-
Notifications
You must be signed in to change notification settings - Fork 0
/
normal_lstm.py
160 lines (143 loc) · 7.4 KB
/
normal_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
import sys
import mxnet as mx
import numpy as np
from collections import namedtuple
from copy import copy
LSTMState = namedtuple("LSTMState", ["c", "h"])
LSTMParam = namedtuple("LSTMParam", ["i2h_weight", "i2h_bias",
"h2h_weight", "h2h_bias"])
HyperPara = namedtuple('HyperPara', ['num_lstm_layer', 'seq_len', 'input_size',
'num_hidden', 'num_embed', 'num_label',
'dropout'])
def lstm(num_hidden, indata, prev_state, param, seqidx, layeridx, dropout=0.):
"""LSTM Cell symbol"""
if dropout > 0.:
indata = mx.sym.Dropout(data=indata, p=dropout)
i2h = mx.sym.FullyConnected(data=indata,
weight=param.i2h_weight,
bias=param.i2h_bias,
num_hidden=num_hidden * 4,
name="t%d_l%d_i2h" % (seqidx, layeridx))
h2h = mx.sym.FullyConnected(data=prev_state.h,
weight=param.h2h_weight,
bias=param.h2h_bias,
num_hidden=num_hidden * 4,
name="t%d_l%d_h2h" % (seqidx, layeridx))
gates = i2h + h2h
slice_gates = mx.sym.SliceChannel(gates, num_outputs=4,
name="t%d_l%d_slice" % (seqidx, layeridx))
in_gate = mx.sym.Activation(slice_gates[0], act_type="sigmoid")
in_transform = mx.sym.Activation(slice_gates[1], act_type="tanh")
forget_gate = mx.sym.Activation(slice_gates[2], act_type="sigmoid")
out_gate = mx.sym.Activation(slice_gates[3], act_type="sigmoid")
next_c = (forget_gate * prev_state.c) + (in_gate * in_transform)
next_h = out_gate * mx.sym.Activation(next_c, act_type="tanh")
return LSTMState(c=next_c, h=next_h)
def lstm_encoder(data_wv, seq_len, num_lstm_layer, input_size, num_hidden, num_embed, dropout=0.):
#multilayer lstm
param_cells = []
last_states = []
for i in range(num_lstm_layer):
with mx.AttrScope(ctx_group='enc_layers'):
param_cells.append(LSTMParam(i2h_weight=mx.sym.Variable("enc_l%d_i2h_weight" % i),
i2h_bias=mx.sym.Variable("enc_l%d_i2h_bias" % i),
h2h_weight=mx.sym.Variable("enc_l%d_h2h_weight" % i),
h2h_bias=mx.sym.Variable("enc_l%d_h2h_bias" % i)))
state = LSTMState(c=mx.sym.Variable("enc_l%d_init_c" % i),
h=mx.sym.Variable("enc_l%d_init_h" % i))
last_states.append(state)
assert(len(last_states) == num_lstm_layer)
for seqidx in range(seq_len):
hidden = data_wv[seqidx]
# stack LSTM
for i in range(num_lstm_layer):
if i == 0:
dp_ratio = 0.
else:
dp_ratio = dropout
with mx.AttrScope(ctx_group='enc_layers'):
next_state = lstm(num_hidden, indata=hidden,
prev_state=last_states[i],
param=param_cells[i],
seqidx=seqidx, layeridx=i, dropout=dp_ratio)
hidden = next_state.h
last_states[i] = next_state
# decoder
if dropout > 0.:
hidden = mx.sym.Dropout(data=hidden, p=dropout)
return last_states
def lstm_decoder(label_wv, in_lstm_state, num_lstm_layer, seq_len, num_hidden, num_label, dropout=0.):
# pass the state
with mx.AttrScope(ctx_group='decode'):
cls_weight = mx.sym.Variable("cls_weight")
cls_bias = mx.sym.Variable("cls_bias")
param_cells = []
last_states = copy(in_lstm_state)
for i in range(num_lstm_layer):
with mx.AttrScope(ctx_group='dec_layers'):
param_cells.append(LSTMParam(i2h_weight=mx.sym.Variable("dec_l%d_i2h_weight" % i),
i2h_bias=mx.sym.Variable("dec_l%d_i2h_bias" % i),
h2h_weight=mx.sym.Variable("dec_l%d_h2h_weight" % i),
h2h_bias=mx.sym.Variable("dec_l%d_h2h_bias" % i)))
hidden_all = []
hidden = mx.sym.Variable('dec_start')
for seqidx in range(seq_len):
# stack LSTM
for i in range(num_lstm_layer):
if i == 0:
dp_ratio = 0.
else:
dp_ratio = dropout
with mx.AttrScope(ctx_group='dec_layers'):
next_state = lstm(num_hidden, indata=hidden,
prev_state=last_states[i],
param=param_cells[i],
seqidx=seqidx, layeridx=i, dropout=dp_ratio)
hidden = next_state.h
last_states[i] = next_state
# decoder
if dropout > 0.:
hidden = mx.sym.Dropout(data=hidden, p=dropout)
hidden_all.append(hidden)
if seqidx == seq_len - 2:
hidden = label_wv[seqidx] + label_wv[seqidx + 1]
else:
hidden = label_wv[seqidx]
with mx.AttrScope(ctx_group='decode'):
hidden_concat = mx.sym.Concat(*hidden_all, dim=0)
pred = mx.sym.FullyConnected(data=hidden_concat, num_hidden=num_label,
weight=cls_weight, bias=cls_bias, name='pred')
return pred
def seq_softmax(label, pred):
with mx.AttrScope(ctx_group='loss'):
label = mx.sym.transpose(data=label)
label = mx.sym.Reshape(data=label, target_shape=(0,))
sm = mx.sym.SoftmaxOutput(data=pred, label=label, name='softmax')
return sm
def lstm_model(data_name, label_name, enc_para, dec_para):
data = mx.sym.Variable(data_name)
label = mx.sym.Variable(label_name)
with mx.AttrScope(ctx_group='embed'):
embed_weight = mx.sym.Variable("embed_weight")
# data
data_embed = mx.sym.Embedding(data=data, input_dim=enc_para.input_size, weight=embed_weight,
output_dim=enc_para.num_embed, name='embed')
data_wv = mx.sym.SliceChannel(data=data_embed, num_outputs=enc_para.seq_len, squeeze_axis=1)
# label
label_embed = mx.sym.Embedding(data=label, input_dim=enc_para.input_size, weight=embed_weight,
output_dim=enc_para.num_embed, name='embed')
label_wv = mx.sym.SliceChannel(data=label_embed, num_outputs=dec_para.seq_len, squeeze_axis=1)
enc_state = lstm_encoder(data_wv, enc_para.seq_len, enc_para.num_lstm_layer, enc_para.input_size,
enc_para.num_hidden, enc_para.num_embed, enc_para.dropout)
pred = lstm_decoder(label_wv, enc_state, dec_para.num_lstm_layer, dec_para.seq_len,
dec_para.num_hidden, dec_para.num_label, dec_para.dropout)
loss = seq_softmax(label, pred)
enc = [enc_state[i].h for i in range(enc_para.num_lstm_layer)]
return mx.sym.Group([loss] + enc)
def get_input_shapes(enc_para, dec_para, batch_size):
init_state_shapes = {}
for i in range(enc_para.num_lstm_layer):
init_state_shapes['enc_l{}_init_h'.format(i)] = (batch_size, enc_para.num_hidden)
init_state_shapes['enc_l{}_init_c'.format(i)] = (batch_size, enc_para.num_hidden)
init_state_shapes['dec_start'] = (batch_size, enc_para.num_embed)
return init_state_shapes