def testCreate(self):
with self.test_session():
tree_ensemble_config = tree_config_pb2.DecisionTreeEnsembleConfig()
tree = tree_ensemble_config.trees.add()
_append_to_leaf(tree.nodes.add().leaf, 0, -0.4)
tree_ensemble_config.tree_weights.append(1.0)
# Prepare learner config.
learner_config = learner_pb2.LearnerConfig()
learner_config.num_classes = 2
tree_ensemble_handle = model_ops.tree_ensemble_variable(
stamp_token=3,
tree_ensemble_config=tree_ensemble_config.SerializeToString(),
name="create_tree")
resources.initialize_resources(resources.shared_resources()).run()
result, _, _ = prediction_ops.gradient_trees_prediction(
tree_ensemble_handle,
self._seed, [self._dense_float_tensor],
[self._sparse_float_indices1, self._sparse_float_indices2],
[self._sparse_float_values1, self._sparse_float_values2],
[self._sparse_float_shape1, self._sparse_float_shape2],
[self._sparse_int_indices1], [self._sparse_int_values1],
[self._sparse_int_shape1],
learner_config=learner_config.SerializeToString(),
apply_dropout=False,
apply_averaging=False,
center_bias=False,
reduce_dim=True)
self.assertAllClose(result.eval(), [[-0.4], [-0.4]])
stamp_token = model_ops.tree_ensemble_stamp_token(
tree_ensemble_handle)
self.assertEqual(stamp_token.eval(), 3)
def testCreate(self):
with self.cached_session():
tree_ensemble_config = tree_config_pb2.DecisionTreeEnsembleConfig()
tree = tree_ensemble_config.trees.add()
_append_to_leaf(tree.nodes.add().leaf, 0, -0.4)
tree_ensemble_config.tree_weights.append(1.0)
# Prepare learner config.
learner_config = learner_pb2.LearnerConfig()
learner_config.num_classes = 2
tree_ensemble_handle = model_ops.tree_ensemble_variable(
stamp_token=3,
tree_ensemble_config=tree_ensemble_config.SerializeToString(),
name="create_tree")
resources.initialize_resources(resources.shared_resources()).run()
result, _ = prediction_ops.gradient_trees_prediction(
tree_ensemble_handle,
self._seed, [self._dense_float_tensor], [
self._sparse_float_indices1, self._sparse_float_indices2
], [self._sparse_float_values1, self._sparse_float_values2],
[self._sparse_float_shape1,
self._sparse_float_shape2], [self._sparse_int_indices1],
[self._sparse_int_values1], [self._sparse_int_shape1],
learner_config=learner_config.SerializeToString(),
apply_dropout=False,
apply_averaging=False,
center_bias=False,
reduce_dim=True)
self.assertAllClose(result.eval(), [[-0.4], [-0.4]])
stamp_token = model_ops.tree_ensemble_stamp_token(tree_ensemble_handle)
self.assertEqual(stamp_token.eval(), 3)
def _predict_and_return_dict(self, ensemble_handle, ensemble_stamp, mode):
"""Runs prediction and returns a dictionary of the prediction results.
Args:
ensemble_handle: ensemble resource handle.
ensemble_stamp: stamp of ensemble resource.
mode: learn.ModeKeys.TRAIN or EVAL or INFER.
Returns:
a dictionary of prediction results -
ENSEMBLE_STAMP, PREDICTION, PARTITION_IDS,
NUM_LAYER_ATTEMPTED, NUM_TREES_ATTEMPED.
"""
ensemble_stats = training_ops.tree_ensemble_stats(
ensemble_handle, ensemble_stamp)
num_handlers = (len(self._dense_floats) +
len(self._sparse_float_shapes) +
len(self._sparse_int_shapes))
# Used during feature selection.
used_handlers = model_ops.tree_ensemble_used_handlers(
ensemble_handle, ensemble_stamp, num_all_handlers=num_handlers)
# We don't need dropout info - we can always restore it based on the
# seed.
apply_dropout, seed = _dropout_params(mode, ensemble_stats)
# Make sure ensemble stats run. This will check that the ensemble has
# the right stamp.
with ops.control_dependencies(ensemble_stats):
predictions, _ = prediction_ops.gradient_trees_prediction(
ensemble_handle,
seed,
self._dense_floats,
self._sparse_float_indices,
self._sparse_float_values,
self._sparse_float_shapes,
self._sparse_int_indices,
self._sparse_int_values,
self._sparse_int_shapes,
learner_config=self._learner_config_serialized,
apply_dropout=apply_dropout,
apply_averaging=mode != learn.ModeKeys.TRAIN,
use_locking=True,
center_bias=self._center_bias,
reduce_dim=self._reduce_dim)
partition_ids = prediction_ops.gradient_trees_partition_examples(
ensemble_handle,
self._dense_floats,
self._sparse_float_indices,
self._sparse_float_values,
self._sparse_float_shapes,
self._sparse_int_indices,
self._sparse_int_values,
self._sparse_int_shapes,
use_locking=True)
return _make_predictions_dict(ensemble_stamp, predictions,
partition_ids, ensemble_stats,
used_handlers)
def _predict_and_return_dict(self, ensemble_handle, ensemble_stamp, mode):
"""Runs prediction and returns a dictionary of the prediction results.
Args:
ensemble_handle: ensemble resource handle.
ensemble_stamp: stamp of ensemble resource.
mode: learn.ModeKeys.TRAIN or EVAL or INFER.
Returns:
a dictionary of prediction results -
ENSEMBLE_STAMP, PREDICTION, PARTITION_IDS,
NUM_LAYER_ATTEMPTED, NUM_TREES_ATTEMPED.
"""
ensemble_stats = training_ops.tree_ensemble_stats(ensemble_handle,
ensemble_stamp)
num_handlers = (
len(self._dense_floats) + len(self._sparse_float_shapes) +
len(self._sparse_int_shapes))
# Used during feature selection.
used_handlers = model_ops.tree_ensemble_used_handlers(
ensemble_handle, ensemble_stamp, num_all_handlers=num_handlers)
# We don't need dropout info - we can always restore it based on the
# seed.
apply_dropout, seed = _dropout_params(mode, ensemble_stats)
# Make sure ensemble stats run. This will check that the ensemble has
# the right stamp.
with ops.control_dependencies(ensemble_stats):
predictions, _ = prediction_ops.gradient_trees_prediction(
ensemble_handle,
seed,
self._dense_floats,
self._sparse_float_indices,
self._sparse_float_values,
self._sparse_float_shapes,
self._sparse_int_indices,
self._sparse_int_values,
self._sparse_int_shapes,
learner_config=self._learner_config_serialized,
apply_dropout=apply_dropout,
apply_averaging=mode != learn.ModeKeys.TRAIN,
use_locking=True,
center_bias=self._center_bias,
reduce_dim=self._reduce_dim)
partition_ids = prediction_ops.gradient_trees_partition_examples(
ensemble_handle,
self._dense_floats,
self._sparse_float_indices,
self._sparse_float_values,
self._sparse_float_shapes,
self._sparse_int_indices,
self._sparse_int_values,
self._sparse_int_shapes,
use_locking=True)
return _make_predictions_dict(ensemble_stamp, predictions, partition_ids,
ensemble_stats, used_handlers)
def _get_predictions(self,
tree_ensemble_handle,
learner_config,
apply_dropout=False,
apply_averaging=False,
center_bias=False,
reduce_dim=False):
return prediction_ops.gradient_trees_prediction(
tree_ensemble_handle,
self._seed, [self._dense_float_tensor],
[self._sparse_float_indices1, self._sparse_float_indices2],
[self._sparse_float_values1, self._sparse_float_values2],
[self._sparse_float_shape1, self._sparse_float_shape2],
[self._sparse_int_indices1], [self._sparse_int_values1],
[self._sparse_int_shape1],
learner_config=learner_config,
apply_dropout=apply_dropout,
apply_averaging=apply_averaging,
center_bias=center_bias,
reduce_dim=reduce_dim)
def predict(self, mode):
"""Returns predictions given the features and mode.
Args:
mode: Mode the graph is running in (train|predict|eval).
Returns:
A dict of predictions tensors.
Raises:
ValueError: if features is not valid.
"""
apply_averaging = mode != learn.ModeKeys.TRAIN
# Use the current ensemble to predict on the current batch of input.
# For faster prediction we check if the inputs are on the same device
# as the model. If not, we create a copy of the model on the worker.
input_deps = (self._dense_floats + self._sparse_float_indices +
self._sparse_int_indices)
if not input_deps:
raise ValueError("No input tensors for prediction.")
if any(i.device != input_deps[0].device for i in input_deps):
raise ValueError("All input tensors should be on the same device.")
# Get most current model stamp.
ensemble_stamp = model_ops.tree_ensemble_stamp_token(self._ensemble_handle)
# Determine if ensemble is colocated with the inputs.
if self._ensemble_handle.device != input_deps[0].device:
# Create a local ensemble and get its local stamp.
with ops.name_scope("local_ensemble", "TreeEnsembleVariable") as name:
local_ensemble_handle = (
gen_model_ops.decision_tree_ensemble_resource_handle_op(name=name))
create_op = gen_model_ops.create_tree_ensemble_variable(
local_ensemble_handle, stamp_token=-1, tree_ensemble_config="")
with ops.control_dependencies([create_op]):
local_stamp = model_ops.tree_ensemble_stamp_token(
local_ensemble_handle)
# Determine whether the local ensemble is stale and update it if needed.
def _refresh_local_ensemble_fn():
# Serialize the model from parameter server after reading all inputs.
with ops.control_dependencies(input_deps):
(ensemble_stamp, serialized_model) = (
model_ops.tree_ensemble_serialize(self._ensemble_handle))
# Update local ensemble with the serialized model from parameter server.
with ops.control_dependencies([create_op]):
return model_ops.tree_ensemble_deserialize(
local_ensemble_handle,
stamp_token=ensemble_stamp,
tree_ensemble_config=serialized_model), ensemble_stamp
refresh_local_ensemble, ensemble_stamp = control_flow_ops.cond(
math_ops.not_equal(ensemble_stamp,
local_stamp), _refresh_local_ensemble_fn,
lambda: (control_flow_ops.no_op(), ensemble_stamp))
# Once updated, Use the the local model for prediction.
with ops.control_dependencies([refresh_local_ensemble]):
ensemble_stats = training_ops.tree_ensemble_stats(
local_ensemble_handle, ensemble_stamp)
apply_dropout, seed = _dropout_params(mode, ensemble_stats)
# We don't need dropout info - we can always restore it based on the
# seed.
predictions, predictions_no_dropout, _ = (
prediction_ops.gradient_trees_prediction(
local_ensemble_handle,
seed,
self._dense_floats,
self._sparse_float_indices,
self._sparse_float_values,
self._sparse_float_shapes,
self._sparse_int_indices,
self._sparse_int_values,
self._sparse_int_shapes,
learner_config=self._learner_config_serialized,
apply_dropout=apply_dropout,
apply_averaging=apply_averaging,
use_locking=False,
center_bias=self._center_bias,
reduce_dim=self._reduce_dim))
partition_ids = prediction_ops.gradient_trees_partition_examples(
local_ensemble_handle,
self._dense_floats,
self._sparse_float_indices,
self._sparse_float_values,
self._sparse_float_shapes,
self._sparse_int_indices,
self._sparse_int_values,
self._sparse_int_shapes,
use_locking=False)
else:
with ops.device(self._ensemble_handle.device):
ensemble_stats = training_ops.tree_ensemble_stats(
self._ensemble_handle, ensemble_stamp)
apply_dropout, seed = _dropout_params(mode, ensemble_stats)
# We don't need dropout info - we can always restore it based on the
# seed.
predictions, predictions_no_dropout, _ = (
prediction_ops.gradient_trees_prediction(
self._ensemble_handle,
seed,
self._dense_floats,
self._sparse_float_indices,
self._sparse_float_values,
self._sparse_float_shapes,
self._sparse_int_indices,
self._sparse_int_values,
self._sparse_int_shapes,
learner_config=self._learner_config_serialized,
apply_dropout=apply_dropout,
apply_averaging=apply_averaging,
use_locking=False,
center_bias=self._center_bias,
reduce_dim=self._reduce_dim))
partition_ids = prediction_ops.gradient_trees_partition_examples(
self._ensemble_handle,
self._dense_floats,
self._sparse_float_indices,
self._sparse_float_values,
self._sparse_float_shapes,
self._sparse_int_indices,
self._sparse_int_values,
self._sparse_int_shapes,
use_locking=False)
return _make_predictions_dict(ensemble_stamp, predictions,
predictions_no_dropout, partition_ids,
ensemble_stats)
def testRestore(self):
# Calling self.test_session() without a graph specified results in
# TensorFlowTestCase caching the session and returning the same one
# every time. In this test, we need to create two different sessions
# which is why we also create a graph and pass it to self.test_session()
# to ensure no caching occurs under the hood.
save_path = os.path.join(self.get_temp_dir(), "restore-test")
with ops.Graph().as_default() as graph:
with self.test_session(graph) as sess:
tree_ensemble_config = tree_config_pb2.DecisionTreeEnsembleConfig(
)
tree = tree_ensemble_config.trees.add()
tree_ensemble_config.tree_metadata.add().is_finalized = True
tree_ensemble_config.tree_weights.append(1.0)
_append_to_leaf(tree.nodes.add().leaf, 0, -0.1)
tree_ensemble_config2 = tree_config_pb2.DecisionTreeEnsembleConfig(
)
tree2 = tree_ensemble_config2.trees.add()
tree_ensemble_config.tree_weights.append(1.0)
_append_to_leaf(tree2.nodes.add().leaf, 0, -1.0)
tree_ensemble_config3 = tree_config_pb2.DecisionTreeEnsembleConfig(
)
tree3 = tree_ensemble_config3.trees.add()
tree_ensemble_config.tree_weights.append(1.0)
_append_to_leaf(tree3.nodes.add().leaf, 0, -10.0)
# Prepare learner config.
learner_config = learner_pb2.LearnerConfig()
learner_config.num_classes = 2
tree_ensemble_handle = model_ops.tree_ensemble_variable(
stamp_token=3,
tree_ensemble_config=tree_ensemble_config.
SerializeToString(),
name="restore_tree")
feature_usage_counts = variables.Variable(
initial_value=array_ops.zeros([1], dtypes.int64),
name="feature_usage_counts",
trainable=False)
feature_gains = variables.Variable(
initial_value=array_ops.zeros([1], dtypes.float32),
name="feature_gains",
trainable=False)
resources.initialize_resources(
resources.shared_resources()).run()
variables.initialize_all_variables().run()
my_saver = saver.Saver()
with ops.control_dependencies([
ensemble_optimizer_ops.add_trees_to_ensemble(
tree_ensemble_handle,
tree_ensemble_config2.SerializeToString(),
feature_usage_counts, [0],
feature_gains, [0], [[]],
learning_rate=1)
]):
result, _, _ = prediction_ops.gradient_trees_prediction(
tree_ensemble_handle,
self._seed, [self._dense_float_tensor], [
self._sparse_float_indices1,
self._sparse_float_indices2
], [
self._sparse_float_values1,
self._sparse_float_values2
],
[self._sparse_float_shape1, self._sparse_float_shape2],
[self._sparse_int_indices1],
[self._sparse_int_values1], [self._sparse_int_shape1],
learner_config=learner_config.SerializeToString(),
apply_dropout=False,
apply_averaging=False,
center_bias=False,
reduce_dim=True)
self.assertAllClose([[-1.1], [-1.1]], result.eval())
# Save before adding other trees.
val = my_saver.save(sess, save_path)
self.assertEqual(save_path, val)
# Add more trees after saving.
with ops.control_dependencies([
ensemble_optimizer_ops.add_trees_to_ensemble(
tree_ensemble_handle,
tree_ensemble_config3.SerializeToString(),
feature_usage_counts, [0],
feature_gains, [0], [[]],
learning_rate=1)
]):
result, _, _ = prediction_ops.gradient_trees_prediction(
tree_ensemble_handle,
self._seed, [self._dense_float_tensor], [
self._sparse_float_indices1,
self._sparse_float_indices2
], [
self._sparse_float_values1,
self._sparse_float_values2
],
[self._sparse_float_shape1, self._sparse_float_shape2],
[self._sparse_int_indices1],
[self._sparse_int_values1], [self._sparse_int_shape1],
learner_config=learner_config.SerializeToString(),
apply_dropout=False,
apply_averaging=False,
center_bias=False,
reduce_dim=True)
self.assertAllClose(result.eval(), [[-11.1], [-11.1]])
# Start a second session. In that session the parameter nodes
# have not been initialized either.
with ops.Graph().as_default() as graph:
with self.test_session(graph) as sess:
tree_ensemble_handle = model_ops.tree_ensemble_variable(
stamp_token=0,
tree_ensemble_config="",
name="restore_tree")
my_saver = saver.Saver()
my_saver.restore(sess, save_path)
result, _, _ = prediction_ops.gradient_trees_prediction(
tree_ensemble_handle,
self._seed, [self._dense_float_tensor],
[self._sparse_float_indices1, self._sparse_float_indices2],
[self._sparse_float_values1, self._sparse_float_values2],
[self._sparse_float_shape1, self._sparse_float_shape2],
[self._sparse_int_indices1], [self._sparse_int_values1],
[self._sparse_int_shape1],
learner_config=learner_config.SerializeToString(),
apply_dropout=False,
apply_averaging=False,
center_bias=False,
reduce_dim=True)
# Make sure we only have the first and second tree.
# The third tree was added after the save.
self.assertAllClose(result.eval(), [[-1.1], [-1.1]])
def testSerialization(self):
with ops.Graph().as_default() as graph:
with self.test_session(graph):
tree_ensemble_config = tree_config_pb2.DecisionTreeEnsembleConfig(
)
# Bias tree only for second class.
tree1 = tree_ensemble_config.trees.add()
_append_to_leaf(tree1.nodes.add().leaf, 1, -0.2)
tree_ensemble_config.tree_weights.append(1.0)
# Depth 2 tree.
tree2 = tree_ensemble_config.trees.add()
tree_ensemble_config.tree_weights.append(1.0)
_set_float_split(
tree2.nodes.add().sparse_float_binary_split_default_right.
split, 1, 4.0, 1, 2)
_set_float_split(tree2.nodes.add().dense_float_binary_split, 0,
9.0, 3, 4)
_append_to_leaf(tree2.nodes.add().leaf, 0, 0.5)
_append_to_leaf(tree2.nodes.add().leaf, 1, 1.2)
_append_to_leaf(tree2.nodes.add().leaf, 0, -0.9)
tree_ensemble_handle = model_ops.tree_ensemble_variable(
stamp_token=7,
tree_ensemble_config=tree_ensemble_config.
SerializeToString(),
name="saver_tree")
stamp_token, serialized_config = model_ops.tree_ensemble_serialize(
tree_ensemble_handle)
resources.initialize_resources(
resources.shared_resources()).run()
self.assertEqual(stamp_token.eval(), 7)
serialized_config = serialized_config.eval()
with ops.Graph().as_default() as graph:
with self.test_session(graph):
tree_ensemble_handle2 = model_ops.tree_ensemble_variable(
stamp_token=9,
tree_ensemble_config=serialized_config,
name="saver_tree2")
resources.initialize_resources(
resources.shared_resources()).run()
# Prepare learner config.
learner_config = learner_pb2.LearnerConfig()
learner_config.num_classes = 3
result, _, _ = prediction_ops.gradient_trees_prediction(
tree_ensemble_handle2,
self._seed, [self._dense_float_tensor],
[self._sparse_float_indices1, self._sparse_float_indices2],
[self._sparse_float_values1, self._sparse_float_values2],
[self._sparse_float_shape1, self._sparse_float_shape2],
[self._sparse_int_indices1], [self._sparse_int_values1],
[self._sparse_int_shape1],
learner_config=learner_config.SerializeToString(),
apply_dropout=False,
apply_averaging=False,
center_bias=False,
reduce_dim=True)
# Re-serialize tree.
stamp_token2, serialized_config2 = model_ops.tree_ensemble_serialize(
tree_ensemble_handle2)
# The first example will get bias class 1 -0.2 from first tree and
# leaf 2 payload (sparse feature missing) of 0.5 hence [0.5, -0.2],
# the second example will get the same bias class 1 -0.2 and leaf 3
# payload of class 1 1.2 hence [0.0, 1.0].
self.assertEqual(stamp_token2.eval(), 9)
# Class 2 does have scores in the leaf => it gets score 0.
self.assertEqual(serialized_config2.eval(), serialized_config)
self.assertAllClose(result.eval(), [[0.5, -0.2], [0, 1.0]])
def testRestore(self):
# Calling self.cached_session() without a graph specified results in
# TensorFlowTestCase caching the session and returning the same one
# every time. In this test, we need to create two different sessions
# which is why we also create a graph and pass it to self.cached_session()
# to ensure no caching occurs under the hood.
save_path = os.path.join(self.get_temp_dir(), "restore-test")
with ops.Graph().as_default() as graph:
with self.session(graph) as sess:
# Prepare learner config.
learner_config = learner_pb2.LearnerConfig()
learner_config.num_classes = 2
# Add the first tree and save.
tree_ensemble_config = tree_config_pb2.DecisionTreeEnsembleConfig()
tree = tree_ensemble_config.trees.add()
tree_ensemble_config.tree_metadata.add().is_finalized = True
tree_ensemble_config.tree_weights.append(1.0)
_append_to_leaf(tree.nodes.add().leaf, 0, -0.1)
tree_ensemble_handle = model_ops.tree_ensemble_variable(
stamp_token=3,
tree_ensemble_config=tree_ensemble_config.SerializeToString(),
name="restore_tree")
resources.initialize_resources(resources.shared_resources()).run()
variables.global_variables_initializer().run()
my_saver = saver.Saver()
# Add the second tree and replace the ensemble of the handle.
tree2 = tree_ensemble_config.trees.add()
tree_ensemble_config.tree_weights.append(1.0)
_append_to_leaf(tree2.nodes.add().leaf, 0, -1.0)
# Predict to confirm.
with ops.control_dependencies([
model_ops.tree_ensemble_deserialize(
tree_ensemble_handle,
stamp_token=3,
tree_ensemble_config=tree_ensemble_config.SerializeToString())
]):
result, _ = prediction_ops.gradient_trees_prediction(
tree_ensemble_handle,
self._seed, [self._dense_float_tensor], [
self._sparse_float_indices1, self._sparse_float_indices2
], [self._sparse_float_values1, self._sparse_float_values2],
[self._sparse_float_shape1,
self._sparse_float_shape2], [self._sparse_int_indices1],
[self._sparse_int_values1], [self._sparse_int_shape1],
learner_config=learner_config.SerializeToString(),
apply_dropout=False,
apply_averaging=False,
center_bias=False,
reduce_dim=True)
self.assertAllClose([[-1.1], [-1.1]], result.eval())
# Save before adding other trees.
val = my_saver.save(sess, save_path)
self.assertEqual(save_path, val)
# Add more trees after saving.
tree3 = tree_ensemble_config.trees.add()
tree_ensemble_config.tree_weights.append(1.0)
_append_to_leaf(tree3.nodes.add().leaf, 0, -10.0)
# Predict to confirm.
with ops.control_dependencies([
model_ops.tree_ensemble_deserialize(
tree_ensemble_handle,
stamp_token=3,
tree_ensemble_config=tree_ensemble_config.SerializeToString())
]):
result, _ = prediction_ops.gradient_trees_prediction(
tree_ensemble_handle,
self._seed, [self._dense_float_tensor], [
self._sparse_float_indices1, self._sparse_float_indices2
], [self._sparse_float_values1, self._sparse_float_values2],
[self._sparse_float_shape1,
self._sparse_float_shape2], [self._sparse_int_indices1],
[self._sparse_int_values1], [self._sparse_int_shape1],
learner_config=learner_config.SerializeToString(),
apply_dropout=False,
apply_averaging=False,
center_bias=False,
reduce_dim=True)
self.assertAllClose(result.eval(), [[-11.1], [-11.1]])
# Start a second session. In that session the parameter nodes
# have not been initialized either.
with ops.Graph().as_default() as graph:
with self.session(graph) as sess:
tree_ensemble_handle = model_ops.tree_ensemble_variable(
stamp_token=0, tree_ensemble_config="", name="restore_tree")
my_saver = saver.Saver()
my_saver.restore(sess, save_path)
result, _ = prediction_ops.gradient_trees_prediction(
tree_ensemble_handle,
self._seed, [self._dense_float_tensor], [
self._sparse_float_indices1, self._sparse_float_indices2
], [self._sparse_float_values1, self._sparse_float_values2],
[self._sparse_float_shape1,
self._sparse_float_shape2], [self._sparse_int_indices1],
[self._sparse_int_values1], [self._sparse_int_shape1],
learner_config=learner_config.SerializeToString(),
apply_dropout=False,
apply_averaging=False,
center_bias=False,
reduce_dim=True)
# Make sure we only have the first and second tree.
# The third tree was added after the save.
self.assertAllClose(result.eval(), [[-1.1], [-1.1]])
def testSerialization(self):
with ops.Graph().as_default() as graph:
with self.session(graph):
tree_ensemble_config = tree_config_pb2.DecisionTreeEnsembleConfig()
# Bias tree only for second class.
tree1 = tree_ensemble_config.trees.add()
_append_to_leaf(tree1.nodes.add().leaf, 1, -0.2)
tree_ensemble_config.tree_weights.append(1.0)
# Depth 2 tree.
tree2 = tree_ensemble_config.trees.add()
tree_ensemble_config.tree_weights.append(1.0)
_set_float_split(tree2.nodes.add()
.sparse_float_binary_split_default_right.split, 1, 4.0,
1, 2)
_set_float_split(tree2.nodes.add().dense_float_binary_split, 0, 9.0, 3,
4)
_append_to_leaf(tree2.nodes.add().leaf, 0, 0.5)
_append_to_leaf(tree2.nodes.add().leaf, 1, 1.2)
_append_to_leaf(tree2.nodes.add().leaf, 0, -0.9)
tree_ensemble_handle = model_ops.tree_ensemble_variable(
stamp_token=7,
tree_ensemble_config=tree_ensemble_config.SerializeToString(),
name="saver_tree")
stamp_token, serialized_config = model_ops.tree_ensemble_serialize(
tree_ensemble_handle)
resources.initialize_resources(resources.shared_resources()).run()
self.assertEqual(stamp_token.eval(), 7)
serialized_config = serialized_config.eval()
with ops.Graph().as_default() as graph:
with self.session(graph):
tree_ensemble_handle2 = model_ops.tree_ensemble_variable(
stamp_token=9,
tree_ensemble_config=serialized_config,
name="saver_tree2")
resources.initialize_resources(resources.shared_resources()).run()
# Prepare learner config.
learner_config = learner_pb2.LearnerConfig()
learner_config.num_classes = 3
result, _ = prediction_ops.gradient_trees_prediction(
tree_ensemble_handle2,
self._seed, [self._dense_float_tensor], [
self._sparse_float_indices1, self._sparse_float_indices2
], [self._sparse_float_values1, self._sparse_float_values2],
[self._sparse_float_shape1,
self._sparse_float_shape2], [self._sparse_int_indices1],
[self._sparse_int_values1], [self._sparse_int_shape1],
learner_config=learner_config.SerializeToString(),
apply_dropout=False,
apply_averaging=False,
center_bias=False,
reduce_dim=True)
# Re-serialize tree.
stamp_token2, serialized_config2 = model_ops.tree_ensemble_serialize(
tree_ensemble_handle2)
# The first example will get bias class 1 -0.2 from first tree and
# leaf 2 payload (sparse feature missing) of 0.5 hence [0.5, -0.2],
# the second example will get the same bias class 1 -0.2 and leaf 3
# payload of class 1 1.2 hence [0.0, 1.0].
self.assertEqual(stamp_token2.eval(), 9)
# Class 2 does have scores in the leaf => it gets score 0.
self.assertEqual(serialized_config2.eval(), serialized_config)
self.assertAllClose(result.eval(), [[0.5, -0.2], [0, 1.0]])
def testDropoutSeed(self):
with self.test_session():
tree_ensemble_config = tree_config_pb2.DecisionTreeEnsembleConfig()
# Add 10 trees with some weights.
for i in range(0, 999):
tree = tree_ensemble_config.trees.add()
tree_ensemble_config.tree_metadata.add().is_finalized = True
_append_to_leaf(tree.nodes.add().leaf, 0, -0.4)
tree_ensemble_config.tree_weights.append(i + 1)
# Prepare learner/dropout config.
learner_config = learner_pb2.LearnerConfig()
learner_config.learning_rate_tuner.dropout.dropout_probability = 0.5
learner_config.learning_rate_tuner.dropout.learning_rate = 1.0
learner_config.num_classes = 2
tree_ensemble_handle = model_ops.tree_ensemble_variable(
stamp_token=0,
tree_ensemble_config=tree_ensemble_config.SerializeToString(),
name="empty")
resources.initialize_resources(resources.shared_resources()).run()
_, dropout_info_1 = self._get_predictions(
tree_ensemble_handle,
learner_config=learner_config.SerializeToString(),
apply_dropout=True,
apply_averaging=False,
center_bias=False,
reduce_dim=True)
_, dropout_info_2 = self._get_predictions(
tree_ensemble_handle,
learner_config=learner_config.SerializeToString(),
apply_dropout=True,
apply_averaging=False,
center_bias=False,
reduce_dim=True)
# Different seed.
_, dropout_info_3 = prediction_ops.gradient_trees_prediction(
tree_ensemble_handle,
112314, [self._dense_float_tensor],
[self._sparse_float_indices1, self._sparse_float_indices2],
[self._sparse_float_values1, self._sparse_float_values2],
[self._sparse_float_shape1, self._sparse_float_shape2],
[self._sparse_int_indices1], [self._sparse_int_values1],
[self._sparse_int_shape1],
learner_config=learner_config.SerializeToString(),
apply_dropout=True,
apply_averaging=False,
center_bias=False,
reduce_dim=True)
# First seed with centering bias.
_, dropout_info_4 = self._get_predictions(
tree_ensemble_handle,
learner_config=learner_config.SerializeToString(),
apply_dropout=True,
apply_averaging=False,
center_bias=True,
reduce_dim=True)
# The same seed returns the same results.
self.assertAllEqual(dropout_info_1.eval(), dropout_info_2.eval())
# Different seeds give diff results.
self.assertNotEqual(dropout_info_3.eval().shape,
dropout_info_2.eval().shape)
# With centering bias and the same seed does not give the same result.
self.assertNotEqual(dropout_info_4.eval(), dropout_info_1.eval())
# With centering bias has 1 less tree dropped (bias tree is not dropped).
self.assertEqual(
len(dropout_info_4.eval()[0]) + 1, len(dropout_info_1.eval()[0]))
