Optimazation is done

main.cpp

@@ -64,6 +64,7 @@
 #include <math/equal.hpp>
 #include <math/mean.hpp>
 #include <neural_network/activation_loss/activation_softmax_loss_categorical_crossentropy.hpp>
+#include <math/sqrt.hpp>
 
 #include <math.h>
 
@@ -709,8 +710,9 @@ int main(void) {
     panic::types::uint_t classes = 3;
 
 
-    // create spiral data
+    // create spiral data 
     panic::neural_network::spiral_data(samples, classes, X, y);
+    //panic::neural_network::vertical_data(samples, classes, X, y);
 
     // Initilise my model
     panic::neural_network::model mymodel;
@@ -741,10 +743,42 @@ int main(void) {
         return false;
     }
 
-    if (! mymodel.add_optimizer_sgd()){
+    /*
+    panic::types::real_t learning_rate = 1;
+    panic::types::real_t decay = 1e-3;
+    panic::types::real_t momentum = 0.9;
+    if (! mymodel.add_optimizer_sgd(learning_rate, decay, momentum)){
         return false;
     }
-   
+    */
+    /*
+    panic::types::real_t learning_rate = 1;
+    panic::types::real_t decay = 1e-4;
+    panic::types::real_t epsilon = 1e-7;
+    if (! mymodel.add_optimizer_adagrad(learning_rate, decay, epsilon)){
+        return false;
+    }
+    */
+    /*
+    panic::types::real_t learning_rate = 0.001;
+    panic::types::real_t decay = 1e-4;
+    panic::types::real_t epsilon = 1e-7;
+    panic::types::real_t rho = 0.999;
+    if (! mymodel.add_optimizer_rmsprop(learning_rate, decay, epsilon, rho)){
+        return false;
+    }
+    */
+
+    panic::types::real_t learning_rate = 0.02;
+    panic::types::real_t decay = 1e-5;
+    panic::types::real_t epsilon = 1e-7;
+    panic::types::real_t beta_1 = 0.9;
+    panic::types::real_t beta_2 = 0.999;
+    
+    if (! mymodel.add_optimizer_adam(learning_rate, decay, epsilon, beta_1, beta_2)){
+        return false;
+    }
+
     if (!mymodel.finalize()){
         return false;
     }
@@ -752,15 +786,13 @@ int main(void) {
     
     
     panic::types::uint_t epochs = 10000;
-    panic::types::uint_t print_every = 100;
+    panic::types::uint_t print_every = 250;
 
     if (!mymodel.train(X, y, epochs, print_every)){
         std::cout << "Training failed" << std::endl;
         return false;
     }
 
-    
-
 
 	return 0;
 }