.. raw:: html
.. code:: python
%matplotlib inline
from mxnet import np, npx
from d2l import mxnet as d2l
npx.set_np()
def init_adadelta_states(feature_dim):
s_w, s_b = np.zeros((feature_dim, 1)), np.zeros(1)
delta_w, delta_b = np.zeros((feature_dim, 1)), np.zeros(1)
return ((s_w, delta_w), (s_b, delta_b))
def adadelta(params, states, hyperparams):
rho, eps = hyperparams['rho'], 1e-5
for p, (s, delta) in zip(params, states):
# In-place updates via [:]
s[:] = rho * s + (1 - rho) * np.square(p.grad)
g = (np.sqrt(delta + eps) / np.sqrt(s + eps)) * p.grad
p[:] -= g
delta[:] = rho * delta + (1 - rho) * g * g
.. raw:: html
.. raw:: html
.. code:: python
%matplotlib inline
import torch
from d2l import torch as d2l
def init_adadelta_states(feature_dim):
s_w, s_b = torch.zeros((feature_dim, 1)), torch.zeros(1)
delta_w, delta_b = torch.zeros((feature_dim, 1)), torch.zeros(1)
return ((s_w, delta_w), (s_b, delta_b))
def adadelta(params, states, hyperparams):
rho, eps = hyperparams['rho'], 1e-5
for p, (s, delta) in zip(params, states):
with torch.no_grad():
# In-place updates via [:]
s[:] = rho * s + (1 - rho) * torch.square(p.grad)
g = (torch.sqrt(delta + eps) / torch.sqrt(s + eps)) * p.grad
p[:] -= g
delta[:] = rho * delta + (1 - rho) * g * g
p.grad.data.zero_()
.. raw:: html
.. raw:: html
.. code:: python
%matplotlib inline
import tensorflow as tf
from d2l import tensorflow as d2l
def init_adadelta_states(feature_dim):
s_w = tf.Variable(tf.zeros((feature_dim, 1)))
s_b = tf.Variable(tf.zeros(1))
delta_w = tf.Variable(tf.zeros((feature_dim, 1)))
delta_b = tf.Variable(tf.zeros(1))
return ((s_w, delta_w), (s_b, delta_b))
def adadelta(params, grads, states, hyperparams):
rho, eps = hyperparams['rho'], 1e-5
for p, (s, delta), grad in zip(params, states, grads):
s[:].assign(rho * s + (1 - rho) * tf.math.square(grad))
g = (tf.math.sqrt(delta + eps) / tf.math.sqrt(s + eps)) * grad
p[:].assign(p - g)
delta[:].assign(rho * delta + (1 - rho) * g * g)
.. raw:: html
.. raw:: html
.. raw:: html
.. code:: python
data_iter, feature_dim = d2l.get_data_ch11(batch_size=10)
d2l.train_ch11(adadelta, init_adadelta_states(feature_dim),
{'rho': 0.9}, data_iter, feature_dim);
.. parsed-literal::
:class: output
loss: 0.244, 0.120 sec/epoch
.. figure:: output_adadelta_0b41cb_15_1.svg
.. raw:: html
.. raw:: html
.. code:: python
data_iter, feature_dim = d2l.get_data_ch11(batch_size=10)
d2l.train_ch11(adadelta, init_adadelta_states(feature_dim),
{'rho': 0.9}, data_iter, feature_dim);
.. parsed-literal::
:class: output
loss: 0.244, 0.016 sec/epoch
.. figure:: output_adadelta_0b41cb_18_1.svg
.. raw:: html
.. raw:: html
.. code:: python
data_iter, feature_dim = d2l.get_data_ch11(batch_size=10)
d2l.train_ch11(adadelta, init_adadelta_states(feature_dim),
{'rho': 0.9}, data_iter, feature_dim);
.. parsed-literal::
:class: output
loss: 0.243, 0.151 sec/epoch
.. figure:: output_adadelta_0b41cb_21_1.svg
.. raw:: html
.. raw:: html
.. raw:: html
.. code:: python
d2l.train_concise_ch11('adadelta', {'rho': 0.9}, data_iter)
.. parsed-literal::
:class: output
loss: 0.245, 0.120 sec/epoch
.. figure:: output_adadelta_0b41cb_27_1.svg
.. raw:: html
.. raw:: html
.. code:: python
trainer = torch.optim.Adadelta
d2l.train_concise_ch11(trainer, {'rho': 0.9}, data_iter)
.. parsed-literal::
:class: output
loss: 0.244, 0.015 sec/epoch
.. figure:: output_adadelta_0b41cb_30_1.svg
.. raw:: html
.. raw:: html
.. code:: python
# adadelta is not converging at default learning rate
# but it's converging at lr = 5.0
trainer = tf.keras.optimizers.Adadelta
d2l.train_concise_ch11(trainer, {'learning_rate':5.0, 'rho': 0.9}, data_iter)
.. parsed-literal::
:class: output
loss: 0.244, 0.105 sec/epoch
.. figure:: output_adadelta_0b41cb_33_1.svg
.. raw:: html
.. raw:: html