-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathwrapup_rnn.py
182 lines (167 loc) · 7.24 KB
/
wrapup_rnn.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
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
import csv
import zipfile
import util_boto3
import os.path
import pickle
import WFA
import json
import itertools
import numpy as np
import eval_rnn
import precalc_rnn
import Dataset
import tensorflow as tf
# rnns = """20190513093734_rnn_no5_abcdef_10_3.zip
# 20190513115614_rnn_no3_abcdef_10_3.zip
# 20190513094050_rnn_no5_abcd_10_3.zip
# 20190513120311_rnn_no2_abcdef_10_3.zip
# 20190515054726_rnn_no4_abcdef_10_3.zip
# 20190515063038_rnn_no3_abcd_10_3.zip
# 20190513094540_rnn_no4_abcd_10_3.zip
# 20190513121140_rnn_no2_abcd_10_3.zip
# 20190513135055_rnn_no1_abcdef_10_3.zip
# 20190513135613_rnn_no1_abcd_10_3.zip
# 20190515060412_rnn_no3_abcdefghij_10_3.zip
# 20190515084348_rnn_no2_abcdefghij_20_3.zip
# 20190515084609_rnn_no2_abcdefghij_10_3.zip
# 20190515112429_rnn_no1_abcdefghijklmno_10_3.zip
# 20190514142823_rnn_no4_abcdefghij_20_3.zip
# 20190515084307_rnn_no2_abcdefghijklmno_10_3.zip
# 20190514171225_rnn_no3_abcdefghijklmno_10_3.zip
# 20190514091252_rnn_no5_abcdefghijklmno_20_3.zip
# 20190514091232_rnn_no5_abcdefghijklmno_15_3.zip
# 20190514142922_rnn_no4_abcdefghijklmno_10_3.zip
# 20190514170722_rnn_no3_abcdefghijklmno_20_3.zip
# 20190514143253_rnn_no4_abcdefghij_10_3.zip
# 20190515112326_rnn_no1_abcdefghijklmno_15_3.zip
# 20190515084323_rnn_no2_abcdefghijklmno_20_3.zip
# 20190515084246_rnn_no2_abcdefghijklmno_15_3.zip
# 20190515060703_rnn_no3_abcdefghij_20_3.zip
# 20190514142303_rnn_no4_abcdefghijklmno_15_3.zip
# 20190514114707_rnn_no5_abcdefghij_15_3.zip
# 20190514114621_rnn_no5_abcdefghij_20_3.zip
# 20190515112145_rnn_no1_abcdefghijklmno_20_3.zip
# 20190514091009_rnn_no5_abcdefghijklmno_10_3.zip
# 20190515060653_rnn_no3_abcdefghij_15_3.zip
# 20190514115036_rnn_no5_abcdefghij_10_3.zip
# 20190514142246_rnn_no4_abcdefghijklmno_20_3.zip
# 20190515131517_rnn_no1_abcdefghij_10_3.zip
# 20190514170952_rnn_no3_abcdefghijklmno_15_3.zip
# 20190514143034_rnn_no4_abcdefghij_15_3.zip
# 20190515084532_rnn_no2_abcdefghij_15_3.zip
# 20190515135758_rnn_no1_abcdefghij_20_3.zip
# 20190515135922_rnn_no1_abcdefghij_15_3.zip"""
# rnns = """20190827110441-172-31-28-104_rnn_no1_abcdefghijklmno_10_3.zip
# 20190827125600-172-31-27-60_rnn_no4_abcd_10_3.zip
# 20190827085359-172-31-27-42_rnn_no4_abcdefghij_15_3.zip
# 20190827125514-172-31-20-198_rnn_no3_abcd_10_3.zip
# 20190827090320-172-31-27-60_rnn_no3_abcdefghijklmno_20_3.zip
# 20190827090310-172-31-18-182_rnn_no3_abcdefghijklmno_15_3.zip
# 20190827070625-172-31-29-101_rnn_no4_abcdefghijklmno_10_3.zip
# 20190827130614-172-31-29-75_rnn_no1_abcd_10_3.zip
# 20190827070655-172-31-18-182_rnn_no5_abcdefghijklmno_15_3.zip
# 20190827090330-172-31-25-112_rnn_no3_abcdefghijklmno_10_3.zip
# 20190904144447-192-168-121-1_rnn_no3_abcdef_10_3.zip
# 20190827090530-172-31-20-198_rnn_no3_abcdefghij_15_3.zip
# 20190827125536-172-31-18-182_rnn_no4_abcdef_10_3.zip
# 20190827105954-172-31-27-60_rnn_no2_abcdefghij_10_3.zip
# 20190827130150-172-31-17-80_rnn_no1_abcdef_10_3.zip
# 20190827130301-172-31-28-104_rnn_no2_abcd_10_3.zip
# 20190827110040-172-31-20-198_rnn_no1_abcdefghijklmno_15_3.zip
# 20190827090911-172-31-17-80_rnn_no2_abcdefghijklmno_15_3.zip
# 20190827070707-172-31-25-112_rnn_no4_abcdefghij_20_3.zip
# 20190827125343-172-31-29-101_rnn_no5_abcd_10_3.zip
# 20190827070217-172-31-27-42_rnn_no4_abcdefghijklmno_20_3.zip
# 20190827104555-172-31-27-42_rnn_no2_abcdefghijklmno_10_3.zip
# 20190827071125-172-31-17-80_rnn_no5_abcdefghij_15_3.zip
# 20190827090650-172-31-23-42_rnn_no2_abcdefghijklmno_20_3.zip
# 20190827090609-172-31-28-104_rnn_no3_abcdefghij_20_3.zip
# 20190827090831-172-31-29-75_rnn_no3_abcdefghij_10_3.zip
# 20190827105956-172-31-25-112_rnn_no1_abcdefghijklmno_20_3.zip
# 20190827070825-172-31-28-104_rnn_no5_abcdefghij_10_3.zip
# 20190827105920-172-31-18-182_rnn_no2_abcdefghij_15_3.zip
# 20190827071031-172-31-23-42_rnn_no4_abcdefghijklmno_15_3.zip
# 20190827070909-172-31-20-198_rnn_no5_abcdefghijklmno_10_3.zip
# 20190827105821-172-31-29-101_rnn_no2_abcdefghij_20_3.zip
# 20190827090218-172-31-29-101_rnn_no4_abcdefghij_10_3.zip
# 20190827125741-172-31-23-42_rnn_no2_abcdef_10_3.zip
# 20190827110535-172-31-17-80_rnn_no1_abcdefghij_10_3.zip
# 20190827070645-172-31-27-60_rnn_no5_abcdefghijklmno_20_3.zip
# 20190827110218-172-31-23-42_rnn_no1_abcdefghij_20_3.zip
# 20190827123757-172-31-27-42_rnn_no5_abcdef_10_3.zip
# 20190827070915-172-31-29-75_rnn_no5_abcdefghij_20_3.zip
# 20190827110716-172-31-29-75_rnn_no1_abcdefghij_15_3.zip"""
rnns = "20190825235324-192-168-121-1_rnn_paren_42_10.zip"
rnns = rnns.split("\n")
rnns = [r.strip() for r in rnns]
print(len(set(rnns)), len(rnns))
input()
rows = []
def split_line(x):
splitted = x.split()
# print(splitted)
if len(splitted) == 1:
return "", float(splitted[0])
elif len(splitted) == 2:
return splitted[0], float(splitted[1])
else:
assert False
xs = []
for e in rnns:
tf.reset_default_graph()
flag_paren = ("paren" in e)
print(e)
util_boto3.download(e)
task_name = "_".join(e.split("_")[2:])
with zipfile.ZipFile(e) as z:
dirname = os.path.splitext(e)[0]
z.extractall(dirname)
if flag_paren:
alphabet = "()0123456789"
wfa_orig = WFA.WFA("a", np.array([[0]]), np.array([[0]]), {"a": np.array([[0]])})
wfa_orig.callings = set()
alphabet_size = len(alphabet)
alphabet2id, max_length, regr, words = precalc_rnn.set_for_paren(e, dirname)
else:
with open(os.path.join(dirname, "wfa.pickle"), "rb") as f:
wfa_orig: WFA.WFA = pickle.load(f)
wfa_orig.callings = set()
alphabet = wfa_orig.alphabet
states_orig = wfa_orig.get_size()
alphabet2id, max_length, regr, words = precalc_rnn.set_for_artificial(e, dirname)
with open(os.path.join(dirname, "args.json"), "r") as f:
j = json.load(f)
batch_size = j["batch_size"]
if flag_paren:
with open(os.path.join(dirname, "test.txt"), "r") as f:
words_test = [split_line(x) for x in f.readlines()]
words_test = [(Dataset.word2vec(x, alphabet2id, max_length), y) for x, y in words_test]
else:
with open(os.path.join(dirname, "test.txt"), "r") as f:
words_test = []
for l in f.readlines():
w = l.strip()
words_test.append((Dataset.word2vec(w, alphabet2id, max_length), wfa_orig.classify_word(w)))
print("Testing: ")
for i in words_test[:5]:
print(f"{i[0]} : {i[1]}")
for i in words_test:
assert len(i[0]) <= max_length
print("Starting evaluation")
ev = eval_rnn.evaluate(regr, max_length, batch_size, words_test, dirname, lambda p: eval_rnn.eps_equality_absolute(p[0], p[1], 0.05))
xs.append((e, ev.n_equals, ev.ave_loss))
rows.append({"task name": task_name, "alphabet size": len(alphabet), "original wfa's size": wfa_orig.get_size(),
"mse": ev.ave_loss, "acc005": ev.n_equals/1000, "file": e})
print(xs[-1])
with open("wrapup_rnn.csv", "w") as f:
fieldnames = ["task name", "alphabet size", "original wfa's size",
"mse", "acc005", "file"]
writer = csv.DictWriter(f, fieldnames=fieldnames)
writer.writeheader()
for r in rows:
writer.writerow(r)
with open("wrapup_rnn.csv", "r") as f:
x = "".join(f.readlines())
x = x.replace("\n\n", "\n")
with open("wrapup_rnn.csv", "w") as f:
f.write(x)