tf.summary.scalar("accuracy", accurate)
summ = tf.summary.merge_all()
sess.run(tf.global_variables_initializer())
def trans(i):
if i == 0:
return [0, 1]
if i == 1:
return [1, 0]
for items in rd(max_records=2):
X = [item["features"].toArray() for item in items]
Y = [trans(item["label"]) for item in items]
if len(X) > 0:
_, gs = sess.run([train_step, global_step],
feed_dict={input_x: X, input_y: Y})
[train_accuracy, s, loss] = sess.run([accurate, summ, xent],
feed_dict={input_x: X, input_y: Y})
print('train_accuracy %g, loss: %g, global step: %d' % (
train_accuracy,
loss,
gs))
sys.stdout.flush()
p = mlsql.params()
mlsql_model.save_model(p["internalSystemParam"]["tempModelLocalPath"], sess, input_x, input_y, True)
sess.close()
TEST_X, TEST_Y = mlsql.get_validate_data()
TEST_Y = [item.toArray() for item in TEST_Y]
for ep in range(epochs):
for items in rd(max_records=batch_size):
X = [item[input_col].toArray() for item in items]
Y = [item[label_col].toArray() for item in items]
_, gs = sess.run([train_step, global_step],
feed_dict={
input_x: X,
input_y: Y
})
if gs % print_interval == 0:
[train_accuracy, s, loss] = sess.run([accurate, summ, xent],
feed_dict={
input_x: X,
input_y: Y
})
[test_accuracy, test_s,
test_lost] = sess.run([accurate, summ, xent],
feed_dict={
input_x: TEST_X,
input_y: TEST_Y
})
print(
'train_accuracy %g,test_accuracy %g, loss: %g,test_lost: %g, global step: %d, ep:%d'
% (train_accuracy, test_accuracy, loss, test_lost, gs, ep))
sys.stdout.flush()
mlsql_model.save_model(tempModelLocalPath, sess, input_x, _logits, True)
sess.close()
sess.run(tf.global_variables_initializer())
TEST_X, TEST_Y = mlsql.get_validate_data()
TEST_Y = [item.toArray() for item in TEST_Y]
for ep in range(epochs):
for items in rd(max_records=batch_size):
X = [item[input_col].toArray() for item in items]
Y = [item[label_col].toArray() for item in items]
if len(X) == 0:
print("bad news , this round no message fetched")
if len(X) > 0:
_, gs = sess.run([train_step, global_step],
feed_dict={input_x: X, input_y: Y})
if gs % print_interval == 0:
[train_accuracy, s, loss] = sess.run([accurate, summ, xent],
feed_dict={input_x: X, input_y: Y})
[test_accuracy, test_s, test_lost] = sess.run([accurate, summ, xent],
feed_dict={input_x: TEST_X, input_y: TEST_Y})
print('train_accuracy %g,test_accuracy %g, loss: %g, test_lost: %g,global step: %d, ep:%d' % (
train_accuracy,
test_accuracy,
loss,
test_lost,
gs, ep))
sys.stdout.flush()
mlsql_model.save_model(tempModelLocalPath, sess, input_x, _logits, True)
sess.close()
def trans(i):
if i == 0:
return [0, 1]
if i == 1:
return [1, 0]
for items in rd(max_records=2):
X = [item["features"].toArray() for item in items]
Y = [trans(item["label"]) for item in items]
if len(X) > 0:
_, gs = sess.run([train_step, global_step],
feed_dict={
input_x: X,
input_y: Y
})
[train_accuracy, s, loss] = sess.run([accurate, summ, xent],
feed_dict={
input_x: X,
input_y: Y
})
print('train_accuracy %g, loss: %g, global step: %d' %
(train_accuracy, loss, gs))
sys.stdout.flush()
p = mlsql.params()
mlsql_model.save_model(p["internalSystemParam"]["tempModelLocalPath"], sess,
input_x, input_y, True)
sess.close()