diff --git a/Makefile b/Makefile
--- a/Makefile
+++ b/Makefile
@@ -1,8 +1,7 @@
 INCLUDES=lib
 
 cabal:
-	runghc Setup.lhs configure --user --prefix=`pwd` --bindir=`pwd`/bin/ &&\
-        runghc Setup.lhs build && runghc Setup.lhs install 
+	cabal install --user --bindir=`pwd`/bin/
 
 cabal-static:
 	runghc Setup.lhs configure --user --prefix=`pwd` --bindir=`pwd`/bin/ \
diff --git a/README.rst b/README.rst
--- a/README.rst
+++ b/README.rst
@@ -138,7 +138,7 @@
 .. [Collins_2002] Collins, Michael. 2002. Discriminative training
    methods for Hidden Markov Models: Theory and experiments with
    perceptron
-   algorithms. EMNLP 2002. http://ir.hit.edu.cn/~car/research/dishmm.pdf
+   algorithms. EMNLP 2002. http://www.clic.cs.columbia.edu/~mcollins/papers/tagperc.pdf
 
 
 .. [Chrupala_and_Klakow_2010] Grzegorz Chrupała and Dietrich
diff --git a/lib/Helper/Text.hs b/lib/Helper/Text.hs
--- a/lib/Helper/Text.hs
+++ b/lib/Helper/Text.hs
@@ -53,8 +53,8 @@
 type Txt = Text
 
 instance Binary Txt where
-    put = put . encodeUtf8
-    get = fmap decodeUtf8 get
+   put = put . encodeUtf8
+   get = fmap decodeUtf8 get
 
 splitOn :: Char -> Txt -> [Txt]
 splitOn c = Prelude.map pack . Utils.splitOn c . unpack
diff --git a/sequor.cabal b/sequor.cabal
--- a/sequor.cabal
+++ b/sequor.cabal
@@ -1,5 +1,5 @@
 Name:                sequor
-Version:             0.7.2
+Version:             0.7.5
 Description:         A sequence labeler based on Collins's sequence perceptron.
 Synopsis:	     A sequence labeler based on Collins's sequence perceptron.
 Homepage:	     https://bitbucket.org/gchrupala/sequor
@@ -34,7 +34,7 @@
                      vector >= 0.5, 
                      array >= 0.2, 
                      pretty >= 1.0,
-                     text >= 0.10, 
+                     text >= 0.10 && < 1.0, 
                      split >= 0.2,
                      nlp-scores >= 0.6.0
   Exposed-modules:   NLP.Sequor, NLP.Sequor.CoNLL, NLP.Sequor.FeatureTemplate, NLP.Sequor.Config
diff --git a/src/NLP/Perceptron/Sequence.hs b/src/NLP/Perceptron/Sequence.hs
--- a/src/NLP/Perceptron/Sequence.hs
+++ b/src/NLP/Perceptron/Sequence.hs
@@ -54,6 +54,7 @@
                        , oYs         :: [Yi]
                        , oBeam       :: !Int 
                        , oRate       :: !Float
+                       , oRateDecay  :: !Float
                        , oEpochs     :: !Int 
                        , oFeatBounds     :: Maybe (Int,Int)
                        , oStopWinSize    :: !Int
@@ -87,9 +88,11 @@
       return $ Model os ws
 
 instance Binary.Binary Options where
-    put (Options a b c d e f g h i j) = Binary.put a >> Binary.put b >> Binary.put c 
-                               >> Binary.put d >>  Binary.put e >> Binary.put f
-                               >> Binary.put g >> Binary.put h >> Binary.put i >> Binary.put j
+    put (Options a b c d e f g h i j k) = 
+        Binary.put a >> Binary.put b >> Binary.put c 
+       >> Binary.put d >>  Binary.put e >> Binary.put f
+       >> Binary.put g >> Binary.put h >> Binary.put i 
+       >> Binary.put j >> Binary.put k
     get = {-# SCC "get2" #-} do
       a <- Binary.get
       a == a `seq` return ()
@@ -111,7 +114,9 @@
       i == i `seq` return ()
       j <- Binary.get
       j == j `seq` return ()
-      return $ Options a b c d e f g h i j
+      k <- Binary.get
+      k == k `seq` return ()
+      return $ Options a b c d e f g h i j k
 
 yDictFind :: Options -> Xi -> [Yi]
 yDictFind opts fs = 
@@ -195,13 +200,14 @@
         -> [(X,Y)]
         -> (STRef s Int, WeightsST s, WeightsST s)
         -> ST s ()
-iter opts _ ss (c,params,params_a) = do
+iter opts i ss (c,params,params_a) = do
     for_ ss $ \ (x,y) -> do
       params' <- AU.unsafeFreeze params
       let (y',phi_xy') = decode' opts (params'`dot`) x
       when (y' /= y) $ do 
         let phi_xy = phi opts x y 
-            update = (phi_xy `minus` phi_xy') `scale` oRate opts
+            update = (phi_xy `minus` phi_xy') 
+                          `scale` (oRate opts * fromIntegral i ** (- oRateDecay opts))
         params `plus_` update
         c' <- readSTRef c
         params_a `plus_` (update `scale` fromIntegral c')
diff --git a/src/NLP/Sequor.hs b/src/NLP/Sequor.hs
--- a/src/NLP/Sequor.hs
+++ b/src/NLP/Sequor.hs
@@ -52,8 +52,8 @@
 
 
   
--- | @predict model sentence@ returns the best label sequence for
---  sentence. A sentence is a sequence of 'Token's.
+-- | @predict model xs@ returns the best label sequence for
+--  each sentence in xs. A sentence is a sequence of 'Token's.
 predict :: ModelData -> [[Token]] -> [[Label]]
 predict m testdat = 
     let bounds = oFeatBounds . P.options . model $ m
@@ -90,6 +90,7 @@
 
 defaultFlags :: Flags
 defaultFlags = Flags { flagRate         = 0.01
+                     , flagRateDecay    = 0.0
                      , flagBeam         = 10
                      , flagIter         = 10
                      , flagMinFeatCount = 100
@@ -172,6 +173,7 @@
                      , oYs   = ys'
                      , oBeam = flagBeam . flags $ conf
                      , oRate = flagRate . flags $ conf
+                     , oRateDecay = flagRateDecay . flags $ conf
                      , oEpochs = flagIter . flags $ conf
                      , oFeatBounds = bounds
                      , oStopWinSize = flagStopWinSize . flags $ conf
diff --git a/src/NLP/Sequor/Config.hs b/src/NLP/Sequor/Config.hs
--- a/src/NLP/Sequor/Config.hs
+++ b/src/NLP/Sequor/Config.hs
@@ -10,6 +10,7 @@
 
 
 data Flags = Flags { flagRate          :: !Float
+                   , flagRateDecay     :: !Float
                    , flagBeam          :: !Int
                    , flagIter          :: !Int
                    , flagMinFeatCount  :: !Int
@@ -28,9 +29,10 @@
                      }
 
 instance B.Binary Flags where
-    get = do (f1,f2,f3,f4,f5,f6,f7,f8,f9,f10) <- B.get
-             return $ Flags f1 f2 f3 f4 f5 f6 f7 f8 f9 f10
-    put (Flags f1 f2 f3 f4 f5 f6 f7 f8 f9 f10) = B.put (f1,f2,f3,f4,f5,f6,f7,f8,f9,f10)
+    get = do ((f1,f2,f3,f4,f5,f6,f7,f8,f9,f10),f11) <- B.get
+             return $ Flags f1 f2 f3 f4 f5 f6 f7 f8 f9 f10 f11
+    put (Flags f1 f2 f3 f4 f5 f6 f7 f8 f9 f10 f11) = 
+        B.put ((f1,f2,f3,f4,f5,f6,f7,f8,f9,f10),f11)
 
 instance B.Binary Config where
     get = let g = B.get
diff --git a/src/sequor.hs b/src/sequor.hs
--- a/src/sequor.hs
+++ b/src/sequor.hs
@@ -13,6 +13,8 @@
                 , OptDescr(Option), ArgDescr(ReqArg,NoArg))
 import NLP.Sequor.Config(Flags(..))
 import Text.Printf
+import qualified Paths_sequor as Paths 
+import Data.Version (showVersion)
 
 commands :: [(String, CommandSpec Flags)] 
 commands = 
@@ -20,6 +22,9 @@
              [ Option [] ["rate"] 
                       (ReqArg (\a o -> o { flagRate = read a }) "NUM (0.01)")
                       "learning rate"
+             , Option [] ["rate-decay"]
+                      (ReqArg (\a o -> o { flagRateDecay = read a }) "NUM (0.0)")
+                      "learning rate decay"
              , Option [] ["beam"] 
                       (ReqArg (\a o -> o { flagBeam = read a }) "INT (10)")
                       "beam size"
@@ -87,7 +92,7 @@
               | (i,(err_train, err_dev, ch)) <- zip [(1::Int) ..] scores ]
   
 version :: Command Flags 
-version _ _ = putStrLn "sequor-0.2.2"
+version _ _ = putStrLn $ "sequor " ++ showVersion Paths.version
 
 help :: Command Flags
 help _ _ = usage commands msg []