Remove Show constraint on UpdateLayer

mafia

@@ -120,4 +120,4 @@ case "$MODE" in
 upgrade) shift; run_upgrade "$@" ;;
 *) exec_mafia "$@"
 esac
-# Version: 3044e63eb472fb9e16926d4ab2ca9dd9e255829c
+# Version: 360716306a06db842ec022a7b9f161d2208483f0

main/mnist.hs

@@ -33,8 +33,8 @@ import           Grenade.Utils.OneHot
 
 -- With the mnist data from Kaggle normalised to doubles between 0 and 1, learning rate of 0.01 and 15 iterations,
 -- this network should get down to about a 1.3% error rate.
-type MNIST = Network '[ Convolution 1 10 5 5 1 1, Pooling 2 2 2 2, Relu, Convolution 10 16 5 5 1 1, Pooling 2 2 2 2, FlattenLayer, Relu, FullyConnected 256 80, Logit, FullyConnected 80 10, Logit]
-                     '[ 'D2 28 28, 'D3 24 24 10, 'D3 12 12 10, 'D3 12 12 10, 'D3 8 8 16, 'D3 4 4 16, 'D1 256, 'D1 256, 'D1 80, 'D1 80, 'D1 10, 'D1 10]
+type MNIST = Network '[ Convolution 1 10 5 5 1 1, Pooling 2 2 2 2, Relu, Convolution 10 16 5 5 1 1, Pooling 2 2 2 2, Relu, FlattenLayer, FullyConnected 256 80, Logit, FullyConnected 80 10, Logit]
+                     '[ 'D2 28 28, 'D3 24 24 10, 'D3 12 12 10, 'D3 12 12 10, 'D3 8 8 16, 'D3 4 4 16, 'D3 4 4 16, 'D1 256, 'D1 80, 'D1 80, 'D1 10, 'D1 10]
 
 randomMnist :: MonadRandom m => m MNIST
 randomMnist = randomNetwork

src/Grenade.hs

@@ -13,6 +13,7 @@ import           Grenade.Layers.Flatten as X
 import           Grenade.Layers.Fuse as X
 import           Grenade.Layers.FullyConnected as X
 import           Grenade.Layers.Logit as X
-import           Grenade.Layers.Convolution   as X
-import           Grenade.Layers.Relu   as X
-import           Grenade.Layers.Tanh   as X
+import           Grenade.Layers.Convolution as X
+import           Grenade.Layers.Relu as X
+import           Grenade.Layers.Tanh as X
+import           Grenade.Layers.Softmax as X

src/Grenade/Core/Network.hs

@@ -42,7 +42,7 @@ data LearningParameters = LearningParameters {
 
 -- | Class for updating a layer. All layers implement this, and it is
 --   shape independent.
-class Show x => UpdateLayer x where
+class UpdateLayer x where
   {-# MINIMAL runUpdate, createRandom #-}
   -- | The type for the gradient for this layer.
   --   Unit if there isn't a gradient to pass back.
@@ -86,9 +86,11 @@ data Network :: [*] -> [Shape] -> * where
     (:~>) :: (SingI i, SingI h, Layer x i h) => !x -> !(Network xs (h ': hs)) -> Network (x ': xs) (i ': h ': hs)
 infixr 5 :~>
 
-instance Show (Network l h) where
+instance Show (Network '[] '[i]) where
   show NNil = "NNil"
-  show (i :~> o) = show i ++ "\n:~>\n" ++ show o
+instance (Show x, Show (Network xs rs)) => Show (Network (x ': xs) (i ': rs)) where
+  show (x :~> xs) = show x ++ "\n~>\n" ++ show xs
+
 
 -- | Gradients of a network.
 --   Parameterised on the layers of a Network.

src/Grenade/Core/Runner.hs

@@ -14,6 +14,9 @@ Stability   : experimental
 This module defines simple back propagation and training functions
 for a network.
 -}
+
+-- GHC 7.10 doesn't think that go is complete
+{-# OPTIONS_GHC -fno-warn-incomplete-patterns #-}
 module Grenade.Core.Runner (
     train
   , backPropagate
@@ -48,7 +51,6 @@ backPropagate network input target =
               (n', dWs')    = go y n
               -- calculate the gradient for this layer to pass down,
               (layer', dWs) = runBackwards layer tape dWs'
-
           in (layer' :/> n', dWs)
 
     -- Bouncing the derivatives back down.

src/Grenade/Layers/Flatten.hs

@@ -31,7 +31,7 @@ instance UpdateLayer FlattenLayer where
   runUpdate _ _ _ = FlattenLayer
   createRandom = return FlattenLayer
 
-instance (KnownNat a, KnownNat x, KnownNat y, a ~ (x * z)) => Layer FlattenLayer ('D2 x y) ('D1 a) where
+instance (KnownNat a, KnownNat x, KnownNat y, a ~ (x * y)) => Layer FlattenLayer ('D2 x y) ('D1 a) where
   type Tape FlattenLayer ('D2 x y) ('D1 a) = ()
   runForwards _ (S2D y)   =  ((), fromJust' . fromStorable . flatten . extract $ y)
   runBackwards _ _ (S1D y) = ((), fromJust' . fromStorable . extract $ y)

src/Grenade/Recurrent/Core/Network.hs

@@ -59,11 +59,12 @@ data RecurrentNetwork :: [*] -> [Shape] -> * where
 infixr 5 :~~>
 infixr 5 :~@>
 
-instance Show (RecurrentNetwork l h) where
-  show RNil       = "RNil"
-  show (i :~~> o) = show i ++ "\n:~~>\n" ++ show o
-  show (i :~@> o) = show i ++ "\n:~@>\n" ++ show o
-
+instance Show (RecurrentNetwork '[] '[i]) where
+  show RNil = "NNil"
+instance (Show x, Show (RecurrentNetwork xs rs)) => Show (RecurrentNetwork (FeedForward x ': xs) (i ': rs)) where
+  show (x :~~> xs) = show x ++ "\n~~>\n" ++ show xs
+instance (Show x, Show (RecurrentNetwork xs rs)) => Show (RecurrentNetwork (Recurrent x ': xs) (i ': rs)) where
+  show (x :~@> xs) = show x ++ "\n~~>\n" ++ show xs
 
 -- | Recurrent inputs (sideways shapes on an imaginary unrolled graph)
 --   Parameterised on the layers of a Network.

src/Grenade/Recurrent/Core/Runner.hs

@@ -7,6 +7,9 @@
 {-# LANGUAGE FlexibleContexts      #-}
 {-# LANGUAGE RankNTypes            #-}
 {-# LANGUAGE RecordWildCards       #-}
+
+-- GHC 7.10 doesn't think that go is complete
+{-# OPTIONS_GHC -fno-warn-incomplete-patterns #-}
 module Grenade.Recurrent.Core.Runner (
     trainRecurrent
   , runRecurrent