Update app.py

app.py

@@ -56,17 +56,17 @@ lr = float(st.text_input('Learning rate', value=0.5))
 
 prob_A = float(st.text_input('Click probability of ad A', value=0.3))
 
-prob_B = float(st.text_input('Click probability of ad B', value=0.54
+prob_B = float(st.text_input('Click probability of ad B', value=0.4))
 
-epochs = int(st.text_input('Number of ad impressions (epochs)', value=1000))
+steps = int(st.text_input('Number of ad impressions (steps)', value=1000))
 
-information_for_plotting = np.zeros((epochs, 10))
+information_for_plotting = np.zeros((steps, 10))
 
 if st.button('Run the ad campaign and display the results'):
 
     with st.spinner('Simulating the ad campaign may take a few seconds ...'):
 
-        for epoch in range(epochs):
+        for step in range(steps):
 
             # The neural network is used to choose the action
             # To display the learning progress, we also record the 
@@ -98,16 +98,16 @@ if st.button('Run the ad campaign and display the results'):
             model.trainable_variables[1].assign(
                 model.trainable_variables[1]-lr*grads_adjusted[1])
 
-            information_for_plotting[epoch, 0] = output.numpy()[0]
-            information_for_plotting[epoch, 1] = action.numpy()[0].astype(int)
-            information_for_plotting[epoch, 2] = loss
-            information_for_plotting[epoch, 3] = grads[0]
-            information_for_plotting[epoch, 4] = grads[1]
-            information_for_plotting[epoch, 5] = reward
-            information_for_plotting[epoch, 6] = grads_adjusted[0]
-            information_for_plotting[epoch, 7] = grads_adjusted[1]
-            information_for_plotting[epoch, 8] = copy.deepcopy(model.trainable_variables[0])
-            information_for_plotting[epoch, 9] = copy.deepcopy(model.trainable_variables[1])
+            information_for_plotting[step, 0] = output.numpy()[0]
+            information_for_plotting[step, 1] = action.numpy()[0].astype(int)
+            information_for_plotting[step, 2] = loss
+            information_for_plotting[step, 3] = grads[0]
+            information_for_plotting[step, 4] = grads[1]
+            information_for_plotting[step, 5] = reward
+            information_for_plotting[step, 6] = grads_adjusted[0]
+            information_for_plotting[step, 7] = grads_adjusted[1]
+            information_for_plotting[step, 8] = copy.deepcopy(model.trainable_variables[0])
+            information_for_plotting[step, 9] = copy.deepcopy(model.trainable_variables[1])
             
         # Plot the results
         titles = ['Model Output', 'Action', 'Loss', 'Gradients', 'Rewards',
@@ -132,7 +132,7 @@ if st.button('Run the ad campaign and display the results'):
                 plt.gca().yaxis.set_major_formatter(plt.FormatStrFormatter('%.2f'))
             plt.ylabel(titles[i])
 
-        plt.xlabel('Epoch')
+        plt.xlabel('Step')
         plt.show()
 
         # Sum of the total clicks obtained