-level-api/understand_sentiment/test_understand_sentiment_conv.py
浏览文件 @ 559d3632
......@@ -12,8 +12,6 @@
# See the License for the specific language governing permissions and
# limitations under the License.
from __future__ import print_function
import paddle
import paddle.fluid as fluid
from functools import partial
......@@ -84,21 +82,21 @@ def train(use_cuda, train_program, params_dirname):
avg_cost, acc = trainer.test(
reader=test_reader, feed_order=['words', 'label'])
print("avg_cost: %s" % avg_cost)
print("acc : %s" % acc)
print(("avg_cost: %s" % avg_cost))
print(("acc : %s" % acc))
if acc > 0.2: # Smaller value to increase CI speed
trainer.save_params(params_dirname)
trainer.stop()
else:
print('BatchID {0}, Test Loss {1:0.2}, Acc {2:0.2}'.format(
event.epoch + 1, avg_cost, acc))
print(('BatchID {0}, Test Loss {1:0.2}, Acc {2:0.2}'.format(
event.epoch + 1, avg_cost, acc)))
if math.isnan(avg_cost):
sys.exit("got NaN loss, training failed.")
elif isinstance(event, fluid.EndStepEvent):
print("Step {0}, Epoch {1} Metrics {2}".format(
event.step, event.epoch, map(np.array, event.metrics)))
print(("Step {0}, Epoch {1} Metrics {2}".format(
event.step, event.epoch, list(map(np.array, event.metrics)))))
if event.step == 1: # Run 2 iterations to speed CI
trainer.save_params(params_dirname)
trainer.stop()
......@@ -140,7 +138,7 @@ def infer(use_cuda, inference_program, params_dirname=None):
tensor_words = fluid.create_random_int_lodtensor(
recursive_seq_lens, base_shape, place, low=0, high=len(word_dict) - 1)
results = inferencer.infer({'words': tensor_words})
print("infer results: ", results)
print(("infer results: ", results))
def main(use_cuda):
......
python/paddle/fluid/tests/book/high-level-api/understand_sentiment/test_understand_sentiment_dynamic_rnn.py
浏览文件 @ 559d3632
......@@ -12,8 +12,6 @@
# See the License for the specific language governing permissions and
# limitations under the License.
from __future__ import print_function
import paddle
import paddle.fluid as fluid
from functools import partial
......@@ -99,21 +97,21 @@ def train(use_cuda, train_program, params_dirname):
avg_cost, acc = trainer.test(
reader=test_reader, feed_order=['words', 'label'])
print("avg_cost: %s" % avg_cost)
print("acc : %s" % acc)
print(("avg_cost: %s" % avg_cost))
print(("acc : %s" % acc))
if acc > 0.2: # Smaller value to increase CI speed
trainer.save_params(params_dirname)
trainer.stop()
else:
print('BatchID {0}, Test Loss {1:0.2}, Acc {2:0.2}'.format(
event.epoch + 1, avg_cost, acc))
print(('BatchID {0}, Test Loss {1:0.2}, Acc {2:0.2}'.format(
event.epoch + 1, avg_cost, acc)))
if math.isnan(avg_cost):
sys.exit("got NaN loss, training failed.")
elif isinstance(event, fluid.EndStepEvent):
print("Step {0}, Epoch {1} Metrics {2}".format(
event.step, event.epoch, map(np.array, event.metrics)))
print(("Step {0}, Epoch {1} Metrics {2}".format(
event.step, event.epoch, list(map(np.array, event.metrics)))))
if event.step == 1: # Run 2 iterations to speed CI
trainer.save_params(params_dirname)
trainer.stop()
......@@ -155,7 +153,7 @@ def infer(use_cuda, inference_program, params_dirname=None):
tensor_words = fluid.create_random_int_lodtensor(
recursive_seq_lens, base_shape, place, low=0, high=len(word_dict) - 1)
results = inferencer.infer({'words': tensor_words})
print("infer results: ", results)
print(("infer results: ", results))
def main(use_cuda):
......
python/paddle/fluid/tests/book/high-level-api/understand_sentiment/test_understand_sentiment_stacked_lstm.py
浏览文件 @ 559d3632
......@@ -12,8 +12,6 @@
# See the License for the specific language governing permissions and
# limitations under the License.
from __future__ import print_function
import paddle
import paddle.fluid as fluid
from functools import partial
......@@ -93,21 +91,21 @@ def train(use_cuda, train_program, params_dirname):
avg_cost, acc = trainer.test(
reader=test_reader, feed_order=['words', 'label'])
print("avg_cost: %s" % avg_cost)
print("acc : %s" % acc)
print(("avg_cost: %s" % avg_cost))
print(("acc : %s" % acc))
if acc > 0.2: # Smaller value to increase CI speed
trainer.save_params(params_dirname)
trainer.stop()
else:
print('BatchID {0}, Test Loss {1:0.2}, Acc {2:0.2}'.format(
event.epoch + 1, avg_cost, acc))
print(('BatchID {0}, Test Loss {1:0.2}, Acc {2:0.2}'.format(
event.epoch + 1, avg_cost, acc)))
if math.isnan(avg_cost):
sys.exit("got NaN loss, training failed.")
elif isinstance(event, fluid.EndStepEvent):
print("Step {0}, Epoch {1} Metrics {2}".format(
event.step, event.epoch, map(np.array, event.metrics)))
print(("Step {0}, Epoch {1} Metrics {2}".format(
event.step, event.epoch, list(map(np.array, event.metrics)))))
if event.step == 1: # Run 2 iterations to speed CI
trainer.save_params(params_dirname)
trainer.stop()
......@@ -150,7 +148,7 @@ def infer(use_cuda, inference_program, params_dirname=None):
tensor_words = fluid.create_random_int_lodtensor(
recursive_seq_lens, base_shape, place, low=0, high=len(word_dict) - 1)
results = inferencer.infer({'words': tensor_words})
print("infer results: ", results)
print(("infer results: ", results))
def main(use_cuda):
......
python/paddle/fluid/tests/book/high-level-api/word2vec/test_word2vec_new_api.py
浏览文件 @ 559d3632
......@@ -98,7 +98,7 @@ def train(use_cuda, train_program, params_dirname):
reader=test_reader,
feed_order=['firstw', 'secondw', 'thirdw', 'forthw', 'nextw'])
avg_cost = outs[0]
print("loss= ", avg_cost)
print(("loss= ", avg_cost))
if avg_cost < 10.0:
trainer.save_params(params_dirname)
......@@ -149,7 +149,7 @@ def infer(use_cuda, inference_program, params_dirname=None):
'forthw': fourth_word
},
return_numpy=False)
print(np.array(result[0]))
print((np.array(result[0])))
def main(use_cuda, is_sparse):
......
python/paddle/fluid/tests/book/notest_understand_sentiment.py
浏览文件 @ 559d3632
......@@ -11,7 +11,7 @@
# 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
from paddle.fluid.layers.device import get_places
import unittest
import paddle.fluid as fluid
......@@ -175,12 +175,12 @@ def train(word_dict,
def train_loop(main_program):
exe.run(fluid.default_startup_program())
for pass_id in xrange(PASS_NUM):
for pass_id in range(PASS_NUM):
for data in train_data():
cost_val, acc_val = exe.run(main_program,
feed=feeder.feed(data),
fetch_list=[cost, acc_out])
print("cost=" + str(cost_val) + " acc=" + str(acc_val))
print(("cost=" + str(cost_val) + " acc=" + str(acc_val)))
if cost_val < 0.4 and acc_val > 0.8:
if save_dirname is not None:
fluid.io.save_inference_model(save_dirname, ["words"],
......@@ -261,10 +261,10 @@ def infer(word_dict, use_cuda, save_dirname=None):
feed={feed_target_names[0]: tensor_words},
fetch_list=fetch_targets,
return_numpy=False)
print(results[0].recursive_sequence_lengths())
print((results[0].recursive_sequence_lengths()))
np_data = np.array(results[0])
print("Inference Shape: ", np_data.shape)
print("Inference results: ", np_data)
print(("Inference Shape: ", np_data.shape))
print(("Inference results: ", np_data))
def main(word_dict, net_method, use_cuda, parallel=False, save_dirname=None):
......
python/paddle/fluid/tests/book/test_fit_a_line.py
浏览文件 @ 559d3632
......@@ -114,7 +114,7 @@ def infer(use_cuda, save_dirname=None):
test_reader = paddle.batch(
paddle.dataset.uci_housing.test(), batch_size=batch_size)
test_data = test_reader().next()
test_data = next(test_reader())
test_feat = numpy.array(
[data[0] for data in test_data]).astype("float32")
test_label = numpy.array(
......@@ -124,9 +124,9 @@ def infer(use_cuda, save_dirname=None):
results = exe.run(inference_program,
feed={feed_target_names[0]: numpy.array(test_feat)},
fetch_list=fetch_targets)
print("infer shape: ", results[0].shape)
print("infer results: ", results[0])
print("ground truth: ", test_label)
print(("infer shape: ", results[0].shape))
print(("infer results: ", results[0]))
print(("ground truth: ", test_label))
def main(use_cuda, is_local=True):
......
python/paddle/fluid/tests/book/test_image_classification.py
浏览文件 @ 559d3632
......@@ -12,8 +12,6 @@
# See the License for the specific language governing permissions and
# limitations under the License.
from __future__ import print_function
import paddle
import paddle.fluid as fluid
import contextlib
......@@ -165,10 +163,10 @@ def train(net_type, use_cuda, save_dirname, is_local):
acc_value = numpy.array(acc_list).mean()
avg_loss_value = numpy.array(avg_loss_list).mean()
print(
print((
'PassID {0:1}, BatchID {1:04}, Test Loss {2:2.2}, Acc {3:2.2}'.
format(pass_id, batch_id + 1,
float(avg_loss_value), float(acc_value)))
float(avg_loss_value), float(acc_value))))
if acc_value > 0.01: # Low threshold for speeding up CI
fluid.io.save_inference_model(save_dirname, ["pixel"],
......@@ -241,7 +239,7 @@ def infer(use_cuda, save_dirname=None):
np.testing.assert_almost_equal(
results[0][i], transpiler_results[0][i], decimal=5)
print("infer results: ", results[0])
print(("infer results: ", results[0]))
fluid.io.save_inference_model(save_dirname, feed_target_names,
fetch_targets, exe,
......
python/paddle/fluid/tests/book/test_label_semantic_roles.py
浏览文件 @ 559d3632
......@@ -181,7 +181,7 @@ def train(use_cuda, save_dirname=None, is_local=True):
start_time = time.time()
batch_id = 0
for pass_id in xrange(PASS_NUM):
for pass_id in range(PASS_NUM):
for data in train_data():
cost = exe.run(main_program,
feed=feeder.feed(data),
......@@ -189,10 +189,10 @@ def train(use_cuda, save_dirname=None, is_local=True):
cost = cost[0]
if batch_id % 10 == 0:
print("avg_cost:" + str(cost))
print(("avg_cost:" + str(cost)))
if batch_id != 0:
print("second per batch: " + str((time.time(
) - start_time) / batch_id))
print(("second per batch: " + str(
(time.time() - start_time) / batch_id)))
# Set the threshold low to speed up the CI test
if float(cost) < 60.0:
if save_dirname is not None:
......@@ -333,9 +333,9 @@ def infer(use_cuda, save_dirname=None):
},
fetch_list=fetch_targets,
return_numpy=False)
print(results[0].recursive_sequence_lengths())
print((results[0].recursive_sequence_lengths()))
np_data = np.array(results[0])
print("Inference Shape: ", np_data.shape)
print(("Inference Shape: ", np_data.shape))
def main(use_cuda, is_local=True):
......
python/paddle/fluid/tests/book/test_machine_translation.py
浏览文件 @ 559d3632
......@@ -199,14 +199,14 @@ def train_main(use_cuda, is_sparse, is_local=True):
feeder = fluid.DataFeeder(feed_list, place)
batch_id = 0
for pass_id in xrange(1):
for pass_id in range(1):
for data in train_data():
outs = exe.run(main_program,
feed=feeder.feed(data),
fetch_list=[avg_cost])
avg_cost_val = np.array(outs[0])
print('pass_id=' + str(pass_id) + ' batch=' + str(batch_id) +
" avg_cost=" + str(avg_cost_val))
print(('pass_id=' + str(pass_id) + ' batch=' + str(batch_id) +
" avg_cost=" + str(avg_cost_val)))
if batch_id > 3:
break
batch_id += 1
......@@ -273,7 +273,7 @@ def decode_main(use_cuda, is_sparse):
feeder = fluid.DataFeeder(feed_list, place)
for data in train_data():
feed_dict = feeder.feed(map(lambda x: [x[0]], data))
feed_dict = feeder.feed([[x[0]] for x in data])
feed_dict['init_ids'] = init_ids
feed_dict['init_scores'] = init_scores
......@@ -282,7 +282,7 @@ def decode_main(use_cuda, is_sparse):
feed=feed_dict,
fetch_list=[translation_ids, translation_scores],
return_numpy=False)
print result_ids.recursive_sequence_lengths()
print((result_ids.recursive_sequence_lengths()))
break
......
python/paddle/fluid/tests/book/test_recognize_digits.py
浏览文件 @ 559d3632
......@@ -11,7 +11,6 @@
# 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 paddle.fluid.core as core
import math
......@@ -143,10 +142,10 @@ def train(nn_type,
params_filename=params_filename)
return
else:
print(
print((
'PassID {0:1}, BatchID {1:04}, Test Loss {2:2.2}, Acc {3:2.2}'.
format(pass_id, batch_id + 1,
float(avg_loss_val), float(acc_val)))
float(avg_loss_val), float(acc_val))))
if math.isnan(float(avg_loss_val)):
sys.exit("got NaN loss, training failed.")
raise AssertionError("Loss of recognize digits is too large")
......@@ -207,7 +206,7 @@ def infer(use_cuda,
results = exe.run(inference_program,
feed={feed_target_names[0]: tensor_img},
fetch_list=fetch_targets)
print("infer results: ", results[0])
print(("infer results: ", results[0]))
def main(use_cuda, parallel, nn_type, combine):
......
python/paddle/fluid/tests/book/test_recommender_system.py
浏览文件 @ 559d3632
......@@ -304,7 +304,7 @@ def infer(use_cuda, save_dirname=None):
},
fetch_list=fetch_targets,
return_numpy=False)
print("inferred score: ", np.array(results[0]))
print(("inferred score: ", np.array(results[0])))
def main(use_cuda):
......
python/paddle/fluid/tests/book/test_rnn_encoder_decoder.py
浏览文件 @ 559d3632
......@@ -175,15 +175,15 @@ def train(use_cuda, save_dirname=None):
feeder = fluid.DataFeeder(feed_list, place)
batch_id = 0
for pass_id in xrange(2):
for pass_id in range(2):
for data in train_data():
outs = exe.run(framework.default_main_program(),
feed=feeder.feed(data),
fetch_list=[avg_cost])
avg_cost_val = np.array(outs[0])
print('pass_id=' + str(pass_id) + ' batch=' + str(batch_id) +
" avg_cost=" + str(avg_cost_val))
print(('pass_id=' + str(pass_id) + ' batch=' + str(batch_id) +
" avg_cost=" + str(avg_cost_val)))
if math.isnan(float(avg_cost_val[0])):
sys.exit("got NaN loss, training failed.")
if batch_id > 3:
......@@ -241,10 +241,10 @@ def infer(use_cuda, save_dirname=None):
},
fetch_list=fetch_targets,
return_numpy=False)
print(results[0].recursive_sequence_lengths())
print((results[0].recursive_sequence_lengths()))
np_data = np.array(results[0])
print("Inference shape: ", np_data.shape)
print("Inference results: ", np_data)
print(("Inference shape: ", np_data.shape))
print(("Inference results: ", np_data))
def main(use_cuda):
......
python/paddle/fluid/tests/book/test_word2vec.py
浏览文件 @ 559d3632
......@@ -85,9 +85,11 @@ def train(use_cuda, is_sparse, is_parallel, save_dirname, is_local=True):
pd = fluid.layers.ParallelDo(places)
with pd.do():
avg_cost, predict_word = __network__(
map(pd.read_input, [
first_word, second_word, third_word, forth_word, next_word
]))
list(
map(pd.read_input, [
first_word, second_word, third_word, forth_word,
next_word
])))
pd.write_output(avg_cost)
avg_cost = fluid.layers.mean(pd())
......@@ -202,9 +204,9 @@ def infer(use_cuda, save_dirname=None):
},
fetch_list=fetch_targets,
return_numpy=False)
print(results[0].recursive_sequence_lengths())
print((results[0].recursive_sequence_lengths()))
np_data = np.array(results[0])
print("Inference Shape: ", np_data.shape)
print(("Inference Shape: ", np_data.shape))
def main(use_cuda, is_sparse, is_parallel):
......
python/paddle/fluid/tests/book_memory_optimization/test_memopt_fit_a_line.py
浏览文件 @ 559d3632
......@@ -78,7 +78,7 @@ for pass_id in range(PASS_NUM):
if avg_loss_value[0] < 10.0:
exit(0) # if avg cost less than 10.0, we think our code is good.
print avg_loss_value[0]
print((avg_loss_value[0]))
if math.isnan(float(avg_loss_value)):
sys.exit("got NaN loss, training failed.")
exit(1)
python/paddle/fluid/tests/book_memory_optimization/test_memopt_image_classification_train.py
浏览文件 @ 559d3632
......@@ -12,8 +12,6 @@
# See the License for the specific language governing permissions and
# limitations under the License.
from __future__ import print_function
import sys
import paddle
......@@ -157,8 +155,8 @@ for pass_id in range(PASS_NUM):
fetch_list=[avg_cost, batch_acc, batch_size])
accuracy.add(value=acc, weight=weight)
pass_acc = accuracy.eval()
print("loss:" + str(loss) + " acc:" + str(acc) + " pass_acc:" + str(
pass_acc))
print(("loss:" + str(loss) + " acc:" + str(acc) + " pass_acc:" +
str(pass_acc)))
# this model is slow, so if we can train two mini batch, we think it works properly.
if i > 0:
exit(0)
......
python/paddle/fluid/tests/book_memory_optimization/test_memopt_machine_translation.py
浏览文件 @ 559d3632
......@@ -118,14 +118,14 @@ def main():
feeder = fluid.DataFeeder(feed_list, place)
batch_id = 0
for pass_id in xrange(10):
for pass_id in range(10):
for data in train_data():
outs = exe.run(fluid.default_main_program(),
feed=feeder.feed(data),
fetch_list=[avg_cost])
avg_cost_val = np.array(outs[0])
print('pass_id=' + str(pass_id) + ' batch=' + str(batch_id) +
" avg_cost=" + str(avg_cost_val))
print(('pass_id=' + str(pass_id) + ' batch=' + str(batch_id) +
" avg_cost=" + str(avg_cost_val)))
if batch_id > 2:
exit(0)
if math.isnan(float(avg_cost_val)):
......
python/paddle/fluid/tests/demo/fc_gan.py
浏览文件 @ 559d3632
......@@ -137,7 +137,7 @@ def main():
generated_img = exe.run(g_program,
feed={'noise': n},
fetch_list={g_img})[0]
real_data = numpy.array(map(lambda x: x[0], data)).astype('float32')
real_data = numpy.array([x[0] for x in data]).astype('float32')
real_data = real_data.reshape(num_true, 784)
total_data = numpy.concatenate([real_data, generated_img])
total_label = numpy.concatenate([
......@@ -150,7 +150,7 @@ def main():
feed={'img': total_data,
'label': total_label},
fetch_list={d_loss})[0]
for _ in xrange(NUM_TRAIN_TIMES_OF_DG):
for _ in range(NUM_TRAIN_TIMES_OF_DG):
n = numpy.random.uniform(
low=-1.0, high=1.0,
size=[2 * num_true * NOISE_SIZE]).astype('float32').reshape(
......@@ -158,8 +158,8 @@ def main():
dg_loss_np = exe.run(dg_program,
feed={'noise': n},
fetch_list={dg_loss})[0]
print("Pass ID={0}, Batch ID={1}, D-Loss={2}, DG-Loss={3}".format(
pass_id, batch_id, d_loss_np, dg_loss_np))
print(("Pass ID={0}, Batch ID={1}, D-Loss={2}, DG-Loss={3}".format(
pass_id, batch_id, d_loss_np, dg_loss_np)))
# generate image each batch
fig = plot(generated_img)
plt.savefig(
......
python/paddle/fluid/tests/demo/file_reader/convert_data_to_recordio.py
浏览文件 @ 559d3632
......@@ -36,7 +36,7 @@ if len(sys.argv) == 1:
else:
word_dict = load_vocab(sys.argv[1])
word_dict["<unk>"] = len(word_dict)
print "Dict dim = ", len(word_dict)
print("Dict dim = ", len(word_dict))
# input text data
data = fluid.layers.data(name="words", shape=[1], dtype="int64", lod_level=1)
......
python/paddle/fluid/tests/no_test_concurrency.py
浏览文件 @ 559d3632
......@@ -194,7 +194,7 @@ class TestRoutineOp(unittest.TestCase):
quit_ch = fluid.make_channel(dtype=core.VarDesc.VarType.LOD_TENSOR)
with fluid.Go():
for i in xrange(10):
for i in range(10):
fluid.channel_recv(ch1, result)
Print(result)
......
python/paddle/fluid/tests/test_detection.py
浏览文件 @ 559d3632
......@@ -12,7 +12,6 @@
# See the License for the specific language governing permissions and
# limitations under the License.
from __future__ import print_function
import paddle.fluid as fluid
import paddle.fluid.layers as layers
from paddle.fluid.framework import Program, program_guard
......@@ -47,7 +46,7 @@ class TestDetection(unittest.TestCase):
scores=scores, loc=loc, prior_box=pb, prior_box_var=pbv)
self.assertIsNotNone(out)
self.assertEqual(out.shape[-1], 6)
print(str(program))
print((str(program)))
def test_detection_api(self):
program = Program()
......@@ -82,7 +81,7 @@ class TestDetection(unittest.TestCase):
self.assertIsNotNone(trg)
self.assertIsNotNone(trg_weight)
print(str(program))
print((str(program)))
def test_ssd_loss(self):
program = Program()
......@@ -106,7 +105,7 @@ class TestDetection(unittest.TestCase):
loss = layers.ssd_loss(loc, scores, gt_box, gt_label, pb, pbv)
self.assertIsNotNone(loss)
self.assertEqual(loss.shape[-1], 1)
print(str(program))
print((str(program)))
class TestPriorBox(unittest.TestCase):
......@@ -197,7 +196,7 @@ class TestDetectionMAP(unittest.TestCase):
map_out = layers.detection_map(detect_res, label, 21)
self.assertIsNotNone(map_out)
self.assertEqual(map_out.shape, (1, ))
print(str(program))
print((str(program)))
if __name__ == '__main__':
......
python/paddle/fluid/tests/test_error_clip.py
浏览文件 @ 559d3632
......@@ -12,7 +12,6 @@
# See the License for the specific language governing permissions and
# limitations under the License.
from __future__ import print_function
import numpy as np
import paddle
import paddle.fluid as fluid
......
python/paddle/fluid/tests/test_if_else_op.py
浏览文件 @ 559d3632
......@@ -76,15 +76,15 @@ class TestMNISTIfElseOp(unittest.TestCase):
PASS_NUM = 100
for pass_id in range(PASS_NUM):
for data in train_reader():
x_data = np.array(map(lambda x: x[0], data)).astype("float32")
y_data = np.array(map(lambda x: x[1], data)).astype("int64")
x_data = np.array([x[0] for x in data]).astype("float32")
y_data = np.array([x[1] for x in data]).astype("int64")
y_data = np.expand_dims(y_data, axis=1)
outs = exe.run(prog,
feed={'x': x_data,
'y': y_data},
fetch_list=[avg_loss])
print outs[0]
print((outs[0]))
if outs[0] < 1.0:
return
self.assertFalse(True)
......@@ -131,15 +131,15 @@ class TestMNISTIfElseOp(unittest.TestCase):
PASS_NUM = 100
for pass_id in range(PASS_NUM):
for data in train_reader():
x_data = np.array(map(lambda x: x[0], data)).astype("float32")
y_data = np.array(map(lambda x: x[1], data)).astype("int64")
x_data = np.array([x[0] for x in data]).astype("float32")
y_data = np.array([x[1] for x in data]).astype("int64")
y_data = y_data.reshape((y_data.shape[0], 1))
outs = exe.run(prog,
feed={'x': x_data,
'y': y_data},
fetch_list=[avg_loss])
print outs[0]
print((outs[0]))
if outs[0] < 1.0:
return
self.assertFalse(True)
......
python/paddle/fluid/tests/unittests/benchmark.py
浏览文件 @ 559d3632
......@@ -20,7 +20,7 @@ import itertools
import paddle.fluid as fluid
import paddle.fluid.core as core
from paddle.fluid.op import Operator
from op_test import OpTest
from .op_test import OpTest
class BenchmarkSuite(OpTest):
......@@ -40,8 +40,7 @@ class BenchmarkSuite(OpTest):
expect_t = np.array(item_cpu_out)
actual = item_gpu_out
actual_t = np.array(item_gpu_out)
var_name = variable if isinstance(variable,
basestring) else variable.name
var_name = variable if isinstance(variable, str) else variable.name
self.assertTrue(
np.allclose(
actual_t, expect_t, atol=atol),
......@@ -53,7 +52,7 @@ class BenchmarkSuite(OpTest):
def _get_input_names(self):
inputs = []
for name, value in self.inputs.iteritems():
for name, value in list(self.inputs.items()):
if isinstance(value, list):
inputs.extend([sub_name for sub_name, _ in value])
inputs.append(name)
......@@ -61,7 +60,7 @@ class BenchmarkSuite(OpTest):
def _get_output_names(self):
outputs = []
for var_name, var in self.outputs.iteritems():
for var_name, var in list(self.outputs.items()):
if isinstance(var, list):
for sub_var_name, sub_var in var:
outputs.append(sub_var_name)
......@@ -89,8 +88,8 @@ class BenchmarkSuite(OpTest):
for place in places:
elapses.append(self.timeit_output_with_place(place, iters))
for place, elapse in zip(places, elapses):
print("One pass of ({2}_op) at {0} cost {1}".format(
str(place), elapse, self.op_type))
print(("One pass of ({2}_op) at {0} cost {1}".format(
str(place), elapse, self.op_type)))
def timeit_grad_with_place(self, place, iters=100):
inputs_to_check = self._get_input_names()
......@@ -109,5 +108,5 @@ class BenchmarkSuite(OpTest):
for place in places:
elapses.append(self.timeit_grad_with_place(place, iters))
for place, elapse in zip(places, elapses):
print("One pass of ({2}_grad_op) at {0} cost {1}".format(
str(place), elapse, self.op_type))
print(("One pass of ({2}_grad_op) at {0} cost {1}".format(
str(place), elapse, self.op_type)))
python/paddle/fluid/tests/unittests/benchmark_sum_op.py
浏览文件 @ 559d3632
......@@ -16,8 +16,8 @@ import unittest
import numpy as np
import paddle.fluid as fluid
from benchmark import BenchmarkSuite
from op_test import OpTest
from .benchmark import BenchmarkSuite
from .op_test import OpTest
# This is a demo op test case for operator benchmarking and high resolution number stability alignment.
......
python/paddle/fluid/tests/unittests/parallel_executor_test_base.py
浏览文件 @ 559d3632
......@@ -91,7 +91,7 @@ class TestParallelExecutorBase(unittest.TestCase):
first_loss, = run_executor(
exe=exe, feed=feed_dict, fetch_list=[loss.name])
for i in xrange(iter):
for i in range(iter):
run_executor(exe=exe, feed=feed_dict, fetch_list=[])
last_loss, = run_executor(
......@@ -99,8 +99,8 @@ class TestParallelExecutorBase(unittest.TestCase):
end = time.time()
if batch_size is not None:
print "%.4f Instance per second" % (
(batch_size * iter + 2) / (end - begin))
print(("%.4f Instance per second" % (
(batch_size * iter + 2) / (end - begin))))
avg_last_loss_val = np.array(last_loss).mean()
avg_first_loss_val = np.array(first_loss).mean()
......@@ -108,6 +108,6 @@ class TestParallelExecutorBase(unittest.TestCase):
float(avg_first_loss_val)):
sys.exit("got NaN loss, training failed.")
print first_loss, last_loss
print((first_loss, last_loss))
# self.assertGreater(first_loss[0], last_loss[0])
return first_loss, last_loss
python/paddle/fluid/tests/unittests/testsuite.py
浏览文件 @ 559d3632
......@@ -142,7 +142,7 @@ def append_input_output(block, op_proto, np_list, is_input, dtype):
def append_loss_ops(block, output_names):
mean_inputs = map(block.var, output_names)
mean_inputs = list(map(block.var, output_names))
# for item in mean_inputs:
# print(item)
# print("Item", item.dtype)
......
python/paddle/fluid/tests/unittests/transformer_model.py
浏览文件 @ 559d3632
......@@ -118,8 +118,9 @@ def multi_head_attention(queries,
# FIXME(guosheng): Decouple the program desc with batch_size.
return layers.reshape(
x=trans_x,
shape=map(int,
[batch_size, -1, trans_x.shape[2] * trans_x.shape[3]]))
shape=list(
map(int, [batch_size, -1, trans_x.shape[2] * trans_x.shape[3]
])))
def scaled_dot_product_attention(q, k, v, attn_bias, d_model, dropout_rate):
"""
......
python/paddle/fluid/trainer.py
浏览文件 @ 559d3632
......@@ -18,16 +18,15 @@ import errno
import shutil
import time
import core
import data_feeder
import executor
import framework
import io
from . import core
from . import data_feeder
from . import executor
from . import framework
from . import io
# optimizer is same as the parameter of Trainer.__init__. Rename it to opt_module
import optimizer as opt_module
import parallel_executor
from transpiler import distribute_transpiler
from . import optimizer as opt_module
from . import parallel_executor
from .transpiler import distribute_transpiler
__all__ = [
'Trainer', 'BeginEpochEvent', 'EndEpochEvent', 'BeginStepEvent',
......@@ -73,7 +72,7 @@ class BeginStepEvent(object):
self.step = step_id
self.fetch_metrics = True
"""
If fetch_metrics is true, the metrics will be fetched at the
If fetch_metrics is true, the metrics will be fetched at the
EndStepEvent. Default is True.
"""
......@@ -614,11 +613,12 @@ def build_feed_var_list(program, feed_order):
if not isinstance(feed_order, dict):
raise TypeError(
"The 'feed_order' should be either None, list or dict.")
if not sorted(feed_order.values()) == range(len(feed_order)):
if not sorted(feed_order.values()) == list(range(len(feed_order))):
raise ValueError(
"The values of 'feed_order' should be a permutation of [0, len(feed_order))"
)
sorted_pair_list = sorted(feed_order.items(), key=lambda item: item[1])
sorted_pair_list = sorted(
list(feed_order.items()), key=lambda item: item[1])
feed_var_list = [
program.global_block().var(pair[0]) for pair in sorted_pair_list
]
......@@ -644,14 +644,14 @@ def save_checkpoint(executor,
pserver_endpoints=None):
"""
This function filters out all checkpoint variables from the give
main_program and then saves these variables to the `checkpoint_dir`
main_program and then saves these variables to the `checkpoint_dir`
directory.
In the training precess, we generally save a checkpoint in each
iteration. So there might be a lot of checkpoints in the
`checkpoint_dir`. To avoid them taking too much disk space, the
`max_num_checkpoints` are introduced to limit the total number of
checkpoints. If the number of existing checkpints is greater than
iteration. So there might be a lot of checkpoints in the
`checkpoint_dir`. To avoid them taking too much disk space, the
`max_num_checkpoints` are introduced to limit the total number of
checkpoints. If the number of existing checkpints is greater than
the `max_num_checkpoints`, oldest ones will be scroll deleted.
A variable is a checkpoint variable and will be saved if it meets
......@@ -663,21 +663,21 @@ def save_checkpoint(executor,
Args:
executor(Executor): The executor to run for save checkpoint.
checkpoint_dir(str): The folder where to save checkpoints.
trainer_id(int): currect trainer id, if id is equal to 0, the trainer
trainer_id(int): currect trainer id, if id is equal to 0, the trainer
is chief.
trainer_args(dict|None): Current training arguments. Such as 'epoch_id'
trainer_args(dict|None): Current training arguments. Such as 'epoch_id'
and 'step_id'.
Defaut: None
main_program(Program): The program whose checkpoint variables will
be saved.
max_num_checkpoints(int): The max number of total number of existing
max_num_checkpoints(int): The max number of total number of existing
checkpoints.
Default: 3
lookup_table(string|None): the lookup table name, when use distribute
lookup table, we can get lookup table name by DistributeTranspiler.
table_name
pserver_endpoints(list|None): the parameter server ip:port list.
when use distribute lookup table, we can get pserver_endpoints by
table_name
pserver_endpoints(list|None): the parameter server ip:port list.
when use distribute lookup table, we can get pserver_endpoints by
distribute arguments.
Returns:
......@@ -747,8 +747,8 @@ def load_checkpoint(executor,
`checkpoint_dir` directory.
In the training precess, we generally save a checkpoint in each
iteration. So there are more than one checkpoint in the
`checkpoint_dir` (each checkpoint has its own sub folder), use
iteration. So there are more than one checkpoint in the
`checkpoint_dir` (each checkpoint has its own sub folder), use
`serial` to specify which serial of checkpoint you would like to
load.
......@@ -819,9 +819,9 @@ def load_checkpoint(executor,
def clean_checkpoint(checkpoint_dir, delete_dir=False):
"""
clean the checkpoint dir, when the train exits normally,
clean the checkpoint dir, when the train exits normally,
the trainer will call clean_checkpoint to delete checkpoint directory saved before.
delete_dir only works when the directory is empty, otherwise, OSError is raised.
delete_dir only works when the directory is empty, otherwise, OSError is raised.
: param checkpoint_dir
: param delete_dir
......@@ -889,7 +889,7 @@ def _load_persist_vars_without_grad(executor,
def _load_lookup_table_vars(executor, dirname, program, pserver_id, table_name):
"""
The parameter server will load lookup table's local file in
The parameter server will load lookup table's local file in
selectedrows variable.
Args:
......@@ -940,7 +940,7 @@ def _load_lookup_table_vars(executor, dirname, program, pserver_id, table_name):
def _save_persist_vars_without_grad(executor, dirname, program):
"""
This function filters out all checkpoint variables from the give
program and then save these variables to a sub-folder '__model__' of
program and then save these variables to a sub-folder '__model__' of
the given directory.
A variable is a checkpoint variable if it meets all following
......@@ -969,7 +969,7 @@ def _save_persist_vars_without_grad(executor, dirname, program):
# In this example, `_save_persist_vars_without_grad` function
# will first filters out all checkpoint variables in the default
# main program, and then saves these variables to the folder
# main program, and then saves these variables to the folder
# "./my_paddle_model/__model__".
"""
cur_dir = _get_model_dir(dirname)
......@@ -988,7 +988,7 @@ def _save_pserver_vars_by_notify(executor, dirname, lookup_table,
"""
This function will send checkpoint notify message from Trainer 0
to all the pservers.
The checkpoint notify message contains lookup table name,
The checkpoint notify message contains lookup table name,
the absolute path on pserver to save lookup_table.
Args:
......@@ -996,13 +996,13 @@ def _save_pserver_vars_by_notify(executor, dirname, lookup_table,
dirname(str): The folder where to save checkpoints.
lookup_table(string): the lookup table name, when use distribute
lookup table, we can get lookup table name by DistributeTranspiler.
table_name
ps_endpoint_list(list): the parameter server ip:port list.
when use distribute lookup table, we can get ps_endpoint_list by
table_name
ps_endpoint_list(list): the parameter server ip:port list.
when use distribute lookup table, we can get ps_endpoint_list by
distribute arguments.
Return:
None
Examples:
.. code-block:: python
......@@ -1013,7 +1013,7 @@ def _save_pserver_vars_by_notify(executor, dirname, lookup_table,
ps_endpoints = ["127.0.0.1:6000","127.0.0.1:6001"]
_save_pserver_vars_by_notify(executor=exe,
dirname=param_path, lookup_table=table_name,
dirname=param_path, lookup_table=table_name,
ps_endpoint_list=ps_endpoints)
"""
cur_dir = _get_lookuptable_dir(dirname)
......@@ -1036,7 +1036,7 @@ def _save_trainer_args(dirname, trainer_id, trainer_args):
cur_dir = _get_trainer_dir(dirname, trainer_id)
for name, value in trainer_args.iteritems():
for name, value in list(trainer_args.items()):
args_file = os.path.join(cur_dir, name)
with open(args_file, 'w') as f:
f.write(str(value))
......@@ -1045,7 +1045,7 @@ def _save_trainer_args(dirname, trainer_id, trainer_args):
def _load_trainer_args(checkpoint_dir, serial, trainer_id, trainer_args):
"""
trainer will load some args from it's independent directory,
trainer will load some args from it's independent directory,
such as epoch_id and step_id.
Args:
......@@ -1168,10 +1168,10 @@ def _scroll_delete(dirname, max_num_checkpoints=3):
serial_num = _get_dir_serial(serial)
serial_map[serial_num] = serial
if len(serial_map.keys()) <= max_num_checkpoints:
if len(list(serial_map.keys())) <= max_num_checkpoints:
return
serials = serial_map.keys()
serials = list(serial_map.keys())
serials.sort(reverse=True)
serials = serials[max_num_checkpoints:]
for serial in serials:
......
python/paddle/fluid/transpiler/__init__.py
浏览文件 @ 559d3632
......@@ -12,10 +12,10 @@
# See the License for the specific language governing permissions and
# limitations under the License.
from distribute_transpiler import DistributeTranspiler, DistributeTranspilerConfig
from inference_transpiler import InferenceTranspiler
from memory_optimization_transpiler import memory_optimize, release_memory
from ps_dispatcher import HashName, RoundRobin
from .distribute_transpiler import DistributeTranspiler, DistributeTranspilerConfig
from .inference_transpiler import InferenceTranspiler
from .memory_optimization_transpiler import memory_optimize, release_memory
from .ps_dispatcher import HashName, RoundRobin
__all__ = [
"DistributeTranspiler", "InferenceTranspiler", "memory_optimize",
......
python/paddle/fluid/transpiler/details/__init__.py
浏览文件 @ 559d3632
......@@ -12,5 +12,5 @@
# See the License for the specific language governing permissions and
# limitations under the License.
from program_utils import *
from ufind import *
from .program_utils import *
from .ufind import *
python/paddle/fluid/transpiler/details/program_utils.py
浏览文件 @ 559d3632
......@@ -17,8 +17,8 @@ def delete_ops(block, ops):
try:
start = list(block.ops).index(ops[0])
end = list(block.ops).index(ops[-1])
[block._remove_op(start) for _ in xrange(end - start + 1)]
except Exception, e:
[block._remove_op(start) for _ in range(end - start + 1)]
except Exception as e:
raise e
block.program._sync_with_cpp()
......
python/paddle/fluid/transpiler/distribute_transpiler.py
浏览文件 @ 559d3632
......@@ -28,18 +28,17 @@ Steps to transpile pserver:
5. add listen_and_serv op
"""
from __future__ import print_function
import math
import random
import numpy as np
from ps_dispatcher import RoundRobin, HashName, PSDispatcher
from .ps_dispatcher import RoundRobin, HashName, PSDispatcher
from .. import core, framework
from ..framework import Program, default_main_program, \
default_startup_program, Block, \
Parameter, grad_var_name
from details import *
from .details import *
from functools import reduce
LOOKUP_TABLE_TYPE = "lookup_table"
LOOKUP_TABLE_GRAD_TYPE = "lookup_table_grad"
......@@ -102,7 +101,7 @@ def slice_variable(var_list, slice_count, min_block_size):
block_size += dim1 - remains
# update split_count after aligning
split_count = int(math.ceil(var_numel / float(block_size)))
for block_id in xrange(split_count):
for block_id in range(split_count):
curr_block_size = min(block_size, var_numel - (
(block_id) * block_size))
block = VarBlock(var.name, block_id, curr_block_size)
......@@ -117,7 +116,7 @@ class DistributeTranspilerConfig(object):
try to choose the best method to balance loads for pservers.
min_block_size (int): Minimum splitted element number in block.
According:https://github.com/PaddlePaddle/Paddle/issues/8638#issuecomment-369912156
We can use bandwidth effiently when data size is larger than 2MB.If you
We can use bandwidth effiently when data size is larger than 2MB.If you
want to change it, please be sure you see the slice_variable function.
"""
......@@ -218,7 +217,7 @@ class DistributeTranspiler(object):
# fc_w@GRAD_trainer_0, fc_w@GRAD_trainer_1 --> pserver1
# fc_b@GRAD_trainer_0, fc_b@GRAD_trainer_1 --> pserver2
# shuffle the map will avoid the uneven distribution above
grad_var_mapping_items = self.grad_var_mapping.items()
grad_var_mapping_items = list(self.grad_var_mapping.items())
if not self.config.slice_var_up:
random.seed(self.trainer_num)
random.shuffle(grad_var_mapping_items)
......@@ -278,7 +277,7 @@ class DistributeTranspiler(object):
self.param_grad_ep_mapping[ep]["grads"].append(send_vars[i])
# step4: Concat the parameters splits together after recv.
for varname, splited_var in self.param_var_mapping.iteritems():
for varname, splited_var in list(self.param_var_mapping.items()):
eps = []
for var in splited_var:
index = [v.name for v in recv_vars].index(var.name)
......@@ -302,7 +301,7 @@ class DistributeTranspiler(object):
RPC_OP_ROLE_ATTR_NAME: RPC_OP_ROLE_ATTR_VALUE
})
for varname, splited_var in self.param_var_mapping.iteritems():
for varname, splited_var in list(self.param_var_mapping.items()):
if len(splited_var) <= 1:
continue
orig_param = program.global_block().vars[varname]
......@@ -371,7 +370,7 @@ class DistributeTranspiler(object):
dtype=v.dtype,
shape=v.shape)
if self.sync_mode and self.trainer_num > 1:
for trainer_id in xrange(self.trainer_num):
for trainer_id in range(self.trainer_num):
var = pserver_program.global_block().create_var(
name="%s.trainer_%d" % (orig_var_name, trainer_id),
persistable=False,
......@@ -461,7 +460,7 @@ class DistributeTranspiler(object):
per_opt_block = pserver_program.create_block(pre_block_idx)
optimize_blocks.append(per_opt_block)
# append grad merging ops before clip and weight decay
# cases may like:
# cases may like:
# L2Decay op -> clip op -> optimize
for _, op in enumerate(self.optimize_ops):
# find the origin @GRAD var before clipping
......@@ -556,7 +555,7 @@ class DistributeTranspiler(object):
# 1. create vars in pserver program to startup program
pserver_vars = pserver_program.global_block().vars
created_var_map = dict()
for _, var in pserver_vars.iteritems():
for _, var in list(pserver_vars.items()):
tmpvar = s_prog.global_block()._clone_variable(var)
created_var_map[var.name] = tmpvar
......@@ -989,11 +988,11 @@ class DistributeTranspiler(object):
var_mapping = dict()
for block_str in block_list:
varname, offset, size = block_str.split(":")
if not block_map.has_key(varname):
if varname not in block_map:
block_map[varname] = []
block_map[varname].append((long(offset), long(size)))
block_map[varname].append((int(offset), int(size)))
for varname, splited in block_map.iteritems():
for varname, splited in list(block_map.items()):
orig_var = program.global_block().var(varname)
if len(splited) == 1:
if self.sync_mode and add_trainer_suffix:
......@@ -1156,7 +1155,7 @@ class DistributeTranspiler(object):
grad_to_block_id.append(merged_var.name + ":" + str(optimize_block.idx))
if self.sync_mode and self.trainer_num > 1:
vars2merge = []
for i in xrange(self.trainer_num):
for i in range(self.trainer_num):
per_trainer_name = "%s.trainer_%d" % \
(merged_var_name, i)
vars2merge.append(pserver_block.vars[per_trainer_name])
......@@ -1204,7 +1203,7 @@ class DistributeTranspiler(object):
# learning rate variable has already be created by non-optimize op,
# don't create it once again.
lr_varname = opt_op.input(key)[0]
if pserver_block.vars.has_key(lr_varname):
if lr_varname in pserver_block.vars:
new_inputs[key] = pserver_block.vars[opt_op.input(key)[0]]
else:
origin_var = origin_program.global_block().vars[lr_varname]
......@@ -1244,7 +1243,7 @@ class DistributeTranspiler(object):
def _is_splited_grad_var(self, var, var_dict):
grad_block = None
for _, g in var_dict.iteritems():
for _, g in list(var_dict.items()):
if self._orig_varname(g.name) == self._orig_varname(var.name):
if g.name.find(".trainer_") == -1:
grad_block = g
......@@ -1254,7 +1253,7 @@ class DistributeTranspiler(object):
def _clone_lr_op(self, program, block, op):
inputs = self._get_input_map_from_op(
self.origin_program.global_block().vars, op)
for key, varlist in inputs.iteritems():
for key, varlist in list(inputs.items()):
if not isinstance(varlist, list):
varlist = [varlist]
for var in varlist:
......@@ -1263,7 +1262,7 @@ class DistributeTranspiler(object):
outputs = self._get_output_map_from_op(
self.origin_program.global_block().vars, op)
for key, varlist in outputs.iteritems():
for key, varlist in list(outputs.items()):
if not isinstance(varlist, list):
varlist = [varlist]
for var in varlist:
......@@ -1278,7 +1277,7 @@ class DistributeTranspiler(object):
# Append the ops for parameters that do not need to be optimized/updated
inputs = self._get_input_map_from_op(
self.origin_program.global_block().vars, opt_op)
for key, varlist in inputs.iteritems():
for key, varlist in list(inputs.items()):
if not isinstance(varlist, list):
varlist = [varlist]
for var in varlist:
......@@ -1288,7 +1287,7 @@ class DistributeTranspiler(object):
var, program.global_block().vars)
if grad_block:
inputs[key] = grad_block
elif not program.global_block().vars.has_key(var.name):
elif var.name not in program.global_block().vars:
program.global_block().create_var(
name=var.name,
persistable=var.persistable,
......@@ -1297,7 +1296,7 @@ class DistributeTranspiler(object):
outputs = self._get_output_map_from_op(
self.origin_program.global_block().vars, opt_op)
for key, varlist in outputs.iteritems():
for key, varlist in list(outputs.items()):
if not isinstance(varlist, list):
varlist = [varlist]
for var in varlist:
......@@ -1305,7 +1304,7 @@ class DistributeTranspiler(object):
var, program.global_block().vars)
if grad_block:
outputs[key] = grad_block
elif not program.global_block().vars.has_key(var.name):
elif var.name not in program.global_block().vars:
program.global_block()._clone_variable(var)
return optimize_block.append_op(
......@@ -1326,8 +1325,8 @@ class DistributeTranspiler(object):
def _create_ufind(self, optimize_ops):
# Create a unit find data struct by optimize ops
ufind = UnionFind(optimize_ops)
for i in xrange(len(optimize_ops)):
for j in xrange(i, len(optimize_ops)):
for i in range(len(optimize_ops)):
for j in range(i, len(optimize_ops)):
op1 = optimize_ops[i]
op2 = optimize_ops[j]
if self._is_op_connected(op1, op2):
......
python/paddle/fluid/transpiler/inference_transpiler.py
浏览文件 @ 559d3632
......@@ -305,6 +305,6 @@ class InferenceTranspiler(object):
args += current_op.output_arg_names
args = list(set(args)) # unique the input and output arguments
for var in self.block.vars.keys():
for var in list(self.block.vars.keys()):
if var not in args:
self.block._remove_var(var)
python/paddle/fluid/transpiler/memory_optimization_transpiler.py
浏览文件 @ 559d3632
......@@ -16,6 +16,7 @@ from collections import defaultdict
from .. import core
from ..framework import Program, default_main_program, Parameter
from ..backward import _rename_arg_
from functools import reduce
dtype_to_size = {
core.VarDesc.VarType.FP16: 2,
......@@ -107,7 +108,7 @@ class ControlFlowGraph(object):
# Repeatedly apply liveness updates until the algorithm stablize
# on a complete set live input vars and live output vars.
while True:
for i in reversed(range(self.op_size)):
for i in reversed(list(range(self.op_size))):
live_in[i] = set(self._live_in[i])
live_out[i] = set(self._live_out[i])
for s in self._successors[i]:
......@@ -172,9 +173,10 @@ class ControlFlowGraph(object):
is_forward = i < self._forward_num
in_diff, out_diff = self._get_diff(self._live_in[i],
self._live_out[i])
can_optimize = filter(
lambda x: self._check_var_validity(block_desc, x, is_forward),
in_diff)
can_optimize = [
x for x in in_diff
if self._check_var_validity(block_desc, x, is_forward)
]
if can_optimize:
index = i + fwd_id + 1 if is_forward else i - self._forward_num + bwd_id + 1
delete_op = block_desc._insert_op(index)
......@@ -213,9 +215,10 @@ class ControlFlowGraph(object):
block_desc = op.block()
is_forward = i < self._forward_num
if self.pool:
defs_can_optimize = filter(
lambda x: self._check_var_validity(block_desc, x, is_forward),
self._defs[i])
defs_can_optimize = [
x for x in self._defs[i]
if self._check_var_validity(block_desc, x, is_forward)
]
out_pair = [
(x, self._find_var(block_desc, x, is_forward).shape())
for x in defs_can_optimize
......@@ -243,11 +246,11 @@ class ControlFlowGraph(object):
continue
if PRINT_LOG:
print(("Hit Cache !!!! cache pool index "
"is %d, var name is %s, "
"cached var name is %s, "
"var shape is %s ") % (index, x, cache_var,
str(cache_shape)))
print((("Hit Cache !!!! cache pool index "
"is %d, var name is %s, "
"cached var name is %s, "
"var shape is %s ") % (index, x, cache_var,
str(cache_shape))))
self.pool.pop(index)
if x == cache_var:
break
......@@ -261,9 +264,10 @@ class ControlFlowGraph(object):
break
in_diff, _ = self._get_diff(self._live_in[i], self._live_out[i])
can_optimize = filter(
lambda x: self._check_var_validity(block_desc, x, is_forward),
in_diff)
can_optimize = [
x for x in in_diff
if self._check_var_validity(block_desc, x, is_forward)
]
if can_optimize:
for var_name in can_optimize:
self.pool.append((var_name, self._find_var(
......
python/paddle/fluid/unique_name.py
浏览文件 @ 559d3632
......@@ -67,7 +67,7 @@ def switch(new_generator=None):
@contextlib.contextmanager
def guard(new_generator=None):
if isinstance(new_generator, basestring):
if isinstance(new_generator, str):
new_generator = UniqueNameGenerator(new_generator)
old = switch(new_generator)
yield
......
python/paddle/reader/creator.py
浏览文件 @ 559d3632
......@@ -67,10 +67,10 @@ def recordio(paths, buf_size=100):
import recordio as rec
import paddle.reader.decorator as dec
import cPickle as pickle
import pickle as pickle
def reader():
if isinstance(paths, basestring):
if isinstance(paths, str):
path = paths
else:
path = ",".join(paths)
......
python/paddle/reader/decorator.py
浏览文件 @ 559d3632
......@@ -21,6 +21,7 @@ from threading import Thread
import subprocess
from six.moves.queue import Queue
from six.moves import zip_longest
import itertools
import random
import zlib
......@@ -42,7 +43,7 @@ def map_readers(func, *readers):
rs = []
for r in readers:
rs.append(r())
for e in itertools.imap(func, *rs):
for e in map(func, *rs):
yield e
return reader
......@@ -148,16 +149,16 @@ def compose(*readers, **kwargs):
for r in readers:
rs.append(r())
if not check_alignment:
for outputs in itertools.izip(*rs):
yield sum(map(make_tuple, outputs), ())
for outputs in zip(*rs):
yield sum(list(map(make_tuple, outputs)), ())
else:
for outputs in itertools.izip_longest(*rs):
for outputs in zip_longest(*rs):
for o in outputs:
if o is None:
# None will be not be present if compose is aligned
raise ComposeNotAligned(
"outputs of readers are not aligned.")
yield sum(map(make_tuple, outputs), ())
yield sum(list(map(make_tuple, outputs)), ())
return reader
......@@ -306,7 +307,7 @@ def xmap_readers(mapper, reader, process_num, buffer_size, order=False):
args = (in_queue, out_queue, mapper, out_order) if order else (
in_queue, out_queue, mapper)
workers = []
for i in xrange(process_num):
for i in range(process_num):
worker = Thread(target=target, args=args)
worker.daemon = True
workers.append(worker)
......
python/paddle/reader/tests/decorator_test.py
浏览文件 @ 559d3632
......@@ -136,7 +136,7 @@ class TestXmap(unittest.TestCase):
reader = paddle.reader.xmap_readers(mapper,
reader_creator_10(0),
tNum, size, order)
for n in xrange(3):
for n in range(3):
result = []
for i in reader():
result.append(i)
......@@ -156,7 +156,7 @@ class TestPipeReader(unittest.TestCase):
import tempfile
records = [str(i) for i in xrange(5)]
records = [str(i) for i in range(5)]
temp = tempfile.NamedTemporaryFile()
try:
with open(temp.name, 'w') as f:
......
tools/test_runner.py
浏览文件 @ 559d3632
......@@ -12,19 +12,20 @@
# See the License for the specific language governing permissions and
# limitations under the License.
from __future__ import print_function
import unittest
import os
import sys
import paddle.fluid as fluid
import importlib
import cStringIO
from six.moves import cStringIO
def main():
sys.path.append(os.getcwd())
some_test_failed = False
for module_name in sys.argv[1:]:
buffer = cStringIO.StringIO()
buffer = cStringIO()
main = fluid.Program()
startup = fluid.Program()
scope = fluid.core.Scope()
......@@ -37,8 +38,11 @@ def main():
res = unittest.TextTestRunner(stream=buffer).run(tests)
if not res.wasSuccessful():
some_test_failed = True
print >> sys.stderr, module_name, 'failed\n', buffer.getvalue(
)
print(
module_name,
'failed\n',
buffer.getvalue(),
file=sys.stderr)
if some_test_failed:
exit(1)
......
Markdown is supported
0% 或 .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑！
想要评论请 注册 或