/
draft4.py
118 lines (85 loc) · 2.42 KB
/
draft4.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
import torch
import torch.nn as nn
import numpy as np
# a = torch.range(1, 32, 1).view(2,2,2,4)
# print(a,'\n', a.dtype)
# b= torch.range(33,64,1).view(2,2,2,4)
# print(b,'\n', b.dtype)
# # cos = nn.CosineSimilarity(dim=1, eps=1e-6)
# # print("=======================")
# # print(cos(a,b))
# # print(cos(a,b).sum())
# # a[range(2)]
# # a = torch.range(1,16).view(2,2,4)
# # b= torch.range(17,32).view(2,2,4)
# a.requires_grad_(True)
# b.requires_grad_(True)
# c= torch.range(1,6).view(2,3)
# c.requires_grad_(True)
# d = torch.range(7,12).view(2,3)
# d.requires_grad_(True)
# # print("c : \n",c)
# # print("d : \n",d)
# # print(torch.dot(a[range(2),range(2)], b[range(2),range(2)]))
# print("a : \n",a)
# print("b : \n",b)
# print("testing")
# tem = a*b
# print(tem)
# print(tem.requires_grad)
# a = torch.arange(1,17).view(2,2,4)
# b = torch.arange(17,41).view(3,2,4)
# print("true value: ")
# print(torch.sum(a[0]*b[0]))
# print(torch.sum(a[0]*b[1]))
# print(torch.sum(a[0]*b[2]))
# print("a : \n",a)
# print("b : \n",b)
# a = a.reshape(2,8)
# b = torch.transpose(b.reshape(3,8),0,1)
# print("test1: ")
# print("reshaping :")
# print(a)
# print(b)
# print("============")
# print(a@b)
# print("===============")
# import numpy as np
# a = np.array([[0, 1, 2],
# [4, 5, 6]])
# b = np.array(
# [
# [0, 1, 2],
# [4, 5, 6],
# [-1, 0, 1],
# [-3, -2, 1]
# ])
# print(np.einsum('ij,hj->hi', a, b))
# a = torch.arange(1, 33, 1).view(2,2,2,4)
# print( 'a\n', a,'\n', a.dtype)
# b= torch.arange(33,65,1).view(2,2,2,4)
# print( 'b\n', b,'\n', b.dtype)
# def FSP_matrix(f1: torch.tensor, f2: torch.tensor) : # (f size: b, c, h, w)
# _b, _c, _h, _w = f1.shape[0], f1.shape[1], f1.shape[2], f1.shape[3]
# # f1 = f1.view(_b,_c,_h*_w)
# # f2 = f2.view(_b,_c,_h*_w)
# # print(f"f1: \n{f1}\n\nf2: {f2}\n")
# # f2 = f2.transpose(1,2)
# # print(f"f2: {f2}\n")
# # return torch.bmm(f1,f2)
# return torch.bmm(f1.view(_b,_c,_h*_w), f2.view(_b,_c,_h*_w).transpose(1,2))
# print("FSP_matrix:\n", FSP_matrix(a, b))
# b = b.view(2,16)
# b = b.transpose(0, 1)
# print("b tranpose: ", b)
# a = torch.arange(0,36).view(2,2,3,3).type('torch.DoubleTensor')
# print("a:\n", a)
# print(a.mean(dim=1))
# print(a.mean(dim=1).mean(dim=0))
x = torch.arange(0,24).view(4,2,3)
x = x.type(torch.DoubleTensor)
print("1:\n", x)
print("2:\n", x.dtype)
# print(x.repeat(2,2,2,3))
# print(x.mean(dim=0))
print(x.repeat(2,1,1))