nlp_ruby 0.3; learning rate

2 files changed, 34 insertions, 43 deletions

diff --git a/bold_reranking.rb b/bold_reranking.rb
index 3041ced..8f8cfab 100755
--- a/bold_reranking.rb
+++ b/bold_reranking.rb
@@ -29,7 +29,7 @@ class FeatureFactory
     @filter_features = false
     if cfg['filter_features']
       @filter_features = true
-      @stopwords_target = ReadFile.new(cfg['filter_features']).readlines.map{ |i| i.strip.downcase }
+      @stopwords_target = ReadFile.readlines(cfg['filter_features']).map{ |i| i.strip.downcase }
     end
   end
 
@@ -107,7 +107,7 @@ class MosesKbestEntryWithPhraseAlignment < Translation
 
   def initialize
     super
-    @other_score = -1.0/0
+    @scores[:rr] = -1.0/0
   end
 
   def get_phrases
@@ -126,11 +126,8 @@ class MosesKbestEntryWithPhraseAlignment < Translation
     @raw.scan(/\|-?\d+\||\|\d+-\d+\|/).map{ |i| i[1..-2] }.map{ |i| _span i }
   end
 
-  def other_score model=nil
-    if model
-      @other_score = model.dot(@f)
-    end
-    return @other_score
+  def score model
+    @scores[:rr] = model.dot(@f)
   end
 end
 
@@ -140,34 +137,27 @@ class ConstrainedSearchOracle < MosesKbestEntryWithPhraseAlignment
     @id = -1
     @raw = s.strip.split(' : ', 2)[1].gsub(/(\[|\])/, '|')
     @s = @raw.gsub(/\s*\|\d+-\d+\||\|-?\d+\|\s*/, ' ').gsub(/\s+/, ' ')
-    @score = 1.0/0
-    @other_score = -1.0/0
+    @scores[:rr] = -1.0/0
   end
 end
 
-def structured_update model, hypothesis, oracle
+def structured_update model, hypothesis, oracle, learning_rate
   if hypothesis.s != oracle.s
-    model += oracle.f - hypothesis.f
+    model += (oracle.f - hypothesis.f) * learning_rate
     return [model, 1]
   end
   return [model, 0]
 end
 
-def ranking_update w, hypothesis, oracle
-  if oracle.other_score <= hypothesis.other_score \
+def ranking_update w, hypothesis, oracle, learning_rate
+  if oracle.scores[:rr] <= hypothesis.scores[:rr] \
       && oracle.s != hypothesis.s
-    model += oracle.f - hypothesis.f
+    model += (oracle.f - hypothesis.f) * learning_rate
     return [model, 1]
   end
   return [model, 0]
 end
 
-def write_model fn, w
-  f = WriteFile.new fn
-  f.write w.to_s+"\n"
-  f.close
-end
-
 def read_additional_phrase_pairs fn
   f = ReadFile.new fn
   add = {}
@@ -192,17 +182,19 @@ end
 
 def main
   usage if ARGV.size != 1
-  cfg = read_cfg ARGV[0]
-
-  sources      = ReadFile.new(cfg['sources']).readlines
-  oracles      = ReadFile.new(cfg['oracles']).readlines
-  kbest_lists  = read_kbest_lists cfg['kbest_lists'], MosesKbestEntryWithPhraseAlignment
-  iterations   = cfg['iterate'].to_i
-  output       = WriteFile.new cfg['output']
-  output_model = cfg['output_model']
-  silent       = true if cfg['silent']
-  verbose      = true if cfg['verbose']
-  cheat        = true if cfg['cheat']
+  cfg = read_config ARGV[0]
+
+  sources       = ReadFile.readlines cfg['sources']
+  oracles       = ReadFile.readlines cfg['oracles']
+  kbest_lists   = read_kbest_lists cfg['kbest_lists'], MosesKbestEntryWithPhraseAlignment
+  learning_rate = cfg['learning_rate'].to_f
+  learning_rate = 1.0 if !learning_rate
+  iterations    = cfg['iterate'].to_i
+  output        = WriteFile.new cfg['output']
+  output_model  = cfg['output_model']
+  silent        = true if cfg['silent']
+  verbose       = true if cfg['verbose']
+  cheat         = true if cfg['cheat']
 
   additional_phrase_pairs = nil
   if cfg['additional_phrase_pairs']
@@ -218,7 +210,7 @@ def main
 
   model = SparseVector.new
   if cfg['init_model']
-    model.from_s ReadFile.new(cfg['init_model']).read
+    model.from_s ReadFile.read cfg['init_model']
   end
 
   sz = sources.size
@@ -235,26 +227,25 @@ def main
     kbest = kbest_lists[j]
     kbest.each { |k|
       k.f = ff.produce k, sources[j]
-      k.other_score model
+      k.score model
     }
 
-    hypothesis = kbest[ kbest.map{ |k| k.other_score }.max_index ]
+    hypothesis = kbest[ kbest.map{ |k| k.scores[:rr] }.max_index ]
 
     if !cheat
       output.write "#{hypothesis.s}\n"
     end
 
-    oracle = ConstrainedSearchOracle.new
-    oracle.from_s oracles[j]
+    oracle = ConstrainedSearchOracle.from_s oracles[j]
     oracle.f = ff.produce oracle, sources[j]
-    oracle.other_score model
+    oracle.score model
 
     err = 0
     case cfg['update']
     when 'structured'
-      model, err = structured_update model, hypothesis, oracle
+      model, err = structured_update model, hypothesis, oracle, learning_rate
     when 'ranking'
-      model, err = ranking_update model, hypothesis, oracle
+      model, err = ranking_update model, hypothesis, oracle, learning_rate
     else
       STDERR.write "Don't know update method '#{cfg['update']}', exiting.\n"
       exit 1
@@ -262,8 +253,8 @@ def main
     overall_errors += err
 
     if cheat
-      kbest.each { |k| k.other_score model }
-      hypothesis = kbest[ kbest.map{ |k| k.other_score }.max_index ]
+      kbest.each { |k| k.score model }
+      hypothesis = kbest[ kbest.map{ |k| k.scores[:rr] }.max_index ]
       output.write "#{hypothesis.s}\n"
     end
 
@@ -279,7 +270,7 @@ def main
 
   elapsed = Time.now - start
   STDERR.write"#{elapsed.round 2} s, #{(elapsed/Float(sz)).round 2} s per kbest; model size: #{model.size}\n\n" if !silent
-  write_model(output_model, model) if output_model
+  WriteFile.write model.to_s+"\n", output_model if output_model
   output.close
 end
 
diff --git a/example/example.ini b/example/example.ini
index 46686bd..8c08a5d 100644
--- a/example/example.ini
+++ b/example/example.ini
@@ -7,7 +7,7 @@ ff_target_ngrams      = 4          # 4 fix
 ff_phrase_pairs       = true       # true /path/to/phrase_table
 #filter_features      = /path/to/target/stopwords_file
 binary_feature_values = true
-iterate               = 3
+iterate               = 1
 
 output                   = - 
 output_model             = /dev/null