1
00:00:00,000 --> 00:00:00,000
Hello guys.

2
00:00:00,000 --> 00:00:04,000
So in this video I'm going to talk about NLP in deep learning.

3
00:00:04,000 --> 00:00:11,000
Till now if I talk about machine learning with respect to NLP, we have discussed about various topics,

4
00:00:11,000 --> 00:00:19,000
you know, and all these topics were involved wherein we were specifically working with text data,

5
00:00:19,000 --> 00:00:19,000
right?

6
00:00:19,000 --> 00:00:26,000
And with respect to this particular text data, if we really wanted to convert this into some vectors,

7
00:00:26,000 --> 00:00:29,000
which is some numerical representation, right?

8
00:00:30,000 --> 00:00:32,000
Numerical representation.

9
00:00:32,000 --> 00:00:35,000
We used various techniques.

10
00:00:36,000 --> 00:00:41,000
The first technique, uh, if you remember, in NLP, in machine learning we discussed about one hot

11
00:00:41,000 --> 00:00:42,000
encoding.

12
00:00:42,000 --> 00:00:43,000
Right.

13
00:00:43,000 --> 00:00:47,000
Then we went up with Bag of Words.

14
00:00:47,000 --> 00:00:51,000
And the third topic that we discussed was with respect to TF-IDF.

15
00:00:51,000 --> 00:00:52,000
Right.

16
00:00:52,000 --> 00:00:56,000
Coming to the fourth one, we discussed about word two vec.

17
00:00:57,000 --> 00:01:01,000
Along with this, we also saw about average word two vec.

18
00:01:02,000 --> 00:01:05,000
And we saw a lot of practical application.

19
00:01:05,000 --> 00:01:13,000
Uh, we took a specific data set and we tried to solve problems like sentiment analysis, sentiment

20
00:01:13,000 --> 00:01:16,000
analysis and text classification.

21
00:01:17,000 --> 00:01:17,000
Right.

22
00:01:17,000 --> 00:01:20,000
We did this kind of task.

23
00:01:20,000 --> 00:01:23,000
Text classification in practicals.

24
00:01:23,000 --> 00:01:23,000
Okay.

25
00:01:25,000 --> 00:01:32,000
Now, now when we move one step ahead, that is nothing but NLP in deep learning.

26
00:01:33,000 --> 00:01:36,000
Now, as you all know, this text data are nothing.

27
00:01:36,000 --> 00:01:38,000
But these are sequential data.

28
00:01:38,000 --> 00:01:42,000
And if I just give you a brief information about an.

29
00:01:43,000 --> 00:01:49,000
We learned about artificial neural network and this artificial neural network.

30
00:01:50,000 --> 00:01:52,000
Let me just go ahead and write it down again.

31
00:01:53,000 --> 00:01:55,000
This artificial neural network.

32
00:01:57,000 --> 00:02:06,000
Was used to solve various problem statements, specifically supervised problem statements like classification

33
00:02:06,000 --> 00:02:08,000
and regression.

34
00:02:09,000 --> 00:02:09,000
Right.

35
00:02:09,000 --> 00:02:13,000
It was used to solve classification and regression.

36
00:02:14,000 --> 00:02:21,000
Now, when we use this classification or regression here, you can consider that I had some kind of

37
00:02:21,000 --> 00:02:21,000
data.

38
00:02:22,000 --> 00:02:25,000
And this data was nothing but it was a tabular data.

39
00:02:25,000 --> 00:02:27,000
What do I mean by tabular data.

40
00:02:28,000 --> 00:02:31,000
So this was nothing, but it was a kind of tabular data.

41
00:02:32,000 --> 00:02:35,000
Now when I say tabular data, let's consider an example.

42
00:02:35,000 --> 00:02:38,000
Suppose I have a house prediction.

43
00:02:39,000 --> 00:02:43,000
House price prediction data set okay.

44
00:02:43,000 --> 00:02:50,000
In this specific data set you will be seeing that I may have some features like what is my house size,

45
00:02:51,000 --> 00:02:57,000
what is my number of rooms in the house and what is the price of the house?

46
00:02:58,000 --> 00:03:01,000
So when we have this kind of tabular data.

47
00:03:02,000 --> 00:03:05,000
Okay, I'll be having a lot of key points over here.

48
00:03:06,000 --> 00:03:10,000
I will be considering this as my output variable, my dependent variable.

49
00:03:10,000 --> 00:03:15,000
And then I can probably go ahead and solve a regression problem statement for this because price is

50
00:03:15,000 --> 00:03:16,000
a continuous value.

51
00:03:16,000 --> 00:03:21,000
If my output feature is a categorical value, then I can probably solve it with the help of classification.

52
00:03:21,000 --> 00:03:26,000
Over here, all these features that you will be seeing, right?

53
00:03:26,000 --> 00:03:31,000
If I probably change the order of this particular feature, let's say.

54
00:03:31,000 --> 00:03:35,000
And for this right, if I really want to solve with the help of artificial neural network, since I

55
00:03:35,000 --> 00:03:39,000
have two inputs over here, I may go ahead and create this kind of Ann.

56
00:03:39,000 --> 00:03:39,000
Right.

57
00:03:39,000 --> 00:03:43,000
So let's say this is my input layer where I will be giving my two features.

58
00:03:43,000 --> 00:03:45,000
Let's consider this is my f one.

59
00:03:45,000 --> 00:03:48,000
This is my f two and this is my y right.

60
00:03:48,000 --> 00:03:51,000
So here I will be giving my f one feature.

61
00:03:51,000 --> 00:03:52,000
I'll give my f two feature.

62
00:03:52,000 --> 00:03:56,000
And then along with this I may have some hidden layers.

63
00:03:56,000 --> 00:03:58,000
Let's say this is my first hidden layer.

64
00:03:58,000 --> 00:04:06,000
And as discussed all this input feature will be connected to this particular hidden layer over here,

65
00:04:06,000 --> 00:04:08,000
over here and over here.

66
00:04:08,000 --> 00:04:09,000
Right.

67
00:04:09,000 --> 00:04:13,000
And let's say this is my output feature and this will also get connected.

68
00:04:14,000 --> 00:04:14,000
Right.

69
00:04:14,000 --> 00:04:16,000
And finally I get my y hat.

70
00:04:17,000 --> 00:04:23,000
So in the end what we do is that we give this feature f1 and f2.

71
00:04:23,000 --> 00:04:25,000
And then we initialize some weights.

72
00:04:25,000 --> 00:04:31,000
Let's say this is W1W2W3W4, W5W6 weights.

73
00:04:31,000 --> 00:04:35,000
And here also we have W7W8W9.

74
00:04:35,000 --> 00:04:35,000
Right.

75
00:04:35,000 --> 00:04:41,000
So based on this particular weights we do forward propagation okay.

76
00:04:41,000 --> 00:04:43,000
And then we do backward propagation.

77
00:04:43,000 --> 00:04:48,000
Now in the forward propagation what we do we basically take the input multiply the weights.

78
00:04:48,000 --> 00:04:51,000
And then along with this we also add a bias.

79
00:04:51,000 --> 00:04:52,000
So let's say this is b one.

80
00:04:52,000 --> 00:04:55,000
This is b two and this is b three.

81
00:04:55,000 --> 00:04:57,000
And then we apply an activation function.

82
00:04:57,000 --> 00:05:00,000
And similarly we do all the steps in the all the nodes.

83
00:05:00,000 --> 00:05:02,000
And finally we get the y hat.

84
00:05:02,000 --> 00:05:05,000
Then we calculate our loss.

85
00:05:05,000 --> 00:05:07,000
Loss is nothing but y minus y hat.

86
00:05:07,000 --> 00:05:11,000
Let's consider this as an example y of I minus y of I hat.

87
00:05:11,000 --> 00:05:15,000
And then based on this particular loss, our main aim is to reduce this loss.

88
00:05:15,000 --> 00:05:18,000
So in order to reduce this loss we need to change all these weights.

89
00:05:18,000 --> 00:05:21,000
And that usually happens in the back propagation.

90
00:05:22,000 --> 00:05:25,000
And that is how we basically solve this problem.

91
00:05:25,000 --> 00:05:32,000
Now one very important point in an is that if I interchange this feature, let's say that if I change

92
00:05:32,000 --> 00:05:36,000
the order of this feature, let's say that if I initially I had this f one, f two, and y.

93
00:05:36,000 --> 00:05:40,000
Let's say my first feature was house price and the second feature was number of room.

94
00:05:40,000 --> 00:05:46,000
Even though I make this feature as F2F1, then also we will be able to train this particular model with

95
00:05:46,000 --> 00:05:47,000
the help of Ann.

96
00:05:47,000 --> 00:05:53,000
Now, one very important information about tabular data is that here tabular data you have features.

97
00:05:53,000 --> 00:05:57,000
You have some values like how we have our data in our CSV format and all.

98
00:05:57,000 --> 00:06:02,000
And the second thing is that over here your sequence of data does not matter okay.

99
00:06:02,000 --> 00:06:04,000
So this is very much important point.

100
00:06:04,000 --> 00:06:09,000
Your sequence of data does not matter.

101
00:06:10,000 --> 00:06:12,000
That basically means you can change the sequence right?

102
00:06:12,000 --> 00:06:13,000
While training it.

103
00:06:13,000 --> 00:06:16,000
You can keep F2 first, F1 first.

104
00:06:16,000 --> 00:06:23,000
Let's say if you have F3 feature again, keep f2 f3 first because at a time this entire row will be

105
00:06:23,000 --> 00:06:24,000
sent for training, right?

106
00:06:24,000 --> 00:06:26,000
So here we are giving our F1 and F2.

107
00:06:26,000 --> 00:06:28,000
So this will be for my first row.

108
00:06:28,000 --> 00:06:30,000
Then for my second row right.

109
00:06:30,000 --> 00:06:35,000
Based on the batch size we will be giving this entire data and doing the forward and the backward propagation.

110
00:06:36,000 --> 00:06:37,000
Now this is fine right.

111
00:06:37,000 --> 00:06:39,000
And this is with respect to an.

112
00:06:40,000 --> 00:06:46,000
Now if I consider the second example that we have seen already in our session, that is nothing.

113
00:06:46,000 --> 00:06:47,000
But it is about CNN.

114
00:06:47,000 --> 00:06:54,000
CNN basically means convolution convolutional neural network.

115
00:06:56,000 --> 00:06:57,000
Okay.

116
00:06:57,000 --> 00:07:01,000
And here with respect to convolutional neural network you use this for.

117
00:07:02,000 --> 00:07:06,000
So basically over here your data will be in the form of images.

118
00:07:06,000 --> 00:07:08,000
It can be in the form of images.

119
00:07:08,000 --> 00:07:11,000
It can be in the form of video frames.

120
00:07:11,000 --> 00:07:12,000
Right.

121
00:07:13,000 --> 00:07:17,000
And here most of the tasks that you specifically do right.

122
00:07:17,000 --> 00:07:22,000
Some of the examples are like image classification.

123
00:07:22,000 --> 00:07:23,000
right?

124
00:07:23,000 --> 00:07:25,000
You have image classification.

125
00:07:25,000 --> 00:07:27,000
Let's say you want to do object detection.

126
00:07:27,000 --> 00:07:31,000
So you use some higher variants of CNN like our CNN.

127
00:07:32,000 --> 00:07:33,000
You know you have YOLO.

128
00:07:33,000 --> 00:07:35,000
All this kind of algorithms you have.

129
00:07:35,000 --> 00:07:36,000
Right.

130
00:07:36,000 --> 00:07:39,000
So this is kind of task you will be able to do with the help of CNN.

131
00:07:39,000 --> 00:07:40,000
Okay.

132
00:07:41,000 --> 00:07:42,000
Now this is fine.

133
00:07:42,000 --> 00:07:44,000
And this is also fine over here.

134
00:07:44,000 --> 00:07:47,000
Uh, now let's talk about the third thing.

135
00:07:47,000 --> 00:07:48,000
And and that is where we are going to come up with.

136
00:07:48,000 --> 00:07:50,000
That is nothing but an LP in deep learning.

137
00:07:50,000 --> 00:07:59,000
The third thing that we will be specifically talking about is for a data that is of sequential type,

138
00:08:01,000 --> 00:08:02,000
sequential type.

139
00:08:03,000 --> 00:08:07,000
Or I can also say this is nothing, but this is my sequential data.

140
00:08:08,000 --> 00:08:11,000
Now what exactly is sequential data?

141
00:08:11,000 --> 00:08:14,000
Some of the examples that I will be considering over here.

142
00:08:14,000 --> 00:08:23,000
Let's say this is one of the example that I will say, um, I have a data set which involves some kind

143
00:08:23,000 --> 00:08:24,000
of text.

144
00:08:24,000 --> 00:08:25,000
Okay.

145
00:08:25,000 --> 00:08:29,000
And uh, or let me just take one example over here.

146
00:08:29,000 --> 00:08:31,000
I'll just go ahead and write it down.

147
00:08:31,000 --> 00:08:33,000
So let's take this example.

148
00:08:33,000 --> 00:08:41,000
Text generation Now text generation is a kind of application where I will be having some kind of input.

149
00:08:41,000 --> 00:08:42,000
Right.

150
00:08:42,000 --> 00:08:56,000
Let's say if I say hey, how are or I'll say this is a apple juice right now, juice, I'm not giving

151
00:08:56,000 --> 00:08:57,000
the word right.

152
00:08:57,000 --> 00:09:02,000
So what my model should be able to do is that it should be able to generate this specific word.

153
00:09:02,000 --> 00:09:07,000
It can say, hey, it is an apple juice or it is an apple pudding something.

154
00:09:07,000 --> 00:09:07,000
Right?

155
00:09:07,000 --> 00:09:09,000
So it should be generated.

156
00:09:09,000 --> 00:09:11,000
It should be able to generate this particular output.

157
00:09:11,000 --> 00:09:16,000
Now in this kind of scenario, the data that you specifically have is a sequential data.

158
00:09:16,000 --> 00:09:22,000
The sequence is very much important to understand the context of this sentence.

159
00:09:22,000 --> 00:09:23,000
Right.

160
00:09:23,000 --> 00:09:26,000
So this kind of data is basically called as sequential data.

161
00:09:26,000 --> 00:09:27,000
Let's take one more example.

162
00:09:28,000 --> 00:09:32,000
The second example that I will be probably taking is nothing but your chat bot conversation.

163
00:09:34,000 --> 00:09:41,000
Now in a chat bot conversation, what you do is that you basically have a queue and a chat bot right

164
00:09:41,000 --> 00:09:48,000
now in Q and a chat bot, let's say if a user is asking any question and based on this question, the

165
00:09:48,000 --> 00:09:50,000
chat chatbot should be able to give an answer.

166
00:09:50,000 --> 00:09:56,000
So here, both on this question and answer, the sequence of information is very much important.

167
00:09:56,000 --> 00:09:58,000
The sequence of text is very much important.

168
00:09:58,000 --> 00:10:03,000
And based on the question since the chatbot is answering, the context of the question is also very

169
00:10:03,000 --> 00:10:04,000
much important.

170
00:10:04,000 --> 00:10:05,000
Okay.

171
00:10:05,000 --> 00:10:08,000
So that is where you can consider this as a chatbot.

172
00:10:08,000 --> 00:10:14,000
Conversation is another one example where our data set will be a sequential data itself, right?

173
00:10:14,000 --> 00:10:17,000
Because see what what is a sequential data?

174
00:10:17,000 --> 00:10:19,000
I'll just give you a basic definition.

175
00:10:20,000 --> 00:10:20,000
Okay.

176
00:10:20,000 --> 00:10:22,000
Sequential data.

177
00:10:24,000 --> 00:10:26,000
A sequential data is something like this.

178
00:10:26,000 --> 00:10:32,000
If I take one example I write, I go and say, hey, the food is good, right?

179
00:10:34,000 --> 00:10:40,000
Now when I give this and when I see this data right here, the sequence of this data is very much important

180
00:10:40,000 --> 00:10:46,000
because based on the sequence, the the sentence is meaning making some meaningful meaning over here.

181
00:10:46,000 --> 00:10:47,000
Right.

182
00:10:47,000 --> 00:10:56,000
If I change the order, if I just go ahead and say, hey, good, the food is if I go ahead and write

183
00:10:56,000 --> 00:11:01,000
this kind of conversation and if I probably see the sentence over here, right over here, the see sea.

184
00:11:01,000 --> 00:11:05,000
Based on the sequence of these words, the meaning may change, right?

185
00:11:05,000 --> 00:11:07,000
The meaning may change.

186
00:11:08,000 --> 00:11:09,000
Right.

187
00:11:09,000 --> 00:11:14,000
So this is just one example of sequential data, where ordering of this word is very much important

188
00:11:14,000 --> 00:11:17,000
because it makes sense of the entire sentence itself.

189
00:11:17,000 --> 00:11:18,000
Okay.

190
00:11:18,000 --> 00:11:23,000
So if I probably consider one more example with respect to the sequential data, one more application,

191
00:11:24,000 --> 00:11:25,000
uh, let's go ahead and see.

192
00:11:25,000 --> 00:11:28,000
This is nothing, but this is called as language translation.

193
00:11:29,000 --> 00:11:31,000
Language translation.

194
00:11:31,000 --> 00:11:34,000
Now see, in language translation, what do we exactly do?

195
00:11:34,000 --> 00:11:37,000
Let's say I have a text in English.

196
00:11:38,000 --> 00:11:41,000
I want to convert this into French.

197
00:11:41,000 --> 00:11:42,000
Okay.

198
00:11:43,000 --> 00:11:44,000
I want to convert this into French.

199
00:11:44,000 --> 00:11:47,000
And I hope everybody has seen Google translation.

200
00:11:47,000 --> 00:11:52,000
As soon as I keep on writing the word in English, automatically, you'll be able to see that it'll

201
00:11:52,000 --> 00:11:54,000
be able to generate the sentence in French.

202
00:11:54,000 --> 00:11:58,000
So over here also sequence of information is very much necessary.

203
00:11:58,000 --> 00:12:00,000
Okay, one more example.

204
00:12:00,000 --> 00:12:02,000
I'll be giving you tons of examples.

205
00:12:02,000 --> 00:12:05,000
Now this is nothing but auto suggestion.

206
00:12:05,000 --> 00:12:15,000
I hope everybody has been using LinkedIn has been using Gmail right over there Whenever you are writing

207
00:12:15,000 --> 00:12:22,000
a mail, whenever you are writing a message, you will be able to get auto suggestion with respect to

208
00:12:22,000 --> 00:12:25,000
the sentences with respect to the mistakes that you do.

209
00:12:25,000 --> 00:12:31,000
Okay, so this is also one very good example, the fifth kind of example that I'm actually going to

210
00:12:31,000 --> 00:12:32,000
take.

211
00:12:32,000 --> 00:12:34,000
Let's consider my sales data.

212
00:12:34,000 --> 00:12:37,000
Now what is so amazing about this sales data.

213
00:12:37,000 --> 00:12:42,000
Your sales data is based on date time, right?

214
00:12:42,000 --> 00:12:51,000
So if I go ahead and plot this entire data, let's say this is my date and this is my sales, right?

215
00:12:51,000 --> 00:12:53,000
So I may have this kind of graph.

216
00:12:54,000 --> 00:12:54,000
Right.

217
00:12:54,000 --> 00:12:56,000
It may be increasing or it may be decreasing.

218
00:12:56,000 --> 00:12:57,000
Right.

219
00:12:57,000 --> 00:12:59,000
So this is basically my sales data.

220
00:12:59,000 --> 00:12:59,000
Right.

221
00:12:59,000 --> 00:13:03,000
And over here also this is a sequence of information.

222
00:13:03,000 --> 00:13:04,000
Right.

223
00:13:04,000 --> 00:13:04,000
Why I'm saying you.

224
00:13:04,000 --> 00:13:10,000
Because one of the use cases that you'll be seeing is something called a sales forecasting.

225
00:13:10,000 --> 00:13:15,000
And probably we can solve this particular use case by just taking the previous timestamp.

226
00:13:15,000 --> 00:13:20,000
And we need to predict the future future date what will be the sales.

227
00:13:21,000 --> 00:13:21,000
Right.

228
00:13:21,000 --> 00:13:25,000
So that kind of forecasting can also be done okay.

229
00:13:25,000 --> 00:13:30,000
So in short I've given you so many different examples to just make you understand about sequential data.

230
00:13:30,000 --> 00:13:40,000
Now, in order to work with a neural network and to or in order to work with sorry, in order to work

231
00:13:40,000 --> 00:13:46,000
with this kind of sequential data and probably use a neural network we cannot directly use.

232
00:13:46,000 --> 00:13:47,000
Okay, here I'll write.

233
00:13:47,000 --> 00:13:55,000
Can we use an to solve this problem?

234
00:13:57,000 --> 00:13:58,000
Okay.

235
00:13:58,000 --> 00:14:01,000
Can we use an to solve this problem?

236
00:14:01,000 --> 00:14:02,000
Right.

237
00:14:02,000 --> 00:14:04,000
This is what I'm going to discuss in my next video.

238
00:14:04,000 --> 00:14:08,000
But I hope in this particular video you have understood, right?

239
00:14:08,000 --> 00:14:11,000
Since we are discussing about an LP in machine learning.

240
00:14:11,000 --> 00:14:17,000
Okay, sorry, an LP in deep learning, I had to give you a brief idea about what does exactly n n do

241
00:14:17,000 --> 00:14:18,000
c n n do?

242
00:14:18,000 --> 00:14:20,000
And then we discussed about sequential data.

243
00:14:20,000 --> 00:14:31,000
Now my next question is that can we use N to solve this problem which has sequential data which has

244
00:14:31,000 --> 00:14:32,000
sequential data.

245
00:14:32,000 --> 00:14:37,000
And this is what I am going to probably discuss about in my next video.

246
00:14:38,000 --> 00:14:44,000
Uh, then, uh, you know, while we are discussing this, I'll also give you a brief idea.

247
00:14:44,000 --> 00:14:45,000
Like what?

248
00:14:45,000 --> 00:14:47,000
All things we really need to learn in NLP, in deep learning.

249
00:14:47,000 --> 00:14:51,000
Okay, so just to give you an idea what all things to learn over here.

250
00:14:51,000 --> 00:14:59,000
So NLP in deep learning involves lot of amazing topic topics right.

251
00:15:00,000 --> 00:15:01,000
Lot of amazing topics.

252
00:15:01,000 --> 00:15:06,000
And this will turn lead to right now generative AI is in trend, right?

253
00:15:06,000 --> 00:15:08,000
Generative AI is in trend.

254
00:15:08,000 --> 00:15:09,000
Llms are in trend.

255
00:15:09,000 --> 00:15:10,000
Right.

256
00:15:10,000 --> 00:15:12,000
Multi models are in trend, right?

257
00:15:12,000 --> 00:15:17,000
If you really want to learn this right, this is the prerequisite.

258
00:15:17,000 --> 00:15:22,000
You really need to know each and every architecture with respect to all the neural networks that we

259
00:15:22,000 --> 00:15:26,000
specifically use for NLP techniques in deep learning.

260
00:15:26,000 --> 00:15:26,000
Okay.

261
00:15:27,000 --> 00:15:33,000
So first uh topic that we will be learning in NLP in deep learning is nothing, but it will be about

262
00:15:33,000 --> 00:15:34,000
simple RNN.

263
00:15:35,000 --> 00:15:36,000
Right?

264
00:15:36,000 --> 00:15:42,000
After learning simple RNN we will be seeing what is LSTM and GRU RNN.

265
00:15:43,000 --> 00:15:47,000
Okay, so this will be my next RNN that we will be discussing.

266
00:15:47,000 --> 00:15:52,000
Then after this you will be seeing that we will also be learning about bidirectional RNN.

267
00:15:53,000 --> 00:15:56,000
Bidirectional RNN.

268
00:15:57,000 --> 00:16:03,000
Then once we complete this, the next step will be that we will go ahead one step more and then we will

269
00:16:03,000 --> 00:16:08,000
be discussing about something called as encoders and decoders.

270
00:16:09,000 --> 00:16:10,000
Right.

271
00:16:10,000 --> 00:16:11,000
Encoder and decoder.

272
00:16:11,000 --> 00:16:17,000
And finally after doing this uh we'll be seeing about what is self-attention.

273
00:16:19,000 --> 00:16:27,000
And then we'll also be seeing transformers right now after we see all these things.

274
00:16:27,000 --> 00:16:32,000
Right now, most of the variants of LLM models that you will currently see in the generative AI field,

275
00:16:33,000 --> 00:16:36,000
specifically of different types of transformers, right.

276
00:16:36,000 --> 00:16:42,000
And that is where, uh, you will get a brief idea about how to work in generative AI, right?

277
00:16:42,000 --> 00:16:47,000
Specifically with the help of LLM models like, you know, about OpenAI LLM models.

278
00:16:47,000 --> 00:16:47,000
Right.

279
00:16:47,000 --> 00:16:49,000
So we'll be discussing this step by step.

280
00:16:49,000 --> 00:16:51,000
We'll be understanding the architecture.

281
00:16:51,000 --> 00:16:54,000
We'll be solving some problem statements and we'll be learning this okay.

282
00:16:55,000 --> 00:16:59,000
So this was a brief idea uh, about uh NLP in deep learning.

283
00:16:59,000 --> 00:17:06,000
And trust me guys, these are like a prerequisite to probably enter into the generative AI field and

284
00:17:06,000 --> 00:17:07,000
work with LLM models.

285
00:17:07,000 --> 00:17:11,000
Even in interviews, they still focus on all these kind of questions.

286
00:17:11,000 --> 00:17:13,000
So we will be solving step by step.

287
00:17:13,000 --> 00:17:17,000
Each and every thing will be discussing, will be understanding the theoretical intuition and will also

288
00:17:17,000 --> 00:17:20,000
be understanding the practical implementation.

289
00:17:20,000 --> 00:17:24,000
So yes, now in our next video we will try to answer this question.

290
00:17:24,000 --> 00:17:28,000
Can we use Ann to solve this problem statement with sequential data.

291
00:17:28,000 --> 00:17:28,000
Right.

292
00:17:28,000 --> 00:17:30,000
So yes this was it for my side.

293
00:17:30,000 --> 00:17:31,000
I will see you all in the next video.

294
00:17:31,000 --> 00:17:32,000
Thank you.