http://caffe.berkeleyvision.org/gathered/examples/mnist.html
jwy@jwy:~/caffe$ ./data/mnist/get_mnist.sh
jwy@jwy:~/caffe$ ./examples/mnist/create_mnist.sh
Creating lmdb...
I0526 20:47:44.249351 2884 db_lmdb.cpp:35] Opened lmdb examples/mnist/mnist_train_lmdb
I0526 20:47:44.249537 2884 convert_mnist_data.cpp:88] A total of 60000 items.
I0526 20:47:44.249545 2884 convert_mnist_data.cpp:89] Rows: 28 Cols: 28
I0526 20:47:49.688385 2884 convert_mnist_data.cpp:108] Processed 60000 files.
I0526 20:47:50.344051 2908 db_lmdb.cpp:35] Opened lmdb examples/mnist/mnist_test_lmdb
I0526 20:47:50.344348 2908 convert_mnist_data.cpp:88] A total of 10000 items.
I0526 20:47:50.344357 2908 convert_mnist_data.cpp:89] Rows: 28 Cols: 28
I0526 20:47:51.249704 2908 convert_mnist_data.cpp:108] Processed 10000 files.
Done.
jwy@jwy:~/caffe$ ./examples/mnist/train_lenet.sh
I0526 20:53:03.452682 2966 caffe.cpp:204] Using GPUs 0
I0526 20:53:03.653373 2966 caffe.cpp:209] GPU 0: GeForce GTX 1050 Ti
I0526 20:53:04.087340 2966 solver.cpp:45] Initializing solver from parameters:
test_iter: 100
test_interval: 500
base_lr: 0.01
display: 100
max_iter: 10000
lr_policy: "inv"
gamma: 0.0001
power: 0.75
momentum: 0.9
weight_decay: 0.0005
snapshot: 5000
snapshot_prefix: "examples/mnist/lenet"
solver_mode: GPU
device_id: 0
net: "examples/mnist/lenet_train_test.prototxt"
train_state {
level: 0
stage: ""
}
I0526 20:53:04.087513 2966 solver.cpp:102] Creating training net from net file: examples/mnist/lenet_train_test.prototxt
I0526 20:53:04.095394 2966 net.cpp:294] The NetState phase (0) differed from the phase (1) specified by a rule in layer mnist
I0526 20:53:04.095438 2966 net.cpp:294] The NetState phase (0) differed from the phase (1) specified by a rule in layer accuracy
I0526 20:53:04.095557 2966 net.cpp:51] Initializing net from parameters:
name: "LeNet"
state {
phase: TRAIN
level: 0
stage: ""
}
layer {
name: "mnist"
type: "Data"
top: "data"
top: "label"
include {
phase: TRAIN
}
transform_param {
scale: 0.00390625
}
data_param {
source: "examples/mnist/mnist_train_lmdb"
batch_size: 64
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: "loss"
type: "SoftmaxWithLoss"
bottom: "ip2"
bottom: "label"
top: "loss"
}
I0526 20:53:04.095737 2966 layer_factory.hpp:77] Creating layer mnist
I0526 20:53:04.095877 2966 db_lmdb.cpp:35] Opened lmdb examples/mnist/mnist_train_lmdb
I0526 20:53:04.095918 2966 net.cpp:84] Creating Layer mnist
I0526 20:53:04.095927 2966 net.cpp:380] mnist -> data
I0526 20:53:04.095945 2966 net.cpp:380] mnist -> label
I0526 20:53:04.102393 2966 data_layer.cpp:45] output data size: 64,1,28,28
I0526 20:53:04.104282 2966 net.cpp:122] Setting up mnist
I0526 20:53:04.104310 2966 net.cpp:129] Top shape: 64 1 28 28 (50176)
I0526 20:53:04.104315 2966 net.cpp:129] Top shape: 64 (64)
I0526 20:53:04.104317 2966 net.cpp:137] Memory required for data: 200960
I0526 20:53:04.104326 2966 layer_factory.hpp:77] Creating layer conv1
I0526 20:53:04.104351 2966 net.cpp:84] Creating Layer conv1
I0526 20:53:04.104357 2966 net.cpp:406] conv1 <- data
I0526 20:53:04.104368 2966 net.cpp:380] conv1 -> conv1
I0526 20:53:05.566699 2966 net.cpp:122] Setting up conv1
I0526 20:53:05.566740 2966 net.cpp:129] Top shape: 64 20 24 24 (737280)
I0526 20:53:05.566743 2966 net.cpp:137] Memory required for data: 3150080
I0526 20:53:05.566781 2966 layer_factory.hpp:77] Creating layer pool1
I0526 20:53:05.566792 2966 net.cpp:84] Creating Layer pool1
I0526 20:53:05.566882 2966 net.cpp:406] pool1 <- conv1
I0526 20:53:05.566889 2966 net.cpp:380] pool1 -> pool1
I0526 20:53:05.566968 2966 net.cpp:122] Setting up pool1
I0526 20:53:05.566973 2966 net.cpp:129] Top shape: 64 20 12 12 (184320)
I0526 20:53:05.566975 2966 net.cpp:137] Memory required for data: 3887360
I0526 20:53:05.566978 2966 layer_factory.hpp:77] Creating layer conv2
I0526 20:53:05.566998 2966 net.cpp:84] Creating Layer conv2
I0526 20:53:05.567000 2966 net.cpp:406] conv2 <- pool1
I0526 20:53:05.567021 2966 net.cpp:380] conv2 -> conv2
I0526 20:53:05.595108 2966 net.cpp:122] Setting up conv2
I0526 20:53:05.595130 2966 net.cpp:129] Top shape: 64 50 8 8 (204800)
I0526 20:53:05.595134 2966 net.cpp:137] Memory required for data: 4706560
I0526 20:53:05.595166 2966 layer_factory.hpp:77] Creating layer pool2
I0526 20:53:05.595178 2966 net.cpp:84] Creating Layer pool2
I0526 20:53:05.595183 2966 net.cpp:406] pool2 <- conv2
I0526 20:53:05.595190 2966 net.cpp:380] pool2 -> pool2
I0526 20:53:05.595263 2966 net.cpp:122] Setting up pool2
I0526 20:53:05.595269 2966 net.cpp:129] Top shape: 64 50 4 4 (51200)
I0526 20:53:05.595273 2966 net.cpp:137] Memory required for data: 4911360
I0526 20:53:05.595293 2966 layer_factory.hpp:77] Creating layer ip1
I0526 20:53:05.595299 2966 net.cpp:84] Creating Layer ip1
I0526 20:53:05.595304 2966 net.cpp:406] ip1 <- pool2
I0526 20:53:05.595309 2966 net.cpp:380] ip1 -> ip1
I0526 20:53:05.597863 2966 net.cpp:122] Setting up ip1
I0526 20:53:05.597903 2966 net.cpp:129] Top shape: 64 500 (32000)
I0526 20:53:05.597905 2966 net.cpp:137] Memory required for data: 5039360
I0526 20:53:05.597939 2966 layer_factory.hpp:77] Creating layer relu1
I0526 20:53:05.597962 2966 net.cpp:84] Creating Layer relu1
I0526 20:53:05.597966 2966 net.cpp:406] relu1 <- ip1
I0526 20:53:05.597971 2966 net.cpp:367] relu1 -> ip1 (in-place)
I0526 20:53:05.598219 2966 net.cpp:122] Setting up relu1
I0526 20:53:05.598227 2966 net.cpp:129] Top shape: 64 500 (32000)
I0526 20:53:05.598246 2966 net.cpp:137] Memory required for data: 5167360
I0526 20:53:05.598249 2966 layer_factory.hpp:77] Creating layer ip2
I0526 20:53:05.598275 2966 net.cpp:84] Creating Layer ip2
I0526 20:53:05.598279 2966 net.cpp:406] ip2 <- ip1
I0526 20:53:05.598285 2966 net.cpp:380] ip2 -> ip2
I0526 20:53:05.599239 2966 net.cpp:122] Setting up ip2
I0526 20:53:05.599251 2966 net.cpp:129] Top shape: 64 10 (640)
I0526 20:53:05.599256 2966 net.cpp:137] Memory required for data: 5169920
I0526 20:53:05.599261 2966 layer_factory.hpp:77] Creating layer loss
I0526 20:53:05.599269 2966 net.cpp:84] Creating Layer loss
I0526 20:53:05.599272 2966 net.cpp:406] loss <- ip2
I0526 20:53:05.599275 2966 net.cpp:406] loss <- label
I0526 20:53:05.599282 2966 net.cpp:380] loss -> loss
I0526 20:53:05.599297 2966 layer_factory.hpp:77] Creating layer loss
I0526 20:53:05.599552 2966 net.cpp:122] Setting up loss
I0526 20:53:05.599560 2966 net.cpp:129] Top shape: (1)
I0526 20:53:05.599565 2966 net.cpp:132] with loss weight 1
I0526 20:53:05.599584 2966 net.cpp:137] Memory required for data: 5169924
I0526 20:53:05.599588 2966 net.cpp:198] loss needs backward computation.
I0526 20:53:05.599594 2966 net.cpp:198] ip2 needs backward computation.
I0526 20:53:05.599597 2966 net.cpp:198] relu1 needs backward computation.
I0526 20:53:05.599601 2966 net.cpp:198] ip1 needs backward computation.
I0526 20:53:05.599603 2966 net.cpp:198] pool2 needs backward computation.
I0526 20:53:05.599606 2966 net.cpp:198] conv2 needs backward computation.
I0526 20:53:05.599609 2966 net.cpp:198] pool1 needs backward computation.
I0526 20:53:05.599612 2966 net.cpp:198] conv1 needs backward computation.
I0526 20:53:05.599615 2966 net.cpp:200] mnist does not need backward computation.
I0526 20:53:05.599618 2966 net.cpp:242] This network produces output loss
I0526 20:53:05.599627 2966 net.cpp:255] Network initialization done.
I0526 20:53:05.599763 2966 solver.cpp:190] Creating test net (#0) specified by net file: examples/mnist/lenet_train_test.prototxt
I0526 20:53:05.599830 2966 net.cpp:294] The NetState phase (1) differed from the phase (0) specified by a rule in layer mnist
I0526 20:53:05.599902 2966 net.cpp:51] Initializing net from parameters:
name: "LeNet"
state {
phase: TEST
}
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"
}
I0526 20:53:05.600005 2966 layer_factory.hpp:77] Creating layer mnist
I0526 20:53:05.600055 2966 db_lmdb.cpp:35] Opened lmdb examples/mnist/mnist_test_lmdb
I0526 20:53:05.600070 2966 net.cpp:84] Creating Layer mnist
I0526 20:53:05.600075 2966 net.cpp:380] mnist -> data
I0526 20:53:05.600100 2966 net.cpp:380] mnist -> label
I0526 20:53:05.600188 2966 data_layer.cpp:45] output data size: 100,1,28,28
I0526 20:53:05.601981 2966 net.cpp:122] Setting up mnist
I0526 20:53:05.602020 2966 net.cpp:129] Top shape: 100 1 28 28 (78400)
I0526 20:53:05.602025 2966 net.cpp:129] Top shape: 100 (100)
I0526 20:53:05.602027 2966 net.cpp:137] Memory required for data: 314000
I0526 20:53:05.602035 2966 layer_factory.hpp:77] Creating layer label_mnist_1_split
I0526 20:53:05.602051 2966 net.cpp:84] Creating Layer label_mnist_1_split
I0526 20:53:05.602054 2966 net.cpp:406] label_mnist_1_split <- label
I0526 20:53:05.602061 2966 net.cpp:380] label_mnist_1_split -> label_mnist_1_split_0
I0526 20:53:05.602071 2966 net.cpp:380] label_mnist_1_split -> label_mnist_1_split_1
I0526 20:53:05.602120 2966 net.cpp:122] Setting up label_mnist_1_split
I0526 20:53:05.602126 2966 net.cpp:129] Top shape: 100 (100)
I0526 20:53:05.602130 2966 net.cpp:129] Top shape: 100 (100)
I0526 20:53:05.602133 2966 net.cpp:137] Memory required for data: 314800
I0526 20:53:05.602135 2966 layer_factory.hpp:77] Creating layer conv1
I0526 20:53:05.602149 2966 net.cpp:84] Creating Layer conv1
I0526 20:53:05.602152 2966 net.cpp:406] conv1 <- data
I0526 20:53:05.602159 2966 net.cpp:380] conv1 -> conv1
I0526 20:53:05.605353 2966 net.cpp:122] Setting up conv1
I0526 20:53:05.605406 2966 net.cpp:129] Top shape: 100 20 24 24 (1152000)
I0526 20:53:05.605412 2966 net.cpp:137] Memory required for data: 4922800
I0526 20:53:05.605425 2966 layer_factory.hpp:77] Creating layer pool1
I0526 20:53:05.605515 2966 net.cpp:84] Creating Layer pool1
I0526 20:53:05.605525 2966 net.cpp:406] pool1 <- conv1
I0526 20:53:05.605532 2966 net.cpp:380] pool1 -> pool1
I0526 20:53:05.605584 2966 net.cpp:122] Setting up pool1
I0526 20:53:05.605592 2966 net.cpp:129] Top shape: 100 20 12 12 (288000)
I0526 20:53:05.605594 2966 net.cpp:137] Memory required for data: 6074800
I0526 20:53:05.605597 2966 layer_factory.hpp:77] Creating layer conv2
I0526 20:53:05.605614 2966 net.cpp:84] Creating Layer conv2
I0526 20:53:05.605618 2966 net.cpp:406] conv2 <- pool1
I0526 20:53:05.605625 2966 net.cpp:380] conv2 -> conv2
I0526 20:53:05.606948 2966 net.cpp:122] Setting up conv2
I0526 20:53:05.606976 2966 net.cpp:129] Top shape: 100 50 8 8 (320000)
I0526 20:53:05.606981 2966 net.cpp:137] Memory required for data: 7354800
I0526 20:53:05.606994 2966 layer_factory.hpp:77] Creating layer pool2
I0526 20:53:05.607010 2966 net.cpp:84] Creating Layer pool2
I0526 20:53:05.607015 2966 net.cpp:406] pool2 <- conv2
I0526 20:53:05.607024 2966 net.cpp:380] pool2 -> pool2
I0526 20:53:05.607066 2966 net.cpp:122] Setting up pool2
I0526 20:53:05.607072 2966 net.cpp:129] Top shape: 100 50 4 4 (80000)
I0526 20:53:05.607076 2966 net.cpp:137] Memory required for data: 7674800
I0526 20:53:05.607079 2966 layer_factory.hpp:77] Creating layer ip1
I0526 20:53:05.607089 2966 net.cpp:84] Creating Layer ip1
I0526 20:53:05.607092 2966 net.cpp:406] ip1 <- pool2
I0526 20:53:05.607100 2966 net.cpp:380] ip1 -> ip1
I0526 20:53:05.610023 2966 net.cpp:122] Setting up ip1
I0526 20:53:05.610049 2966 net.cpp:129] Top shape: 100 500 (50000)
I0526 20:53:05.610051 2966 net.cpp:137] Memory required for data: 7874800
I0526 20:53:05.610064 2966 layer_factory.hpp:77] Creating layer relu1
I0526 20:53:05.610082 2966 net.cpp:84] Creating Layer relu1
I0526 20:53:05.610088 2966 net.cpp:406] relu1 <- ip1
I0526 20:53:05.610095 2966 net.cpp:367] relu1 -> ip1 (in-place)
I0526 20:53:05.610596 2966 net.cpp:122] Setting up relu1
I0526 20:53:05.610606 2966 net.cpp:129] Top shape: 100 500 (50000)
I0526 20:53:05.610610 2966 net.cpp:137] Memory required for data: 8074800
I0526 20:53:05.610615 2966 layer_factory.hpp:77] Creating layer ip2
I0526 20:53:05.610625 2966 net.cpp:84] Creating Layer ip2
I0526 20:53:05.610628 2966 net.cpp:406] ip2 <- ip1
I0526 20:53:05.610635 2966 net.cpp:380] ip2 -> ip2
I0526 20:53:05.610752 2966 net.cpp:122] Setting up ip2
I0526 20:53:05.610759 2966 net.cpp:129] Top shape: 100 10 (1000)
I0526 20:53:05.610762 2966 net.cpp:137] Memory required for data: 8078800
I0526 20:53:05.610767 2966 layer_factory.hpp:77] Creating layer ip2_ip2_0_split
I0526 20:53:05.610774 2966 net.cpp:84] Creating Layer ip2_ip2_0_split
I0526 20:53:05.610776 2966 net.cpp:406] ip2_ip2_0_split <- ip2
I0526 20:53:05.610781 2966 net.cpp:380] ip2_ip2_0_split -> ip2_ip2_0_split_0
I0526 20:53:05.610787 2966 net.cpp:380] ip2_ip2_0_split -> ip2_ip2_0_split_1
I0526 20:53:05.610816 2966 net.cpp:122] Setting up ip2_ip2_0_split
I0526 20:53:05.610821 2966 net.cpp:129] Top shape: 100 10 (1000)
I0526 20:53:05.610824 2966 net.cpp:129] Top shape: 100 10 (1000)
I0526 20:53:05.610827 2966 net.cpp:137] Memory required for data: 8086800
I0526 20:53:05.610831 2966 layer_factory.hpp:77] Creating layer accuracy
I0526 20:53:05.610837 2966 net.cpp:84] Creating Layer accuracy
I0526 20:53:05.610841 2966 net.cpp:406] accuracy <- ip2_ip2_0_split_0
I0526 20:53:05.610844 2966 net.cpp:406] accuracy <- label_mnist_1_split_0
I0526 20:53:05.610849 2966 net.cpp:380] accuracy -> accuracy
I0526 20:53:05.610857 2966 net.cpp:122] Setting up accuracy
I0526 20:53:05.610860 2966 net.cpp:129] Top shape: (1)
I0526 20:53:05.610863 2966 net.cpp:137] Memory required for data: 8086804
I0526 20:53:05.610867 2966 layer_factory.hpp:77] Creating layer loss
I0526 20:53:05.610872 2966 net.cpp:84] Creating Layer loss
I0526 20:53:05.610875 2966 net.cpp:406] loss <- ip2_ip2_0_split_1
I0526 20:53:05.610879 2966 net.cpp:406] loss <- label_mnist_1_split_1
I0526 20:53:05.610904 2966 net.cpp:380] loss -> loss
I0526 20:53:05.610911 2966 layer_factory.hpp:77] Creating layer loss
I0526 20:53:05.611115 2966 net.cpp:122] Setting up loss
I0526 20:53:05.611124 2966 net.cpp:129] Top shape: (1)
I0526 20:53:05.611126 2966 net.cpp:132] with loss weight 1
I0526 20:53:05.611135 2966 net.cpp:137] Memory required for data: 8086808
I0526 20:53:05.611140 2966 net.cpp:198] loss needs backward computation.
I0526 20:53:05.611143 2966 net.cpp:200] accuracy does not need backward computation.
I0526 20:53:05.611147 2966 net.cpp:198] ip2_ip2_0_split needs backward computation.
I0526 20:53:05.611151 2966 net.cpp:198] ip2 needs backward computation.
I0526 20:53:05.611155 2966 net.cpp:198] relu1 needs backward computation.
I0526 20:53:05.611158 2966 net.cpp:198] ip1 needs backward computation.
I0526 20:53:05.611161 2966 net.cpp:198] pool2 needs backward computation.
I0526 20:53:05.611166 2966 net.cpp:198] conv2 needs backward computation.
I0526 20:53:05.611168 2966 net.cpp:198] pool1 needs backward computation.
I0526 20:53:05.611172 2966 net.cpp:198] conv1 needs backward computation.
I0526 20:53:05.611176 2966 net.cpp:200] label_mnist_1_split does not need backward computation.
I0526 20:53:05.611181 2966 net.cpp:200] mnist does not need backward computation.
I0526 20:53:05.611183 2966 net.cpp:242] This network produces output accuracy
I0526 20:53:05.611187 2966 net.cpp:242] This network produces output loss
I0526 20:53:05.611197 2966 net.cpp:255] Network initialization done.
I0526 20:53:05.611235 2966 solver.cpp:57] Solver scaffolding done.
I0526 20:53:05.611456 2966 caffe.cpp:239] Starting Optimization
I0526 20:53:05.611460 2966 solver.cpp:293] Solving LeNet
I0526 20:53:05.611464 2966 solver.cpp:294] Learning Rate Policy: inv
I0526 20:53:05.612161 2966 solver.cpp:351] Iteration 0, Testing net (#0)
I0526 20:53:05.638193 2966 blocking_queue.cpp:49] Waiting for data
I0526 20:53:05.727545 2980 data_layer.cpp:73] Restarting data prefetching from start.
I0526 20:53:05.731555 2966 solver.cpp:418] Test net output #0: accuracy = 0.1216
I0526 20:53:05.731581 2966 solver.cpp:418] Test net output #1: loss = 2.36362 (* 1 = 2.36362 loss)
I0526 20:53:05.735931 2966 solver.cpp:239] Iteration 0 (-0.273247 iter/s, 0.124447s/100 iters), loss = 2.32661
I0526 20:53:05.735957 2966 solver.cpp:258] Train net output #0: loss = 2.32661 (* 1 = 2.32661 loss)
I0526 20:53:05.735975 2966 sgd_solver.cpp:112] Iteration 0, lr = 0.01
I0526 20:53:06.072301 2966 solver.cpp:239] Iteration 100 (297.317 iter/s, 0.336342s/100 iters), loss = 0.213958
I0526 20:53:06.072345 2966 solver.cpp:258] Train net output #0: loss = 0.213958 (* 1 = 0.213958 loss)
I0526 20:53:06.072353 2966 sgd_solver.cpp:112] Iteration 100, lr = 0.00992565
I0526 20:53:06.394886 2966 solver.cpp:239] Iteration 200 (310.036 iter/s, 0.322544s/100 iters), loss = 0.151827
I0526 20:53:06.394940 2966 solver.cpp:258] Train net output #0: loss = 0.151827 (* 1 = 0.151827 loss)
I0526 20:53:06.394951 2966 sgd_solver.cpp:112] Iteration 200, lr = 0.00985258
I0526 20:53:06.702833 2966 solver.cpp:239] Iteration 300 (324.786 iter/s, 0.307895s/100 iters), loss = 0.199652
I0526 20:53:06.702881 2966 solver.cpp:258] Train net output #0: loss = 0.199652 (* 1 = 0.199652 loss)
I0526 20:53:06.702906 2966 sgd_solver.cpp:112] Iteration 300, lr = 0.00978075
I0526 20:53:07.018010 2966 solver.cpp:239] Iteration 400 (317.336 iter/s, 0.315124s/100 iters), loss = 0.0894907
I0526 20:53:07.018054 2966 solver.cpp:258] Train net output #0: loss = 0.0894907 (* 1 = 0.0894907 loss)
I0526 20:53:07.018064 2966 sgd_solver.cpp:112] Iteration 400, lr = 0.00971013
I0526 20:53:07.324208 2966 solver.cpp:351] Iteration 500, Testing net (#0)
I0526 20:53:07.428937 2980 data_layer.cpp:73] Restarting data prefetching from start.
I0526 20:53:07.430645 2966 solver.cpp:418] Test net output #0: accuracy = 0.9724
I0526 20:53:07.430685 2966 solver.cpp:418] Test net output #1: loss = 0.0885262 (* 1 = 0.0885262 loss)
I0526 20:53:07.433795 2966 solver.cpp:239] Iteration 500 (240.528 iter/s, 0.415753s/100 iters), loss = 0.186715
I0526 20:53:07.433852 2966 solver.cpp:258] Train net output #0: loss = 0.186715 (* 1 = 0.186715 loss)
I0526 20:53:07.433861 2966 sgd_solver.cpp:112] Iteration 500, lr = 0.00964069
I0526 20:53:07.748772 2966 solver.cpp:239] Iteration 600 (317.54 iter/s, 0.314921s/100 iters), loss = 0.100644
I0526 20:53:07.748816 2966 solver.cpp:258] Train net output #0: loss = 0.100644 (* 1 = 0.100644 loss)
I0526 20:53:07.748826 2966 sgd_solver.cpp:112] Iteration 600, lr = 0.0095724
I0526 20:53:08.069672 2966 solver.cpp:239] Iteration 700 (311.664 iter/s, 0.320859s/100 iters), loss = 0.161007
I0526 20:53:08.069715 2966 solver.cpp:258] Train net output #0: loss = 0.161007 (* 1 = 0.161007 loss)
I0526 20:53:08.069725 2966 sgd_solver.cpp:112] Iteration 700, lr = 0.00950522
I0526 20:53:08.391367 2966 solver.cpp:239] Iteration 800 (310.892 iter/s, 0.321655s/100 iters), loss = 0.236455
I0526 20:53:08.391410 2966 solver.cpp:258] Train net output #0: loss = 0.236456 (* 1 = 0.236456 loss)
I0526 20:53:08.391420 2966 sgd_solver.cpp:112] Iteration 800, lr = 0.00943913
I0526 20:53:08.708053 2966 solver.cpp:239] Iteration 900 (315.809 iter/s, 0.316647s/100 iters), loss = 0.177113
I0526 20:53:08.708115 2966 solver.cpp:258] Train net output #0: loss = 0.177113 (* 1 = 0.177113 loss)
I0526 20:53:08.708128 2966 sgd_solver.cpp:112] Iteration 900, lr = 0.00937411
I0526 20:53:08.817097 2979 data_layer.cpp:73] Restarting data prefetching from start.
I0526 20:53:09.029856 2966 solver.cpp:351] Iteration 1000, Testing net (#0)
I0526 20:53:09.156496 2980 data_layer.cpp:73] Restarting data prefetching from start.
I0526 20:53:09.158327 2966 solver.cpp:418] Test net output #0: accuracy = 0.9804
I0526 20:53:09.158361 2966 solver.cpp:418] Test net output #1: loss = 0.0615653 (* 1 = 0.0615653 loss)
I0526 20:53:09.161300 2966 solver.cpp:239] Iteration 1000 (220.655 iter/s, 0.453195s/100 iters), loss = 0.123391
I0526 20:53:09.161393 2966 solver.cpp:258] Train net output #0: loss = 0.123391 (* 1 = 0.123391 loss)
I0526 20:53:09.161434 2966 sgd_solver.cpp:112] Iteration 1000, lr = 0.00931012
I0526 20:53:09.510118 2966 solver.cpp:239] Iteration 1100 (286.763 iter/s, 0.34872s/100 iters), loss = 0.00871656
I0526 20:53:09.510197 2966 solver.cpp:258] Train net output #0: loss = 0.00871659 (* 1 = 0.00871659 loss)
I0526 20:53:09.510212 2966 sgd_solver.cpp:112] Iteration 1100, lr = 0.00924715
I0526 20:53:09.855741 2966 solver.cpp:239] Iteration 1200 (289.396 iter/s, 0.345548s/100 iters), loss = 0.0137676
I0526 20:53:09.855787 2966 solver.cpp:258] Train net output #0: loss = 0.0137676 (* 1 = 0.0137676 loss)
I0526 20:53:09.855795 2966 sgd_solver.cpp:112] Iteration 1200, lr = 0.00918515
I0526 20:53:10.174860 2966 solver.cpp:239] Iteration 1300 (313.404 iter/s, 0.319077s/100 iters), loss = 0.0125752
I0526 20:53:10.174901 2966 solver.cpp:258] Train net output #0: loss = 0.0125753 (* 1 = 0.0125753 loss)
I0526 20:53:10.174909 2966 sgd_solver.cpp:112] Iteration 1300, lr = 0.00912412
I0526 20:53:10.491436 2966 solver.cpp:239] Iteration 1400 (315.917 iter/s, 0.316539s/100 iters), loss = 0.0048969
I0526 20:53:10.491477 2966 solver.cpp:258] Train net output #0: loss = 0.00489691 (* 1 = 0.00489691 loss)
I0526 20:53:10.491487 2966 sgd_solver.cpp:112] Iteration 1400, lr = 0.00906403
I0526 20:53:10.801805 2966 solver.cpp:351] Iteration 1500, Testing net (#0)
I0526 20:53:10.910562 2980 data_layer.cpp:73] Restarting data prefetching from start.
I0526 20:53:10.913337 2966 solver.cpp:418] Test net output #0: accuracy = 0.9831
I0526 20:53:10.913365 2966 solver.cpp:418] Test net output #1: loss = 0.0542353 (* 1 = 0.0542353 loss)
I0526 20:53:10.916178 2966 solver.cpp:239] Iteration 1500 (235.455 iter/s, 0.42471s/100 iters), loss = 0.0902566
I0526 20:53:10.916203 2966 solver.cpp:258] Train net output #0: loss = 0.0902567 (* 1 = 0.0902567 loss)
I0526 20:53:10.916213 2966 sgd_solver.cpp:112] Iteration 1500, lr = 0.00900485
I0526 20:53:11.231348 2966 solver.cpp:239] Iteration 1600 (317.312 iter/s, 0.315147s/100 iters), loss = 0.123736
I0526 20:53:11.231379 2966 solver.cpp:258] Train net output #0: loss = 0.123736 (* 1 = 0.123736 loss)
I0526 20:53:11.231386 2966 sgd_solver.cpp:112] Iteration 1600, lr = 0.00894657
I0526 20:53:11.538249 2966 solver.cpp:239] Iteration 1700 (325.867 iter/s, 0.306873s/100 iters), loss = 0.0267345
I0526 20:53:11.538297 2966 solver.cpp:258] Train net output #0: loss = 0.0267346 (* 1 = 0.0267346 loss)
I0526 20:53:11.538327 2966 sgd_solver.cpp:112] Iteration 1700, lr = 0.00888916
I0526 20:53:11.854401 2966 solver.cpp:239] Iteration 1800 (316.345 iter/s, 0.31611s/100 iters), loss = 0.0102682
I0526 20:53:11.854444 2966 solver.cpp:258] Train net output #0: loss = 0.0102682 (* 1 = 0.0102682 loss)
I0526 20:53:11.854454 2966 sgd_solver.cpp:112] Iteration 1800, lr = 0.0088326
I0526 20:53:12.081883 2979 data_layer.cpp:73] Restarting data prefetching from start.
I0526 20:53:12.174082 2966 solver.cpp:239] Iteration 1900 (312.847 iter/s, 0.319645s/100 iters), loss = 0.131612
I0526 20:53:12.174130 2966 solver.cpp:258] Train net output #0: loss = 0.131612 (* 1 = 0.131612 loss)
I0526 20:53:12.174139 2966 sgd_solver.cpp:112] Iteration 1900, lr = 0.00877687
I0526 20:53:12.488847 2966 solver.cpp:351] Iteration 2000, Testing net (#0)
I0526 20:53:12.595489 2980 data_layer.cpp:73] Restarting data prefetching from start.
I0526 20:53:12.597185 2966 solver.cpp:418] Test net output #0: accuracy = 0.9852
I0526 20:53:12.597208 2966 solver.cpp:418] Test net output #1: loss = 0.0458715 (* 1 = 0.0458715 loss)
I0526 20:53:12.600069 2966 solver.cpp:239] Iteration 2000 (234.768 iter/s, 0.425952s/100 iters), loss = 0.0130547
I0526 20:53:12.600131 2966 solver.cpp:258] Train net output #0: loss = 0.0130547 (* 1 = 0.0130547 loss)
I0526 20:53:12.600138 2966 sgd_solver.cpp:112] Iteration 2000, lr = 0.00872196
I0526 20:53:12.911358 2966 solver.cpp:239] Iteration 2100 (321.262 iter/s, 0.311272s/100 iters), loss = 0.0510326
I0526 20:53:12.911401 2966 solver.cpp:258] Train net output #0: loss = 0.0510326 (* 1 = 0.0510326 loss)
I0526 20:53:12.911411 2966 sgd_solver.cpp:112] Iteration 2100, lr = 0.00866784
I0526 20:53:13.226441 2966 solver.cpp:239] Iteration 2200 (317.416 iter/s, 0.315044s/100 iters), loss = 0.0169382
I0526 20:53:13.226485 2966 solver.cpp:258] Train net output #0: loss = 0.0169383 (* 1 = 0.0169383 loss)
I0526 20:53:13.226495 2966 sgd_solver.cpp:112] Iteration 2200, lr = 0.0086145
I0526 20:53:13.541839 2966 solver.cpp:239] Iteration 2300 (317.101 iter/s, 0.315357s/100 iters), loss = 0.116517
I0526 20:53:13.541883 2966 solver.cpp:258] Train net output #0: loss = 0.116517 (* 1 = 0.116517 loss)
I0526 20:53:13.541893 2966 sgd_solver.cpp:112] Iteration 2300, lr = 0.00856192
I0526 20:53:13.856551 2966 solver.cpp:239] Iteration 2400 (317.793 iter/s, 0.31467s/100 iters), loss = 0.0162969
I0526 20:53:13.856595 2966 solver.cpp:258] Train net output #0: loss = 0.0162969 (* 1 = 0.0162969 loss)
I0526 20:53:13.856604 2966 sgd_solver.cpp:112] Iteration 2400, lr = 0.00851008
I0526 20:53:14.175319 2966 solver.cpp:351] Iteration 2500, Testing net (#0)
I0526 20:53:14.326673 2980 data_layer.cpp:73] Restarting data prefetching from start.
I0526 20:53:14.329370 2966 solver.cpp:418] Test net output #0: accuracy = 0.9847
I0526 20:53:14.329412 2966 solver.cpp:418] Test net output #1: loss = 0.0517607 (* 1 = 0.0517607 loss)
I0526 20:53:14.332283 2966 solver.cpp:239] Iteration 2500 (210.216 iter/s, 0.475701s/100 iters), loss = 0.0305351
I0526 20:53:14.332320 2966 solver.cpp:258] Train net output #0: loss = 0.0305351 (* 1 = 0.0305351 loss)
I0526 20:53:14.332332 2966 sgd_solver.cpp:112] Iteration 2500, lr = 0.00845897
I0526 20:53:14.678824 2966 solver.cpp:239] Iteration 2600 (288.597 iter/s, 0.346504s/100 iters), loss = 0.056123
I0526 20:53:14.678854 2966 solver.cpp:258] Train net output #0: loss = 0.056123 (* 1 = 0.056123 loss)
I0526 20:53:14.678879 2966 sgd_solver.cpp:112] Iteration 2600, lr = 0.00840857
I0526 20:53:14.998436 2966 solver.cpp:239] Iteration 2700 (312.908 iter/s, 0.319583s/100 iters), loss = 0.0520153
I0526 20:53:14.998476 2966 solver.cpp:258] Train net output #0: loss = 0.0520154 (* 1 = 0.0520154 loss)
I0526 20:53:14.998486 2966 sgd_solver.cpp:112] Iteration 2700, lr = 0.00835886
I0526 20:53:15.315748 2966 solver.cpp:239] Iteration 2800 (315.184 iter/s, 0.317275s/100 iters), loss = 0.00551482
I0526 20:53:15.315791 2966 solver.cpp:258] Train net output #0: loss = 0.00551484 (* 1 = 0.00551484 loss)
I0526 20:53:15.315800 2966 sgd_solver.cpp:112] Iteration 2800, lr = 0.00830984
I0526 20:53:15.345508 2979 data_layer.cpp:73] Restarting data prefetching from start.
I0526 20:53:15.648234 2966 solver.cpp:239] Iteration 2900 (300.799 iter/s, 0.332447s/100 iters), loss = 0.021568
I0526 20:53:15.648273 2966 solver.cpp:258] Train net output #0: loss = 0.0215681 (* 1 = 0.0215681 loss)
I0526 20:53:15.648283 2966 sgd_solver.cpp:112] Iteration 2900, lr = 0.00826148
I0526 20:53:15.964462 2966 solver.cpp:351] Iteration 3000, Testing net (#0)
I0526 20:53:16.078405 2980 data_layer.cpp:73] Restarting data prefetching from start.
I0526 20:53:16.085981 2966 solver.cpp:418] Test net output #0: accuracy = 0.9861
I0526 20:53:16.086019 2966 solver.cpp:418] Test net output #1: loss = 0.0425739 (* 1 = 0.0425739 loss)
I0526 20:53:16.089234 2966 solver.cpp:239] Iteration 3000 (226.777 iter/s, 0.440963s/100 iters), loss = 0.00921144
I0526 20:53:16.090874 2966 solver.cpp:258] Train net output #0: loss = 0.00921145 (* 1 = 0.00921145 loss)
I0526 20:53:16.090970 2966 sgd_solver.cpp:112] Iteration 3000, lr = 0.00821377
I0526 20:53:16.443243 2966 solver.cpp:239] Iteration 3100 (283.748 iter/s, 0.352426s/100 iters), loss = 0.00934141
I0526 20:53:16.443294 2966 solver.cpp:258] Train net output #0: loss = 0.0093414 (* 1 = 0.0093414 loss)
I0526 20:53:16.443300 2966 sgd_solver.cpp:112] Iteration 3100, lr = 0.0081667
I0526 20:53:16.759610 2966 solver.cpp:239] Iteration 3200 (316.119 iter/s, 0.316336s/100 iters), loss = 0.0120144
I0526 20:53:16.759650 2966 solver.cpp:258] Train net output #0: loss = 0.0120144 (* 1 = 0.0120144 loss)
I0526 20:53:16.759660 2966 sgd_solver.cpp:112] Iteration 3200, lr = 0.00812025
I0526 20:53:17.101521 2966 solver.cpp:239] Iteration 3300 (292.52 iter/s, 0.341857s/100 iters), loss = 0.0458635
I0526 20:53:17.101608 2966 solver.cpp:258] Train net output #0: loss = 0.0458635 (* 1 = 0.0458635 loss)
I0526 20:53:17.101622 2966 sgd_solver.cpp:112] Iteration 3300, lr = 0.00807442
I0526 20:53:17.443616 2966 solver.cpp:239] Iteration 3400 (292.383 iter/s, 0.342017s/100 iters), loss = 0.0129833
I0526 20:53:17.443660 2966 solver.cpp:258] Train net output #0: loss = 0.0129833 (* 1 = 0.0129833 loss)
I0526 20:53:17.443670 2966 sgd_solver.cpp:112] Iteration 3400, lr = 0.00802918
I0526 20:53:17.761767 2966 solver.cpp:351] Iteration 3500, Testing net (#0)
I0526 20:53:17.897162 2980 data_layer.cpp:73] Restarting data prefetching from start.
I0526 20:53:17.898878 2966 solver.cpp:418] Test net output #0: accuracy = 0.9857
I0526 20:53:17.899029 2966 solver.cpp:418] Test net output #1: loss = 0.0437298 (* 1 = 0.0437298 loss)
I0526 20:53:17.902154 2966 solver.cpp:239] Iteration 3500 (218.099 iter/s, 0.458507s/100 iters), loss = 0.00534368
I0526 20:53:17.902294 2966 solver.cpp:258] Train net output #0: loss = 0.00534369 (* 1 = 0.00534369 loss)
I0526 20:53:17.902312 2966 sgd_solver.cpp:112] Iteration 3500, lr = 0.00798454
I0526 20:53:18.232357 2966 solver.cpp:239] Iteration 3600 (302.994 iter/s, 0.330039s/100 iters), loss = 0.0281394
I0526 20:53:18.232447 2966 solver.cpp:258] Train net output #0: loss = 0.0281394 (* 1 = 0.0281394 loss)
I0526 20:53:18.232465 2966 sgd_solver.cpp:112] Iteration 3600, lr = 0.00794046
I0526 20:53:18.607868 2966 solver.cpp:239] Iteration 3700 (266.36 iter/s, 0.375431s/100 iters), loss = 0.0125563
I0526 20:53:18.607975 2966 solver.cpp:258] Train net output #0: loss = 0.0125563 (* 1 = 0.0125563 loss)
I0526 20:53:18.608063 2966 sgd_solver.cpp:112] Iteration 3700, lr = 0.00789695
I0526 20:53:18.771580 2979 data_layer.cpp:73] Restarting data prefetching from start.
I0526 20:53:18.973671 2966 solver.cpp:239] Iteration 3800 (273.446 iter/s, 0.365703s/100 iters), loss = 0.00594559
I0526 20:53:18.973739 2966 solver.cpp:258] Train net output #0: loss = 0.00594559 (* 1 = 0.00594559 loss)
I0526 20:53:18.973745 2966 sgd_solver.cpp:112] Iteration 3800, lr = 0.007854
I0526 20:53:19.298178 2966 solver.cpp:239] Iteration 3900 (308.201 iter/s, 0.324463s/100 iters), loss = 0.0212954
I0526 20:53:19.298220 2966 solver.cpp:258] Train net output #0: loss = 0.0212954 (* 1 = 0.0212954 loss)
I0526 20:53:19.298229 2966 sgd_solver.cpp:112] Iteration 3900, lr = 0.00781158
I0526 20:53:19.611858 2966 solver.cpp:351] Iteration 4000, Testing net (#0)
I0526 20:53:19.726984 2980 data_layer.cpp:73] Restarting data prefetching from start.
I0526 20:53:19.728912 2966 solver.cpp:418] Test net output #0: accuracy = 0.9891
I0526 20:53:19.728941 2966 solver.cpp:418] Test net output #1: loss = 0.0322071 (* 1 = 0.0322071 loss)
I0526 20:53:19.731956 2966 solver.cpp:239] Iteration 4000 (230.55 iter/s, 0.433746s/100 iters), loss = 0.0188879
I0526 20:53:19.731986 2966 solver.cpp:258] Train net output #0: loss = 0.0188879 (* 1 = 0.0188879 loss)
I0526 20:53:19.731994 2966 sgd_solver.cpp:112] Iteration 4000, lr = 0.0077697
I0526 20:53:20.042589 2966 solver.cpp:239] Iteration 4100 (321.962 iter/s, 0.310596s/100 iters), loss = 0.0131814
I0526 20:53:20.042637 2966 solver.cpp:258] Train net output #0: loss = 0.0131814 (* 1 = 0.0131814 loss)
I0526 20:53:20.042644 2966 sgd_solver.cpp:112] Iteration 4100, lr = 0.00772833
I0526 20:53:20.361515 2966 solver.cpp:239] Iteration 4200 (313.599 iter/s, 0.318879s/100 iters), loss = 0.0151852
I0526 20:53:20.361557 2966 solver.cpp:258] Train net output #0: loss = 0.0151852 (* 1 = 0.0151852 loss)
I0526 20:53:20.361567 2966 sgd_solver.cpp:112] Iteration 4200, lr = 0.00768748
I0526 20:53:20.674365 2966 solver.cpp:239] Iteration 4300 (319.681 iter/s, 0.312812s/100 iters), loss = 0.0477351
I0526 20:53:20.674412 2966 solver.cpp:258] Train net output #0: loss = 0.0477351 (* 1 = 0.0477351 loss)
I0526 20:53:20.674419 2966 sgd_solver.cpp:112] Iteration 4300, lr = 0.00764712
I0526 20:53:20.994652 2966 solver.cpp:239] Iteration 4400 (312.263 iter/s, 0.320242s/100 iters), loss = 0.00818447
I0526 20:53:20.994691 2966 solver.cpp:258] Train net output #0: loss = 0.00818447 (* 1 = 0.00818447 loss)
I0526 20:53:20.994699 2966 sgd_solver.cpp:112] Iteration 4400, lr = 0.00760726
I0526 20:53:21.305419 2966 solver.cpp:351] Iteration 4500, Testing net (#0)
I0526 20:53:21.413980 2980 data_layer.cpp:73] Restarting data prefetching from start.
I0526 20:53:21.415732 2966 solver.cpp:418] Test net output #0: accuracy = 0.9865
I0526 20:53:21.415756 2966 solver.cpp:418] Test net output #1: loss = 0.0391329 (* 1 = 0.0391329 loss)
I0526 20:53:21.418720 2966 solver.cpp:239] Iteration 4500 (235.827 iter/s, 0.424039s/100 iters), loss = 0.00663482
I0526 20:53:21.418745 2966 solver.cpp:258] Train net output #0: loss = 0.00663481 (* 1 = 0.00663481 loss)
I0526 20:53:21.418756 2966 sgd_solver.cpp:112] Iteration 4500, lr = 0.00756788
I0526 20:53:21.738775 2966 solver.cpp:239] Iteration 4600 (312.481 iter/s, 0.320019s/100 iters), loss = 0.0199677
I0526 20:53:21.738818 2966 solver.cpp:258] Train net output #0: loss = 0.0199676 (* 1 = 0.0199676 loss)
I0526 20:53:21.738826 2966 sgd_solver.cpp:112] Iteration 4600, lr = 0.00752897
I0526 20:53:22.012771 2979 data_layer.cpp:73] Restarting data prefetching from start.
I0526 20:53:22.069990 2966 solver.cpp:239] Iteration 4700 (301.953 iter/s, 0.331177s/100 iters), loss = 0.00672691
I0526 20:53:22.070024 2966 solver.cpp:258] Train net output #0: loss = 0.00672689 (* 1 = 0.00672689 loss)
I0526 20:53:22.070032 2966 sgd_solver.cpp:112] Iteration 4700, lr = 0.00749052
I0526 20:53:22.381690 2966 solver.cpp:239] Iteration 4800 (320.855 iter/s, 0.311668s/100 iters), loss = 0.0105764
I0526 20:53:22.381767 2966 solver.cpp:258] Train net output #0: loss = 0.0105764 (* 1 = 0.0105764 loss)
I0526 20:53:22.381775 2966 sgd_solver.cpp:112] Iteration 4800, lr = 0.00745253
I0526 20:53:22.707139 2966 solver.cpp:239] Iteration 4900 (307.336 iter/s, 0.325377s/100 iters), loss = 0.0102207
I0526 20:53:22.707185 2966 solver.cpp:258] Train net output #0: loss = 0.0102206 (* 1 = 0.0102206 loss)
I0526 20:53:22.707195 2966 sgd_solver.cpp:112] Iteration 4900, lr = 0.00741498
I0526 20:53:23.023452 2966 solver.cpp:468] Snapshotting to binary proto file examples/mnist/lenet_iter_5000.caffemodel
I0526 20:53:23.032524 2966 sgd_solver.cpp:280] Snapshotting solver state to binary proto file examples/mnist/lenet_iter_5000.solverstate
I0526 20:53:23.035620 2966 solver.cpp:351] Iteration 5000, Testing net (#0)
I0526 20:53:23.145226 2980 data_layer.cpp:73] Restarting data prefetching from start.
I0526 20:53:23.146982 2966 solver.cpp:418] Test net output #0: accuracy = 0.9886
I0526 20:53:23.147011 2966 solver.cpp:418] Test net output #1: loss = 0.0334444 (* 1 = 0.0334444 loss)
I0526 20:53:23.150020 2966 solver.cpp:239] Iteration 5000 (225.812 iter/s, 0.442847s/100 iters), loss = 0.0506286
I0526 20:53:23.150048 2966 solver.cpp:258] Train net output #0: loss = 0.0506285 (* 1 = 0.0506285 loss)
I0526 20:53:23.150058 2966 sgd_solver.cpp:112] Iteration 5000, lr = 0.00737788
I0526 20:53:23.465286 2966 solver.cpp:239] Iteration 5100 (317.216 iter/s, 0.315242s/100 iters), loss = 0.0253424
I0526 20:53:23.465337 2966 solver.cpp:258] Train net output #0: loss = 0.0253424 (* 1 = 0.0253424 loss)
I0526 20:53:23.465343 2966 sgd_solver.cpp:112] Iteration 5100, lr = 0.0073412
I0526 20:53:23.784103 2966 solver.cpp:239] Iteration 5200 (313.721 iter/s, 0.318755s/100 iters), loss = 0.00687777
I0526 20:53:23.784148 2966 solver.cpp:258] Train net output #0: loss = 0.00687772 (* 1 = 0.00687772 loss)
I0526 20:53:23.784157 2966 sgd_solver.cpp:112] Iteration 5200, lr = 0.00730495
I0526 20:53:24.097098 2966 solver.cpp:239] Iteration 5300 (319.533 iter/s, 0.312957s/100 iters), loss = 0.00152222
I0526 20:53:24.097147 2966 solver.cpp:258] Train net output #0: loss = 0.00152217 (* 1 = 0.00152217 loss)
I0526 20:53:24.097156 2966 sgd_solver.cpp:112] Iteration 5300, lr = 0.00726911
I0526 20:53:24.409426 2966 solver.cpp:239] Iteration 5400 (320.225 iter/s, 0.31228s/100 iters), loss = 0.00699226
I0526 20:53:24.409498 2966 solver.cpp:258] Train net output #0: loss = 0.0069922 (* 1 = 0.0069922 loss)
I0526 20:53:24.409510 2966 sgd_solver.cpp:112] Iteration 5400, lr = 0.00723368
I0526 20:53:24.718976 2966 solver.cpp:351] Iteration 5500, Testing net (#0)
I0526 20:53:24.834223 2980 data_layer.cpp:73] Restarting data prefetching from start.
I0526 20:53:24.837232 2966 solver.cpp:418] Test net output #0: accuracy = 0.9883
I0526 20:53:24.837262 2966 solver.cpp:418] Test net output #1: loss = 0.0351423 (* 1 = 0.0351423 loss)
I0526 20:53:24.840135 2966 solver.cpp:239] Iteration 5500 (232.207 iter/s, 0.43065s/100 iters), loss = 0.00880486
I0526 20:53:24.840165 2966 solver.cpp:258] Train net output #0: loss = 0.00880479 (* 1 = 0.00880479 loss)
I0526 20:53:24.840175 2966 sgd_solver.cpp:112] Iteration 5500, lr = 0.00719865
I0526 20:53:25.150647 2966 solver.cpp:239] Iteration 5600 (322.075 iter/s, 0.310487s/100 iters), loss = 0.000624476
I0526 20:53:25.150697 2966 solver.cpp:258] Train net output #0: loss = 0.000624411 (* 1 = 0.000624411 loss)
I0526 20:53:25.150703 2966 sgd_solver.cpp:112] Iteration 5600, lr = 0.00716402
I0526 20:53:25.219350 2979 data_layer.cpp:73] Restarting data prefetching from start.
I0526 20:53:25.475374 2966 solver.cpp:239] Iteration 5700 (307.996 iter/s, 0.32468s/100 iters), loss = 0.00480867
I0526 20:53:25.475419 2966 solver.cpp:258] Train net output #0: loss = 0.00480861 (* 1 = 0.00480861 loss)
I0526 20:53:25.475426 2966 sgd_solver.cpp:112] Iteration 5700, lr = 0.00712977
I0526 20:53:25.797672 2966 solver.cpp:239] Iteration 5800 (310.308 iter/s, 0.322261s/100 iters), loss = 0.0414849
I0526 20:53:25.797731 2966 solver.cpp:258] Train net output #0: loss = 0.0414848 (* 1 = 0.0414848 loss)
I0526 20:53:25.797739 2966 sgd_solver.cpp:112] Iteration 5800, lr = 0.0070959
I0526 20:53:26.110270 2966 solver.cpp:239] Iteration 5900 (319.955 iter/s, 0.312544s/100 iters), loss = 0.0081392
I0526 20:53:26.110317 2966 solver.cpp:258] Train net output #0: loss = 0.00813913 (* 1 = 0.00813913 loss)
I0526 20:53:26.110325 2966 sgd_solver.cpp:112] Iteration 5900, lr = 0.0070624
I0526 20:53:26.423714 2966 solver.cpp:351] Iteration 6000, Testing net (#0)
I0526 20:53:26.531909 2980 data_layer.cpp:73] Restarting data prefetching from start.
I0526 20:53:26.534649 2966 solver.cpp:418] Test net output #0: accuracy = 0.9903
I0526 20:53:26.534677 2966 solver.cpp:418] Test net output #1: loss = 0.0299221 (* 1 = 0.0299221 loss)
I0526 20:53:26.537436 2966 solver.cpp:239] Iteration 6000 (234.12 iter/s, 0.427131s/100 iters), loss = 0.00366409
I0526 20:53:26.537466 2966 solver.cpp:258] Train net output #0: loss = 0.00366403 (* 1 = 0.00366403 loss)
I0526 20:53:26.537473 2966 sgd_solver.cpp:112] Iteration 6000, lr = 0.00702927
I0526 20:53:26.851928 2966 solver.cpp:239] Iteration 6100 (317.997 iter/s, 0.314468s/100 iters), loss = 0.00321985
I0526 20:53:26.851976 2966 solver.cpp:258] Train net output #0: loss = 0.00321979 (* 1 = 0.00321979 loss)
I0526 20:53:26.852001 2966 sgd_solver.cpp:112] Iteration 6100, lr = 0.0069965
I0526 20:53:27.172431 2966 solver.cpp:239] Iteration 6200 (312.051 iter/s, 0.32046s/100 iters), loss = 0.010021
I0526 20:53:27.172482 2966 solver.cpp:258] Train net output #0: loss = 0.0100209 (* 1 = 0.0100209 loss)
I0526 20:53:27.172489 2966 sgd_solver.cpp:112] Iteration 6200, lr = 0.00696408
I0526 20:53:27.490736 2966 solver.cpp:239] Iteration 6300 (314.195 iter/s, 0.318274s/100 iters), loss = 0.0118692
I0526 20:53:27.490779 2966 solver.cpp:258] Train net output #0: loss = 0.0118691 (* 1 = 0.0118691 loss)
I0526 20:53:27.490787 2966 sgd_solver.cpp:112] Iteration 6300, lr = 0.00693201
I0526 20:53:27.810791 2966 solver.cpp:239] Iteration 6400 (312.481 iter/s, 0.320019s/100 iters), loss = 0.00745714
I0526 20:53:27.810840 2966 solver.cpp:258] Train net output #0: loss = 0.00745708 (* 1 = 0.00745708 loss)
I0526 20:53:27.810847 2966 sgd_solver.cpp:112] Iteration 6400, lr = 0.00690029
I0526 20:53:28.127219 2966 solver.cpp:351] Iteration 6500, Testing net (#0)
I0526 20:53:28.235361 2980 data_layer.cpp:73] Restarting data prefetching from start.
I0526 20:53:28.238287 2966 solver.cpp:418] Test net output #0: accuracy = 0.9898
I0526 20:53:28.238328 2966 solver.cpp:418] Test net output #1: loss = 0.033051 (* 1 = 0.033051 loss)
I0526 20:53:28.241364 2966 solver.cpp:239] Iteration 6500 (232.262 iter/s, 0.430548s/100 iters), loss = 0.0104185
I0526 20:53:28.241413 2966 solver.cpp:258] Train net output #0: loss = 0.0104185 (* 1 = 0.0104185 loss)
I0526 20:53:28.241423 2966 sgd_solver.cpp:112] Iteration 6500, lr = 0.0068689
I0526 20:53:28.453863 2979 data_layer.cpp:73] Restarting data prefetching from start.
I0526 20:53:28.598322 2966 solver.cpp:239] Iteration 6600 (280.18 iter/s, 0.356913s/100 iters), loss = 0.0180983
I0526 20:53:28.598382 2966 solver.cpp:258] Train net output #0: loss = 0.0180983 (* 1 = 0.0180983 loss)
I0526 20:53:28.598393 2966 sgd_solver.cpp:112] Iteration 6600, lr = 0.00683784
I0526 20:53:28.924755 2966 solver.cpp:239] Iteration 6700 (306.396 iter/s, 0.326375s/100 iters), loss = 0.0056943
I0526 20:53:28.924806 2966 solver.cpp:258] Train net output #0: loss = 0.00569423 (* 1 = 0.00569423 loss)
I0526 20:53:28.924818 2966 sgd_solver.cpp:112] Iteration 6700, lr = 0.00680711
I0526 20:53:29.277735 2966 solver.cpp:239] Iteration 6800 (283.344 iter/s, 0.352928s/100 iters), loss = 0.00439319
I0526 20:53:29.277822 2966 solver.cpp:258] Train net output #0: loss = 0.00439311 (* 1 = 0.00439311 loss)
I0526 20:53:29.277834 2966 sgd_solver.cpp:112] Iteration 6800, lr = 0.0067767
I0526 20:53:29.596837 2966 solver.cpp:239] Iteration 6900 (313.454 iter/s, 0.319026s/100 iters), loss = 0.00596661
I0526 20:53:29.596886 2966 solver.cpp:258] Train net output #0: loss = 0.00596654 (* 1 = 0.00596654 loss)
I0526 20:53:29.596894 2966 sgd_solver.cpp:112] Iteration 6900, lr = 0.0067466
I0526 20:53:29.944209 2966 solver.cpp:351] Iteration 7000, Testing net (#0)
I0526 20:53:30.065789 2980 data_layer.cpp:73] Restarting data prefetching from start.
I0526 20:53:30.067525 2966 solver.cpp:418] Test net output #0: accuracy = 0.9895
I0526 20:53:30.067554 2966 solver.cpp:418] Test net output #1: loss = 0.0316801 (* 1 = 0.0316801 loss)
I0526 20:53:30.070441 2966 solver.cpp:239] Iteration 7000 (211.164 iter/s, 0.473566s/100 iters), loss = 0.00476039
I0526 20:53:30.070467 2966 solver.cpp:258] Train net output #0: loss = 0.00476033 (* 1 = 0.00476033 loss)
I0526 20:53:30.070477 2966 sgd_solver.cpp:112] Iteration 7000, lr = 0.00671681
I0526 20:53:30.427279 2966 solver.cpp:239] Iteration 7100 (280.256 iter/s, 0.356817s/100 iters), loss = 0.0145102
I0526 20:53:30.427325 2966 solver.cpp:258] Train net output #0: loss = 0.0145101 (* 1 = 0.0145101 loss)
I0526 20:53:30.427336 2966 sgd_solver.cpp:112] Iteration 7100, lr = 0.00668733
I0526 20:53:30.771559 2966 solver.cpp:239] Iteration 7200 (290.495 iter/s, 0.34424s/100 iters), loss = 0.0032593
I0526 20:53:30.771603 2966 solver.cpp:258] Train net output #0: loss = 0.00325925 (* 1 = 0.00325925 loss)
I0526 20:53:30.771613 2966 sgd_solver.cpp:112] Iteration 7200, lr = 0.00665815
I0526 20:53:31.105365 2966 solver.cpp:239] Iteration 7300 (299.609 iter/s, 0.333769s/100 iters), loss = 0.0249073
I0526 20:53:31.105412 2966 solver.cpp:258] Train net output #0: loss = 0.0249072 (* 1 = 0.0249072 loss)
I0526 20:53:31.105419 2966 sgd_solver.cpp:112] Iteration 7300, lr = 0.00662927
I0526 20:53:31.426230 2966 solver.cpp:239] Iteration 7400 (311.699 iter/s, 0.320822s/100 iters), loss = 0.00565549
I0526 20:53:31.426262 2966 solver.cpp:258] Train net output #0: loss = 0.00565543 (* 1 = 0.00565543 loss)
I0526 20:53:31.426270 2966 sgd_solver.cpp:112] Iteration 7400, lr = 0.00660067
I0526 20:53:31.739367 2979 data_layer.cpp:73] Restarting data prefetching from start.
I0526 20:53:31.751914 2966 solver.cpp:351] Iteration 7500, Testing net (#0)
I0526 20:53:31.870637 2980 data_layer.cpp:73] Restarting data prefetching from start.
I0526 20:53:31.872447 2966 solver.cpp:418] Test net output #0: accuracy = 0.9893
I0526 20:53:31.872550 2966 solver.cpp:418] Test net output #1: loss = 0.0326281 (* 1 = 0.0326281 loss)
I0526 20:53:31.875960 2966 solver.cpp:239] Iteration 7500 (222.371 iter/s, 0.449699s/100 iters), loss = 0.00179423
I0526 20:53:31.876109 2966 solver.cpp:258] Train net output #0: loss = 0.00179417 (* 1 = 0.00179417 loss)
I0526 20:53:31.876158 2966 sgd_solver.cpp:112] Iteration 7500, lr = 0.00657236
I0526 20:53:32.204651 2966 solver.cpp:239] Iteration 7600 (304.35 iter/s, 0.328569s/100 iters), loss = 0.00562589
I0526 20:53:32.204694 2966 solver.cpp:258] Train net output #0: loss = 0.00562585 (* 1 = 0.00562585 loss)
I0526 20:53:32.204702 2966 sgd_solver.cpp:112] Iteration 7600, lr = 0.00654433
I0526 20:53:32.530794 2966 solver.cpp:239] Iteration 7700 (306.648 iter/s, 0.326107s/100 iters), loss = 0.0345423
I0526 20:53:32.530840 2966 solver.cpp:258] Train net output #0: loss = 0.0345422 (* 1 = 0.0345422 loss)
I0526 20:53:32.530869 2966 sgd_solver.cpp:112] Iteration 7700, lr = 0.00651658
I0526 20:53:32.903666 2966 solver.cpp:239] Iteration 7800 (268.217 iter/s, 0.372833s/100 iters), loss = 0.00340463
I0526 20:53:32.903729 2966 solver.cpp:258] Train net output #0: loss = 0.00340458 (* 1 = 0.00340458 loss)
I0526 20:53:32.903745 2966 sgd_solver.cpp:112] Iteration 7800, lr = 0.00648911
I0526 20:53:33.268342 2966 solver.cpp:239] Iteration 7900 (274.247 iter/s, 0.364634s/100 iters), loss = 0.00657078
I0526 20:53:33.268391 2966 solver.cpp:258] Train net output #0: loss = 0.00657073 (* 1 = 0.00657073 loss)
I0526 20:53:33.268440 2966 sgd_solver.cpp:112] Iteration 7900, lr = 0.0064619
I0526 20:53:33.611172 2966 solver.cpp:351] Iteration 8000, Testing net (#0)
I0526 20:53:33.722162 2980 data_layer.cpp:73] Restarting data prefetching from start.
I0526 20:53:33.723891 2966 solver.cpp:418] Test net output #0: accuracy = 0.991
I0526 20:53:33.723929 2966 solver.cpp:418] Test net output #1: loss = 0.0299871 (* 1 = 0.0299871 loss)
I0526 20:53:33.726871 2966 solver.cpp:239] Iteration 8000 (218.104 iter/s, 0.458498s/100 iters), loss = 0.0046433
I0526 20:53:33.726907 2966 solver.cpp:258] Train net output #0: loss = 0.00464324 (* 1 = 0.00464324 loss)
I0526 20:53:33.726915 2966 sgd_solver.cpp:112] Iteration 8000, lr = 0.00643496
I0526 20:53:34.063024 2966 solver.cpp:239] Iteration 8100 (297.524 iter/s, 0.336107s/100 iters), loss = 0.00935549
I0526 20:53:34.063122 2966 solver.cpp:258] Train net output #0: loss = 0.00935543 (* 1 = 0.00935543 loss)
I0526 20:53:34.063135 2966 sgd_solver.cpp:112] Iteration 8100, lr = 0.00640827
I0526 20:53:34.396461 2966 solver.cpp:239] Iteration 8200 (299.986 iter/s, 0.333349s/100 iters), loss = 0.00689925
I0526 20:53:34.396505 2966 solver.cpp:258] Train net output #0: loss = 0.0068992 (* 1 = 0.0068992 loss)
I0526 20:53:34.396514 2966 sgd_solver.cpp:112] Iteration 8200, lr = 0.00638185
I0526 20:53:34.730932 2966 solver.cpp:239] Iteration 8300 (299.018 iter/s, 0.334428s/100 iters), loss = 0.0434783
I0526 20:53:34.730994 2966 solver.cpp:258] Train net output #0: loss = 0.0434782 (* 1 = 0.0434782 loss)
I0526 20:53:34.731003 2966 sgd_solver.cpp:112] Iteration 8300, lr = 0.00635567
I0526 20:53:35.075024 2966 solver.cpp:239] Iteration 8400 (290.667 iter/s, 0.344036s/100 iters), loss = 0.00595309
I0526 20:53:35.075070 2966 solver.cpp:258] Train net output #0: loss = 0.00595304 (* 1 = 0.00595304 loss)
I0526 20:53:35.075081 2966 sgd_solver.cpp:112] Iteration 8400, lr = 0.00632975
I0526 20:53:35.198303 2979 data_layer.cpp:73] Restarting data prefetching from start.
I0526 20:53:35.417398 2966 solver.cpp:351] Iteration 8500, Testing net (#0)
I0526 20:53:35.527163 2980 data_layer.cpp:73] Restarting data prefetching from start.
I0526 20:53:35.530733 2966 solver.cpp:418] Test net output #0: accuracy = 0.9895
I0526 20:53:35.530761 2966 solver.cpp:418] Test net output #1: loss = 0.0312877 (* 1 = 0.0312877 loss)
I0526 20:53:35.533597 2966 solver.cpp:239] Iteration 8500 (218.083 iter/s, 0.458541s/100 iters), loss = 0.00686427
I0526 20:53:35.533622 2966 solver.cpp:258] Train net output #0: loss = 0.00686422 (* 1 = 0.00686422 loss)
I0526 20:53:35.533639 2966 sgd_solver.cpp:112] Iteration 8500, lr = 0.00630407
I0526 20:53:35.851109 2966 solver.cpp:239] Iteration 8600 (314.971 iter/s, 0.31749s/100 iters), loss = 0.000872626
I0526 20:53:35.851148 2966 solver.cpp:258] Train net output #0: loss = 0.000872567 (* 1 = 0.000872567 loss)
I0526 20:53:35.851156 2966 sgd_solver.cpp:112] Iteration 8600, lr = 0.00627864
I0526 20:53:36.169440 2966 solver.cpp:239] Iteration 8700 (314.174 iter/s, 0.318295s/100 iters), loss = 0.00289198
I0526 20:53:36.169492 2966 solver.cpp:258] Train net output #0: loss = 0.00289192 (* 1 = 0.00289192 loss)
I0526 20:53:36.169502 2966 sgd_solver.cpp:112] Iteration 8700, lr = 0.00625344
I0526 20:53:36.483927 2966 solver.cpp:239] Iteration 8800 (318.025 iter/s, 0.314441s/100 iters), loss = 0.000743419
I0526 20:53:36.483978 2966 solver.cpp:258] Train net output #0: loss = 0.000743358 (* 1 = 0.000743358 loss)
I0526 20:53:36.483983 2966 sgd_solver.cpp:112] Iteration 8800, lr = 0.00622847
I0526 20:53:36.799194 2966 solver.cpp:239] Iteration 8900 (317.221 iter/s, 0.315237s/100 iters), loss = 0.000569501
I0526 20:53:36.799244 2966 solver.cpp:258] Train net output #0: loss = 0.000569436 (* 1 = 0.000569436 loss)
I0526 20:53:36.799257 2966 sgd_solver.cpp:112] Iteration 8900, lr = 0.00620374
I0526 20:53:37.115048 2966 solver.cpp:351] Iteration 9000, Testing net (#0)
I0526 20:53:37.223479 2980 data_layer.cpp:73] Restarting data prefetching from start.
I0526 20:53:37.225188 2966 solver.cpp:418] Test net output #0: accuracy = 0.9906
I0526 20:53:37.225229 2966 solver.cpp:418] Test net output #1: loss = 0.0301926 (* 1 = 0.0301926 loss)
I0526 20:53:37.228317 2966 solver.cpp:239] Iteration 9000 (233.053 iter/s, 0.429087s/100 iters), loss = 0.0162924
I0526 20:53:37.228353 2966 solver.cpp:258] Train net output #0: loss = 0.0162924 (* 1 = 0.0162924 loss)
I0526 20:53:37.228361 2966 sgd_solver.cpp:112] Iteration 9000, lr = 0.00617924
I0526 20:53:37.545066 2966 solver.cpp:239] Iteration 9100 (315.737 iter/s, 0.316719s/100 iters), loss = 0.00710421
I0526 20:53:37.545107 2966 solver.cpp:258] Train net output #0: loss = 0.00710414 (* 1 = 0.00710414 loss)
I0526 20:53:37.545115 2966 sgd_solver.cpp:112] Iteration 9100, lr = 0.00615496
I0526 20:53:37.865255 2966 solver.cpp:239] Iteration 9200 (312.351 iter/s, 0.320152s/100 iters), loss = 0.00479173
I0526 20:53:37.865294 2966 solver.cpp:258] Train net output #0: loss = 0.00479167 (* 1 = 0.00479167 loss)
I0526 20:53:37.865303 2966 sgd_solver.cpp:112] Iteration 9200, lr = 0.0061309
I0526 20:53:38.185989 2966 solver.cpp:239] Iteration 9300 (311.819 iter/s, 0.320699s/100 iters), loss = 0.00574932
I0526 20:53:38.186033 2966 solver.cpp:258] Train net output #0: loss = 0.00574926 (* 1 = 0.00574926 loss)
I0526 20:53:38.186043 2966 sgd_solver.cpp:112] Iteration 9300, lr = 0.00610706
I0526 20:53:38.417049 2979 data_layer.cpp:73] Restarting data prefetching from start.
I0526 20:53:38.515170 2966 solver.cpp:239] Iteration 9400 (303.817 iter/s, 0.329145s/100 iters), loss = 0.0248899
I0526 20:53:38.515221 2966 solver.cpp:258] Train net output #0: loss = 0.0248899 (* 1 = 0.0248899 loss)
I0526 20:53:38.515228 2966 sgd_solver.cpp:112] Iteration 9400, lr = 0.00608343
I0526 20:53:38.829449 2966 solver.cpp:351] Iteration 9500, Testing net (#0)
I0526 20:53:38.939790 2980 data_layer.cpp:73] Restarting data prefetching from start.
I0526 20:53:38.942500 2966 solver.cpp:418] Test net output #0: accuracy = 0.9892
I0526 20:53:38.942539 2966 solver.cpp:418] Test net output #1: loss = 0.0346093 (* 1 = 0.0346093 loss)
I0526 20:53:38.945417 2966 solver.cpp:239] Iteration 9500 (232.435 iter/s, 0.430228s/100 iters), loss = 0.00292056
I0526 20:53:38.945453 2966 solver.cpp:258] Train net output #0: loss = 0.00292051 (* 1 = 0.00292051 loss)
I0526 20:53:38.945461 2966 sgd_solver.cpp:112] Iteration 9500, lr = 0.00606002
I0526 20:53:39.259587 2966 solver.cpp:239] Iteration 9600 (318.317 iter/s, 0.314152s/100 iters), loss = 0.00238432
I0526 20:53:39.259625 2966 solver.cpp:258] Train net output #0: loss = 0.00238427 (* 1 = 0.00238427 loss)
I0526 20:53:39.259632 2966 sgd_solver.cpp:112] Iteration 9600, lr = 0.00603682
I0526 20:53:39.582947 2966 solver.cpp:239] Iteration 9700 (309.285 iter/s, 0.323326s/100 iters), loss = 0.00316796
I0526 20:53:39.582990 2966 solver.cpp:258] Train net output #0: loss = 0.00316791 (* 1 = 0.00316791 loss)
I0526 20:53:39.582999 2966 sgd_solver.cpp:112] Iteration 9700, lr = 0.00601382
I0526 20:53:39.913578 2966 solver.cpp:239] Iteration 9800 (302.487 iter/s, 0.330592s/100 iters), loss = 0.0155842
I0526 20:53:39.913622 2966 solver.cpp:258] Train net output #0: loss = 0.0155842 (* 1 = 0.0155842 loss)
I0526 20:53:39.913633 2966 sgd_solver.cpp:112] Iteration 9800, lr = 0.00599102
I0526 20:53:40.224758 2966 solver.cpp:239] Iteration 9900 (321.395 iter/s, 0.311143s/100 iters), loss = 0.00502401
I0526 20:53:40.224807 2966 solver.cpp:258] Train net output #0: loss = 0.00502396 (* 1 = 0.00502396 loss)
I0526 20:53:40.224841 2966 sgd_solver.cpp:112] Iteration 9900, lr = 0.00596843
I0526 20:53:40.542547 2966 solver.cpp:468] Snapshotting to binary proto file examples/mnist/lenet_iter_10000.caffemodel
I0526 20:53:40.550356 2966 sgd_solver.cpp:280] Snapshotting solver state to binary proto file examples/mnist/lenet_iter_10000.solverstate
I0526 20:53:40.554579 2966 solver.cpp:331] Iteration 10000, loss = 0.00530681
I0526 20:53:40.554625 2966 solver.cpp:351] Iteration 10000, Testing net (#0)
I0526 20:53:40.663066 2980 data_layer.cpp:73] Restarting data prefetching from start.
I0526 20:53:40.664521 2966 solver.cpp:418] Test net output #0: accuracy = 0.99
I0526 20:53:40.664681 2966 solver.cpp:418] Test net output #1: loss = 0.0302442 (* 1 = 0.0302442 loss)
I0526 20:53:40.664695 2966 solver.cpp:336] Optimization Done.
I0526 20:53:40.664700 2966 caffe.cpp:250] Optimization Done.