1
00:00:05,850 --> 00:00:11,610
In this lesson we are going to talk about the foundation LM models.

2
00:00:17,360 --> 00:00:28,940
So Foundation LM models, Foundation LMS and what we call LMS in this program are the same thing are

3
00:00:29,180 --> 00:00:33,710
LMS train with all the internet data.

4
00:00:34,580 --> 00:00:38,270
So this is not 100% accurate.

5
00:00:38,270 --> 00:00:45,890
They are not trained with all the internet data, but almost a we can say they are trained with all

6
00:00:45,890 --> 00:00:47,780
the public internet data.

7
00:00:47,780 --> 00:00:58,580
Some people like Elon Musk say that this is not true, that really llms are trained with private data

8
00:00:58,580 --> 00:00:59,300
as well.

9
00:00:59,300 --> 00:01:08,570
But let's say that we go with the official story, and we assume that Llms are trained with all the

10
00:01:08,570 --> 00:01:12,200
public internet data is a lot of data.

11
00:01:12,350 --> 00:01:19,670
You may have a two different kinds of foundation LM models.

12
00:01:19,670 --> 00:01:28,730
You have the private ones like ChatGPT or anthropic, and you have the open source one like lamassu,

13
00:01:28,730 --> 00:01:31,940
Falcon and Mistral Way.

14
00:01:31,940 --> 00:01:35,030
Will they remain open source?

15
00:01:35,090 --> 00:01:36,170
We don't know.

16
00:01:36,170 --> 00:01:44,660
In the case of llama two, we know that Facebook is behind, but right now they are free to use both

17
00:01:44,660 --> 00:01:46,460
private and open source.

18
00:01:46,550 --> 00:01:54,590
Uh, LM models have advantages and disadvantages and is very important to understand them well in order

19
00:01:54,590 --> 00:02:00,020
to make the right choice when you are building your LM application.

20
00:02:01,430 --> 00:02:05,000
We are going to use one analogy.

21
00:02:05,000 --> 00:02:14,570
During this program, we are going to understand that llms like ChatGPT are like the engine we use for

22
00:02:14,570 --> 00:02:16,610
our LLM applications.

23
00:02:16,610 --> 00:02:24,620
And our LLM applications are going to be like the vehicles and the machines that use that engine, which

24
00:02:24,620 --> 00:02:25,790
is the LLM.

25
00:02:26,300 --> 00:02:33,650
So remember the foundation LLM model is what we have been calling just LLM.

26
00:02:33,650 --> 00:02:37,160
So we will use just LLM during the program.

27
00:02:37,160 --> 00:02:46,400
There is a it is so new what is happening that there are still a little bit there is still a little

28
00:02:46,430 --> 00:02:50,480
bit confusion uh around some names.

29
00:02:50,870 --> 00:02:58,220
So we are going to use LLM when we want to uh refer to foundation LLM models.

30
00:02:58,220 --> 00:02:58,820
Okay.

31
00:03:01,780 --> 00:03:10,840
In the next lesson, we are going to start talking about some very important basic concepts of LMS,

32
00:03:10,840 --> 00:03:19,660
and you will see that LM limitations are in the origins of LM applications.

33
00:03:19,750 --> 00:03:22,510
We will see that in the next lesson.