MNIST 模型测试

来源：互联网发布：js 0 100两位小数编辑：程序博客网时间：2024/06/04 20:04

网上很多训练mnist模型的教程，但是没有将怎么测试，其实在caffe根目录输入：

:/caffe-master$ ./build/tools/caffe.bin test \

-model examples/mnist/lenet_train_test.prototxt \

-weights examples/mnist/lenet_iter_10000.caffemodel \

-iterations 100

命令行解释：

./build/tools/caffe.bin test ,表示只作预测（前向传播计算），不进行参数更新（后向传播计算）

-model examples/mnist/lenet_train_test.prototxt ，指定模型描述文本文件

-weights examples/mnist/lenet_iter_10000.caffemodel ，指定模型预先训练好的权值文件

-iterations 100 ，指定测试迭代次数，参与测试的样例数目为（iterations * batch_size）,

batch_size在model prototxt 里定义，设为100时刚好覆盖10000个测试样本

以下是输出结果：

I1116 10:44:50.146291 3247 caffe.cpp:279] Use CPU.

I1116 10:44:51.683002 3247 net.cpp:322] The NetState phase (1) differed from the phase (0) specified by a rule in layer mnist //只测试
I1116 10:44:51.683157 3247 net.cpp:58] Initializing net from parameters:
name: "LeNet"
state {
phase: TEST
level: 0
stage: ""
}
layer {
name: "mnist"
type: "Data"
top: "data"
top: "label"
include {
    phase: TEST
}
transform_param {
    scale: 0.00390625
}
data_param {
    source: "examples/mnist/mnist_test_lmdb"
    batch_size: 100
    backend: LMDB
}
}
layer {
name: "conv1"
type: "Convolution"
bottom: "data"
top: "conv1"
param {
    lr_mult: 1
}
param {
    lr_mult: 2
}
convolution_param {
    num_output: 20
    kernel_size: 5
    stride: 1
    weight_filler {
      type: "xavier"
    }
    bias_filler {
      type: "constant"
    }
}
}
layer {
name: "pool1"
type: "Pooling"
bottom: "conv1"
top: "pool1"
pooling_param {
    pool: MAX
    kernel_size: 2
    stride: 2
}
}
layer {
name: "conv2"
type: "Convolution"
bottom: "pool1"
top: "conv2"
param {
    lr_mult: 1
}
param {
    lr_mult: 2
}
convolution_param {
    num_output: 50
    kernel_size: 5
    stride: 1
    weight_filler {
      type: "xavier"
    }
    bias_filler {
      type: "constant"
    }
}
}
layer {
name: "pool2"
type: "Pooling"
bottom: "conv2"
top: "pool2"
pooling_param {
    pool: MAX
    kernel_size: 2
    stride: 2
}
}
layer {
name: "ip1"
type: "InnerProduct"
bottom: "pool2"
top: "ip1"
param {
    lr_mult: 1
}
param {
    lr_mult: 2
}
inner_product_param {
    num_output: 500
    weight_filler {
      type: "xavier"
    }
    bias_filler {
      type: "constant"
    }
}
}
layer {
name: "relu1"
type: "ReLU"
bottom: "ip1"
top: "ip1"
}
layer {
name: "ip2"
type: "InnerProduct"
bottom: "ip1"
top: "ip2"
param {
    lr_mult: 1
}
param {
    lr_mult: 2
}
inner_product_param {
    num_output: 10
    weight_filler {
      type: "xavier"
    }
    bias_filler {
      type: "constant"
    }
}
}
layer {
name: "accuracy"
type: "Accuracy"
bottom: "ip2"
bottom: "label"
top: "accuracy"
include {
    phase: TEST
}
}
layer {
name: "loss"
type: "SoftmaxWithLoss"
bottom: "ip2"
bottom: "label"
top: "loss"
}
I1116 10:44:51.683599 3247 layer_factory.hpp:77] Creating layer mnist
I1116 10:44:51.709772 3247 net.cpp:100] Creating Layer mnist
I1116 10:44:51.709861 3247 net.cpp:408] mnist -> data
I1116 10:44:51.709889 3247 net.cpp:408] mnist -> label
I1116 10:44:51.734688 3260 db_lmdb.cpp:35] Opened lmdb examples/mnist/mnist_test_lmdb   //打开测试数据

I1116 10:44:51.741736 3247 data_layer.cpp:41] output data size: 100,1,28,28 //data四维数组尺寸（100,1,28,28）

I1116 10:44:51.742224 3247 net.cpp:150] Setting up mnist

I1116 10:44:51.742274 3247 net.cpp:157] Top shape: 100 1 28 28 (78400)I1116 10:44:51.742281 3247net.cpp:157] Top shape: 100 (100)

I1116 10:44:51.742300 3247 net.cpp:165] Memory required for data: 314000

I1116 10:44:51.742311 3247 layer_factory.hpp:77] Creating layer label_mnist_1_split
I1116 10:44:51.742331 3247 net.cpp:100] Creating Layer label_mnist_1_split
I1116 10:44:51.742339 3247 net.cpp:434] label_mnist_1_split <- label
I1116 10:44:51.742365 3247 net.cpp:408] label_mnist_1_split -> label_mnist_1_split_0
I1116 10:44:51.742377 3247 net.cpp:408] label_mnist_1_split -> label_mnist_1_split_1
I1116 10:44:51.742401 3247 net.cpp:150] Setting up label_mnist_1_split
I1116 10:44:51.742409 3247 net.cpp:157] Top shape: 100 (100)
I1116 10:44:51.742415 3247 net.cpp:157] Top shape: 100 (100)
I1116 10:44:51.742420 3247 net.cpp:165] Memory required for data: 314800
I1116 10:44:51.742426 3247 layer_factory.hpp:77] Creating layer conv1
I1116 10:44:51.742442 3247 net.cpp:100] Creating Layer conv1
I1116 10:44:51.742449 3247 net.cpp:434] conv1 <- data
I1116 10:44:51.742458 3247 net.cpp:408] conv1 -> conv1
I1116 10:44:51.743266 3261 blocking_queue.cpp:50] Waiting for data
I1116 10:44:53.022564 3247 net.cpp:150] Setting up conv1
I1116 10:44:53.022596 3247 net.cpp:157] Top shape: 100 20 24 24 (1152000)
I1116 10:44:53.022616 3247 net.cpp:165] Memory required for data: 4922800
I1116 10:44:53.022687 3247 layer_factory.hpp:77] Creating layer pool1
I1116 10:44:53.022701 3247 net.cpp:100] Creating Layer pool1
I1116 10:44:53.022711 3247 net.cpp:434] pool1 <- conv1
I1116 10:44:53.022718 3247 net.cpp:408] pool1 -> pool1
I1116 10:44:53.022737 3247 net.cpp:150] Setting up pool1
I1116 10:44:53.022745 3247 net.cpp:157] Top shape: 100 20 12 12 (288000)
I1116 10:44:53.022750 3247 net.cpp:165] Memory required for data: 6074800
I1116 10:44:53.022756 3247 layer_factory.hpp:77] Creating layer conv2
I1116 10:44:53.022768 3247 net.cpp:100] Creating Layer conv2
I1116 10:44:53.022774 3247 net.cpp:434] conv2 <- pool1
I1116 10:44:53.022797 3247 net.cpp:408] conv2 -> conv2
I1116 10:44:53.023576 3247 net.cpp:150] Setting up conv2
I1116 10:44:53.023591 3247 net.cpp:157] Top shape: 100 50 8 8 (320000)
I1116 10:44:53.023597 3247 net.cpp:165] Memory required for data: 7354800
I1116 10:44:53.023622 3247 layer_factory.hpp:77] Creating layer pool2
I1116 10:44:53.023629 3247 net.cpp:100] Creating Layer pool2
I1116 10:44:53.023635 3247 net.cpp:434] pool2 <- conv2
I1116 10:44:53.023643 3247 net.cpp:408] pool2 -> pool2
I1116 10:44:53.023665 3247 net.cpp:150] Setting up pool2
I1116 10:44:53.023674 3247 net.cpp:157] Top shape: 100 50 4 4 (80000)
I1116 10:44:53.023679 3247 net.cpp:165] Memory required for data: 7674800
I1116 10:44:53.023685 3247 layer_factory.hpp:77] Creating layer ip1
I1116 10:44:53.023696 3247 net.cpp:100] Creating Layer ip1
I1116 10:44:53.023702 3247 net.cpp:434] ip1 <- pool2
I1116 10:44:53.023710 3247 net.cpp:408] ip1 -> ip1
I1116 10:44:53.026619 3247 net.cpp:150] Setting up ip1
I1116 10:44:53.026635 3247 net.cpp:157] Top shape: 100 500 (50000)
I1116 10:44:53.026641 3247 net.cpp:165] Memory required for data: 7874800
I1116 10:44:53.026666 3247 layer_factory.hpp:77] Creating layer relu1
I1116 10:44:53.026675 3247 net.cpp:100] Creating Layer relu1
I1116 10:44:53.026681 3247 net.cpp:434] relu1 <- ip1
I1116 10:44:53.026688 3247 net.cpp:395] relu1 -> ip1 (in-place)
I1116 10:44:53.026942 3247 net.cpp:150] Setting up relu1
I1116 10:44:53.026954 3247 net.cpp:157] Top shape: 100 500 (50000)
I1116 10:44:53.026960 3247 net.cpp:165] Memory required for data: 8074800
I1116 10:44:53.026980 3247 layer_factory.hpp:77] Creating layer ip2
I1116 10:44:53.026989 3247 net.cpp:100] Creating Layer ip2
I1116 10:44:53.026995 3247 net.cpp:434] ip2 <- ip1
I1116 10:44:53.027017 3247 net.cpp:408] ip2 -> ip2
I1116 10:44:53.027065 3247 net.cpp:150] Setting up ip2
I1116 10:44:53.027086 3247 net.cpp:157] Top shape: 100 10 (1000)
I1116 10:44:53.027091 3247 net.cpp:165] Memory required for data: 8078800
I1116 10:44:53.027112 3247 layer_factory.hpp:77] Creating layer ip2_ip2_0_split
I1116 10:44:53.027120 3247 net.cpp:100] Creating Layer ip2_ip2_0_split
I1116 10:44:53.027125 3247 net.cpp:434] ip2_ip2_0_split <- ip2
I1116 10:44:53.027132 3247 net.cpp:408] ip2_ip2_0_split -> ip2_ip2_0_split_0
I1116 10:44:53.027140 3247 net.cpp:408] ip2_ip2_0_split -> ip2_ip2_0_split_1
I1116 10:44:53.027153 3247 net.cpp:150] Setting up ip2_ip2_0_split
I1116 10:44:53.027173 3247 net.cpp:157] Top shape: 100 10 (1000)
I1116 10:44:53.027179 3247 net.cpp:157] Top shape: 100 10 (1000)
I1116 10:44:53.027184 3247 net.cpp:165] Memory required for data: 8086800
I1116 10:44:53.027189 3247 layer_factory.hpp:77] Creating layer accuracy
I1116 10:44:53.027199 3247 net.cpp:100] Creating Layer accuracy

I1116 10:44:53.027205 3247 net.cpp:434] accuracy <- ip2_ip2_0_split_0 //兵分两路~split_0给了accuracy层

I1116 10:44:53.027211 3247 net.cpp:434] accuracy <- label_mnist_1_split_0

I1116 10:44:53.027218 3247 net.cpp:408] accuracy -> accuracy
I1116 10:44:53.027775 3247 net.cpp:150] Setting up accuracy
I1116 10:44:53.027791 3247 net.cpp:157] Top shape: (1)
I1116 10:44:53.027797 3247 net.cpp:165] Memory required for data: 8086804
I1116 10:44:53.027817 3247 layer_factory.hpp:77] Creating layer loss
I1116 10:44:53.027825 3247 net.cpp:100] Creating Layer loss
I1116 10:44:53.027832 3247 net.cpp:434] loss <- ip2_ip2_0_split_1 //兵分两路~split_1给了loss层
I1116 10:44:53.027859 3247 net.cpp:434] loss <- label_mnist_1_split_1
I1116 10:44:53.027868 3247 net.cpp:408] loss -> loss
I1116 10:44:53.027899 3247 layer_factory.hpp:77] Creating layer loss
I1116 10:44:53.028553 3247 net.cpp:150] Setting up loss
I1116 10:44:53.028569 3247 net.cpp:157] Top shape: (1)
I1116 10:44:53.028575 3247 net.cpp:160]     with loss weight 1
I1116 10:44:53.028594 3247 net.cpp:165] Memory required for data: 8086808
I1116 10:44:53.028599 3247 net.cpp:226] loss needs backward computation.
I1116 10:44:53.028605 3247 net.cpp:228] accuracy does not need backward computation.
I1116 10:44:53.028611 3247 net.cpp:226] ip2_ip2_0_split needs backward computation.
I1116 10:44:53.028616 3247 net.cpp:226] ip2 needs backward computation.
I1116 10:44:53.028621 3247 net.cpp:226] relu1 needs backward computation.
I1116 10:44:53.028626 3247 net.cpp:226] ip1 needs backward computation.
I1116 10:44:53.028631 3247 net.cpp:226] pool2 needs backward computation.
I1116 10:44:53.028636 3247 net.cpp:226] conv2 needs backward computation.
I1116 10:44:53.028641 3247 net.cpp:226] pool1 needs backward computation.
I1116 10:44:53.028646 3247 net.cpp:226] conv1 needs backward computation.
I1116 10:44:53.028651 3247 net.cpp:228] label_mnist_1_split does not need backward computation.
I1116 10:44:53.028657 3247 net.cpp:228] mnist does not need backward computation.
I1116 10:44:53.028661 3247 net.cpp:270] This network produces output accuracy
I1116 10:44:53.028667 3247 net.cpp:270] This network produces output loss
I1116 10:44:53.028678 3247 net.cpp:283] Network initialization done.
I1116 10:44:53.057978 3247 caffe.cpp:285] Running for 100 iterations.
I1116 10:44:53.126924 3247 caffe.cpp:308] Batch 0, accuracy = 1
I1116 10:44:53.126963 3247 caffe.cpp:308] Batch 0, loss = 0.0061212
I1116 10:44:53.168745 3247 caffe.cpp:308] Batch 1, accuracy = 0.99
I1116 10:44:53.168772 3247 caffe.cpp:308] Batch 1, loss = 0.0133515
I1116 10:44:53.210995 3247 caffe.cpp:308] Batch 2, accuracy = 0.99
I1116 10:44:53.211024 3247 caffe.cpp:308] Batch 2, loss = 0.0117903
I1116 10:44:53.253181 3247 caffe.cpp:308] Batch 3, accuracy = 0.99
I1116 10:44:53.253211 3247 caffe.cpp:308] Batch 3, loss = 0.0317178
I1116 10:44:53.294770 3247 caffe.cpp:308] Batch 4, accuracy = 0.99
I1116 10:44:53.294795 3247 caffe.cpp:308] Batch 4, loss = 0.0568775
I1116 10:44:53.336119 3247 caffe.cpp:308] Batch 5, accuracy = 0.99
I1116 10:44:53.336146 3247 caffe.cpp:308] Batch 5, loss = 0.0253471
I1116 10:44:53.377841 3247 caffe.cpp:308] Batch 6, accuracy = 0.97
I1116 10:44:53.377868 3247 caffe.cpp:308] Batch 6, loss = 0.0566137
I1116 10:44:53.419960 3247 caffe.cpp:308] Batch 7, accuracy = 0.99
I1116 10:44:53.419988 3247 caffe.cpp:308] Batch 7, loss = 0.0153878
I1116 10:44:53.462306 3247 caffe.cpp:308] Batch 8, accuracy = 1
I1116 10:44:53.462332 3247 caffe.cpp:308] Batch 8, loss = 0.00515131
I1116 10:44:53.504509 3247 caffe.cpp:308] Batch 9, accuracy = 0.99
I1116 10:44:53.504535 3247 caffe.cpp:308] Batch 9, loss = 0.0175177
I1116 10:44:53.547013 3247 caffe.cpp:308] Batch 10, accuracy = 0.98
I1116 10:44:53.547040 3247 caffe.cpp:308] Batch 10, loss = 0.0777676
I1116 10:44:53.589557 3247 caffe.cpp:308] Batch 11, accuracy = 0.98
I1116 10:44:53.589584 3247 caffe.cpp:308] Batch 11, loss = 0.0444185
I1116 10:44:53.632716 3247 caffe.cpp:308] Batch 12, accuracy = 0.95
I1116 10:44:53.632742 3247 caffe.cpp:308] Batch 12, loss = 0.148465
I1116 10:44:53.674947 3247 caffe.cpp:308] Batch 13, accuracy = 0.98
I1116 10:44:53.674973 3247 caffe.cpp:308] Batch 13, loss = 0.0589911
I1116 10:44:53.717653 3247 caffe.cpp:308] Batch 14, accuracy = 1
I1116 10:44:53.717679 3247 caffe.cpp:308] Batch 14, loss = 0.00966478
I1116 10:44:53.759479 3247 caffe.cpp:308] Batch 15, accuracy = 0.99
I1116 10:44:53.759534 3247 caffe.cpp:308] Batch 15, loss = 0.0354297
I1116 10:44:53.802633 3247 caffe.cpp:308] Batch 16, accuracy = 0.99
I1116 10:44:53.802661 3247 caffe.cpp:308] Batch 16, loss = 0.0254676
I1116 10:44:53.845255 3247 caffe.cpp:308] Batch 17, accuracy = 0.99
I1116 10:44:53.845299 3247 caffe.cpp:308] Batch 17, loss = 0.0211856
I1116 10:44:53.886400 3247 caffe.cpp:308] Batch 18, accuracy = 0.99
I1116 10:44:53.886426 3247 caffe.cpp:308] Batch 18, loss = 0.0162245
I1116 10:44:53.927749 3247 caffe.cpp:308] Batch 19, accuracy = 0.98
I1116 10:44:53.927774 3247 caffe.cpp:308] Batch 19, loss = 0.0700603
I1116 10:44:53.969600 3247 caffe.cpp:308] Batch 20, accuracy = 0.98
I1116 10:44:53.969645 3247 caffe.cpp:308] Batch 20, loss = 0.0812266
I1116 10:44:54.011916 3247 caffe.cpp:308] Batch 21, accuracy = 0.97
I1116 10:44:54.011941 3247 caffe.cpp:308] Batch 21, loss = 0.0836177
I1116 10:44:54.054926 3247 caffe.cpp:308] Batch 22, accuracy = 0.99
I1116 10:44:54.054952 3247 caffe.cpp:308] Batch 22, loss = 0.0409367
I1116 10:44:54.096271 3247 caffe.cpp:308] Batch 23, accuracy = 0.98
I1116 10:44:54.096295 3247 caffe.cpp:308] Batch 23, loss = 0.0364728
I1116 10:44:54.137509 3247 caffe.cpp:308] Batch 24, accuracy = 0.99
I1116 10:44:54.137537 3247 caffe.cpp:308] Batch 24, loss = 0.0299302
I1116 10:44:54.179723 3247 caffe.cpp:308] Batch 25, accuracy = 0.99
I1116 10:44:54.179762 3247 caffe.cpp:308] Batch 25, loss = 0.0700943
I1116 10:44:54.222487 3247 caffe.cpp:308] Batch 26, accuracy = 0.99
I1116 10:44:54.222512 3247 caffe.cpp:308] Batch 26, loss = 0.110037
I1116 10:44:54.264148 3247 caffe.cpp:308] Batch 27, accuracy = 1
I1116 10:44:54.264174 3247 caffe.cpp:308] Batch 27, loss = 0.0183853
I1116 10:44:54.305537 3247 caffe.cpp:308] Batch 28, accuracy = 0.99
I1116 10:44:54.305567 3247 caffe.cpp:308] Batch 28, loss = 0.0425091
I1116 10:44:54.347069 3247 caffe.cpp:308] Batch 29, accuracy = 0.96
I1116 10:44:54.347095 3247 caffe.cpp:308] Batch 29, loss = 0.137452
I1116 10:44:54.388613 3247 caffe.cpp:308] Batch 30, accuracy = 0.99
I1116 10:44:54.388655 3247 caffe.cpp:308] Batch 30, loss = 0.0188697
I1116 10:44:54.431311 3247 caffe.cpp:308] Batch 31, accuracy = 1
I1116 10:44:54.431339 3247 caffe.cpp:308] Batch 31, loss = 0.00298686
I1116 10:44:54.474185 3247 caffe.cpp:308] Batch 32, accuracy = 1
I1116 10:44:54.474213 3247 caffe.cpp:308] Batch 32, loss = 0.00986821
I1116 10:44:54.516254 3247 caffe.cpp:308] Batch 33, accuracy = 1
I1116 10:44:54.516280 3247 caffe.cpp:308] Batch 33, loss = 0.00496284
I1116 10:44:54.558568 3247 caffe.cpp:308] Batch 34, accuracy = 0.98
I1116 10:44:54.558593 3247 caffe.cpp:308] Batch 34, loss = 0.0732583
I1116 10:44:54.600500 3247 caffe.cpp:308] Batch 35, accuracy = 0.95
I1116 10:44:54.600528 3247 caffe.cpp:308] Batch 35, loss = 0.159537
I1116 10:44:54.643375 3247 caffe.cpp:308] Batch 36, accuracy = 1
I1116 10:44:54.643406 3247 caffe.cpp:308] Batch 36, loss = 0.00401762
I1116 10:44:54.685214 3247 caffe.cpp:308] Batch 37, accuracy = 0.99
I1116 10:44:54.685256 3247 caffe.cpp:308] Batch 37, loss = 0.0490689
I1116 10:44:54.727855 3247 caffe.cpp:308] Batch 38, accuracy = 0.99
I1116 10:44:54.727882 3247 caffe.cpp:308] Batch 38, loss = 0.0382032
I1116 10:44:54.769639 3247 caffe.cpp:308] Batch 39, accuracy = 0.99
I1116 10:44:54.769665 3247 caffe.cpp:308] Batch 39, loss = 0.0403106
I1116 10:44:54.811323 3247 caffe.cpp:308] Batch 40, accuracy = 1
I1116 10:44:54.811364 3247 caffe.cpp:308] Batch 40, loss = 0.0177203
I1116 10:44:54.852814 3247 caffe.cpp:308] Batch 41, accuracy = 0.98
I1116 10:44:54.852843 3247 caffe.cpp:308] Batch 41, loss = 0.0884625
I1116 10:44:54.894551 3247 caffe.cpp:308] Batch 42, accuracy = 1
I1116 10:44:54.894604 3247 caffe.cpp:308] Batch 42, loss = 0.0325108
I1116 10:44:54.936128 3247 caffe.cpp:308] Batch 43, accuracy = 0.99
I1116 10:44:54.936153 3247 caffe.cpp:308] Batch 43, loss = 0.0148244
I1116 10:44:54.978384 3247 caffe.cpp:308] Batch 44, accuracy = 0.99
I1116 10:44:54.978411 3247 caffe.cpp:308] Batch 44, loss = 0.0352546
I1116 10:44:55.020856 3247 caffe.cpp:308] Batch 45, accuracy = 0.98
I1116 10:44:55.020906 3247 caffe.cpp:308] Batch 45, loss = 0.0425254
I1116 10:44:55.063021 3247 caffe.cpp:308] Batch 46, accuracy = 1
I1116 10:44:55.063052 3247 caffe.cpp:308] Batch 46, loss = 0.00386924
I1116 10:44:55.105479 3247 caffe.cpp:308] Batch 47, accuracy = 0.99
I1116 10:44:55.105521 3247 caffe.cpp:308] Batch 47, loss = 0.0165543
I1116 10:44:55.147413 3247 caffe.cpp:308] Batch 48, accuracy = 0.95
I1116 10:44:55.147451 3247 caffe.cpp:308] Batch 48, loss = 0.0724568
I1116 10:44:55.189774 3247 caffe.cpp:308] Batch 49, accuracy = 0.99
I1116 10:44:55.189800 3247 caffe.cpp:308] Batch 49, loss = 0.0166493
I1116 10:44:55.231638 3247 caffe.cpp:308] Batch 50, accuracy = 1
I1116 10:44:55.231664 3247 caffe.cpp:308] Batch 50, loss = 0.000193432
I1116 10:44:55.273180 3247 caffe.cpp:308] Batch 51, accuracy = 1
I1116 10:44:55.273211 3247 caffe.cpp:308] Batch 51, loss = 0.00549245
I1116 10:44:55.315142 3247 caffe.cpp:308] Batch 52, accuracy = 1
I1116 10:44:55.315170 3247 caffe.cpp:308] Batch 52, loss = 0.00735993
I1116 10:44:55.357081 3247 caffe.cpp:308] Batch 53, accuracy = 1
I1116 10:44:55.357106 3247 caffe.cpp:308] Batch 53, loss = 0.00249736
I1116 10:44:55.398986 3247 caffe.cpp:308] Batch 54, accuracy = 1
I1116 10:44:55.399014 3247 caffe.cpp:308] Batch 54, loss = 0.00380324
I1116 10:44:55.441495 3247 caffe.cpp:308] Batch 55, accuracy = 1
I1116 10:44:55.441521 3247 caffe.cpp:308] Batch 55, loss = 0.000834018
I1116 10:44:55.484167 3247 caffe.cpp:308] Batch 56, accuracy = 1
I1116 10:44:55.484194 3247 caffe.cpp:308] Batch 56, loss = 0.0137293
I1116 10:44:55.526473 3247 caffe.cpp:308] Batch 57, accuracy = 1
I1116 10:44:55.526501 3247 caffe.cpp:308] Batch 57, loss = 0.00453061
I1116 10:44:55.567912 3247 caffe.cpp:308] Batch 58, accuracy = 1
I1116 10:44:55.567937 3247 caffe.cpp:308] Batch 58, loss = 0.00564778
I1116 10:44:55.610164 3247 caffe.cpp:308] Batch 59, accuracy = 0.97
I1116 10:44:55.610190 3247 caffe.cpp:308] Batch 59, loss = 0.1065
I1116 10:44:55.651887 3247 caffe.cpp:308] Batch 60, accuracy = 1
I1116 10:44:55.651913 3247 caffe.cpp:308] Batch 60, loss = 0.00507791
I1116 10:44:55.693601 3247 caffe.cpp:308] Batch 61, accuracy = 1
I1116 10:44:55.693630 3247 caffe.cpp:308] Batch 61, loss = 0.00784779
I1116 10:44:55.735671 3247 caffe.cpp:308] Batch 62, accuracy = 1
I1116 10:44:55.735697 3247 caffe.cpp:308] Batch 62, loss = 2.75084e-05
I1116 10:44:55.778391 3247 caffe.cpp:308] Batch 63, accuracy = 1
I1116 10:44:55.778417 3247 caffe.cpp:308] Batch 63, loss = 8.72722e-05
I1116 10:44:55.821347 3247 caffe.cpp:308] Batch 64, accuracy = 1
I1116 10:44:55.821372 3247 caffe.cpp:308] Batch 64, loss = 0.000635555
I1116 10:44:55.863615 3247 caffe.cpp:308] Batch 65, accuracy = 0.96
I1116 10:44:55.863641 3247 caffe.cpp:308] Batch 65, loss = 0.118228
I1116 10:44:55.905485 3247 caffe.cpp:308] Batch 66, accuracy = 0.98
I1116 10:44:55.905514 3247 caffe.cpp:308] Batch 66, loss = 0.0594917
I1116 10:44:55.947041 3247 caffe.cpp:308] Batch 67, accuracy = 0.99
I1116 10:44:55.947067 3247 caffe.cpp:308] Batch 67, loss = 0.0321053
I1116 10:44:55.988890 3247 caffe.cpp:308] Batch 68, accuracy = 1
I1116 10:44:55.988915 3247 caffe.cpp:308] Batch 68, loss = 0.00137083
I1116 10:44:56.030824 3247 caffe.cpp:308] Batch 69, accuracy = 1
I1116 10:44:56.030849 3247 caffe.cpp:308] Batch 69, loss = 0.00270546
I1116 10:44:56.073659 3247 caffe.cpp:308] Batch 70, accuracy = 1
I1116 10:44:56.073686 3247 caffe.cpp:308] Batch 70, loss = 0.00114084
I1116 10:44:56.116312 3247 caffe.cpp:308] Batch 71, accuracy = 1
I1116 10:44:56.116336 3247 caffe.cpp:308] Batch 71, loss = 0.000395572
I1116 10:44:56.158524 3247 caffe.cpp:308] Batch 72, accuracy = 1
I1116 10:44:56.158550 3247 caffe.cpp:308] Batch 72, loss = 0.00495989
I1116 10:44:56.200970 3247 caffe.cpp:308] Batch 73, accuracy = 1
I1116 10:44:56.200997 3247 caffe.cpp:308] Batch 73, loss = 0.000111445
I1116 10:44:56.243125 3247 caffe.cpp:308] Batch 74, accuracy = 1
I1116 10:44:56.243151 3247 caffe.cpp:308] Batch 74, loss = 0.00227753
I1116 10:44:56.284889 3247 caffe.cpp:308] Batch 75, accuracy = 1
I1116 10:44:56.284915 3247 caffe.cpp:308] Batch 75, loss = 0.00349699
I1116 10:44:56.327323 3247 caffe.cpp:308] Batch 76, accuracy = 1
I1116 10:44:56.327347 3247 caffe.cpp:308] Batch 76, loss = 0.000260342
I1116 10:44:56.370226 3247 caffe.cpp:308] Batch 77, accuracy = 1
I1116 10:44:56.370260 3247 caffe.cpp:308] Batch 77, loss = 0.000231701
I1116 10:44:56.413427 3247 caffe.cpp:308] Batch 78, accuracy = 1
I1116 10:44:56.413452 3247 caffe.cpp:308] Batch 78, loss = 0.0027974
I1116 10:44:56.455852 3247 caffe.cpp:308] Batch 79, accuracy = 1
I1116 10:44:56.455878 3247 caffe.cpp:308] Batch 79, loss = 0.00245169
I1116 10:44:56.497237 3247 caffe.cpp:308] Batch 80, accuracy = 0.99
I1116 10:44:56.497263 3247 caffe.cpp:308] Batch 80, loss = 0.0126951
I1116 10:44:56.538885 3247 caffe.cpp:308] Batch 81, accuracy = 1
I1116 10:44:56.538910 3247 caffe.cpp:308] Batch 81, loss = 0.00183047
I1116 10:44:56.580886 3247 caffe.cpp:308] Batch 82, accuracy = 1
I1116 10:44:56.580912 3247 caffe.cpp:308] Batch 82, loss = 0.00443107
I1116 10:44:56.624069 3247 caffe.cpp:308] Batch 83, accuracy = 1
I1116 10:44:56.624095 3247 caffe.cpp:308] Batch 83, loss = 0.0138792
I1116 10:44:56.666620 3247 caffe.cpp:308] Batch 84, accuracy = 0.99
I1116 10:44:56.666646 3247 caffe.cpp:308] Batch 84, loss = 0.018512
I1116 10:44:56.709748 3247 caffe.cpp:308] Batch 85, accuracy = 0.99
I1116 10:44:56.709771 3247 caffe.cpp:308] Batch 85, loss = 0.0235319
I1116 10:44:56.752324 3247 caffe.cpp:308] Batch 86, accuracy = 1
I1116 10:44:56.752351 3247 caffe.cpp:308] Batch 86, loss = 7.07973e-05
I1116 10:44:56.795102 3247 caffe.cpp:308] Batch 87, accuracy = 1
I1116 10:44:56.795130 3247 caffe.cpp:308] Batch 87, loss = 9.06585e-05
I1116 10:44:56.836699 3247 caffe.cpp:308] Batch 88, accuracy = 1
I1116 10:44:56.836726 3247 caffe.cpp:308] Batch 88, loss = 0.000124428
I1116 10:44:56.879022 3247 caffe.cpp:308] Batch 89, accuracy = 1
I1116 10:44:56.879047 3247 caffe.cpp:308] Batch 89, loss = 5.27109e-05
I1116 10:44:56.920341 3247 caffe.cpp:308] Batch 90, accuracy = 0.96
I1116 10:44:56.920367 3247 caffe.cpp:308] Batch 90, loss = 0.0960243
I1116 10:44:56.962316 3247 caffe.cpp:308] Batch 91, accuracy = 1
I1116 10:44:56.962340 3247 caffe.cpp:308] Batch 91, loss = 3.48874e-05
I1116 10:44:57.004240 3247 caffe.cpp:308] Batch 92, accuracy = 1
I1116 10:44:57.004266 3247 caffe.cpp:308] Batch 92, loss = 0.000362797
I1116 10:44:57.046540 3247 caffe.cpp:308] Batch 93, accuracy = 1
I1116 10:44:57.046566 3247 caffe.cpp:308] Batch 93, loss = 0.000916503
I1116 10:44:57.088037 3247 caffe.cpp:308] Batch 94, accuracy = 1
I1116 10:44:57.088065 3247 caffe.cpp:308] Batch 94, loss = 0.00034051
I1116 10:44:57.129601 3247 caffe.cpp:308] Batch 95, accuracy = 1
I1116 10:44:57.129624 3247 caffe.cpp:308] Batch 95, loss = 0.0044833
I1116 10:44:57.171982 3247 caffe.cpp:308] Batch 96, accuracy = 0.98
I1116 10:44:57.172008 3247 caffe.cpp:308] Batch 96, loss = 0.0463236
I1116 10:44:57.213846 3247 caffe.cpp:308] Batch 97, accuracy = 0.98
I1116 10:44:57.213872 3247 caffe.cpp:308] Batch 97, loss = 0.076892
I1116 10:44:57.255475 3247 caffe.cpp:308] Batch 98, accuracy = 1
I1116 10:44:57.255499 3247 caffe.cpp:308] Batch 98, loss = 0.00295341
I1116 10:44:57.297261 3247 caffe.cpp:308] Batch 99, accuracy = 1
I1116 10:44:57.297286 3247 caffe.cpp:308] Batch 99, loss = 0.0056333   //100批，每批100数。刚好10000
I1116 10:44:57.297293 3247 caffe.cpp:313] Loss: 0.0284559
I1116 10:44:57.297319 3247 caffe.cpp:325] accuracy = 0.9912     //最终精度
I1116 10:44:57.297343 3247 caffe.cpp:325] loss = 0.0284559 (* 1 = 0.0284559 loss)

0 0