Added NFData instances for networks and layers

src/Grenade/Core/Layer.hs

@@ -48,6 +48,7 @@ import           Data.Kind (Type)
 
 import           Grenade.Core.Shape
 import           Grenade.Core.LearningParameters
+import           Control.DeepSeq (NFData)
 
 -- | Class for updating a layer. All layers implement this, as it
 --   describes how to create and update the layer.
@@ -73,7 +74,7 @@ class UpdateLayer x where
 --   need to implement it for all shapes, only ones which are
 --   appropriate.
 --
-class UpdateLayer x => Layer x (i :: Shape) (o :: Shape) where
+class (UpdateLayer x, NFData x) => Layer x (i :: Shape) (o :: Shape) where
   -- | The Wengert tape for this layer. Includes all that is required
   --   to generate the back propagated gradients efficiently. As a
   --   default, `S i` is fine.

src/Grenade/Core/Network.hs

@@ -44,6 +44,7 @@ import           Data.Kind (Type)
 import           Grenade.Core.Layer
 import           Grenade.Core.LearningParameters
 import           Grenade.Core.Shape
+import           Control.DeepSeq
 
 -- | Type of a network.
 --
@@ -68,6 +69,12 @@ instance Show (Network '[] '[i]) where
 instance (Show x, Show (Network xs rs)) => Show (Network (x ': xs) (i ': rs)) where
   show (x :~> xs) = show x ++ "\n~>\n" ++ show xs
 
+instance NFData (Network '[] '[i]) where
+  rnf NNil = ()
+
+instance NFData (Network xs rs) => NFData (Network (x ': xs) (i ': rs)) where
+  rnf (x :~> xs) = x `deepseq` xs `deepseq` ()
+
 -- | Gradient of a network.
 --
 --   Parameterised on the layers of the network.
@@ -191,7 +198,7 @@ instance CreatableNetwork sublayers subshapes => UpdateLayer (Network sublayers
 -- | Ultimate composition.
 --
 --   This allows a complete network to be treated as a layer in a larger network.
-instance (CreatableNetwork sublayers subshapes, i ~ (Head subshapes), o ~ (Last subshapes)) => Layer (Network sublayers subshapes) i o where
+instance (CreatableNetwork sublayers subshapes, i ~ (Head subshapes), o ~ (Last subshapes), NFData (Network sublayers subshapes)) => Layer (Network sublayers subshapes) i o where
   type Tape (Network sublayers subshapes) i o = Tapes sublayers subshapes
   runForwards  = runNetwork
   runBackwards = runGradient

src/Grenade/Layers/Concat.hs

@@ -35,6 +35,7 @@ import           Data.Kind (Type)
 import           Grenade.Core
 
 import           Numeric.LinearAlgebra.Static ( row, (===), splitRows, unrow, (#), split, R )
+import           Control.DeepSeq
 
 -- | A Concatentating Layer.
 --
@@ -52,6 +53,9 @@ import           Numeric.LinearAlgebra.Static ( row, (===), splitRows, unrow, (#
 data Concat :: Shape -> Type -> Shape -> Type -> Type where
   Concat :: x -> y -> Concat m x n y
 
+instance (NFData x, NFData y) => NFData (Concat m x n y) where
+  rnf (Concat x y) = rnf x `deepseq` rnf y `deepseq` ()
+
 instance (Show x, Show y) => Show (Concat m x n y) where
   show (Concat x y) = "Concat\n" ++ show x ++ "\n" ++ show y
 

src/Grenade/Layers/Convolution.hs

@@ -46,6 +46,7 @@ import           Numeric.LinearAlgebra.Static hiding ((|||), build, toRows)
 import           Grenade.Core
 import           Grenade.Layers.Internal.Convolution
 import           Grenade.Layers.Internal.Update
+import           Control.DeepSeq
 
 -- | A convolution layer for a neural network.
 --   This uses the im2col convolution trick popularised by Caffe, which essentially turns the
@@ -113,6 +114,9 @@ instance Show (Convolution c f k k' s s') where
             px = (fmap . fmap . fmap) render ms
         in unlines $ foldl1 (zipWith (\a' b' -> a' ++ "   |   " ++ b')) $ px
 
+instance NFData (Convolution c f k k' s s') where
+  rnf (Convolution a b) = rnf a `deepseq` rnf b `deepseq` ()
+
 randomConvolution :: ( MonadRandom m
                      , KnownNat channels
                      , KnownNat filters

src/Grenade/Layers/Crop.hs

@@ -37,6 +37,7 @@ import           Grenade.Layers.Internal.Pad
 
 import           Numeric.LinearAlgebra (konst, subMatrix, diagBlock)
 import           Numeric.LinearAlgebra.Static (extract, create)
+import           Control.DeepSeq
 
 -- | A cropping layer for a neural network.
 data Crop :: Nat
@@ -45,6 +46,9 @@ data Crop :: Nat
           -> Nat -> Type where
   Crop :: Crop cropLeft cropTop cropRight cropBottom
 
+instance NFData (Crop cropLeft cropTop cropRight cropBottom) where
+  rnf Crop = ()
+
 instance Show (Crop cropLeft cropTop cropRight cropBottom) where
   show Crop = "Crop"
 

src/Grenade/Layers/Deconvolution.hs

@@ -50,6 +50,7 @@ import           Numeric.LinearAlgebra.Static hiding ((|||), build, toRows)
 import           Grenade.Core
 import           Grenade.Layers.Internal.Convolution
 import           Grenade.Layers.Internal.Update
+import           Control.DeepSeq
 
 -- | A Deconvolution layer for a neural network.
 --   This uses the im2col Convolution trick popularised by Caffe.
@@ -94,6 +95,9 @@ data Deconvolution' :: Nat -- Number of channels, for the first layer this could
                => !(L kernelFlattened channels) -- The kernel filter gradient
                -> Deconvolution' channels filters kernelRows kernelColumns strideRows strideColumns
 
+instance NFData (Deconvolution c f k k' s s') where
+  rnf (Deconvolution a b) = rnf a `deepseq` rnf b `deepseq` ()
+
 instance Show (Deconvolution c f k k' s s') where
   show (Deconvolution a _) = renderConv a
     where

src/Grenade/Layers/Dropout.hs

@@ -11,6 +11,7 @@ import           Control.Monad.Random hiding (fromList)
 
 import           GHC.TypeLits
 import           Grenade.Core
+import           Control.DeepSeq
 
 -- Dropout layer help to reduce overfitting.
 -- Idea here is that the vector is a shape of 1s and 0s, which we multiply the input by.
@@ -20,7 +21,10 @@ import           Grenade.Core
 data Dropout = Dropout {
     dropoutRate :: Double
   , dropoutSeed :: Int
-  } deriving Show
+  } deriving (Show)
+
+instance NFData Dropout where
+  rnf (Dropout r s) = rnf r `deepseq` rnf s `deepseq` ()
 
 instance UpdateLayer Dropout where
   type Gradient Dropout = ()

src/Grenade/Layers/Elu.hs

@@ -19,13 +19,17 @@ import           GHC.TypeLits
 import           Grenade.Core
 
 import qualified Numeric.LinearAlgebra.Static as LAS
+import           Control.DeepSeq
 
 -- | An exponential linear unit.
 --   A layer which can act between any shape of the same dimension, acting as a
 --   diode on every neuron individually.
 data Elu = Elu
   deriving Show
 
+instance NFData Elu where
+  rnf Elu = ()
+
 instance UpdateLayer Elu where
   type Gradient Elu = ()
   runUpdate _ _ _ = Elu

src/Grenade/Layers/FullyConnected.hs

@@ -21,6 +21,7 @@ import           Numeric.LinearAlgebra.Static
 import           Grenade.Core
 
 import           Grenade.Layers.Internal.Update
+import           Control.DeepSeq
 
 -- | A basic fully connected (or inner product) neural network layer.
 data FullyConnected i o = FullyConnected
@@ -31,6 +32,12 @@ data FullyConnected' i o = FullyConnected'
                          !(R o)   -- Bias
                          !(L o i) -- Activations
 
+instance NFData (FullyConnected' i o) where
+  rnf (FullyConnected' a b) = a `deepseq` b `deepseq` ()
+
+instance NFData (FullyConnected i o) where
+  rnf (FullyConnected a b) = a `deepseq` b `deepseq` ()
+
 instance Show (FullyConnected i o) where
   show FullyConnected {} = "FullyConnected"
 

src/Grenade/Layers/Logit.hs

@@ -19,6 +19,7 @@ import           Data.Serialize
 import           Data.Singletons
 
 import           Grenade.Core
+import           Control.DeepSeq
 
 -- | A Logit layer.
 --
@@ -28,6 +29,9 @@ import           Grenade.Core
 data Logit = Logit
   deriving Show
 
+instance NFData Logit where
+  rnf Logit = ()
+
 instance UpdateLayer Logit where
   type Gradient Logit = ()
   runUpdate _ _ _ = Logit

src/Grenade/Layers/Merge.hs

@@ -29,6 +29,7 @@ import           Data.Kind (Type)
 #endif
 
 import           Grenade.Core
+import           Control.DeepSeq 
 
 -- | A Merging layer.
 --
@@ -37,6 +38,9 @@ import           Grenade.Core
 data Merge :: Type -> Type -> Type where
   Merge :: x -> y -> Merge x y
 
+instance (NFData x, NFData y) => NFData (Merge x y) where
+  rnf (Merge x y) = rnf x `deepseq` rnf y `deepseq` ()
+
 instance (Show x, Show y) => Show (Merge x y) where
   show (Merge x y) = "Merge\n" ++ show x ++ "\n" ++ show y
 

src/Grenade/Layers/Pad.hs

@@ -37,6 +37,7 @@ import           Grenade.Layers.Internal.Pad
 
 import           Numeric.LinearAlgebra (konst, subMatrix, diagBlock)
 import           Numeric.LinearAlgebra.Static (extract, create)
+import           Control.DeepSeq
 
 -- | A padding layer for a neural network.
 --
@@ -47,6 +48,9 @@ data Pad  :: Nat
           -> Nat -> Type where
   Pad  :: Pad padLeft padTop padRight padBottom
 
+instance NFData (Pad padLeft padTop padRight padBottom) where
+  rnf Pad = ()
+
 instance Show (Pad padLeft padTop padRight padBottom) where
   show Pad = "Pad"
 

src/Grenade/Layers/Pooling.hs

@@ -37,6 +37,7 @@ import           Grenade.Core
 import           Grenade.Layers.Internal.Pooling
 
 import           Numeric.LinearAlgebra.Static as LAS hiding ((|||), build, toRows)
+import           Control.DeepSeq
 
 -- | A pooling layer for a neural network.
 --
@@ -49,6 +50,9 @@ import           Numeric.LinearAlgebra.Static as LAS hiding ((|||), build, toRow
 data Pooling :: Nat -> Nat -> Nat -> Nat -> Type where
   Pooling :: Pooling kernelRows kernelColumns strideRows strideColumns
 
+instance NFData (Pooling k k' s s') where
+  rnf Pooling = ()
+
 instance Show (Pooling k k' s s') where
   show Pooling = "Pooling"
 

src/Grenade/Layers/Relu.hs

@@ -19,13 +19,17 @@ import           GHC.TypeLits
 import           Grenade.Core
 
 import qualified Numeric.LinearAlgebra.Static as LAS
+import           Control.DeepSeq
 
 -- | A rectifying linear unit.
 --   A layer which can act between any shape of the same dimension, acting as a
 --   diode on every neuron individually.
 data Relu = Relu
   deriving Show
 
+instance NFData Relu where
+  rnf Relu = ()
+
 instance UpdateLayer Relu where
   type Gradient Relu = ()
   runUpdate _ _ _ = Relu

src/Grenade/Layers/Reshape.hs

@@ -24,6 +24,7 @@ import           Numeric.LinearAlgebra.Static
 import           Numeric.LinearAlgebra.Data as LA ( flatten )
 
 import           Grenade.Core
+import           Control.DeepSeq
 
 -- | Reshape Layer
 --
@@ -36,6 +37,9 @@ import           Grenade.Core
 data Reshape = Reshape
   deriving Show
 
+instance NFData Reshape where
+  rnf Reshape = ()
+
 instance UpdateLayer Reshape where
   type Gradient Reshape = ()
   runUpdate _ _ _ = Reshape

src/Grenade/Layers/Sinusoid.hs

@@ -18,12 +18,16 @@ import           Data.Serialize
 import           Data.Singletons
 
 import           Grenade.Core
+import           Control.DeepSeq
 
 -- | A Sinusoid layer.
 --   A layer which can act between any shape of the same dimension, performing a sin function.
 data Sinusoid = Sinusoid
   deriving Show
 
+instance NFData Sinusoid where
+  rnf Sinusoid = ()
+
 instance UpdateLayer Sinusoid where
   type Gradient Sinusoid  = ()
   runUpdate _ _ _ = Sinusoid

src/Grenade/Layers/Softmax.hs

@@ -22,6 +22,7 @@ import           GHC.TypeLits
 import           Grenade.Core
 
 import           Numeric.LinearAlgebra.Static as LAS
+import           Control.DeepSeq
 
 -- | A Softmax layer
 --
@@ -33,6 +34,9 @@ import           Numeric.LinearAlgebra.Static as LAS
 data Softmax = Softmax
   deriving Show
 
+instance NFData Softmax where
+  rnf Softmax = ()
+
 instance UpdateLayer Softmax where
   type Gradient Softmax = ()
   runUpdate _ _ _ = Softmax

src/Grenade/Layers/Tanh.hs

@@ -18,12 +18,16 @@ import           Data.Serialize
 import           Data.Singletons
 
 import           Grenade.Core
+import           Control.DeepSeq
 
 -- | A Tanh layer.
 --   A layer which can act between any shape of the same dimension, performing a tanh function.
 data Tanh = Tanh
   deriving Show
 
+instance NFData Tanh where
+  rnf Tanh = ()
+
 instance UpdateLayer Tanh where
   type Gradient Tanh = ()
   runUpdate _ _ _ = Tanh

src/Grenade/Layers/Trivial.hs

@@ -17,6 +17,7 @@ module Grenade.Layers.Trivial (
 import           Data.Serialize
 
 import           Grenade.Core
+import           Control.DeepSeq
 
 -- | A Trivial layer.
 --
@@ -25,6 +26,9 @@ import           Grenade.Core
 data Trivial = Trivial
   deriving Show
 
+instance NFData Trivial where
+  rnf Trivial = ()
+
 instance Serialize Trivial where
   put _ = return ()
   get = return Trivial

src/Grenade/Recurrent/Core/Layer.hs

@@ -14,6 +14,7 @@ import           Data.Kind (Type)
 #endif
 
 import           Grenade.Core
+import           Control.DeepSeq (NFData)
 
 -- | Class for a recurrent layer.
 --   It's quite similar to a normal layer but for the input and output
@@ -22,7 +23,7 @@ class UpdateLayer x => RecurrentUpdateLayer x where
   -- | Shape of data that is passed between each subsequent run of the layer
   type RecurrentShape x   :: Type
 
-class (RecurrentUpdateLayer x, Num (RecurrentShape x)) => RecurrentLayer x (i :: Shape) (o :: Shape) where
+class (RecurrentUpdateLayer x, Num (RecurrentShape x), NFData x) => RecurrentLayer x (i :: Shape) (o :: Shape) where
   -- | Wengert Tape
   type RecTape x i o :: Type
   -- | Used in training and scoring. Take the input from the previous