def get_settings(self):
"""
Initialize and customize settings.
"""
translation_settings = TranslationSettings()
translation_settings.input = open('../../en-de/in')
translation_settings.output = open('../../en-de/out', 'w')
translation_settings.models = ["model.npz"]
translation_settings.beam_size = 12
translation_settings.normalization_alpha = 1.0
return translation_settings
def get_settings(self):
"""
Initialize and customize settings.
"""
translation_settings = TranslationSettings()
translation_settings.models = ["model.npz"]
translation_settings.num_processes = 1
translation_settings.beam_width = 12
translation_settings.normalization_alpha = 1.0
translation_settings.suppress_unk = True
return translation_settings
def get_settings(model, beam):
"""
Initialize and customize settings.
"""
translation_settings = TranslationSettings()
translation_settings.models = [model]
translation_settings.num_processes = 1
translation_settings.beam_width = beam
translation_settings.normalization_alpha = 1.0
translation_settings.suppress_unk = True
translation_settings.get_word_probs = False
return translation_settings
def test_ende(self):
with open('en-de/in', 'r', encoding='utf-8') as in_file, \
open('en-de/out', 'w', encoding='utf-8') as out_file:
os.chdir('models/en-de/')
settings = TranslationSettings()
settings.input = in_file
settings.output = out_file
settings.models = ["model.npz"]
settings.beam_size = 12
settings.normalization_alpha = 1.0
translate(settings=settings)
os.chdir('../..')
self.outputEqual('en-de/ref2','en-de/out')
def test_ende(self):
with open('en-de/in', 'r', encoding='utf-8') as in_file, \
open('en-de/out', 'w', encoding='utf-8') as out_file:
os.chdir('models/en-de/')
settings = TranslationSettings()
settings.input = in_file
settings.output = out_file
settings.models = ["model.npz"]
settings.beam_size = 12
settings.normalization_alpha = 1.0
translate(settings=settings)
os.chdir('../..')
self.outputEqual('en-de/ref2', 'en-de/out')
def outputEqual(output1, output2):
"""given two translation outputs, check that output string is identical
"""
for i, (line, line2) in enumerate(
zip(open(output1).readlines(),
open(output2).readlines())):
#assertEqual(line.strip(), line2.strip())
print "translate {}".format(i)
print line
print line2
# English-German WMT16 system, no dropout
"""
Initialize and customize settings.
"""
translation_settings = TranslationSettings()
translation_settings.models = ["model_test/model.npz-80000"]
#translation_settings.num_processes = 8
translation_settings.beam_width = 10
translation_settings.normalization_alpha = 1.0
translation_settings.verbose = True
translation_settings.n_best = True
translation_settings.suppress_unk = True
translate(input_file=open('data/translated/fr.00'),
output_file=open('data/fr.00.8.out', 'w'),
translation_settings=translation_settings)
print "everyting ok"
#outputEqual('en-de/ref2','en-de/out')