def test_MLBL_implementation():
""" Test Your Implementation of Forward and Backward """
import coco_proc
import trainer
import cPickle as pickle
z, zd, zt = coco_proc.process(context=5)
d = {}
d['name'] = 'testrun'
d['loc'] = 'models/mlbl_model.pkl'
d['context'] = 5
d['learning_rate'] = 0.43
d['momentum'] = 0.23
d['batch_size'] = 40
d['maxepoch'] = 10
d['hidden_size'] = 441
d['word_decay'] = 3e-7
d['context_decay'] = 1e-8
d['factors'] = 50
# Load the word embeddings
embed_map = trainer.load_embeddings()
# Unpack some stuff from the data
train_ngrams = z['ngrams']
train_labels = z['labels']
train_instances = z['instances']
word_dict = z['word_dict']
index_dict = z['index_dict']
context = z['context']
vocabsize = len(z['word_dict'])
trainIM = z['IM']
train_index = z['index']
net = MLBL(name=d['name'],
loc=d['loc'],
seed=1234,
V=vocabsize,
K=100,
D=trainIM.shape[1],
h=d['hidden_size'],
context=d['context'],
batchsize=1,
maxepoch=d['maxepoch'],
eta_t=d['learning_rate'],
gamma_r=d['word_decay'],
gamma_c=d['context_decay'],
f=0.99,
p_i=d['momentum'],
p_f=d['momentum'],
T=20.0,
verbose=1)
# Train the network
X = train_instances
indX = train_index
Y = train_labels
net.init_params(embed_map, index_dict)
context_size = d['context']
batchX = X[0:context_size]
batchY = Y[0:context_size].toarray()
batchindX = indX[0:context_size].astype(int).flatten()
batchindX = np.floor(batchindX / 5).astype(int)
batchIm = trainIM[batchindX]
# check forward implementation
ft = net.forward(net.params, batchX, batchIm, test=True)
# load gt feature
ft_gt = pickle.load(open("data/val_implementation.p", "rb"))
# it should be less than 1.0e-5
print 'Difference (L2 norm) between implemented and ground truth feature = {}'.format(
np.linalg.norm(ft_gt - ft))
def response():
print("RESPONSE ...")
# Loading model
model = torch.load(model_path)
model.eval()
# Pre-processing inputs
input_text = request.form.get('message')
print(input_text)
input_text = input_text.strip()
if input_text == '':
print("Oops! You input nothing!")
emotions = ['NONE']
movies = [['NONE']]
songs = [['NONE']]
return render_template('response.html',
emotions=emotions,
movies=movies,
songs=songs)
else:
pro_sent = preprocessor(input_text)
# Embedding and vectorize
word2idx, _, embeddings = load_embeddings(model_conf)
sample = vectorize(pro_sent, word2idx, max_length)
# Processing to get model inputs
samples = []
lengths = []
samples.append(sample)
lengths.append(len(pro_sent))
# tensor([17, 4, 37, 42, 21, 16, 13, 44, 29, 10, 15, 22, 21, 23, 18, 23, 25, 10,
# 23, 9, 22, 18, 14, 15, 6, 33, 14, 30, 13, 22, 26, 17],
# device='cuda:0')
# torch.ones([2, 4], dtype=torch.float64, device=cuda0)
samples = np.asarray(samples)
lengths = np.asarray(lengths)
samples = torch.tensor(samples)
lengths = torch.tensor(lengths)
samples = samples.to(DEVICE)
lengths = lengths.to(DEVICE)
# Running model
outputs, attentions = model(samples, lengths)
# print(attentions)
# tensor([ 2.1146, -1.7304, 2.0117, -1.3296, -3.1048, -5.9759, -2.7536, -2.7494, print(outputs)
# -1.8445, -4.1412, -4.7449], device='cuda:0', grad_fn=<ThAddBackward>)
# tensor([0.0521, 0.0631, 0.0632, 0.0632, 0.0632, 0.0632, 0.0631, 0.0632, 0.0632, print(attentions)
# 0.0632, 0.0631, 0.0632, 0.0632, 0.0632, 0.0632, 0.0632],
# device='cuda:0', grad_fn=<DivBackward1>)
# gold: 1 1 0 0 0 0 1 0 0 0 1
# [ 2.11464429 -1.7303592 2.01172185 -1.32956982 -3.10483193 -5.97593689 print(posts)
# -2.75357819 -2.74935317 -1.84453487 -4.14115143 -4.74489594]
posts = outputs.data.cpu().numpy()
predicted = np.clip(
np.sign(posts), a_min=0,
a_max=None) # 1 1 0 0 0 0 1 0 0 0 1
predicted = predicted.astype(np.int32)
emotions = []
ids = set()
sum_item = 0
for idx, item in enumerate(predicted):
if item == 1:
emotions.append(labels[idx])
ids.add(recommend_id[idx])
sum_item += item
if sum_item == 0:
emotions.append(labels[11]) # neutral
ids.add(recommend_id[11])
if len(emotions) > 6:
print(
"Hey, there are more than 6 predicted emotions. Below are original emotions and emotions for displaying."
)
print(emotions)
emotions = emotions[:6]
print(emotions)
else:
print(emotions)
print(ids)
# movies and music matching
all_movies = load_movies(movie_path)
all_music = load_music(music_path)
movies = []
songs = []
for i in ids:
for m in all_movies[i]:
movies.append(m)
for s in all_music[i]:
songs.append(s)
print(songs)
print(movies)
if len(movies) > 3:
random_int = set()
while len(random_int) < 3:
index = np.random.randint(low=0, high=len(movies))
random_int.add(index)
new_movies = []
for i in random_int:
new_movies.append(movies[i])
movies = new_movies
print(movies)
if len(songs) > 6:
random_int = set()
while len(random_int) < 6:
index = np.random.randint(low=0, high=len(songs))
random_int.add(index)
# indexs = np.random.randint(low=0, high=len(songs), size=6)
new_songs = []
for i in random_int:
new_songs.append(songs[i])
songs = new_songs
print(songs)
return render_template('response.html',
emotions=emotions,
movies=movies,
songs=songs)