-
Notifications
You must be signed in to change notification settings - Fork 77
/
attention_lstm.py
executable file
·219 lines (177 loc) · 8.77 KB
/
attention_lstm.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
#!/usr/bin/python
# -*- coding: utf-8 -*-
# coding=utf-8
from keras.layers import activations, Wrapper
from keras.engine import InputSpec
from keras import backend as K
from keras.layers import LSTM
class AttentionLSTM(LSTM):
def __init__(self, output_dim, attention_vec, **kwargs):
self.attention_vec = attention_vec
super(AttentionLSTM, self).__init__(output_dim, **kwargs)
def build(self, input_shape):
'''
this method initializes all of the weight matrices we need for the attentional component
:param input_shape:
:return:
'''
super(AttentionLSTM, self).build(input_shape)
assert hasattr(self.attention_vec, '_keras_shape')
attention_dim = self.attention_vec._keras_shape[1]
self.U_a = self.inner_init((self.output_dim, self.output_dim),
name='{}_U_a'.format(self.name))
self.b_a = K.zeros((self.output_dim,), name='{}_b_a'.format(self.name))
self.U_m = self.inner_init((attention_dim, self.output_dim),
name='{}_U_m'.format(self.name))
self.b_m = K.zeros((self.output_dim,), name='{}_b_m'.format(self.name))
self.U_s = self.inner_init((self.output_dim, self.output_dim),
name='{}_U_s'.format(self.name))
self.b_s = K.zeros((self.output_dim,), name='{}_b_s'.format(self.name))
self.trainable_weights += [self.U_a, self.U_m, self.U_s,
self.b_a, self.b_m, self.b_s]
if self.initial_weights is not None:
self.set_weights(self.initial_weights)
del self.initial_weights
def step(self, x, states):
'''
This method is used by the RNN superclass, and tells the function what to do on each timestep.
:param x:
:param states:
:return:
'''
h, [h, c] = super(AttentionLSTM, self).step(x, states)
attention = states[4]
m = K.tanh(K.dot(h, self.U_a) + attention + self.b_a)
s = K.exp(K.dot(m, self.U_s) + self.b_s)
h = h * s
return h, [h, c]
def get_constants(self, x):
'''
This method is used by the LSTM superclass to define components outside of the step function,
so that they don’t need to be recomputed every time step.
:param x:
:return:
'''
constants = super(AttentionLSTM, self).get_constants(x)
constants.append(K.dot(self.attention_vec, self.U_m) + self.b_m)
return constants
class AttentionLSTM_t(LSTM):
def __init__(self, output_dim, attn_activation='tanh', **kwargs):
self.attn_activation = activations.get(attn_activation)
super(AttentionLSTM_t, self).__init__(output_dim, **kwargs)
def build(self, input_shape):
super(AttentionLSTM_t, self).build(input_shape)
# assert hasattr(self.attention_vec, '_keras_shape')
# attention_dim = self.attention_vec._keras_shape[1]
self.U_a = self.inner_init((self.output_dim, self.output_dim),
name='{}_U_a'.format(self.name))
self.b_a = K.zeros((self.output_dim,), name='{}_b_a'.format(self.name))
self.U_s = self.inner_init((self.output_dim, self.output_dim),
name='{}_U_s'.format(self.name))
self.b_s = K.zeros((self.output_dim,), name='{}_b_s'.format(self.name))
self.trainable_weights += [self.U_a, self.b_a, self.U_s, self.b_s]
if self.initial_weights is not None:
self.set_weights(self.initial_weights)
del self.initial_weights
def step(self, x, states):
'''
This method is used by the RNN superclass, and tells the function what to do on each timestep.
:param x:
:param states:
:return:
'''
h, [h, c] = super(AttentionLSTM_t, self).step(x, states)
m = K.tanh(K.dot(h, self.U_a) + self.b_a)
alpha = K.exp(K.dot(m, self.U_s) + self.b_s)
h = h * alpha
return h, [h, c]
def get_constants(self, x):
constants = super(AttentionLSTM_t, self).get_constants(x)
return constants
class AttentionLSTMWrapper(Wrapper):
def __init__(self, layer, attn_activation='tanh', single_attention_param=False, **kwargs):
assert isinstance(layer, LSTM)
self.supports_masking = True
self.attn_activation = activations.get(attn_activation)
self.single_attention_param = single_attention_param
super(AttentionLSTMWrapper, self).__init__(layer, **kwargs)
def build(self, input_shape):
assert len(input_shape) >= 3
self.input_spec = [InputSpec(shape=input_shape)]
if not self.layer.built:
self.layer.build(input_shape)
self.layer.built = True
super(AttentionLSTMWrapper, self).build()
self.U_a = self.layer.inner_init((self.layer.output_dim, self.layer.output_dim),
name='{}_U_a'.format(self.name))
self.b_a = K.zeros((self.layer.output_dim,), name='{}_b_a'.format(self.name))
if self.single_attention_param:
self.U_s = self.layer.inner_init((self.layer.output_dim, 1), name='{}_U_s'.format(self.name))
self.b_s = K.zeros((1,), name='{}_b_s'.format(self.name))
else:
self.U_s = self.layer.inner_init((self.layer.output_dim, self.layer.output_dim),
name='{}_U_s'.format(self.name))
self.b_s = K.zeros((self.layer.output_dim,), name='{}_b_s'.format(self.name))
self.trainable_weights = [self.U_a, self.U_s, self.b_a, self.b_s]
def get_output_shape_for(self, input_shape):
return self.layer.get_output_shape_for(input_shape)
def step(self, x, states):
h, [h, c] = self.layer.step(x, states)
m = self.attn_activation(h)
s = K.softmax(K.dot(m, self.U_s))
if self.single_attention_param:
h = h * K.repeat_elements(s, self.layer.output_dim, axis=1)
else:
h = h * s
# attention = states[4]
#
# m = self.attn_activation(K.dot(h, self.U_a) * attention + self.b_a)
# s = K.sigmoid(K.dot(m, self.U_s) + self.b_s)
#
# if self.single_attention_param:
# h = h * K.repeat_elements(s, self.layer.output_dim, axis=1)
# else:
# h = h * s
return h, [h, c]
def get_constants(self, x):
constants = self.layer.get_constants(x)
return constants
def call(self, x, mask=None):
# input shape: (nb_samples, time (padded with zeros), input_dim)
# note that the .build() method of subclasses MUST define
# self.input_spec with a complete input shape.
input_shape = self.input_spec[0].shape
if K._BACKEND == 'tensorflow':
if not input_shape[1]:
raise Exception('When using TensorFlow, you should define '
'explicitly the number of timesteps of '
'your sequences.\n'
'If your first layer is an Embedding, '
'make sure to pass it an "input_length" '
'argument. Otherwise, make sure '
'the first layer has '
'an "input_shape" or "batch_input_shape" '
'argument, including the time axis. '
'Found input shape at layer ' + self.name +
': ' + str(input_shape))
if self.layer.stateful:
initial_states = self.layer.states
else:
initial_states = self.layer.get_initial_states(x)
constants = self.get_constants(x)
preprocessed_input = self.layer.preprocess_input(x)
last_output, outputs, states = K.rnn(self.step, preprocessed_input,
initial_states,
go_backwards=self.layer.go_backwards,
mask=mask,
constants=constants,
unroll=self.layer.unroll,
input_length=input_shape[1])
if self.layer.stateful:
self.updates = []
for i in range(len(states)):
self.updates.append((self.layer.states[i], states[i]))
if self.layer.return_sequences:
return outputs
else:
return last_output