forked from amidos2006/gym-pcgrl
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathmodel.py
167 lines (147 loc) · 8.22 KB
/
model.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
import tensorflow as tf
from tensorflow.keras import layers
from gym import spaces
import numpy as np
from stable_baselines.common.policies import ActorCriticPolicy, FeedForwardPolicy
from stable_baselines.common.distributions import CategoricalProbabilityDistributionType, ProbabilityDistributionType, CategoricalProbabilityDistribution, ProbabilityDistribution
from stable_baselines.a2c.utils import conv, linear, conv_to_fc
def Cnn1(image, **kwargs):
activ = tf.nn.relu
layer_1 = activ(conv(image, 'c1', n_filters=32, filter_size=3, stride=1, init_scale=np.sqrt(2), **kwargs))
layer_2 = activ(conv(layer_1, 'c2', n_filters=64, filter_size=3, stride=1, init_scale=np.sqrt(2), **kwargs))
layer_3 = activ(conv(layer_2, 'c3', n_filters=64, filter_size=3, stride=1, init_scale=np.sqrt(2), **kwargs))
layer_3 = conv_to_fc(layer_3)
return activ(linear(layer_3, 'fc1', n_hidden=512, init_scale=np.sqrt(2)))
def Cnn2(image, **kwargs):
activ = tf.nn.relu
layer_1 = activ(conv(image, 'c1', n_filters=32, filter_size=3, stride=2, init_scale=np.sqrt(2), **kwargs))
layer_2 = activ(conv(layer_1, 'c2', n_filters=64, filter_size=3, stride=2, init_scale=np.sqrt(2), **kwargs))
layer_3 = activ(conv(layer_2, 'c3', n_filters=64, filter_size=3, stride=1, init_scale=np.sqrt(2), **kwargs))
layer_3 = conv_to_fc(layer_3)
return activ(linear(layer_3, 'fc1', n_hidden=512, init_scale=np.sqrt(2)))
def FullyConv1(image, n_tools, **kwargs):
activ = tf.nn.relu
x = activ(conv(image, 'c1', n_filters=32, filter_size=3, stride=1,
pad='SAME', init_scale=np.sqrt(2)))
x = activ(conv(x, 'c2', n_filters=64, filter_size=3, stride=1,
pad='SAME', init_scale=np.sqrt(2)))
x = activ(conv(x, 'c3', n_filters=64, filter_size=3, stride=1,
pad='SAME', init_scale=np.sqrt(2)))
x = activ(conv(x, 'c4', n_filters=64, filter_size=3, stride=1,
pad='SAME', init_scale=np.sqrt(2)))
x = activ(conv(x, 'c5', n_filters=64, filter_size=3, stride=1,
pad='SAME', init_scale=np.sqrt(2)))
x = activ(conv(x, 'c6', n_filters=64, filter_size=3, stride=1,
pad='SAME', init_scale=np.sqrt(2)))
x = activ(conv(x, 'c7', n_filters=64, filter_size=3, stride=1,
pad='SAME', init_scale=np.sqrt(2)))
x = activ(conv(x, 'c8', n_filters=n_tools, filter_size=3, stride=1,
pad='SAME', init_scale=np.sqrt(2)))
act = conv_to_fc(x)
val = activ(conv(x, 'v1', n_filters=64, filter_size=3, stride=2,
init_scale=np.sqrt(2)))
val = activ(conv(val, 'v4', n_filters=64, filter_size=1, stride=1,
init_scale=np.sqrt(2)))
val = conv_to_fc(val)
return act, val
def FullyConv2(image, n_tools, **kwargs):
activ = tf.nn.relu
x = activ(conv(image, 'c1', n_filters=32, filter_size=3, stride=1,
pad='SAME', init_scale=np.sqrt(2)))
x = activ(conv(x, 'c2', n_filters=64, filter_size=3, stride=1,
pad='SAME', init_scale=np.sqrt(2)))
x = activ(conv(x, 'c3', n_filters=64, filter_size=3, stride=1,
pad='SAME', init_scale=np.sqrt(2)))
x = activ(conv(x, 'c4', n_filters=64, filter_size=3, stride=1,
pad='SAME', init_scale=np.sqrt(2)))
x = activ(conv(x, 'c5', n_filters=64, filter_size=3, stride=1,
pad='SAME', init_scale=np.sqrt(2)))
x = activ(conv(x, 'c6', n_filters=64, filter_size=3, stride=1,
pad='SAME', init_scale=np.sqrt(2)))
x = activ(conv(x, 'c7', n_filters=64, filter_size=3, stride=1,
pad='SAME', init_scale=np.sqrt(2)))
x = activ(conv(x, 'c8', n_filters=n_tools, filter_size=3, stride=1,
pad='SAME', init_scale=np.sqrt(2)))
act = conv_to_fc(x)
val = activ(conv(x, 'v1', n_filters=64, filter_size=3, stride=2,
init_scale=np.sqrt(2)))
val = activ(conv(val, 'v2', n_filters=64, filter_size=3, stride=2,
init_scale=np.sqrt(3)))
val = activ(conv(val, 'v4', n_filters=64, filter_size=1, stride=1,
init_scale=np.sqrt(2)))
val = conv_to_fc(val)
return act, val
class NoDenseCategoricalProbabilityDistributionType(ProbabilityDistributionType):
def __init__(self, n_cat):
"""
The probability distribution type for categorical input
:param n_cat: (int) the number of categories
"""
self.n_cat = n_cat
def probability_distribution_class(self):
return CategoricalProbabilityDistribution
def proba_distribution_from_latent(self, pi_latent_vector, vf_latent_vector, init_scale=1.0,
init_bias=0.0):
pdparam = pi_latent_vector
q_values = vf_latent_vector
return self.proba_distribution_from_flat(pdparam), pdparam, q_values
def param_shape(self):
return [self.n_cat]
def sample_shape(self):
return []
def sample_dtype(self):
return tf.int64
#used for "binary" and "zelda"
class FullyConvPolicyBigMap(ActorCriticPolicy):
def __init__(self, sess, ob_space, ac_space, n_env, n_steps, n_batch, **kwargs):
super(FullyConvPolicyBigMap, self).__init__(sess, ob_space, ac_space, n_env, n_steps, n_batch, **kwargs)
n_tools = int(ac_space.n / (ob_space.shape[0] * ob_space.shape[1]))
self._pdtype = NoDenseCategoricalProbabilityDistributionType(ac_space.n)
with tf.variable_scope("model", reuse=kwargs['reuse']):
pi_latent, vf_latent = FullyConv2(self.processed_obs, n_tools, **kwargs)
self._value_fn = linear(vf_latent, 'vf', 1)
self._proba_distribution, self._policy, self.q_value = \
self.pdtype.proba_distribution_from_latent(pi_latent, vf_latent, init_scale=0.01)
self._setup_init()
def step(self, obs, state=None, mask=None, deterministic=False):
if deterministic:
action, value, neglogp = self.sess.run([self.deterministic_action, self.value_flat, self.neglogp],
{self.obs_ph: obs})
else:
action, value, neglogp = self.sess.run([self.action, self.value_flat, self.neglogp],
{self.obs_ph: obs})
return action, value, self.initial_state, neglogp
def proba_step(self, obs, state=None, mask=None):
return self.sess.run(self.policy_proba, {self.obs_ph: obs})
def value(self, obs, state=None, mask=None):
return self.sess.run(self.value_flat, {self.obs_ph: obs})
#used for sokoban
class FullyConvPolicySmallMap(ActorCriticPolicy):
def __init__(self, sess, ob_space, ac_space, n_env, n_steps, n_batch, **kwargs):
super(FullyConvPolicySmallMap, self).__init__(sess, ob_space, ac_space, n_env, n_steps, n_batch, **kwargs)
n_tools = int(ac_space.n / (ob_space.shape[0] * ob_space.shape[1]))
self._pdtype = NoDenseCategoricalProbabilityDistributionType(ac_space.n)
with tf.variable_scope("model", reuse=kwargs['reuse']):
pi_latent, vf_latent = FullyConv1(self.processed_obs, n_tools, **kwargs)
self._value_fn = linear(vf_latent, 'vf', 1)
self._proba_distribution, self._policy, self.q_value = \
self.pdtype.proba_distribution_from_latent(pi_latent, vf_latent, init_scale=0.01)
self._setup_init()
def step(self, obs, state=None, mask=None, deterministic=False):
if deterministic:
action, value, neglogp = self.sess.run([self.deterministic_action, self.value_flat, self.neglogp],
{self.obs_ph: obs})
else:
action, value, neglogp = self.sess.run([self.action, self.value_flat, self.neglogp],
{self.obs_ph: obs})
return action, value, self.initial_state, neglogp
def proba_step(self, obs, state=None, mask=None):
return self.sess.run(self.policy_proba, {self.obs_ph: obs})
def value(self, obs, state=None, mask=None):
return self.sess.run(self.value_flat, {self.obs_ph: obs})
class CustomPolicyBigMap(FeedForwardPolicy):
def __init__(self, *args, **kwargs):
super(CustomPolicyBigMap, self).__init__(*args, **kwargs, cnn_extractor=Cnn2, feature_extraction="cnn")
class CustomPolicySmallMap(FeedForwardPolicy):
def __init__(self, *args, **kwargs):
super(CustomPolicySmallMap, self).__init__(*args, **kwargs, cnn_extractor=Cnn1, feature_extraction="cnn")