cxxnet/example/ImageNet/ImageNet.conf at master · voidException/cxxnet · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
# Configuration for ImageNet
# Acknowledgement:
#  Ref: http://www.cs.toronto.edu/~fritz/absps/imagenet.pdf
#  The scheduling parameters is adapted from Caffe(http://caffe.berkeleyvision.org/)

data = train
iter = imgbin
  image_list = "../../NameList.train"
  image_bin  = "../../TRAIN.BIN"
  image_root = "../../data/resize256/"
  image_mean = "models/image_net_mean.bin"
  rand_crop=1
  rand_mirror=1
iter = threadbuffer
iter = end

eval = test
iter = imgbin
  image_list = "../../NameList.test"
  image_bin = "../../TEST.BIN"
  image_root = "../../data/resize256/"
  image_mean = "models/image_net_mean.bin"
# no random crop and mirror in test
iter = end

netconfig=start
layer[0->1] = conv
  kernel_size = 11
  stride = 4
  nchannel = 96
layer[1->2] = relu
layer[2->3] = max_pooling
  kernel_size = 3
  stride = 2
layer[3->4] = lrn
  local_size = 5
  alpha = 0.001
  beta = 0.75
  knorm = 1
###############
layer[4->5] = conv
  ngroup = 2
  nchannel = 256
  kernel_size = 5
  pad = 2
layer[5->6] = relu
layer[6->7] = max_pooling
  kernel_size = 3
  stride = 2
layer[7->8] = lrn
  local_size = 5
  alpha = 0.001
  beta = 0.75
  knorm = 1
#############
layer[8->9] = conv
  nchannel = 384
  kernel_size = 3
  pad = 1
layer[9->10]= relu
layer[10->11] = conv
  nchannel = 384
  ngroup = 2
  kernel_size = 3
  pad = 1
layer[11->12] = relu
layer[12->13] = conv
  nchannel = 256
  ngroup = 2
  kernel_size = 3
  pad = 1
  init_bias = 1.0
layer[13->14] = relu
layer[14->15] = max_pooling
  kernel_size = 3
  stride = 2
layer[15->16] = flatten
layer[16->17] = fullc
  nhidden = 4096
  init_sigma = 0.005
  init_bias = 1.0
layer[17->18] = relu
layer[18->18] = dropout
  threshold = 0.5
layer[18->19] = fullc
  nhidden = 4096
  init_sigma = 0.005
  init_bias = 1.0
layer[19->20] = relu
layer[20->20] = dropout
  threshold = 0.5
layer[20->21] = fullc
  nhidden = 1000
layer[21->21] = softmax
netconfig=end

# evaluation metric
metric = error
metric = rec@1
metric = rec@5

max_round = 45
num_round = 45

# input shape not including batch
input_shape = 3,227,227

batch_size = 256

# global parameters in any sectiion outside netconfig, and iter
momentum = 0.9
wmat:lr  = 0.01
wmat:wd  = 0.0005

bias:wd  = 0.000
bias:lr  = 0.02

# all the learning rate schedule starts with lr
lr:schedule = expdecay
lr:gamma = 0.1
lr:step = 100000

save_model=1
model_dir=models

# random config
random_type = xavier


# new line