a lot of stuff, fast_sparse_vector, perceptron, removed sofia, sample [...]

13 files changed, 106 insertions, 7 deletions

diff --git a/dtrain/test/EXAMPLE/cdec.ini b/dtrain/test/EXAMPLE/cdec.ini
index b6e92b5f..e57138b0 100644
--- a/dtrain/test/EXAMPLE/cdec.ini
+++ b/dtrain/test/EXAMPLE/cdec.ini
@@ -2,5 +2,6 @@ formalism=scfg
 add_pass_through_rules=true
 feature_function=WordPenalty
 cubepruning_pop_limit=30
+feature_function=KLanguageModel data/nc-wmt11.en.srilm.gz
 scfg_max_span_limit=15
 
diff --git a/dtrain/test/EXAMPLE/dtrain.ini b/dtrain/test/EXAMPLE/dtrain.ini
index 1467b332..ffafd0b8 100644
--- a/dtrain/test/EXAMPLE/dtrain.ini
+++ b/dtrain/test/EXAMPLE/dtrain.ini
@@ -1,10 +1,10 @@
 decoder_config=test/EXAMPLE/cdec.ini
 kbest=100
 ngrams=3
-epochs=22
+epochs=8
 input=test/EXAMPLE/dtrain.nc-1k
 scorer=approx_bleu
 output=test/EXAMPLE/weights.gz
-stop_after=5
-wprint=Glue WordPenalty LanguageModel LanguageModel_OOV PhraseModel_0 PhraseModel_1 PhraseModel_2 PhraseModel_3 PhraseModel_4
+stop_after=1000
+wprint=Glue WordPenalty LanguageModel LanguageModel_OOV PhraseModel_0 PhraseModel_1 PhraseModel_2 PhraseModel_3 PhraseModel_4 PassThrough
 
diff --git a/dtrain/test/log_reg/bin_class.cc b/dtrain/test/log_reg/bin_class.cc
new file mode 100644
index 00000000..19bcde25
--- /dev/null
+++ b/dtrain/test/log_reg/bin_class.cc
@@ -0,0 +1,4 @@
+#include "bin_class.h"
+
+Objective::~Objective() {}
+
diff --git a/dtrain/test/log_reg/bin_class.h b/dtrain/test/log_reg/bin_class.h
new file mode 100644
index 00000000..3466109a
--- /dev/null
+++ b/dtrain/test/log_reg/bin_class.h
@@ -0,0 +1,22 @@
+#ifndef _BIN_CLASS_H_
+#define _BIN_CLASS_H_
+
+#include <vector>
+#include "sparse_vector.h"
+
+struct TrainingInstance {
+  // TODO add other info? loss for MIRA-type updates?
+  SparseVector<double> x_feature_map;
+  bool y;
+};
+
+struct Objective {
+  virtual ~Objective();
+
+  // returns f(x) and f'(x)
+  virtual double ObjectiveAndGradient(const SparseVector<double>& x,
+                  const std::vector<TrainingInstance>& training_instances,
+                  SparseVector<double>* g) const = 0;
+};
+
+#endif
diff --git a/dtrain/test/log_reg/log_reg.cc b/dtrain/test/log_reg/log_reg.cc
new file mode 100644
index 00000000..ec2331fe
--- /dev/null
+++ b/dtrain/test/log_reg/log_reg.cc
@@ -0,0 +1,39 @@
+#include "log_reg.h"
+
+#include <vector>
+#include <cmath>
+
+#include "sparse_vector.h"
+
+using namespace std;
+
+double LogisticRegression::ObjectiveAndGradient(const SparseVector<double>& x,
+                              const vector<TrainingInstance>& training_instances,
+                              SparseVector<double>* g) const {
+  double cll = 0;
+  for (int i = 0; i < training_instances.size(); ++i) {
+    const double dotprod = training_instances[i].x_feature_map.dot(x); // TODO no bias, if bias, add x[0]
+    double lp_false = dotprod;
+    double lp_true = -dotprod;
+    if (0 < lp_true) {
+      lp_true += log1p(exp(-lp_true));
+      lp_false = log1p(exp(lp_false));
+    } else {
+      lp_true = log1p(exp(lp_true));
+      lp_false += log1p(exp(-lp_false));
+    }
+    lp_true *= -1;
+    lp_false *= -1;
+    if (training_instances[i].y) {  // true label
+      cll -= lp_true;
+      (*g) -= training_instances[i].x_feature_map * exp(lp_false);
+      // (*g)[0] -= exp(lp_false); // bias
+    } else {                  // false label
+      cll -= lp_false;
+      (*g) += training_instances[i].x_feature_map * exp(lp_true);
+      // g += corpus[i].second * exp(lp_true);
+    }
+  }
+  return cll;
+}
+
diff --git a/dtrain/test/log_reg/log_reg.h b/dtrain/test/log_reg/log_reg.h
new file mode 100644
index 00000000..ecc560b8
--- /dev/null
+++ b/dtrain/test/log_reg/log_reg.h
@@ -0,0 +1,14 @@
+#ifndef _LOG_REG_H_
+#define _LOG_REG_H_
+
+#include <vector>
+#include "sparse_vector.h"
+#include "bin_class.h"
+
+struct LogisticRegression : public Objective {
+  double ObjectiveAndGradient(const SparseVector<double>& x,
+                              const std::vector<TrainingInstance>& training_instances,
+                              SparseVector<double>* g) const;
+};
+
+#endif
diff --git a/dtrain/test/nc-wmt11/dtrain.ini b/dtrain/test/nc-wmt11/dtrain.ini
index 51033f2d..ddbf5da7 100644
--- a/dtrain/test/nc-wmt11/dtrain.ini
+++ b/dtrain/test/nc-wmt11/dtrain.ini
@@ -2,7 +2,7 @@ decoder_config=test/nc-wmt11/cdec.ini
 kbest=100
 ngrams=3
 epochs=8
-input=data/nc-wmt11.loo.localf.p0.500.rule-id #nc-wmt11-de-en-dyer-cs-joshua.tok.lc.fixamp1.loo.psg.dtrain.localf.p0
+input=data/nc-wmt11.loo.localf.p0.500.rule-id
 scorer=approx_bleu
 output=data/w/nc-wmt11.loo.p0.weights.gz
 #stop_after=100
diff --git a/dtrain/test/toy.dtrain.ini b/dtrain/test/toy.dtrain.ini
index cacb3a2c..35f76281 100644
--- a/dtrain/test/toy.dtrain.ini
+++ b/dtrain/test/toy.dtrain.ini
@@ -2,8 +2,9 @@ decoder_config=test/cdec.ini
 kbest=4
 ngrams=1
 epochs=3
-input=data/in.toy
+input=test/toy.in
 scorer=bleu
 output=toy.gz
 #stop_after=1000
+wprint=logp use_shell use_house PassThrough
 
diff --git a/dtrain/test/toy.in b/dtrain/test/toy.in
index 63f97158..989a1f77 100644
--- a/dtrain/test/toy.in
+++ b/dtrain/test/toy.in
@@ -1,2 +1,2 @@
-0	ich sah ein kleines haus	i saw a little house	[S] ||| [NP,1] [VP,2] ||| [1] [2] ||| logp=0 __NEXT_RULE__ [NP] ||| ich ||| i ||| logp=0 __NEXT_RULE__ [NP] ||| ein [NN,1] ||| a [1] ||| logp=0 __NEXT_RULE__ [NN] ||| [JJ,1] haus ||| [1] house ||| logp=0 use_house=1 __NEXT_RULE__ [NN] ||| [JJ,1] haus ||| [1] shell ||| logp=0 use_shell=1 __NEXT_RULE__ [JJ] ||| kleines ||| small ||| logp=0 __NEXT_RULE__ [JJ] ||| kleines ||| little ||| logp=0 __NEXT_RULE__ [JJ] ||| grosses ||| big ||| logp=0 __NEXT_RULE__ [JJ] ||| grosses ||| large ||| logp=0 __NEXT_RULE__ [VP] ||| [V,1] [NP,2] ||| [1] [2] ||| logp=0 __NEXT_RULE__ [V] ||| sah ||| saw ||| logp=0 __NEXT_RULE__ [V] ||| fand ||| found ||| logp=0
-1	ich fand ein grosses haus	i found a large house	[S] ||| [NP,1] [VP,2] ||| [1] [2] ||| logp=0 __NEXT_RULE__ [NP] ||| ich ||| i ||| logp=0 __NEXT_RULE__ [NP] ||| ein [NN,1] ||| a [1] ||| logp=0 __NEXT_RULE__ [NN] ||| [JJ,1] haus ||| [1] house ||| logp=0 use_house=1 __NEXT_RULE__ [NN] ||| [JJ,1] haus ||| [1] shell ||| logp=0 use_shell=1 __NEXT_RULE__ [JJ] ||| kleines ||| small ||| logp=0 __NEXT_RULE__ [JJ] ||| kleines ||| little ||| logp=0 __NEXT_RULE__ [JJ] ||| grosses ||| big ||| logp=0 __NEXT_RULE__ [JJ] ||| grosses ||| large ||| logp=0 __NEXT_RULE__ [VP] ||| [V,1] [NP,2] ||| [1] [2] ||| logp=0 __NEXT_RULE__ [V] ||| sah ||| saw ||| logp=0 __NEXT_RULE__ [V] ||| fand ||| found ||| logp=0
+0	ich sah ein kleines haus	i saw a little house	[S] ||| [NP,1] [VP,2] ||| [1] [2] ||| logp=0 __NEXT__RULE__ [NP] ||| ich ||| i ||| logp=0 __NEXT__RULE__ [NP] ||| ein [NN,1] ||| a [1] ||| logp=0 __NEXT__RULE__ [NN] ||| [JJ,1] haus ||| [1] house ||| logp=0 use_house=1 __NEXT__RULE__ [NN] ||| [JJ,1] haus ||| [1] shell ||| logp=0 use_shell=1 __NEXT__RULE__ [JJ] ||| kleines ||| small ||| logp=0 __NEXT__RULE__ [JJ] ||| kleines ||| little ||| logp=0 __NEXT__RULE__ [JJ] ||| grosses ||| big ||| logp=0 __NEXT__RULE__ [JJ] ||| grosses ||| large ||| logp=0 __NEXT__RULE__ [VP] ||| [V,1] [NP,2] ||| [1] [2] ||| logp=0 __NEXT__RULE__ [V] ||| sah ||| saw ||| logp=0 __NEXT__RULE__ [V] ||| fand ||| found ||| logp=0
+1	ich fand ein grosses haus	i found a large house	[S] ||| [NP,1] [VP,2] ||| [1] [2] ||| logp=0 __NEXT__RULE__ [NP] ||| ich ||| i ||| logp=0 __NEXT__RULE__ [NP] ||| ein [NN,1] ||| a [1] ||| logp=0 __NEXT__RULE__ [NN] ||| [JJ,1] haus ||| [1] house ||| logp=0 use_house=1 __NEXT__RULE__ [NN] ||| [JJ,1] haus ||| [1] shell ||| logp=0 use_shell=1 __NEXT__RULE__ [JJ] ||| kleines ||| small ||| logp=0 __NEXT__RULE__ [JJ] ||| kleines ||| little ||| logp=0 __NEXT__RULE__ [JJ] ||| grosses ||| big ||| logp=0 __NEXT__RULE__ [JJ] ||| grosses ||| large ||| logp=0 __NEXT__RULE__ [VP] ||| [V,1] [NP,2] ||| [1] [2] ||| logp=0 __NEXT__RULE__ [V] ||| sah ||| saw ||| logp=0 __NEXT__RULE__ [V] ||| fand ||| found ||| logp=0
diff --git a/dtrain/test/toy_cdec/cdec.ini b/dtrain/test/toy_cdec/cdec.ini
new file mode 100644
index 00000000..3a6bab68
--- /dev/null
+++ b/dtrain/test/toy_cdec/cdec.ini
@@ -0,0 +1,3 @@
+formalism=scfg
+grammar=../dtrain/test/toy_cdec/grammar
+add_pass_through_rules=true
diff --git a/dtrain/test/toy_cdec/grammar b/dtrain/test/toy_cdec/grammar
new file mode 100644
index 00000000..aeed75ef
--- /dev/null
+++ b/dtrain/test/toy_cdec/grammar
@@ -0,0 +1,12 @@
+[S] ||| [NP,1] [VP,2] ||| [1] [2] ||| logp=0
+[NP] ||| ich ||| i ||| logp=0
+[NP] ||| ein [NN,1] ||| a [1] ||| logp=0
+[NN] ||| [JJ,1] haus ||| [1] house ||| logp=0 use_house=1
+[NN] ||| [JJ,1] haus ||| [1] shell ||| logp=0 use_shell=1
+[JJ] ||| kleines ||| small ||| logp=0
+[JJ] ||| kleines ||| little ||| logp=0
+[JJ] ||| grosses ||| big ||| logp=0
+[JJ] ||| grosses ||| large ||| logp=0
+[VP] ||| [V,1] [NP,2] ||| [1] [2] ||| logp=0
+[V] ||| sah ||| saw ||| logp=0
+[V] ||| fand ||| found ||| logp=0
diff --git a/dtrain/test/toy_cdec/in b/dtrain/test/toy_cdec/in
new file mode 100644
index 00000000..e6df9275
--- /dev/null
+++ b/dtrain/test/toy_cdec/in
@@ -0,0 +1 @@
+ich sah ein kleines haus
diff --git a/dtrain/test/toy_cdec/weights b/dtrain/test/toy_cdec/weights
new file mode 100644
index 00000000..10d7ed83
--- /dev/null
+++ b/dtrain/test/toy_cdec/weights
@@ -0,0 +1,2 @@
+logp 1
+use_shell 1