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
|
#!/usr/bin/env python3
import argparse
import pickle
import math
import numpy as np
import re
import sys
import sloika.layers
parser = argparse.ArgumentParser()
parser.add_argument('--id', default='' , help='Identifier for model names')
parser.add_argument('model', help='Pickle to read model from')
EMBEDDING_MATRIX = np.array([[1.0, 0.0, -1.0 / np.sqrt(2.0)],
[-1.0, 0.0, -1.0 / np.sqrt(2.0)],
[0.0, 1.0, 1.0 / np.sqrt(2.0)],
[0.0, -1.0, 1.0 / np.sqrt(2.0)]], dtype=np.float32)
trim_trailing_zeros = re.compile('0+p')
def small_hex(f):
hf = float(f).hex()
return trim_trailing_zeros.sub('p', hf)
def process_column(v, pad):
""" process and pad """
return [small_hex(f) for f in v] + [small_hex(0.0)] * pad
def cformatM(fh, name, X, nr=None, nc=None):
nrq = int(math.ceil(X.shape[1] / 4.0))
pad = nrq * 4 - X.shape[1]
lines = map(lambda v: ', '.join(process_column(v, pad)), X)
if nr is None:
nr = X.shape[1]
else:
nrq = int(math.ceil(nr / 4.0))
if nc is None:
nc = X.shape[0]
nelt = 4 * nrq * nc
fh.write('float {}[{}] = {}\n'.format('__' + name, nelt, '{'))
fh.write('\t' + ',\n\t'.join(lines))
fh.write('};\n')
fh.write('_Mat {} = {}\n\t.nr = {},\n\t.nrq = {},\n\t.nc = {},\n\t.stride = {},\n\t.data.f = {}\n{};\n'.format('_' + name, '{', nr, nrq, nc, nrq * 4, '__' + name, '}'))
fh.write('const scrappie_matrix {} = &{};\n\n'.format(name, '_' + name))
def cformatV(fh, name, X):
nrq = int(math.ceil(X.shape[0] / 4.0))
pad = nrq * 4 - X.shape[0]
lines = ', '.join(list(map(lambda f: small_hex(f), X)) + [small_hex(0.0)] * pad)
fh.write('float {}[] = {}\n'.format( '__' + name, '{'))
fh.write('\t' + lines)
fh.write('};\n')
fh.write('_Mat {} = {}\n\t.nr = {},\n\t.nrq = {},\n\t.nc = {},\n\t.stride = {},\n\t.data.f = {}\n{};\n'.format('_' + name, '{', X.shape[0], nrq, 1, nrq * 4, '__' + name, '}'))
fh.write('const scrappie_matrix {} = &{};\n\n'.format(name, '_' + name))
def process_convolution(layer, layerid):
assert isinstance(layer, sloika.layers.Convolution)
filterW = layer.W.get_value()
nfilter, nfeature , winlen = filterW.shape
nfeatureCeil4 = 4 * int(math.ceil(nfeature / 4.0))
cformatM(sys.stdout, '{}W'.format(layerid), filterW.transpose(0, 2, 1).reshape(-1, nfeature),
nr = (winlen - 1) * nfeatureCeil4 + nfeature, nc=nfilter)
cformatV(sys.stdout, '{}b'.format(layerid), layer.b.get_value().reshape(-1))
sys.stdout.write("const int {}stride = {};\n".format(layerid, layer.stride))
sys.stdout.write("""const size_t _{}nfilter = {};
const size_t _{}winlen = {};
""".format(layerid, nfilter, layerid, winlen))
if __name__ == '__main__':
args = parser.parse_args()
modelid = args.id + '_'
with open(args.model, 'rb') as fh:
network = pickle.load(fh)
assert network.major_version >= 2
sys.stdout.write("""#pragma once
#ifndef NANONET_SQUIGGLE_{}MODEL_H
#define NANONET_SQUIGGLE_{}MODEL_H
#include <assert.h>
#include "../util.h"
""".format(modelid.upper(), modelid.upper()))
# Embedding layer
cformatM(sys.stdout, 'embed_squiggle_{}W'.format(modelid), EMBEDDING_MATRIX)
""" first convolution
"""
process_convolution(network.sublayers[0], 'conv1_squiggle_{}'.format(modelid))
assert isinstance(network.sublayers[1], sloika.layers.Residual)
process_convolution(network.sublayers[1].sublayers[0], 'conv2_squiggle_{}'.format(modelid))
assert isinstance(network.sublayers[2], sloika.layers.Residual)
process_convolution(network.sublayers[2].sublayers[0], 'conv3_squiggle_{}'.format(modelid))
assert isinstance(network.sublayers[3], sloika.layers.Residual)
process_convolution(network.sublayers[3].sublayers[0], 'conv4_squiggle_{}'.format(modelid))
assert isinstance(network.sublayers[4], sloika.layers.Residual)
process_convolution(network.sublayers[4].sublayers[0], 'conv5_squiggle_{}'.format(modelid))
process_convolution(network.sublayers[5], 'conv6_squiggle_{}'.format(modelid))
sys.stdout.write('\n#endif /* NANONET_SQUIGGLE_{}MODEL_H */\n'.format(modelid.upper()))
|
