File size: 4,662 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 | 1
00:00:00,430 --> 00:00:00,850
OK.
2
00:00:00,880 --> 00:00:04,340
So let's talk about how we save our models that we just trained.
3
00:00:04,380 --> 00:00:08,020
We spend some time training a model and we don't want to lose this model in clothes.
4
00:00:08,040 --> 00:00:09,000
I buy the book.
5
00:00:09,000 --> 00:00:13,350
So let's figure out how we save and load our models.
6
00:00:13,390 --> 00:00:15,790
OK so we just spend so much time printing our model.
7
00:00:15,810 --> 00:00:17,890
So let's get to saving it now.
8
00:00:17,930 --> 00:00:24,780
So if you scroll down below these charts to see if she's part of the model object you just specified
9
00:00:24,830 --> 00:00:27,000
directory and file him as you want.
10
00:00:27,000 --> 00:00:33,570
So basically we can just call this chapter 8 which is from 8 here eminence simple CNN Tennie box can
11
00:00:33,570 --> 00:00:36,080
actually put on display here make it look a little bit nicer.
12
00:00:36,510 --> 00:00:39,780
And voila it saved instantly.
13
00:00:39,780 --> 00:00:46,770
So now if you want to redo it on what all this is into and you found him here and we use Chris's model
14
00:00:46,890 --> 00:00:48,720
which is part of the models library here.
15
00:00:49,200 --> 00:00:55,890
And this basically this is what's strange to most people loading isn't as instantaneous as saving and
16
00:00:55,890 --> 00:01:00,990
that's because it has to actually unpack everything now and put it back into the model format.
17
00:01:01,020 --> 00:01:06,150
It's a bit tricky to understand I actually spend some time trying to figure it out because I had a project
18
00:01:06,150 --> 00:01:11,840
where I was loading 50 different models at once and it was taking maybe about five 10 minutes to live
19
00:01:11,840 --> 00:01:13,130
with all of them together.
20
00:01:13,530 --> 00:01:14,940
So let's just run this quickly.
21
00:01:15,600 --> 00:01:16,510
And there we go.
22
00:01:16,520 --> 00:01:17,930
Stick what 15 seconds.
23
00:01:18,340 --> 00:01:22,860
And so there's no snow or model which was called the model before.
24
00:01:22,890 --> 00:01:24,280
It's called classifier.
25
00:01:24,690 --> 00:01:30,270
And now we can use it here classify it just like we were using model before they are interchangeable
26
00:01:30,270 --> 00:01:33,080
now because our model is effectively classified.
27
00:01:33,080 --> 00:01:34,660
It's just a different name.
28
00:01:34,680 --> 00:01:40,510
What I mean by that is you see how we declared this as a model equals sequential No what we've related
29
00:01:40,510 --> 00:01:41,220
to our model.
30
00:01:41,220 --> 00:01:44,160
I didn't want to overwrite what we had see if it was a model.
31
00:01:44,250 --> 00:01:48,160
So effectively model is now as we have seen it and loaded it.
32
00:01:48,180 --> 00:01:52,730
It's now effectively Plus a classified called classify.
33
00:01:52,770 --> 00:01:53,600
Should say.
34
00:01:53,880 --> 00:01:57,630
And we can load it here and basically run some tests.
35
00:01:57,640 --> 00:01:58,800
This has an error in it.
36
00:01:58,800 --> 00:02:01,280
Him just to leave this quickly.
37
00:02:02,560 --> 00:02:04,940
And it should work fine now.
38
00:02:04,980 --> 00:02:05,490
There we go.
39
00:02:05,570 --> 00:02:07,940
So we're actually running some test animals here.
40
00:02:07,940 --> 00:02:09,630
So this is a zero a tree.
41
00:02:09,650 --> 00:02:13,740
It can see how accurate this model is.
42
00:02:13,760 --> 00:02:14,530
It's quite cool.
43
00:02:15,500 --> 00:02:19,050
And it actually looks and what it actually is because it's a bit tilted.
44
00:02:19,430 --> 00:02:20,130
So that's good.
45
00:02:20,200 --> 00:02:20,660
It's great.
46
00:02:20,660 --> 00:02:27,740
Actually we've just trained we've just loaded and trained and tested and plotted and analyzed and done
47
00:02:27,800 --> 00:02:30,470
all of the settings for this model.
48
00:02:30,470 --> 00:02:34,940
So if you wanted to put it all together because you don't actually need to have every block separate
49
00:02:34,940 --> 00:02:40,130
like this this was basically for teaching purposes I actually never separate these things that much
50
00:02:42,060 --> 00:02:47,780
but let's put it all together so quickly you can see this is everything here and you run this and you
51
00:02:47,780 --> 00:02:49,990
basically start training instantaneously.
|