def main():
with open(CONFIG_FN) as f:
conf = json.load(f)
global sent_groups
with open(conf["sent_groups"]) as f:
sent_groups = json.load(f)["groups"]
kb = load_kb(conf["kb"], 'name')
sys_vocab, sys_word2idx = load_sys_vocab(conf["sys_vocab"])
sys_codec = Codec(sys_vocab, sys_word2idx)
onto, onto_idx = load_ontology(conf["ontology"])
word2idx, embed = load_embed(**conf)
usr_codec = Codec([], word2idx)
trk_model, slot_len_sum = load_tracker_model(onto, embed, conf, kb)
trk_model.eval()
hidden = trk_model.state_tracker.init_hidden()
kb_vec = Variable(torch.zeros(1, conf["kb_indicator_len"]))
sentence_generator = SentenceGenerator(kb, onto, sent_groups)
for line in iter(sys.stdin.readline, ''):
inp = usr_codec.encode(line.strip())
inp = Variable(torch.LongTensor([
inp,
]))
sentvecs, states_reps, states_preds, hidden, sent_grp_preds = trk_model(
inp, None, hidden)
criteria = to_search_criteria(states_preds, onto)
ret, kb_vec = get_kb_result(kb, criteria, conf["kb_indicator_len"])
# print criteria, kb_vec
sentvecs = sentvecs.view(1, -1)
states_reps = states_reps.view(1, -1)
print_ret(states_preds, sent_grp_preds, onto, sentence_generator)
def test_get_sentence(self):
# arrange
gen = SentenceGenerator()
# act, assert
self.assertEqual('es ist zwölf uhr'.split(' '),
gen.get_sentence(time(12, 0)))
self.assertEqual('es ist zwölf uhr'.split(' '),
gen.get_sentence(time(0, 0)))
self.assertEqual('es ist sechs uhr'.split(' '),
gen.get_sentence(time(6, 0)))
self.assertEqual('es ist sechs uhr'.split(' '),
gen.get_sentence(time(18, 0)))
self.assertEqual('es ist fünf nach eins'.split(' '),
gen.get_sentence(time(1, 5)))
self.assertEqual('es ist zehn nach zwei'.split(' '),
gen.get_sentence(time(2, 11)))
self.assertEqual('es ist viertel nach drei'.split(' '),
gen.get_sentence(time(3, 14)))
self.assertEqual('es ist zwanzig nach vier'.split(' '),
gen.get_sentence(time(4, 22)))
self.assertEqual('es ist fünf vor halb fünf2'.split(' '),
gen.get_sentence(time(4, 25)))
self.assertEqual('es ist halb sechs'.split(' '),
gen.get_sentence(time(5, 30)))
self.assertEqual('es ist fünf nach halb sieben'.split(' '),
gen.get_sentence(time(6, 34)))
self.assertEqual('es ist zwanzig vor acht'.split(' '),
gen.get_sentence(time(7, 42)))
self.assertEqual('es ist viertel vor neun'.split(' '),
gen.get_sentence(time(20, 45)))
self.assertEqual('es ist zehn vor zehn2'.split(' '),
gen.get_sentence(time(21, 49)))
self.assertEqual('es ist fünf vor elf'.split(' '),
gen.get_sentence(time(10, 55)))
self.assertEqual('es ist fünf vor eins'.split(' '),
gen.get_sentence(time(12, 55)))
embed_model = SentenceTransformer('bert-base-nli-mean-tokens')
embedding_fn = lambda s: embed_model.encode(
[s.replace("@@ ", "").replace("@@", "")])[0]
# Load or create the model.
model_save_path = "drive/My Drive/sentence_generator-{}.pickle".format(
model_id)
if os.path.isfile(model_save_path):
print("Loading sentence generator.")
sentence_generator = pickle.load(open(model_save_path, "rb"))
# Flattening is different depending on the decoder used in the sentence generator.
# sentence_generator._decoder.decoder.lstm.flatten_parameters() # VanillaRNNDecoder
sentence_generator._decoder.decoder.rnn.flatten_parameters() # DecoderRNN
print("Loaded sentence generator.")
else:
sentence_generator = SentenceGenerator(embedding_fn, id=model_id)
if train:
# Note: pickled vocab is only used if a vocab does not already exist (i.e.
# when training a model from scratch).
if not use_pickled_vocab:
pickled_vocab = ""
# Note: the training data used is either pickled_shards, pickled_pairs,
# or all_sentences (in that priority).
if not use_pickled_shards:
pickled_shards = []
if not use_pickled_pairs:
pickled_pairs = ""
sentence_generator.train_generator(all_sentences,
num_train_iters,
pickled_pairs=pickled_pairs,
