Stable, needs to be way faster

src/cnn.c

@@ -105,7 +105,7 @@ void train_detection_net(char *cfgfile)
         time=clock();
         float loss = train_network(net, train);
         avg_loss = avg_loss*.9 + loss*.1;
-        printf("%d: %f, %f avg, %lf seconds, %d images\n", i, loss, avg_loss, sec(clock()-time), i*imgs*net.batch);
+        printf("%d: %f, %f avg, %lf seconds, %d images\n", i, loss, avg_loss, sec(clock()-time), i*imgs);
         if(i%100==0){
             char buff[256];
             sprintf(buff, "/home/pjreddie/imagenet_backup/detnet_%d.cfg", i);
@@ -213,7 +213,7 @@ void train_imagenet(char *cfgfile)
     set_learning_network(&net, net.learning_rate, 0, net.decay);
     printf("Learning Rate: %g, Momentum: %g, Decay: %g\n", net.learning_rate, net.momentum, net.decay);
     int imgs = 1024;
-    int i = 77700;
+    int i = 0;
     char **labels = get_labels("/home/pjreddie/data/imagenet/cls.labels.list");
     list *plist = get_paths("/data/imagenet/cls.train.list");
     char **paths = (char **)list_to_array(plist);
@@ -240,7 +240,7 @@ void train_imagenet(char *cfgfile)
         free_data(train);
         if(i%100==0){
             char buff[256];
-            sprintf(buff, "/home/pjreddie/imagenet_backup/net_%d.cfg", i);
+            sprintf(buff, "/home/pjreddie/imagenet_backup/alexnet_%d.cfg", i);
             save_network(net, buff);
         }
     }

src/network.c

@@ -15,6 +15,35 @@
 #include "softmax_layer.h"
 #include "dropout_layer.h"
 
+char *get_layer_string(LAYER_TYPE a)
+{
+    switch(a){
+        case CONVOLUTIONAL:
+            return "convolutional";
+        case CONNECTED:
+            return "connected";
+        case MAXPOOL:
+            return "maxpool";
+        case SOFTMAX:
+            return "softmax";
+        case NORMALIZATION:
+            return "normalization";
+        case DROPOUT:
+            return "dropout";
+        case FREEWEIGHT:
+            return "freeweight";
+        case CROP:
+            return "crop";
+        case COST:
+            return "cost";
+        default:
+            break;
+    }
+    return "none";
+}
+
+
+
 network make_network(int n, int batch)
 {
     network net;

src/network.h

@@ -41,6 +41,7 @@ float *network_predict_gpu(network net, float *input);
 #endif
 
 void compare_networks(network n1, network n2, data d);
+char *get_layer_string(LAYER_TYPE a);
 
 network make_network(int n, int batch);
 void forward_network(network net, float *input, float *truth, int train);

src/network_gpu.c

@@ -24,7 +24,7 @@ void forward_network_gpu(network net, cl_mem input, cl_mem truth, int train)
 {
     int i;
     for(i = 0; i < net.n; ++i){
-        clock_t time = clock();
+        //clock_t time = clock();
         if(net.types[i] == CONVOLUTIONAL){
             convolutional_layer layer = *(convolutional_layer *)net.layers[i];
             forward_convolutional_layer_gpu(layer, input);
@@ -61,7 +61,7 @@ void forward_network_gpu(network net, cl_mem input, cl_mem truth, int train)
             input = layer.output_cl;
         }
         check_error(cl);
-        //printf("Forw %d %f\n", i, sec(clock() - time));
+        //printf("Forward %d %s %f\n", i, get_layer_string(net.types[i]), sec(clock() - time));
     }
 }
 
@@ -71,7 +71,7 @@ void backward_network_gpu(network net, cl_mem input)
     cl_mem prev_input;
     cl_mem prev_delta;
     for(i = net.n-1; i >= 0; --i){
-        clock_t time = clock();
+        //clock_t time = clock();
         if(i == 0){
             prev_input = input;
             prev_delta = 0;
@@ -104,7 +104,7 @@ void backward_network_gpu(network net, cl_mem input)
             backward_softmax_layer_gpu(layer, prev_delta);
         }
         check_error(cl);
-        //printf("Back %d %f\n", i, sec(clock() - time));
+        //printf("Backward %d %s %f\n", i, get_layer_string(net.types[i]), sec(clock() - time));
     }
 }