diff --git a/README.md b/README.md
index 9b7c7fb..c95208a 100644
--- a/README.md
+++ b/README.md
@@ -11,8 +11,9 @@ A minimal usage example would look like this:
 ```haskell
 main :: IO ()
 main = do
-  {- Creates a new network with two inputs, two hidden layers and one output -}
-  network <- newIO [2, 2, 1]
+  {- Creates a new network with two inputs,
+     two hidden layers and one output -}
+  network <- new [2, 2, 1]
 
   {- Train the network for a common logical AND,
      until the maximum error of 0.01 is reached -}
diff --git a/package.yaml b/package.yaml
index 7857cac..a633e83 100644
--- a/package.yaml
+++ b/package.yaml
@@ -1,6 +1,6 @@
 ---
 name: nn
-version: 0.1.0
+version: 0.2.0
 github: "saschagrunert/nn"
 license: MIT
 author: "Sascha Grunert"
diff --git a/src/AI/Nn.hs b/src/AI/Nn.hs
index c7b9e6d..2af2adb 100644
--- a/src/AI/Nn.hs
+++ b/src/AI/Nn.hs
@@ -6,7 +6,6 @@ module AI.Nn
   ( Network
   , predict
   , new
-  , newIO
   , train
   ) where
 
@@ -97,6 +96,12 @@ sigmoidLayer :: [Neuron'] -> Layer ()
 sigmoidLayer n = (biasNeuron x, ()) : createLayer n sigmoidNeuron
   where x = length $ head n
 
+-- | Create a new standard network for a number of layer and neurons
+--
+-- @since 0.1.0
+new :: [Int] -> IO Network
+new n = newGen n <$> getStdGen
+
 -- | Create a new output Layer from a list of Neuron'
 --
 -- @since 0.1.0
@@ -106,8 +111,8 @@ outputLayer n = createLayer n outputNeuron
 -- | Create a new network for a StdGen and a number of layer and neurons
 --
 -- @since 0.1.0
-new :: [Int] -> StdGen -> Network
-new n g = (sigmoidLayer <$> init wss) ++ [outputLayer (last wss)]
+newGen :: [Int] -> StdGen -> Network
+newGen n g = (sigmoidLayer <$> init wss) ++ [outputLayer (last wss)]
  where
   rest                 = init n
   hiddenIcsNcs         = zip ((+ 1) <$> rest) (tail rest)
@@ -123,12 +128,6 @@ new n g = (sigmoidLayer <$> init wss) ++ [outputLayer (last wss)]
   wss            = hidden ++ [outputWss]
   pack ic nc ws = (take nc $ chunksOf ic ws, drop (ic * nc) ws)
 
--- | Create a new standard network for a number of layer and neurons
---
--- @since 0.1.0
-newIO :: [Int] -> IO Network
-newIO n = new n <$> getStdGen
-
 -- | Do the complete back propagation
 --
 -- @since 0.1.0
@@ -199,7 +198,6 @@ updateNeuron (n, fpi) d = (n { inputWeights = ws' }, e)
   e   = activate' n (sumInputWeight fpi) * d
   ws' = zipWith (\x w -> w + (rate * e * x)) (inputs fpi) (inputWeights n)
 
-
 -- | Trains a network with a set of vector pairs until the global error is
 -- smaller than epsilon
 --
diff --git a/test/Spec.hs b/test/Spec.hs
index e87ae85..c76f75d 100644
--- a/test/Spec.hs
+++ b/test/Spec.hs
@@ -6,7 +6,7 @@ module Main
   ( main
   ) where
 
-import AI.Nn                 (newIO
+import AI.Nn                 (new
                              ,predict
                              ,train)
 import Test.Tasty            (TestTree
@@ -37,7 +37,7 @@ unitTests = do
 nnSpec :: Spec
 nnSpec = parallel $ do
   it "should succeed to train logical AND" $ do
-    n <- newIO [2, 2, 1]
+    n <- new [2, 2, 1]
     let
       nw = train 0.001
                  n
@@ -48,7 +48,7 @@ nnSpec = parallel $ do
     round (head $ predict nw [0, 0]) `shouldBe` (0 :: Int)
 
   it "should succeed to train logical OR" $ do
-    n <- newIO [2, 2, 1]
+    n <- new [2, 2, 1]
     let
       nw = train 0.001
                  n
@@ -59,7 +59,7 @@ nnSpec = parallel $ do
     round (head $ predict nw [0, 0]) `shouldBe` (0 :: Int)
 
   it "should succeed to train addition" $ do
-    n <- newIO [2, 2, 1]
+    n <- new [2, 2, 1]
     let
       nw = train 0.001
                  n