Source code for sugartensor.sg_optimize

from __future__ import absolute_import
import tensorflow as tf

__author__ = 'buriburisuri@gmail.com'


# noinspection PyAbstractClass
class AdaMaxOptimizer(tf.train.Optimizer):
    r"""Optimizer that implements the Adamax algorithm.
    See [Kingma et. al., 2014](http://arxiv.org/abs/1412.6980)
    ([pdf](http://arxiv.org/pdf/1412.6980.pdf)).

    excerpted from https://github.com/openai/iaf/blob/master/tf_utils/adamax.py

    @@__init__
    """

    def __init__(self, learning_rate=0.001, beta1=0.9, beta2=0.999, use_locking=False, name="Adamax"):
        super(AdaMaxOptimizer, self).__init__(use_locking, name)
        self._lr = learning_rate
        self._beta1 = beta1
        self._beta2 = beta2

        # Tensor versions of the constructor arguments, created in _prepare().
        self._lr_t = None
        self._beta1_t = None
        self._beta2_t = None

    def _prepare(self):
        self._lr_t = tf.convert_to_tensor(self._lr, name="learning_rate")
        self._beta1_t = tf.convert_to_tensor(self._beta1, name="beta1")
        self._beta2_t = tf.convert_to_tensor(self._beta2, name="beta2")

    def _create_slots(self, var_list):
        # Create slots for the first and second moments.
        for v in var_list:
            self._zeros_slot(v, "m", self._name)
            self._zeros_slot(v, "v", self._name)

    def _apply_dense(self, grad, var):
        lr_t = tf.cast(self._lr_t, var.dtype.base_dtype)
        beta1_t = tf.cast(self._beta1_t, var.dtype.base_dtype)
        beta2_t = tf.cast(self._beta2_t, var.dtype.base_dtype)
        if var.dtype.base_dtype == tf.float16:
            eps = 1e-7  # Can't use 1e-8 due to underflow -- not sure if it makes a big difference.
        else:
            eps = 1e-8

        v = self.get_slot(var, "v")
        v_t = v.assign(beta1_t * v + (1. - beta1_t) * grad)
        m = self.get_slot(var, "m")
        m_t = m.assign(tf.maximum(beta2_t * m + eps, tf.abs(grad)))
        g_t = v_t / m_t

        var_update = tf.assign_sub(var, lr_t * g_t)
        return tf.group(*[var_update, m_t, v_t])

    def _apply_sparse(self, grad, var):
        return self._apply_dense(grad, var)


# noinspection PyAbstractClass
class MaxPropOptimizer(tf.train.Optimizer):
    r"""Optimizer that implements the MaxProp algorithm by buriburisuri@gmail.com.
    """
    def __init__(self, learning_rate=0.001, beta2=0.999, use_locking=False, name="MaxProp"):
        super(MaxPropOptimizer, self).__init__(use_locking, name)
        self._lr = learning_rate
        self._beta2 = beta2

        # Tensor versions of the constructor arguments, created in _prepare().
        self._lr_t = None
        self._beta2_t = None

    def _prepare(self):
        self._lr_t = tf.convert_to_tensor(self._lr, name="learning_rate")
        self._beta2_t = tf.convert_to_tensor(self._beta2, name="beta2")

    def _create_slots(self, var_list):
        # Create slots for the second moments.
        for v in var_list:
            self._zeros_slot(v, "m", self._name)

    def _apply_dense(self, grad, var):
        lr_t = tf.cast(self._lr_t, var.dtype.base_dtype)
        beta2_t = tf.cast(self._beta2_t, var.dtype.base_dtype)
        if var.dtype.base_dtype == tf.float16:
            eps = 1e-7  # Can't use 1e-8 due to underflow -- not sure if it makes a big difference.
        else:
            eps = 1e-8

        m = self.get_slot(var, "m")
        m_t = m.assign(tf.maximum(beta2_t * m + eps, tf.abs(grad)))
        g_t = grad / m_t

        var_update = tf.assign_sub(var, lr_t * g_t)
        return tf.group(*[var_update, m_t])

    def _apply_sparse(self, grad, var):
        return self._apply_dense(grad, var)