fix(word2vec): simple example of word2vec

examples/word2vec.hs

@@ -27,13 +27,14 @@ module Main where
               "the dwarf poisons the king", "the dwarf poisons the queen"]
 
     let session = def { learningRate = 0.1
-                      , batchSize = 16
+                      , batchSize = 1
                       , epochs = 100
+                      , debug = True
                       } :: Session
         w2v = def { docs = ds
                   , dimensions = 50
                   , method = SkipGram
-                  , window = 3
+                  , window = 2
                   } :: Word2Vec
 
 

src/Sibe.hs

@@ -86,6 +86,7 @@ module Sibe
                               , batchSize    :: Int
                               , chart        :: [(Int, Double, Double)]
                               , momentum     :: Double
+                              , debug        :: Bool
                               } deriving (Show)
 
       emptyNetwork = randomNetwork 0 (0, 0) 0 [] (0, (id, id))
@@ -99,6 +100,7 @@ module Sibe
                       , batchSize    = 0
                       , chart        = []
                       , momentum     = 0
+                      , debug        = False
                       }
 
       saveNetwork :: Network -> String -> IO ()

src/Sibe/Word2Vec.hs

@@ -13,6 +13,7 @@ module Sibe.Word2Vec
     import qualified Data.Vector.Storable as V
     import Data.Default.Class
     import Data.Function (on)
+    import Control.Monad
 
     data W2VMethod = SkipGram | CBOW
     data Word2Vec = Word2Vec { docs :: [String]
@@ -30,11 +31,12 @@ module Sibe.Word2Vec
                       , network = randomNetwork 0 (-1, 1) v [(dimensions w2v, (id, one))] (v, (softmax, one))
                       }
 
-      putStr "vocabulary size: "
+      when (debug s) $ do
-      print v
+        putStr "vocabulary size: "
+        print v
 
-      putStr "trainingData length: "
+        putStr "trainingData length: "
-      print . length $ trainingData
+        print . length $ trainingData
 
       -- biases are not used in skipgram/cbow
       newses <- run (sgd . ignoreBiases) s
@@ -73,7 +75,7 @@ module Sibe.Word2Vec
                     | i == length vocvec - 1 = before
                     | otherwise = before ++ after
                   vectorized = map (\w -> snd . fromJust $ find ((== w) . fst) vocvec) ns
-                  new = foldl1 (+) vectorized
+                  new = cmap (max 1) $ foldl1 (+) vectorized
               in
                 case method w2v of
                   SkipGram -> zip (repeat v) vectorized