diff --git a/Makefile b/Makefile
index 1f310448..b1f93c4c 100644
--- a/Makefile
+++ b/Makefile
@@ -68,8 +68,8 @@ EXECOBJ = $(addprefix $(OBJDIR), $(EXECOBJA))
 OBJS = $(addprefix $(OBJDIR), $(OBJ))
 DEPS = $(wildcard src/*.h) Makefile include/darknet.h
 
-#all: obj backup results $(SLIB) $(ALIB) $(EXEC)
-all: obj  results $(SLIB) $(ALIB) $(EXEC)
+all: obj backup results $(SLIB) $(ALIB) $(EXEC)
+#all: obj  results $(SLIB) $(ALIB) $(EXEC)
 
 
 $(EXEC): $(EXECOBJ) $(ALIB)
diff --git a/cfg/alexnet.cfg b/cfg/alexnet.cfg
index 7e5a9b26..0458735f 100644
--- a/cfg/alexnet.cfg
+++ b/cfg/alexnet.cfg
@@ -90,6 +90,3 @@ activation=linear
 [softmax]
 groups=1
 
-[cost]
-type=sse
-
diff --git a/cfg/cifar.cfg b/cfg/cifar.cfg
index c10fae57..b2f69f53 100644
--- a/cfg/cifar.cfg
+++ b/cfg/cifar.cfg
@@ -119,6 +119,3 @@ activation=leaky
 
 [softmax]
 groups=1
-
-[cost]
-
diff --git a/cfg/cifar.test.cfg b/cfg/cifar.test.cfg
index d3afcdd7..18b6c54c 100644
--- a/cfg/cifar.test.cfg
+++ b/cfg/cifar.test.cfg
@@ -115,5 +115,3 @@ activation=leaky
 groups=1
 temperature=3
 
-[cost]
-
diff --git a/cfg/darknet19.cfg b/cfg/darknet19.cfg
index f56a46e2..28ac9669 100644
--- a/cfg/darknet19.cfg
+++ b/cfg/darknet19.cfg
@@ -203,6 +203,3 @@ activation=linear
 [softmax]
 groups=1
 
-[cost]
-type=sse
-
diff --git a/cfg/darknet19_448.cfg b/cfg/darknet19_448.cfg
index 133c688e..c6df7306 100644
--- a/cfg/darknet19_448.cfg
+++ b/cfg/darknet19_448.cfg
@@ -195,6 +195,3 @@ activation=linear
 [softmax]
 groups=1
 
-[cost]
-type=sse
-
diff --git a/cfg/densenet201.cfg b/cfg/densenet201.cfg
index 5e1e7dd1..65b4aecc 100644
--- a/cfg/densenet201.cfg
+++ b/cfg/densenet201.cfg
@@ -1949,6 +1949,3 @@ activation=linear
 [softmax]
 groups=1
 
-[cost]
-type=sse
-
diff --git a/cfg/extraction.cfg b/cfg/extraction.cfg
index 94e10675..e18d5c0d 100644
--- a/cfg/extraction.cfg
+++ b/cfg/extraction.cfg
@@ -201,6 +201,3 @@ activation=leaky
 [softmax]
 groups=1
 
-[cost]
-type=sse
-
diff --git a/cfg/extraction22k.cfg b/cfg/extraction22k.cfg
index 4cec6da9..b5f54090 100644
--- a/cfg/extraction22k.cfg
+++ b/cfg/extraction22k.cfg
@@ -204,6 +204,3 @@ activation=leaky
 [softmax]
 groups=1
 
-[cost]
-type=sse
-
diff --git a/cfg/go.cfg b/cfg/go.cfg
index 77d20c4d..c730092f 100644
--- a/cfg/go.cfg
+++ b/cfg/go.cfg
@@ -130,6 +130,3 @@ stride=1
 
 [softmax]
 
-[cost]
-type=sse
-
diff --git a/cfg/go.test.cfg b/cfg/go.test.cfg
index ba15c2e5..1e4e4380 100644
--- a/cfg/go.test.cfg
+++ b/cfg/go.test.cfg
@@ -129,6 +129,4 @@ stride=1
 
 [softmax]
 
-[cost]
-type=sse
 
diff --git a/cfg/gru.cfg b/cfg/gru.cfg
index a68d3fc5..4ce9b180 100644
--- a/cfg/gru.cfg
+++ b/cfg/gru.cfg
@@ -27,6 +27,4 @@ activation=linear
 
 [softmax]
 
-[cost]
-type=sse
 
diff --git a/cfg/resnet152.cfg b/cfg/resnet152.cfg
index d5fe9094..e8e3297a 100644
--- a/cfg/resnet152.cfg
+++ b/cfg/resnet152.cfg
@@ -1458,6 +1458,3 @@ activation=linear
 [softmax]
 groups=1
 
-[cost]
-type=sse
-
diff --git a/cfg/resnet50.cfg b/cfg/resnet50.cfg
index bfe69b18..8c4d01fe 100644
--- a/cfg/resnet50.cfg
+++ b/cfg/resnet50.cfg
@@ -506,6 +506,4 @@ activation=linear
 [softmax]
 groups=1
 
-[cost]
-type=sse
 
diff --git a/cfg/rnn.cfg b/cfg/rnn.cfg
index 68c032d2..61b202f3 100644
--- a/cfg/rnn.cfg
+++ b/cfg/rnn.cfg
@@ -35,6 +35,4 @@ activation=leaky
 
 [softmax]
 
-[cost]
-type=sse
 
diff --git a/cfg/rnn.train.cfg b/cfg/rnn.train.cfg
index 9139757f..b9748990 100644
--- a/cfg/rnn.train.cfg
+++ b/cfg/rnn.train.cfg
@@ -35,6 +35,4 @@ activation=leaky
 
 [softmax]
 
-[cost]
-type=sse
 
diff --git a/cfg/strided.cfg b/cfg/strided.cfg
index a52700b4..2f745085 100644
--- a/cfg/strided.cfg
+++ b/cfg/strided.cfg
@@ -180,6 +180,3 @@ activation=ramp
 
 [softmax]
 
-[cost]
-type=sse
-
diff --git a/cfg/tiny.cfg b/cfg/tiny.cfg
index ba686f56..f97327cf 100644
--- a/cfg/tiny.cfg
+++ b/cfg/tiny.cfg
@@ -171,6 +171,4 @@ activation=linear
 [softmax]
 groups=1
 
-[cost]
-type=sse
 
diff --git a/cfg/vgg-16.cfg b/cfg/vgg-16.cfg
index 2b6f7029..79694b87 100644
--- a/cfg/vgg-16.cfg
+++ b/cfg/vgg-16.cfg
@@ -148,6 +148,4 @@ activation=linear
 [softmax]
 groups=1
 
-[cost]
-type=sse
 
diff --git a/src/convolutional_kernels.cu b/src/convolutional_kernels.cu
index 8fa2ab2e..4a1047b8 100644
--- a/src/convolutional_kernels.cu
+++ b/src/convolutional_kernels.cu
@@ -111,9 +111,13 @@ void forward_convolutional_layer_gpu(convolutional_layer l, network net)
             float *a = l.weights_gpu + j*l.nweights/l.groups;
             float *b = net.workspace;
             float *c = l.output_gpu + (i*l.groups + j)*n*m;
+            float *im = net.input_gpu + (i*l.groups + j)*l.c/l.groups*l.h*l.w;
 
-            im2col_gpu(net.input_gpu + (i*l.groups + j)*l.c/l.groups*l.h*l.w,
-                l.c/l.groups, l.h, l.w, l.size, l.stride, l.pad, b);
+            if (l.size == 1){
+                b = im;
+            } else {
+                im2col_gpu(im, l.c/l.groups, l.h, l.w, l.size, l.stride, l.pad, b);
+            }
             gemm_gpu(0,0,m,n,k,1,a,k,b,n,1,c,n);
         }
     }
@@ -236,22 +240,26 @@ void backward_convolutional_layer_gpu(convolutional_layer l, network net)
             float *b = net.workspace;
             float *c = l.weight_updates_gpu + j*l.nweights/l.groups;
 
-            float *im = net.input_gpu+(i*l.groups + j)*l.c/l.groups*l.h*l.w;
+            float *im  = net.input_gpu+(i*l.groups + j)*l.c/l.groups*l.h*l.w;
+            float *imd = net.delta_gpu+(i*l.groups + j)*l.c/l.groups*l.h*l.w;
 
-            im2col_gpu(im, l.c/l.groups, l.h, l.w,
-                    l.size, l.stride, l.pad, b);
+            im2col_gpu(im, l.c/l.groups, l.h, l.w, l.size, l.stride, l.pad, b);
             gemm_gpu(0,1,m,n,k,1,a,k,b,k,1,c,n);
 
-            if(net.delta_gpu){
-                if(l.binary || l.xnor) swap_binary(&l);
+            if (net.delta_gpu) {
+                if (l.binary || l.xnor) swap_binary(&l);
                 a = l.weights_gpu + j*l.nweights/l.groups;
                 b = l.delta_gpu + (i*l.groups + j)*m*k;
                 c = net.workspace;
+                if (l.size == 1) {
+                    c = imd;
+                }
 
                 gemm_gpu(1,0,n,k,m,1,a,n,b,k,0,c,k);
 
-                col2im_gpu(net.workspace, l.c/l.groups, l.h, l.w, l.size, l.stride, 
-                    l.pad, net.delta_gpu + (i*l.groups + j)*l.c/l.groups*l.h*l.w);
+                if (l.size != 1) {
+                    col2im_gpu(net.workspace, l.c/l.groups, l.h, l.w, l.size, l.stride, l.pad, imd);
+                }
                 if(l.binary || l.xnor) {
                     swap_binary(&l);
                 }
diff --git a/src/convolutional_layer.c b/src/convolutional_layer.c
index e4fb9bde..5ac9ef0d 100644
--- a/src/convolutional_layer.c
+++ b/src/convolutional_layer.c
@@ -463,9 +463,13 @@ void forward_convolutional_layer(convolutional_layer l, network net)
             float *a = l.weights + j*l.nweights/l.groups;
             float *b = net.workspace;
             float *c = l.output + (i*l.groups + j)*n*m;
+            float *im =  net.input + (i*l.groups + j)*l.c/l.groups*l.h*l.w;
 
-            im2col_cpu(net.input + (i*l.groups + j)*l.c/l.groups*l.h*l.w,
-                l.c/l.groups, l.h, l.w, l.size, l.stride, l.pad, b);
+            if (l.size == 1) {
+                b = im;
+            } else {
+                im2col_cpu(im, l.c/l.groups, l.h, l.w, l.size, l.stride, l.pad, b);
+            }
             gemm(0,0,m,n,k,1,a,k,b,n,1,c,n);
         }
     }
@@ -501,21 +505,31 @@ void backward_convolutional_layer(convolutional_layer l, network net)
             float *b = net.workspace;
             float *c = l.weight_updates + j*l.nweights/l.groups;
 
-            float *im = net.input+(i*l.groups + j)*l.c/l.groups*l.h*l.w;
+            float *im  = net.input + (i*l.groups + j)*l.c/l.groups*l.h*l.w;
+            float *imd = net.delta + (i*l.groups + j)*l.c/l.groups*l.h*l.w;
+
+            if(l.size == 1){
+                b = im;
+            } else {
+                im2col_cpu(im, l.c/l.groups, l.h, l.w, 
+                        l.size, l.stride, l.pad, b);
+            }
 
-            im2col_cpu(im, l.c/l.groups, l.h, l.w, 
-                    l.size, l.stride, l.pad, b);
             gemm(0,1,m,n,k,1,a,k,b,k,1,c,n);
 
-            if(net.delta){
+            if (net.delta) {
                 a = l.weights + j*l.nweights/l.groups;
                 b = l.delta + (i*l.groups + j)*m*k;
                 c = net.workspace;
+                if (l.size == 1) {
+                    c = imd;
+                }
 
                 gemm(1,0,n,k,m,1,a,n,b,k,0,c,k);
 
-                col2im_cpu(net.workspace, l.c/l.groups, l.h, l.w, l.size, l.stride, 
-                    l.pad, net.delta + (i*l.groups + j)*l.c/l.groups*l.h*l.w);
+                if (l.size != 1) {
+                    col2im_cpu(net.workspace, l.c/l.groups, l.h, l.w, l.size, l.stride, l.pad, imd);
+                }
             }
         }
     }