324 lines
12 KiB
Python
324 lines
12 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.
|
|
# ==============================================================================
|
|
"""One-line documentation for rmsprop module.
|
|
|
|
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_grad{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 tensorflow.python.framework import ops
|
|
from tensorflow.python.ops import array_ops
|
|
from tensorflow.python.ops import gen_training_ops
|
|
from tensorflow.python.ops import init_ops
|
|
from tensorflow.python.ops import math_ops
|
|
from tensorflow.python.training import optimizer
|
|
from tensorflow.python.util.tf_export import tf_export
|
|
|
|
|
|
@tf_export(v1=["train.RMSPropOptimizer"])
|
|
class RMSPropOptimizer(optimizer.Optimizer):
|
|
"""Optimizer that implements the RMSProp algorithm (Tielemans et al.
|
|
|
|
2012).
|
|
|
|
References:
|
|
Coursera slide 29:
|
|
Hinton, 2012
|
|
([pdf](http://www.cs.toronto.edu/~tijmen/csc321/slides/lecture_slides_lec6.pdf))
|
|
|
|
@compatibility(TF2)
|
|
tf.compat.v1.train.RMSPropOptimizer is compatible with eager mode and
|
|
`tf.function`.
|
|
When eager execution is enabled, `learning_rate`, `decay`, `momentum`,
|
|
and `epsilon` can each be a callable that
|
|
takes no arguments and returns the actual value to use. This can be useful
|
|
for changing these values across different invocations of optimizer
|
|
functions.
|
|
|
|
To switch to native TF2 style, use [`tf.keras.optimizers.RMSprop`]
|
|
(https://www.tensorflow.org/api_docs/python/tf/keras/optimizers/RMSprop)
|
|
instead. Please notice that due to the implementation differences,
|
|
`tf.keras.optimizers.RMSprop` and
|
|
`tf.compat.v1.train.RMSPropOptimizer` may have slight differences in
|
|
floating point numerics even though the formula used for the variable
|
|
updates still matches.
|
|
|
|
#### Structural mapping to native TF2
|
|
|
|
Before:
|
|
|
|
```python
|
|
optimizer = tf.compat.v1.train.RMSPropOptimizer(
|
|
learning_rate=learning_rate,
|
|
decay=decay,
|
|
momentum=momentum,
|
|
epsilon=epsilon)
|
|
```
|
|
|
|
After:
|
|
|
|
```python
|
|
optimizer = tf.keras.optimizers.RMSprop(
|
|
learning_rate=learning_rate,
|
|
rho=decay,
|
|
momentum=momentum,
|
|
epsilon=epsilon)
|
|
```
|
|
|
|
#### How to map arguments
|
|
| TF1 Arg Name | TF2 Arg Name | Note |
|
|
| ------------------ | ------------- | ------------------------------- |
|
|
| `learning_rate` | `learning_rate`| Be careful of setting |
|
|
: : : learning_rate tensor value computed from the global step. :
|
|
: : : In TF1 this was usually meant to imply a dynamic learning rate and :
|
|
: : : would recompute in each step. In TF2 (eager + function) it will :
|
|
: : : treat it as a scalar value that only gets computed once instead of :
|
|
: : : a symbolic placeholder to be computed each time. :
|
|
| `decay` | `rho` | - |
|
|
| `momentum` | `momentum` | - |
|
|
| `epsilon` | `epsilon` | Default value is 1e-10 in TF1, |
|
|
: : : but 1e-07 in TF2. :
|
|
| `use_locking` | - | Not applicable in TF2. |
|
|
|
|
#### Before & after usage example
|
|
Before:
|
|
|
|
```python
|
|
x = tf.Variable([1,2,3], dtype=tf.float32)
|
|
grad = tf.constant([0.1, 0.2, 0.3])
|
|
optimizer = tf.compat.v1.train.RMSPropOptimizer(learning_rate=0.001)
|
|
optimizer.apply_gradients(zip([grad], [x]))
|
|
```
|
|
|
|
After:
|
|
|
|
```python
|
|
x = tf.Variable([1,2,3], dtype=tf.float32)
|
|
grad = tf.constant([0.1, 0.2, 0.3])
|
|
optimizer = tf.keras.optimizers.RMSprop(learning_rate=0.001)
|
|
optimizer.apply_gradients(zip([grad], [x]))
|
|
```
|
|
|
|
@end_compatibility
|
|
"""
|
|
|
|
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.
|
|
|
|
Args:
|
|
learning_rate: A Tensor or a floating point value. The learning rate.
|
|
decay: Discounting factor for the history/coming gradient
|
|
momentum: A scalar tensor.
|
|
epsilon: 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._learning_rate = learning_rate
|
|
self._decay = decay
|
|
self._momentum = momentum
|
|
self._epsilon = epsilon
|
|
self._centered = centered
|
|
|
|
# Tensors for learning rate and momentum. Created in _prepare.
|
|
self._learning_rate_tensor = None
|
|
self._decay_tensor = None
|
|
self._momentum_tensor = None
|
|
self._epsilon_tensor = None
|
|
|
|
def _create_slots(self, var_list):
|
|
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)
|
|
self._get_or_make_slot_with_initializer(v, init_rms, v.get_shape(),
|
|
v.dtype.base_dtype, "rms",
|
|
self._name)
|
|
if self._centered:
|
|
self._zeros_slot(v, "mg", self._name)
|
|
self._zeros_slot(v, "momentum", self._name)
|
|
|
|
def _prepare(self):
|
|
lr = self._call_if_callable(self._learning_rate)
|
|
decay = self._call_if_callable(self._decay)
|
|
momentum = self._call_if_callable(self._momentum)
|
|
epsilon = self._call_if_callable(self._epsilon)
|
|
|
|
self._learning_rate_tensor = ops.convert_to_tensor(lr, name="learning_rate")
|
|
self._decay_tensor = ops.convert_to_tensor(decay, name="decay")
|
|
self._momentum_tensor = ops.convert_to_tensor(momentum, name="momentum")
|
|
self._epsilon_tensor = ops.convert_to_tensor(epsilon, name="epsilon")
|
|
|
|
def _apply_dense(self, grad, var):
|
|
rms = self.get_slot(var, "rms")
|
|
mom = self.get_slot(var, "momentum")
|
|
if self._centered:
|
|
mg = self.get_slot(var, "mg")
|
|
return gen_training_ops.apply_centered_rms_prop(
|
|
var,
|
|
mg,
|
|
rms,
|
|
mom,
|
|
math_ops.cast(self._learning_rate_tensor, var.dtype.base_dtype),
|
|
math_ops.cast(self._decay_tensor, var.dtype.base_dtype),
|
|
math_ops.cast(self._momentum_tensor, var.dtype.base_dtype),
|
|
math_ops.cast(self._epsilon_tensor, var.dtype.base_dtype),
|
|
grad,
|
|
use_locking=self._use_locking).op
|
|
else:
|
|
return gen_training_ops.apply_rms_prop(
|
|
var,
|
|
rms,
|
|
mom,
|
|
math_ops.cast(self._learning_rate_tensor, var.dtype.base_dtype),
|
|
math_ops.cast(self._decay_tensor, var.dtype.base_dtype),
|
|
math_ops.cast(self._momentum_tensor, var.dtype.base_dtype),
|
|
math_ops.cast(self._epsilon_tensor, var.dtype.base_dtype),
|
|
grad,
|
|
use_locking=self._use_locking).op
|
|
|
|
def _resource_apply_dense(self, grad, var):
|
|
rms = self.get_slot(var, "rms")
|
|
mom = self.get_slot(var, "momentum")
|
|
if self._centered:
|
|
mg = self.get_slot(var, "mg")
|
|
return gen_training_ops.resource_apply_centered_rms_prop(
|
|
var.handle,
|
|
mg.handle,
|
|
rms.handle,
|
|
mom.handle,
|
|
math_ops.cast(self._learning_rate_tensor, grad.dtype.base_dtype),
|
|
math_ops.cast(self._decay_tensor, grad.dtype.base_dtype),
|
|
math_ops.cast(self._momentum_tensor, grad.dtype.base_dtype),
|
|
math_ops.cast(self._epsilon_tensor, grad.dtype.base_dtype),
|
|
grad,
|
|
use_locking=self._use_locking)
|
|
else:
|
|
return gen_training_ops.resource_apply_rms_prop(
|
|
var.handle,
|
|
rms.handle,
|
|
mom.handle,
|
|
math_ops.cast(self._learning_rate_tensor, grad.dtype.base_dtype),
|
|
math_ops.cast(self._decay_tensor, grad.dtype.base_dtype),
|
|
math_ops.cast(self._momentum_tensor, grad.dtype.base_dtype),
|
|
math_ops.cast(self._epsilon_tensor, grad.dtype.base_dtype),
|
|
grad,
|
|
use_locking=self._use_locking)
|
|
|
|
def _apply_sparse(self, grad, var):
|
|
rms = self.get_slot(var, "rms")
|
|
mom = self.get_slot(var, "momentum")
|
|
if self._centered:
|
|
mg = self.get_slot(var, "mg")
|
|
return gen_training_ops.sparse_apply_centered_rms_prop(
|
|
var,
|
|
mg,
|
|
rms,
|
|
mom,
|
|
math_ops.cast(self._learning_rate_tensor, var.dtype.base_dtype),
|
|
math_ops.cast(self._decay_tensor, var.dtype.base_dtype),
|
|
math_ops.cast(self._momentum_tensor, var.dtype.base_dtype),
|
|
math_ops.cast(self._epsilon_tensor, var.dtype.base_dtype),
|
|
grad.values,
|
|
grad.indices,
|
|
use_locking=self._use_locking)
|
|
else:
|
|
return gen_training_ops.sparse_apply_rms_prop(
|
|
var,
|
|
rms,
|
|
mom,
|
|
math_ops.cast(self._learning_rate_tensor, var.dtype.base_dtype),
|
|
math_ops.cast(self._decay_tensor, var.dtype.base_dtype),
|
|
math_ops.cast(self._momentum_tensor, var.dtype.base_dtype),
|
|
math_ops.cast(self._epsilon_tensor, var.dtype.base_dtype),
|
|
grad.values,
|
|
grad.indices,
|
|
use_locking=self._use_locking)
|
|
|
|
def _resource_apply_sparse(self, grad, var, indices):
|
|
rms = self.get_slot(var, "rms")
|
|
mom = self.get_slot(var, "momentum")
|
|
if self._centered:
|
|
mg = self.get_slot(var, "mg")
|
|
return gen_training_ops.resource_sparse_apply_centered_rms_prop(
|
|
var.handle,
|
|
mg.handle,
|
|
rms.handle,
|
|
mom.handle,
|
|
math_ops.cast(self._learning_rate_tensor, grad.dtype),
|
|
math_ops.cast(self._decay_tensor, grad.dtype),
|
|
math_ops.cast(self._momentum_tensor, grad.dtype),
|
|
math_ops.cast(self._epsilon_tensor, grad.dtype),
|
|
grad,
|
|
indices,
|
|
use_locking=self._use_locking)
|
|
else:
|
|
return gen_training_ops.resource_sparse_apply_rms_prop(
|
|
var.handle,
|
|
rms.handle,
|
|
mom.handle,
|
|
math_ops.cast(self._learning_rate_tensor, grad.dtype),
|
|
math_ops.cast(self._decay_tensor, grad.dtype),
|
|
math_ops.cast(self._momentum_tensor, grad.dtype),
|
|
math_ops.cast(self._epsilon_tensor, grad.dtype),
|
|
grad,
|
|
indices,
|
|
use_locking=self._use_locking)
|