def __init__(self, args, vocab):
dim_emb = args.dim_emb
filter_sizes = [int(x) for x in args.filter_sizes.split(',')]
n_filters = args.n_filters
self.dropout = tf.placeholder(tf.float32, name='dropout')
self.learning_rate = tf.placeholder(tf.float32, name='learning_rate')
self.x = tf.placeholder(
tf.int32,
[None, None], # batch_size * max_len
name='x')
self.y = tf.placeholder(tf.float32, [None], name='y')
# embeding
embedding = tf.get_variable('embedding', [vocab.size, dim_emb])
x = tf.nn.embedding_lookup(embedding, self.x)
# import CNN
self.logits = cnn(x, filter_sizes, n_filters, self.dropout, 'cnn')
# Sigmoid binomiial ditribution?
self.probs = tf.sigmoid(self.logits)
# cross entropy loss
loss = tf.nn.sigmoid_cross_entropy_with_logits(labels=self.y,
logits=self.logits)
# mean over all samples
self.loss = tf.reduce_mean(loss)
# optimize
self.optimizer = tf.train.AdamOptimizer(self.learning_rate) \
.minimize(self.loss)
self.saver = tf.train.Saver()
def __init__(self, args, vocab):
dim_emb = args.dim_emb
filter_sizes = [int(x) for x in args.filter_sizes.split(',')]
n_filters = args.n_filters
self.dropout = tf.placeholder(tf.float32,
name='dropout')
self.learning_rate = tf.placeholder(tf.float32,
name='learning_rate')
self.x = tf.placeholder(tf.int32, [None, None], #batch_size * max_len
name='x')
self.y = tf.placeholder(tf.float32, [None],
name='y')
embedding = tf.get_variable('embedding', [vocab.size, dim_emb])
x = tf.nn.embedding_lookup(embedding, self.x)
self.logits = cnn(x, filter_sizes, n_filters, self.dropout, 'cnn')
self.probs = tf.sigmoid(self.logits)
loss = tf.nn.sigmoid_cross_entropy_with_logits(
labels=self.y, logits=self.logits)
self.loss = tf.reduce_mean(loss)
self.optimizer = tf.train.AdamOptimizer(self.learning_rate) \
.minimize(self.loss)
self.saver = tf.train.Saver()
def classify_cnn(model_path, text, dp, t = 0):
f = open(dp, 'rb')
p = pickle.load(f)
f.close()
v = list(p['vocab'])
c = list(p['categories'])
v.sort()
c.sort()
v = { w : i + 1 for i, w in enumerate(v)}
c = { w : i for i, w in enumerate(c)}
ms = p['max_sequence']
ms = 10 * int(ms / 10) + 10
model = cnn((ms,), len(v) + 1, ms, pretrained_weights=model_path, t = t)
x = preprocessing(text)
x = compute_input(x, v, ms)
x = np.reshape(x, (1, 5430))
y = model.predict_on_batch(x)
return read_output(y, c)
def __init__(self, args, vocab):
dim_emb = args.dim_emb
filter_sizes = [int(x) for x in args.filter_sizes.split(',')]
n_filters = args.n_filters
self.dropout = tf.compat.v1.placeholder(tf.float32,
name='dropout')
self.learning_rate = tf.compat.v1.placeholder(tf.float32,
name='learning_rate')
self.x = tf.compat.v1.placeholder(tf.int32, [None, None], #batch_size * max_len
name='x')
self.y = tf.compat.v1.placeholder(tf.float32, [None],
name='y')
embedding = tf.compat.v1.get_variable('embedding', [vocab.size, dim_emb])
x = tf.nn.embedding_lookup(params=embedding, ids=self.x)
self.logits = cnn(x, filter_sizes, n_filters, self.dropout, 'cnn')
self.probs = tf.sigmoid(self.logits)
loss = tf.nn.sigmoid_cross_entropy_with_logits(
labels=self.y, logits=self.logits)
self.loss = tf.reduce_mean(input_tensor=loss)
self.optimizer = tf.compat.v1.train.AdamOptimizer(self.learning_rate) \
.minimize(self.loss)
self.saver = tf.compat.v1.train.Saver()
def __init__(self, args, vocab):
dim_emb = args.dim_emb
filter_sizes = range(1, 1 + args.max_filter_width)
n_filters = args.n_filters
self.dropout = tf.placeholder(tf.float32, name='dropout')
self.learning_rate = tf.placeholder(tf.float32, name='learning_rate')
self.x = tf.placeholder(
tf.int32,
[None, None], #batch_size * max_len
name='x')
self.y = tf.placeholder(tf.float32, [None], name='y')
embedding = tf.get_variable('embedding', [vocab.size, dim_emb])
x = tf.nn.embedding_lookup(embedding, self.x)
self.logits = cnn(x, filter_sizes, n_filters, self.dropout, 'cnn')
self.probs = tf.sigmoid(self.logits)
loss = tf.nn.sigmoid_cross_entropy_with_logits(labels=self.y,
logits=self.logits)
self.loss = tf.reduce_mean(loss)
self.optimizer = tf.train.AdamOptimizer(self.learning_rate) \
.minimize(self.loss)
self.saver = tf.train.Saver()
image = tf.decode_raw(features['image_raw'], tf.uint8)
# height and width of image is 28 pixels, 28*28 = 784
image.set_shape([784])
image = tf.cast(image, tf.float32) * (1. / 255)
label = tf.cast(features['label'], tf.int32)
images_batch, labels_batch = tf.train.shuffle_batch([image, label],
batch_size=128,
capacity=2000,
min_after_dequeue=1000)
# keep probability just the inverse of 'dropout' rate used in some other frameworks
# keep out probability of 0.75 == dropout rate of 0.25
keep_prob = 0.75
y_pred = cnn(images_batch, keep_prob)
loss = tf.nn.sparse_softmax_cross_entropy_with_logits(logits=y_pred,
labels=labels_batch)
loss_mean = tf.reduce_mean(loss)
train_op = tf.train.AdamOptimizer().minimize(loss)
sess = tf.Session()
init = tf.global_variables_initializer()
sess.run(init)
init = tf.local_variables_initializer()
sess.run(init)
coord = tf.train.Coordinator()
threads = tf.train.start_queue_runners(sess=sess, coord=coord)
