-
Notifications
You must be signed in to change notification settings - Fork 0
/
experiments.py
66 lines (46 loc) · 1.1 KB
/
experiments.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
from math import floor, ceil
import numpy as np
def round_number(num):
c = ceil(num)
f = floor(num)
uni = np.random.uniform(0.0, 1.0)
if uni < (num - f):
return c
else:
return f
print(round_number(1.2))
# arr = np.random.randint(10, size=20)
# arr[0] = 1
# arr[6] = 1
# arr[11] = 1
# print(arr)
counts = np.array([1, 7, 1, 8, 7, 3, 1, 4, 7, 3, 1, 1, 7, 9, 2, 0, 1, 5, 9, 3])
print(counts)
s = set()
reduced_vocab_size = 0
for i, num in enumerate(counts):
if 1 < num:
s.add(i)
reduced_vocab_size += 1
print(s)
print(reduced_vocab_size)
total_counts = 0
new_word_index = 0
for i, num in enumerate(counts):
if 1 < num:
counts[new_word_index] = counts[i] - 1
total_counts += counts[new_word_index]
new_word_index += 1
print(counts)
for i in range(reduced_vocab_size, len(counts)):
counts[i] = 0
print(counts)
# unigram table
pow_counts = np.power(counts, 0.75)
z = np.sum(pow_counts)
max_size = 3000
nums = (max_size * pow_counts / z)
nums = np.vectorize(round_number)(nums)
print(nums)
sum_nums = np.sum(nums)
print(sum_nums)