gpu_options = tf.GPUOptions(per_process_gpu_memory_fraction=config.GPU_USAGE)
with tf.Graph().as_default():
sess = tf.Session(config=tf.ConfigProto(gpu_options=gpu_options))
with sess.as_default():
model = DetGP(data.num_vocab, data.num_nodes, data.text,
data.train_edges)
opt = tf.train.AdamOptimizer(config.lr)
train_op = opt.minimize(model.total_loss)
sess.run(tf.global_variables_initializer())
inducing_points = get_initial_inducing(sess, model,
config.inducing_num)
model.load_inducing_points(inducing_points)
#training
log.write('start training : {0}'.format(model_name))
for epoch in range(config.num_epoch):
loss_epoch = 0
batches = data.generate_batches()
num_batch = len(batches)
for i in range(num_batch):
batch = batches[i]
node1, node2, node3 = zip(*batch)
node1, node2, node3 = np.array(node1), np.array(
node2), np.array(node3)
#text1, text2,text3=data.text[node1],data.text[node2],data.text[node3]
feed_dict = {
# model.edges: data.edges,
# model.text_all: data.text,
gpu_options = tf.GPUOptions(per_process_gpu_memory_fraction=config.GPU_USAGE)
with tf.Graph().as_default():
sess = tf.Session(config=tf.ConfigProto(gpu_options=gpu_options))
with sess.as_default():
model = DetGP(data.num_vocab, data.num_nodes, data.text,
data.train_edges)
opt = tf.train.AdamOptimizer(config.lr)
train_op = opt.minimize(model.total_loss)
sess.run(tf.global_variables_initializer())
inducing_points = get_initial_inducing(sess, model,
config.inducing_num)
model.load_inducing_points(inducing_points)
#training
log.write('start training : {0}'.format(model_name))
auc_best = 0.
for epoch in range(config.num_epoch):
loss_epoch = 0
batches = data.generate_batches()
num_batch = len(batches)
for i in range(num_batch):
batch = batches[i]
node1, node2, node3 = zip(*batch)
node1, node2, node3 = np.array(node1), np.array(
node2), np.array(node3)
feed_dict = {
model.node_a_ids: node1,
model.node_b_ids: node2,
model.node_n_ids: node3