Skip to first unread message
Junjie Zhang
Aug 3, 2015, 5:22:19 PM8/3/15
to Caffe Users
Hi, everyone
I met a problem. I tried to finetune a new dataset with VGG 16 layer network.
But I got the following error information.
I0804 06:48:29.185379 24930 caffe.cpp:113] Use GPU with device ID 0
I0804 06:48:29.380897 24930 caffe.cpp:121] Starting Optimization
I0804 06:48:29.380971 24930 solver.cpp:32] Initializing solver from parameters:
test_iter: 100
test_interval: 1000
base_lr: 0.001
display: 20
max_iter: 100000
lr_policy: "step"
gamma: 0.1
momentum: 0.9
weight_decay: 0.0005
stepsize: 20000
snapshot: 10000
snapshot_prefix: "models/vgg/vgg"
net: "models/vgg/vgg_train_val.prototxt"
I0804 06:48:29.381000 24930 solver.cpp:70] Creating training net from net file: models/vgg/vgg_train_val.prototxt
[libprotobuf ERROR google/protobuf/text_format.cc:245] Error parsing text-format caffe.NetParameter: 345:19: Message type "caffe.InnerProductParameter" has no field named "weight_filter".
F0804 06:48:29.381695 24930 upgrade_proto.cpp:928] Check failed: ReadProtoFromTextFile(param_file, param) Failed to parse NetParameter file: models/vgg/vgg_train_val.prototxt
*** Check failure stack trace: ***
@ 0x7f92eadfe9fd google::LogMessage::Fail()
@ 0x7f92eae0089d google::LogMessage::SendToLog()
@ 0x7f92eadfe5ec google::LogMessage::Flush()
@ 0x7f92eae011be google::LogMessageFatal::~LogMessageFatal()
@ 0x7f92eb24d3de caffe::ReadNetParamsFromTextFileOrDie()
@ 0x7f92eb14b882 caffe::Solver<>::InitTrainNet()
@ 0x7f92eb14c773 caffe::Solver<>::Init()
@ 0x7f92eb14c946 caffe::Solver<>::Solver()
@ 0x40cae0 caffe::GetSolver<>()
@ 0x406661 train()
@ 0x404be1 main
@ 0x7f92ea315ec5 (unknown)
@ 0x40518d (unknown)
It shows that I can't use weight_filter in inner_product-param, but it can be used in other training example.
The following is part of my configuration:
layers {
bottom: "conv5_2"
top: "conv5_3"
name: "conv5_3"
type: CONVOLUTION
convolution_param {
num_output: 512
pad: 1
kernel_size: 3
}
blobs_lr: 0
blobs_lr: 0
}
layers {
bottom: "conv5_3"
top: "conv5_3"
name: "relu5_3"
type: RELU
}
layers {
bottom: "conv5_3"
top: "pool5"
name: "pool5"
type: POOLING
pooling_param {
pool: MAX
kernel_size: 2
stride: 2
}
}
layers {
bottom: "pool5"
top: "vgg_fc6-pes"
name: "vgg_fc6-pes"
type: INNER_PRODUCT
inner_product_param {
num_output: 2048
weight_filter {
type: "gaussian"
std: 0.001
}
bias_filter {
type: "constant"
value: 0
}
}
blobs_lr: 0
blobs_lr: 0
}
layers {
bottom: "vgg_fc6-pes"
top: "vgg_fc6-pes"
name: "relu6"
type: RELU
}
layers {
bottom: "vgg_fc6-pes"
top: "vgg_fc6-pes"
name: "drop6"
type: DROPOUT
dropout_param {
dropout_ratio: 0.5
}
}
layers {
bottom: "vgg_fc6-pes"
top: "vgg_fc7-pes"
name: "vgg_fc7-pes"
type: INNER_PRODUCT
inner_product_param {
num_output: 2048
weight_filter {
type: "gaussian"
std: 0.001
}
bias_filter {
type: "constant"
value: 0
}
}
blobs_lr: 0
blobs_lr: 0
}
layers {
bottom: "vgg_fc7-pes"
top: "vgg_fc7-pes"
name: "relu7"
type: RELU
}
layers {
bottom: "vgg_fc7-pes"
top: "vgg_fc7-pes"
name: "drop7"
type: DROPOUT
dropout_param {
dropout_ratio: 0.5
}
}
layers {
name: "vgg_fc8-pes"
bottom: "vgg_fc7-pes"
top: "vgg_fc8-pes"
type: INNER_PRODUCT
inner_product_param {
num_output: 2
weight_filter {
type: "gaussian"
std: 0.001
}
bias_filter {
type: "constant"
value: 0
}
}
blobs_lr: 0
blobs_lr: 0
}
Any suggestion?
Thanks.
Yi Liu
Aug 4, 2015, 5:54:06 PM8/4/15
to Caffe Users
Not sure what weight_filter. Tentatively, maybe you meant weight_filler?
Reply all
Reply to author
Forward
0 new messages