NEXT

5 files changed, 13 insertions, 6 deletions

diff --git a/dtrain/NEXT b/dtrain/NEXT
new file mode 100644
index 00000000..24939cf3
--- /dev/null
+++ b/dtrain/NEXT
@@ -0,0 +1,6 @@
+cuda vecs?
+target side rule ngrams
+decoder meta-parameters testing
+cdyer -> sa-extract -> loo?
+reranking while sgd
+
diff --git a/dtrain/README.md b/dtrain/README.md
index e28bebe7..f4e1abed 100644
--- a/dtrain/README.md
+++ b/dtrain/README.md
@@ -1,4 +1,4 @@
-This is a really fast (parallelizable) tuning method for cdec as used here:
+This is a simple (but parallelizable) tuning method for cdec, as used here:
   "Joint Feature Selection in Distributed Stochastic
    Learning for Large-Scale Discriminative Training in
    SMT" Simianer, Riezler, Dyer
diff --git a/dtrain/dtrain.h b/dtrain/dtrain.h
index 59ceb6f6..783aa179 100644
--- a/dtrain/dtrain.h
+++ b/dtrain/dtrain.h
@@ -13,7 +13,7 @@
 
 #include "filelib.h"
 
-//#define DTRAIN_LOCAL
+#define DTRAIN_LOCAL
 
 #define DTRAIN_DOTS 10 // when to display a '.'
 #define DTRAIN_GRAMMAR_DELIM "########EOS########"
diff --git a/dtrain/hstreaming/avg.rb b/dtrain/hstreaming/avg.rb
index 91d4e29a..5deb62e4 100755
--- a/dtrain/hstreaming/avg.rb
+++ b/dtrain/hstreaming/avg.rb
@@ -22,10 +22,11 @@ else
 end
 w.each_key { |k|
   if k == shard_count_key
-    puts "# shard count: #{shard_count.to_i}"
+    #puts "# shard count: #{shard_count.to_i}"
+    next
   else
     puts "#{k}\t#{w[k]/shard_count}"
-    puts "# #{c[k]}"
+    #puts "# #{c[k]}"
   end
 }
 
diff --git a/dtrain/hstreaming/lplp.rb b/dtrain/hstreaming/lplp.rb
index 40409bbd..57353adb 100755
--- a/dtrain/hstreaming/lplp.rb
+++ b/dtrain/hstreaming/lplp.rb
@@ -82,7 +82,7 @@ def _test()
   cut(w, method(:mean), 1, 2.05)
  exit
 end
-_test()
+#_test()
 
 # actually do something
 def usage()
@@ -107,7 +107,7 @@ STDOUT.set_encoding 'utf-8'
 w = {}
 shard_count = 0
 while line = STDIN.gets
-  key, val = line.split /\t/
+  key, val = line.split /\s+/
   if key == shard_count_key
     shard_count += 1
     next