From b6a8fdc981daa3b01826a3be4c2355ee5cd61f0a Mon Sep 17 00:00:00 2001
From: Patrick Simianer <p@simianer.de>
Date: Mon, 28 Nov 2011 11:34:41 +0100
Subject: hstreaming stuff

---
 dtrain/hstreaming/cdec.ini                | 17 +++++++++++++++--
 dtrain/hstreaming/dtrain.ini              | 17 +++++++++++------
 dtrain/hstreaming/dtrain.sh               |  4 ++--
 dtrain/hstreaming/hadoop-streaming-job.sh | 16 ++++++++--------
 dtrain/hstreaming/red-avg.rb              | 10 ++++++----
 dtrain/hstreaming/red-test                |  1 +
 6 files changed, 43 insertions(+), 22 deletions(-)

(limited to 'dtrain/hstreaming')

diff --git a/dtrain/hstreaming/cdec.ini b/dtrain/hstreaming/cdec.ini
index 5afa89a9..ce1e1ae2 100644
--- a/dtrain/hstreaming/cdec.ini
+++ b/dtrain/hstreaming/cdec.ini
@@ -4,5 +4,18 @@ scfg_max_span_limit=15
 intersection_strategy=cube_pruning
 cubepruning_pop_limit=200
 feature_function=WordPenalty
-feature_function=KLanguageModel nc-wmt11.en.srilm.3.gz
-feature_function=RuleIdentityFeatures
+feature_function=KLanguageModel test/example/nc-wmt11.en.srilm.gz
+#feature_function=ArityPenalty
+#feature_function=CMR2008ReorderingFeatures
+#feature_function=InputIndicator
+#feature_function=LexNullJump
+#feature_function=NewJump
+#feature_function=NgramFeatures
+#feature_function=NonLatinCount
+#feature_function=OutputIndicator
+#feature_function=RuleIdentityFeatures
+#feature_function=RuleNgramFeatures
+#feature_function=RuleShape
+#feature_function=SourceSpanSizeFeatures
+#feature_function=SourceWordPenalty
+#feature_function=SpanFeatures
diff --git a/dtrain/hstreaming/dtrain.ini b/dtrain/hstreaming/dtrain.ini
index 708bbe46..118a27c5 100644
--- a/dtrain/hstreaming/dtrain.ini
+++ b/dtrain/hstreaming/dtrain.ini
@@ -1,10 +1,15 @@
+input=-
+output=-
 decoder_config=cdec.ini
+tmp=/var/hadoop/mapred/local/
+epochs=10
 k=100
 N=4
-epochs=10
-input=-
-output=-
+learning_rate=0.0001
+gamma=0.00001
 scorer=stupid_bleu
-sample_from=forest
-pair_sampling=all
-tmp=/var/hadoop/mapred/local
+sample_from=kbest
+filter=uniq
+pair_sampling=108010
+pair_threshold=0
+select_weights=last
diff --git a/dtrain/hstreaming/dtrain.sh b/dtrain/hstreaming/dtrain.sh
index b6847591..ea0276dd 100755
--- a/dtrain/hstreaming/dtrain.sh
+++ b/dtrain/hstreaming/dtrain.sh
@@ -1,8 +1,8 @@
 #!/bin/bash
 
-pushd .
+pushd . &>/dev/null
 cd ..
 ID=$(basename $(pwd)) # attempt_...
-popd
+popd &>/dev/null
 ./dtrain -c dtrain.ini --hstreaming $ID 
 
diff --git a/dtrain/hstreaming/hadoop-streaming-job.sh b/dtrain/hstreaming/hadoop-streaming-job.sh
index 788c9fd1..4c0238f3 100755
--- a/dtrain/hstreaming/hadoop-streaming-job.sh
+++ b/dtrain/hstreaming/hadoop-streaming-job.sh
@@ -1,26 +1,26 @@
 #!/bin/bash
 
-ID=
 EXP=test
 
 HADOOP_HOME=/usr/lib/hadoop-0.20
 JAR=contrib/streaming/hadoop-streaming-0.20.2-cdh3u1.jar
 HSTREAMING="$HADOOP_HOME/bin/hadoop jar $HADOOP_HOME/$JAR"
 
-IN=nc-v6.de-en/nc-v6.de-en-dtrain.1500m
-OUT=nc-v6.de-en/nc-v6.de-en-dtrain.1500m-weights
+ IN=nc-v6.de-en.cs.giza.loo/nc-v6.de-en.cs.giza.loo-dtrain1.sz2
+OUT=out/$EXP-weights
 
 $HSTREAMING \
-    -mapper "dtrain -c dtrain.ini --hstreaming" \
+    -mapper "dtrain.sh" \
     -reducer "red-avg.rb" \
     -input $IN \
     -output $OUT \
+    -file dtrain.sh \
     -file red-avg.rb \
-    -file ../dtrain \
+    -file ~/exp/cdec-dtrain-ro/dtrain/dtrain \
     -file dtrain.ini \
     -file cdec.ini \
-    -file nc-wmt11.en.srilm.3.gz \
+    -file ~/exp/data/nc-v6.en.3.unk.probing.kenv5 \
     -jobconf mapred.reduce.tasks=1 \
-    -jobconf mapred.max.map.failures.percent=100 \
-    -jobconf mapred.job.name="dtrain $ID $EXP"
+    -jobconf mapred.max.map.failures.percent=0 \
+    -jobconf mapred.job.name="dtrain $EXP"
 
diff --git a/dtrain/hstreaming/red-avg.rb b/dtrain/hstreaming/red-avg.rb
index 771f4c0e..c0b69eb4 100755
--- a/dtrain/hstreaming/red-avg.rb
+++ b/dtrain/hstreaming/red-avg.rb
@@ -1,6 +1,5 @@
 #!/usr/bin/env ruby1.9.1
 
-
 shard_count_key = "__SHARD_COUNT__"
 
 STDIN.set_encoding 'utf-8'
@@ -8,7 +7,7 @@ STDOUT.set_encoding 'utf-8'
 
 w = {}
 c = {}
-w.default = 0
+w.default = 0.
 c.default = 0
 while line = STDIN.gets
   key, val = line.split /\t/
@@ -19,7 +18,10 @@ end
 puts "# dtrain reducer: average"
 shard_count = w["__SHARD_COUNT__"]
 w.each_key { |k|
-  if k == shard_count_key then next end
-  puts "#{k}\t#{w[k]/shard_count}"
+  if k == shard_count_key
+    puts "# shard count: #{shard_count.to_i}"
+  else
+    puts "#{k}\t#{w[k]/shard_count}\t# #{c[k]}"
+  end
 }
 
diff --git a/dtrain/hstreaming/red-test b/dtrain/hstreaming/red-test
index a2a0edb1..2623d697 100644
--- a/dtrain/hstreaming/red-test
+++ b/dtrain/hstreaming/red-test
@@ -5,4 +5,5 @@ a	1
 b	2
 c	3.5
 d	1
+e	2
 __SHARD_COUNT__	2
-- 
cgit v1.2.3