233 lines
9 KiB
Python
233 lines
9 KiB
Python
# Copyright 2015 The TensorFlow Authors. All Rights Reserved.
|
|
#
|
|
# Licensed under the Apache License, Version 2.0 (the "License");
|
|
# you may not use this file except in compliance with the License.
|
|
# You may obtain a copy of the License at
|
|
#
|
|
# http://www.apache.org/licenses/LICENSE-2.0
|
|
#
|
|
# Unless required by applicable law or agreed to in writing, software
|
|
# distributed under the License is distributed on an "AS IS" BASIS,
|
|
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
# See the License for the specific language governing permissions and
|
|
# limitations under the License.
|
|
# ==============================================================================
|
|
"""RMSprop optimizer for Tensorflow.
|
|
|
|
rmsprop algorithm [tieleman2012rmsprop]
|
|
|
|
A detailed description of rmsprop.
|
|
|
|
- maintain a moving (discounted) average of the square of gradients
|
|
- divide gradient by the root of this average
|
|
|
|
mean_square = decay * mean_square{t-1} + (1-decay) * gradient ** 2
|
|
mom = momentum * mom{t-1} + learning_rate * g_t / sqrt(mean_square + epsilon)
|
|
delta = - mom
|
|
|
|
This implementation of RMSProp uses plain momentum, not Nesterov momentum.
|
|
|
|
The centered version additionally maintains a moving (discounted) average of the
|
|
gradients, and uses that average to estimate the variance:
|
|
|
|
mean_grad = decay * mean_square{t-1} + (1-decay) * gradient
|
|
mean_square = decay * mean_square{t-1} + (1-decay) * gradient ** 2
|
|
mom = momentum * mom{t-1} + learning_rate * g_t /
|
|
sqrt(mean_square - mean_grad**2 + epsilon)
|
|
delta = - mom
|
|
"""
|
|
|
|
from __future__ import absolute_import
|
|
from __future__ import division
|
|
from __future__ import print_function
|
|
|
|
from tensorflow.contrib.optimizer_v2 import optimizer_v2
|
|
from tensorflow.python.ops import array_ops
|
|
from tensorflow.python.ops import init_ops
|
|
|
|
from tensorflow.python.training import training_ops
|
|
|
|
|
|
class RMSPropOptimizer(optimizer_v2.OptimizerV2):
|
|
"""Optimizer that implements the RMSProp algorithm.
|
|
|
|
See the
|
|
[paper](http://www.cs.toronto.edu/~tijmen/csc321/slides/lecture_slides_lec6.pdf).
|
|
"""
|
|
|
|
def __init__(self,
|
|
learning_rate,
|
|
decay=0.9,
|
|
momentum=0.0,
|
|
epsilon=1e-10,
|
|
use_locking=False,
|
|
centered=False,
|
|
name="RMSProp"):
|
|
"""Construct a new RMSProp optimizer.
|
|
|
|
Note that in the dense implementation of this algorithm, variables and their
|
|
corresponding accumulators (momentum, gradient moving average, square
|
|
gradient moving average) will be updated even if the gradient is zero
|
|
(i.e. accumulators will decay, momentum will be applied). The sparse
|
|
implementation (used when the gradient is an `IndexedSlices` object,
|
|
typically because of `tf.gather` or an embedding lookup in the forward pass)
|
|
will not update variable slices or their accumulators unless those slices
|
|
were used in the forward pass (nor is there an "eventual" correction to
|
|
account for these omitted updates). This leads to more efficient updates for
|
|
large embedding lookup tables (where most of the slices are not accessed in
|
|
a particular graph execution), but differs from the published algorithm.
|
|
|
|
Some of the args below are hyperparameters, where a hyperparameter is
|
|
defined as a scalar Tensor, a regular Python value or a callable (which
|
|
will be evaluated when `apply_gradients` is called) returning a scalar
|
|
Tensor or a Python value.
|
|
|
|
Args:
|
|
learning_rate: A float hyperparameter. The learning rate.
|
|
decay: A float hyperparameter. Discounting factor for the history/coming
|
|
gradient.
|
|
momentum: A float hyperparameter.
|
|
epsilon: A float hyperparameter. Small value to avoid zero denominator.
|
|
use_locking: If True use locks for update operation.
|
|
centered: If True, gradients are normalized by the estimated variance of
|
|
the gradient; if False, by the uncentered second moment. Setting this to
|
|
True may help with training, but is slightly more expensive in terms of
|
|
computation and memory. Defaults to False.
|
|
name: Optional name prefix for the operations created when applying
|
|
gradients. Defaults to "RMSProp".
|
|
"""
|
|
super(RMSPropOptimizer, self).__init__(use_locking, name)
|
|
self._set_hyper("learning_rate", learning_rate)
|
|
self._set_hyper("decay", decay)
|
|
self._set_hyper("momentum", momentum)
|
|
self._set_hyper("epsilon", epsilon)
|
|
|
|
self._centered = centered
|
|
|
|
def _create_vars(self, var_list, state):
|
|
for v in var_list:
|
|
if v.get_shape().is_fully_defined():
|
|
init_rms = init_ops.ones_initializer(dtype=v.dtype.base_dtype)
|
|
else:
|
|
init_rms = array_ops.ones_like(v)
|
|
state.create_slot_with_initializer(v, init_rms, v.get_shape(),
|
|
v.dtype.base_dtype, "rms")
|
|
if self._centered:
|
|
state.zeros_slot(v, "mg")
|
|
state.zeros_slot(v, "momentum")
|
|
|
|
def _apply_dense(self, grad, var, state):
|
|
rms = state.get_slot(var, "rms")
|
|
mom = state.get_slot(var, "momentum")
|
|
if self._centered:
|
|
mg = state.get_slot(var, "mg")
|
|
return training_ops.apply_centered_rms_prop(
|
|
var,
|
|
mg,
|
|
rms,
|
|
mom,
|
|
state.get_hyper("learning_rate", var.dtype.base_dtype),
|
|
state.get_hyper("decay", var.dtype.base_dtype),
|
|
state.get_hyper("momentum", var.dtype.base_dtype),
|
|
state.get_hyper("epsilon", var.dtype.base_dtype),
|
|
grad,
|
|
use_locking=self._use_locking).op
|
|
else:
|
|
return training_ops.apply_rms_prop(
|
|
var,
|
|
rms,
|
|
mom,
|
|
state.get_hyper("learning_rate", var.dtype.base_dtype),
|
|
state.get_hyper("decay", var.dtype.base_dtype),
|
|
state.get_hyper("momentum", var.dtype.base_dtype),
|
|
state.get_hyper("epsilon", var.dtype.base_dtype),
|
|
grad,
|
|
use_locking=self._use_locking).op
|
|
|
|
def _resource_apply_dense(self, grad, var, state):
|
|
rms = state.get_slot(var, "rms")
|
|
mom = state.get_slot(var, "momentum")
|
|
if self._centered:
|
|
mg = state.get_slot(var, "mg")
|
|
return training_ops.resource_apply_centered_rms_prop(
|
|
var.handle,
|
|
mg.handle,
|
|
rms.handle,
|
|
mom.handle,
|
|
state.get_hyper("learning_rate", var.dtype.base_dtype),
|
|
state.get_hyper("decay", var.dtype.base_dtype),
|
|
state.get_hyper("momentum", var.dtype.base_dtype),
|
|
state.get_hyper("epsilon", var.dtype.base_dtype),
|
|
grad,
|
|
use_locking=self._use_locking)
|
|
else:
|
|
return training_ops.resource_apply_rms_prop(
|
|
var.handle,
|
|
rms.handle,
|
|
mom.handle,
|
|
state.get_hyper("learning_rate", var.dtype.base_dtype),
|
|
state.get_hyper("decay", var.dtype.base_dtype),
|
|
state.get_hyper("momentum", var.dtype.base_dtype),
|
|
state.get_hyper("epsilon", var.dtype.base_dtype),
|
|
grad,
|
|
use_locking=self._use_locking)
|
|
|
|
def _apply_sparse(self, grad, var, state):
|
|
rms = state.get_slot(var, "rms")
|
|
mom = state.get_slot(var, "momentum")
|
|
if self._centered:
|
|
mg = state.get_slot(var, "mg")
|
|
return training_ops.sparse_apply_centered_rms_prop(
|
|
var,
|
|
mg,
|
|
rms,
|
|
mom,
|
|
state.get_hyper("learning_rate", var.dtype.base_dtype),
|
|
state.get_hyper("decay", var.dtype.base_dtype),
|
|
state.get_hyper("momentum", var.dtype.base_dtype),
|
|
state.get_hyper("epsilon", var.dtype.base_dtype),
|
|
grad.values,
|
|
grad.indices,
|
|
use_locking=self._use_locking)
|
|
else:
|
|
return training_ops.sparse_apply_rms_prop(
|
|
var,
|
|
rms,
|
|
mom,
|
|
state.get_hyper("learning_rate", var.dtype.base_dtype),
|
|
state.get_hyper("decay", var.dtype.base_dtype),
|
|
state.get_hyper("momentum", var.dtype.base_dtype),
|
|
state.get_hyper("epsilon", var.dtype.base_dtype),
|
|
grad.values,
|
|
grad.indices,
|
|
use_locking=self._use_locking)
|
|
|
|
def _resource_apply_sparse(self, grad, var, indices, state):
|
|
rms = state.get_slot(var, "rms")
|
|
mom = state.get_slot(var, "momentum")
|
|
if self._centered:
|
|
mg = self.get_slot(var, "mg")
|
|
return training_ops.resource_sparse_apply_centered_rms_prop(
|
|
var.handle,
|
|
mg.handle,
|
|
rms.handle,
|
|
mom.handle,
|
|
state.get_hyper("learning_rate", var.dtype.base_dtype),
|
|
state.get_hyper("decay", var.dtype.base_dtype),
|
|
state.get_hyper("momentum", var.dtype.base_dtype),
|
|
state.get_hyper("epsilon", var.dtype.base_dtype),
|
|
grad,
|
|
indices,
|
|
use_locking=self._use_locking)
|
|
else:
|
|
return training_ops.resource_sparse_apply_rms_prop(
|
|
var.handle,
|
|
rms.handle,
|
|
mom.handle,
|
|
state.get_hyper("learning_rate", var.dtype.base_dtype),
|
|
state.get_hyper("decay", var.dtype.base_dtype),
|
|
state.get_hyper("momentum", var.dtype.base_dtype),
|
|
state.get_hyper("epsilon", var.dtype.base_dtype),
|
|
grad,
|
|
indices,
|
|
use_locking=self._use_locking)
|