File size: 4,442 Bytes
d157f08
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
1
00:00:01,580 --> 00:00:07,310
And welcome to chapter 8 point 1 0 where we're actually going to use curus to display a visual output

2
00:00:07,460 --> 00:00:08,420
or model.

3
00:00:08,650 --> 00:00:12,950
So remember before previously I was drawing all this nice visualization of our model.

4
00:00:13,220 --> 00:00:18,640
Well carrots can actually do something not quite as nice as this but it produces a pretty decent model

5
00:00:18,660 --> 00:00:22,730
visualization that helps you basically show people and explain your model.

6
00:00:22,910 --> 00:00:24,240
So let's see how we do it.

7
00:00:24,290 --> 00:00:26,220
Let's go back to what I thought in the book.

8
00:00:26,500 --> 00:00:26,840
OK.

9
00:00:26,870 --> 00:00:32,630
So now we're about to visualize our model so to visualize our model we need to import this library from

10
00:00:32,680 --> 00:00:34,200
Carousel dysfunction.

11
00:00:34,280 --> 00:00:39,680
It's called plot model and it's found in Cara's utilities thought visualization utilities vid's underscore

12
00:00:39,680 --> 00:00:41,220
utilities for short.

13
00:00:41,240 --> 00:00:43,980
So what we do we create or recreate or model first.

14
00:00:44,030 --> 00:00:49,430
We don't necessarily have to do this but it's good practice just in case we didn't do it before and

15
00:00:49,450 --> 00:00:53,370
it previously as in the previous cells and this I Pitre notebook.

16
00:00:53,390 --> 00:00:59,800
So let's go ahead and run the slime on this block and we get this table which is our same model output

17
00:00:59,810 --> 00:01:00,610
from before.

18
00:01:00,770 --> 00:01:01,740
All right.

19
00:01:01,970 --> 00:01:04,370
This in itself is a pretty decent visualization.

20
00:01:04,370 --> 00:01:06,970
However it's not like a visual diagram.

21
00:01:07,220 --> 00:01:09,380
What we're going to do is produce a visual Vaga now.

22
00:01:09,440 --> 00:01:15,320
So to do this we use a plot model function and a plot modeled function basically takes a model that

23
00:01:15,340 --> 00:01:20,820
we find here we take we enter a pot for the file to be saved.

24
00:01:20,870 --> 00:01:25,550
So we just use this path here which is the way our train models are saved and we give it a phylum model

25
00:01:25,550 --> 00:01:30,720
and scale plot PNB we can be more descriptive and give it like this model.

26
00:01:30,740 --> 00:01:31,510
All right.

27
00:01:32,060 --> 00:01:34,860
And then after we use matplotlib to actually show.

28
00:01:34,880 --> 00:01:40,220
So we just point matplotlib here that image directory of where we see that and we plotted here and it

29
00:01:40,220 --> 00:01:42,140
comes up below right here.

30
00:01:42,200 --> 00:01:43,500
So let's see how this works.

31
00:01:43,520 --> 00:01:48,990
Let's run this block and then we go here's a nice model visualization.

32
00:01:49,040 --> 00:01:54,530
What's cool about this is that we have inputs and outputs coming in and out is actually quite nice.

33
00:01:54,530 --> 00:01:55,970
This is basically a random number.

34
00:01:55,970 --> 00:02:00,990
Never have been able to figure out what this number actually is pretty much ignored for now.

35
00:02:01,040 --> 00:02:01,820
What school is that.

36
00:02:01,820 --> 00:02:04,160
We actually have Olias coming in here.

37
00:02:04,280 --> 00:02:07,780
We have all kinds of layers tool is here Max pooling.

38
00:02:07,840 --> 00:02:13,050
It shows input outputs what dropout does doesn't change a thing just drops out some layers.

39
00:02:13,250 --> 00:02:19,330
While training Flaten which we know what it does now are dense connections here or drop out again and

40
00:02:19,330 --> 00:02:21,130
are fully connectedly the end here.

41
00:02:21,200 --> 00:02:23,390
Which is also for the connectedly here.

42
00:02:23,930 --> 00:02:25,260
And basically this is it.

43
00:02:25,340 --> 00:02:30,190
So if you were to go to this directory here let's go to the planning directory.

44
00:02:30,290 --> 00:02:32,110
Let's go to train models.

45
00:02:32,330 --> 00:02:41,420
We can see here it is and it's clearer and sharper than I in the book being a DNG file and that's it.

46
00:02:41,420 --> 00:02:43,130
So we've just successfully saved.