diff --git a/.gitignore b/.gitignore
index 2299a477..bea19ff4 100644
--- a/.gitignore
+++ b/.gitignore
@@ -4,6 +4,7 @@
 *.out
 *.png
 *.jpg
+*.pyc
 old/
 mnist/
 data/
diff --git a/Makefile b/Makefile
index 36a451cc..5c226af7 100644
--- a/Makefile
+++ b/Makefile
@@ -10,21 +10,21 @@ ARCH= -gencode arch=compute_20,code=[sm_20,sm_21] \
       -gencode arch=compute_52,code=[sm_52,compute_52]
 
 # This is what I use, uncomment if you know your arch and want to specify
-# ARCH=  -gencode arch=compute_52,code=compute_52
+ARCH= -gencode arch=compute_52,code=compute_52
 
 VPATH=./src/:./examples
-LIB=libdarknet.a
+LIB=libdarknet.so
 EXEC=darknet
 OBJDIR=./obj/
 
 CC=gcc
-NVCC=nvcc 
+NVCC=nvcc --compiler-options '-fPIC'
 AR=ar
 ARFLAGS=-rv
 OPTS=-Ofast
 LDFLAGS= -lm -pthread 
 COMMON= -Iinclude/ -Isrc/
-CFLAGS=-Wall -Wfatal-errors 
+CFLAGS=-Wall -Wfatal-errors -fPIC
 
 ifeq ($(DEBUG), 1) 
 OPTS=-O0 -g
@@ -69,7 +69,7 @@ $(EXEC): $(EXECOBJ) $(LIB)
 	$(CC) $(COMMON) $(CFLAGS) $^ -o $@ $(LDFLAGS) $(LIB)
 
 $(LIB): $(OBJS)
-	$(AR) $(ARFLAGS) $@ $^ 
+	$(CC) $(CFLAGS) -shared $^ -o $@ $(LDFLAGS)
 
 $(OBJDIR)%.o: %.c $(DEPS)
 	$(CC) $(COMMON) $(CFLAGS) -c $< -o $@
diff --git a/cfg/cifar.cfg b/cfg/cifar.cfg
index f2c801a5..c10fae57 100644
--- a/cfg/cifar.cfg
+++ b/cfg/cifar.cfg
@@ -1,25 +1,23 @@
 [net]
 batch=128
 subdivisions=1
-height=32
-width=32
+height=28
+width=28
 channels=3
-momentum=0.9
-decay=0.0005
+max_crop=32
+min_crop=32
+
+hue=.1
+saturation=.75
+exposure=.75
 
 learning_rate=0.4
 policy=poly
 power=4
-max_batches = 50000
+max_batches = 5000
+momentum=0.9
+decay=0.0005
 
-[crop]
-crop_width=28
-crop_height=28
-flip=1
-angle=0
-saturation = 1
-exposure = 1
-noadjust=1
 
 [convolutional]
 batch_normalize=1
diff --git a/examples/classifier.c b/examples/classifier.c
index 112b913d..645c80d4 100644
--- a/examples/classifier.c
+++ b/examples/classifier.c
@@ -33,7 +33,6 @@ void train_classifier(char *datacfg, char *cfgfile, char *weightfile, int *gpus,
         cuda_set_device(gpus[i]);
 #endif
         nets[i] = load_network(cfgfile, weightfile, clear);
-        nets[i].learning_rate *= ngpus;
     }
     srand(time(0));
     network net = nets[0];
diff --git a/examples/coco.c b/examples/coco.c
index bab507d9..a07906e4 100644
--- a/examples/coco.c
+++ b/examples/coco.c
@@ -313,7 +313,7 @@ void test_coco(char *cfgfile, char *weightfile, char *filename, float thresh)
     if(weightfile){
         load_weights(&net, weightfile);
     }
-    detection_layer l = net.layers[net.n-1];
+    layer l = net.layers[net.n-1];
     set_batch_network(&net, 1);
     srand(2222222);
     float nms = .4;
diff --git a/examples/detector.c b/examples/detector.c
index 0a31fc27..e6b57b8c 100644
--- a/examples/detector.c
+++ b/examples/detector.c
@@ -22,12 +22,7 @@ void train_detector(char *datacfg, char *cfgfile, char *weightfile, int *gpus, i
 #ifdef GPU
         cuda_set_device(gpus[i]);
 #endif
-        nets[i] = parse_network_cfg(cfgfile);
-        if(weightfile){
-            load_weights(&nets[i], weightfile);
-        }
-        if(clear) *nets[i].seen = 0;
-        nets[i].learning_rate *= ngpus;
+        nets[i] = load_network(cfgfile, weightfile, clear);
     }
     srand(time(0));
     network net = nets[0];
diff --git a/examples/lsd.c b/examples/lsd.c
index 85641ecd..1b2b0b5f 100644
--- a/examples/lsd.c
+++ b/examples/lsd.c
@@ -451,7 +451,7 @@ void train_dcgan(char *cfg, char *weight, char *acfg, char *aweight, int clear,
     printf("%s\n", base);
     network gnet = load_network(cfg, weight, clear);
     network anet = load_network(acfg, aweight, clear);
-    float orig_rate = anet.learning_rate;
+    //float orig_rate = anet.learning_rate;
 
     int start = 0;
     int i, j, k;
@@ -494,7 +494,7 @@ void train_dcgan(char *cfg, char *weight, char *acfg, char *aweight, int clear,
     int y_size = gnet.truths*gnet.batch;
     float *imerror = cuda_make_array(0, y_size);
 
-    int ay_size = anet.truths*anet.batch;
+    //int ay_size = anet.truths*anet.batch;
 
     float aloss_avg = -1;
 
@@ -664,14 +664,14 @@ void train_colorizer(char *cfg, char *weight, char *acfg, char *aweight, int cle
     clock_t time;
 
     int x_size = net.inputs*net.batch;
-    int y_size = x_size;
+    //int y_size = x_size;
     net.delta = 0;
     net.train = 1;
     float *pixs = calloc(x_size, sizeof(float));
     float *graypixs = calloc(x_size, sizeof(float));
-    float *y = calloc(y_size, sizeof(float));
+    //float *y = calloc(y_size, sizeof(float));
 
-    int ay_size = anet.outputs*anet.batch;
+    //int ay_size = anet.outputs*anet.batch;
     anet.delta = 0;
     anet.train = 1;
 
diff --git a/examples/nightmare.c b/examples/nightmare.c
index 9538238d..fe7c5f7f 100644
--- a/examples/nightmare.c
+++ b/examples/nightmare.c
@@ -1,5 +1,7 @@
 #include "darknet.h"
 
+#include <math.h>
+
 // ./darknet nightmare cfg/extractor.recon.cfg ~/trained/yolo-coco.conv frame6.png -reconstruct -iters 500 -i 3 -lambda .1 -rate .01 -smooth 2
 
 float abs_mean(float *x, int n)
@@ -128,11 +130,11 @@ void smooth(image recon, image update, float lambda, int num)
 void reconstruct_picture(network net, float *features, image recon, image update, float rate, float momentum, float lambda, int smooth_size, int iters)
 {
     int iter = 0;
-    layer l = get_network_output_layer(net);
     for (iter = 0; iter < iters; ++iter) {
         image delta = make_image(recon.w, recon.h, recon.c);
 
 #ifdef GPU
+        layer l = get_network_output_layer(net);
         cuda_push_array(net.input_gpu, recon.data, recon.w*recon.h*recon.c);
         //cuda_push_array(net.truth_gpu, features, net.truths);
         net.delta_gpu = cuda_make_array(delta.data, delta.w*delta.h*delta.c);
diff --git a/examples/rnn.c b/examples/rnn.c
index 0ea4daab..05c8f37b 100644
--- a/examples/rnn.c
+++ b/examples/rnn.c
@@ -1,5 +1,7 @@
 #include "darknet.h"
 
+#include <math.h>
+
 typedef struct {
     float *x;
     float *y;
diff --git a/examples/segmenter.c b/examples/segmenter.c
index f24bbdaa..fe3d498e 100644
--- a/examples/segmenter.c
+++ b/examples/segmenter.c
@@ -24,7 +24,6 @@ void train_segmenter(char *datacfg, char *cfgfile, char *weightfile, int *gpus,
             load_weights(&nets[i], weightfile);
         }
         if(clear) *nets[i].seen = 0;
-        nets[i].learning_rate *= ngpus;
     }
     srand(time(0));
     network net = nets[0];
@@ -76,6 +75,15 @@ void train_segmenter(char *datacfg, char *cfgfile, char *weightfile, int *gpus,
         pthread_join(load_thread, 0);
         train = buffer;
         load_thread = load_data(args);
+        image tr = float_to_image(net.w, net.h, 81, train.y.vals[0]);
+        image im = float_to_image(net.w, net.h, net.c, train.X.vals[0]);
+        image mask = mask_to_rgb(tr);
+        show_image(im, "input");
+        show_image(mask, "truth");
+#ifdef OPENCV
+        cvWaitKey(100);
+#endif
+        free_image(mask);
 
         printf("Loaded: %lf seconds\n", sec(clock()-time));
         time=clock();
diff --git a/examples/yolo.c b/examples/yolo.c
index e6ac873e..5b3fd163 100644
--- a/examples/yolo.c
+++ b/examples/yolo.c
@@ -279,7 +279,7 @@ void test_yolo(char *cfgfile, char *weightfile, char *filename, float thresh)
     if(weightfile){
         load_weights(&net, weightfile);
     }
-    detection_layer l = net.layers[net.n-1];
+    layer l = net.layers[net.n-1];
     set_batch_network(&net, 1);
     srand(2222222);
     clock_t time;
diff --git a/include/darknet.h b/include/darknet.h
index 986d4309..c78ea10f 100644
--- a/include/darknet.h
+++ b/include/darknet.h
@@ -1,7 +1,11 @@
 #ifndef DARKNET_API
 #define DARKNET_API
 #include <stdlib.h>
+#include <stdio.h>
+#include <string.h>
+#include <pthread.h>
 
+#define SECRET_NUM -1234
 extern int gpu_index;
 
 #ifdef GPU
@@ -27,6 +31,13 @@ extern int gpu_index;
     #endif
 #endif
 
+typedef struct{
+    int classes;
+    char **names;
+} metadata;
+
+metadata get_metadata(char *file);
+
 typedef struct{
     int *leaf;
     int n;
@@ -42,7 +53,7 @@ typedef struct{
 
 typedef enum{
     LOGISTIC, RELU, RELIE, LINEAR, RAMP, TANH, PLSE, LEAKY, ELU, LOGGY, STAIR, HARDTAN, LHTAN
-}ACTIVATION;
+} ACTIVATION;
 
 typedef enum {
     CONVOLUTIONAL,
@@ -255,7 +266,7 @@ struct layer{
 
     size_t workspace_size;
 
-    #ifdef GPU
+#ifdef GPU
     int *indexes_gpu;
 
     float *z_gpu;
@@ -281,8 +292,8 @@ struct layer{
     float * concat_gpu;
     float * concat_delta_gpu;
 
-    float *binary_input_gpu;
-    float *binary_weights_gpu;
+    float * binary_input_gpu;
+    float * binary_weights_gpu;
 
     float * mean_gpu;
     float * variance_gpu;
@@ -297,19 +308,22 @@ struct layer{
     float * x_norm_gpu;
     float * weights_gpu;
     float * weight_updates_gpu;
+    float * weight_change_gpu;
 
     float * biases_gpu;
     float * bias_updates_gpu;
+    float * bias_change_gpu;
 
     float * scales_gpu;
     float * scale_updates_gpu;
+    float * scale_change_gpu;
 
     float * output_gpu;
     float * delta_gpu;
     float * rand_gpu;
     float * squared_gpu;
     float * norms_gpu;
-    #ifdef CUDNN
+#ifdef CUDNN
     cudnnTensorDescriptor_t srcTensorDesc, dstTensorDesc;
     cudnnTensorDescriptor_t dsrcTensorDesc, ddstTensorDesc;
     cudnnTensorDescriptor_t normTensorDesc;
@@ -319,8 +333,8 @@ struct layer{
     cudnnConvolutionFwdAlgo_t fw_algo;
     cudnnConvolutionBwdDataAlgo_t bd_algo;
     cudnnConvolutionBwdFilterAlgo_t bf_algo;
-    #endif
-    #endif
+#endif
+#endif
 };
 
 void free_layer(layer);
@@ -383,12 +397,12 @@ typedef struct network{
     int index;
     float *cost;
 
-    #ifdef GPU
+#ifdef GPU
     float *input_gpu;
     float *truth_gpu;
     float *delta_gpu;
     float *output_gpu;
-    #endif
+#endif
 
 } network;
 
@@ -403,8 +417,8 @@ typedef struct {
 } augment_args;
 
 typedef struct {
-    int h;
     int w;
+    int h;
     int c;
     float *data;
 } image;
@@ -472,6 +486,7 @@ typedef struct{
 
 
 network load_network(char *cfg, char *weights, int clear);
+network *load_network_p(char *cfg, char *weights, int clear);
 load_args get_base_args(network net);
 
 void free_data(data d);
@@ -492,47 +507,171 @@ pthread_t load_data(load_args args);
 list *read_data_cfg(char *filename);
 list *read_cfg(char *filename);
 
+void forward_network(network net);
+void backward_network(network net);
+void update_network(network net);
+
+
+void axpy_cpu(int N, float ALPHA, float *X, int INCX, float *Y, int INCY);
+void copy_cpu(int N, float *X, int INCX, float *Y, int INCY);
+void scal_cpu(int N, float ALPHA, float *X, int INCX);
+void normalize_cpu(float *x, float *mean, float *variance, int batch, int filters, int spatial);
+
+int best_3d_shift_r(image a, image b, int min, int max);
+#ifdef GPU
+void axpy_ongpu(int N, float ALPHA, float * X, int INCX, float * Y, int INCY);
+void fill_ongpu(int N, float ALPHA, float * X, int INCX);
+void scal_ongpu(int N, float ALPHA, float * X, int INCX);
+void copy_ongpu(int N, float * X, int INCX, float * Y, int INCY);
+
+void cuda_set_device(int n);
+void cuda_free(float *x_gpu);
+float *cuda_make_array(float *x, size_t n);
+void cuda_pull_array(float *x_gpu, float *x, size_t n);
+float cuda_mag_array(float *x_gpu, size_t n);
+void cuda_push_array(float *x_gpu, float *x, size_t n);
+
+void forward_network_gpu(network net);
+void backward_network_gpu(network net);
+void update_network_gpu(network net);
+
+float train_networks(network *nets, int n, data d, int interval);
+void sync_nets(network *nets, int n, int interval);
+void harmless_update_network_gpu(network net);
+#endif
+void save_image_png(image im, const char *name);
+void get_next_batch(data d, int n, int offset, float *X, float *y);
+void grayscale_image_3c(image im);
+void normalize_image(image p);
+void matrix_to_csv(matrix m);
+float train_network_sgd(network net, data d, int n);
+void rgbgr_image(image im);
+data copy_data(data d);
+data concat_data(data d1, data d2);
+data load_cifar10_data(char *filename);
+float matrix_topk_accuracy(matrix truth, matrix guess, int k);
+void matrix_add_matrix(matrix from, matrix to);
+void scale_matrix(matrix m, float scale);
+matrix csv_to_matrix(char *filename);
+float *network_accuracies(network net, data d, int n);
+float train_network_datum(network net);
+image make_random_image(int w, int h, int c);
+
+void denormalize_connected_layer(layer l);
+void denormalize_convolutional_layer(layer l);
+void statistics_connected_layer(layer l);
+void rescale_weights(layer l, float scale, float trans);
+void rgbgr_weights(layer l);
+image *get_weights(layer l);
+
+void demo(char *cfgfile, char *weightfile, float thresh, int cam_index, const char *filename, char **names, int classes, int frame_skip, char *prefix, int avg, float hier_thresh, int w, int h, int fps, int fullscreen);
+void get_detection_boxes(layer l, int w, int h, float thresh, float **probs, box *boxes, int only_objectness);
+
+char *option_find_str(list *l, char *key, char *def);
+int option_find_int(list *l, char *key, int def);
+
+network parse_network_cfg(char *filename);
+void save_weights(network net, char *filename);
+void load_weights(network *net, char *filename);
+void save_weights_upto(network net, char *filename, int cutoff);
+void load_weights_upto(network *net, char *filename, int start, int cutoff);
+
+void zero_objectness(layer l);
+void get_region_boxes(layer l, int w, int h, int netw, int neth, float thresh, float **probs, box *boxes, int only_objectness, int *map, float tree_thresh, int relative);
+void free_network(network net);
+void set_batch_network(network *net, int b);
+image load_image(char *filename, int w, int h, int c);
+image load_image_color(char *filename, int w, int h);
+image make_image(int w, int h, int c);
+image resize_image(image im, int w, int h);
+image letterbox_image(image im, int w, int h);
+image crop_image(image im, int dx, int dy, int w, int h);
+image resize_min(image im, int min);
+image threshold_image(image im, float thresh);
+image mask_to_rgb(image mask);
+int resize_network(network *net, int w, int h);
+void free_matrix(matrix m);
+void test_resize(char *filename);
+void save_image(image p, const char *name);
+void show_image(image p, const char *name);
+image copy_image(image p);
+void draw_box_width(image a, int x1, int y1, int x2, int y2, int w, float r, float g, float b);
+float get_current_rate(network net);
+void composite_3d(char *f1, char *f2, char *out, int delta);
+data load_data_old(char **paths, int n, int m, char **labels, int k, int w, int h);
+int get_current_batch(network net);
+void constrain_image(image im);
+image get_network_image_layer(network net, int i);
+layer get_network_output_layer(network net);
+void top_predictions(network net, int n, int *index);
+void flip_image(image a);
+image float_to_image(int w, int h, int c, float *data);
+void ghost_image(image source, image dest, int dx, int dy);
+float network_accuracy(network net, data d);
+void random_distort_image(image im, float hue, float saturation, float exposure);
+void fill_image(image m, float s);
+image grayscale_image(image im);
+void rotate_image_cw(image im, int times);
+image rotate_image(image m, float rad);
+void visualize_network(network net);
+float box_iou(box a, box b);
+void do_nms(box *boxes, float **probs, int total, int classes, float thresh);
+data load_all_cifar10();
+box_label *read_boxes(char *filename, int *n);
+void draw_detections(image im, int num, float thresh, box *boxes, float **probs, char **names, image **labels, int classes);
+
+matrix network_predict_data(network net, data test);
+image **load_alphabet();
+image get_network_image(network net);
+float *network_predict(network net, float *input);
+float *network_predict_p(network *net, float *input);
+
+int network_width(network *net);
+int network_height(network *net);
+float *network_predict_image(network *net, image im);
+
+char **get_labels(char *filename);
+void do_nms_sort(box *boxes, float **probs, int total, int classes, float thresh);
+void do_nms_obj(box *boxes, float **probs, int total, int classes, float thresh);
+
+matrix make_matrix(int rows, int cols);
+
+#ifndef __cplusplus
+#ifdef OPENCV
+image get_image_from_stream(CvCapture *cap);
+#endif
+#endif
+void free_image(image m);
+float train_network(network net, data d);
+pthread_t load_data_in_thread(load_args args);
+list *get_paths(char *filename);
+void hierarchy_predictions(float *predictions, int n, tree *hier, int only_leaves, int stride);
+void change_leaves(tree *t, char *leaf_list);
+
+int find_int_arg(int argc, char **argv, char *arg, int def);
+float find_float_arg(int argc, char **argv, char *arg, float def);
+int find_arg(int argc, char* argv[], char *arg);
+char *find_char_arg(int argc, char **argv, char *arg, char *def);
+char *basecfg(char *cfgfile);
+void find_replace(char *str, char *orig, char *rep, char *output);
+void free_ptrs(void **ptrs, int n);
+char *fgetl(FILE *fp);
+void strip(char *s);
+float sec(clock_t clocks);
+void **list_to_array(list *l);
+void top_k(float *a, int n, int k, int *index);
+int *read_map(char *filename);
+void error(const char *s);
+int max_index(float *a, int n);
+int sample_array(float *a, int n);
+void free_list(list *l);
+float mse_array(float *a, int n);
+float variance_array(float *a, int n);
+float mag_array(float *a, int n);
+float mean_array(float *a, int n);
+void normalize_array(float *a, int n);
+int *read_intlist(char *s, int *n, int d);
+size_t rand_size_t();
+float rand_normal();
 
-#include "activation_layer.h"
-#include "activations.h"
-#include "avgpool_layer.h"
-#include "batchnorm_layer.h"
-#include "blas.h"
-#include "box.h"
-#include "classifier.h"
-#include "col2im.h"
-#include "connected_layer.h"
-#include "convolutional_layer.h"
-#include "cost_layer.h"
-#include "crnn_layer.h"
-#include "crop_layer.h"
-#include "cuda.h"
-#include "data.h"
-#include "deconvolutional_layer.h"
-#include "demo.h"
-#include "detection_layer.h"
-#include "dropout_layer.h"
-#include "gemm.h"
-#include "gru_layer.h"
-#include "im2col.h"
-#include "image.h"
-#include "layer.h"
-#include "list.h"
-#include "local_layer.h"
-#include "matrix.h"
-#include "maxpool_layer.h"
-#include "network.h"
-#include "normalization_layer.h"
-#include "option_list.h"
-#include "parser.h"
-#include "region_layer.h"
-#include "reorg_layer.h"
-#include "rnn_layer.h"
-#include "route_layer.h"
-#include "shortcut_layer.h"
-#include "softmax_layer.h"
-#include "stb_image.h"
-#include "stb_image_write.h"
-#include "tree.h"
-#include "utils.h"
 #endif
diff --git a/python/darknet.py b/python/darknet.py
new file mode 100644
index 00000000..e23e5383
--- /dev/null
+++ b/python/darknet.py
@@ -0,0 +1,70 @@
+from ctypes import *
+
+class IMAGE(Structure):
+    _fields_ = [("w", c_int),
+                ("h", c_int),
+                ("c", c_int),
+                ("data", POINTER(c_float))]
+
+class METADATA(Structure):
+    _fields_ = [("classes", c_int),
+                ("names", POINTER(c_char_p))]
+
+lib = CDLL("/home/pjreddie/documents/darknet/libdarknet.so", RTLD_GLOBAL)
+lib.network_width.argtypes = [c_void_p]
+lib.network_width.restype = c_int
+lib.network_height.argtypes = [c_void_p]
+lib.network_height.restype = c_int
+
+def load_meta(f):
+    lib.get_metadata.argtypes = [c_char_p]
+    lib.get_metadata.restype = METADATA
+    return lib.get_metadata(f)
+
+def load_net(cfg, weights):
+    load_network = lib.load_network_p
+    load_network.argtypes = [c_char_p, c_char_p, c_int]
+    load_network.restype = c_void_p
+    return load_network(cfg, weights, 0)
+
+def load_img(f):
+    load_image = lib.load_image_color
+    load_image.argtypes = [c_char_p, c_int, c_int]
+    load_image.restype = IMAGE
+    return load_image(f, 0, 0)
+
+def letterbox_img(im, w, h):
+    letterbox_image = lib.letterbox_image
+    letterbox_image.argtypes = [IMAGE, c_int, c_int]
+    letterbox_image.restype = IMAGE
+    return letterbox_image(im, w, h)
+
+def predict(net, im):
+    pred = lib.network_predict_image
+    pred.argtypes = [c_void_p, IMAGE]
+    pred.restype = POINTER(c_float)
+    return pred(net, im)
+
+def classify(net, meta, im):
+    out = predict(net, im)
+    res = []
+    for i in range(meta.classes):
+        res.append((meta.names[i], out[i]))
+    res = sorted(res, key=lambda x: -x[1])
+    return res
+
+def detect(net, meta, im):
+    out = predict(net, im)
+    res = []
+    for i in range(meta.classes):
+        res.append((meta.names[i], out[i]))
+    res = sorted(res, key=lambda x: -x[1])
+    return res
+
+if __name__ == "__main__":
+    net = load_net("cfg/densenet.cfg", "/home/pjreddie/trained/densenet201.weights")
+    im = load_img("data/wolf.jpg")
+    meta = load_meta("cfg/imagenet1k.data")
+    r = classify(net, meta, im)
+    print r[:10]
+
diff --git a/src/blas.h b/src/blas.h
index 49ff0875..7b0218f6 100644
--- a/src/blas.h
+++ b/src/blas.h
@@ -1,5 +1,7 @@
 #ifndef BLAS_H
 #define BLAS_H
+#include "darknet.h"
+
 void flatten(float *x, int size, int layers, int batch, int forward);
 void pm(int M, int N, float *A);
 float *random_matrix(int rows, int cols);
@@ -13,9 +15,6 @@ void constrain_ongpu(int N, float ALPHA, float * X, int INCX);
 void pow_cpu(int N, float ALPHA, float *X, int INCX, float *Y, int INCY);
 void mul_cpu(int N, float *X, int INCX, float *Y, int INCY);
 
-void axpy_cpu(int N, float ALPHA, float *X, int INCX, float *Y, int INCY);
-void copy_cpu(int N, float *X, int INCX, float *Y, int INCY);
-void scal_cpu(int N, float ALPHA, float *X, int INCX);
 void fill_cpu(int N, float ALPHA, float * X, int INCX);
 float dot_cpu(int N, float *X, int INCX, float *Y, int INCY);
 int test_gpu_blas();
@@ -23,7 +22,6 @@ void shortcut_cpu(int batch, int w1, int h1, int c1, float *add, int w2, int h2,
 
 void mean_cpu(float *x, int batch, int filters, int spatial, float *mean);
 void variance_cpu(float *x, float *mean, int batch, int filters, int spatial, float *variance);
-void normalize_cpu(float *x, float *mean, float *variance, int batch, int filters, int spatial);
 
 void scale_bias(float *output, float *scales, int batch, int n, int size);
 void backward_scale_cpu(float *x_norm, float *delta, int batch, int n, int size, float *scale_updates);
@@ -47,14 +45,12 @@ void axpy_ongpu(int N, float ALPHA, float * X, int INCX, float * Y, int INCY);
 void axpy_ongpu_offset(int N, float ALPHA, float * X, int OFFX, int INCX, float * Y, int OFFY, int INCY);
 void copy_ongpu(int N, float * X, int INCX, float * Y, int INCY);
 void copy_ongpu_offset(int N, float * X, int OFFX, int INCX, float * Y, int OFFY, int INCY);
-void scal_ongpu(int N, float ALPHA, float * X, int INCX);
 void add_ongpu(int N, float ALPHA, float * X, int INCX);
 void supp_ongpu(int N, float ALPHA, float * X, int INCX);
 void mask_ongpu(int N, float * X, float mask_num, float * mask);
 void const_ongpu(int N, float ALPHA, float *X, int INCX);
 void pow_ongpu(int N, float ALPHA, float *X, int INCX, float *Y, int INCY);
 void mul_ongpu(int N, float *X, int INCX, float *Y, int INCY);
-void fill_ongpu(int N, float ALPHA, float * X, int INCX);
 
 void mean_gpu(float *x, int batch, int filters, int spatial, float *mean);
 void variance_gpu(float *x, float *mean, int batch, int filters, int spatial, float *variance);
diff --git a/src/box.h b/src/box.h
index 59b38a58..35e8bdc9 100644
--- a/src/box.h
+++ b/src/box.h
@@ -7,12 +7,8 @@ typedef struct{
 } dbox;
 
 box float_to_box(float *f, int stride);
-float box_iou(box a, box b);
 float box_rmse(box a, box b);
 dbox diou(box a, box b);
-void do_nms(box *boxes, float **probs, int total, int classes, float thresh);
-void do_nms_sort(box *boxes, float **probs, int total, int classes, float thresh);
-void do_nms_obj(box *boxes, float **probs, int total, int classes, float thresh);
 box decode_box(box b, box anchor);
 box encode_box(box b, box anchor);
 
diff --git a/src/connected_layer.h b/src/connected_layer.h
index 62dd497f..067b2566 100644
--- a/src/connected_layer.h
+++ b/src/connected_layer.h
@@ -12,8 +12,6 @@ connected_layer make_connected_layer(int batch, int inputs, int outputs, ACTIVAT
 void forward_connected_layer(connected_layer layer, network net);
 void backward_connected_layer(connected_layer layer, network net);
 void update_connected_layer(connected_layer layer, int batch, float learning_rate, float momentum, float decay);
-void denormalize_connected_layer(layer l);
-void statistics_connected_layer(layer l);
 
 #ifdef GPU
 void forward_connected_layer_gpu(connected_layer layer, network net);
diff --git a/src/convolutional_layer.h b/src/convolutional_layer.h
index d25ef649..91394c38 100644
--- a/src/convolutional_layer.h
+++ b/src/convolutional_layer.h
@@ -26,7 +26,6 @@ void cudnn_convolutional_setup(layer *l);
 #endif
 
 convolutional_layer make_convolutional_layer(int batch, int h, int w, int c, int n, int size, int stride, int padding, ACTIVATION activation, int batch_normalize, int binary, int xnor, int adam);
-void denormalize_convolutional_layer(convolutional_layer l);
 void resize_convolutional_layer(convolutional_layer *layer, int w, int h);
 void forward_convolutional_layer(const convolutional_layer layer, network net);
 void update_convolutional_layer(convolutional_layer layer, int batch, float learning_rate, float momentum, float decay);
@@ -40,15 +39,12 @@ void backward_convolutional_layer(convolutional_layer layer, network net);
 void add_bias(float *output, float *biases, int batch, int n, int size);
 void backward_bias(float *bias_updates, float *delta, int batch, int n, int size);
 
-image *get_weights(convolutional_layer l);
 image get_convolutional_image(convolutional_layer layer);
 image get_convolutional_delta(convolutional_layer layer);
 image get_convolutional_weight(convolutional_layer layer, int i);
 
 int convolutional_out_height(convolutional_layer layer);
 int convolutional_out_width(convolutional_layer layer);
-void rescale_weights(convolutional_layer l, float scale, float trans);
-void rgbgr_weights(convolutional_layer l);
 
 #endif
 
diff --git a/src/cuda.c b/src/cuda.c
index e011b89d..e5114037 100644
--- a/src/cuda.c
+++ b/src/cuda.c
@@ -96,6 +96,8 @@ float *cuda_make_array(float *x, size_t n)
     if(x){
         status = cudaMemcpy(x_gpu, x, size, cudaMemcpyHostToDevice);
         check_error(status);
+    } else {
+        fill_ongpu(n, 0, x_gpu, 1);
     }
     if(!x_gpu) error("Cuda malloc failed\n");
     return x_gpu;
diff --git a/src/cuda.h b/src/cuda.h
index cdaba208..a1bc2160 100644
--- a/src/cuda.h
+++ b/src/cuda.h
@@ -7,16 +7,10 @@
 
 void check_error(cudaError_t status);
 cublasHandle_t blas_handle();
-float *cuda_make_array(float *x, size_t n);
 int *cuda_make_int_array(int *x, size_t n);
-void cuda_push_array(float *x_gpu, float *x, size_t n);
-void cuda_pull_array(float *x_gpu, float *x, size_t n);
-void cuda_set_device(int n);
-void cuda_free(float *x_gpu);
 void cuda_random(float *x_gpu, size_t n);
 float cuda_compare(float *x_gpu, float *x, size_t n, char *s);
 dim3 cuda_gridsize(size_t n);
-float cuda_mag_array(float *x_gpu, size_t n);
 
 #ifdef CUDNN
 cudnnHandle_t cudnn_handle();
diff --git a/src/data.c b/src/data.c
index dcc5089b..c227b6f8 100644
--- a/src/data.c
+++ b/src/data.c
@@ -535,16 +535,17 @@ void or_image(image src, image dest, int c)
     }
 }
 
-void fill_bg_mask(image m)
+void exclusive_image(image src)
 {
-    int i,k;
-    int index = m.w*m.h*(m.c-1);
-    for(i = 0; i < m.w*m.h; ++i){
-        m.data[index + i] = 1;
-    }
-    for(k = 0; k < m.c-1; ++k){
-        for(i = 0; i < m.w*m.h; ++i){
-            if(m.data[index + i] && m.data[k*m.w*m.h + i]) m.data[index + i] = 0; 
+    int k, j, i;
+    int s = src.w*src.h;
+    for(k = 0; k < src.c-1; ++k){
+        for(i = 0; i < s; ++i){
+            if (src.data[k*s + i]){
+                for(j = k+1; j < src.c; ++j){
+                    src.data[j*s + i] = 0;
+                }
+            }
         }
     }
 }
@@ -558,6 +559,10 @@ image get_segmentation_image(char *path, int w, int h, int classes)
     find_replace(labelpath, ".JPG", ".txt", labelpath);
     find_replace(labelpath, ".JPEG", ".txt", labelpath);
     image mask = make_image(w, h, classes+1);
+    int i;
+    for(i = 0; i < w*h; ++i){
+        mask.data[w*h*classes + i] = 1;
+    }
     FILE *file = fopen(labelpath, "r");
     if(!file) file_error(labelpath);
     char buff[32788];
@@ -568,9 +573,12 @@ image get_segmentation_image(char *path, int w, int h, int classes)
         int *rle = read_intlist(buff, &n, 0);
         load_rle(part, rle, n);
         or_image(part, mask, id);
+        for(i = 0; i < w*h; ++i){
+            if(part.data[i]) mask.data[w*h*classes + i] = 0;
+        }
         free(rle);
     }
-    fill_bg_mask(mask);
+    //exclusive_image(mask);
     fclose(file);
     free_image(part);
     return mask;
diff --git a/src/data.h b/src/data.h
index 8f18d9f6..931b99b2 100644
--- a/src/data.h
+++ b/src/data.h
@@ -20,12 +20,10 @@ static inline float distance_from_edge(int x, int max)
 }
 void load_data_blocking(load_args args);
 
-pthread_t load_data_in_thread(load_args args);
 
 void print_letters(float *pred, int n);
 data load_data_captcha(char **paths, int n, int m, int k, int w, int h);
 data load_data_captcha_encode(char **paths, int n, int m, int w, int h);
-data load_data_old(char **paths, int n, int m, char **labels, int k, int w, int h);
 data load_data_detection(int n, char **paths, int m, int w, int h, int boxes, int classes, float jitter, float hue, float saturation, float exposure);
 data load_data_tag(char **paths, int n, int m, int k, int min, int max, int size, float angle, float aspect, float hue, float saturation, float exposure);
 matrix load_image_augment_paths(char **paths, int n, int min, int max, int size, float angle, float aspect, float hue, float saturation, float exposure, int center);
@@ -34,27 +32,19 @@ data load_data_augment(char **paths, int n, int m, char **labels, int k, tree *h
 data load_data_regression(char **paths, int n, int m, int min, int max, int size, float angle, float aspect, float hue, float saturation, float exposure);
 data load_go(char *filename);
 
-box_label *read_boxes(char *filename, int *n);
-data load_cifar10_data(char *filename);
-data load_all_cifar10();
 
 data load_data_writing(char **paths, int n, int m, int w, int h, int out_w, int out_h);
 
-list *get_paths(char *filename);
-char **get_labels(char *filename);
 void get_random_batch(data d, int n, float *X, float *y);
 data get_data_part(data d, int part, int total);
 data get_random_data(data d, int num);
-void get_next_batch(data d, int n, int offset, float *X, float *y);
 data load_categorical_data_csv(char *filename, int target, int k);
 void normalize_data_rows(data d);
 void scale_data_rows(data d, float s);
 void translate_data_rows(data d, float s);
 void randomize_data(data d);
 data *split_data(data d, int part, int total);
-data concat_data(data d1, data d2);
 data concat_datas(data *d, int n);
 void fill_truth(char *path, char **labels, int k, float *truth);
-data copy_data(data d);
 
 #endif
diff --git a/src/demo.h b/src/demo.h
index d920759c..86e46541 100644
--- a/src/demo.h
+++ b/src/demo.h
@@ -2,6 +2,5 @@
 #define DEMO_H
 
 #include "image.h"
-void demo(char *cfgfile, char *weightfile, float thresh, int cam_index, const char *filename, char **names, int classes, int frame_skip, char *prefix, int avg, float hier_thresh, int w, int h, int fps, int fullscreen);
 
 #endif
diff --git a/src/detection_layer.h b/src/detection_layer.h
index fecfed04..1c818535 100644
--- a/src/detection_layer.h
+++ b/src/detection_layer.h
@@ -9,7 +9,6 @@ typedef layer detection_layer;
 detection_layer make_detection_layer(int batch, int inputs, int n, int size, int classes, int coords, int rescore);
 void forward_detection_layer(const detection_layer l, network net);
 void backward_detection_layer(const detection_layer l, network net);
-void get_detection_boxes(layer l, int w, int h, float thresh, float **probs, box *boxes, int only_objectness);
 
 #ifdef GPU
 void forward_detection_layer_gpu(const detection_layer l, network net);
diff --git a/src/image.c b/src/image.c
index 1a8c984a..146ec355 100644
--- a/src/image.c
+++ b/src/image.c
@@ -36,9 +36,9 @@ image mask_to_rgb(image mask)
         float green = get_color(1,offset,n);
         float blue = get_color(0,offset,n);
         for(i = 0; i < im.w*im.h; ++i){
-            im.data[i + 0*im.w*im.h] = mask.data[j*im.h*im.w + i]*red;
-            im.data[i + 1*im.w*im.h] = mask.data[j*im.h*im.w + i]*green;
-            im.data[i + 2*im.w*im.h] = mask.data[j*im.h*im.w + i]*blue;
+            im.data[i + 0*im.w*im.h] += mask.data[j*im.h*im.w + i]*red;
+            im.data[i + 1*im.w*im.h] += mask.data[j*im.h*im.w + i]*green;
+            im.data[i + 2*im.w*im.h] += mask.data[j*im.h*im.w + i]*blue;
         }
     }
     return im;
diff --git a/src/image.h b/src/image.h
index 8b516c14..02c79f08 100644
--- a/src/image.h
+++ b/src/image.h
@@ -11,7 +11,6 @@
 
 #ifndef __cplusplus
 #ifdef OPENCV
-image get_image_from_stream(CvCapture *cap);
 int fill_image_from_stream(CvCapture *cap, image im);
 image ipl_to_image(IplImage* src);
 void ipl_into_image(IplImage* src, image im);
@@ -20,36 +19,22 @@ void show_image_cv(image p, const char *name, IplImage *disp);
 #endif
 #endif
 
-image mask_to_rgb(image mask);
 float get_color(int c, int x, int max);
-void flip_image(image a);
 void draw_box(image a, int x1, int y1, int x2, int y2, float r, float g, float b);
-void draw_box_width(image a, int x1, int y1, int x2, int y2, int w, float r, float g, float b);
 void draw_bbox(image a, box bbox, int w, float r, float g, float b);
 void draw_label(image a, int r, int c, image label, const float *rgb);
 void write_label(image a, int r, int c, image *characters, char *string, float *rgb);
-void draw_detections(image im, int num, float thresh, box *boxes, float **probs, char **names, image **labels, int classes);
 image image_distance(image a, image b);
 void scale_image(image m, float s);
-image crop_image(image im, int dx, int dy, int w, int h);
 image rotate_crop_image(image im, float rad, float s, int w, int h, float dx, float dy, float aspect);
 image center_crop_image(image im, int w, int h);
 image random_crop_image(image im, int w, int h);
 image random_augment_image(image im, float angle, float aspect, int low, int high, int w, int h);
 augment_args random_augment_args(image im, float angle, float aspect, int low, int high, int w, int h);
-void random_distort_image(image im, float hue, float saturation, float exposure);
-image letterbox_image(image im, int w, int h);
 void letterbox_image_into(image im, int w, int h, image boxed);
-image resize_image(image im, int w, int h);
-image resize_min(image im, int min);
 image resize_max(image im, int max);
-void fill_image(image m, float s);
 void translate_image(image m, float s);
-void normalize_image(image p);
-image rotate_image(image m, float rad);
-void rotate_image_cw(image im, int times);
 void embed_image(image source, image dest, int dx, int dy);
-void ghost_image(image source, image dest, int dx, int dy);
 void place_image(image im, int w, int h, int dx, int dy, image canvas);
 void saturate_image(image im, float sat);
 void exposure_image(image im, float sat);
@@ -59,38 +44,21 @@ void rgb_to_hsv(image im);
 void hsv_to_rgb(image im);
 void yuv_to_rgb(image im);
 void rgb_to_yuv(image im);
-void rgbgr_image(image im);
-void constrain_image(image im);
-void composite_3d(char *f1, char *f2, char *out, int delta);
-int best_3d_shift_r(image a, image b, int min, int max);
 
-image grayscale_image(image im);
-void grayscale_image_3c(image im);
-image threshold_image(image im, float thresh);
 
 image collapse_image_layers(image source, int border);
 image collapse_images_horz(image *ims, int n);
 image collapse_images_vert(image *ims, int n);
 
-void show_image(image p, const char *name);
 void show_image_normalized(image im, const char *name);
-void save_image_png(image im, const char *name);
-void save_image(image p, const char *name);
 void show_images(image *ims, int n, char *window);
 void show_image_layers(image p, char *name);
 void show_image_collapsed(image p, char *name);
 
 void print_image(image m);
 
-image make_image(int w, int h, int c);
-image make_random_image(int w, int h, int c);
 image make_empty_image(int w, int h, int c);
-image float_to_image(int w, int h, int c, float *data);
-image copy_image(image p);
 void copy_image_into(image src, image dest);
-image load_image(char *filename, int w, int h, int c);
-image load_image_color(char *filename, int w, int h);
-image **load_alphabet();
 
 float get_pixel(image m, int x, int y, int c);
 float get_pixel_extend(image m, int x, int y, int c);
@@ -100,7 +68,5 @@ float bilinear_interpolate(image im, float x, float y, int c);
 
 image get_image_layer(image m, int l);
 
-void free_image(image m);
-void test_resize(char *filename);
 #endif
 
diff --git a/src/list.h b/src/list.h
index 214b2805..6b445c71 100644
--- a/src/list.h
+++ b/src/list.h
@@ -7,9 +7,7 @@ int list_find(list *l, void *val);
 
 void list_insert(list *, void *);
 
-void **list_to_array(list *l);
 
-void free_list(list *l);
 void free_list_contents(list *l);
 
 #endif
diff --git a/src/matrix.h b/src/matrix.h
index 14afcd75..879acd70 100644
--- a/src/matrix.h
+++ b/src/matrix.h
@@ -2,17 +2,10 @@
 #define MATRIX_H
 #include "darknet.h"
 
-matrix make_matrix(int rows, int cols);
 matrix copy_matrix(matrix m);
-void free_matrix(matrix m);
 void print_matrix(matrix m);
 
-matrix csv_to_matrix(char *filename);
-void matrix_to_csv(matrix m);
 matrix hold_out_matrix(matrix *m, int n);
-float matrix_topk_accuracy(matrix truth, matrix guess, int k);
-void matrix_add_matrix(matrix from, matrix to);
-void scale_matrix(matrix m, float scale);
 matrix resize_matrix(matrix m, int size);
 
 float *pop_column(matrix *m, int c);
diff --git a/src/network.c b/src/network.c
index 2b21338d..c5a44b5a 100644
--- a/src/network.c
+++ b/src/network.c
@@ -58,6 +58,13 @@ network load_network(char *cfg, char *weights, int clear)
     return net;
 }
 
+network *load_network_p(char *cfg, char *weights, int clear)
+{
+    network *net = calloc(1, sizeof(network));
+    *net = load_network(cfg, weights, clear);
+    return net;
+}
+
 int get_current_batch(network net)
 {
     int batch_num = (*net.seen)/(net.batch*net.subdivisions);
@@ -439,6 +446,23 @@ float *network_predict(network net, float *input)
     return net.output;
 }
 
+float *network_predict_p(network *net, float *input)
+{
+    return network_predict(*net, input);
+}
+
+float *network_predict_image(network *net, image im)
+{
+    image imr = letterbox_image(im, net->w, net->h);
+    set_batch_network(net, 1);
+    float *p = network_predict(*net, imr.data);
+    free_image(imr);
+    return p;
+}
+
+int network_width(network *net){return net->w;}
+int network_height(network *net){return net->h;}
+
 matrix network_predict_data_multi(network net, data test, int n)
 {
     int i,j,b,m;
diff --git a/src/network.h b/src/network.h
index 15a644e5..4e956cbf 100644
--- a/src/network.h
+++ b/src/network.h
@@ -10,46 +10,21 @@
 
 
 #ifdef GPU
-float train_networks(network *nets, int n, data d, int interval);
-void sync_nets(network *nets, int n, int interval);
 float train_network_datum_gpu(network net);
 float *network_predict_gpu(network net, float *input);
 void pull_network_output(network net);
-void forward_network_gpu(network net);
-void backward_network_gpu(network net);
-void update_network_gpu(network net);
-void harmless_update_network_gpu(network net);
 #endif
 
-float get_current_rate(network net);
-int get_current_batch(network net);
-void free_network(network net);
 void compare_networks(network n1, network n2, data d);
 char *get_layer_string(LAYER_TYPE a);
 
 network make_network(int n);
-void forward_network(network net);
-void backward_network(network net);
-void update_network(network net);
 
-float train_network(network net, data d);
-float train_network_sgd(network net, data d, int n);
-float train_network_datum(network net);
 
-matrix network_predict_data(network net, data test);
-float *network_predict(network net, float *input);
-float network_accuracy(network net, data d);
-float *network_accuracies(network net, data d, int n);
 float network_accuracy_multi(network net, data d, int n);
-void top_predictions(network net, int n, int *index);
-image get_network_image(network net);
-image get_network_image_layer(network net, int i);
-layer get_network_output_layer(network net);
 int get_predicted_class_network(network net);
 void print_network(network net);
-void visualize_network(network net);
 int resize_network(network *net, int w, int h);
-void set_batch_network(network *net, int b);
 void calc_network_cost(network net);
 
 #endif
diff --git a/src/network_kernels.cu b/src/network_kernels.cu
index 2fbc2dd4..28aa20ec 100644
--- a/src/network_kernels.cu
+++ b/src/network_kernels.cu
@@ -149,6 +149,74 @@ pthread_t train_network_in_thread(network net, data d, float *err)
     return thread;
 }
 
+void merge_weights(layer l, layer base)
+{
+    if (l.type == CONVOLUTIONAL) {
+        axpy_cpu(l.n, 1, l.bias_updates, 1, base.biases, 1);
+        axpy_cpu(l.n*l.size*l.size*l.c, 1, l.weight_updates, 1, base.weights, 1);
+        if (l.scales) {
+            axpy_cpu(l.n, 1, l.scale_updates, 1, base.scales, 1);
+        }
+    } else if(l.type == CONNECTED) {
+        axpy_cpu(l.outputs, 1, l.bias_updates, 1, base.biases, 1);
+        axpy_cpu(l.outputs*l.inputs, 1, l.weight_updates, 1, base.weights, 1);
+    }
+}
+
+void scale_weights(layer l, float s)
+{
+    if (l.type == CONVOLUTIONAL) {
+        scal_cpu(l.n, s, l.biases, 1);
+        scal_cpu(l.n*l.size*l.size*l.c, s, l.weights, 1);
+        if (l.scales) {
+            scal_cpu(l.n, s, l.scales, 1);
+        }
+    } else if(l.type == CONNECTED) {
+        scal_cpu(l.outputs, s, l.biases, 1);
+        scal_cpu(l.outputs*l.inputs, s, l.weights, 1);
+    }
+}
+
+
+void pull_weights(layer l)
+{
+    if(l.type == CONVOLUTIONAL || l.type == DECONVOLUTIONAL){
+        cuda_pull_array(l.biases_gpu, l.bias_updates, l.n);
+        cuda_pull_array(l.weights_gpu, l.weight_updates, l.n*l.size*l.size*l.c);
+        if(l.scales) cuda_pull_array(l.scales_gpu, l.scale_updates, l.n);
+    } else if(l.type == CONNECTED){
+        cuda_pull_array(l.biases_gpu, l.bias_updates, l.outputs);
+        cuda_pull_array(l.weights_gpu, l.weight_updates, l.outputs*l.inputs);
+    }
+}
+
+void push_weights(layer l)
+{
+    if(l.type == CONVOLUTIONAL || l.type == DECONVOLUTIONAL){
+        cuda_push_array(l.biases_gpu, l.biases, l.n);
+        cuda_push_array(l.weights_gpu, l.weights, l.n*l.size*l.size*l.c);
+        if(l.scales) cuda_push_array(l.scales_gpu, l.scales, l.n);
+    } else if(l.type == CONNECTED){
+        cuda_push_array(l.biases_gpu, l.biases, l.outputs);
+        cuda_push_array(l.weights_gpu, l.weights, l.outputs*l.inputs);
+    }
+}
+
+void distribute_weights(layer l, layer base)
+{
+    if (l.type == CONVOLUTIONAL || l.type == DECONVOLUTIONAL) {
+        cuda_push_array(l.biases_gpu, base.biases, l.n);
+        cuda_push_array(l.weights_gpu, base.weights, l.n*l.size*l.size*l.c);
+        if (base.scales) cuda_push_array(l.scales_gpu, base.scales, l.n);
+    } else if (l.type == CONNECTED) {
+        cuda_push_array(l.biases_gpu, base.biases, l.outputs);
+        cuda_push_array(l.weights_gpu, base.weights, l.outputs*l.inputs);
+    }
+}
+
+
+/*
+
 void pull_updates(layer l)
 {
     if(l.type == CONVOLUTIONAL){
@@ -182,73 +250,6 @@ void update_layer(layer l, network net)
         l.update_gpu(l, update_batch, rate*l.learning_rate_scale, net.momentum, net.decay);
     }
 }
-
-void merge_weights(layer l, layer base)
-{
-    if (l.type == CONVOLUTIONAL) {
-        axpy_cpu(l.n, 1, l.biases, 1, base.biases, 1);
-        axpy_cpu(l.n*l.size*l.size*l.c, 1, l.weights, 1, base.weights, 1);
-        if (l.scales) {
-            axpy_cpu(l.n, 1, l.scales, 1, base.scales, 1);
-        }
-    } else if(l.type == CONNECTED) {
-        axpy_cpu(l.outputs, 1, l.biases, 1, base.biases, 1);
-        axpy_cpu(l.outputs*l.inputs, 1, l.weights, 1, base.weights, 1);
-    }
-}
-
-void scale_weights(layer l, float s)
-{
-    if (l.type == CONVOLUTIONAL) {
-        scal_cpu(l.n, s, l.biases, 1);
-        scal_cpu(l.n*l.size*l.size*l.c, s, l.weights, 1);
-        if (l.scales) {
-            scal_cpu(l.n, s, l.scales, 1);
-        }
-    } else if(l.type == CONNECTED) {
-        scal_cpu(l.outputs, s, l.biases, 1);
-        scal_cpu(l.outputs*l.inputs, s, l.weights, 1);
-    }
-}
-
-
-void pull_weights(layer l)
-{
-    if(l.type == CONVOLUTIONAL){
-        cuda_pull_array(l.biases_gpu, l.biases, l.n);
-        cuda_pull_array(l.weights_gpu, l.weights, l.n*l.size*l.size*l.c);
-        if(l.scales) cuda_pull_array(l.scales_gpu, l.scales, l.n);
-    } else if(l.type == CONNECTED){
-        cuda_pull_array(l.biases_gpu, l.biases, l.outputs);
-        cuda_pull_array(l.weights_gpu, l.weights, l.outputs*l.inputs);
-    }
-}
-
-void push_weights(layer l)
-{
-    if(l.type == CONVOLUTIONAL){
-        cuda_push_array(l.biases_gpu, l.biases, l.n);
-        cuda_push_array(l.weights_gpu, l.weights, l.n*l.size*l.size*l.c);
-        if(l.scales) cuda_push_array(l.scales_gpu, l.scales, l.n);
-    } else if(l.type == CONNECTED){
-        cuda_push_array(l.biases_gpu, l.biases, l.outputs);
-        cuda_push_array(l.weights_gpu, l.weights, l.outputs*l.inputs);
-    }
-}
-
-void distribute_weights(layer l, layer base)
-{
-    if(l.type == CONVOLUTIONAL){
-        cuda_push_array(l.biases_gpu, base.biases, l.n);
-        cuda_push_array(l.weights_gpu, base.weights, l.n*l.size*l.size*l.c);
-        if(base.scales) cuda_push_array(l.scales_gpu, base.scales, l.n);
-    } else if(l.type == CONNECTED){
-        cuda_push_array(l.biases_gpu, base.biases, l.outputs);
-        cuda_push_array(l.weights_gpu, base.weights, l.outputs*l.inputs);
-    }
-}
-
-
 void merge_updates(layer l, layer base)
 {
     if (l.type == CONVOLUTIONAL) {
@@ -265,7 +266,7 @@ void merge_updates(layer l, layer base)
 
 void distribute_updates(layer l, layer base)
 {
-    if(l.type == CONVOLUTIONAL){
+    if(l.type == CONVOLUTIONAL || l.type == DECONVOLUTIONAL){
         cuda_push_array(l.bias_updates_gpu, base.bias_updates, l.n);
         cuda_push_array(l.weight_updates_gpu, base.weight_updates, l.n*l.size*l.size*l.c);
         if(base.scale_updates) cuda_push_array(l.scale_updates_gpu, base.scale_updates, l.n);
@@ -274,16 +275,37 @@ void distribute_updates(layer l, layer base)
         cuda_push_array(l.weight_updates_gpu, base.weight_updates, l.outputs*l.inputs);
     }
 }
+*/
 
+/*
 void sync_layer(network *nets, int n, int j)
 {
-    //printf("Syncing layer %d\n", j);
     int i;
     network net = nets[0];
     layer base = net.layers[j];
-    cuda_set_device(net.gpu_index);
-    pull_weights(base);
-    for (i = 1; i < n; ++i) {
+    scale_weights(base, 0);
+    for (i = 0; i < n; ++i) {
+        cuda_set_device(nets[i].gpu_index);
+        layer l = nets[i].layers[j];
+        pull_weights(l);
+        merge_weights(l, base);
+    }
+    scale_weights(base, 1./n);
+    for (i = 0; i < n; ++i) {
+        cuda_set_device(nets[i].gpu_index);
+        layer l = nets[i].layers[j];
+        distribute_weights(l, base);
+    }
+}
+*/
+
+void sync_layer(network *nets, int n, int j)
+{
+    int i;
+    network net = nets[0];
+    layer base = net.layers[j];
+    scale_weights(base, 0);
+    for (i = 0; i < n; ++i) {
         cuda_set_device(nets[i].gpu_index);
         layer l = nets[i].layers[j];
         pull_weights(l);
@@ -295,7 +317,6 @@ void sync_layer(network *nets, int n, int j)
         layer l = nets[i].layers[j];
         distribute_weights(l, base);
     }
-    //printf("Done syncing layer %d\n", j);
 }
 
 typedef struct{
diff --git a/src/option_list.c b/src/option_list.c
index f935af30..2f52781f 100644
--- a/src/option_list.c
+++ b/src/option_list.c
@@ -32,6 +32,23 @@ list *read_data_cfg(char *filename)
     return options;
 }
 
+metadata get_metadata(char *file)
+{
+    metadata m = {0};
+    list *options = read_data_cfg(file);
+
+    char *name_list = option_find_str(options, "names", 0);
+    if(!name_list) name_list = option_find_str(options, "labels", 0);
+    if(!name_list) {
+        fprintf(stderr, "No names or labels found\n");
+    } else {
+        m.names = get_labels(name_list);
+    }
+    m.classes = option_find_int(options, "classes", 2);
+    free_list(options);
+    return m;
+}
+
 int read_option(char *s, list *options)
 {
     size_t i;
diff --git a/src/option_list.h b/src/option_list.h
index d0417aa9..3b8b7da1 100644
--- a/src/option_list.h
+++ b/src/option_list.h
@@ -12,8 +12,6 @@ typedef struct{
 int read_option(char *s, list *options);
 void option_insert(list *l, char *key, char *val);
 char *option_find(list *l, char *key);
-char *option_find_str(list *l, char *key, char *def);
-int option_find_int(list *l, char *key, int def);
 int option_find_int_quiet(list *l, char *key, int def);
 float option_find_float(list *l, char *key, float def);
 float option_find_float_quiet(list *l, char *key, float def);
diff --git a/src/parser.h b/src/parser.h
index 473f21a9..81aef2c8 100644
--- a/src/parser.h
+++ b/src/parser.h
@@ -1,13 +1,9 @@
 #ifndef PARSER_H
 #define PARSER_H
+#include "darknet.h"
 #include "network.h"
 
-network parse_network_cfg(char *filename);
 void save_network(network net, char *filename);
-void save_weights(network net, char *filename);
-void save_weights_upto(network net, char *filename, int cutoff);
 void save_weights_double(network net, char *filename);
-void load_weights(network *net, char *filename);
-void load_weights_upto(network *net, char *filename, int start, int cutoff);
 
 #endif
diff --git a/src/region_layer.c b/src/region_layer.c
index 1f88bf77..d2f73029 100644
--- a/src/region_layer.c
+++ b/src/region_layer.c
@@ -445,6 +445,15 @@ void forward_region_layer_gpu(const layer l, network net)
         }
     }
     if (l.softmax_tree){
+        int mmin = 9000;
+        int mmax = 0;
+        int i;
+        for(i = 0; i < l.softmax_tree->groups; ++i){
+            int group_size = l.softmax_tree->group_size[i];
+            if (group_size < mmin) mmin = group_size;
+            if (group_size > mmax) mmax = group_size;
+        }
+        printf("%d %d %d \n", l.softmax_tree->groups, mmin, mmax);
         int index = entry_index(l, 0, 0, 5);
         softmax_tree(net.input_gpu + index, l.w*l.h, l.batch*l.n, l.inputs/l.n, 1, l.output_gpu + index, *l.softmax_tree);
         /*
diff --git a/src/region_layer.h b/src/region_layer.h
index 03b1a3d2..f67f9016 100644
--- a/src/region_layer.h
+++ b/src/region_layer.h
@@ -1,15 +1,14 @@
 #ifndef REGION_LAYER_H
 #define REGION_LAYER_H
 
+#include "darknet.h"
 #include "layer.h"
 #include "network.h"
 
 layer make_region_layer(int batch, int h, int w, int n, int classes, int coords);
 void forward_region_layer(const layer l, network net);
 void backward_region_layer(const layer l, network net);
-void get_region_boxes(layer l, int w, int h, int netw, int neth, float thresh, float **probs, box *boxes, int only_objectness, int *map, float tree_thresh, int relative);
 void resize_region_layer(layer *l, int w, int h);
-void zero_objectness(layer l);
 
 #ifdef GPU
 void forward_region_layer_gpu(const layer l, network net);
diff --git a/src/tree.h b/src/tree.h
index ec74057b..100858a3 100644
--- a/src/tree.h
+++ b/src/tree.h
@@ -3,8 +3,6 @@
 #include "darknet.h"
 
 tree *read_tree(char *filename);
-void hierarchy_predictions(float *predictions, int n, tree *hier, int only_leaves, int stride);
-void change_leaves(tree *t, char *leaf_list);
 int hierarchy_top_prediction(float *predictions, tree *hier, float thresh, int stride);
 float get_hierarchy_probability(float *x, tree *hier, int c, int stride);
 
diff --git a/src/utils.h b/src/utils.h
index 27206533..1593e62b 100644
--- a/src/utils.h
+++ b/src/utils.h
@@ -2,18 +2,15 @@
 #define UTILS_H
 #include <stdio.h>
 #include <time.h>
+#include "darknet.h"
 #include "list.h"
 
-#define SECRET_NUM -1234
 #define TWO_PI 6.2831853071795864769252866
 
 double what_time_is_it_now();
-int *read_intlist(char *s, int *n, int d);
-int *read_map(char *filename);
 void shuffle(void *arr, size_t n, size_t size);
 void sorta_shuffle(void *arr, size_t n, size_t size, size_t sections);
 void free_ptrs(void **ptrs, int n);
-char *basecfg(char *cfgfile);
 int alphanum_to_int(char c);
 char int_to_alphanum(int i);
 int read_int(int fd);
@@ -23,43 +20,28 @@ void write_all(int fd, char *buffer, size_t bytes);
 int read_all_fail(int fd, char *buffer, size_t bytes);
 int write_all_fail(int fd, char *buffer, size_t bytes);
 void find_replace(char *str, char *orig, char *rep, char *output);
-void error(const char *s);
 void malloc_error();
 void file_error(char *s);
 void strip(char *s);
 void strip_char(char *s, char bad);
-void top_k(float *a, int n, int k, int *index);
 list *split_str(char *s, char delim);
 char *fgetl(FILE *fp);
 list *parse_csv_line(char *line);
 char *copy_string(char *s);
 int count_fields(char *line);
 float *parse_fields(char *line, int n);
-void normalize_array(float *a, int n);
 void scale_array(float *a, int n, float s);
 void translate_array(float *a, int n, float s);
-int max_index(float *a, int n);
 float constrain(float min, float max, float a);
 int constrain_int(int a, int min, int max);
-float mse_array(float *a, int n);
-float rand_normal();
-size_t rand_size_t();
 float rand_uniform(float min, float max);
 float rand_scale(float s);
 int rand_int(int min, int max);
 float sum_array(float *a, int n);
-float mean_array(float *a, int n);
 void mean_arrays(float **a, int n, int els, float *avg);
-float variance_array(float *a, int n);
-float mag_array(float *a, int n);
 float dist_array(float *a, float *b, int n, int sub);
 float **one_hot_encode(float *a, int n, int k);
 float sec(clock_t clocks);
-int find_int_arg(int argc, char **argv, char *arg, int def);
-float find_float_arg(int argc, char **argv, char *arg, float def);
-int find_arg(int argc, char* argv[], char *arg);
-char *find_char_arg(int argc, char **argv, char *arg, char *def);
-int sample_array(float *a, int n);
 void print_statistics(float *a, int n);
 
 #endif