diff --git a/session-2/session-2.ipynb b/session-2/session-2.ipynb
index 72933a12..ec5e1818 100644
--- a/session-2/session-2.ipynb
+++ b/session-2/session-2.ipynb
@@ -288,7 +288,7 @@
    "source": [
     "Now multiply the tensor using a new variable, $\\textbf{W}$, which has 2 rows and 20 columns, so that when it is left mutiplied by $\\textbf{X}$, the output of the multiplication is None x 20, giving you 20 output neurons.  Recall that the `tf.matmul` function takes two arguments, the left hand ($\\textbf{X}$) and right hand side ($\\textbf{W}$) of a matrix multiplication.\n",
     "\n",
-    "To create $\\textbf{W}$, you will use `tf.get_variable` to create a matrix which is `2 x 20` in dimension.  Look up the docstrings of functions `tf.get_variable` and `tf.random_normal_initializer` to get familiar with these functions.  There are many options we will ignore for now.  Just be sure to set the `name`, `shape` (this is the one that has to be [2, 20]), `dtype` (i.e. tf.float32), and `initializer` (the `tf.random_normal_intializer` you should create) when creating your $\\textbf{W}$ variable with `tf.get_variable(...)`.\n",
+    "To create $\\textbf{W}$, you will use `tf.get_variable` to create a matrix which is `2 x 20` in dimension.  Look up the docstrings of functions `tf.get_variable` and `tf.random_normal_initializer` to get familiar with these functions.  There are many options we will ignore for now.  Just be sure to set the `name`, `shape` (this is the one that has to be [2, 20]), `dtype` (i.e. tf.float32), and `initializer` (the `tf.random_normal_initializer` you should create) when creating your $\\textbf{W}$ variable with `tf.get_variable(...)`.\n",
     "\n",
     "For the random normal initializer, often the mean is set to 0, and the standard deviation is set based on the number of neurons.  But that really depends on the input and outputs of your network, how you've \"normalized\" your dataset, what your nonlinearity/activation function is, and what your expected range of inputs/outputs are.  Don't worry about the values for the initializer for now, as this part will take a bit more experimentation to understand better!\n",
     "\n",