Skip to content

Navigation Menu

Sign in
Appearance settings

Search code, repositories, users, issues, pull requests...

Provide feedback

We read every piece of feedback, and take your input very seriously.

Saved searches

Use saved searches to filter your results more quickly
Sign up
Appearance settings

Commit e8a153b

Browse files
committed
PyTorch for Numpy users
1 parent ec2d652 commit e8a153b

File tree

2 files changed

+203
-0
lines changed

2 files changed

+203
-0
lines changed
Lines changed: 199 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,199 @@
1+
- [PyTorch for Numpy users.](#pytorch-for-numpy-users)
2+
- [Types](#types)
3+
- [Constructors](#constructors)
4+
- [Ones and zeros](#ones-and-zeros)
5+
- [From existing data](#from-existing-data)
6+
- [Numerical ranges](#numerical-ranges)
7+
- [Linear algebra](#linear-algebra)
8+
- [Building matrices](#building-matrices)
9+
- [Attributes](#attributes)
10+
- [Indexing](#indexing)
11+
- [Shape manipulation](#shape-manipulation)
12+
- [Item selection and manipulation](#item-selection-and-manipulation)
13+
- [Calculation](#calculation)
14+
- [Arithmetic and comparison operations](#arithmetic-and-comparison-operations)
15+
- [Random numbers](#random-numbers)
16+
- [Numerical operations](#numerical-operations)
17+
18+
19+
# PyTorch for Numpy users.
20+
21+
22+
## Types
23+
24+
| Numpy | PyTorch |
25+
| ---------- | ----------------------------- |
26+
| np.ndarray | torch.Tensor |
27+
| np.float32 | torch.float32<br>torch.float |
28+
| np.float64 | torch.float64<br>torch.double |
29+
| np.float16 | torch.float16<br>torch.half |
30+
| np.int8 | torch.int8 |
31+
| np.uint8 | torch.uint8 |
32+
| np.int16 | torch.int16<br>torch.short |
33+
| np.int32 | torch.int32<br>torch.int |
34+
| np.int64 | torch.int64<br>torch.long |
35+
36+
37+
38+
## Constructors
39+
40+
### Ones and zeros
41+
42+
| Numpy | PyTorch |
43+
| ---------------- | ------------------- |
44+
| np.empty((2, 3)) | torch.empty(2, 3) |
45+
| np.empty_like(x) | torch.empty_like(x) |
46+
| np.eye | torch.eye |
47+
| np.identity | torch.eye |
48+
| np.ones | torch.ones |
49+
| np.ones_like | torch.ones_like |
50+
| np.zeros | torch.zeros |
51+
| np.zeros_like | torch.zeros_like |
52+
53+
54+
### From existing data
55+
56+
| Numpy | PyTorch |
57+
| ------------------------------------------------------------- | --------------------------------------------- |
58+
| np.array([[1, 2], [3, 4]]) | torch.tensor([[1, 2], [3, 4]]) |
59+
| np.array([3.2, 4.3], dtype=np.float16)<br>np.float16([3.2, 4.3]) | torch.tensor([3.2, 4.3], dtype=torch.float16) |
60+
| x.copy() | x.clone() |
61+
| np.fromfile(file) | torch.tensor(torch.Storage(file)) |
62+
| np.frombuffer | |
63+
| np.fromfunction | |
64+
| np.fromiter | |
65+
| np.fromstring | |
66+
| np.load | torch.load |
67+
| np.loadtxt | |
68+
| np.concatenate | torch.cat |
69+
70+
71+
### Numerical ranges
72+
73+
| Numpy | PyTorch |
74+
| -------------------- | ----------------------- |
75+
| np.arange(10) | torch.arange(10) |
76+
| np.arange(2, 3, 0.1) | torch.arange(2, 3, 0.1) |
77+
| np.linspace | torch.linspace |
78+
| np.logspace | torch.logspace |
79+
80+
81+
### Linear algebra
82+
83+
| Numpy | PyTorch |
84+
| ------ | -------- |
85+
| np.dot | torch.mm |
86+
87+
88+
### Building matrices
89+
90+
| Numpy | PyTorch |
91+
| ------- | ---------- |
92+
| np.diag | torch.diag |
93+
| np.tril | torch.tril |
94+
| np.triu | torch.triu |
95+
96+
97+
### Attributes
98+
99+
| Numpy | PyTorch |
100+
| --------- | ------------ |
101+
| x.shape | x.shape |
102+
| x.strides | x.stride() |
103+
| x.ndim | x.dim() |
104+
| x.data | x.data |
105+
| x.size | x.nelement() |
106+
| x.dtype | x.dtype |
107+
108+
109+
### Indexing
110+
111+
| Numpy | PyTorch |
112+
| ------------------- | ---------------------------------------- |
113+
| x[0] | x[0] |
114+
| x[:, 0] | x[:, 0] |
115+
| x[indices] | x[indices] |
116+
| np.take(x, indices) | torch.take(x, torch.LongTensor(indices)) |
117+
| x[x != 0] | x[x != 0] |
118+
119+
120+
### Shape manipulation
121+
122+
| Numpy | PyTorch |
123+
| -------------------------------------- | ------------------------ |
124+
| x.reshape | x.reshape; x.view |
125+
| x.resize() | x.resize_ |
126+
| | x.resize_as_ |
127+
| x.transpose | x.transpose<br> x.permute |
128+
| x.flatten | x.view(-1) |
129+
| x.squeeze() | x.squeeze() |
130+
| x[:, np.newaxis]<br>np.expand_dims(x, 1) | x.unsqueeze(1) |
131+
132+
133+
### Item selection and manipulation
134+
135+
| Numpy | PyTorch |
136+
| -------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------- |
137+
| np.put | - |
138+
| x.put | x.put_ |
139+
| x = np.array([1, 2, 3])<br>x.repeat(2)<br># [1, 1, 2, 2, 3, 3] | x = torch.tensor([1, 2, 3])<br>x.repeat(2)<br># [1, 2, 3, 1, 2, 3]<br>x.repeat(2).reshape(2, -1).transpose(1, 0).reshape(-1)<br># [1, 1, 2, 2, 3, 3] |
140+
| np.tile(x, (3, 2)) | x.repeat(3, 2) |
141+
| np.choose | |
142+
| np.sort | sorted, indices = torch.sort(x, [dim]) |
143+
| np.argsort | sorted, indices = torch.sort(x, [dim]) |
144+
| np.nonzero | torch.nonzero |
145+
| np.where | torch.where |
146+
| x[::-1] | torch.flip(x, [0]) |
147+
148+
149+
### Calculation
150+
151+
| Numpy | PyTorch |
152+
| ------------- | ------------------------------ |
153+
| x.min | x.min |
154+
| x.argmin | x.argmin |
155+
| x.max | x.max |
156+
| x.argmax | x.argmax |
157+
| x.clip | x.clamp |
158+
| x.round | x.round |
159+
| np.floor(x) | torch.floor(x); x.floor() |
160+
| np.ceil(x) | torch.ceil(x); x.ceil() |
161+
| x.trace | x.trace |
162+
| x.sum | x.sum |
163+
| x.sum(axis=0) | x.sum(0) |
164+
| x.cumsum | x.cumsum |
165+
| x.mean | x.mean |
166+
| x.std | x.std |
167+
| x.prod | x.prod |
168+
| x.cumprod | x.cumprod |
169+
| x.all | (x == 1).sum() == x.nelement() |
170+
| x.any | (x == 1).sum() > 0 |
171+
172+
173+
### Arithmetic and comparison operations
174+
175+
| Numpy | PyTorch |
176+
| ---------------- | ------- |
177+
| np.less | x.lt |
178+
| np.less_equal | x.le |
179+
| np.greater | x.gt |
180+
| np.greater_equal | x.ge |
181+
| np.equal | x.eq |
182+
| np.not_equal | x.ne |
183+
184+
### Random numbers
185+
186+
| Numpy | PyTorch |
187+
| ------------------------ | ----------------- |
188+
| np.random.seed | torch.manual_seed |
189+
| np.random.permutation(5) | torch.randperm(5) |
190+
191+
192+
### Numerical operations
193+
194+
| Numpy | PyTorch |
195+
| ------- | ---------- |
196+
| np.sign | torch.sign |
197+
| np.sqrt | torch.sqrt |
198+
199+

‎README.md‎

Lines changed: 4 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -5,6 +5,7 @@
55
>Table of Contents / 目录:
66
- [PyTorch tutorials, examples and books](#pytorch-tutorials-examples-and-books)
77
- [PyTorch 版本变化及迁移指南](#pytorch-版本变化及迁移指南)
8+
- [PyTorch for Numpy users 给Numpy用户的PyTorch指南](#pytorch-for-numpy-users-给numpy用户的pytorch指南)
89
- [PyTorch 1.0 tutorials and examples](#pytorch-10-tutorials-and-examples)
910
- [Books and slides about PyTorch](#books-and-slides-about-pytorch)
1011
- [PyTorch深度学习:60分钟入门与实战](#pytorch深度学习60分钟入门与实战)
@@ -22,6 +23,9 @@
2223

2324
* 版本变化及迁移指南见[**这里**](PyTorch版本变化及迁移指南/README.md)。
2425

26+
## PyTorch for Numpy users 给Numpy用户的PyTorch指南
27+
28+
2529
## PyTorch 1.0 tutorials and examples
2630

2731
* [PyTorch-basics PyTorch基础](https://github.com/bat67/pytorch-tutorials-examples-and-books/tree/master/PyTorch-basics%20PyTorch基础)

0 commit comments

Comments
(0)

AltStyle によって変換されたページ (->オリジナル) /