1
00:00:05,370 --> 00:00:06,570
In this lesson.

2
00:00:07,250 --> 00:00:15,140
We will talk very quickly about some other interesting notes regarding the advanced architecture of

3
00:00:15,140 --> 00:00:16,820
LLM applications.

4
00:00:17,480 --> 00:00:22,730
Remember that you have a lot of documents in the attached materials.

5
00:00:22,730 --> 00:00:29,990
And in this lesson, I just wanted to tell you a little bit about how you are going to find in this

6
00:00:29,990 --> 00:00:32,210
attached materials for these blocks.

7
00:00:32,210 --> 00:00:42,950
So you will see some information about LLM ops frameworks, about evaluation frameworks, about a optimization

8
00:00:42,950 --> 00:00:53,750
techniques for uh, the RAC system, the microservices uh alternative, the microservices architecture

9
00:00:53,750 --> 00:01:02,090
for LLM applications, and also a very interesting document, just highlighting the most frequent challenges

10
00:01:02,090 --> 00:01:05,630
when deploying LLM applications to production.

11
00:01:05,720 --> 00:01:15,980
I am not going, uh, to go into detail, uh, I think that at this point it is not a the best moment

12
00:01:15,980 --> 00:01:21,290
to go in detail, uh, with this, uh, topics.

13
00:01:21,500 --> 00:01:26,600
The only thing I will tell you about is about the last, uh, point.

14
00:01:26,600 --> 00:01:36,170
The top challenges when deploying LLM applications to production right now are cost evaluation, reliability

15
00:01:36,170 --> 00:01:42,170
of the infrastructure, and quality of the foundational model.

16
00:01:42,170 --> 00:01:56,360
So with that, you can understand that advanced architecture of LLM application doesn't end with back

17
00:01:56,360 --> 00:02:00,050
end front end LLM.

18
00:02:00,890 --> 00:02:03,530
APIs and cloud storage.

19
00:02:03,530 --> 00:02:13,190
We also have many other issues that we need to solve in the process of going into production, and you

20
00:02:13,190 --> 00:02:18,320
will see that the top challenges when deploying applications to production are.

21
00:02:18,710 --> 00:02:19,550
Cost.

22
00:02:20,670 --> 00:02:30,780
Evaluation, which is testing the reliability of the infrastructure and the quality of the foundation

23
00:02:30,780 --> 00:02:34,800
LM model that we have, uh, selected.

24
00:02:34,800 --> 00:02:49,890
So just wanted to give you a quick a note of the complexity that it is involved in developing and deploying

25
00:02:49,890 --> 00:02:51,120
LM applications.

26
00:02:51,120 --> 00:02:56,790
This is the the kind of things we are going to see in the next lessons.

27
00:02:56,790 --> 00:03:01,920
So we are going to be, uh, far from the toy demos.

28
00:03:01,920 --> 00:03:06,840
We are going to be focused on professional applications.

29
00:03:08,860 --> 00:03:17,830
In the next lesson, we are going to talk about a very, very important topic in this program, which

30
00:03:17,830 --> 00:03:19,240
is the Rag technique.

31
00:03:19,240 --> 00:03:25,690
The Rag technique is the main technique of the LM applications right now.

32
00:03:25,690 --> 00:03:31,450
So it is very important for us to understand what it is and how to master it.