Minor edit

bayesian_neural_networks_pymc4.ipynb

@@ -1843,7 +1843,7 @@
     "\n",
     "### Model definition\n",
     "\n",
-    "To model the non-linear relationship between `x` and `y` in the dataset we use a ReLU neural network with two hidden layers, 5 neurons each. The weights of the neural network are random variables instead of deterministic variables. This is what makes a neural network a Bayesian neural network. Here, we assume that the weights follow a normal distribution and are independent. \n",
+    "To model the non-linear relationship between `x` and `y` in the dataset we use a ReLU neural network with two hidden layers, 5 neurons each. The weights of the neural network are random variables instead of deterministic variables. This is what makes a neural network a Bayesian neural network. Here, we assume that the weights are independent random variables. \n",
     "\n",
     "The neural network defines a prior over the mean `loc` of the data likelihood `obs` which is represented by a normal distribution. For simplicity, the aleatoric uncertainty (`noise`) in the data is assumed to be known. Thanks to PyMC4's model composition support, priors can be defined layer-wise using the `layer` generator function and composed to a neural network as shown in function `model`. During inference, a posterior distribution over the neural network weights is obtained. "
    ]