1
00:00:00,000 --> 00:00:06,000
Uh, now, as you know, from the past two years, like from 2022 and until now, right?

2
00:00:06,000 --> 00:00:12,000
Generative AI is really the talk of the town, you know, and, uh, with respect to generative AI,

3
00:00:12,000 --> 00:00:17,000
if I talk about large language models, open source, large language models, open AI, doing some such

4
00:00:17,000 --> 00:00:18,000
an amazing word.

5
00:00:18,000 --> 00:00:21,000
Now cloudy three is also there to compete.

6
00:00:21,000 --> 00:00:24,000
Open AI, you know, and Amazon is definitely supporting them.

7
00:00:24,000 --> 00:00:25,000
So you should know right?

8
00:00:25,000 --> 00:00:27,000
What exactly is generative AI?

9
00:00:27,000 --> 00:00:30,000
How is it different from all these terms?

10
00:00:30,000 --> 00:00:34,000
Or what is the exact difference between AI versus ML versus DL versus generative AI?

11
00:00:35,000 --> 00:00:38,000
So I'll make you understand in this specific video.

12
00:00:38,000 --> 00:00:42,000
Probably I'll try to make this video from somewhere between 15 to 20 minutes.

13
00:00:42,000 --> 00:00:49,000
Um, I'll be including all simplistic terms, very easy definitions, so that you'll be able to understand

14
00:00:49,000 --> 00:00:51,000
it, because once you have that clarity.

15
00:00:51,000 --> 00:00:56,000
Right, uh, at the end of the day, right, you really want to probably become an AI engineer, you

16
00:00:56,000 --> 00:01:00,000
know, which uses all these technologies to create that AI apps.

17
00:01:00,000 --> 00:01:04,000
So, uh, let's go ahead and let's understand this thing.

18
00:01:04,000 --> 00:01:04,000
Okay.

19
00:01:04,000 --> 00:01:11,000
So as usual, when I probably start the session, you know, let us consider the entire universe.

20
00:01:11,000 --> 00:01:17,000
And, uh, in this field of universe, I would definitely like to call this as AI.

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

22
00:01:18,000 --> 00:01:19,000
now what is AI?

23
00:01:20,000 --> 00:01:22,000
Simple artificial intelligence.

24
00:01:22,000 --> 00:01:43,000
The main aim is to build applications that can perform, that can perform its own task without human

25
00:01:43,000 --> 00:01:44,000
intervention.

26
00:01:44,000 --> 00:01:49,000
Now, this is the most important definition, right?

27
00:01:49,000 --> 00:01:52,000
Without human intervention.

28
00:01:52,000 --> 00:01:57,000
So it will be able to perform its entire task without human intervention.

29
00:01:57,000 --> 00:01:59,000
So let me just talk about it.

30
00:01:59,000 --> 00:02:02,000
Some of the examples as we know, are Netflix, right.

31
00:02:02,000 --> 00:02:05,000
It has a amazing AI recommendation system.

32
00:02:05,000 --> 00:02:09,000
Recommendation system which recommends movies, right?

33
00:02:09,000 --> 00:02:15,000
Now here, human intervention is not required to probably provide you some kind of recommendation.

34
00:02:15,000 --> 00:02:19,000
Whatever things you are using, you're browsing the movies, whatever movies you are specifically seeing,

35
00:02:20,000 --> 00:02:25,000
you know, at that point of time, this AI model is actually helping you to provide you good recommendations

36
00:02:25,000 --> 00:02:28,000
so that you can stay for a longer period of time.

37
00:02:28,000 --> 00:02:30,000
Self driving car is another example.

38
00:02:30,000 --> 00:02:34,000
Okay, so self driving car is another example.

39
00:02:34,000 --> 00:02:35,000
Right?

40
00:02:35,000 --> 00:02:36,000
Self driving car.

41
00:02:36,000 --> 00:02:38,000
Now it is able to probably drive itself.

42
00:02:38,000 --> 00:02:41,000
You know whenever a turning is coming everything it is able to do.

43
00:02:41,000 --> 00:02:45,000
See at the end of the day whichever field you specifically work right.

44
00:02:45,000 --> 00:02:49,000
Uh, if I talk about AI engineers, this is very much important.

45
00:02:49,000 --> 00:02:56,000
If I talk about AI engineers, at the end of the day, at the end of the day, you are actually creating

46
00:02:56,000 --> 00:02:58,000
an AI product, right?

47
00:02:58,000 --> 00:03:03,000
And this product may be integrated with the software product itself, right?

48
00:03:03,000 --> 00:03:07,000
If I say Netflix, it is a streaming platform, movie streaming platform.

49
00:03:07,000 --> 00:03:12,000
And if I talk about AI engineers, you know, they are trying to integrate some amazing models, you

50
00:03:12,000 --> 00:03:17,000
know, which involves fine tuning, which involves multiple things so that you integrate in such a way

51
00:03:17,000 --> 00:03:19,000
that it is quite scalable.

52
00:03:19,000 --> 00:03:21,000
The machine learning engineer task will also come right?

53
00:03:21,000 --> 00:03:27,000
So at the end of the day, we are specifically creating an AI product, and that product is getting

54
00:03:27,000 --> 00:03:35,000
seamlessly integrated with some, um, you can probably say web app or Android app or mobile app or

55
00:03:35,000 --> 00:03:36,000
edge devices, something as such.

56
00:03:36,000 --> 00:03:39,000
So I hope you got an idea with respect to AI.

57
00:03:40,000 --> 00:03:45,000
So it is mostly about building applications that can perform its own tasks without human intervention.

58
00:03:45,000 --> 00:03:47,000
Now let's talk about machine learning.

59
00:03:47,000 --> 00:03:52,000
Now, whenever I talk about machine learning, machine learning is a subset of AI.

60
00:03:52,000 --> 00:03:54,000
So this is basically machine learning.

61
00:03:54,000 --> 00:03:56,000
It is a subset of AI.

62
00:03:56,000 --> 00:04:06,000
And the main aim of machine learning is to provide you stats, tools to perform to perform various tasks

63
00:04:06,000 --> 00:04:08,000
such as statistical analysis.

64
00:04:08,000 --> 00:04:10,000
Statistical analysis.

65
00:04:11,000 --> 00:04:12,000
Visualization.

66
00:04:13,000 --> 00:04:15,000
Visualization.

67
00:04:15,000 --> 00:04:19,000
Prediction and forecasting.

68
00:04:19,000 --> 00:04:19,000
Right.

69
00:04:20,000 --> 00:04:23,000
So these are some of the examples that I have just written it over here.

70
00:04:23,000 --> 00:04:26,000
But at the end of the day what is machine learning.

71
00:04:26,000 --> 00:04:31,000
It provides a lot of stats tools to properly perform the complete life cycle of a data science project

72
00:04:31,000 --> 00:04:38,000
that is specifically required, uh, in every life cycle, from data ingestion to probably data transformation

73
00:04:38,000 --> 00:04:43,000
feature engineering, you will be required some or the other concepts with respect to ML techniques,

74
00:04:43,000 --> 00:04:47,000
where this is a very important term.

75
00:04:47,000 --> 00:04:50,000
Specifically, if I say about that is nothing but stats tools.

76
00:04:50,000 --> 00:04:53,000
Okay, so machine learning is a subset of AI, right?

77
00:04:53,000 --> 00:04:57,000
And here it is providing the stats tools to perform all this task.

78
00:04:57,000 --> 00:05:00,000
And this task is usually performed on what.

79
00:05:00,000 --> 00:05:01,000
On data.

80
00:05:01,000 --> 00:05:02,000
Now why we do this.

81
00:05:02,000 --> 00:05:04,000
Why we do this.

82
00:05:04,000 --> 00:05:08,000
So that so that we understand about the data.

83
00:05:10,000 --> 00:05:13,000
We understand about the data.

84
00:05:13,000 --> 00:05:13,000
Right.

85
00:05:13,000 --> 00:05:18,000
So that a data will be meaningful, it will be able to convey some information to you.

86
00:05:18,000 --> 00:05:18,000
Right.

87
00:05:18,000 --> 00:05:21,000
So that is where machine learning basically comes into existence.

88
00:05:21,000 --> 00:05:26,000
Now coming to the next part, that is nothing but deep learning.

89
00:05:26,000 --> 00:05:30,000
So if I talk about deep learning again, deep learning is a subset of machine learning.

90
00:05:31,000 --> 00:05:36,000
And if I talk about deep learning from 1950s, it's not like deep learning has become just famous right

91
00:05:36,000 --> 00:05:42,000
now is it has become famous right now because of amazing things like GPUs, technological advancement,

92
00:05:42,000 --> 00:05:44,000
open source libraries, and many more things.

93
00:05:44,000 --> 00:05:52,000
But this, this entire deep learning was built to mimic human brain, right?

94
00:05:52,000 --> 00:05:53,000
Human brain.

95
00:05:53,000 --> 00:06:01,000
We wanted we wanted AI or we wanted machines to perform like how we human beings used to perform, right?

96
00:06:01,000 --> 00:06:03,000
How we human being used to learn.

97
00:06:03,000 --> 00:06:07,000
So that is the reason, as I said, mimicking human brain.

98
00:06:07,000 --> 00:06:12,000
And that is where we have something called as multilayered neural networks.

99
00:06:12,000 --> 00:06:13,000
Right.

100
00:06:13,000 --> 00:06:15,000
Multi-layered neural networks.

101
00:06:15,000 --> 00:06:17,000
So I hope you got a clear idea about this.

102
00:06:17,000 --> 00:06:23,000
And here whenever I talk about deep learning, the three important things that we specifically learn,

103
00:06:23,000 --> 00:06:24,000
right.

104
00:06:24,000 --> 00:06:29,000
Please understand this because after this only will be able to understand about generative AI.

105
00:06:29,000 --> 00:06:35,000
So with respect to deep learning, the three important things that we specifically learn, write and

106
00:06:36,000 --> 00:06:37,000
write CNN.

107
00:06:37,000 --> 00:06:41,000
And then we basically say RNN and its variants.

108
00:06:41,000 --> 00:06:42,000
Right.

109
00:06:42,000 --> 00:06:49,000
And this is where when I say RNN and its variants, CNN and object detection, CNN object detection

110
00:06:49,000 --> 00:06:52,000
is completely for computer vision purpose.

111
00:06:52,000 --> 00:06:52,000
Right.

112
00:06:52,000 --> 00:06:59,000
So if I talk about task over here specifically you use computer vision over here.

113
00:06:59,000 --> 00:07:05,000
You specifically use for what kind of use cases for text related use cases, write text related use

114
00:07:05,000 --> 00:07:06,000
cases.

115
00:07:07,000 --> 00:07:08,000
Right?

116
00:07:08,000 --> 00:07:16,000
Or you can also use it for time series use cases, because that is how RNN and its variants are designed,

117
00:07:16,000 --> 00:07:16,000
right?

118
00:07:16,000 --> 00:07:18,000
Time series use cases.

119
00:07:18,000 --> 00:07:23,000
So an is definitely like how machine learning is basically trained.

120
00:07:23,000 --> 00:07:27,000
Similarly, you can train machine learning problem statement with the help of a and also.

121
00:07:27,000 --> 00:07:28,000
Right.

122
00:07:28,000 --> 00:07:33,000
So overall most of the things that we specifically learn in deep learning are based on these three things.

123
00:07:33,000 --> 00:07:34,000
Okay.

124
00:07:34,000 --> 00:07:38,000
When I say CNN object detection there are techniques like R-cnn and many more things.

125
00:07:38,000 --> 00:07:44,000
And all right similarly RNN you have RNN, LSTM, RNN and GRU, then you have encoder decoder.

126
00:07:44,000 --> 00:07:45,000
Attention is all you need.

127
00:07:46,000 --> 00:07:48,000
Then you have Transformers and Berts right?

128
00:07:48,000 --> 00:07:52,000
So I hope everybody has learned from my playlist till here.

129
00:07:53,000 --> 00:07:57,000
Everything has been explained along with theoretical intuition and practical intuition.

130
00:07:57,000 --> 00:08:03,000
Transformer and Bert are something very advanced and from here only we will be deriving the next thing

131
00:08:03,000 --> 00:08:05,000
that is nothing but generative AI.

132
00:08:06,000 --> 00:08:13,000
Okay, because this is the backbone used in most of the LLM models in generative AI transformers and

133
00:08:13,000 --> 00:08:13,000
birds.

134
00:08:13,000 --> 00:08:17,000
Okay, so deep learning is mostly about this specific thing.

135
00:08:17,000 --> 00:08:20,000
At the end of the day, we are trying to mimic the human brain.

136
00:08:20,000 --> 00:08:22,000
We are trying to understand how human beings specifically learn.

137
00:08:22,000 --> 00:08:24,000
We also learn in the same way.

138
00:08:24,000 --> 00:08:27,000
Now coming to the next one, that is nothing but generative AI.

139
00:08:27,000 --> 00:08:31,000
Obviously, generative AI is a subset of deep learning again.

140
00:08:32,000 --> 00:08:35,000
And here advanced things okay.

141
00:08:35,000 --> 00:08:39,000
So I will specifically talk about two types of model that we use in data science.

142
00:08:39,000 --> 00:08:43,000
Industry models are two types of model training.

143
00:08:43,000 --> 00:08:44,000
We specifically say right.

144
00:08:44,000 --> 00:08:51,000
One is the discriminative, uh, if I if I just want to say it is that it's mostly like discriminative

145
00:08:51,000 --> 00:08:52,000
and generative models.

146
00:08:52,000 --> 00:08:52,000
Okay.

147
00:08:52,000 --> 00:08:59,000
So if I talk about deep learning models, they are mostly of two types okay.

148
00:08:59,000 --> 00:09:02,000
This this is one and this is the other one okay.

149
00:09:03,000 --> 00:09:07,000
Here you have something called as discriminative models.

150
00:09:07,000 --> 00:09:10,000
Here you have something called as generative models.

151
00:09:10,000 --> 00:09:16,000
Now you should understand whenever I talk about generative AI you should understand one very important

152
00:09:16,000 --> 00:09:16,000
thing.

153
00:09:16,000 --> 00:09:20,000
Generative AI is more about generating content okay.

154
00:09:21,000 --> 00:09:26,000
So these are specific models which will help you to generate new content okay.

155
00:09:26,000 --> 00:09:29,000
Based on whatever content it has already been trained on.

156
00:09:29,000 --> 00:09:30,000
Okay.

157
00:09:30,000 --> 00:09:39,000
If I talk about discriminative AI, it is mostly about task like classification, classification, prediction,

158
00:09:39,000 --> 00:09:41,000
all this regression prediction.

159
00:09:41,000 --> 00:09:47,000
So all the stocks you can basically do in short over here your data set that you have.

160
00:09:47,000 --> 00:09:48,000
Right.

161
00:09:48,000 --> 00:09:54,000
These are this entire discriminative models are trained on labeled data set.

162
00:09:54,000 --> 00:09:57,000
You have to really understand this okay.

163
00:09:58,000 --> 00:09:59,000
Labeled data set.

164
00:09:59,000 --> 00:10:05,000
Similarly in the case of generative AI, if I really want to understand about the task, it is nothing,

165
00:10:05,000 --> 00:10:12,000
but it generates new data trained on.

166
00:10:15,000 --> 00:10:16,000
Huge data set.

167
00:10:17,000 --> 00:10:17,000
Okay.

168
00:10:18,000 --> 00:10:20,000
If I really want to make you understand.

169
00:10:20,000 --> 00:10:20,000
Right.

170
00:10:20,000 --> 00:10:23,000
In a simple way, if I just take.

171
00:10:23,000 --> 00:10:25,000
Okay, I, I am a person.

172
00:10:25,000 --> 00:10:32,000
What I've done is that I have probably learnt or read 100 books on cats.

173
00:10:33,000 --> 00:10:33,000
Right?

174
00:10:33,000 --> 00:10:37,000
So I have been trained on this many number of books, huge amount of data.

175
00:10:37,000 --> 00:10:44,000
Right now what I being a person, if anybody asks me any questions with respect to cats, I will be

176
00:10:44,000 --> 00:10:47,000
able to answer all the questions in my own way.

177
00:10:47,000 --> 00:10:48,000
Right?

178
00:10:48,000 --> 00:10:52,000
And obviously the answers will be accurate because I have already read 100 books on the cat.

179
00:10:52,000 --> 00:10:57,000
So similarly, if I talk with respect to generative AI, there are two types of models that we specifically

180
00:10:57,000 --> 00:10:57,000
learn.

181
00:10:57,000 --> 00:10:59,000
One is large language models.

182
00:10:59,000 --> 00:11:00,000
Large image models.

183
00:11:00,000 --> 00:11:03,000
Large language model is with respect to text data.

184
00:11:03,000 --> 00:11:07,000
Large image model is specifically with respect to images and videos.

185
00:11:07,000 --> 00:11:09,000
So here if a text is given, you can convert that into an image.

186
00:11:09,000 --> 00:11:14,000
If a text if a text is given you can convert that into videos here.

187
00:11:15,000 --> 00:11:16,000
Large language model.

188
00:11:16,000 --> 00:11:19,000
If a text is given it will be able to provide some response in terms of text.

189
00:11:19,000 --> 00:11:20,000
Right.

190
00:11:20,000 --> 00:11:25,000
And if I talk about all the various categories of models, just in some time I'll be explaining you

191
00:11:25,000 --> 00:11:25,000
that also.

192
00:11:25,000 --> 00:11:27,000
Okay, so generative AI.

193
00:11:27,000 --> 00:11:34,000
In short, what it is basically doing is that here we have models that is already trained in huge amount

194
00:11:34,000 --> 00:11:35,000
of data.

195
00:11:35,000 --> 00:11:38,000
And the task of that specific models are based on any input.

196
00:11:38,000 --> 00:11:40,000
It will generate a new data itself.

197
00:11:40,000 --> 00:11:41,000
Okay.

198
00:11:41,000 --> 00:11:44,000
Now to make you understand more about LM models.

199
00:11:44,000 --> 00:11:44,000
Right.

200
00:11:44,000 --> 00:11:47,000
So if I talk about more LM models, right.

201
00:11:47,000 --> 00:11:58,000
Obviously, you know, companies like OpenAI, you know, companies like meta, Google, anthropic.

202
00:11:58,000 --> 00:11:59,000
Right.

203
00:11:59,000 --> 00:12:04,000
And everybody is race in, in the race, right to generate the best LM model.

204
00:12:04,000 --> 00:12:07,000
So this, this, this companies are already doing really, really well.

205
00:12:07,000 --> 00:12:11,000
Anthropic basically comes up with uh model which is called as cloud three.

206
00:12:11,000 --> 00:12:11,000
Right.

207
00:12:11,000 --> 00:12:15,000
Right now it is being a fierce competitor of OpenAI, GPT four.

208
00:12:15,000 --> 00:12:16,000
Right.

209
00:12:16,000 --> 00:12:17,000
GPT four.

210
00:12:17,000 --> 00:12:20,000
Then you have in meta, you have open source models like llama two.

211
00:12:20,000 --> 00:12:21,000
Google.

212
00:12:21,000 --> 00:12:25,000
I hope everybody has heard about Gemini, right in Google.

213
00:12:25,000 --> 00:12:29,000
And open source model has also come, which is called as gamma right now.

214
00:12:29,000 --> 00:12:31,000
What are all these specific models?

215
00:12:31,000 --> 00:12:38,000
These models are nothing, but these are specifically called as foundation models okay.

216
00:12:38,000 --> 00:12:41,000
This foundation models are also called as pre-trained models.

217
00:12:41,000 --> 00:12:42,000
Why?

218
00:12:42,000 --> 00:12:44,000
We say it as pre-trained models?

219
00:12:44,000 --> 00:12:47,000
Because these are trained in huge amount of data.

220
00:12:47,000 --> 00:12:49,000
The entire internet data.

221
00:12:49,000 --> 00:12:50,000
Huge data.

222
00:12:52,000 --> 00:12:55,000
It has been trained in huge data.

223
00:12:55,000 --> 00:12:56,000
The entire internet data.

224
00:12:56,000 --> 00:12:57,000
Right.

225
00:12:57,000 --> 00:12:59,000
It may be cold, it may be multiple things.

226
00:12:59,000 --> 00:13:05,000
And all right, now we can use this specific model for domain specific use cases.

227
00:13:05,000 --> 00:13:08,000
Also domain specific use cases also.

228
00:13:08,000 --> 00:13:13,000
And that is where a concept something called as fine tuning is used.

229
00:13:13,000 --> 00:13:13,000
Right.

230
00:13:13,000 --> 00:13:17,000
And I have already created a playlist about fine tuning.

231
00:13:17,000 --> 00:13:23,000
Also, I've explained concepts like Laura, Laura, how you can basically do the fine tuning, how you

232
00:13:23,000 --> 00:13:26,000
can do it with, uh, open source models and many more things.

233
00:13:26,000 --> 00:13:26,000
Right.

234
00:13:26,000 --> 00:13:32,000
The fierce competition right now is basically to get the best foundation model, right?

235
00:13:32,000 --> 00:13:38,000
And right now, GPT four, GPT four turbo, GPT five is also going to come right.

236
00:13:38,000 --> 00:13:42,000
If I talk about open source model, llama three is there in the pipeline, right?

237
00:13:42,000 --> 00:13:45,000
Cloudy has just recently launched cloud three.

238
00:13:45,000 --> 00:13:48,000
Gemini is also coming up with Gemini Pro.

239
00:13:48,000 --> 00:13:48,000
More variants.

240
00:13:48,000 --> 00:13:49,000
It is basically coming up.

241
00:13:49,000 --> 00:13:54,000
So the main race, if I probably talk about is to create the best foundation model.

242
00:13:54,000 --> 00:13:55,000
Later on.

243
00:13:55,000 --> 00:14:01,000
Many companies will be able to use this foundation model in the form of pre-trained models, or it will

244
00:14:01,000 --> 00:14:01,000
also.

245
00:14:01,000 --> 00:14:07,000
They will also be able to fine tune with some own custom data set right to solve their use cases.

246
00:14:08,000 --> 00:14:11,000
So this was the entire idea about generative AI.

247
00:14:12,000 --> 00:14:16,000
Again, uh, this is the video that I really needed to upload, uh, early.

248
00:14:16,000 --> 00:14:19,000
But yes, many people were waiting requesting for this specific video.

249
00:14:19,000 --> 00:14:23,000
So I made you understand this entire thing by writing in front of you.

250
00:14:23,000 --> 00:14:28,000
So for more details with respect to generative AI Lang Chen, now, if I talk about Lang Chen, it is

251
00:14:28,000 --> 00:14:29,000
a framework, right?

252
00:14:29,000 --> 00:14:33,000
Lang Chen is a framework which will be able to work with all the specific models.

253
00:14:33,000 --> 00:14:40,000
At the end of the day, we will be able to generate our application, our application like retrieval,

254
00:14:40,000 --> 00:14:46,000
augmentation query, and along with that, we will also be able to develop some amazing chat bots.

255
00:14:46,000 --> 00:14:48,000
That is the main reason, right?

256
00:14:48,000 --> 00:14:54,000
We why LM becoming very famous because of the chances of getting automated right, we will be able to

257
00:14:54,000 --> 00:14:58,000
automate the entire chat bot response things with respect to it.

258
00:14:58,000 --> 00:15:01,000
Not only that, it has a huge scope in multiple sectors.

259
00:15:01,000 --> 00:15:06,000
Also, with respect to various use cases, uh, this is mostly about LM models.

260
00:15:06,000 --> 00:15:09,000
Lim models is with respect to large image models, stability.

261
00:15:10,000 --> 00:15:13,000
Uh, stability I is a company which is specifically working in this specific thing.

262
00:15:13,000 --> 00:15:14,000
Right.

263
00:15:14,000 --> 00:15:19,000
So I hope you like this particular video for more videos related to long chain generative AI lm models,

264
00:15:19,000 --> 00:15:24,000
you can follow my other playlist of long chain open AI and ah, where you'll be able to see lot of end

265
00:15:24,000 --> 00:15:25,000
to end projects.

266
00:15:25,000 --> 00:15:27,000
So I hope you like this particular video.

267
00:15:27,000 --> 00:15:28,000
I will see you all in the next video.

268
00:15:28,000 --> 00:15:28,000
Have a great day!

269
00:15:28,000 --> 00:15:29,000
Thank you all.

270
00:15:29,000 --> 00:15:29,000
Take care.

271
00:15:29,000 --> 00:15:30,000
Bye bye.