forked from tonywu95/eval_gen
-
Notifications
You must be signed in to change notification settings - Fork 0
/
samplers_32.py
161 lines (137 loc) · 6.23 KB
/
samplers_32.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 time
import numpy as np
from algorithms.hmc import*
import theano
import theano.tensor as T
from algorithms import ais
from lib import utils
np.random.seed(123)
sharedX = (lambda X:
theano.shared(np.asarray(X, dtype=theano.config.floatX)))
class AISPath:
def __init__(self, generator, obs, num_samples, sigma, hdim, L, epsilon, data,prior, init_state=None,recog_mean=None,recog_log_sigma=None):
self.generator = generator
self.batch_size = obs.shape[0]
self.obs_val = sharedX(np.reshape(obs,[1,self.batch_size,32,32]))
self.obs = T.tensor4()
self.t = T.scalar()
self.sigma = sigma
self.n_sam = num_samples
self.hdim = hdim
self.L = L
self.eps = epsilon
self.data = data
self.prior = prior
if self.prior[:5] == 'recog':
self.recog_mean = sharedX(np.reshape(recog_mean,[self.n_sam,self.batch_size,self.hdim]))
self.recog_log_sigma = sharedX(np.reshape(recog_log_sigma,[self.n_sam,self.batch_size,self.hdim]))
if init_state is None:
self.build(self.eps, self.L)
else:
self.build(self.eps, self.L,init_state = init_state)
def build(self,
initial_stepsize,
n_steps,
target_acceptance_rate=.65,
stepsize_dec=0.98,
stepsize_min=0.0001,
stepsize_max=0.5,
stepsize_inc=1.02,
# used in geometric avg. 1.0 would be not moving at all
avg_acceptance_slowness=0.9,
seed=12345,
init_state=None
):
if init_state is None:
init_h = np.random.normal(0,1,size=[self.n_sam*self.batch_size,self.hdim]).astype(np.float32)
else:
init_h = init_state
print ('load init_state')
init_m = np.random.randn(self.n_sam*self.batch_size, self.hdim).astype(np.float32)
# For HMC
# h denotes current states
self.h = sharedX(init_h)
# m denotes momentum
t = T.scalar()
self.generated = self.generate(self.h)
lld = T.reshape(-self.energy_fn(self.h), [self.n_sam,self.batch_size])
self.eval_lld = theano.function([t],lld,givens ={self.obs:self.obs_val,self.t:t})
# allocate shared variables
stepsize = sharedX(initial_stepsize)
avg_acceptance_rate = sharedX(target_acceptance_rate)
s_rng = TT.shared_randomstreams.RandomStreams(seed)
# define graph for an `n_steps` HMC simulation
accept, final_pos = hmc_move(
s_rng,
self.h,
self.energy_fn,
stepsize,
n_steps)
# define the dictionary of updates, to apply on every `simulate` call
simulate_updates = hmc_updates(
self.h,
stepsize,
avg_acceptance_rate,
final_pos=final_pos,
accept=accept,
stepsize_min=stepsize_min,
stepsize_max=stepsize_max,
stepsize_inc=stepsize_inc,
stepsize_dec=stepsize_dec,
target_acceptance_rate=target_acceptance_rate,
avg_acceptance_slowness=avg_acceptance_slowness)
self.step = theano.function([t], [accept], updates=simulate_updates, givens={self.obs:self.obs_val,self.t:t})
def init_partition_function(self):
return 0.
def prior_logpdf(self, state, prior):
if prior == "normal":
return (-T.sum(T.square(state), [-1]) / (2.) - self.hdim/2.*np.log(2 * np.pi))
elif prior == 'recog1':
return -T.sum(T.square(state-self.recog_mean)/(T.exp(self.recog_log_sigma)), [-1]) / (2.) - self.hdim/2.*np.log(2 * np.pi)-T.sum(self.recog_log_sigma,[-1])/2.
elif prior == 'recog2': ##IWAE
return -T.sum(T.square(state-self.recog_mean)/T.square(T.exp(self.recog_log_sigma)), [-1]) / (2.) - self.hdim/2.*np.log(2 * np.pi)-T.sum(self.recog_log_sigma,[-1])
def likelihood(self, state,generated):
k = 32*32
if self.data == "binary":
return - T.sum(T.nnet.binary_crossentropy(generated, T.addbroadcast(self.obs,0)),[-1,-2])
if self.data == "continuous":
return (-T.sum(T.square(generated-T.addbroadcast(self.obs,0)),[-1,-2]) / (2*self.sigma) - k/2.*np.log(2 * np.pi)-k/2.*np.log(self.sigma))
def energy_fn(self, state):
generated = self.generator(state)
generated = T.reshape(generated,[self.n_sam,self.batch_size,32,32])
state = T.reshape(state,[self.n_sam,self.batch_size,self.hdim])
if self.prior =='normal':
energy = - (self.prior_logpdf(state,'normal') + self.t*self.likelihood(state,generated))
else:
energy = - (self.t*(self.prior_logpdf(state,'normal') + self.likelihood(state,generated))+(1-self.t)*self.prior_logpdf(state,self.prior))
return T.reshape(energy,[-1])
def generate(self,state):
generated = self.generator(state)
generated = T.reshape(generated,[self.n_sam,self.batch_size,32,32])
return generated
def run_ais(model, obs, num_samples, num_steps, sigma, hdim, L, epsilon, data, prior, schedule=None):
if schedule is None:
schedule = ais.sigmoid_schedule(num_steps)
mean = None
log_sigma = None
## prior:recog <--If using the recognition nets to predict initial AIS chain.
if prior[:5] == 'recog':
'load encoder net to predict initial dist...'
obs = obs.reshape(obs.shape[0],784)
obs_rep = np.tile(obs,[num_samples,1])
if prior[5] == '1':
rec_net = utils.load_encoder('vae','c4000',eval_np=True)
elif prior[5] == '2':
rec_net = utils.load_encoder('iwae','50',eval_np=True)
state,mean,log_sigma = rec_net(obs_rep)
else:
state = None
path = AISPath(model, obs, num_samples,sigma, hdim, L, epsilon, data, prior,init_state=state,recog_mean=mean,recog_log_sigma=log_sigma)
lld = ais.ais(path, schedule,sigma)
return lld
def run_reverse_ais(model, obs, state, num_steps, sigma, hdim, L, epsilon, data, prior, schedule=None):
if schedule is None:
schedule = ais.sigmoid_schedule(num_steps)
path = AISPath(model, obs, 1, sigma, hdim, L, epsilon,data,prior, init_state = state)
lld = ais.reverse_ais(path, schedule, sigma)
return lld