OrgLion-ML
/
dlib
mirror of https://github.com/davisking/dlib.git
1
0
Fork 0
Code Issues Releases Wiki Activity

Fixed avg pooling filter sizes to avoid errors with the new rules about 

non-one based strides.
This commit is contained in:
Davis King 2016-05-04 21:40:29 -04:00
parent e4d70f8d3d
commit 4a7633056c
1 changed files with 4 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

8
examples/dnn_mnist_advanced_ex.cpp
View File

@ -61,7 +61,7 @@ template <typename SUBNET> using ares_down = base_ares<2,SUBNET>;
// large networks.
const unsigned long number_of_classes = 10;
using net_type = loss_multiclass_log<fc<number_of_classes,
avg_pool<11,11,11,11,
avg_pool<6,6,11,11,
res<res<res<res_down<
repeat<9,res, // repeat this layer 9 times
res_down<
@ -114,7 +114,7 @@ int main(int argc, char** argv) try
// Now, let's imagine we wanted to replace some of the relu layers with
// prelu layers. We might do it like this:
using net_type2 = loss_multiclass_log<fc<number_of_classes,
avg_pool<11,11,11,11,
avg_pool<6,6,11,11,
pres<res<res<res_down< // 2 prelu layers here
tag4<repeat<9,pres, // 9 groups, each containing 2 prelu layers
res_down<
@ -144,7 +144,7 @@ int main(int argc, char** argv) try
The pnet has 125 layers in it.
layer<0> loss_multiclass_log
layer<1> fc (num_outputs=10)
layer<2> avg_pool (nr=11, nc=11, stride_y=11, _stride_x=11)
layer<2> avg_pool (nr=6, nc=6, stride_y=11, _stride_x=11)
layer<3> prelu (initial_param_value=0.2)
layer<4> add_prev
layer<5> bn_con
@ -268,7 +268,7 @@ int main(int argc, char** argv) try
// making a network type which is identical to net_type but with the batch
// normalization layers replaced with affine. For example:
using test_net_type = loss_multiclass_log<fc<number_of_classes,
avg_pool<11,11,11,11,
avg_pool<6,6,11,11,
ares<ares<ares<ares_down<
repeat<9,res,
ares_down<