-
Notifications
You must be signed in to change notification settings - Fork 0
/
MyLoss.py
50 lines (42 loc) · 1.69 KB
/
MyLoss.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
# -*- coding: utf-8 -*-
from __future__ import print_function, division
import torch
from torch.autograd import Variable
from torch.distributions.multivariate_normal import MultivariateNormal
import numpy as np
import torch.nn as nn
def LogLikeLoss(output_mean, output_cov, target):
D = 3
torchType = torch.cuda.FloatTensor
loss = 0
mean = output_mean
L_vect = output_cov
numEl = L_vect.size()[0]
#L = Variable(torch.zeros(L_vect.shape()[0], D, D).type(torchType))
#cov = Variable(torch.ones(L_vect.shape()[0], D, D).type(torchType))
for i in range(numEl):
L = Variable(torch.zeros(D, D).type(torchType))
cov = Variable(torch.ones(D, D).type(torchType))
#L[i, np.tril_indices(D, 1)] = L_vect.clone()
L[0, 0] = torch.exp(L_vect[i, 0]).clone()
L[1, 0] = L_vect[i, 1].clone()
L[1, 1] = torch.exp(L_vect[i, 2]).clone()
L[2, 0] = L_vect[i, 3].clone()
L[2, 1] = L_vect[i, 4].clone()
L[2, 2] = torch.exp(L_vect[i, 5]).clone()
cov = L.mm(L.t())
m = MultivariateNormal(mean[i], cov)
loss -= m.log_prob(target[i])
loss /= numEl
# L = Variable(torch.zeros(D, D).type(torchType))
# cov = Variable(torch.ones(D, D).type(torchType))
# for i in range(ngauss):
# L[i, 0, 0] = torch.exp(cov2[(i * 6) + 0]).clone()
# L[i, 1, 0] = cov2[(i * 6) + 1].clone()
# L[i, 1, 1] = torch.exp(cov2[(i * 6) + 2]).clone()
# L[i, 2, 0] = cov2[(i * 6) + 3].clone()
# L[i, 2, 1] = cov2[(i * 6) + 4].clone()
# L[i, 2, 2] = torch.exp(cov2[(i * 6) + 5]).clone()
# Ltemp = L[i].clone()
# cov[i] = Ltemp.mm(Ltemp.t())
return loss