Added comments for better readability

src/finn/core/multi_thresholding.py

@@ -8,26 +8,35 @@ def compare(value, threshold):
                 res = 0.0
         return res
 
-def execute(inputs,thresholds):
-
-        inputs_reshaped = inputs.reshape((thresholds.shape[1], -1))
-        num_channels = thresholds.shape[0]
-        ret = np.zeros_like(inputs_reshaped)
+def execute(v,thresholds):
+	#reshape inputs to enable channel-wise reading
+        vr = inputs.reshape((thresholds.shape[1], -1))
+        
+	#calculate the channel interval for the for loops
+	num_channels = thresholds.shape[0]
         channel_interval = int(inputs_reshaped.shape[1]/num_channels)
-
-        i=-1
+ 	
+	#iniate output tensor 
+	ret = np.zeros_like(vr)
+        
+	#initiate helper variable i for channel-wise thresholding
+	i = -1
+
+	#iterate over thresholds channel-wise
         for t in thresholds:
-                i+=1
+                i += 1
+		#calculate the lower and upper limit in which elements belong to one channel 
                 if i == 0:
-                        ce1_low_lim=0
+                        ce1_low_lim = 0
                 else:
-                        ce1_low_lim=i*channel_interval
-                ce1_up_lim=(i+1)*channel_interval
+                        ce1_low_lim = i*channel_interval
+                ce1_up_lim = (i+1)*channel_interval
 
+		#iterate in ascending order over the thresholds belonging to one channel
                 for c in range(thresholds.shape[1]):
                         for ce0 in range(inputs_reshaped.shape[0]):
                                 for ce1 in range(ce1_low_lim,ce1_up_lim):
-                                        ret[ce0][ce1] += compare(inputs_reshaped[ce0][ce1], t[c])
+                                        ret[ce0][ce1] += compare(vr[ce0][ce1], t[c])
 
 
         return ret.reshape(inputs.shape)