Skip to content

Commit d5860b0

Browse files
janeyx99janeyx99
committed
Redirect tensor tutorial in blitz to QS
1 parent a551cdf commit d5860b0

File tree

2 files changed

+4
-194
lines changed

2 files changed

+4
-194
lines changed

‎.jenkins/validate_tutorials_built.py‎

Lines changed: 1 addition & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -10,6 +10,7 @@
1010

1111
NOT_RUN = [
1212
"beginner_source/basics/intro", # no code
13+
"beginner_source/blitz/tensor_tutorial", # no code, redirects
1314
"beginner_source/onnx/intro_onnx",
1415
"beginner_source/profiler",
1516
"beginner_source/saving_loading_models",

‎beginner_source/blitz/tensor_tutorial.py‎

Lines changed: 3 additions & 194 deletions
Original file line numberDiff line numberDiff line change
@@ -2,200 +2,9 @@
22
Tensors
33
========
44
5-
Tensors are a specialized data structure that are very similar to arrays
6-
and matrices. In PyTorch, we use tensors to encode the inputs and
7-
outputs of a model, as well as the model’s parameters.
5+
This page has been moved.
86
9-
Tensors are similar to NumPy’s ndarrays, except that tensors can run on
10-
GPUs or other specialized hardware to accelerate computing. If you’re familiar with ndarrays, you’ll
11-
be right at home with the Tensor API. If not, follow along in this quick
12-
API walkthrough.
7+
.. raw:: html
8+
<meta http-equiv="Refresh" content="0; url='https://pytorch.org/tutorials/beginner/basics/tensorqs_tutorial'" />
139
1410
"""
15-
16-
import torch
17-
import numpy as np
18-
19-
20-
######################################################################
21-
# Tensor Initialization
22-
# ~~~~~~~~~~~~~~~~~~~~~
23-
#
24-
# Tensors can be initialized in various ways. Take a look at the following examples:
25-
#
26-
# **Directly from data**
27-
#
28-
# Tensors can be created directly from data. The data type is automatically inferred.
29-
30-
data = [[1, 2], [3, 4]]
31-
x_data = torch.tensor(data)
32-
33-
######################################################################
34-
# **From a NumPy array**
35-
#
36-
# Tensors can be created from NumPy arrays (and vice versa - see :ref:`bridge-to-np-label`).
37-
np_array = np.array(data)
38-
x_np = torch.from_numpy(np_array)
39-
40-
41-
###############################################################
42-
# **From another tensor:**
43-
#
44-
# The new tensor retains the properties (shape, datatype) of the argument tensor, unless explicitly overridden.
45-
46-
x_ones = torch.ones_like(x_data) # retains the properties of x_data
47-
print(f"Ones Tensor: \n {x_ones} \n")
48-
49-
x_rand = torch.rand_like(x_data, dtype=torch.float) # overrides the datatype of x_data
50-
print(f"Random Tensor: \n {x_rand} \n")
51-
52-
53-
######################################################################
54-
# **With random or constant values:**
55-
#
56-
# ``shape`` is a tuple of tensor dimensions. In the functions below, it determines the dimensionality of the output tensor.
57-
58-
shape = (2, 3,)
59-
rand_tensor = torch.rand(shape)
60-
ones_tensor = torch.ones(shape)
61-
zeros_tensor = torch.zeros(shape)
62-
63-
print(f"Random Tensor: \n {rand_tensor} \n")
64-
print(f"Ones Tensor: \n {ones_tensor} \n")
65-
print(f"Zeros Tensor: \n {zeros_tensor}")
66-
67-
68-
69-
70-
######################################################################
71-
# --------------
72-
#
73-
74-
75-
######################################################################
76-
# Tensor Attributes
77-
# ~~~~~~~~~~~~~~~~~
78-
#
79-
# Tensor attributes describe their shape, datatype, and the device on which they are stored.
80-
81-
tensor = torch.rand(3, 4)
82-
83-
print(f"Shape of tensor: {tensor.shape}")
84-
print(f"Datatype of tensor: {tensor.dtype}")
85-
print(f"Device tensor is stored on: {tensor.device}")
86-
87-
88-
######################################################################
89-
# --------------
90-
#
91-
92-
93-
######################################################################
94-
# Tensor Operations
95-
# ~~~~~~~~~~~~~~~~~
96-
#
97-
# Over 100 tensor operations, including transposing, indexing, slicing,
98-
# mathematical operations, linear algebra, random sampling, and more are
99-
# comprehensively described
100-
# `here <https://pytorch.org/docs/stable/torch.html>`__.
101-
#
102-
# Each of them can be run on the GPU (at typically higher speeds than on a
103-
# CPU). If you’re using Colab, allocate a GPU by going to Edit > Notebook
104-
# Settings.
105-
#
106-
107-
# We move our tensor to the GPU if available
108-
if torch.cuda.is_available():
109-
tensor = tensor.to('cuda')
110-
print(f"Device tensor is stored on: {tensor.device}")
111-
112-
113-
######################################################################
114-
# Try out some of the operations from the list.
115-
# If you're familiar with the NumPy API, you'll find the Tensor API a breeze to use.
116-
#
117-
118-
###############################################################
119-
# **Standard numpy-like indexing and slicing:**
120-
121-
tensor = torch.ones(4, 4)
122-
tensor[:,1] = 0
123-
print(tensor)
124-
125-
######################################################################
126-
# **Joining tensors** You can use ``torch.cat`` to concatenate a sequence of tensors along a given dimension.
127-
# See also `torch.stack <https://pytorch.org/docs/stable/generated/torch.stack.html>`__,
128-
# another tensor joining op that is subtly different from ``torch.cat``.
129-
t1 = torch.cat([tensor, tensor, tensor], dim=1)
130-
print(t1)
131-
132-
######################################################################
133-
# **Multiplying tensors**
134-
135-
# This computes the element-wise product
136-
print(f"tensor.mul(tensor) \n {tensor.mul(tensor)} \n")
137-
# Alternative syntax:
138-
print(f"tensor * tensor \n {tensor * tensor}")
139-
140-
######################################################################
141-
#
142-
# This computes the matrix multiplication between two tensors
143-
print(f"tensor.matmul(tensor.T) \n {tensor.matmul(tensor.T)} \n")
144-
# Alternative syntax:
145-
print(f"tensor @ tensor.T \n {tensor @ tensor.T}")
146-
147-
148-
######################################################################
149-
# **In-place operations**
150-
# Operations that have a ``_`` suffix are in-place. For example: ``x.copy_(y)``, ``x.t_()``, will change ``x``.
151-
152-
print(tensor, "\n")
153-
tensor.add_(5)
154-
print(tensor)
155-
156-
######################################################################
157-
# .. note::
158-
# In-place operations save some memory, but can be problematic when computing derivatives because of an immediate loss
159-
# of history. Hence, their use is discouraged.
160-
161-
######################################################################
162-
# --------------
163-
#
164-
165-
166-
######################################################################
167-
# .. _bridge-to-np-label:
168-
#
169-
# Bridge with NumPy
170-
# ~~~~~~~~~~~~~~~~~
171-
# Tensors on the CPU and NumPy arrays can share their underlying memory
172-
# locations, and changing one will change the other.
173-
174-
175-
######################################################################
176-
# Tensor to NumPy array
177-
# ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
178-
t = torch.ones(5)
179-
print(f"t: {t}")
180-
n = t.numpy()
181-
print(f"n: {n}")
182-
183-
######################################################################
184-
# A change in the tensor reflects in the NumPy array.
185-
186-
t.add_(1)
187-
print(f"t: {t}")
188-
print(f"n: {n}")
189-
190-
191-
######################################################################
192-
# NumPy array to Tensor
193-
# ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
194-
n = np.ones(5)
195-
t = torch.from_numpy(n)
196-
197-
######################################################################
198-
# Changes in the NumPy array reflects in the tensor.
199-
np.add(n, 1, out=n)
200-
print(f"t: {t}")
201-
print(f"n: {n}")

0 commit comments

Comments
 (0)