1
00:00:00,360 --> 00:00:00,920
All right.

2
00:00:00,960 --> 00:00:02,440
Now it's the moment of truth.

3
00:00:02,600 --> 00:00:10,080
We now read the LinkedIn profile in as before, and hopefully this is your LinkedIn profile, not mine.

4
00:00:10,320 --> 00:00:13,240
And then we bring in the summary text.

5
00:00:13,280 --> 00:00:15,480
You should change this name to be your name.

6
00:00:15,560 --> 00:00:20,880
We run that and we now have our system prompt and user prompt for the LM.

7
00:00:21,240 --> 00:00:27,160
And the system prompt is going to be the same as last time you're acting as me or you.

8
00:00:27,200 --> 00:00:32,680
You're answering questions on their website, particularly related to their career, etc., etc. but

9
00:00:32,680 --> 00:00:34,120
there's a twist at the end here.

10
00:00:34,120 --> 00:00:39,840
It says if you don't know the answer to any question, use your tool to record the question you couldn't

11
00:00:39,840 --> 00:00:40,400
answer.

12
00:00:40,840 --> 00:00:46,560
And if the user is engaging in discussion, try to steer them towards getting in touch via email.

13
00:00:46,600 --> 00:00:48,760
Record it using your tool.

14
00:00:49,160 --> 00:00:55,160
Now, in theory, this isn't needed because the JSON that we've written to describe the tool already

15
00:00:55,160 --> 00:00:58,840
gives this kind of context and describes the tool and when it should be used.

16
00:00:59,240 --> 00:01:06,440
But it never hurts to be repetitive in prompting, as I'm sure you know, repetition always works well.

17
00:01:06,480 --> 00:01:12,800
Explain things several times over, because you will increase the probability that your model performs

18
00:01:12,800 --> 00:01:13,960
the way you want it to.

19
00:01:14,080 --> 00:01:19,840
You're biasing the model to be outputting tokens consistent with your objective, and it is important

20
00:01:19,840 --> 00:01:21,400
to always keep that in the back of your mind.

21
00:01:21,440 --> 00:01:22,750
We often like to talk about.

22
00:01:22,790 --> 00:01:27,190
You know, the model calls this tool and we make it sound like it almost that it's almost got something

23
00:01:27,190 --> 00:01:28,550
that has true autonomy.

24
00:01:28,550 --> 00:01:30,710
But all that's actually going on here.

25
00:01:30,750 --> 00:01:37,110
And always remind yourself of this from time to time, an LM is just something that's generating the

26
00:01:37,110 --> 00:01:41,550
most likely next tokens to follow a sequence of tokens.

27
00:01:41,550 --> 00:01:45,430
It's got an input and it's generating like predictive text.

28
00:01:45,470 --> 00:01:48,230
What is the most likely text to follow this?

29
00:01:48,670 --> 00:01:54,710
And if we've inserted into this prompt stuff about different tools that it could call and stuff that

30
00:01:54,710 --> 00:02:01,230
directs it to use them, then that just means that the most likely next tokens will be consistent with

31
00:02:01,230 --> 00:02:03,150
that information that we passed in.

32
00:02:03,350 --> 00:02:04,910
And that's all that's going on.

33
00:02:04,910 --> 00:02:07,470
It's about statistical most likely tokens.

34
00:02:07,470 --> 00:02:12,350
And that really I mean, it hurts my mind to try and think that one through, because it seems so unlikely

35
00:02:12,350 --> 00:02:17,750
that predicting next tokens could be so powerful as to do things like calling tools.

36
00:02:17,750 --> 00:02:19,950
But that really is what's happening here.

37
00:02:20,150 --> 00:02:22,550
All right, that was a sidebar, but it's an important sidebar.

38
00:02:22,590 --> 00:02:24,150
You should always try and keep that in mind.

39
00:02:24,190 --> 00:02:26,990
And I have a bunch of YouTube videos about that stuff.

40
00:02:26,990 --> 00:02:32,070
If you want to learn more about how Next token prediction is able to achieve these things.

41
00:02:32,590 --> 00:02:38,470
Okay, so with all of this we get to our chat function again.

42
00:02:38,590 --> 00:02:42,110
And this is where it really comes down to action.

43
00:02:42,110 --> 00:02:44,070
And we'll cover it in the next video.