From 8aa29810bb77611cc20b7a384897ff6703783ea1 Mon Sep 17 00:00:00 2001
From: Chris Dyer <cdyer@allegro.clab.cs.cmu.edu>
Date: Sun, 18 Nov 2012 13:35:42 -0500
Subject: major restructure of the training code

---
 training/dpmert/test_aer/README     |  8 ++++++++
 training/dpmert/test_aer/cdec.ini   |  3 +++
 training/dpmert/test_aer/corpus.src |  3 +++
 training/dpmert/test_aer/grammar    | 12 ++++++++++++
 training/dpmert/test_aer/ref.0      |  3 +++
 training/dpmert/test_aer/weights    | 13 +++++++++++++
 6 files changed, 42 insertions(+)
 create mode 100644 training/dpmert/test_aer/README
 create mode 100644 training/dpmert/test_aer/cdec.ini
 create mode 100644 training/dpmert/test_aer/corpus.src
 create mode 100644 training/dpmert/test_aer/grammar
 create mode 100644 training/dpmert/test_aer/ref.0
 create mode 100644 training/dpmert/test_aer/weights

(limited to 'training/dpmert/test_aer')

diff --git a/training/dpmert/test_aer/README b/training/dpmert/test_aer/README
new file mode 100644
index 00000000..819b2e32
--- /dev/null
+++ b/training/dpmert/test_aer/README
@@ -0,0 +1,8 @@
+To run the test:
+
+../dist-vest.pl --local --metric aer cdec.ini --source-file corpus.src --ref-files=ref.0 --weights weights
+
+This will optimize the parameters of the tiny lexical translation model
+so as to minimize the AER of the Viterbi alignment on the development
+set in corpus.src according to the reference alignments in ref.0.
+
diff --git a/training/dpmert/test_aer/cdec.ini b/training/dpmert/test_aer/cdec.ini
new file mode 100644
index 00000000..08187848
--- /dev/null
+++ b/training/dpmert/test_aer/cdec.ini
@@ -0,0 +1,3 @@
+formalism=lextrans
+grammar=grammar
+aligner=true
diff --git a/training/dpmert/test_aer/corpus.src b/training/dpmert/test_aer/corpus.src
new file mode 100644
index 00000000..31b23971
--- /dev/null
+++ b/training/dpmert/test_aer/corpus.src
@@ -0,0 +1,3 @@
+el gato negro ||| the black cat
+el gato ||| the cat
+el libro ||| the book
diff --git a/training/dpmert/test_aer/grammar b/training/dpmert/test_aer/grammar
new file mode 100644
index 00000000..9d857824
--- /dev/null
+++ b/training/dpmert/test_aer/grammar
@@ -0,0 +1,12 @@
+el ||| cat ||| F1=1
+el ||| the ||| F2=1
+el ||| black ||| F3=1
+el ||| book ||| F11=1
+gato ||| cat ||| F4=1 NN=1
+gato ||| black ||| F5=1
+gato ||| the ||| F6=1
+negro ||| the ||| F7=1
+negro ||| cat ||| F8=1
+negro ||| black ||| F9=1
+libro ||| the ||| F10=1
+libro ||| book ||| F12=1 NN=1
diff --git a/training/dpmert/test_aer/ref.0 b/training/dpmert/test_aer/ref.0
new file mode 100644
index 00000000..734a9c5b
--- /dev/null
+++ b/training/dpmert/test_aer/ref.0
@@ -0,0 +1,3 @@
+0-0 1-2 2-1
+0-0 1-1
+0-0 1-1
diff --git a/training/dpmert/test_aer/weights b/training/dpmert/test_aer/weights
new file mode 100644
index 00000000..afc9282e
--- /dev/null
+++ b/training/dpmert/test_aer/weights
@@ -0,0 +1,13 @@
+F1 0.1
+F2 -.5980815
+F3 0.24235
+F4 0.625
+F5 0.4514
+F6 0.112316
+F7 -0.123415
+F8 -0.25390285
+F9 -0.23852
+F10 0.646
+F11 0.413141
+F12 0.343216
+NN -0.1215
-- 
cgit v1.2.3