diff --git a/modules.py b/modules.py
index ab8ef87..2ff2385 100644
--- a/modules.py
+++ b/modules.py
@@ -143,13 +143,15 @@ def positional_encoding(inputs,
     with tf.variable_scope(scope, reuse = reuse):
 
         input_one = tf.tile(tf.expand_dims(tf.range(tf.shape(inputs)[1]), 0), [tf.shape(inputs)[0], 1])
-        position_block = tf.tile(tf.expand_dims(tf.range(vocab_size), 1), [1, num_units // 2])
-        unit_block = tf.tile(tf.expand_dims(tf.range(num_units // 2), 0), [vocab_size, 1])
-        rad_block = tf.pow(tf.div(position_block, tf.multiply(10000, 1)), tf.div(unit_block, num_units // 2))
-        
-        sin_block = tf.sin(tf.cast(rad_block, tf.float32))
-        cos_block = tf.cos(tf.cast(rad_block, tf.float32))
-        lookup_table = tf.concat([sin_block, cos_block], axis = 1)
+        # First part of the PE function: sin and cos argument
+        position_enc = np.array([
+            [pos / np.power(10000, 2*i/num_units) for i in range(num_units)]
+            for pos in range(max_len)])
+        # Second part, apply the cosine to even columns and sin to odds.
+        position_enc[:, 0::2] = np.sin(position_enc[1:, 0::2]) # dim 2i
+        position_enc[:, 1::2] = np.cos(position_enc[1:, 1::2]) # dim 2i+1
+        # Convert to a tensor
+        lookup_table = tf.convert_to_tensor(position_enc)
 
         if zero_pad:
 
@@ -328,4 +330,4 @@ def label_smoothing(inputs, epsilon=0.1):
     
     
 
-            
\ No newline at end of file
+