-
Notifications
You must be signed in to change notification settings - Fork 8
/
losses.py
166 lines (140 loc) · 5.11 KB
/
losses.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
import torch
import torch.nn as nn
import h5py
import os
import constants
def KLDIVloss(output, target, V, D, loss_cuda):
"""
output (batch, vocab_size)
target (batch,)
criterion (nn.KLDIVLoss)
V (vocab_size, k)
D (vocab_size, k)
"""
# (batch, k) index in vocab_size dimension
# k-nearest neighbors for target
indices = torch.index_select(V, 0, target)
# (batch, k) gather along vocab_size dimension
outputk = torch.gather(output, 1, indices)
# (batch, k) index in vocab_size dimension
targetk = torch.index_select(D, 0, target)
# KLDIVcriterion
criterion = nn.KLDivLoss(reduction='sum').to(loss_cuda)
return criterion(outputk, targetk)
def dist2weight(D, dist_decay_speed=0.8):
'''
D is a matrix recording distances between each vocab and its k nearest vocabs
D(k, vocab_size)
weight: \frac{\exp{-|dis|*scale}}{\sum{\exp{-|dis|*scale}}}
Divide 100
'''
D = D.div(100)
D = torch.exp(-D * dist_decay_speed)
s = D.sum(dim=1, keepdim=True)
D = D / s
# The PAD should not contribute to the decoding loss
D[constants.PAD, :] = 0.0
return D
def load_dis_matrix(args):
assert os.path.isfile(args.knearestvocabs),\
"{} does not exist".format(args.knearestvocabs)
with h5py.File(args.knearestvocabs, 'r') as f:
V, D = f["V"], f["D"]
V, D = torch.LongTensor(V), torch.FloatTensor(D)
D = dist2weight(D, args.dist_decay_speed)
return V, D
def clusterloss(q, p, loss_cuda):
'''
caculate the KL loss for clustering
'''
q, p = q.to(loss_cuda), p.to(loss_cuda)
criterion = nn.KLDivLoss(reduction='sum').to(loss_cuda)
return criterion(q.log(), p)
def reconstructionLoss(gendata,
autoencoder,
rclayer,
lossF,
args,
cuda0,
cuda1,
loss_cuda):
"""
One batch reconstruction loss
cuda0 for autoencoder
cuda1 for rclayer
loss_cuda for reconstruction loss
Input:
gendata: a named tuple contains
gendata.src (seq_len1, batch): input tensor
gendata.lengths (1, batch): lengths of source sequences
gendata.trg (seq_len2, batch): target tensor.
autoencoder: map input to output.
log transform.
lossF: loss function.
---
Output:
loss
context (cuda0)
"""
input, lengths, target = gendata.src, gendata.lengths, gendata.trg
input = input.to(cuda0)
lengths = lengths.to(cuda0)
target = target.to(cuda0)
# print("input size:", input.size())
# print("lengths size:", lengths.size())
# print("target size:", target.size())
# Encoder & decoder
# output (trg_seq_len, batch, hidden_size)
# context (batch, hidden_size * num_directions)
output, context = autoencoder(input, lengths, target)
batch = output.size(1)
loss = 0
# we want to decode target in range [BOS+1:EOS]
target = target[1:]
# generate words from autoencoder output
for o, t in zip(output.split(args.gen_batch),
target.split(args.gen_batch)):
# (seq_len, gen_batch, hidden_size) =>
## (seq_len*gen_batch, hidden_size)
o = o.view(-1, o.size(2)).to(cuda1)
# print("o size:", o.size())
o = rclayer(o)
# (seq_len*gen_batch,)
t = t.view(-1)
o, t = o.to(loss_cuda), t.to(loss_cuda)
loss += lossF(o, t)
return loss.div(batch), context
def clusteringLoss(clusterlayer, context, p, cuda2, loss_cuda):
"""
One batch cluster KL loss
Input:
context: (batch, hidden_size * num_directions) last hidden layer from encoder
clusterlayer: caculate Student’s t-distribution with clustering center
p: (batch_size,n_clusters)target distribution
Output:loss
"""
batch = context.size(0)
assert batch == p.size(0)
q = clusterlayer(context.to(cuda2))
kl_loss = clusterloss(q, p, loss_cuda)
return kl_loss.div(batch)
def triLoss(a, p, n, autoencoder, loss_cuda):
"""
a (named tuple): anchor data
p (named tuple): positive data
n (named tuple): negative data
"""
a_src, a_lengths, a_invp = a.src, a.lengths, a.invp
p_src, p_lengths, p_invp = p.src, p.lengths, p.invp
n_src, n_lengths, n_invp = n.src, n.lengths, n.invp
a_src, a_lengths, a_invp = a_src.to(
loss_cuda), a_lengths.to(loss_cuda), a_invp.to(loss_cuda)
p_src, p_lengths, p_invp = p_src.to(
loss_cuda), p_lengths.to(loss_cuda), p_invp.to(loss_cuda)
n_src, n_lengths, n_invp = n_src.to(
loss_cuda), n_lengths.to(loss_cuda), n_invp.to(loss_cuda)
a_context = autoencoder.encoder_hn(a_src, a_lengths)
p_context = autoencoder.encoder_hn(p_src, p_lengths)
n_context = autoencoder.encoder_hn(n_src, n_lengths)
triplet_loss = nn.TripletMarginLoss(margin=1.0, p=2).to(loss_cuda)
return triplet_loss(a_context[a_invp], p_context[p_invp], n_context[n_invp])