From b738e349be490c24d3604c224f44fc54e16d3d7b Mon Sep 17 00:00:00 2001 From: Victor Chahuneau Date: Sat, 23 Jun 2012 11:59:48 -0400 Subject: Support for sparse/dense vectors in the python extension - SparseVector, DenseVector - improved Lattice - Lattice translation - Hypergraph reweighting, pruning --- python/test.py | 17 ++++++++++++++--- 1 file changed, 14 insertions(+), 3 deletions(-) (limited to 'python/test.py') diff --git a/python/test.py b/python/test.py index 1542dd4f..92403298 100644 --- a/python/test.py +++ b/python/test.py @@ -19,14 +19,16 @@ with gzip.open(grammar_file) as f: # Input sentence sentence = u'澳洲 是 与 北韩 有 邦交 的 少数 国家 之一 。' -print 'Input:', sentence +print ' Input:', sentence.encode('utf8') # Decode forest = decoder.translate(sentence, grammar=grammar) # Get viterbi translation print 'Output[0]:', forest.viterbi().encode('utf8') -print ' Tree[0]:', forest.viterbi_tree().encode('utf8') +print ' ETree[0]:', forest.viterbi_tree().encode('utf8') +print ' FTree[0]:', forest.viterbi_source_tree().encode('utf8') +print 'LgProb[0]:', forest.viterbi_features().dot(decoder.weights) # Get k-best translations for i, (sentence, tree) in enumerate(zip(forest.kbest(5), forest.kbest_tree(5)), 1): @@ -37,6 +39,9 @@ for i, (sentence, tree) in enumerate(zip(forest.kbest(5), forest.kbest_tree(5)), for sentence in forest.sample(5): print 'Sample:', sentence.encode('utf8') +# Get feature vector for 1best +fsrc = forest.viterbi_features() + # Reference lattice lattice = ((('australia',0,1),),(('is',0,1),),(('one',0,1),),(('of',0,1),),(('the',0,4),('a',0,4),('a',0,1),('the',0,1),),(('small',0,1),('tiny',0,1),('miniscule',0,1),('handful',0,2),),(('number',0,1),('group',0,1),),(('of',0,2),),(('few',0,1),),(('countries',0,1),),(('that',0,1),),(('has',0,1),('have',0,1),),(('diplomatic',0,1),),(('relations',0,1),),(('with',0,1),),(('north',0,1),),(('korea',0,1),),(('.',0,1),),) @@ -44,6 +49,12 @@ lat = cdec.Lattice(lattice) assert (lattice == tuple(lat)) # Intersect forest and lattice -forest.intersect(lat) +assert forest.intersect(lat) + # Get best synchronous parse print forest.viterbi_tree() +print forest.viterbi_source_tree() + +# Compare 1best and reference feature vectors +fref = forest.viterbi_features() +print dict(fsrc - fref) -- cgit v1.2.3