-
Notifications
You must be signed in to change notification settings - Fork 2
/
DLF.html
355 lines (285 loc) · 13.8 KB
/
DLF.html
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
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
<!DOCTYPE html>
<html>
<title>Deep Learning Frameworks</title>
<meta charset="UTF-8">
<meta name="viewport" content="width=device-width, initial-scale=1">
<link rel="stylesheet" href="https://www.w3schools.com/w3css/4/w3.css">
<link rel="stylesheet" href="https://fonts.googleapis.com/css?family=Inconsolata">
<script src="https://cdn.rawgit.com/google/code-prettify/master/loader/run_prettify.js?skin=desert"></script>
<style>
body, html {
height: 100%;
font-family: FreightSans,Helvetica Neue,Helvetica,Arial,sans-serif;
/* line-height: 1.5rem; */
/* font-size: 1.2rem; */
color: #6c6c6d;
letter-spacing: .01px;
}
.bgimg {
background-position: center;
background-size: cover;
background-image: url("background.jpeg");
min-height: 75%;
}
.menu {
display: none;
}
h5 {
color: #111111;
}
</style>
<body>
<!-- Links (sit on top) -->
<div class="w3-top">
<div class="w3-row w3-padding w3-black">
<div class="w3-col s2">
<a href="#" class="w3-button w3-block w3-black">Outline</a>
</div>
<div class="w3-col s2">
<a href="#setup" class="w3-button w3-block w3-black">Setup</a>
</div>
<div class="w3-col s2">
<a href="#PT" class="w3-button w3-block w3-black">PyTorch</a>
</div>
<div class="w3-col s2">
<a href="#TK" class="w3-button w3-block w3-black">TensorFlow/Keras</a>
</div>
<div class="w3-col s2">
<a href="#tools" class="w3-button w3-block w3-black">Tools</a>
</div>
</div>
</div>
<!-- Header with image -->
<header class="bgimg w3-display-container w3-grayscale-min" id="home">
<div class="w3-display-middle w3-center w3-padding-large w3-hide-small">
<span class="w3-tag" style="font-size:30px">Deep Learning Frameworks</span>
</div>
</header>
<!-- Add a background color and large text to the whole page -->
<!-- <div class="w3-sand w3-grayscale w3-large"> -->
<div class="w3-grayscale">
<!-- About Container -->
<div class="w3-container" id="description">
<div class="w3-content" style="max-width:700px">
<h5 class="w3-center w3-padding-64"><span class="w3-tag w3-wide">Outline</span></h5>
In this lecture, you will learn how to:
<ul>
<li><a href="#setup">Set Up A Working Environment</a></li>
<li><a href="#PT">Install PyTorch</a></li>
<li><a href="#TK">Install TensorFlow and Keras</a></li>
<li><a href="#tools">Work With Useful Tools</a></li>
</ul>
</div>
</div>
<!-- Set up a working environment -->
<div class="w3-container" id="setup">
<div class="w3-content" style="max-width:700px">
<h5 class="w3-center w3-padding-64"><span class="w3-tag w3-wide">Set Up A Working Environment</span></h5>
The first step before installing any state-of-the-art deep learning frameworks is to set up a proper working environment.
<br></br></br>
Most popular frameworks are in Python. If you are not familiar with it but have a programming background, read this: <a href=https://learnxinyminutes.com/docs/python/>Learn Python in Y Minutes</a>. Otherwise if you want to learn from the beginning, <a href="https://learnpythonthehardway.org/book/">this book</a> has been recommended by many.
<br></br>
<h5 class="w3-center">Python</h5>
In Linux/Mac systems, you should already have Python installed.
<br></br>
Install pip, the convenient package manager, if needed.
<pre class="prettyprint">curl https://bootstrap.pypa.io/get-pip.py -o get-pip.py</pre>
<pre class="prettyprint">python get-pip.py</pre>
You might want to upgrade it:
<pre class="prettyprint">pip install -U pip</pre>
<h5 class="w3-center">Virtualenv</h5>
* Based on TensorFlow's <a href="https://www.tensorflow.org/install/pip?lang=python2#1-install-the-python-development-environment-on-your-system">introduction page</a>
<br></br>
Since there are many versions of Python and its packages, it is highly recommended to have an isolated Python environment: virtualenv is designed for this purpose.
<br></br>
You can have multiple virtualenvs on your machine, for example one for Python2, one for Python3.
<br></br>
It is also useful if you don't have root access to the global Python directory of your system (for example you are using a server).
<br></br>
Install virtualenv:
<pre class="prettyprint">pip install virtualenv</pre>
Create a virtual environment at directory ./venv:
<pre class="prettyprint">virtualenv --system-site-packages -p python2.7 ./venv</pre>
A full python environment is now installed at ./venv. If you take a look inside, you will see directories such as bin/lib.
<br></br>
To use the virtual environment you just created:
<pre class="prettyprint">source ./venv/bin/activate</pre>
You will see a little prefix (venv) to that the environment is activated.
<br></br>
Update pip:
<pre class="prettyprint">(venv) # pip install -U pip</pre>
To exit:
<pre class="prettyprint">(venv) # deactivate</pre>
<h5 class="w3-center">Anaconda</h5>
<a href="https://www.anaconda.com/">Anaconda</a> is a software package manager that provides a lot of popular tools for data science tasks.
Download the <a href="https://www.anaconda.com/download/">installer</a> that is suitable for your system.
For Linux system,
<pre class="prettyprint"># bash ./Anaconda2-5.3.0-Linux-x86_64.sh</pre>
After installation, you might want to create a virtual environment too:
<pre class="prettyprint">conda create -n venv pip python=2.7</pre>
<pre class="prettyprint">source activate venv</pre>
<h5 class="w3-center">Docker</h5>
<!-- ./docker_run.sh -u root -ti --rm ubuntu:18.04 bash -->
"With <a href="https://docs.docker.com/get-started/">Docker</a>, you can just grab a portable Python runtime as an image, no installation necessary. Then, your build can include the base Python image right alongside your app code, ensuring that your app, its dependencies, and the runtime, all travel together."
<br></br>
<img src="https://www.docker.com/sites/default/files/Container%402x.png" class="w3-center" style="max-width:200px"></img>
A good <a href="https://docs.docker.com/get-started/part2/">tutorial</a> to understand the basics.
<h5 class="w3-center">IPython</h5>
IPython is a powerful interactive shell for Python. You will have prettier printing and flexible commands.
<br></br>
To install,
<pre class="prettyprint">pip install ipython</pre>
To run,
<pre class="prettyprint">ipython</pre>
<h5 class="w3-center">Jupyter</h5>
<a href="http://jupyter.org/">Jupyter</a> notebook is a browser based convenient interactive tool for developing Python applications.
<br></br>
To install,
<pre class="prettyprint">pip install jupyter</pre>
To run,
<pre class="prettyprint">jupyter notebook</pre>
A new tab will be opened in your web browser. You can start a new notebook or open the ones you downloaded from the Internet.
</div>
</div>
<!-- Install PyTorch -->
<div class="w3-container" id="PT">
<div class="w3-content" style="max-width:700px">
<h5 class="w3-center w3-padding-64"><span class="w3-tag w3-wide">PyTorch</span></h5>
<a href="https://pytorch.org/">PyTorch</a> is one of the most popular deep learning framework now. It is powerful, flexible, extensible, and easy to use. A nice intro to it can be found <a href="https://blog.algorithmia.com/exploring-the-deep-learning-framework-pytorch/">here</a>.
<br></br>
<h5 class="w3-center">Installation</h5>
Follow the <a href="https://pytorch.org/get-started/locally/">configuration page</a> to pick the installation that is suitable for you. Then run the command at the bottom of that page.
<br></br></br>
For example, if you chose Stable/Linux/Pip/Python2.7/CUDA-None,
<pre class="prettyprint">pip install http://download.pytorch.org/whl/cpu/torch-0.4.1-cp27-cp27mu-linux_x86_64.whl</pre>
<pre class="prettyprint">pip install torchvision</pre>
<h5 class="w3-center">Get Started</h5>
To verify the installation, open an IPython shell:
<pre class="prettyprint">
import torch
print(torch.__version__)
</pre>
A bare-bone learning example in PyTorch (from <a href="https://pytorch.org/tutorials/beginner/examples_tensor/two_layer_net_tensor.html#pytorch-tensors">here</a>):
<pre class="prettyprint lang-py">
import torch
dtype = torch.float
device = torch.device("cpu")
# device = torch.device("cuda:0") # Uncomment this to run on GPU
# N is batch size; D_in is input dimension;
# H is hidden dimension; D_out is output dimension.
N, D_in, H, D_out = 64, 1000, 100, 10
# Create random input and output data
x = torch.randn(N, D_in, device=device, dtype=dtype)
y = torch.randn(N, D_out, device=device, dtype=dtype)
# Randomly initialize weights
w1 = torch.randn(D_in, H, device=device, dtype=dtype)
w2 = torch.randn(H, D_out, device=device, dtype=dtype)
learning_rate = 1e-6
for t in range(500):
# Forward pass: compute predicted y
h = x.mm(w1)
h_relu = h.clamp(min=0)
y_pred = h_relu.mm(w2)
# Compute and print loss
loss = (y_pred - y).pow(2).sum().item()
print(t, loss)
# Backprop to compute gradients of w1 and w2 with respect to loss
grad_y_pred = 2.0 * (y_pred - y)
grad_w2 = h_relu.t().mm(grad_y_pred)
grad_h_relu = grad_y_pred.mm(w2.t())
grad_h = grad_h_relu.clone()
grad_h[h < 0] = 0
grad_w1 = x.t().mm(grad_h)
# Update weights using gradient descent
w1 -= learning_rate * grad_w1
w2 -= learning_rate * grad_w2
</pre>
<h5 class="w3-center">A Real Example</h5>
<a href="https://pytorch.org/tutorials/beginner/blitz/cifar10_tutorial.html">Training A Cifar10 Image Classifier</a>
<h5 class="w3-center">Next Step</h5>
<a href="https://pytorch.org/tutorials/">Tutorials</a> and
<a href="https://github.com/pytorch/examples">Github Examples</a>
</div>
</div>
<!-- Install TensorFlow -->
<div class="w3-container" id="TK">
<div class="w3-content" style="max-width:700px">
<h5 class="w3-center w3-padding-64"><span class="w3-tag w3-wide">TensorFlow and Keras</span></h5>
<a href="https://www.tensorflow.org/">TensorFlow</a> is developed by Google that "allows easy deployment of computation across a variety of platforms (CPUs, GPUs, TPUs), and from desktops to clusters of servers to mobile and edge devices."
<br></br>
<a href="https://keras.io/">Keras</a> is a high-level interface that people found very useful in using TensorFlow in practice.
<h5 class="w3-center">Installation</h5>
If you have pip and virtualenv ready, it is easy to install TensorFlow by running this command:
<pre class="prettyprint">pip install --upgrade tensorflow</pre>
There is a built-in Keras inside TensorFlow, but if you want to use it for other backends such as Theano, you can install it too:
<pre class="prettyprint">pip install keras</pre>
<h5 class="w3-center">Get Started</h5>
To verify the installation,
<pre class="prettyprint">
# TensorFlow and tf.keras
import tensorflow as tf
from tensorflow import keras
print(tf.__version__)
</pre>
A bare-bone TensorFlow code snippets:
<pre class="prettyprint">
import tensorflow as tf
a = tf.constant(5.0)
b = tf.constant(6.0)
c = a*b
with tf.Session() as sess:
print(sess.run(c))
print(c.eval())
</pre>
<h5 class="w3-center">A Keras Example</h5>
<pre class="prettyprint">
import tensorflow as tf
mnist = tf.keras.datasets.mnist
(x_train, y_train),(x_test, y_test) = mnist.load_data()
x_train, x_test = x_train / 255.0, x_test / 255.0
model = tf.keras.models.Sequential([
tf.keras.layers.Flatten(input_shape=(28,28)),
tf.keras.layers.Dense(512, activation=tf.nn.relu),
tf.keras.layers.Dropout(0.2),
tf.keras.layers.Dense(10, activation=tf.nn.softmax)
])
model.compile(optimizer='adam',
loss='sparse_categorical_crossentropy',
metrics=['accuracy'])
model.fit(x_train, y_train, epochs=5)
model.evaluate(x_test, y_test)
</pre>
<h5 class="w3-center">Next Step</h5>
<a href="https://www.tensorflow.org/tutorials/">Tutorials</a>
</div>
</div>
<!-- Use Popular Tools -->
<div class="w3-container" id="tools">
<div class="w3-content" style="max-width:700px">
<h5 class="w3-center w3-padding-64"><span class="w3-tag w3-wide">Useful Tools</span></h5>
<a href="https://www.jetbrains.com/pycharm/"><strong>PyCharm</strong></a> is a powerful Integrated Development Environment (IDE) for Python.
<img src="https://www.jetbrains.com/pycharm/img/screenshots/complexLook@2x.jpg" style="max-width:700px"></img>
<br></br>
<a href="https://matplotlib.org/"><strong>Matplotlib</strong></a> is a plotting library for Python. It is a convenient tool in the interactive shells to quickly visualize your data. You can import it as:
<pre class="prettyprint">
import matplotlib.pyplot as plt
import numpy as np
x = np.random.randn(100)
plt.plot(x)
plt.show()
</pre>
<a href="https://www.tensorflow.org/guide/summaries_and_tensorboard"><strong>Tensorboard</strong></a> is part of the TensorFlow tool chain that people found very useful. You can now use it in Pytorch:
<a href="https://github.com/lanpa/tensorboardX">TensorboardX</a>
<img src="https://github.com/lanpa/tensorboardX/raw/master/screenshots/Demo.gif" style="max-width:700px"></img>
<a href="https://colab.research.google.com/notebooks/welcome.ipynb"><strong>Google Colab</strong></a> is a cloud based *FREE* Jupyter notebook environment that you can just run on your browser from anywhere. It is very useful to learn basics of DL without too much distractions from system specifics.
</div>
</div>
<!-- End page content -->
</div>
<!-- Footer -->
<footer class="w3-center w3-light-grey w3-padding-48 w3-large">
<h4><a href="./index.html">Deep Learning Clinic</a></h4>
<p>Powered by <a href="https://www.w3schools.com/w3css/default.asp" title="W3.CSS" target="_blank" class="w3-hover-text-green">w3.css</a></p>
</footer>
</body>
</html>