尝试使用未初始化的值 InceptionV3/Mixed_6d/Branch_3/Conv2d_0b_1x

Posted 2023-03-12

【中文标题】尝试使用未初始化的值 InceptionV3/Mixed_6d/Branch_3/Conv2d_0b_1x

【英文标题】：Attempting to use uninitialized value InceptionV3/Mixed_6d/Branch_3/Conv2d_0b_1x

【发布时间】：2018-06-18 05:20:01

【问题描述】：

我修改了 Inception V3 网络（删除了一些层模块），并创建了 6 个类训练数据，每个类 1 个图像。当我执行训练时，我得到了错误

tensorflow.python.framework.errors_impl.FailedPreconditionError： 尝试使用未初始化的值 InceptionV3/Mixed_6d/Branch_3/Conv2d_0b_1x1/weights [[节点： InceptionV3/Mixed_6d/Branch_3/Conv2d_0b_1x1/weights/read = 身份[T=DT_FLOAT, _class=["loc:/Branch_3/Conv2d_0

火车代码：

import tensorflow as tf
import inception
import create_record
import numpy as np
import inception_utils
width, height = 299, 299
classes = 6
batch_size = 6
learning_rate = 0.01
max_step = 1
image_dir = '/home/xzy/test/images/'
path = '/home/xzy/test/train.tfrecords'
logs_dir = '/home/xzy/test/logs/'
# %% Training
def train():
    filename_queue = tf.train.string_input_producer([path])
    reader = tf.TFRecordReader()
    _, serialized_example = reader.read(filename_queue)
    features = tf.parse_single_example(serialized_example,
                                       features=
                                           'label': tf.FixedLenFeature([], tf.int64),
                                           'img_raw': tf.FixedLenFeature([], tf.string),
                                       )
    image = tf.decode_raw(features['img_raw'], tf.uint8)
    image = tf.reshape(image, [299, 299, 3])
    label = tf.cast(features['label'], tf.int32)
    image_batch, label_batch = tf.train.batch([image, label],
                                              batch_size=6, num_threads=64, capacity=300)
    label_batch = tf.one_hot(label_batch, depth=classes)
    label_batch = tf.cast(label_batch, dtype=tf.int32)
    label_batch = tf.reshape(label_batch, [batch_size, classes])
    x = tf.placeholder(tf.float32, shape=[batch_size, width, height, 3])
    y_ = tf.placeholder(tf.int16, shape=[batch_size, classes])
    init_op = tf.initialize_all_variables()
    logits = inception.inference(x, num_classes=classes)
    loss = inception.loss(logits, y_)
    my_global_step = tf.Variable(0, name='global_step', trainable=False)
    optimizer = tf.train.GradientDescentOptimizer(learning_rate)
    train_op = optimizer.minimize(loss, global_step=my_global_step)
    saver = tf.train.Saver(tf.global_variables())
    summary_op = tf.summary.merge_all()
    with tf.Session() as sess:
        sess.run(init_op)
        coord = tf.train.Coordinator()
        threads = tf.train.start_queue_runners(coord=coord)
        train_summary_writer = tf.summary.FileWriter(logs_dir, sess.graph)
        try:
            for step in np.arange(max_step):
                if coord.should_stop():
                    break
                example, lab = sess.run([image_batch, label_batch])
                example = tf.to_float(example)
                _, train_loss = sess.run([train_op, loss], feed_dict=x: example.eval(), y_: lab)
                if step == 0 or (step + 1) == max_step:
                    print ('Step: %d, loss: %.4f' % (step, train_loss))
                    summary_str = sess.run(summary_op)
                    train_summary_writer.add_summary(summary_str, step)
                    if step % 2000 == 0 or (step + 1) == max_step:
                        checkpoint_path = os.path.join(train_log_dir, 'model.ckpt')
                        saver.save(sess, checkpoint_path, global_step=step)
        except tf.errors.OutOfRangeError:
            print('Done training -- epoch limit reached')
        coord.request_stop()
        coord.join(threads)
        sess.close()
train()

错误堆栈跟踪：

Traceback（最近一次调用最后一次）：

文件“/home/xzy/PycharmProjects/network/train_inception.py”，第 89 行，在 火车（）

文件“/home/xzy/PycharmProjects/network/train_inception.py”，第 71 行，在 train() _, train_loss = sess.run([train_op, loss], feed_dict=x: example.eval(), y_: lab)

文件“/usr/local/lib/python2.7/dist-packages/tensorflow/python/client/session.py”，第 889 行，运行中 run_metadata_ptr)

文件“/usr/local/lib/python2.7/dist-packages/tensorflow/python/client/session.py”，第 1120 行，在 _run feed_dict_tensor、选项、run_metadata)

文件“/usr/local/lib/python2.7/dist-packages/tensorflow/python/client/session.py”，第 1317 行，在 _do_run 选项，run_metadata）

文件“/usr/local/lib/python2.7/dist-packages/tensorflow/python/client/session.py”，第 1336 行，在 _do_call raise type(e)(node_def, op, message)

tensorflow.python.framework.errors_impl.FailedPreconditionError: Attempting to use uninitialized value InceptionV3/Mixed_6d/Branch_3/Conv2d_0b_1x1/weights [[节点：InceptionV3/Mixed_6d/Branch_3/Conv2d_0b_1x1/weights/read = Identity[T=DT_FLOAT, _class=["loc:@InceptionV3/Mixed_6d/Branch_3/Conv2d_0

怎么了？有人能给我一些想法吗，谢谢？

Tensorflow 版本：1.5.0-dev20171206，python 2.7，Ubuntu 16.04。

【问题讨论】：

你忘了这个 - tensorflow.org/api_docs/python/tf/global_variables_initializer

@Maxim。感谢您的回答。我将代码从init_op = tf.initialize_all_variables() 更改为init_op = tf.global_variables_initializer()。但我得到类似的错误。只有错误提示的差异才会改变未初始化错误与权重未初始化错误的偏差。详情是Attempting to use uninitialized value InceptionV3/Mixed_6d/Branch_3/Conv2d_0b_1x1/biases

如果问题出在 gcloud - 检查这个答案：***.com/a/60384189/4137497

【参考方案1】：

您的init_op 定义过早：

init_op = tf.initialize_all_variables()

# BAD! All the ops below won't get initialized!
logits = inception.inference(x, num_classes=classes)
loss = inception.loss(logits, y_)
my_global_step = tf.Variable(0, name='global_step', trainable=False)
optimizer = tf.train.GradientDescentOptimizer(learning_rate)
train_op = optimizer.minimize(loss, global_step=my_global_step)

解决方案：

logits = inception.inference(x, num_classes=classes)
loss = inception.loss(logits, y_)
my_global_step = tf.Variable(0, name='global_step', trainable=False)
optimizer = tf.train.GradientDescentOptimizer(learning_rate)
train_op = optimizer.minimize(loss, global_step=my_global_step)

# Now it's OK.
init_op = tf.global_variables_initializer()

【讨论】：

非常感谢。你的解决方案很好！现在可以了。谢谢