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
|
#
# Copyright (C) 2020 The Android Open Source Project
#
# 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.
#
# Model: given x = [x0, x1] and n, produces [sum(x0 ^ i), sum(x1 ^ i)] for i in [1, n].
#
# sum = [1, 1] // More generarlly, sum = ones_like(x).
# i = 1
# while i <= n:
# xi = x // x to the power of i (element-wise)
# j = 1
# while j < i:
# xi = xi * x
# j += 1
# sum = sum + xi
# i = i + 1
DataType10 = ["TENSOR_QUANT8_ASYMM_SIGNED", [1, 2], 1.0, 12]
DataType05 = ["TENSOR_QUANT8_ASYMM_SIGNED", [1, 2], 0.5, 12]
CounterType = ["TENSOR_INT32", [1]]
BoolType = ["TENSOR_BOOL8", [1]]
def quantize(data, scale, offset):
return [max(-128, min(127, int(round(x / scale)) + offset)) for x in data]
def MakeInnerConditionModel():
xi = Input("xi", DataType10)
j = Input("j", CounterType)
i = Input("i", CounterType)
x = Input("x", DataType05)
out = Output("out", BoolType)
model = Model()
model.IdentifyInputs(xi, j, i, x)
model.IdentifyOutputs(out)
model.Operation("LESS", j, i).To(out)
return model
def MakeInnerBodyModel():
xi = Input("xi", DataType10)
j = Input("j", CounterType)
i = Input("i", CounterType)
x = Input("x", DataType05)
xi_out = Output("xi_out", DataType10)
j_out = Output("j_out", CounterType)
model = Model()
model.IdentifyInputs(xi, j, i, x)
model.IdentifyOutputs(xi_out, j_out)
model.Operation("MUL", xi, x, 0).To(xi_out)
model.Operation("ADD", j, [1], 0).To(j_out)
return model
def MakeOuterConditionModel():
sum = Input("sum", DataType10)
i = Input("i", CounterType)
n = Input("n", CounterType)
x = Input("x", DataType05)
out = Output("out", BoolType)
model = Model()
model.IdentifyInputs(sum, i, n, x)
model.IdentifyOutputs(out)
model.Operation("LESS_EQUAL", i, n).To(out)
return model
def MakeOuterBodyModel():
sum = Input("sum", DataType10)
i = Input("i", CounterType)
n = Input("n", CounterType)
x = Input("x", DataType05)
sum_out = Output("sum_out", DataType10)
i_out = Output("i_out", CounterType)
xi_init = Internal("xi_init", DataType10)
j_init = [1]
cond = MakeInnerConditionModel()
body = MakeInnerBodyModel()
xi = Internal("xi", DataType10)
zero = Parameter("zero", DataType10, quantize([0, 0], 1.0, 12))
model = Model()
model.IdentifyInputs(sum, i, n, x)
model.IdentifyOutputs(sum_out, i_out)
model.Operation("ADD", x, zero, 0).To(xi_init)
model.Operation("WHILE", cond, body, xi_init, j_init, i, x).To(xi)
model.Operation("ADD", i, [1], 0).To(i_out)
model.Operation("ADD", sum, xi, 0).To(sum_out)
return model
def Test(x_data, n_data, sum_data):
x = Input("x", DataType05)
n = Input("n", CounterType)
sum = Output("sum", DataType10)
cond = MakeOuterConditionModel()
body = MakeOuterBodyModel()
sum_init = Parameter("sum_init", DataType10, quantize([1, 1], 1.0, 12))
i_init = [1]
model = Model().Operation("WHILE", cond, body, sum_init, i_init, n, x).To(sum)
example = Example({
x: quantize(x_data, 0.5, 12),
n: [n_data],
sum: quantize(sum_data, 1.0, 12),
}, name="n_{}".format(n_data))
example.AddVariations(AllOutputsAsInternalCoverter())
Test(x_data=[2, 3], n_data=0, sum_data=[1, 1])
Test(x_data=[2, 3], n_data=1, sum_data=[1 + 2, 1 + 3])
Test(x_data=[2, 3], n_data=2, sum_data=[1 + 2 + 4, 1 + 3 + 9])
Test(x_data=[2, 3], n_data=3, sum_data=[1 + 2 + 4 + 8, 1 + 3 + 9 + 27])
Test(x_data=[2, 3], n_data=4, sum_data=[1 + 2 + 4 + 8 + 16, 1 + 3 + 9 + 27 + 81])
|
