LoginSignup
1
0

More than 5 years have passed since last update.

@7of9

TensorFlow > TFRecords読込み版sine curveの学習 v0.1,v0.2 > OutOfRangeError | current size 0 > tf.initialize_local_variables()を実施する

Last updated at Posted at 2017-07-08
動作環境
GeForce GTX 1070 (8GB)
ASRock Z170M Pro4S [Intel Z170chipset]
Ubuntu 16.04 LTS desktop amd64
TensorFlow v1.1.0
cuDNN v5.1 for Linux
CUDA v8.0
Python 3.5.2
IPython 6.0.0 -- An enhanced Interactive Python.
gcc (Ubuntu 5.4.0-6ubuntu1~16.04.4) 5.4.0 20160609
GNU bash, version 4.3.48(1)-release (x86_64-pc-linux-gnu)

TensorFlow > sine curveの学習 > TensorFlowコードでpredictionをグラフ化してみた > sine curveになっていなかった > sine curveになった ( 誤差:0.01以下)
のTFRecords版を実装中。

ファイル書出し処理

prep_TFRecord_170708.py
import numpy as np
import random
import tensorflow as tf

"""
v0.1, Jul. 08, 2017
   - add _int64_feature()
   - add _bytes_feature()
"""

# on
#   Ubuntu 16.04 LTS
#   TensorFlow v1.1
#   Python 3.5.2

# codingrule: PEP8


def _bytes_feature(value):
    return tf.train.Feature(bytes_list=tf.train.BytesList(value=[value]))


def _int64_feature(value):
    return tf.train.Feature(int64_list=tf.train.Int64List(value=[value]))


OUT_FILE = 'input_170708.tfrecords'
NUM_DATA = 100
#
x_data = np.random.rand(NUM_DATA)
y_data = np.sin(2*np.pi*x_data) + 0.3 * np.random.rand()

with tf.python_io.TFRecordWriter(OUT_FILE) as tf_writer:
    for xs, ys in zip(x_data, y_data):
        x_org = np.array(xs, dtype=np.float32)
        y_org = np.array(ys, dtype=np.float32)
        x_raw = x_org.tostring()
        y_raw = y_org.tostring()
        example = tf.train.Example(features=tf.train.Features(feature={
            'x_raw': _bytes_feature(x_raw),
            'y_raw': _bytes_feature(y_raw)}))
        tf_writer.write(example.SerializeToString())

for xs, ys in zip(x_data, y_data):
    print('%.5f, %.5f' % (xs, ys))

テスト読込み処理

test_readTFRecord_170708.py
import numpy as np
import tensorflow as tf

# codingrule: PEP8

INP_FILE = 'input_170708.tfrecords'

record_iterator = tf.python_io.tf_record_iterator(path=INP_FILE)

for record in record_iterator:
    example = tf.train.Example()
    example.ParseFromString(record)

    x_raw = (example.features.feature['x_raw']
             .bytes_list
             .value[0])
    x_1d = np.fromstring(x_raw, dtype=np.float32)
    x_org = x_1d.reshape([1, -1])

    y_raw = (example.features.feature['y_raw']
             .bytes_list
             .value[0])
    y_1d = np.fromstring(y_raw, dtype=np.float32)
    y_org = y_1d.reshape([1, -1])

    print(x_org, y_org)


$ python3 test_readTFRecord_170708.py | head
[[ 0.90197307]] [[-0.52444661]]
[[ 0.39993107]] [[ 0.64140004]]
[[ 0.75747746]] [[-0.94563216]]
[[ 0.16201401]] [[ 0.9043051]]
[[ 0.34640592]] [[ 0.8753475]]
[[ 0.45442739]] [[ 0.33570865]]
[[ 0.72813314]] [[-0.93731195]]
[[ 0.00909476]] [[ 0.11037734]]
[[ 0.50905269]] [[-0.00358463]]
[[ 0.36261058]] [[ 0.81321895]]

テスト読込みはできた。

v0.1

sessionでの使用 > 失敗

learn_sineCurve_170708.py
import numpy as np
import tensorflow as tf
import tensorflow.contrib.slim as slim

"""
v0.1 Jul. 08, 2017
  - only read [.tfrecord]
    + add inputs_xy()
    + add read_and_decode()
"""

# codingrule: PEP8

IN_FILE = 'input_170708.tfrecords'


def read_and_decode(filename_queue):
    reader = tf.TFRecordReader()
    _, serialized_example = reader.read(filename_queue)
    features = tf.parse_single_example(
        serialized_example,
        features={
           'x_raw': tf.FixedLenFeature([], tf.string),
           'y_raw': tf.FixedLenFeature([], tf.string),
        })
    x_raw = tf.decode_raw(features['x_raw'], tf.float32)
    y_raw = tf.decode_raw(features['y_raw'], tf.float32)

    x_org = tf.reshape(x_raw, [1])
    y_org = tf.reshape(y_raw, [1])

    return x_org, y_org


def inputs_xy():
    filename = IN_FILE
    filequeue = tf.train.string_input_producer(
        [filename], num_epochs=1)

    x_org, y_org = read_and_decode(filequeue)
    return x_org, y_org

x_orgs, y_orgs = inputs_xy()
batch_size = 1  # 4
x_batch, y_batch = tf.train.shuffle_batch([x_orgs, y_orgs],
                                          batch_size,
                                          capacity=10,
                                          min_after_dequeue=7)
init_op = tf.initialize_all_variables()


with tf.Session() as sess:
    sess.run(init_op)
    coord = tf.train.Coordinator()
    threads = tf.train.start_queue_runners(sess=sess, coord=coord)

    try:
        ax, ay = sess.run([x_batch, y_batch])
        print(ax, ay)
    finally:
        coord.request_stop()

以下のようなエラーが出る。


tensorflow.python.framework.errors_impl.OutOfRangeError: RandomShuffleQueue '_0_shuffle_batch/random_shuffle_queue' is closed and has insufficient elements (requested 1, current size 0)
[[Node: shuffle_batch = QueueDequeueManyV2[component_types=[DT_FLOAT, DT_FLOAT], timeout_ms=-1, _device="/job:localhost/replica:0/task:0/cpu:0"](shuffle_batch/random_shuffle_queue, shuffle_batch/n)]]

current size 0となる理由について調べているが、今のところ、よくわからない。

v0.2 > current size 0エラーの対処

変更箇所

TensorFlow ファイルからの入力のいろいろ
を参考に以下の変更をした。

# init_op = tf.initialize_all_variables()
init_op = [tf.initialize_all_variables(), tf.initialize_local_variables()]

code

learn_sineCurve_170708.py
import numpy as np
import tensorflow as tf
import tensorflow.contrib.slim as slim

"""
v0.2 Jul. 08, 2017
  - fix bug > OutOfRangeError (current size 0)
    + use [tf.initialize_local_variables()]
v0.1 Jul. 08, 2017
  - only read [.tfrecord]
    + add inputs_xy()
    + add read_and_decode()
"""

# codingrule: PEP8

IN_FILE = 'input_170708.tfrecords'


def read_and_decode(filename_queue):
    reader = tf.TFRecordReader()
    _, serialized_example = reader.read(filename_queue)
    features = tf.parse_single_example(
        serialized_example,
        features={
           'x_raw': tf.FixedLenFeature([], tf.string),
           'y_raw': tf.FixedLenFeature([], tf.string),
        })
    x_raw = tf.decode_raw(features['x_raw'], tf.float32)
    y_raw = tf.decode_raw(features['y_raw'], tf.float32)

    x_org = tf.reshape(x_raw, [1])
    y_org = tf.reshape(y_raw, [1])

    return x_org, y_org


def inputs_xy():
    filename = IN_FILE
    filequeue = tf.train.string_input_producer(
        [filename], num_epochs=1)

    x_org, y_org = read_and_decode(filequeue)
    return x_org, y_org

x_orgs, y_orgs = inputs_xy()
batch_size = 1  # 4
x_batch, y_batch = tf.train.shuffle_batch([x_orgs, y_orgs],
                                          batch_size,
                                          capacity=10,
                                          min_after_dequeue=7)
# init_op = tf.initialize_all_variables()
init_op = [tf.initialize_all_variables(), tf.initialize_local_variables()]

with tf.Session() as sess:
    sess.run(init_op)
    coord = tf.train.Coordinator()
    threads = tf.train.start_queue_runners(sess=sess, coord=coord)

    try:
        ax, ay = sess.run([x_batch, y_batch])
        print(ax, ay)
    finally:
        coord.request_stop()

実行

$ python3 learn_sineCurve_170708.py 
...
2017-07-08 11:54:04.261826: I tensorflow/core/common_runtime/gpu/gpu_device.cc:908] DMA: 0 
2017-07-08 11:54:04.261844: I tensorflow/core/common_runtime/gpu/gpu_device.cc:918] 0:   Y 
2017-07-08 11:54:04.261853: I tensorflow/core/common_runtime/gpu/gpu_device.cc:977] Creating TensorFlow device (/gpu:0) -> (device: 0, name: GeForce GTX 1070, pci bus id: 0000:01:00.0)
[[ 0.39993107]] [[ 0.64140004]]

current size 0エラーが出なくなり、ax, ayの値がprintされるようになりました。

@learn_tensorflow さん
情報感謝です。

別途、参考にしていたfully_connected_reader.pyにおいてもlocal_variablesの初期化をしていたようだ。

    init_op = tf.group(tf.global_variables_initializer(),
                       tf.local_variables_initializer())
1
0
0

Register as a new user and use Qiita more conveniently

  1. You get articles that match your needs
  2. You can efficiently read back useful information
  3. You can use dark theme
What you can do with signing up
Sign upLogin
1
0