sess.run(tf.global_variables_initializer())
sess.run(tf.local_variables_initializer())
prec, rec, f1, auprc = _eval(sess, model, x_val, y_val)
print('Prec:{:.4f} | Rec:{:.4f} | F1:{:.4f} | Eval_auprc:{:.4f}'.
format(prec, rec, f1, auprc))
sys.stdout.flush()
# Start training
start_time = time.time()
for i in range(nb_epochs):
print('==== Training epoch {} ===='.format(i))
sys.stdout.flush()
# shuffle for each epoch
x_train, y_train = _shuffle(x_train, y_train)
loss_dis_sum, loss_gen_sum, loss_enc_sum = 0., 0., 0.
for j, batch_data in DataInput(x_train, train_batch_size):
# Update discriminator
if (j % d_g_iter != 0) or (j == 0):
loss_dis, loss_gen, loss_enc = model.train(sess,
batch_data,
learning_rate,
train_d=True)
# Update generator and encoder
else:
loss_dis, loss_gen, loss_enc = model.train(sess,
batch_data,
learning_rate,
# model.restore(sess, 'train_logs_5%_0.8623/talkingdata/cross-e/0.9/1/ckpt')
print('Total params: ', np.sum([np.prod(v.get_shape().as_list()) for v in tf.trainable_variables()]))
start_time = time.time()
auroc, prec, rec, f1 = _eval(sess, model, x_val, y_val)
print('Eval_auc:{:.4f} | prec:{:.4f} | rec:{:.4f} | f1:{:.4f}\tTime Cost:{:.4f}'.format(auroc, prec,
rec, f1,
time.time()-start_time))
sys.stdout.flush()
start_time = time.time()
for i in range(nb_epochs):
print('==== Training epoch {} ===='.format(i))
sys.stdout.flush()
x, y = _shuffle(x, y)
loss_dis_sum, loss_gen_sum, loss_enc_sum = 0., 0., 0.
for j, batch_data in DataInput(x, train_batch_size):
# train D
if (j % d_g_iter != 0) or (j == 0):
loss_dis, loss_gen, loss_enc = model.train(sess, batch_data, learning_rate, train_d=True)
# train G, E
else:
loss_dis, loss_gen, loss_enc = model.train(sess, batch_data, learning_rate, train_d=False)
loss_dis_sum += loss_dis
loss_gen_sum += loss_gen
loss_enc_sum += loss_enc
print_iter = 1000 if best_auroc > 0.90 else 2000
if j % print_iter == 0:
print('== Epoch {} Batch {}\tLoss_dis:{:.4f} Loss_gen:{:.4f} loss_enc:{:.4f} =='\
