Repository navigation
Expand file tree
/
Copy pathsymmetries.py
More file actions
196 lines (160 loc) · 5.68 KB
/
Copy pathsymmetries.py
File metadata and controls
196 lines (160 loc) · 5.68 KB
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
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
# Copyright 2018 Google LLC
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import functools
import random
import go
import numpy as np
import tensorflow as tf
"""
Allowable symmetries:
identity [12][34]
rot90 [24][13]
rot180 [43][21]
rot270 [31][42]
flip [13][24]
fliprot90 [34][12]
fliprot180 [42][31]
fliprot270 [21][43]
"""
INVERSES = {
'identity': 'identity',
'rot90': 'rot270',
'rot180': 'rot180',
'rot270': 'rot90',
'flip': 'flip',
'fliprot90': 'fliprot90',
'fliprot180': 'fliprot180',
'fliprot270': 'fliprot270',
}
IMPLS = {
'identity': lambda x: x,
'rot90': np.rot90,
'rot180': functools.partial(np.rot90, k=2),
'rot270': functools.partial(np.rot90, k=3),
'flip': lambda x: np.rot90(np.fliplr(x)),
'fliprot90': np.flipud,
'fliprot180': lambda x: np.rot90(np.flipud(x)),
'fliprot270': np.fliplr,
}
assert set(IMPLS.keys()) == set(INVERSES.keys())
# A symmetry is just a string describing the transformation.
SYMMETRIES = list(INVERSES.keys())
# SYMMETRIES_SANS_IDENTITY = [s for s in SYMMETRIES if s != 'identity']
def invert_symmetry(s):
return INVERSES[s]
def apply_symmetry_feat(sym: str, features):
return IMPLS[sym](features)
def apply_symmetry_pi(s, pi):
"""
pi: policy head output: 1-d array of size N*N+1 with pass move at end
"""
pi = np.copy(pi)
# rotate all moves except for the pass move at end
pi[:-1] = IMPLS[s](pi[:-1].reshape([go.N, go.N])).ravel()
return pi
def randomize_symmetries_feat(features):
symmetries_used = [random.choice(SYMMETRIES) for _ in features]
return symmetries_used, [apply_symmetry_feat(s, f)
for s, f in zip(symmetries_used, features)]
def invert_symmetries_pi(symmetries, pis):
return [apply_symmetry_pi(invert_symmetry(s), pi)
for s, pi in zip(symmetries, pis)]
def rotate_train_nhwc(x, pi):
sym = tf.random_uniform(
[],
minval=0,
maxval=len(SYMMETRIES),
dtype=tf.int32,
seed=123)
def rotate(tensor):
# flipLeftRight
tensor = tf.where(
tf.bitwise.bitwise_and(sym, 1) > 0,
tf.reverse(tensor, axis=[0]),
tensor)
# flipUpDown
tensor = tf.where(
tf.bitwise.bitwise_and(sym, 2) > 0,
tf.reverse(tensor, axis=[1]),
tensor)
# flipDiagonal
tensor = tf.where(
tf.bitwise.bitwise_and(sym, 4) > 0,
tf.transpose(tensor, perm=[1, 0, 2]),
tensor)
return tensor
# TODO(tommadams): use tf.ensure_shape instead of tf.assert_equal.
squares = go.N * go.N
assert_shape_pi = tf.assert_equal(pi.shape.as_list(), [squares + 1])
x_shape = x.shape.as_list()
assert_shape_x = tf.assert_equal(x_shape, [go.N, go.N, x_shape[2]])
pi_move = tf.slice(pi, [0], [squares], name="slice_moves")
pi_pass = tf.slice(pi, [squares], [1], name="slice_pass")
# Add a final dim so that x and pi have same shape: [N,N,num_features].
pi_n_by_n = tf.reshape(pi_move, [go.N, go.N, 1])
with tf.control_dependencies([assert_shape_x, assert_shape_pi]):
pi_rot = tf.concat(
[tf.reshape(rotate(pi_n_by_n), [squares]), pi_pass],
axis=0)
return rotate(x), pi_rot
def rotate_train_nchw(x, pi):
sym = tf.random_uniform(
[],
minval=0,
maxval=len(SYMMETRIES),
dtype=tf.int32,
seed=123)
def rotate(tensor):
# flipLeftRight
tensor = tf.where(
tf.bitwise.bitwise_and(sym, 1) > 0,
tf.reverse(tensor, axis=[1]),
tensor)
# flipUpDown
tensor = tf.where(
tf.bitwise.bitwise_and(sym, 2) > 0,
tf.reverse(tensor, axis=[2]),
tensor)
# flipDiagonal
tensor = tf.where(
tf.bitwise.bitwise_and(sym, 4) > 0,
tf.transpose(tensor, perm=[0, 2, 1]),
tensor)
return tensor
# TODO(tommadams): use tf.ensure_shape instead of tf.assert_equal.
squares = go.N * go.N
assert_shape_pi = tf.assert_equal(pi.shape.as_list(), [squares + 1])
x_shape = x.shape.as_list()
assert_shape_x = tf.assert_equal(x_shape, [x_shape[0], go.N, go.N])
pi_move = tf.slice(pi, [0], [squares], name="slice_moves")
pi_pass = tf.slice(pi, [squares], [1], name="slice_pass")
# Add a dim so that x and pi have same shape: [num_features,N,N].
pi_n_by_n = tf.reshape(pi_move, [1, go.N, go.N])
with tf.control_dependencies([assert_shape_x, assert_shape_pi]):
pi_rot = tf.concat(
[tf.reshape(rotate(pi_n_by_n), [squares]), pi_pass],
axis=0)
return rotate(x), pi_rot
def apply_symmetry_dual(X0, y0, v0, num_symmetries=8):
"""
to transform on the fly, just need to use tensor ops
# return tf.repeat(X0, repeats=2, axis=0), tf.repeat(y0, repeats=2, axis=0), tf.repeat(v0, repeats=2, axis=0)
# return tf.experimental.numpy.rot90(X0, axes=(1, 2)), y0, v0
"""
Xs, ys, vs = [], [], []
for s in random.sample(SYMMETRIES, num_symmetries):
Xs.append(apply_symmetry_feat(s, X0))
ys.append(apply_symmetry_pi(s, y0))
vs.append(v0)
return np.stack(Xs), np.stack(ys), np.stack(vs)