def build_graph(self):
self.train_inputs = tf.placeholder(tf.int32, shape=[self.batch_size])
self.train_labels = tf.placeholder(tf.int32, shape=[self.batch_size])
self.learning_rate = tf.placeholder(tf.float32)
# variables
embeddings = tf.Variable(
tf.random_uniform([self.vocabulary_size, self.embedding_size],
-1.0, 1.0))
softmax_weights = tf.Variable(
tf.truncated_normal([self.embedding_size, self.vocabulary_size],
stddev=1.0 / math.sqrt(self.embedding_size)))
softmax_biases = tf.Variable(tf.zeros([self.vocabulary_size]))
self.gathered = tf.gather(embeddings, self.train_inputs)
prediction = tf.matmul(self.gathered, softmax_weights) + softmax_biases
self.loss = tf.reduce_mean(
tf.nn.sparse_softmax_cross_entropy_with_logits(
labels=self.train_labels, logits=prediction))
self.optimizer = tf.train.GradientDescentOptimizer(
self.learning_rate).minimize(self.loss)
self.sess = tf.Session()
self.sess.run(tf.global_variables_initializer())
self.saver = tf.train.Saver()
def make_node(self):
return gof.Apply(
self,
[],
[
theano.Variable(Generic()),
tensor(self.dtype, broadcastable=self.broadcastable),
],
)
def make_node(self, request, data):
return gof.Apply(self, [request, data], [theano.Variable(Generic())])
def make_node(self, data):
return gof.Apply(self, [data],
[theano.Variable(Generic()),
data.type()])
def test_get_output_input_is_variable(self, layer):
variable = theano.Variable("myvariable")
assert layer.get_output(variable) is variable
def test_get_output_input_is_variable(self, layer, get_output):
variable = theano.Variable("myvariable")
assert get_output(layer, variable) is variable
def test_get_output_input_is_variable(self, layer, get_outputs):
variable = theano.Variable("myvariable")
assert get_outputs(layer, variable) == (variable, )
