2016年11月18日星期五

MXNET符号API

MXNET的符号API

本节中，我将介绍MXNET主要的符号API。符号是神经网络的构件。每个符号可以被视为具有前向和后向操作的功能对象。单个符号可以组成更复杂的符号，从而成为神经网络。
有关Symbolic API的更详细的教程，请参阅：http://mxnet.io/api/python/symbol.html

符号变量

mxnet.symbol.Variable(name,attr=None,shape=None)
作用：创建具有指定名称的符号变量。

参数：

name (str) - 变量的名称。

attr (dict of string - > string) - 要在变量上设置的其他属性。

shape (tuple) - 可以选择指定变量的形状。这将在形状推理期间使用。如果用户在调用形状推断时使用关键字参数为此变量指定了不同的形状，则此形状信息将被忽略。

返回：

variable - 创建的变量符号。
返回类型：Symbol

全连接层和卷积层

全连接层

mxnet.symbol.FullyConnected(*args,**kwargs)
作用：将矩阵乘法应用于输入，然后添加偏置。它将形状(batch_size，input_dim)的输入映射到(batch_size，num_hidden)的形状。可学习参数包括线性变换的权重和可选的偏置向量。

参数：

data (Symbol) - 将数据输入到FullyConnectedOp。

weight (Symbol) - 权重矩阵。

bias (Symbol) - 偏置参数。

num_hidden (int，必须) - 输出的隐藏节点数。

no_bias (boolean，可选，default = False) - 是否禁用偏置参数。

name (string，可选.) - 结果符号的名称。

返回：

symbol - 结果符号。
返回类型：Symbol

卷积层

mxnet.symbol.Convolution(*args,**kwargs)
作用：将卷积应用于输入，然后添加偏置。

参数：

data (Symbol) - 输入数据到ConvolutionOp。

weight (Symbol) - 权重矩阵。

bias (Symbol) - 偏置参数。

kernel(Shape(tuple)，必须) - 卷积核大小：(y，x)或(d，y，x)

stride (Shape(tuple)，可选，default =(1,1)) - 卷积窗口移动的步长：(y，x)or(d，y，x)

dilate (Shape(tuple)，可选，default =(1,1)) - 卷积扩张：(y，x)

pad (Shape(tuple)，可选，默认=(0,0)) - 卷积填充：(y，x)或(d，y，

num_filter (int(非负)，必须) - 卷积过滤器(通道)号

num_group (int(非负数)，可选，默认值= 1) - 组数。这个选项不支持CuDNN，你可以使用SliceChannel到num_group，应用卷积和concat来实现相同的需要。

workspace(long(非负)，可选，默认值= 1024) - 卷积的MB工作空间(MB)。

no_bias (boolean，可选，default = False) - 是否禁用偏置参数。

cudnn_tune (_ {'fastest'，'limitedworkspace'，'off'}，可选，default ='off') - 是否通过运行性能测试找到卷积算法。更好的speed.auto调整被默认关闭。设置环境varialbe MXNET_CUDNN_AUTOTUNE_DEFAULT = 1默认打开。

cudnn_off (boolean，可选，default = False) - 关闭cudnn。

name (string，可选.) - 结果符号的名称。

返回：

symbol - 结果符号。
返回类型：Symbol

反卷积层

mxnet.symbol.Deconvolution(*args,**kwargs)
作用：将反卷积应用于输入，然后添加偏置。

参数：

data (Symbol) - 输入数据到DeconvolutionOp。

weight (Symbol) - 权重矩阵。

bias (Symbol) - 偏置参数。

kernel (Shape(tuple)，必须) - 反卷积核大小：(y，x)

stride (Shape(tuple)，可选，default =(1,1)) - 反卷积窗移动步长：(y，x)

pad (Shape(tuple)，可选，默认=(0,0)) - 用于反卷积的pad：(y，x)，好的数字是：(kernel-1)/ 2，如果target_shape set ，pad将被忽略并自动计算

adj (Shape(tuple)，可选，default =(0,0)) - 输出形状的调整：(y，x)，如果target_shape set，adj将被忽略，

target_shape (Shape(tuple)，可选，default =(0,0)) - 输出形状为targe shape：(y，x)

num_filter (int(非负数)，必须) - 反卷积滤波器(通道)数

num_group (int(非负数)，可选，默认= 1) - 组分区数

workspace(long(非负)，可选，默认值= 512) - 反褶积工作空间(MB)

no_bias (boolean，可选，default = True) - 是否禁用偏置参数。

name (string，可选.) - 结果符号的名称。

返回：

symbol - 结果符号。
返回类型：Symbol

池化层

mxnet.symbol.Pooling(*args，**kwargs) 作用：对输入执行空间合并。

参数：

data (Symbol) - 输入数据到池操作符。

global_pool (boolean，可选，default = False) - 忽略内核大小，根据当前输入要素图进行全局池化。这对于具有不同形状的输入有用

kernel(Shape(tuple)，required) - 合并内核大小：(y，x)或(d，y，x)

pool_type ( {'avg'，'max'，'sum'}，required) - 要应用的池化类型。

pooling_convention ( {'full'，'valid'}，可选，default ='valid') - 要应用的池规则.kValid是Mxnet的默认设置，并舍弃输出池大小.kFull是兼容Caffe和舍入输出池大小。

stride (Shape(tuple)，optional，default =(1,1)) - 池化窗口移动的步长(y，x)or(d，y，x)

pad (Shape(tuple)，optional，default =(0,0)) - 池化窗的填充(y，x)or(d，y，x)

name (string，optional.) - 结果符号的名称。

返回：

symbol - 结果符号。返回类型：Symbol

循环神经网络层

mxnet.symbol.RNN(*args，**kwargs)
作用：应用循环层来输入。

参数：

data (Symbol) - 将数据输入到RNN

parameter(Symbol) - 矢量所有RNN可训练参数连接

state (Symbol) - RNN的初始隐藏状态

state_cell (Symbol) - LSTM网络的初始单元格状态(仅限对于LSTM)

state_size (int(non-negative)，required) - 每个层的状态大小

num_layers (int(非负数)，required) - 堆叠层数

bidirectional (boolean，可选，default = False) - 是否使用双向循环层

mode({'gru'，'lstm'，'rnn_relu'，'rnntanh'}，required) - 计算的RNN类型

p (float，可选，默认值= 0) - 丢弃概率，在训练时被丢弃的输入的分数

state_outputs (boolean，可选，default = False) - 是否将状态作为符号输出。

name (string，optional.) - 结果符号的名称。

返回：

symbol - 结果符号。
返回类型：Symbol

激活层

mxnet.symbol.Activation(*args,**kwargs)
作用：元素激活函数。

支持以下激活类型(操作按元素应用于输入张量的每个标量)：

relu ：整流线性单位， $y = m a x (x ， 0)$

sigmoid ：

y = 1 / (1 + e x p (- x))

tanh：双曲正切，

y = (e x p (x) - e x p (- x)) / (e x p (x) + e x p (- x))

softreLU或SoftPlus，

y = l o g (1 + e x p (x))

有关参数的其他激活，请参见 LeakyReLU 。

参数：

act_type ( {'relu'，'sigmoid'，'softrelu'，'tanh'}，必须) - 要应用的激活函数。

name (string，可选.) - 结果符号的名称。

返回：

symbol - 结果符号。
返回类型：Symbol

softmax激活层

mxnet.symbol.SoftmaxActivation(*args,**kwargs)
作用：应用softmax激活输入。这是为内部层。对于输出(损耗层)，请使用SoftmaxOutput。如果mode = instance，此运算符将为批处理中的每个实例计算一个softmax;这是默认模式。如果mode = channel，此运算符将在每个实例的每个位置计算num_channel类softmax;这可以用于完全卷积网络，图像分割等。

参数：

data (Symbol) - 输入数据到激活功能。

mode( {'channel'，'instance'}，可选，默认='instance') - Softmax模式。如果设置为instance，此运算符将为batch中的每个实例计算一个softmax;这是默认模式。如果设置为channel，此运算符将在每个实例的每个位置计算num_channel类softmax;这可以用于完全卷积网络，图像分割等。

name (string，可选.) - 结果符号的名称。

返回：

symbol - 结果符号。
返回类型：Symbol

LeakyReLU层

mxnet.symbol.LeakyReLU(*args,**kwargs)
作用：应用激活功能输入。

参数：

data (Symbol) - 输入数据到激活功能。

act_type ( {'elu'，'leaky'，'prelu'，'rrelu'}，可选，default ='leaky') - 要应用的激活函数。

slope (float，可选，默认值= 0.25) - 激活的初始斜率。 (仅适用于泄漏和elu)

lower_bound (float，可选，默认值= 0.125) - 随机斜率的下限。 (仅适用于rrelu)

upper_bound (float，可选，default = 0.334) - 随机斜率的上限。 (仅适用于rrelu)

name (string，可选.) - 结果符号的名称。

返回：

symbol - 结果符号。
返回类型：Symbol

输出层

线性回归输出层

mxnet.symbol.LinearRegressionOutput(*args，**kwargs) 作用：对最终输出使用线性回归，这用于网的最终输出。

参数：

data (Symbol) - 输入数据到函数。

label (Symbol) - 输入标签到函数。

grad_scale (float，optional，default = 1) - 梯度缩放因子

name (string，optional.) - 结果符号的名称。

返回：

symbol - 结果符号。返回类型：Symbol

逻辑回归输出层

mxnet.symbol.LogisticRegressionOutput(*args，**kwargs) 作用：对最终输出使用Logistic回归，这用于网络的最终输出。逻辑回归适用于二进制分类或概率预测任务。

参数：

data (Symbol) - 输入数据到函数。

label (Symbol) - 输入标签到函数。

grad_scale (float，optional，default = 1) - 梯度缩放因子

name (string，optional.) - 结果符号的名称。

返回：

symbol - 结果符号。返回类型：Symbol

MAE输出层

mxnet.symbol.MAERegressionOutput(*args，**kwargs) 作用：对最终输出使用平均绝对误差回归，这用于网的最终输出。

参数：

data (Symbol) - 输入数据到函数。

label (Symbol) - 输入标签到函数。

grad_scale (float，optional，default = 1) - 梯度缩放因子

name (string，optional.) - 结果符号的名称。

返回：

symbol - 结果符号。返回类型：Symbol

softmax输出层

mxnet.symbol.SoftmaxOutput(*args,**kwargs)
作用：对输入执行softmax变换，使用logloss执行backprop。

参数：

data (Symbol) - 输入数据到softmax。

label (Symbol) - 标签数据，也可以是概率值与数据形状相同

grad_scale (float，可选，default = 1) - 梯度缩放因子

ignore_label (float，可选，默认值= -1) - 标签值将在向后被忽略(仅在use_ignore设置为true时有效)。

multi_output (boolean，可选，默认值= False) - 如果设置为true，对于(n，k，x_1，...，x_n)维输入张量，softmax将生成n x_1 x_n输出，每个都有k个类

use_ignore (boolean，可选，default = False) - 如果设置为true，ignore_label值将不会影响向后梯度

preserve_shape (boolean，可选，默认= False) - 如果为真，对于(n_1，n_2，...，n_d，k)维输入张量，softmax将生成(n1，n2， n_d，k)输出，将k个类标准化为最后一个维度。

normalization ( {'batch'，'null'，'valid'}，可选，default ='null') - 如果设置为null，op将不对输出梯度执行任何操作。如果设置为batch，op将通过除以batch大小以归一化梯度。如果设置为valid，op将通过除以样本数来归一化梯度

out_grad (boolean，可选，default = False) - 对输出梯度应用加权

name (string，可选.) - 结果符号的名称。

返回：

symbol - 结果符号。
返回类型：Symbol

SVM输出层

mxnet.symbol.SVMOutput(*args,**kwargs)
作用：支持向量机输出层，L1-SVM或L2-SVM

参数：

data (Symbol) - 将数据输入到svm。

label (Symbol) - 标签数据。

margin (float，可选，default = 1) - 调整DType(param_.margin)的激活大小

regularization_coefficient (float，可选，default = 1) - 正则系数

use_linear (boolean，可选，default = False) - 如果设置为true，则使用L1-SVM目标函数。默认使用L2-SVM目标

name (string，可选.) - 结果符号的名称。

返回：

symbol - 结果符号。
返回类型：Symbol

Reshape层

mxnet.symbol.Rechape(*args，**kwargs) 作用：重新塑造目标形状的输入

参数：

data (Symbol) - 输入数据以重塑。

target_shape (Shape(tuple)，可选，default =(0,0)) - (不推荐，请使用shape。)目标新形状。一个并且只有一个dim可以是0，在这种情况下，将从剩余的dims中推断出

keep_highest (boolean，可选，默认值= False) - (不推荐，请使用shape。)是否保持最高的dim不变。如果设置为true，则target_shape中的第一个dim被忽略，

shape(，可选，默认=()) - 目标新形状。如果dim是相同的，将其设置为0 \。如果dim设置为-1，则将从其余dims中推断出。一个且只有一个dim可以是-1

reverse (boolean，可选，default = False) - 是否匹配形状从向后。如果reverse为true，则将从后面搜索 shape 参数中的0个值。例如，原始形状是(10,5,4)，形状参数是(-1,0)。如果reverse为true，则新形状应为(50，4)。否则将是(40，5)。

name (string，optional.) - 结果符号的名称。

返回：

symbol - 结果符号。返回类型：Symbol

Cast层

mxnet.symbol.Cast(*args,**kwargs)
作用：将数组转换为不同的数据类型。

参数：

data (Symbol) - 输入要投射的数据。

dtype ( {'float16'，'float32'，'float64'，'int32'，'uint8'}，必须) - 目标数据类型。

name (string，可选.) - 结果符号的名称。

返回：

symbol - 结果符号。
返回类型：Symbol

Concat层

mxnet.symbol.Concat(*args,**kwargs)
作用：在指定的坐标轴上(defaut为1)上执行特征拼接，此函数支持可变长度的位置输入。

参数：

data (Symbol [] ) - 连接的张量列表

num_args (int，必须) - 要拼接的输入数。

dim (int，可选，默认值为'1') - 要拼接的维。

name (string，可选.) - 结果符号的名称。

返回：

symbol - 结果符号。
返回类型：Symbol

例子：

>>> import mxnet as mx
>>> data = mx.nd.array(range(6)).reshape((2,1,3))
>>> print "input shape = %s" % data.shape
>>> print "data = %s" % (data.asnumpy(), )
input shape = (2L, 1L, 3L)
data = [[[ 0\.  1\.  2.]]
 [[ 3\.  4\.  5.]]]

>>> # concat two variables on different dimensions
>>> a = mx.sym.Variable('a')
>>> b = mx.sym.Variable('b')
>>> for dim in range(3):
... cat = mx.sym.Concat(a, b, dim=dim)
... exe = cat.bind(ctx=mx.cpu(), args={'a':data, 'b':data})
... exe.forward()
... out = exe.outputs[0]
... print "concat at dim = %d" % dim
... print "shape = %s" % (out.shape, )
... print "results = %s" % (out.asnumpy(), )
concat at dim = 0
shape = (4L, 1L, 3L)
results = [[[ 0\.  1\.  2.]]
 [[ 3\.  4\.  5.]]
 [[ 0\.  1\.  2.]]
 [[ 3\.  4\.  5.]]]
concat at dim = 1
shape = (2L, 2L, 3L)
results = [[[ 0\.  1\.  2.]
 [ 0\.  1\.  2.]]
 [[ 3\.  4\.  5.]
 [ 3\.  4\.  5.]]]
concat at dim = 2
shape = (2L, 1L, 6L)
results = [[[ 0\.  1\.  2\.  0\.  1\.  2.]]
 [[ 3\.  4\.  5\.  3\.  4\.  5.]]]

ElementWiseSum层

mxnet.symbol.ElementWiseSum(*args,**kwargs)
作用：对所有输入执行逐元素求和。此功能支持可变长度的位置输入。

参数：

num_args (int，必须) - 要求和的输入数。

name (string，可选.) - 结果符号的名称。

返回：

symbol - 结果符号。
返回类型：Symbol

Embedding层

mxnet.symbol.Embedding(*args，**kwargs) 作用：将整数索引映射到向量表示(嵌入)。这些嵌入是可学习的参数。对于形状(d1，...，dK)的输入，输出形状为(d1，...，dK，output_dim)。所有输入值应为 [0，input_dim)范围内的整数。

参数：

data (Symbol) - 输入数据到EmbeddingOp。

weight (Symbol) - 包含权重矩阵。

input_dim (int，required) - 输入索引的词汇大小。

output_dim (int，required) - 嵌入向量的维数。

name (string，optional.) - 结果符号的名称。

返回：

symbol - 结果符号。返回类型：Symbol

Flatten层

mxnet.symbol.Flatten(*args,**kwargs)
作用：通过高维tensor输入转换为2D矩阵。比如 (d1，d2，...，dK)的张量被展平为(d1，d2 ... dK)矩阵。

参数：

data (Symbol) - 输入要展平的数据。

name (string，可选.) - 结果符号的名称。

返回：

symbol - 结果符号。
返回类型：Symbol

例子：

>>> data = Variable('data')  # say this is 4D from some conv/pool
>>> flatten = Flatten(data=data, name='flat')  # now this is 2D
>>> SymbolDoc.get_output_shape(flatten, data=(2, 3, 4, 5))
{'flat_output': (2L, 60L)}

>>> test_dims = [(2, 3, 4, 5), (2, 3), (2,)]
>>> op = Flatten(name='flat')
>>> for dims in test_dims:
... x = test_utils.random_arrays(dims)
... y = test_utils.simple_forward(op, flat_data=x)
... y_np = x.reshape((dims[0], numpy.prod(dims[1:])))
... print('%s: %s' % (dims, test_utils.almost_equal(y, y_np)))
(2, 3, 4, 5): True
(2, 3): True
(2,): True

Group层

mxnet.symbol.Group(symbols)
作用：创建将符号分组在一起的符号。

参数：

symbol(list) - 要分组的符号列表。

返回：

sym - 创建的组符号。
返回类型： Symbol

SliceChannel层

mxnet.symbol.SliceChannel(*args,**kwargs)
作用：沿指定轴均等地切割输入

参数：

num_outputs (int，必须) - 要分割的输出数。

axis(int，可选，默认值='1') - 沿切片的尺寸。

squeeze_axis (boolean，可选，default = False) - 如果为true并且切片维度变为1，则挤压该维度。

name (string，可选.) - 结果符号的名称。

返回：

symbol - 结果符号。
返回类型：Symbol

SwapAxis层

mxnet.symbol.SwapAxis(*args,**kwargs)
作用：交换Symbol的两个axis，其作用相当于弱化版的theano的dimshuffle。

参数：

data (Symbol) - 将数据输入到SwapAxisOp。

dim1 (int(非负数)，可选，默认值= 0) - 要交换的第一个轴。

dim2 (int(非负数)，可选，默认值= 0) - 要交换的第二个轴。

name (string，可选.) - 结果符号的名称。

返回：

symbol - 结果符号。
返回类型：Symbol

Upsampling层

mxnet.symbol.UpSampling(*args,**kwargs)
作用：执行最近邻域/双线性采样到输入此功能支持位置输入的可变长度。

参数：

data (Symbol [] ) - 上采样的张量数组

scale (int(非负数)，必须) - 向上抽样标度

num_filter (int(非负数)，可选，默认值= 0) - 输入过滤器。仅由双线性sample_type使用。

sample_type ( {'bilinear'，'nearest'}，必须) - 上采样方法

multi_input_mode ( {'concat'，'sum'}，可选，default ='concat') - 如何处理多个输入。连接意味着沿着通道维度连接上采样图像。求和装置将所有图像相加在一起，仅用于最近邻上采样。

num_args (int，必须) - 要上采样的输入数。对于最近邻上采样，这可以是1-N;输出的大小将是(scale h_0，scale w_0)，所有其他输入将被上采样到相同的大小。对于双线性上采样，这必须是2; 1个输入和1个权重。

workspace(long(非负)，可选，默认值= 512) - 反褶积工作空间(MB)

name (string，可选.) - 结果符号的名称。

返回：

symbol - 结果符号。
返回类型：Symbol

Regularization层

BatchNorm层

mxnet.symbol.BatchNorm(*args, **kwargs) 作用：将批次标准化应用于输入。

参数：

data (Symbol) - 输入数据以进行批次标准化

eps (float，可选，默认值= 0.001) - Epsilon防止div 0

momentum (float，可选，默认值= 0.9) - 移动平均值的动量

fix_gamma (boolean，optional，default = True) - 训练时修正gamma

use_global_stats (boolean，可选，default = False) - 是否使用全局移动统计，而不是使用本地批处理。这将强制改变批处理规范为缩放移位算子。

name (string，optional.) - 结果符号的名称。

返回：

symbol - 结果符号。返回类型：Symbol

Dropout层

mxnet.symbol.Dropout(*args，**kwargs) 作用：对输入应用dropout。在训练期间，输入的每个元素随机地以概率p设置为零。然后整个张量被重新缩放1 /（1-p），以保持期望与应用dropout之前相同。在测试时间期间，这表现为恒等映射。

参数：

data（符号） - 输入数据到dropout。

p（float，optional，default = 0.5） - 在训练时丢弃的比例

name（字符串，可选。） - 结果符号的名称。

返回：

symbol - 结果符号。返回类型：Symbol

参考

1.http://mxnet.io/api/python/symbol.html#mxnet.symbol.Activation 2.https://turi.com/products/create/docs/graphlab.mxnet.html

Win10安装MXNET

安装MXNET

由于公司需要，近期需要快速精通mxnet，接下来的几个星期会陆续更新关于mxnet的笔记，提供参考和备忘。第一篇介绍mxnet的安装，mxnet的安装过程十分蛋疼，个人也是摸索了许久才安装成功，期间也是遇到了各种奇奇怪怪的坑，为了避免新人少走弯路，遂将经验总结于此。

windows上的安装

本人机器配置为Win10 + Cuda 7.5，后续的安装以此为准。

1.mxnet需要VS2013支持C++ 11特性

在Visual C++ Compiler Nov 2013 CTP下载C++ 11版本的编译器，接着将C:\Program Files (x86)\Microsoft Visual C++ Compiler Nov 2013 CTP下所有同名目录中的文件覆盖到C:\Program Files (x86)\Microsoft Visual Studio 12.0\VC下所有同名目录下对应的文件（覆盖前记得备份）

2.从github克隆源码

git clone --recursive https://github.com/dmlc/mxnet 这里提醒注意一定不要忘记--recursive参数，因为mxnet依赖于DMLC通用工具包，--recursive参数可以自动加载mshadow等依赖。

3.用Cmake生成项目工程文件，并编译项目

打开cmake，Where is the source code栏里打开刚才下好的mxnet源代码目录，Where to build the binaries栏里指定生成工程文件和编译结果的路径，这里我填的是C:/mxnet/build，如图所示：

接着点击configure，生成配置。

然后我们点击generate，生成.sln项目文件

找到生成的工程文件mxnet.sln，用vs2013打开

最后，我们在项目mxnet上点击右键->生成，开始编译。

经过漫长的等待后，mxnet终于编译完成。

编译完成后，在C:\mxnet\build\Release目录下会生成三个文件：libmxnet.dll，libmxnet.exp，libmxnet.lib。

4.安装mxnet的python接口接下来我们到mxnet的源代码目录：G:\OpenSource\mxnet\python，运行

python setup.py install

来安装mxnet的python包。

我们将libmxnet.dll 接着，导入mxnet的时候发生了如下的错误：

通过调试发现问题出在打开libmxnet.dll的时候，问题应该出在没有导入依赖的dll文件，但蛋疼的是我也不知道它到底依赖哪一些dll文件。

5.安装依赖

通过一番搜索，我找到一个名为dependency walker的软件，用它打开libmxnet.dll，我们看到还缺少的dll文件有哪些（图中的问号）

这些均能dll在mxnet的release tab下找到，下载完成后将其解压到mxnet的pthon安装目录C:\Anaconda2\Lib\site-packages\mxnet-0.7.0-py2.7.egg\mxnet下。将这些文件放入目录后，我们测试一下能不能导入

import ctypes
ctypes._dlopen(r"C:\Anaconda2\Lib\site-packages\mxnet-0.7.0-py2.7.egg\mxnet\cudart64_75.dll")
ctypes._dlopen(r"C:\Anaconda2\Lib\site-packages\mxnet-0.7.0-py2.7.egg\mxnet\cublas64_75.dll")
ctypes._dlopen(r"cudnn64_5.dll")
ctypes._dlopen(r"C:\Anaconda2\Lib\site-packages\mxnet-0.7.0-py2.7.egg\mxnet\libopenblas.dll")
ctypes._dlopen(r"C:\Anaconda2\Lib\site-packages\mxnet-0.7.0-py2.7.egg\mxnet\opencv_world300.dll")
ctypes._dlopen(r"C:\Anaconda2\Lib\site-packages\mxnet-0.7.0-py2.7.egg\mxnet\opencv_core2413.dll")
ctypes._dlopen(r"vcomp120.dll")
ctypes._dlopen(r"kernel32.dll")
import mxnet as mx
print "mxnet version is:%s"%mx.__version__

mxnet version is:0.7.0

上面的代码中，我们需要手动的载入mxnet依赖的动态链接库才能导入，目前还不清楚为什么它不会自动载入，这个问题留待以后解决，目前可以先把上段代码加入到mxnet的初始化代码中。接着我们跑一跑examples/image-classification/train_mnist这个例子

import ctypes
ctypes._dlopen(r"C:\Anaconda2\Lib\site-packages\mxnet-0.7.0-py2.7.egg\mxnet\cudart64_75.dll")
ctypes._dlopen(r"C:\Anaconda2\Lib\site-packages\mxnet-0.7.0-py2.7.egg\mxnet\cublas64_75.dll")
ctypes._dlopen(r"cudnn64_5.dll")
ctypes._dlopen(r"C:\Anaconda2\Lib\site-packages\mxnet-0.7.0-py2.7.egg\mxnet\libopenblas.dll")
ctypes._dlopen(r"C:\Anaconda2\Lib\site-packages\mxnet-0.7.0-py2.7.egg\mxnet\opencv_world300.dll")
ctypes._dlopen(r"C:\Anaconda2\Lib\site-packages\mxnet-0.7.0-py2.7.egg\mxnet\opencv_core2413.dll")
ctypes._dlopen(r"vcomp120.dll")
ctypes._dlopen(r"kernel32.dll")
import mxnet as mx
import argparse
import os, sys
import logging

def _download(data_dir):
    if not os.path.isdir(data_dir):
        os.system("mkdir " + data_dir)
    os.chdir(data_dir)
    if (not os.path.exists('train-images-idx3-ubyte')) or \
       (not os.path.exists('train-labels-idx1-ubyte')) or \
       (not os.path.exists('t10k-images-idx3-ubyte')) or \
       (not os.path.exists('t10k-labels-idx1-ubyte')):
        os.system("wget http://data.dmlc.ml/mxnet/data/mnist.zip")
        os.system("unzip -u mnist.zip; rm mnist.zip")
    os.chdir("..")

def get_loc(data, attr={'lr_mult':'0.01'}):
    """
    the localisation network in lenet-stn, it will increase acc about more than 1%,
    when num-epoch >=15
    """
    loc = mx.symbol.Convolution(data=data, num_filter=30, kernel=(5, 5), stride=(2,2))
    loc = mx.symbol.Activation(data = loc, act_type='relu')
    loc = mx.symbol.Pooling(data=loc, kernel=(2, 2), stride=(2, 2), pool_type='max')
    loc = mx.symbol.Convolution(data=loc, num_filter=60, kernel=(3, 3), stride=(1,1), pad=(1, 1))
    loc = mx.symbol.Activation(data = loc, act_type='relu')
    loc = mx.symbol.Pooling(data=loc, global_pool=True, kernel=(2, 2), pool_type='avg')
    loc = mx.symbol.Flatten(data=loc)
    loc = mx.symbol.FullyConnected(data=loc, num_hidden=6, name="stn_loc", attr=attr)
    return loc

def get_mlp():
    """
    multi-layer perceptron
    """
    data = mx.symbol.Variable('data')
    fc1  = mx.symbol.FullyConnected(data = data, name='fc1', num_hidden=128)
    act1 = mx.symbol.Activation(data = fc1, name='relu1', act_type="relu")
    fc2  = mx.symbol.FullyConnected(data = act1, name = 'fc2', num_hidden = 64)
    act2 = mx.symbol.Activation(data = fc2, name='relu2', act_type="relu")
    fc3  = mx.symbol.FullyConnected(data = act2, name='fc3', num_hidden=10)
    mlp  = mx.symbol.SoftmaxOutput(data = fc3, name = 'softmax')
    return mlp

def get_lenet(add_stn=False):
    """
    LeCun, Yann, Leon Bottou, Yoshua Bengio, and Patrick
    Haffner. "Gradient-based learning applied to document recognition."
    Proceedings of the IEEE (1998)
    """
    data = mx.symbol.Variable('data')
    if(add_stn):
        data = mx.sym.SpatialTransformer(data=data, loc=get_loc(data), target_shape = (28,28),
                                         transform_type="affine", sampler_type="bilinear")
    # first conv
    conv1 = mx.symbol.Convolution(data=data, kernel=(5,5), num_filter=20)
    tanh1 = mx.symbol.Activation(data=conv1, act_type="tanh")
    pool1 = mx.symbol.Pooling(data=tanh1, pool_type="max",
                              kernel=(2,2), stride=(2,2))
    # second conv
    conv2 = mx.symbol.Convolution(data=pool1, kernel=(5,5), num_filter=50)
    tanh2 = mx.symbol.Activation(data=conv2, act_type="tanh")
    pool2 = mx.symbol.Pooling(data=tanh2, pool_type="max",
                              kernel=(2,2), stride=(2,2))
    # first fullc
    flatten = mx.symbol.Flatten(data=pool2)
    fc1 = mx.symbol.FullyConnected(data=flatten, num_hidden=500)
    tanh3 = mx.symbol.Activation(data=fc1, act_type="tanh")
    # second fullc
    fc2 = mx.symbol.FullyConnected(data=tanh3, num_hidden=10)
    # loss
    lenet = mx.symbol.SoftmaxOutput(data=fc2, name='softmax')
    return lenet

def get_iterator(data_shape):
    def get_iterator_impl(args, kv):
        data_dir = args.data_dir
        """
        if '://' not in args.data_dir:
            _download(args.data_dir)
        """
        flat = False if len(data_shape) == 3 else True

        train           = mx.io.MNISTIter(
            image       = data_dir + "train-images-idx3-ubyte",
            label       = data_dir + "train-labels-idx1-ubyte",
            input_shape = data_shape,
            batch_size  = args.batch_size,
            shuffle     = True,
            flat        = flat,
            num_parts   = kv.num_workers,
            part_index  = kv.rank)

        val = mx.io.MNISTIter(
            image       = data_dir + "t10k-images-idx3-ubyte",
            label       = data_dir + "t10k-labels-idx1-ubyte",
            input_shape = data_shape,
            batch_size  = args.batch_size,
            flat        = flat,
            num_parts   = kv.num_workers,
            part_index  = kv.rank)

        return (train, val)
    return get_iterator_impl

def parse_args():
    parser = argparse.ArgumentParser(description='train an image classifer on mnist')
    parser.add_argument('--network', type=str, default='mlp',
                        choices = ['mlp', 'lenet', 'lenet-stn'],
                        help = 'the cnn to use')
    parser.add_argument('--data-dir', type=str, default='mnist/',
                        help='the input data directory')
    parser.add_argument('--gpus', type=str,
                        help='the gpus will be used, e.g "0,1,2,3"')
    parser.add_argument('--num-examples', type=int, default=60000,
                        help='the number of training examples')
    parser.add_argument('--batch-size', type=int, default=128,
                        help='the batch size')
    parser.add_argument('--lr', type=float, default=.1,
                        help='the initial learning rate')
    parser.add_argument('--model-prefix', type=str,
                        help='the prefix of the model to load/save')
    parser.add_argument('--save-model-prefix', type=str,
                        help='the prefix of the model to save')
    parser.add_argument('--num-epochs', type=int, default=10,
                        help='the number of training epochs')
    parser.add_argument('--load-epoch', type=int,
                        help="load the model on an epoch using the model-prefix")
    parser.add_argument('--kv-store', type=str, default='local',
                        help='the kvstore type')
    parser.add_argument('--lr-factor', type=float, default=1,
                        help='times the lr with a factor for every lr-factor-epoch epoch')
    parser.add_argument('--lr-factor-epoch', type=float, default=1,
                        help='the number of epoch to factor the lr, could be .5')
    return parser.parse_args(['--gpus', '0', '--data-dir', 'G:/OpenSource/mxnet/example/image-classification/mnist/'])

if __name__ == '__main__':
    args = parse_args()

    if args.network == 'mlp':
        data_shape = (784, )
        net = get_mlp()
    elif args.network == 'lenet-stn':
        data_shape = (1, 28, 28)
        net = get_lenet(True)
    else:
        data_shape = (1, 28, 28)
        net = get_lenet()

    # kvstore
    kv = mx.kvstore.create(args.kv_store)

    # logging
    head = '%(asctime)-15s Node[' + str(kv.rank) + '] %(message)s'
    
    logger = logging.getLogger()
    formatter = logging.Formatter(head)
    stdout_handler = logging.StreamHandler(sys.stdout)
    stdout_handler.setFormatter(formatter)

    logger.addHandler(stdout_handler)

    logger.setLevel(logging.INFO)
    logger.info('start with arguments %s', args)
    
    # load model
    model_prefix = args.model_prefix
    if model_prefix is not None:
        model_prefix += "-%d" % (kv.rank)
    model_args = {}
    if args.load_epoch is not None:
        assert model_prefix is not None
        tmp = mx.model.FeedForward.load(model_prefix, args.load_epoch)
        model_args = {'arg_params' : tmp.arg_params,
                      'aux_params' : tmp.aux_params,
                      'begin_epoch' : args.load_epoch}
    # save model
    save_model_prefix = args.save_model_prefix
    if save_model_prefix is None:
        save_model_prefix = model_prefix
    checkpoint = None if save_model_prefix is None else mx.callback.do_checkpoint(save_model_prefix)

    # data
    (train, val) = get_iterator(data_shape)(args, kv)

    # train
    devs = mx.cpu() if args.gpus is None else [
        mx.gpu(int(i)) for i in args.gpus.split(',')]

    epoch_size = args.num_examples / args.batch_size

    if args.kv_store == 'dist_sync':
        epoch_size /= kv.num_workers
        model_args['epoch_size'] = epoch_size

    if 'lr_factor' in args and args.lr_factor < 1:
        model_args['lr_scheduler'] = mx.lr_scheduler.FactorScheduler(
            step = max(int(epoch_size * args.lr_factor_epoch), 1),
            factor = args.lr_factor)

    if 'clip_gradient' in args and args.clip_gradient is not None:
        model_args['clip_gradient'] = args.clip_gradient

    # disable kvstore for single device
    if 'local' in kv.type and (
            args.gpus is None or len(args.gpus.split(',')) is 1):
        kv = None

    model = mx.model.FeedForward(
        ctx                = devs,
        symbol             = net,
        num_epoch          = args.num_epochs,
        learning_rate      = args.lr,
        momentum           = 0.9,
        wd                 = 0.00001,
        initializer        = mx.init.Xavier(factor_type="in", magnitude=2.34),
        **model_args)

    eval_metrics = ['accuracy']
    ## TopKAccuracy only allows top_k > 1
    for top_k in [5, 10, 20]:
        eval_metrics.append(mx.metric.create('top_k_accuracy', top_k = top_k))

    model.fit(
        X                  = train,
        eval_data          = val,
        eval_metric        = eval_metrics,
        kvstore            = kv,
        batch_end_callback = [mx.callback.Speedometer(args.batch_size, 50)],
        epoch_end_callback = checkpoint)

INFO:root:start with arguments Namespace(batch_size=128, data_dir='G:/OpenSource/mxnet/example/image-classification/mnist/', gpus='0', kv_store='local', load_epoch=None, lr=0.1, lr_factor=1, lr_factor_epoch=1, model_prefix=None, network='mlp', num_epochs=10, num_examples=60000, save_model_prefix=None)


2016-10-26 19:37:38,994 Node[0] start with arguments Namespace(batch_size=128, data_dir='G:/OpenSource/mxnet/example/image-classification/mnist/', gpus='0', kv_store='local', load_epoch=None, lr=0.1, lr_factor=1, lr_factor_epoch=1, model_prefix=None, network='mlp', num_epochs=10, num_examples=60000, save_model_prefix=None)


INFO:root:Start training with [gpu(0)]


2016-10-26 19:37:42,038 Node[0] Start training with [gpu(0)]


INFO:root:Epoch[0] Batch [50]   Speed: 23104.70 samples/sec Train-accuracy=0.687344


2016-10-26 19:37:45,351 Node[0] Epoch[0] Batch [50] Speed: 23104.70 samples/sec Train-accuracy=0.687344


INFO:root:Epoch[0] Batch [50]   Speed: 23104.70 samples/sec Train-top_k_accuracy_5=0.935937


2016-10-26 19:37:45,354 Node[0] Epoch[0] Batch [50] Speed: 23104.70 samples/sec Train-top_k_accuracy_5=0.935937


INFO:root:Epoch[0] Batch [50]   Speed: 23104.70 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:37:45,357 Node[0] Epoch[0] Batch [50] Speed: 23104.70 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[0] Batch [50]   Speed: 23104.70 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:37:45,361 Node[0] Epoch[0] Batch [50] Speed: 23104.70 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[0] Batch [100]  Speed: 22535.22 samples/sec Train-accuracy=0.897188


2016-10-26 19:37:45,648 Node[0] Epoch[0] Batch [100]    Speed: 22535.22 samples/sec Train-accuracy=0.897188


INFO:root:Epoch[0] Batch [100]  Speed: 22535.22 samples/sec Train-top_k_accuracy_5=0.992812


2016-10-26 19:37:45,650 Node[0] Epoch[0] Batch [100]    Speed: 22535.22 samples/sec Train-top_k_accuracy_5=0.992812


INFO:root:Epoch[0] Batch [100]  Speed: 22535.22 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:37:45,651 Node[0] Epoch[0] Batch [100]    Speed: 22535.22 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[0] Batch [100]  Speed: 22535.22 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:37:45,654 Node[0] Epoch[0] Batch [100]    Speed: 22535.22 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[0] Batch [150]  Speed: 23443.22 samples/sec Train-accuracy=0.919687


2016-10-26 19:37:45,930 Node[0] Epoch[0] Batch [150]    Speed: 23443.22 samples/sec Train-accuracy=0.919687


INFO:root:Epoch[0] Batch [150]  Speed: 23443.22 samples/sec Train-top_k_accuracy_5=0.995469


2016-10-26 19:37:45,930 Node[0] Epoch[0] Batch [150]    Speed: 23443.22 samples/sec Train-top_k_accuracy_5=0.995469


INFO:root:Epoch[0] Batch [150]  Speed: 23443.22 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:37:45,933 Node[0] Epoch[0] Batch [150]    Speed: 23443.22 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[0] Batch [150]  Speed: 23443.22 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:37:45,934 Node[0] Epoch[0] Batch [150]    Speed: 23443.22 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[0] Batch [200]  Speed: 24150.96 samples/sec Train-accuracy=0.927656


2016-10-26 19:37:46,200 Node[0] Epoch[0] Batch [200]    Speed: 24150.96 samples/sec Train-accuracy=0.927656


INFO:root:Epoch[0] Batch [200]  Speed: 24150.96 samples/sec Train-top_k_accuracy_5=0.997031


2016-10-26 19:37:46,203 Node[0] Epoch[0] Batch [200]    Speed: 24150.96 samples/sec Train-top_k_accuracy_5=0.997031


INFO:root:Epoch[0] Batch [200]  Speed: 24150.96 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:37:46,206 Node[0] Epoch[0] Batch [200]    Speed: 24150.96 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[0] Batch [200]  Speed: 24150.96 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:37:46,210 Node[0] Epoch[0] Batch [200]    Speed: 24150.96 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[0] Batch [250]  Speed: 22145.33 samples/sec Train-accuracy=0.942031


2016-10-26 19:37:46,502 Node[0] Epoch[0] Batch [250]    Speed: 22145.33 samples/sec Train-accuracy=0.942031


INFO:root:Epoch[0] Batch [250]  Speed: 22145.33 samples/sec Train-top_k_accuracy_5=0.996875


2016-10-26 19:37:46,503 Node[0] Epoch[0] Batch [250]    Speed: 22145.33 samples/sec Train-top_k_accuracy_5=0.996875


INFO:root:Epoch[0] Batch [250]  Speed: 22145.33 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:37:46,509 Node[0] Epoch[0] Batch [250]    Speed: 22145.33 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[0] Batch [250]  Speed: 22145.33 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:37:46,513 Node[0] Epoch[0] Batch [250]    Speed: 22145.33 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[0] Batch [300]  Speed: 25600.00 samples/sec Train-accuracy=0.940781


2016-10-26 19:37:46,766 Node[0] Epoch[0] Batch [300]    Speed: 25600.00 samples/sec Train-accuracy=0.940781


INFO:root:Epoch[0] Batch [300]  Speed: 25600.00 samples/sec Train-top_k_accuracy_5=0.997656


2016-10-26 19:37:46,767 Node[0] Epoch[0] Batch [300]    Speed: 25600.00 samples/sec Train-top_k_accuracy_5=0.997656


INFO:root:Epoch[0] Batch [300]  Speed: 25600.00 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:37:46,769 Node[0] Epoch[0] Batch [300]    Speed: 25600.00 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[0] Batch [300]  Speed: 25600.00 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:37:46,770 Node[0] Epoch[0] Batch [300]    Speed: 25600.00 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[0] Batch [350]  Speed: 25497.99 samples/sec Train-accuracy=0.943750


2016-10-26 19:37:47,025 Node[0] Epoch[0] Batch [350]    Speed: 25497.99 samples/sec Train-accuracy=0.943750


INFO:root:Epoch[0] Batch [350]  Speed: 25497.99 samples/sec Train-top_k_accuracy_5=0.998594


2016-10-26 19:37:47,026 Node[0] Epoch[0] Batch [350]    Speed: 25497.99 samples/sec Train-top_k_accuracy_5=0.998594


INFO:root:Epoch[0] Batch [350]  Speed: 25497.99 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:37:47,028 Node[0] Epoch[0] Batch [350]    Speed: 25497.99 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[0] Batch [350]  Speed: 25497.99 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:37:47,029 Node[0] Epoch[0] Batch [350]    Speed: 25497.99 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[0] Batch [400]  Speed: 23970.04 samples/sec Train-accuracy=0.952344


2016-10-26 19:37:47,296 Node[0] Epoch[0] Batch [400]    Speed: 23970.04 samples/sec Train-accuracy=0.952344


INFO:root:Epoch[0] Batch [400]  Speed: 23970.04 samples/sec Train-top_k_accuracy_5=0.998594


2016-10-26 19:37:47,298 Node[0] Epoch[0] Batch [400]    Speed: 23970.04 samples/sec Train-top_k_accuracy_5=0.998594


INFO:root:Epoch[0] Batch [400]  Speed: 23970.04 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:37:47,299 Node[0] Epoch[0] Batch [400]    Speed: 23970.04 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[0] Batch [400]  Speed: 23970.04 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:37:47,302 Node[0] Epoch[0] Batch [400]    Speed: 23970.04 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[0] Batch [450]  Speed: 27350.40 samples/sec Train-accuracy=0.952969


2016-10-26 19:37:47,539 Node[0] Epoch[0] Batch [450]    Speed: 27350.40 samples/sec Train-accuracy=0.952969


INFO:root:Epoch[0] Batch [450]  Speed: 27350.40 samples/sec Train-top_k_accuracy_5=0.998906


2016-10-26 19:37:47,542 Node[0] Epoch[0] Batch [450]    Speed: 27350.40 samples/sec Train-top_k_accuracy_5=0.998906


INFO:root:Epoch[0] Batch [450]  Speed: 27350.40 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:37:47,546 Node[0] Epoch[0] Batch [450]    Speed: 27350.40 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[0] Batch [450]  Speed: 27350.40 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:37:47,548 Node[0] Epoch[0] Batch [450]    Speed: 27350.40 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[0] Resetting Data Iterator


2016-10-26 19:37:47,634 Node[0] Epoch[0] Resetting Data Iterator


INFO:root:Epoch[0] Time cost=3.100


2016-10-26 19:37:47,637 Node[0] Epoch[0] Time cost=3.100


INFO:root:Epoch[0] Validation-accuracy=0.960036


2016-10-26 19:37:47,826 Node[0] Epoch[0] Validation-accuracy=0.960036


INFO:root:Epoch[0] Validation-top_k_accuracy_5=0.998698


2016-10-26 19:37:47,828 Node[0] Epoch[0] Validation-top_k_accuracy_5=0.998698


INFO:root:Epoch[0] Validation-top_k_accuracy_10=1.000000


2016-10-26 19:37:47,829 Node[0] Epoch[0] Validation-top_k_accuracy_10=1.000000


INFO:root:Epoch[0] Validation-top_k_accuracy_20=1.000000


2016-10-26 19:37:47,832 Node[0] Epoch[0] Validation-top_k_accuracy_20=1.000000


INFO:root:Epoch[1] Batch [50]   Speed: 25806.46 samples/sec Train-accuracy=0.955156


2016-10-26 19:37:48,085 Node[0] Epoch[1] Batch [50] Speed: 25806.46 samples/sec Train-accuracy=0.955156


INFO:root:Epoch[1] Batch [50]   Speed: 25806.46 samples/sec Train-top_k_accuracy_5=0.998594


2016-10-26 19:37:48,088 Node[0] Epoch[1] Batch [50] Speed: 25806.46 samples/sec Train-top_k_accuracy_5=0.998594


INFO:root:Epoch[1] Batch [50]   Speed: 25806.46 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:37:48,091 Node[0] Epoch[1] Batch [50] Speed: 25806.46 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[1] Batch [50]   Speed: 25806.46 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:37:48,095 Node[0] Epoch[1] Batch [50] Speed: 25806.46 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[1] Batch [100]  Speed: 27004.19 samples/sec Train-accuracy=0.957969


2016-10-26 19:37:48,334 Node[0] Epoch[1] Batch [100]    Speed: 27004.19 samples/sec Train-accuracy=0.957969


INFO:root:Epoch[1] Batch [100]  Speed: 27004.19 samples/sec Train-top_k_accuracy_5=0.998281


2016-10-26 19:37:48,335 Node[0] Epoch[1] Batch [100]    Speed: 27004.19 samples/sec Train-top_k_accuracy_5=0.998281


INFO:root:Epoch[1] Batch [100]  Speed: 27004.19 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:37:48,336 Node[0] Epoch[1] Batch [100]    Speed: 27004.19 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[1] Batch [100]  Speed: 27004.19 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:37:48,338 Node[0] Epoch[1] Batch [100]    Speed: 27004.19 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[1] Batch [150]  Speed: 23443.22 samples/sec Train-accuracy=0.962969


2016-10-26 19:37:48,612 Node[0] Epoch[1] Batch [150]    Speed: 23443.22 samples/sec Train-accuracy=0.962969


INFO:root:Epoch[1] Batch [150]  Speed: 23443.22 samples/sec Train-top_k_accuracy_5=0.999062


2016-10-26 19:37:48,615 Node[0] Epoch[1] Batch [150]    Speed: 23443.22 samples/sec Train-top_k_accuracy_5=0.999062


INFO:root:Epoch[1] Batch [150]  Speed: 23443.22 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:37:48,618 Node[0] Epoch[1] Batch [150]    Speed: 23443.22 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[1] Batch [150]  Speed: 23443.22 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:37:48,619 Node[0] Epoch[1] Batch [150]    Speed: 23443.22 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[1] Batch [200]  Speed: 26446.27 samples/sec Train-accuracy=0.964688


2016-10-26 19:37:48,864 Node[0] Epoch[1] Batch [200]    Speed: 26446.27 samples/sec Train-accuracy=0.964688


INFO:root:Epoch[1] Batch [200]  Speed: 26446.27 samples/sec Train-top_k_accuracy_5=0.999531


2016-10-26 19:37:48,865 Node[0] Epoch[1] Batch [200]    Speed: 26446.27 samples/sec Train-top_k_accuracy_5=0.999531


INFO:root:Epoch[1] Batch [200]  Speed: 26446.27 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:37:48,867 Node[0] Epoch[1] Batch [200]    Speed: 26446.27 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[1] Batch [200]  Speed: 26446.27 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:37:48,868 Node[0] Epoch[1] Batch [200]    Speed: 26446.27 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[1] Batch [250]  Speed: 28193.83 samples/sec Train-accuracy=0.967656


2016-10-26 19:37:49,096 Node[0] Epoch[1] Batch [250]    Speed: 28193.83 samples/sec Train-accuracy=0.967656


INFO:root:Epoch[1] Batch [250]  Speed: 28193.83 samples/sec Train-top_k_accuracy_5=0.998906


2016-10-26 19:37:49,098 Node[0] Epoch[1] Batch [250]    Speed: 28193.83 samples/sec Train-top_k_accuracy_5=0.998906


INFO:root:Epoch[1] Batch [250]  Speed: 28193.83 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:37:49,101 Node[0] Epoch[1] Batch [250]    Speed: 28193.83 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[1] Batch [250]  Speed: 28193.83 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:37:49,102 Node[0] Epoch[1] Batch [250]    Speed: 28193.83 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[1] Batch [300]  Speed: 24806.19 samples/sec Train-accuracy=0.962656


2016-10-26 19:37:49,364 Node[0] Epoch[1] Batch [300]    Speed: 24806.19 samples/sec Train-accuracy=0.962656


INFO:root:Epoch[1] Batch [300]  Speed: 24806.19 samples/sec Train-top_k_accuracy_5=0.999687


2016-10-26 19:37:49,365 Node[0] Epoch[1] Batch [300]    Speed: 24806.19 samples/sec Train-top_k_accuracy_5=0.999687


INFO:root:Epoch[1] Batch [300]  Speed: 24806.19 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:37:49,368 Node[0] Epoch[1] Batch [300]    Speed: 24806.19 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[1] Batch [300]  Speed: 24806.19 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:37:49,369 Node[0] Epoch[1] Batch [300]    Speed: 24806.19 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[1] Batch [350]  Speed: 27004.22 samples/sec Train-accuracy=0.966719


2016-10-26 19:37:49,608 Node[0] Epoch[1] Batch [350]    Speed: 27004.22 samples/sec Train-accuracy=0.966719


INFO:root:Epoch[1] Batch [350]  Speed: 27004.22 samples/sec Train-top_k_accuracy_5=0.999062


2016-10-26 19:37:49,609 Node[0] Epoch[1] Batch [350]    Speed: 27004.22 samples/sec Train-top_k_accuracy_5=0.999062


INFO:root:Epoch[1] Batch [350]  Speed: 27004.22 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:37:49,611 Node[0] Epoch[1] Batch [350]    Speed: 27004.22 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[1] Batch [350]  Speed: 27004.22 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:37:49,612 Node[0] Epoch[1] Batch [350]    Speed: 27004.22 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[1] Batch [400]  Speed: 27586.22 samples/sec Train-accuracy=0.970313


2016-10-26 19:37:49,846 Node[0] Epoch[1] Batch [400]    Speed: 27586.22 samples/sec Train-accuracy=0.970313


INFO:root:Epoch[1] Batch [400]  Speed: 27586.22 samples/sec Train-top_k_accuracy_5=0.999219


2016-10-26 19:37:49,848 Node[0] Epoch[1] Batch [400]    Speed: 27586.22 samples/sec Train-top_k_accuracy_5=0.999219


INFO:root:Epoch[1] Batch [400]  Speed: 27586.22 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:37:49,851 Node[0] Epoch[1] Batch [400]    Speed: 27586.22 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[1] Batch [400]  Speed: 27586.22 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:37:49,852 Node[0] Epoch[1] Batch [400]    Speed: 27586.22 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[1] Batch [450]  Speed: 26229.51 samples/sec Train-accuracy=0.969531


2016-10-26 19:37:50,099 Node[0] Epoch[1] Batch [450]    Speed: 26229.51 samples/sec Train-accuracy=0.969531


INFO:root:Epoch[1] Batch [450]  Speed: 26229.51 samples/sec Train-top_k_accuracy_5=0.999844


2016-10-26 19:37:50,101 Node[0] Epoch[1] Batch [450]    Speed: 26229.51 samples/sec Train-top_k_accuracy_5=0.999844


INFO:root:Epoch[1] Batch [450]  Speed: 26229.51 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:37:50,105 Node[0] Epoch[1] Batch [450]    Speed: 26229.51 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[1] Batch [450]  Speed: 26229.51 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:37:50,109 Node[0] Epoch[1] Batch [450]    Speed: 26229.51 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[1] Resetting Data Iterator


2016-10-26 19:37:50,196 Node[0] Epoch[1] Resetting Data Iterator


INFO:root:Epoch[1] Time cost=2.364


2016-10-26 19:37:50,197 Node[0] Epoch[1] Time cost=2.364


INFO:root:Epoch[1] Validation-accuracy=0.968349


2016-10-26 19:37:50,381 Node[0] Epoch[1] Validation-accuracy=0.968349


INFO:root:Epoch[1] Validation-top_k_accuracy_5=0.999099


2016-10-26 19:37:50,382 Node[0] Epoch[1] Validation-top_k_accuracy_5=0.999099


INFO:root:Epoch[1] Validation-top_k_accuracy_10=1.000000


2016-10-26 19:37:50,384 Node[0] Epoch[1] Validation-top_k_accuracy_10=1.000000


INFO:root:Epoch[1] Validation-top_k_accuracy_20=1.000000


2016-10-26 19:37:50,385 Node[0] Epoch[1] Validation-top_k_accuracy_20=1.000000


INFO:root:Epoch[2] Batch [50]   Speed: 27004.22 samples/sec Train-accuracy=0.971875


2016-10-26 19:37:50,635 Node[0] Epoch[2] Batch [50] Speed: 27004.22 samples/sec Train-accuracy=0.971875


INFO:root:Epoch[2] Batch [50]   Speed: 27004.22 samples/sec Train-top_k_accuracy_5=0.999219


2016-10-26 19:37:50,638 Node[0] Epoch[2] Batch [50] Speed: 27004.22 samples/sec Train-top_k_accuracy_5=0.999219


INFO:root:Epoch[2] Batch [50]   Speed: 27004.22 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:37:50,644 Node[0] Epoch[2] Batch [50] Speed: 27004.22 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[2] Batch [50]   Speed: 27004.22 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:37:50,644 Node[0] Epoch[2] Batch [50] Speed: 27004.22 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[2] Batch [100]  Speed: 27234.05 samples/sec Train-accuracy=0.971250


2016-10-26 19:37:50,881 Node[0] Epoch[2] Batch [100]    Speed: 27234.05 samples/sec Train-accuracy=0.971250


INFO:root:Epoch[2] Batch [100]  Speed: 27234.05 samples/sec Train-top_k_accuracy_5=0.999531


2016-10-26 19:37:50,882 Node[0] Epoch[2] Batch [100]    Speed: 27234.05 samples/sec Train-top_k_accuracy_5=0.999531


INFO:root:Epoch[2] Batch [100]  Speed: 27234.05 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:37:50,884 Node[0] Epoch[2] Batch [100]    Speed: 27234.05 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[2] Batch [100]  Speed: 27234.05 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:37:50,887 Node[0] Epoch[2] Batch [100]    Speed: 27234.05 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[2] Batch [150]  Speed: 26778.23 samples/sec Train-accuracy=0.970625


2016-10-26 19:37:51,127 Node[0] Epoch[2] Batch [150]    Speed: 26778.23 samples/sec Train-accuracy=0.970625


INFO:root:Epoch[2] Batch [150]  Speed: 26778.23 samples/sec Train-top_k_accuracy_5=0.999531


2016-10-26 19:37:51,128 Node[0] Epoch[2] Batch [150]    Speed: 26778.23 samples/sec Train-top_k_accuracy_5=0.999531


INFO:root:Epoch[2] Batch [150]  Speed: 26778.23 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:37:51,130 Node[0] Epoch[2] Batch [150]    Speed: 26778.23 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[2] Batch [150]  Speed: 26778.23 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:37:51,131 Node[0] Epoch[2] Batch [150]    Speed: 26778.23 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[2] Batch [200]  Speed: 27705.61 samples/sec Train-accuracy=0.974844


2016-10-26 19:37:51,364 Node[0] Epoch[2] Batch [200]    Speed: 27705.61 samples/sec Train-accuracy=0.974844


INFO:root:Epoch[2] Batch [200]  Speed: 27705.61 samples/sec Train-top_k_accuracy_5=0.999844


2016-10-26 19:37:51,367 Node[0] Epoch[2] Batch [200]    Speed: 27705.61 samples/sec Train-top_k_accuracy_5=0.999844


INFO:root:Epoch[2] Batch [200]  Speed: 27705.61 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:37:51,368 Node[0] Epoch[2] Batch [200]    Speed: 27705.61 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[2] Batch [200]  Speed: 27705.61 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:37:51,371 Node[0] Epoch[2] Batch [200]    Speed: 27705.61 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[2] Batch [250]  Speed: 24902.73 samples/sec Train-accuracy=0.975781


2016-10-26 19:37:51,631 Node[0] Epoch[2] Batch [250]    Speed: 24902.73 samples/sec Train-accuracy=0.975781


INFO:root:Epoch[2] Batch [250]  Speed: 24902.73 samples/sec Train-top_k_accuracy_5=0.999844


2016-10-26 19:37:51,631 Node[0] Epoch[2] Batch [250]    Speed: 24902.73 samples/sec Train-top_k_accuracy_5=0.999844


INFO:root:Epoch[2] Batch [250]  Speed: 24902.73 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:37:51,634 Node[0] Epoch[2] Batch [250]    Speed: 24902.73 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[2] Batch [250]  Speed: 24902.73 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:37:51,635 Node[0] Epoch[2] Batch [250]    Speed: 24902.73 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[2] Batch [300]  Speed: 25600.00 samples/sec Train-accuracy=0.973125


2016-10-26 19:37:51,887 Node[0] Epoch[2] Batch [300]    Speed: 25600.00 samples/sec Train-accuracy=0.973125


INFO:root:Epoch[2] Batch [300]  Speed: 25600.00 samples/sec Train-top_k_accuracy_5=0.999844


2016-10-26 19:37:51,890 Node[0] Epoch[2] Batch [300]    Speed: 25600.00 samples/sec Train-top_k_accuracy_5=0.999844


INFO:root:Epoch[2] Batch [300]  Speed: 25600.00 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:37:51,894 Node[0] Epoch[2] Batch [300]    Speed: 25600.00 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[2] Batch [300]  Speed: 25600.00 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:37:51,895 Node[0] Epoch[2] Batch [300]    Speed: 25600.00 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[2] Batch [350]  Speed: 27118.64 samples/sec Train-accuracy=0.975781


2016-10-26 19:37:52,134 Node[0] Epoch[2] Batch [350]    Speed: 27118.64 samples/sec Train-accuracy=0.975781


INFO:root:Epoch[2] Batch [350]  Speed: 27118.64 samples/sec Train-top_k_accuracy_5=0.999531


2016-10-26 19:37:52,137 Node[0] Epoch[2] Batch [350]    Speed: 27118.64 samples/sec Train-top_k_accuracy_5=0.999531


INFO:root:Epoch[2] Batch [350]  Speed: 27118.64 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:37:52,140 Node[0] Epoch[2] Batch [350]    Speed: 27118.64 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[2] Batch [350]  Speed: 27118.64 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:37:52,141 Node[0] Epoch[2] Batch [350]    Speed: 27118.64 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[2] Batch [400]  Speed: 26337.45 samples/sec Train-accuracy=0.977969


2016-10-26 19:37:52,385 Node[0] Epoch[2] Batch [400]    Speed: 26337.45 samples/sec Train-accuracy=0.977969


INFO:root:Epoch[2] Batch [400]  Speed: 26337.45 samples/sec Train-top_k_accuracy_5=0.999219


2016-10-26 19:37:52,387 Node[0] Epoch[2] Batch [400]    Speed: 26337.45 samples/sec Train-top_k_accuracy_5=0.999219


INFO:root:Epoch[2] Batch [400]  Speed: 26337.45 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:37:52,390 Node[0] Epoch[2] Batch [400]    Speed: 26337.45 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[2] Batch [400]  Speed: 26337.45 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:37:52,391 Node[0] Epoch[2] Batch [400]    Speed: 26337.45 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[2] Batch [450]  Speed: 25196.83 samples/sec Train-accuracy=0.977187


2016-10-26 19:37:52,647 Node[0] Epoch[2] Batch [450]    Speed: 25196.83 samples/sec Train-accuracy=0.977187


INFO:root:Epoch[2] Batch [450]  Speed: 25196.83 samples/sec Train-top_k_accuracy_5=0.999687


2016-10-26 19:37:52,648 Node[0] Epoch[2] Batch [450]    Speed: 25196.83 samples/sec Train-top_k_accuracy_5=0.999687


INFO:root:Epoch[2] Batch [450]  Speed: 25196.83 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:37:52,650 Node[0] Epoch[2] Batch [450]    Speed: 25196.83 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[2] Batch [450]  Speed: 25196.83 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:37:52,651 Node[0] Epoch[2] Batch [450]    Speed: 25196.83 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[2] Resetting Data Iterator


2016-10-26 19:37:52,736 Node[0] Epoch[2] Resetting Data Iterator


INFO:root:Epoch[2] Time cost=2.345


2016-10-26 19:37:52,737 Node[0] Epoch[2] Time cost=2.345


INFO:root:Epoch[2] Validation-accuracy=0.973858


2016-10-26 19:37:52,903 Node[0] Epoch[2] Validation-accuracy=0.973858


INFO:root:Epoch[2] Validation-top_k_accuracy_5=0.999099


2016-10-26 19:37:52,905 Node[0] Epoch[2] Validation-top_k_accuracy_5=0.999099


INFO:root:Epoch[2] Validation-top_k_accuracy_10=1.000000


2016-10-26 19:37:52,907 Node[0] Epoch[2] Validation-top_k_accuracy_10=1.000000


INFO:root:Epoch[2] Validation-top_k_accuracy_20=1.000000


2016-10-26 19:37:52,910 Node[0] Epoch[2] Validation-top_k_accuracy_20=1.000000


INFO:root:Epoch[3] Batch [50]   Speed: 27705.61 samples/sec Train-accuracy=0.977969


2016-10-26 19:37:53,147 Node[0] Epoch[3] Batch [50] Speed: 27705.61 samples/sec Train-accuracy=0.977969


INFO:root:Epoch[3] Batch [50]   Speed: 27705.61 samples/sec Train-top_k_accuracy_5=0.999844


2016-10-26 19:37:53,148 Node[0] Epoch[3] Batch [50] Speed: 27705.61 samples/sec Train-top_k_accuracy_5=0.999844


INFO:root:Epoch[3] Batch [50]   Speed: 27705.61 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:37:53,151 Node[0] Epoch[3] Batch [50] Speed: 27705.61 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[3] Batch [50]   Speed: 27705.61 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:37:53,151 Node[0] Epoch[3] Batch [50] Speed: 27705.61 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[3] Batch [100]  Speed: 26446.30 samples/sec Train-accuracy=0.978281


2016-10-26 19:37:53,395 Node[0] Epoch[3] Batch [100]    Speed: 26446.30 samples/sec Train-accuracy=0.978281


INFO:root:Epoch[3] Batch [100]  Speed: 26446.30 samples/sec Train-top_k_accuracy_5=0.999844


2016-10-26 19:37:53,398 Node[0] Epoch[3] Batch [100]    Speed: 26446.30 samples/sec Train-top_k_accuracy_5=0.999844


INFO:root:Epoch[3] Batch [100]  Speed: 26446.30 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:37:53,400 Node[0] Epoch[3] Batch [100]    Speed: 26446.30 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[3] Batch [100]  Speed: 26446.30 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:37:53,401 Node[0] Epoch[3] Batch [100]    Speed: 26446.30 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[3] Batch [150]  Speed: 26666.69 samples/sec Train-accuracy=0.977969


2016-10-26 19:37:53,642 Node[0] Epoch[3] Batch [150]    Speed: 26666.69 samples/sec Train-accuracy=0.977969


INFO:root:Epoch[3] Batch [150]  Speed: 26666.69 samples/sec Train-top_k_accuracy_5=0.999687


2016-10-26 19:37:53,644 Node[0] Epoch[3] Batch [150]    Speed: 26666.69 samples/sec Train-top_k_accuracy_5=0.999687


INFO:root:Epoch[3] Batch [150]  Speed: 26666.69 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:37:53,647 Node[0] Epoch[3] Batch [150]    Speed: 26666.69 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[3] Batch [150]  Speed: 26666.69 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:37:53,648 Node[0] Epoch[3] Batch [150]    Speed: 26666.69 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[3] Batch [200]  Speed: 26229.51 samples/sec Train-accuracy=0.980781


2016-10-26 19:37:53,894 Node[0] Epoch[3] Batch [200]    Speed: 26229.51 samples/sec Train-accuracy=0.980781


INFO:root:Epoch[3] Batch [200]  Speed: 26229.51 samples/sec Train-top_k_accuracy_5=1.000000


2016-10-26 19:37:53,895 Node[0] Epoch[3] Batch [200]    Speed: 26229.51 samples/sec Train-top_k_accuracy_5=1.000000


INFO:root:Epoch[3] Batch [200]  Speed: 26229.51 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:37:53,898 Node[0] Epoch[3] Batch [200]    Speed: 26229.51 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[3] Batch [200]  Speed: 26229.51 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:37:53,900 Node[0] Epoch[3] Batch [200]    Speed: 26229.51 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[3] Batch [250]  Speed: 26890.77 samples/sec Train-accuracy=0.979531


2016-10-26 19:37:54,141 Node[0] Epoch[3] Batch [250]    Speed: 26890.77 samples/sec Train-accuracy=0.979531


INFO:root:Epoch[3] Batch [250]  Speed: 26890.77 samples/sec Train-top_k_accuracy_5=0.999687


2016-10-26 19:37:54,142 Node[0] Epoch[3] Batch [250]    Speed: 26890.77 samples/sec Train-top_k_accuracy_5=0.999687


INFO:root:Epoch[3] Batch [250]  Speed: 26890.77 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:37:54,144 Node[0] Epoch[3] Batch [250]    Speed: 26890.77 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[3] Batch [250]  Speed: 26890.77 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:37:54,145 Node[0] Epoch[3] Batch [250]    Speed: 26890.77 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[3] Batch [300]  Speed: 27586.22 samples/sec Train-accuracy=0.979844


2016-10-26 19:37:54,378 Node[0] Epoch[3] Batch [300]    Speed: 27586.22 samples/sec Train-accuracy=0.979844


INFO:root:Epoch[3] Batch [300]  Speed: 27586.22 samples/sec Train-top_k_accuracy_5=0.999844


2016-10-26 19:37:54,380 Node[0] Epoch[3] Batch [300]    Speed: 27586.22 samples/sec Train-top_k_accuracy_5=0.999844


INFO:root:Epoch[3] Batch [300]  Speed: 27586.22 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:37:54,381 Node[0] Epoch[3] Batch [300]    Speed: 27586.22 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[3] Batch [300]  Speed: 27586.22 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:37:54,384 Node[0] Epoch[3] Batch [300]    Speed: 27586.22 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[3] Batch [350]  Speed: 27467.81 samples/sec Train-accuracy=0.979375


2016-10-26 19:37:54,618 Node[0] Epoch[3] Batch [350]    Speed: 27467.81 samples/sec Train-accuracy=0.979375


INFO:root:Epoch[3] Batch [350]  Speed: 27467.81 samples/sec Train-top_k_accuracy_5=0.999531


2016-10-26 19:37:54,621 Node[0] Epoch[3] Batch [350]    Speed: 27467.81 samples/sec Train-top_k_accuracy_5=0.999531


INFO:root:Epoch[3] Batch [350]  Speed: 27467.81 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:37:54,621 Node[0] Epoch[3] Batch [350]    Speed: 27467.81 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[3] Batch [350]  Speed: 27467.81 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:37:54,622 Node[0] Epoch[3] Batch [350]    Speed: 27467.81 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[3] Batch [400]  Speed: 27004.22 samples/sec Train-accuracy=0.982656


2016-10-26 19:37:54,862 Node[0] Epoch[3] Batch [400]    Speed: 27004.22 samples/sec Train-accuracy=0.982656


INFO:root:Epoch[3] Batch [400]  Speed: 27004.22 samples/sec Train-top_k_accuracy_5=0.999531


2016-10-26 19:37:54,865 Node[0] Epoch[3] Batch [400]    Speed: 27004.22 samples/sec Train-top_k_accuracy_5=0.999531


INFO:root:Epoch[3] Batch [400]  Speed: 27004.22 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:37:54,867 Node[0] Epoch[3] Batch [400]    Speed: 27004.22 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[3] Batch [400]  Speed: 27004.22 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:37:54,868 Node[0] Epoch[3] Batch [400]    Speed: 27004.22 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[3] Batch [450]  Speed: 27586.19 samples/sec Train-accuracy=0.981094


2016-10-26 19:37:55,101 Node[0] Epoch[3] Batch [450]    Speed: 27586.19 samples/sec Train-accuracy=0.981094


INFO:root:Epoch[3] Batch [450]  Speed: 27586.19 samples/sec Train-top_k_accuracy_5=1.000000


2016-10-26 19:37:55,104 Node[0] Epoch[3] Batch [450]    Speed: 27586.19 samples/sec Train-top_k_accuracy_5=1.000000


INFO:root:Epoch[3] Batch [450]  Speed: 27586.19 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:37:55,105 Node[0] Epoch[3] Batch [450]    Speed: 27586.19 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[3] Batch [450]  Speed: 27586.19 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:37:55,107 Node[0] Epoch[3] Batch [450]    Speed: 27586.19 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[3] Resetting Data Iterator


2016-10-26 19:37:55,191 Node[0] Epoch[3] Resetting Data Iterator


INFO:root:Epoch[3] Time cost=2.283


2016-10-26 19:37:55,194 Node[0] Epoch[3] Time cost=2.283


INFO:root:Epoch[3] Validation-accuracy=0.974359


2016-10-26 19:37:55,359 Node[0] Epoch[3] Validation-accuracy=0.974359


INFO:root:Epoch[3] Validation-top_k_accuracy_5=0.999199


2016-10-26 19:37:55,361 Node[0] Epoch[3] Validation-top_k_accuracy_5=0.999199


INFO:root:Epoch[3] Validation-top_k_accuracy_10=1.000000


2016-10-26 19:37:55,362 Node[0] Epoch[3] Validation-top_k_accuracy_10=1.000000


INFO:root:Epoch[3] Validation-top_k_accuracy_20=1.000000


2016-10-26 19:37:55,364 Node[0] Epoch[3] Validation-top_k_accuracy_20=1.000000


INFO:root:Epoch[4] Batch [50]   Speed: 27586.19 samples/sec Train-accuracy=0.980938


2016-10-26 19:37:55,601 Node[0] Epoch[4] Batch [50] Speed: 27586.19 samples/sec Train-accuracy=0.980938


INFO:root:Epoch[4] Batch [50]   Speed: 27586.19 samples/sec Train-top_k_accuracy_5=0.999844


2016-10-26 19:37:55,605 Node[0] Epoch[4] Batch [50] Speed: 27586.19 samples/sec Train-top_k_accuracy_5=0.999844


INFO:root:Epoch[4] Batch [50]   Speed: 27586.19 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:37:55,608 Node[0] Epoch[4] Batch [50] Speed: 27586.19 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[4] Batch [50]   Speed: 27586.19 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:37:55,611 Node[0] Epoch[4] Batch [50] Speed: 27586.19 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[4] Batch [100]  Speed: 27586.19 samples/sec Train-accuracy=0.981406


2016-10-26 19:37:55,844 Node[0] Epoch[4] Batch [100]    Speed: 27586.19 samples/sec Train-accuracy=0.981406


INFO:root:Epoch[4] Batch [100]  Speed: 27586.19 samples/sec Train-top_k_accuracy_5=0.999687


2016-10-26 19:37:55,845 Node[0] Epoch[4] Batch [100]    Speed: 27586.19 samples/sec Train-top_k_accuracy_5=0.999687


INFO:root:Epoch[4] Batch [100]  Speed: 27586.19 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:37:55,846 Node[0] Epoch[4] Batch [100]    Speed: 27586.19 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[4] Batch [100]  Speed: 27586.19 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:37:55,848 Node[0] Epoch[4] Batch [100]    Speed: 27586.19 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[4] Batch [150]  Speed: 27118.64 samples/sec Train-accuracy=0.981875


2016-10-26 19:37:56,085 Node[0] Epoch[4] Batch [150]    Speed: 27118.64 samples/sec Train-accuracy=0.981875


INFO:root:Epoch[4] Batch [150]  Speed: 27118.64 samples/sec Train-top_k_accuracy_5=0.999687


2016-10-26 19:37:56,088 Node[0] Epoch[4] Batch [150]    Speed: 27118.64 samples/sec Train-top_k_accuracy_5=0.999687


INFO:root:Epoch[4] Batch [150]  Speed: 27118.64 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:37:56,089 Node[0] Epoch[4] Batch [150]    Speed: 27118.64 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[4] Batch [150]  Speed: 27118.64 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:37:56,091 Node[0] Epoch[4] Batch [150]    Speed: 27118.64 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[4] Batch [200]  Speed: 28571.43 samples/sec Train-accuracy=0.987187


2016-10-26 19:37:56,315 Node[0] Epoch[4] Batch [200]    Speed: 28571.43 samples/sec Train-accuracy=0.987187


INFO:root:Epoch[4] Batch [200]  Speed: 28571.43 samples/sec Train-top_k_accuracy_5=1.000000


2016-10-26 19:37:56,318 Node[0] Epoch[4] Batch [200]    Speed: 28571.43 samples/sec Train-top_k_accuracy_5=1.000000


INFO:root:Epoch[4] Batch [200]  Speed: 28571.43 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:37:56,319 Node[0] Epoch[4] Batch [200]    Speed: 28571.43 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[4] Batch [200]  Speed: 28571.43 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:37:56,321 Node[0] Epoch[4] Batch [200]    Speed: 28571.43 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[4] Batch [250]  Speed: 27350.43 samples/sec Train-accuracy=0.982812


2016-10-26 19:37:56,555 Node[0] Epoch[4] Batch [250]    Speed: 27350.43 samples/sec Train-accuracy=0.982812


INFO:root:Epoch[4] Batch [250]  Speed: 27350.43 samples/sec Train-top_k_accuracy_5=0.999531


2016-10-26 19:37:56,558 Node[0] Epoch[4] Batch [250]    Speed: 27350.43 samples/sec Train-top_k_accuracy_5=0.999531


INFO:root:Epoch[4] Batch [250]  Speed: 27350.43 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:37:56,559 Node[0] Epoch[4] Batch [250]    Speed: 27350.43 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[4] Batch [250]  Speed: 27350.43 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:37:56,559 Node[0] Epoch[4] Batch [250]    Speed: 27350.43 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[4] Batch [300]  Speed: 29493.09 samples/sec Train-accuracy=0.982969


2016-10-26 19:37:56,779 Node[0] Epoch[4] Batch [300]    Speed: 29493.09 samples/sec Train-accuracy=0.982969


INFO:root:Epoch[4] Batch [300]  Speed: 29493.09 samples/sec Train-top_k_accuracy_5=1.000000


2016-10-26 19:37:56,780 Node[0] Epoch[4] Batch [300]    Speed: 29493.09 samples/sec Train-top_k_accuracy_5=1.000000


INFO:root:Epoch[4] Batch [300]  Speed: 29493.09 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:37:56,782 Node[0] Epoch[4] Batch [300]    Speed: 29493.09 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[4] Batch [300]  Speed: 29493.09 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:37:56,786 Node[0] Epoch[4] Batch [300]    Speed: 29493.09 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[4] Batch [350]  Speed: 26016.25 samples/sec Train-accuracy=0.981563


2016-10-26 19:37:57,036 Node[0] Epoch[4] Batch [350]    Speed: 26016.25 samples/sec Train-accuracy=0.981563


INFO:root:Epoch[4] Batch [350]  Speed: 26016.25 samples/sec Train-top_k_accuracy_5=0.999531


2016-10-26 19:37:57,039 Node[0] Epoch[4] Batch [350]    Speed: 26016.25 samples/sec Train-top_k_accuracy_5=0.999531


INFO:root:Epoch[4] Batch [350]  Speed: 26016.25 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:37:57,040 Node[0] Epoch[4] Batch [350]    Speed: 26016.25 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[4] Batch [350]  Speed: 26016.25 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:37:57,042 Node[0] Epoch[4] Batch [350]    Speed: 26016.25 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[4] Batch [400]  Speed: 27118.66 samples/sec Train-accuracy=0.985000


2016-10-26 19:37:57,280 Node[0] Epoch[4] Batch [400]    Speed: 27118.66 samples/sec Train-accuracy=0.985000


INFO:root:Epoch[4] Batch [400]  Speed: 27118.66 samples/sec Train-top_k_accuracy_5=0.999687


2016-10-26 19:37:57,282 Node[0] Epoch[4] Batch [400]    Speed: 27118.66 samples/sec Train-top_k_accuracy_5=0.999687


INFO:root:Epoch[4] Batch [400]  Speed: 27118.66 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:37:57,285 Node[0] Epoch[4] Batch [400]    Speed: 27118.66 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[4] Batch [400]  Speed: 27118.66 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:37:57,286 Node[0] Epoch[4] Batch [400]    Speed: 27118.66 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[4] Batch [450]  Speed: 26446.30 samples/sec Train-accuracy=0.983281


2016-10-26 19:37:57,530 Node[0] Epoch[4] Batch [450]    Speed: 26446.30 samples/sec Train-accuracy=0.983281


INFO:root:Epoch[4] Batch [450]  Speed: 26446.30 samples/sec Train-top_k_accuracy_5=1.000000


2016-10-26 19:37:57,532 Node[0] Epoch[4] Batch [450]    Speed: 26446.30 samples/sec Train-top_k_accuracy_5=1.000000


INFO:root:Epoch[4] Batch [450]  Speed: 26446.30 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:37:57,533 Node[0] Epoch[4] Batch [450]    Speed: 26446.30 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[4] Batch [450]  Speed: 26446.30 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:37:57,535 Node[0] Epoch[4] Batch [450]    Speed: 26446.30 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[4] Resetting Data Iterator


2016-10-26 19:37:57,615 Node[0] Epoch[4] Resetting Data Iterator


INFO:root:Epoch[4] Time cost=2.251


2016-10-26 19:37:57,617 Node[0] Epoch[4] Time cost=2.251


INFO:root:Epoch[4] Validation-accuracy=0.972456


2016-10-26 19:37:57,776 Node[0] Epoch[4] Validation-accuracy=0.972456


INFO:root:Epoch[4] Validation-top_k_accuracy_5=0.999299


2016-10-26 19:37:57,776 Node[0] Epoch[4] Validation-top_k_accuracy_5=0.999299


INFO:root:Epoch[4] Validation-top_k_accuracy_10=1.000000


2016-10-26 19:37:57,779 Node[0] Epoch[4] Validation-top_k_accuracy_10=1.000000


INFO:root:Epoch[4] Validation-top_k_accuracy_20=1.000000


2016-10-26 19:37:57,779 Node[0] Epoch[4] Validation-top_k_accuracy_20=1.000000


INFO:root:Epoch[5] Batch [50]   Speed: 27705.63 samples/sec Train-accuracy=0.984219


2016-10-26 19:37:58,019 Node[0] Epoch[5] Batch [50] Speed: 27705.63 samples/sec Train-accuracy=0.984219


INFO:root:Epoch[5] Batch [50]   Speed: 27705.63 samples/sec Train-top_k_accuracy_5=0.999687


2016-10-26 19:37:58,022 Node[0] Epoch[5] Batch [50] Speed: 27705.63 samples/sec Train-top_k_accuracy_5=0.999687


INFO:root:Epoch[5] Batch [50]   Speed: 27705.63 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:37:58,023 Node[0] Epoch[5] Batch [50] Speed: 27705.63 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[5] Batch [50]   Speed: 27705.63 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:37:58,025 Node[0] Epoch[5] Batch [50] Speed: 27705.63 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[5] Batch [100]  Speed: 28193.83 samples/sec Train-accuracy=0.987344


2016-10-26 19:37:58,253 Node[0] Epoch[5] Batch [100]    Speed: 28193.83 samples/sec Train-accuracy=0.987344


INFO:root:Epoch[5] Batch [100]  Speed: 28193.83 samples/sec Train-top_k_accuracy_5=1.000000


2016-10-26 19:37:58,255 Node[0] Epoch[5] Batch [100]    Speed: 28193.83 samples/sec Train-top_k_accuracy_5=1.000000


INFO:root:Epoch[5] Batch [100]  Speed: 28193.83 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:37:58,256 Node[0] Epoch[5] Batch [100]    Speed: 28193.83 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[5] Batch [100]  Speed: 28193.83 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:37:58,257 Node[0] Epoch[5] Batch [100]    Speed: 28193.83 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[5] Batch [150]  Speed: 27947.62 samples/sec Train-accuracy=0.986094


2016-10-26 19:37:58,487 Node[0] Epoch[5] Batch [150]    Speed: 27947.62 samples/sec Train-accuracy=0.986094


INFO:root:Epoch[5] Batch [150]  Speed: 27947.62 samples/sec Train-top_k_accuracy_5=0.999531


2016-10-26 19:37:58,489 Node[0] Epoch[5] Batch [150]    Speed: 27947.62 samples/sec Train-top_k_accuracy_5=0.999531


INFO:root:Epoch[5] Batch [150]  Speed: 27947.62 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:37:58,490 Node[0] Epoch[5] Batch [150]    Speed: 27947.62 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[5] Batch [150]  Speed: 27947.62 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:37:58,492 Node[0] Epoch[5] Batch [150]    Speed: 27947.62 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[5] Batch [200]  Speed: 28828.84 samples/sec Train-accuracy=0.987031


2016-10-26 19:37:58,714 Node[0] Epoch[5] Batch [200]    Speed: 28828.84 samples/sec Train-accuracy=0.987031


INFO:root:Epoch[5] Batch [200]  Speed: 28828.84 samples/sec Train-top_k_accuracy_5=1.000000


2016-10-26 19:37:58,717 Node[0] Epoch[5] Batch [200]    Speed: 28828.84 samples/sec Train-top_k_accuracy_5=1.000000


INFO:root:Epoch[5] Batch [200]  Speed: 28828.84 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:37:58,719 Node[0] Epoch[5] Batch [200]    Speed: 28828.84 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[5] Batch [200]  Speed: 28828.84 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:37:58,720 Node[0] Epoch[5] Batch [200]    Speed: 28828.84 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[5] Batch [250]  Speed: 28571.40 samples/sec Train-accuracy=0.984531


2016-10-26 19:37:58,946 Node[0] Epoch[5] Batch [250]    Speed: 28571.40 samples/sec Train-accuracy=0.984531


INFO:root:Epoch[5] Batch [250]  Speed: 28571.40 samples/sec Train-top_k_accuracy_5=0.999844


2016-10-26 19:37:58,947 Node[0] Epoch[5] Batch [250]    Speed: 28571.40 samples/sec Train-top_k_accuracy_5=0.999844


INFO:root:Epoch[5] Batch [250]  Speed: 28571.40 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:37:58,950 Node[0] Epoch[5] Batch [250]    Speed: 28571.40 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[5] Batch [250]  Speed: 28571.40 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:37:58,953 Node[0] Epoch[5] Batch [250]    Speed: 28571.40 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[5] Batch [300]  Speed: 27826.08 samples/sec Train-accuracy=0.985469


2016-10-26 19:37:59,184 Node[0] Epoch[5] Batch [300]    Speed: 27826.08 samples/sec Train-accuracy=0.985469


INFO:root:Epoch[5] Batch [300]  Speed: 27826.08 samples/sec Train-top_k_accuracy_5=1.000000


2016-10-26 19:37:59,187 Node[0] Epoch[5] Batch [300]    Speed: 27826.08 samples/sec Train-top_k_accuracy_5=1.000000


INFO:root:Epoch[5] Batch [300]  Speed: 27826.08 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:37:59,187 Node[0] Epoch[5] Batch [300]    Speed: 27826.08 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[5] Batch [300]  Speed: 27826.08 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:37:59,188 Node[0] Epoch[5] Batch [300]    Speed: 27826.08 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[5] Batch [350]  Speed: 28444.46 samples/sec Train-accuracy=0.983125


2016-10-26 19:37:59,415 Node[0] Epoch[5] Batch [350]    Speed: 28444.46 samples/sec Train-accuracy=0.983125


INFO:root:Epoch[5] Batch [350]  Speed: 28444.46 samples/sec Train-top_k_accuracy_5=0.999844


2016-10-26 19:37:59,417 Node[0] Epoch[5] Batch [350]    Speed: 28444.46 samples/sec Train-top_k_accuracy_5=0.999844


INFO:root:Epoch[5] Batch [350]  Speed: 28444.46 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:37:59,418 Node[0] Epoch[5] Batch [350]    Speed: 28444.46 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[5] Batch [350]  Speed: 28444.46 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:37:59,421 Node[0] Epoch[5] Batch [350]    Speed: 28444.46 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[5] Batch [400]  Speed: 28318.58 samples/sec Train-accuracy=0.987500


2016-10-26 19:37:59,648 Node[0] Epoch[5] Batch [400]    Speed: 28318.58 samples/sec Train-accuracy=0.987500


INFO:root:Epoch[5] Batch [400]  Speed: 28318.58 samples/sec Train-top_k_accuracy_5=0.999687


2016-10-26 19:37:59,650 Node[0] Epoch[5] Batch [400]    Speed: 28318.58 samples/sec Train-top_k_accuracy_5=0.999687


INFO:root:Epoch[5] Batch [400]  Speed: 28318.58 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:37:59,651 Node[0] Epoch[5] Batch [400]    Speed: 28318.58 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[5] Batch [400]  Speed: 28318.58 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:37:59,653 Node[0] Epoch[5] Batch [400]    Speed: 28318.58 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[5] Batch [450]  Speed: 28571.40 samples/sec Train-accuracy=0.987031


2016-10-26 19:37:59,880 Node[0] Epoch[5] Batch [450]    Speed: 28571.40 samples/sec Train-accuracy=0.987031


INFO:root:Epoch[5] Batch [450]  Speed: 28571.40 samples/sec Train-top_k_accuracy_5=1.000000


2016-10-26 19:37:59,881 Node[0] Epoch[5] Batch [450]    Speed: 28571.40 samples/sec Train-top_k_accuracy_5=1.000000


INFO:root:Epoch[5] Batch [450]  Speed: 28571.40 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:37:59,882 Node[0] Epoch[5] Batch [450]    Speed: 28571.40 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[5] Batch [450]  Speed: 28571.40 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:37:59,884 Node[0] Epoch[5] Batch [450]    Speed: 28571.40 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[5] Resetting Data Iterator


2016-10-26 19:37:59,973 Node[0] Epoch[5] Resetting Data Iterator


INFO:root:Epoch[5] Time cost=2.194


2016-10-26 19:37:59,974 Node[0] Epoch[5] Time cost=2.194


INFO:root:Epoch[5] Validation-accuracy=0.974459


2016-10-26 19:38:00,132 Node[0] Epoch[5] Validation-accuracy=0.974459


INFO:root:Epoch[5] Validation-top_k_accuracy_5=0.999199


2016-10-26 19:38:00,134 Node[0] Epoch[5] Validation-top_k_accuracy_5=0.999199


INFO:root:Epoch[5] Validation-top_k_accuracy_10=1.000000


2016-10-26 19:38:00,135 Node[0] Epoch[5] Validation-top_k_accuracy_10=1.000000


INFO:root:Epoch[5] Validation-top_k_accuracy_20=1.000000


2016-10-26 19:38:00,138 Node[0] Epoch[5] Validation-top_k_accuracy_20=1.000000


INFO:root:Epoch[6] Batch [50]   Speed: 29357.81 samples/sec Train-accuracy=0.990156


2016-10-26 19:38:00,361 Node[0] Epoch[6] Batch [50] Speed: 29357.81 samples/sec Train-accuracy=0.990156


INFO:root:Epoch[6] Batch [50]   Speed: 29357.81 samples/sec Train-top_k_accuracy_5=1.000000


2016-10-26 19:38:00,364 Node[0] Epoch[6] Batch [50] Speed: 29357.81 samples/sec Train-top_k_accuracy_5=1.000000


INFO:root:Epoch[6] Batch [50]   Speed: 29357.81 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:38:00,365 Node[0] Epoch[6] Batch [50] Speed: 29357.81 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[6] Batch [50]   Speed: 29357.81 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:38:00,367 Node[0] Epoch[6] Batch [50] Speed: 29357.81 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[6] Batch [100]  Speed: 26890.77 samples/sec Train-accuracy=0.989219


2016-10-26 19:38:00,605 Node[0] Epoch[6] Batch [100]    Speed: 26890.77 samples/sec Train-accuracy=0.989219


INFO:root:Epoch[6] Batch [100]  Speed: 26890.77 samples/sec Train-top_k_accuracy_5=1.000000


2016-10-26 19:38:00,607 Node[0] Epoch[6] Batch [100]    Speed: 26890.77 samples/sec Train-top_k_accuracy_5=1.000000


INFO:root:Epoch[6] Batch [100]  Speed: 26890.77 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:38:00,608 Node[0] Epoch[6] Batch [100]    Speed: 26890.77 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[6] Batch [100]  Speed: 26890.77 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:38:00,609 Node[0] Epoch[6] Batch [100]    Speed: 26890.77 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[6] Batch [150]  Speed: 27947.62 samples/sec Train-accuracy=0.989219


2016-10-26 19:38:00,839 Node[0] Epoch[6] Batch [150]    Speed: 27947.62 samples/sec Train-accuracy=0.989219


INFO:root:Epoch[6] Batch [150]  Speed: 27947.62 samples/sec Train-top_k_accuracy_5=1.000000


2016-10-26 19:38:00,842 Node[0] Epoch[6] Batch [150]    Speed: 27947.62 samples/sec Train-top_k_accuracy_5=1.000000


INFO:root:Epoch[6] Batch [150]  Speed: 27947.62 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:38:00,842 Node[0] Epoch[6] Batch [150]    Speed: 27947.62 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[6] Batch [150]  Speed: 27947.62 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:38:00,845 Node[0] Epoch[6] Batch [150]    Speed: 27947.62 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[6] Batch [200]  Speed: 26890.74 samples/sec Train-accuracy=0.988281


2016-10-26 19:38:01,084 Node[0] Epoch[6] Batch [200]    Speed: 26890.74 samples/sec Train-accuracy=0.988281


INFO:root:Epoch[6] Batch [200]  Speed: 26890.74 samples/sec Train-top_k_accuracy_5=0.999844


2016-10-26 19:38:01,088 Node[0] Epoch[6] Batch [200]    Speed: 26890.74 samples/sec Train-top_k_accuracy_5=0.999844


INFO:root:Epoch[6] Batch [200]  Speed: 26890.74 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:38:01,088 Node[0] Epoch[6] Batch [200]    Speed: 26890.74 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[6] Batch [200]  Speed: 26890.74 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:38:01,089 Node[0] Epoch[6] Batch [200]    Speed: 26890.74 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[6] Batch [250]  Speed: 27947.59 samples/sec Train-accuracy=0.987969


2016-10-26 19:38:01,322 Node[0] Epoch[6] Batch [250]    Speed: 27947.59 samples/sec Train-accuracy=0.987969


INFO:root:Epoch[6] Batch [250]  Speed: 27947.59 samples/sec Train-top_k_accuracy_5=1.000000


2016-10-26 19:38:01,323 Node[0] Epoch[6] Batch [250]    Speed: 27947.59 samples/sec Train-top_k_accuracy_5=1.000000


INFO:root:Epoch[6] Batch [250]  Speed: 27947.59 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:38:01,325 Node[0] Epoch[6] Batch [250]    Speed: 27947.59 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[6] Batch [250]  Speed: 27947.59 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:38:01,326 Node[0] Epoch[6] Batch [250]    Speed: 27947.59 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[6] Batch [300]  Speed: 28070.18 samples/sec Train-accuracy=0.987187


2016-10-26 19:38:01,555 Node[0] Epoch[6] Batch [300]    Speed: 28070.18 samples/sec Train-accuracy=0.987187


INFO:root:Epoch[6] Batch [300]  Speed: 28070.18 samples/sec Train-top_k_accuracy_5=1.000000


2016-10-26 19:38:01,558 Node[0] Epoch[6] Batch [300]    Speed: 28070.18 samples/sec Train-top_k_accuracy_5=1.000000


INFO:root:Epoch[6] Batch [300]  Speed: 28070.18 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:38:01,559 Node[0] Epoch[6] Batch [300]    Speed: 28070.18 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[6] Batch [300]  Speed: 28070.18 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:38:01,561 Node[0] Epoch[6] Batch [300]    Speed: 28070.18 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[6] Batch [350]  Speed: 27234.03 samples/sec Train-accuracy=0.981719


2016-10-26 19:38:01,798 Node[0] Epoch[6] Batch [350]    Speed: 27234.03 samples/sec Train-accuracy=0.981719


INFO:root:Epoch[6] Batch [350]  Speed: 27234.03 samples/sec Train-top_k_accuracy_5=0.999687


2016-10-26 19:38:01,799 Node[0] Epoch[6] Batch [350]    Speed: 27234.03 samples/sec Train-top_k_accuracy_5=0.999687


INFO:root:Epoch[6] Batch [350]  Speed: 27234.03 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:38:01,802 Node[0] Epoch[6] Batch [350]    Speed: 27234.03 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[6] Batch [350]  Speed: 27234.03 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:38:01,803 Node[0] Epoch[6] Batch [350]    Speed: 27234.03 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[6] Batch [400]  Speed: 28193.83 samples/sec Train-accuracy=0.987187


2016-10-26 19:38:02,032 Node[0] Epoch[6] Batch [400]    Speed: 28193.83 samples/sec Train-accuracy=0.987187


INFO:root:Epoch[6] Batch [400]  Speed: 28193.83 samples/sec Train-top_k_accuracy_5=0.999687


2016-10-26 19:38:02,035 Node[0] Epoch[6] Batch [400]    Speed: 28193.83 samples/sec Train-top_k_accuracy_5=0.999687


INFO:root:Epoch[6] Batch [400]  Speed: 28193.83 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:38:02,036 Node[0] Epoch[6] Batch [400]    Speed: 28193.83 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[6] Batch [400]  Speed: 28193.83 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:38:02,038 Node[0] Epoch[6] Batch [400]    Speed: 28193.83 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[6] Batch [450]  Speed: 28318.58 samples/sec Train-accuracy=0.988750


2016-10-26 19:38:02,265 Node[0] Epoch[6] Batch [450]    Speed: 28318.58 samples/sec Train-accuracy=0.988750


INFO:root:Epoch[6] Batch [450]  Speed: 28318.58 samples/sec Train-top_k_accuracy_5=1.000000


2016-10-26 19:38:02,266 Node[0] Epoch[6] Batch [450]    Speed: 28318.58 samples/sec Train-top_k_accuracy_5=1.000000


INFO:root:Epoch[6] Batch [450]  Speed: 28318.58 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:38:02,269 Node[0] Epoch[6] Batch [450]    Speed: 28318.58 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[6] Batch [450]  Speed: 28318.58 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:38:02,270 Node[0] Epoch[6] Batch [450]    Speed: 28318.58 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[6] Resetting Data Iterator


2016-10-26 19:38:02,349 Node[0] Epoch[6] Resetting Data Iterator


INFO:root:Epoch[6] Time cost=2.214


2016-10-26 19:38:02,354 Node[0] Epoch[6] Time cost=2.214


INFO:root:Epoch[6] Validation-accuracy=0.975160


2016-10-26 19:38:02,510 Node[0] Epoch[6] Validation-accuracy=0.975160


INFO:root:Epoch[6] Validation-top_k_accuracy_5=0.999499


2016-10-26 19:38:02,512 Node[0] Epoch[6] Validation-top_k_accuracy_5=0.999499


INFO:root:Epoch[6] Validation-top_k_accuracy_10=1.000000


2016-10-26 19:38:02,513 Node[0] Epoch[6] Validation-top_k_accuracy_10=1.000000


INFO:root:Epoch[6] Validation-top_k_accuracy_20=1.000000


2016-10-26 19:38:02,515 Node[0] Epoch[6] Validation-top_k_accuracy_20=1.000000


INFO:root:Epoch[7] Batch [50]   Speed: 30188.69 samples/sec Train-accuracy=0.989844


2016-10-26 19:38:02,733 Node[0] Epoch[7] Batch [50] Speed: 30188.69 samples/sec Train-accuracy=0.989844


INFO:root:Epoch[7] Batch [50]   Speed: 30188.69 samples/sec Train-top_k_accuracy_5=0.999844


2016-10-26 19:38:02,736 Node[0] Epoch[7] Batch [50] Speed: 30188.69 samples/sec Train-top_k_accuracy_5=0.999844


INFO:root:Epoch[7] Batch [50]   Speed: 30188.69 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:38:02,736 Node[0] Epoch[7] Batch [50] Speed: 30188.69 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[7] Batch [50]   Speed: 30188.69 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:38:02,739 Node[0] Epoch[7] Batch [50] Speed: 30188.69 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[7] Batch [100]  Speed: 27350.43 samples/sec Train-accuracy=0.990781


2016-10-26 19:38:02,973 Node[0] Epoch[7] Batch [100]    Speed: 27350.43 samples/sec Train-accuracy=0.990781


INFO:root:Epoch[7] Batch [100]  Speed: 27350.43 samples/sec Train-top_k_accuracy_5=1.000000


2016-10-26 19:38:02,974 Node[0] Epoch[7] Batch [100]    Speed: 27350.43 samples/sec Train-top_k_accuracy_5=1.000000


INFO:root:Epoch[7] Batch [100]  Speed: 27350.43 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:38:02,976 Node[0] Epoch[7] Batch [100]    Speed: 27350.43 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[7] Batch [100]  Speed: 27350.43 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:38:02,977 Node[0] Epoch[7] Batch [100]    Speed: 27350.43 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[7] Batch [150]  Speed: 27947.62 samples/sec Train-accuracy=0.988125


2016-10-26 19:38:03,207 Node[0] Epoch[7] Batch [150]    Speed: 27947.62 samples/sec Train-accuracy=0.988125


INFO:root:Epoch[7] Batch [150]  Speed: 27947.62 samples/sec Train-top_k_accuracy_5=0.999687


2016-10-26 19:38:03,210 Node[0] Epoch[7] Batch [150]    Speed: 27947.62 samples/sec Train-top_k_accuracy_5=0.999687


INFO:root:Epoch[7] Batch [150]  Speed: 27947.62 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:38:03,210 Node[0] Epoch[7] Batch [150]    Speed: 27947.62 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[7] Batch [150]  Speed: 27947.62 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:38:03,211 Node[0] Epoch[7] Batch [150]    Speed: 27947.62 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[7] Batch [200]  Speed: 28070.18 samples/sec Train-accuracy=0.985625


2016-10-26 19:38:03,441 Node[0] Epoch[7] Batch [200]    Speed: 28070.18 samples/sec Train-accuracy=0.985625


INFO:root:Epoch[7] Batch [200]  Speed: 28070.18 samples/sec Train-top_k_accuracy_5=1.000000


2016-10-26 19:38:03,444 Node[0] Epoch[7] Batch [200]    Speed: 28070.18 samples/sec Train-top_k_accuracy_5=1.000000


INFO:root:Epoch[7] Batch [200]  Speed: 28070.18 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:38:03,444 Node[0] Epoch[7] Batch [200]    Speed: 28070.18 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[7] Batch [200]  Speed: 28070.18 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:38:03,446 Node[0] Epoch[7] Batch [200]    Speed: 28070.18 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[7] Batch [250]  Speed: 27705.63 samples/sec Train-accuracy=0.989219


2016-10-26 19:38:03,677 Node[0] Epoch[7] Batch [250]    Speed: 27705.63 samples/sec Train-accuracy=0.989219


INFO:root:Epoch[7] Batch [250]  Speed: 27705.63 samples/sec Train-top_k_accuracy_5=1.000000


2016-10-26 19:38:03,680 Node[0] Epoch[7] Batch [250]    Speed: 27705.63 samples/sec Train-top_k_accuracy_5=1.000000


INFO:root:Epoch[7] Batch [250]  Speed: 27705.63 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:38:03,680 Node[0] Epoch[7] Batch [250]    Speed: 27705.63 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[7] Batch [250]  Speed: 27705.63 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:38:03,683 Node[0] Epoch[7] Batch [250]    Speed: 27705.63 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[7] Batch [300]  Speed: 28070.18 samples/sec Train-accuracy=0.988125


2016-10-26 19:38:03,911 Node[0] Epoch[7] Batch [300]    Speed: 28070.18 samples/sec Train-accuracy=0.988125


INFO:root:Epoch[7] Batch [300]  Speed: 28070.18 samples/sec Train-top_k_accuracy_5=1.000000


2016-10-26 19:38:03,914 Node[0] Epoch[7] Batch [300]    Speed: 28070.18 samples/sec Train-top_k_accuracy_5=1.000000


INFO:root:Epoch[7] Batch [300]  Speed: 28070.18 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:38:03,915 Node[0] Epoch[7] Batch [300]    Speed: 28070.18 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[7] Batch [300]  Speed: 28070.18 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:38:03,917 Node[0] Epoch[7] Batch [300]    Speed: 28070.18 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[7] Batch [350]  Speed: 28318.58 samples/sec Train-accuracy=0.989531


2016-10-26 19:38:04,144 Node[0] Epoch[7] Batch [350]    Speed: 28318.58 samples/sec Train-accuracy=0.989531


INFO:root:Epoch[7] Batch [350]  Speed: 28318.58 samples/sec Train-top_k_accuracy_5=1.000000


2016-10-26 19:38:04,144 Node[0] Epoch[7] Batch [350]    Speed: 28318.58 samples/sec Train-top_k_accuracy_5=1.000000


INFO:root:Epoch[7] Batch [350]  Speed: 28318.58 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:38:04,147 Node[0] Epoch[7] Batch [350]    Speed: 28318.58 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[7] Batch [350]  Speed: 28318.58 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:38:04,148 Node[0] Epoch[7] Batch [350]    Speed: 28318.58 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[7] Batch [400]  Speed: 30331.76 samples/sec Train-accuracy=0.987812


2016-10-26 19:38:04,361 Node[0] Epoch[7] Batch [400]    Speed: 30331.76 samples/sec Train-accuracy=0.987812


INFO:root:Epoch[7] Batch [400]  Speed: 30331.76 samples/sec Train-top_k_accuracy_5=0.999844


2016-10-26 19:38:04,362 Node[0] Epoch[7] Batch [400]    Speed: 30331.76 samples/sec Train-top_k_accuracy_5=0.999844


INFO:root:Epoch[7] Batch [400]  Speed: 30331.76 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:38:04,365 Node[0] Epoch[7] Batch [400]    Speed: 30331.76 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[7] Batch [400]  Speed: 30331.76 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:38:04,367 Node[0] Epoch[7] Batch [400]    Speed: 30331.76 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[7] Batch [450]  Speed: 27826.08 samples/sec Train-accuracy=0.990000


2016-10-26 19:38:04,598 Node[0] Epoch[7] Batch [450]    Speed: 27826.08 samples/sec Train-accuracy=0.990000


INFO:root:Epoch[7] Batch [450]  Speed: 27826.08 samples/sec Train-top_k_accuracy_5=1.000000


2016-10-26 19:38:04,599 Node[0] Epoch[7] Batch [450]    Speed: 27826.08 samples/sec Train-top_k_accuracy_5=1.000000


INFO:root:Epoch[7] Batch [450]  Speed: 27826.08 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:38:04,601 Node[0] Epoch[7] Batch [450]    Speed: 27826.08 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[7] Batch [450]  Speed: 27826.08 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:38:04,604 Node[0] Epoch[7] Batch [450]    Speed: 27826.08 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[7] Resetting Data Iterator


2016-10-26 19:38:04,687 Node[0] Epoch[7] Resetting Data Iterator


INFO:root:Epoch[7] Time cost=2.172


2016-10-26 19:38:04,690 Node[0] Epoch[7] Time cost=2.172


INFO:root:Epoch[7] Validation-accuracy=0.977564


2016-10-26 19:38:04,842 Node[0] Epoch[7] Validation-accuracy=0.977564


INFO:root:Epoch[7] Validation-top_k_accuracy_5=0.999599


2016-10-26 19:38:04,845 Node[0] Epoch[7] Validation-top_k_accuracy_5=0.999599


INFO:root:Epoch[7] Validation-top_k_accuracy_10=1.000000


2016-10-26 19:38:04,845 Node[0] Epoch[7] Validation-top_k_accuracy_10=1.000000


INFO:root:Epoch[7] Validation-top_k_accuracy_20=1.000000


2016-10-26 19:38:04,848 Node[0] Epoch[7] Validation-top_k_accuracy_20=1.000000


INFO:root:Epoch[8] Batch [50]   Speed: 29629.65 samples/sec Train-accuracy=0.990469


2016-10-26 19:38:05,069 Node[0] Epoch[8] Batch [50] Speed: 29629.65 samples/sec Train-accuracy=0.990469


INFO:root:Epoch[8] Batch [50]   Speed: 29629.65 samples/sec Train-top_k_accuracy_5=1.000000


2016-10-26 19:38:05,072 Node[0] Epoch[8] Batch [50] Speed: 29629.65 samples/sec Train-top_k_accuracy_5=1.000000


INFO:root:Epoch[8] Batch [50]   Speed: 29629.65 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:38:05,072 Node[0] Epoch[8] Batch [50] Speed: 29629.65 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[8] Batch [50]   Speed: 29629.65 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:38:05,075 Node[0] Epoch[8] Batch [50] Speed: 29629.65 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[8] Batch [100]  Speed: 28699.55 samples/sec Train-accuracy=0.988594


2016-10-26 19:38:05,299 Node[0] Epoch[8] Batch [100]    Speed: 28699.55 samples/sec Train-accuracy=0.988594


INFO:root:Epoch[8] Batch [100]  Speed: 28699.55 samples/sec Train-top_k_accuracy_5=1.000000


2016-10-26 19:38:05,299 Node[0] Epoch[8] Batch [100]    Speed: 28699.55 samples/sec Train-top_k_accuracy_5=1.000000


INFO:root:Epoch[8] Batch [100]  Speed: 28699.55 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:38:05,302 Node[0] Epoch[8] Batch [100]    Speed: 28699.55 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[8] Batch [100]  Speed: 28699.55 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:38:05,303 Node[0] Epoch[8] Batch [100]    Speed: 28699.55 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[8] Batch [150]  Speed: 28070.18 samples/sec Train-accuracy=0.990313


2016-10-26 19:38:05,533 Node[0] Epoch[8] Batch [150]    Speed: 28070.18 samples/sec Train-accuracy=0.990313


INFO:root:Epoch[8] Batch [150]  Speed: 28070.18 samples/sec Train-top_k_accuracy_5=1.000000


2016-10-26 19:38:05,536 Node[0] Epoch[8] Batch [150]    Speed: 28070.18 samples/sec Train-top_k_accuracy_5=1.000000


INFO:root:Epoch[8] Batch [150]  Speed: 28070.18 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:38:05,538 Node[0] Epoch[8] Batch [150]    Speed: 28070.18 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[8] Batch [150]  Speed: 28070.18 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:38:05,539 Node[0] Epoch[8] Batch [150]    Speed: 28070.18 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[8] Batch [200]  Speed: 29223.73 samples/sec Train-accuracy=0.991250


2016-10-26 19:38:05,759 Node[0] Epoch[8] Batch [200]    Speed: 29223.73 samples/sec Train-accuracy=0.991250


INFO:root:Epoch[8] Batch [200]  Speed: 29223.73 samples/sec Train-top_k_accuracy_5=1.000000


2016-10-26 19:38:05,762 Node[0] Epoch[8] Batch [200]    Speed: 29223.73 samples/sec Train-top_k_accuracy_5=1.000000


INFO:root:Epoch[8] Batch [200]  Speed: 29223.73 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:38:05,763 Node[0] Epoch[8] Batch [200]    Speed: 29223.73 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[8] Batch [200]  Speed: 29223.73 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:38:05,766 Node[0] Epoch[8] Batch [200]    Speed: 29223.73 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[8] Batch [250]  Speed: 26016.25 samples/sec Train-accuracy=0.987969


2016-10-26 19:38:06,016 Node[0] Epoch[8] Batch [250]    Speed: 26016.25 samples/sec Train-accuracy=0.987969


INFO:root:Epoch[8] Batch [250]  Speed: 26016.25 samples/sec Train-top_k_accuracy_5=0.999844


2016-10-26 19:38:06,017 Node[0] Epoch[8] Batch [250]    Speed: 26016.25 samples/sec Train-top_k_accuracy_5=0.999844


INFO:root:Epoch[8] Batch [250]  Speed: 26016.25 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:38:06,019 Node[0] Epoch[8] Batch [250]    Speed: 26016.25 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[8] Batch [250]  Speed: 26016.25 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:38:06,022 Node[0] Epoch[8] Batch [250]    Speed: 26016.25 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[8] Batch [300]  Speed: 29223.73 samples/sec Train-accuracy=0.990313


2016-10-26 19:38:06,242 Node[0] Epoch[8] Batch [300]    Speed: 29223.73 samples/sec Train-accuracy=0.990313


INFO:root:Epoch[8] Batch [300]  Speed: 29223.73 samples/sec Train-top_k_accuracy_5=1.000000


2016-10-26 19:38:06,243 Node[0] Epoch[8] Batch [300]    Speed: 29223.73 samples/sec Train-top_k_accuracy_5=1.000000


INFO:root:Epoch[8] Batch [300]  Speed: 29223.73 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:38:06,244 Node[0] Epoch[8] Batch [300]    Speed: 29223.73 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[8] Batch [300]  Speed: 29223.73 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:38:06,246 Node[0] Epoch[8] Batch [300]    Speed: 29223.73 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[8] Batch [350]  Speed: 29223.73 samples/sec Train-accuracy=0.988750


2016-10-26 19:38:06,467 Node[0] Epoch[8] Batch [350]    Speed: 29223.73 samples/sec Train-accuracy=0.988750


INFO:root:Epoch[8] Batch [350]  Speed: 29223.73 samples/sec Train-top_k_accuracy_5=1.000000


2016-10-26 19:38:06,469 Node[0] Epoch[8] Batch [350]    Speed: 29223.73 samples/sec Train-top_k_accuracy_5=1.000000


INFO:root:Epoch[8] Batch [350]  Speed: 29223.73 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:38:06,470 Node[0] Epoch[8] Batch [350]    Speed: 29223.73 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[8] Batch [350]  Speed: 29223.73 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:38:06,471 Node[0] Epoch[8] Batch [350]    Speed: 29223.73 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[8] Batch [400]  Speed: 29223.73 samples/sec Train-accuracy=0.991250


2016-10-26 19:38:06,693 Node[0] Epoch[8] Batch [400]    Speed: 29223.73 samples/sec Train-accuracy=0.991250


INFO:root:Epoch[8] Batch [400]  Speed: 29223.73 samples/sec Train-top_k_accuracy_5=1.000000


2016-10-26 19:38:06,694 Node[0] Epoch[8] Batch [400]    Speed: 29223.73 samples/sec Train-top_k_accuracy_5=1.000000


INFO:root:Epoch[8] Batch [400]  Speed: 29223.73 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:38:06,696 Node[0] Epoch[8] Batch [400]    Speed: 29223.73 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[8] Batch [400]  Speed: 29223.73 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:38:06,697 Node[0] Epoch[8] Batch [400]    Speed: 29223.73 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[8] Batch [450]  Speed: 27350.43 samples/sec Train-accuracy=0.989844


2016-10-26 19:38:06,931 Node[0] Epoch[8] Batch [450]    Speed: 27350.43 samples/sec Train-accuracy=0.989844


INFO:root:Epoch[8] Batch [450]  Speed: 27350.43 samples/sec Train-top_k_accuracy_5=1.000000


2016-10-26 19:38:06,934 Node[0] Epoch[8] Batch [450]    Speed: 27350.43 samples/sec Train-top_k_accuracy_5=1.000000


INFO:root:Epoch[8] Batch [450]  Speed: 27350.43 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:38:06,934 Node[0] Epoch[8] Batch [450]    Speed: 27350.43 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[8] Batch [450]  Speed: 27350.43 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:38:06,937 Node[0] Epoch[8] Batch [450]    Speed: 27350.43 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[8] Resetting Data Iterator


2016-10-26 19:38:07,016 Node[0] Epoch[8] Resetting Data Iterator


INFO:root:Epoch[8] Time cost=2.169


2016-10-26 19:38:07,017 Node[0] Epoch[8] Time cost=2.169


INFO:root:Epoch[8] Validation-accuracy=0.976863


2016-10-26 19:38:07,174 Node[0] Epoch[8] Validation-accuracy=0.976863


INFO:root:Epoch[8] Validation-top_k_accuracy_5=0.999700


2016-10-26 19:38:07,174 Node[0] Epoch[8] Validation-top_k_accuracy_5=0.999700


INFO:root:Epoch[8] Validation-top_k_accuracy_10=1.000000


2016-10-26 19:38:07,177 Node[0] Epoch[8] Validation-top_k_accuracy_10=1.000000


INFO:root:Epoch[8] Validation-top_k_accuracy_20=1.000000


2016-10-26 19:38:07,177 Node[0] Epoch[8] Validation-top_k_accuracy_20=1.000000


INFO:root:Epoch[9] Batch [50]   Speed: 29906.54 samples/sec Train-accuracy=0.990625


2016-10-26 19:38:07,398 Node[0] Epoch[9] Batch [50] Speed: 29906.54 samples/sec Train-accuracy=0.990625


INFO:root:Epoch[9] Batch [50]   Speed: 29906.54 samples/sec Train-top_k_accuracy_5=1.000000


2016-10-26 19:38:07,400 Node[0] Epoch[9] Batch [50] Speed: 29906.54 samples/sec Train-top_k_accuracy_5=1.000000


INFO:root:Epoch[9] Batch [50]   Speed: 29906.54 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:38:07,403 Node[0] Epoch[9] Batch [50] Speed: 29906.54 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[9] Batch [50]   Speed: 29906.54 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:38:07,404 Node[0] Epoch[9] Batch [50] Speed: 29906.54 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[9] Batch [100]  Speed: 27350.43 samples/sec Train-accuracy=0.987969


2016-10-26 19:38:07,641 Node[0] Epoch[9] Batch [100]    Speed: 27350.43 samples/sec Train-accuracy=0.987969


INFO:root:Epoch[9] Batch [100]  Speed: 27350.43 samples/sec Train-top_k_accuracy_5=1.000000


2016-10-26 19:38:07,641 Node[0] Epoch[9] Batch [100]    Speed: 27350.43 samples/sec Train-top_k_accuracy_5=1.000000


INFO:root:Epoch[9] Batch [100]  Speed: 27350.43 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:38:07,644 Node[0] Epoch[9] Batch [100]    Speed: 27350.43 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[9] Batch [100]  Speed: 27350.43 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:38:07,644 Node[0] Epoch[9] Batch [100]    Speed: 27350.43 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[9] Batch [150]  Speed: 28444.46 samples/sec Train-accuracy=0.990469


2016-10-26 19:38:07,871 Node[0] Epoch[9] Batch [150]    Speed: 28444.46 samples/sec Train-accuracy=0.990469


INFO:root:Epoch[9] Batch [150]  Speed: 28444.46 samples/sec Train-top_k_accuracy_5=1.000000


2016-10-26 19:38:07,874 Node[0] Epoch[9] Batch [150]    Speed: 28444.46 samples/sec Train-top_k_accuracy_5=1.000000


INFO:root:Epoch[9] Batch [150]  Speed: 28444.46 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:38:07,875 Node[0] Epoch[9] Batch [150]    Speed: 28444.46 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[9] Batch [150]  Speed: 28444.46 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:38:07,877 Node[0] Epoch[9] Batch [150]    Speed: 28444.46 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[9] Batch [200]  Speed: 28070.18 samples/sec Train-accuracy=0.992969


2016-10-26 19:38:08,105 Node[0] Epoch[9] Batch [200]    Speed: 28070.18 samples/sec Train-accuracy=0.992969


INFO:root:Epoch[9] Batch [200]  Speed: 28070.18 samples/sec Train-top_k_accuracy_5=1.000000


2016-10-26 19:38:08,108 Node[0] Epoch[9] Batch [200]    Speed: 28070.18 samples/sec Train-top_k_accuracy_5=1.000000


INFO:root:Epoch[9] Batch [200]  Speed: 28070.18 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:38:08,111 Node[0] Epoch[9] Batch [200]    Speed: 28070.18 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[9] Batch [200]  Speed: 28070.18 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:38:08,111 Node[0] Epoch[9] Batch [200]    Speed: 28070.18 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[9] Batch [250]  Speed: 27586.22 samples/sec Train-accuracy=0.992344


2016-10-26 19:38:08,345 Node[0] Epoch[9] Batch [250]    Speed: 27586.22 samples/sec Train-accuracy=0.992344


INFO:root:Epoch[9] Batch [250]  Speed: 27586.22 samples/sec Train-top_k_accuracy_5=1.000000


2016-10-26 19:38:08,348 Node[0] Epoch[9] Batch [250]    Speed: 27586.22 samples/sec Train-top_k_accuracy_5=1.000000


INFO:root:Epoch[9] Batch [250]  Speed: 27586.22 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:38:08,349 Node[0] Epoch[9] Batch [250]    Speed: 27586.22 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[9] Batch [250]  Speed: 27586.22 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:38:08,351 Node[0] Epoch[9] Batch [250]    Speed: 27586.22 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[9] Batch [300]  Speed: 28070.16 samples/sec Train-accuracy=0.989062


2016-10-26 19:38:08,581 Node[0] Epoch[9] Batch [300]    Speed: 28070.16 samples/sec Train-accuracy=0.989062


INFO:root:Epoch[9] Batch [300]  Speed: 28070.16 samples/sec Train-top_k_accuracy_5=1.000000


2016-10-26 19:38:08,582 Node[0] Epoch[9] Batch [300]    Speed: 28070.16 samples/sec Train-top_k_accuracy_5=1.000000


INFO:root:Epoch[9] Batch [300]  Speed: 28070.16 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:38:08,584 Node[0] Epoch[9] Batch [300]    Speed: 28070.16 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[9] Batch [300]  Speed: 28070.16 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:38:08,585 Node[0] Epoch[9] Batch [300]    Speed: 28070.16 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[9] Batch [350]  Speed: 28699.55 samples/sec Train-accuracy=0.989531


2016-10-26 19:38:08,809 Node[0] Epoch[9] Batch [350]    Speed: 28699.55 samples/sec Train-accuracy=0.989531


INFO:root:Epoch[9] Batch [350]  Speed: 28699.55 samples/sec Train-top_k_accuracy_5=1.000000


2016-10-26 19:38:08,812 Node[0] Epoch[9] Batch [350]    Speed: 28699.55 samples/sec Train-top_k_accuracy_5=1.000000


INFO:root:Epoch[9] Batch [350]  Speed: 28699.55 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:38:08,813 Node[0] Epoch[9] Batch [350]    Speed: 28699.55 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[9] Batch [350]  Speed: 28699.55 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:38:08,815 Node[0] Epoch[9] Batch [350]    Speed: 28699.55 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[9] Batch [400]  Speed: 28571.43 samples/sec Train-accuracy=0.989375


2016-10-26 19:38:09,040 Node[0] Epoch[9] Batch [400]    Speed: 28571.43 samples/sec Train-accuracy=0.989375


INFO:root:Epoch[9] Batch [400]  Speed: 28571.43 samples/sec Train-top_k_accuracy_5=0.999687


2016-10-26 19:38:09,042 Node[0] Epoch[9] Batch [400]    Speed: 28571.43 samples/sec Train-top_k_accuracy_5=0.999687


INFO:root:Epoch[9] Batch [400]  Speed: 28571.43 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:38:09,043 Node[0] Epoch[9] Batch [400]    Speed: 28571.43 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[9] Batch [400]  Speed: 28571.43 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:38:09,046 Node[0] Epoch[9] Batch [400]    Speed: 28571.43 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[9] Batch [450]  Speed: 28828.81 samples/sec Train-accuracy=0.991406


2016-10-26 19:38:09,269 Node[0] Epoch[9] Batch [450]    Speed: 28828.81 samples/sec Train-accuracy=0.991406


INFO:root:Epoch[9] Batch [450]  Speed: 28828.81 samples/sec Train-top_k_accuracy_5=1.000000


2016-10-26 19:38:09,270 Node[0] Epoch[9] Batch [450]    Speed: 28828.81 samples/sec Train-top_k_accuracy_5=1.000000


INFO:root:Epoch[9] Batch [450]  Speed: 28828.81 samples/sec Train-top_k_accuracy_10=1.000000


2016-10-26 19:38:09,272 Node[0] Epoch[9] Batch [450]    Speed: 28828.81 samples/sec Train-top_k_accuracy_10=1.000000


INFO:root:Epoch[9] Batch [450]  Speed: 28828.81 samples/sec Train-top_k_accuracy_20=1.000000


2016-10-26 19:38:09,273 Node[0] Epoch[9] Batch [450]    Speed: 28828.81 samples/sec Train-top_k_accuracy_20=1.000000


INFO:root:Epoch[9] Resetting Data Iterator


2016-10-26 19:38:09,355 Node[0] Epoch[9] Resetting Data Iterator


INFO:root:Epoch[9] Time cost=2.179


2016-10-26 19:38:09,358 Node[0] Epoch[9] Time cost=2.179


INFO:root:Epoch[9] Validation-accuracy=0.973958


2016-10-26 19:38:09,522 Node[0] Epoch[9] Validation-accuracy=0.973958


INFO:root:Epoch[9] Validation-top_k_accuracy_5=0.999299


2016-10-26 19:38:09,523 Node[0] Epoch[9] Validation-top_k_accuracy_5=0.999299


INFO:root:Epoch[9] Validation-top_k_accuracy_10=1.000000


2016-10-26 19:38:09,525 Node[0] Epoch[9] Validation-top_k_accuracy_10=1.000000


INFO:root:Epoch[9] Validation-top_k_accuracy_20=1.000000


2016-10-26 19:38:09,526 Node[0] Epoch[9] Validation-top_k_accuracy_20=1.000000

可以看到，验证集上的准确率接近于1，这说明我们的安装过程是成功的。

总结一下期间遇到的错误：

1.CMake Error: The following variables are used in this project, but they are set to NOTFOUND.

Please set them or make sure they are set and tested correctly in the CMake files:

CUDA_cublas_LIBRARY (ADVANCED)

linked by target "mxnet" in directory G:/OpenSource/mxnet

linked by target "mxnet" in directory G:/OpenSource/mxnet

CUDA_cublas_device_LIBRARY (ADVANCED)

linked by target "mxnet" in directory G:/OpenSource/mxnet

linked by target "mxnet" in directory G:/OpenSource/mxnet

CUDA_curand_LIBRARY (ADVANCED)

linked by target "mxnet" in directory G:/OpenSource/mxnet

linked by target "mxnet" in directory G:/OpenSource/mxnet

解决方法：参考github上的issue，换成64位编译器。

2.无法打开包括文件:opencv2.hpp 解决方法：

在项目属性页的VC++标签页中的包含目录选项中加入opencv的头文件路径G:\opencv\build\include即可。

3.错误 5373 error LNK2001: 无法解析的外部符号 "int __cdecl cv::_interlockedExchangeAdd(int *,int)"(?_interlockedExchangeAdd@cv@@YAHPEAHH@Z)

解决方法：

在项目属性页的标签页中的链接器下的附加依赖项属性中加入opencv的库文件G:\opencv\build\x64\vc12\lib\opencv_core2413.lib

参考

1.mxnet配置安装

2.https://github.com/dmlc/mxnet/issues/655

订阅：博文 (Atom)