OOM在分配张量时

时间:2017-10-02 16:37:14

标签: python tensorflow neural-network deep-learning conv-neural-network

如何在分配张量时解决ResourceExhaustedError:OOM的问题?

  

ResourceExhaustedError(参见上面的回溯):分配时的OOM   张量形状[10000,32,28,28]

我包含了几乎所有的代码

learning_rate = 0.0001
epochs = 10
batch_size = 50

# declare the training data placeholders
# input x - for 28 x 28 pixels = 784 - this is the flattened image data that is drawn from
# mnist.train.nextbatch()
x = tf.placeholder(tf.float32, [None, 784])
# dynamically reshape the input
x_shaped = tf.reshape(x, [-1, 28, 28, 1])
# now declare the output data placeholder - 10 digits
y = tf.placeholder(tf.float32, [None, 10])
def create_new_conv_layer(input_data, num_input_channels, num_filters, filter_shape, pool_shape, name):
    # setup the filter input shape for tf.nn.conv_2d
    conv_filt_shape = [filter_shape[0], filter_shape[1], num_input_channels,
                      num_filters]

    # initialise weights and bias for the filter
    weights = tf.Variable(tf.truncated_normal(conv_filt_shape, stddev=0.03),
                                      name=name+'_W')
    bias = tf.Variable(tf.truncated_normal([num_filters]), name=name+'_b')

    # setup the convolutional layer operation
    out_layer = tf.nn.conv2d(input_data, weights, [1, 1, 1, 1], padding='SAME')

    # add the bias
    out_layer += bias

    # apply a ReLU non-linear activation
    out_layer = tf.nn.relu(out_layer)

    # now perform max pooling
    ksize = [1, 2, 2, 1]
    strides = [1, 2, 2, 1]
    out_layer = tf.nn.max_pool(out_layer, ksize=ksize, strides=strides,
                               padding='SAME')

    return out_layer
# create some convolutional layers
layer1 = create_new_conv_layer(x_shaped, 1, 32, [5, 5], [2, 2], name='layer1')
layer2 = create_new_conv_layer(layer1, 32, 64, [5, 5], [2, 2], name='layer2')

flattened = tf.reshape(layer2, [-1, 7 * 7 * 64])

# setup some weights and bias values for this layer, then activate with ReLU
wd1 = tf.Variable(tf.truncated_normal([7 * 7 * 64, 1000], stddev=0.03), name='wd1')
bd1 = tf.Variable(tf.truncated_normal([1000], stddev=0.01), name='bd1')
dense_layer1 = tf.matmul(flattened, wd1) + bd1
dense_layer1 = tf.nn.relu(dense_layer1)

# another layer with softmax activations
wd2 = tf.Variable(tf.truncated_normal([1000, 10], stddev=0.03), name='wd2')
bd2 = tf.Variable(tf.truncated_normal([10], stddev=0.01), name='bd2')
dense_layer2 = tf.matmul(dense_layer1, wd2) + bd2
y_ = tf.nn.softmax(dense_layer2)
cross_entropy = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(logits=dense_layer2, labels=y))


# add an optimiser
optimiser = tf.train.AdamOptimizer(learning_rate=learning_rate).minimize(cross_entropy)

# define an accuracy assessment operation
correct_prediction = tf.equal(tf.argmax(y, 1), tf.argmax(y_, 1))
accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))

# setup the initialisation operator
init_op = tf.global_variables_initializer() 



 with tf.Session() as sess:
            # initialise the variables
            sess.run(init_op)
            total_batch = int(len(mnist.train.labels) / batch_size)
            for epoch in range(epochs):
                avg_cost = 0
                for i in range(total_batch):
                    batch_x, batch_y = mnist.train.next_batch(batch_size=batch_size)
                    _, c = sess.run([optimiser, cross_entropy], feed_dict={x: 
         batch_x, 
            y: batch_y})
                    avg_cost += c / total_batch
                test_acc = sess.run(accuracy,feed_dict={x: mnist.test.images, y: 
            mnist.test.labels})
                print("Epoch:", (epoch + 1), "cost =", "{:.3f}".format(avg_cost), "  
            test accuracy: {:.3f}".format(test_acc))

            print("\nTraining complete!")
            print(sess.run(accuracy, feed_dict={x: mnist.test.images, y: 
            mnist.test.labels}))

并且错误中引用的那些行是: create_new_conv_layer - function

sess.run ..在训练循环中

我从调试器输出中复制的更多错误列在下面(有更多行,但我认为这些是主要的,其他是由此引起的。)

  

tensorflow.python.framework.errors_impl.ResourceExhaustedError:分配张量形状的OOM [10000,32,28,28] [[节点:Conv2D =   Conv2D [T = DT_FLOAT,data_format =" NHWC&#34 ;, padding =" SAME&#34 ;, strides = [1,1,   1,1],use_cudnn_on_gpu = true,   _device =" / job:localhost / replica:0 / task:0 / gpu:0"](重塑,layer1_W / read)]]

我第二次运行它是发出以下错误我有cpu和GPU可以在下面的输出中看到,我可以理解一些与cpu问题相关的错误可能是因为我的tensorflow没有编译使用那些功能,我在Windows 10上安装了cuda 8和cudnn 6,python 3.5,tensorflow 1.3.0。

  

2017-10-03 03:53:58.944371:W   C:\ tf_jenkins \家庭\工作区\ REL-WIN \中号\ WINDOWS-GPU \ PY \ 35 \ tensorflow \核心\平台\ cpu_feature_guard.cc:45]   TensorFlow库未编译为使用AVX指令,但是   这些都可以在您的机器上使用,并可以加速CPU   计算。 2017-10-03 03:53:58.945563:W   C:\ tf_jenkins \家庭\工作区\ REL-WIN \中号\ WINDOWS-GPU \ PY \ 35 \ tensorflow \核心\平台\ cpu_feature_guard.cc:45]   TensorFlow库未编译为使用AVX2指令,但是   这些都可以在您的机器上使用,并可以加速CPU   计算。 2017-10-03 03:53:59.230761:我   C:\ tf_jenkins \家庭\工作区\ REL-WIN \中号\ WINDOWS-GPU \ PY \ 35 \ tensorflow \核心\ common_runtime \ GPU \ gpu_device.cc:955]   找到具有属性的设备0:   名称:Quadro K620主要:5次要:0 memoryClockRate(GHz)1.124 pciBusID 0000:01:00.0总内存:2.00GiB可用内存:1.66GiB   2017-10-03 03:53:59.231109:我   C:\ tf_jenkins \家庭\工作区\ REL-WIN \中号\ WINDOWS-GPU \ PY \ 35 \ tensorflow \核心\ common_runtime \ GPU \ gpu_device.cc:976]   DMA:0 2017-10-03 03:53:59.231229:我   C:\ tf_jenkins \家庭\工作区\ REL-WIN \中号\ WINDOWS-GPU \ PY \ 35 \ tensorflow \核心\ common_runtime \ GPU \ gpu_device.cc:986]   0:Y 2017-10-03 03:53:59.231363:我   C:\ tf_jenkins \家庭\工作区\ REL-WIN \中号\ WINDOWS-GPU \ PY \ 35 \ tensorflow \核心\ common_runtime \ GPU \ gpu_device.cc:1045]   创建TensorFlow设备(/ gpu:0) - > (设备:0,名称:Quadro K620,   pci bus id:0000:01:00.0)2017-10-03 03:54:01.511141:E   C:\ tf_jenkins \家庭\工作区\ REL-WIN \中号\ WINDOWS-GPU \ PY \ 35 \ tensorflow \ stream_executor \ CUDA \ cuda_dnn.cc:371]   无法创建cudnn句柄:CUDNN_STATUS_NOT_INITIALIZED 2017-10-03 03:54:01.511372:E   C:\ tf_jenkins \家庭\工作区\ REL-WIN \中号\ WINDOWS-GPU \ PY \ 35 \ tensorflow \ stream_executor \ CUDA \ cuda_dnn.cc:375]   错误检索驱动程序版本:未实现:内核报告的驱动程序版本未在Windows上实现 2017-10-03   03:54:01.511862:E   C:\ tf_jenkins \家庭\工作区\ REL-WIN \中号\ WINDOWS-GPU \ PY \ 35 \ tensorflow \ stream_executor \ CUDA \ cuda_dnn.cc:338]   无法破坏cudnn句柄:CUDNN_STATUS_BAD_PARAM 2017-10-03 03:54:01.512074:F   C:\ tf_jenkins \家庭\工作区\ REL-WIN \中号\ WINDOWS-GPU \ PY \ 35 \ tensorflow \核心\仁\ conv_ops.cc:672]   检查失败:stream-> parent() - > GetConvolveAlgorithms(   conv_parameters.ShouldIncludeWinogradNonfusedAlgo(),& algorithms)

1 个答案:

答案 0 :(得分:2)

由于您一次推送整个测试集以进行评估,因此进程因内存不足(OOM)而失败(请参阅this question)。很容易看出10000 * 32 * 28 * 28 * 4几乎是1Gb,而你的GPU总共只有1.66Gb可用,而且大部分已经被网络本身使用了。

解决方案是为神经网络批次提供饲料,不仅用于培训,还用于测试。结果准确度将是所有批次的平均值。此外,您不需要在每个时代之后执行此操作:您是否真的对所有中间网络的测试结果感兴趣?

您的第二条错误消息很可能是之前失败的结果,因为CUDNN驱动程序似乎不再起作用。我建议重启你的机器。