Se agregó gráfica Training Loss y Pesos finales del modelo

README.md

@@ -347,6 +347,11 @@ MyANN = ann(inputNodes, hiddenNodes, outNodes)
 
 
 ```python
+# Model Training and Loss Tracking
+
+# Initialize a list to keep track of the loss for plotting
+epoch_losses = []
+
 # Iterative training loop
 epochs = 5
 for e in range(epochs):
@@ -355,8 +360,8 @@ for e in range(epochs):
         values = record.split(",")
        
         # Input data normalization
-        data = np.asarray(values[1:], dtype=int)/255*0.99+0.01
+        data = np.asarray(values[1:], dtype=float) / 255.0 * 0.99 + 0.01
-        index = np.asarray(values[0],dtype=int)
+        index = int(values[0])
        
         # Target Vector construction
         target = np.zeros(outNodes) + 0.01
@@ -371,8 +376,22 @@ for e in range(epochs):
         # Train
         MyANN.backpropagation(data, target, learningRate)
         
+    # Calculate and store the average loss for this epoch
     average_loss = total_loss / len(list)
+    epoch_losses.append(average_loss)
     print(f"Epoch {e+1}/{epochs} - Average Loss: {average_loss:.4f}")
+
+# --- Plotting the Learning Curve ---
+plt.figure(figsize=(8, 5))
+plt.plot(range(1, epochs + 1), epoch_losses, marker='o', color='red', linewidth=2)
+
+plt.title("Training Loss Convergence")
+plt.xlabel("Epoch")
+plt.ylabel("Average Loss (SSE)")
+plt.xticks(range(1, epochs + 1))
+plt.grid(True, linestyle='--', alpha=0.7)
+
+plt.show()
 ```
 
     Epoch 1/5 - Average Loss: 0.0972
@@ -382,6 +401,51 @@ for e in range(epochs):
     Epoch 5/5 - Average Loss: 0.0361
     
 
+
+    
+![png](README_files/README_28_1.png)
+    
+
+
+### Final Trained Weights Extraction
+
+After completing the training epochs, the optimal state of the network is fully captured within its weight matrices ($W_{ih}$ and $W_{ho}$). These resulting numerical arrays act as the permanent "memory" of the model. 
+
+
+```python
+# Display the final trained weight matrices
+# Displaying only a representative 3x3 slice to keep the notebook clean
+
+print("FINAL WEIGHTS: HIDDEN TO OUTPUT LAYER (Who)")
+print(f"Matrix Dimensions: {MyANN.who.shape}")
+print("-" * 50)
+print(MyANN.who[:3, :3])
+
+print("\n" + "="*50 + "\n")
+
+print("FINAL WEIGHTS: INPUT TO HIDDEN LAYER (Wih)")
+print(f"Matrix Dimensions: {MyANN.wih.shape}")
+print("-" * 50)
+print(MyANN.wih[:3, :3])
+```
+
+    FINAL WEIGHTS: HIDDEN TO OUTPUT LAYER (Who)
+    Matrix Dimensions: (10, 100)
+    --------------------------------------------------
+    [[-1.67078471  0.15194614 -1.7767362 ]
+     [ 1.62235502 -1.00391943 -0.37989541]
+     [-1.26930144  0.42418591  0.62579789]]
+    
+    ==================================================
+    
+    FINAL WEIGHTS: INPUT TO HIDDEN LAYER (Wih)
+    Matrix Dimensions: (100, 784)
+    --------------------------------------------------
+    [[ 0.46048182 -0.69392989  0.22993549]
+     [-0.94519842  0.42366932  1.41797905]
+     [ 0.36351595 -0.27497497  1.32172583]]
+    
+
 ## 5. Validation & Inference
 Evaluating model performance using unseen test data. A new sample is normalized and processed to extract the final prediction vector, which is then visually compared to the ground truth image.
 
@@ -467,7 +531,7 @@ plt.show()
 
 
     
-![png](README_files/README_33_0.png)
+![png](README_files/README_35_0.png)
     
 
 
@@ -558,7 +622,7 @@ plt.show()
 
 
     
-![png](README_files/README_35_1.png)
+![png](README_files/README_37_1.png)
     
 
 
@@ -645,7 +709,7 @@ plt.show()
 
 
     
-![png](README_files/README_37_1.png)
+![png](README_files/README_39_1.png)