Transcripts
1. Introduction: L Welcome to ChargiBTPms that
work Anywhere. Let me give you this introduction
with a small example. So let's get into
Cha GIPT and I'm giving the instruction as don't answer anything outside the data that I've provided to you. If I ask you anything outside of the
data that I've given, answer with I don't know. And I've given the data as my name is Joshua,
and I'm from Chennai. Okay. So now let's ask
Chai I'm in hagiBti now. Cha Gibt knows everything. Let me ask who is the Who is the president
of the United States? It says, I don't know. And even if I ask anything, it will say, I don't know, except this data which I've
given inside the codes. So if I ask, where is Joshua
from, it should answer it. It's Joshua's from Chennai. So other than this, if you
ask anything, let's say, top ten bikes in
the United States. I don't know. It doesn't know anything other than the
data that I've provided. Similarly, I have given
you a set of proms, which can be used anywhere. You can use it for
your personal life, you can use it for
your official life, and it'll be useful anywhere. And this is not restricted
only for Charge pity. You can use it in any alum. You can use it for grog. You can use it for Gemini. You can use it anywhere,
and it'll work. So and the better way or the efficient way to make use of this course is try to
combine these proms. Like, you'll be going
through a lot of proms. You just have to
combine two or three together to make
it more efficient. I'll give you an
example how to combine and make it more efficient
in the project section.
2. Project: Your project is to combine
the learned prompts, whichever is relevant to
your field and submit it. So first, let me show
you an example of how I'm using these prompts
efficiently in the right way. So let's assume
our task now is to create a promotion for a
brand name Sugar Rush. And the Sugar Rush is
nothing but a juice shop. Okay, so now this
will be my prompt. You are a world class
prompt engineer, specialized in graphic design. This is nothing but role play. Next, your task is to help me create a poster for juice shop, and the shop name is Sugar Rush. They sell only watermelon
and orange juice. Both are priced Rupees 100. I'm just giving
some context about the shop the promotion that
we are going to create. And step one, ask me five questions for your
better understanding. Before doing this task, ask
your questions one by one. So always, we have
something in our mind, but the machine doesn't
understand that exactly, right, because it's not
that every time, whatever is in our mind will be understood by the machine. So in order to articulate
what is there in my mind, I'm giving this prompt.
Ask me five questions. So the machine is going
to ask me five questions, and I'm going to answer that, and the machine is going to
get a better understanding. So this is nothing but
ask for context, right? And I'm also
specifying the steps. Step one, step two, step three. So it'll be going in an
order. Step one, step two. This is nothing but
specifying the steps. So in this single prompt, I'm using role play in
the initial prompt. Then I'm using
specifying the steps, step one step two step three, and in the step one,
I'm asking for context. So I'm using three to
complete this task. And in the second
prompt I'm asking, give me a prompt to create an AI video for 6
seconds for the same. And in the third
step I'm asking, give me a prompt to create a commercial poster for the
same using image generation. So we're going to
create an image, and also we are going to create a small AI video for the
brand name Sugar Rush. So let's try this out. So in the step one, we asked to ask us
five questions, so it's asking the
first question. Where will this poster
mainly be used? Instagram post or
Instagram Story, Whatsap status or let's
go with Instagram Story. Who is the exact
target audience? Let's say, college students, What vibe should
Sugar Rush give? Premium, cool and trendy. Okay, let's go with
the Cool and trendy. Do you want to highlight
any special tag line? Okay. Let me go with
this beat the heat. And the final question is, do you want the design to feel more watermelon
dominant or dominant? Split themed. Okay, let me
go with this split themed. Okay Okay. Step two, it is giving us a prompt
for video generation. And step three, it is giving us a prompt for image generation. Great. Okay. Now let me
just copy this prompt. And I'm getting into
a video generation. There are many very good
video generation model like a V three, grog. For this example, I'm
going to use grog. Let me get it into this image of grog and I'm just
pasting this prompt. That's all. Make sure that
video is checked there. Sometimes it'll be an image, you have to select
the size also. I'm selecting three
by two, three by two, or maybe I'll go 16 by nine
or proper landscape video, 16 by nine, 6 seconds. Okay, let me generate. The video is now ready. Just take a look at this video, and I think it's looking
really good. Sugar Rush. I think this video
is really good. You can always download this and let me play
this for you once. 100% fresh, no added sugars. So now we have got an AI add. Let's now try image
generation for the same. Now it has given us a prompt
for image poster, as well. Again, I'm just copying this. I'm just copying this.
Let me get into Sora. Sora is very good for
image generation, usually. So I'm going with Sora. Yeah, in Sora again, let me choose a landscape
one, which is three by two. And I'm pasting the prompt which is given by hagi but
I didn't do anything. I didn't think anything. I'm
just pasting it. That's all. Let me go with it. Yeah, Sora has now
created as a poster. It has given us two variations. That is only a small a slight variation
between these two, and I think it's
looking really good. We didn't think anything about the creative. We
didn't think anything. Chardy Betty gave us the prompt, and just with the prompt, we have created an
air video we have created a very good poster. This is looking really
good, and I think this can be posted in
social media as well. Yeah. So using the
prompting framework and all the prompts
that we learned, I have done a small project. So similarly, you can
use all these prompts. You can combine them,
combine them together, and you can make it
relevant to a field and paste that as a project in the project section. I'll
meet you in the next.
3. TLDR: In this video, you'll be learning how to
summarize a video. It can be a podcast. It can be a course or whatever. So you don't even
have to watch it, but you can get all the
details which is inside it. So first, I'm getting
into YouTube, this is a long podcast. This podcast is
around 2 hours long. Okay, so what I'm doing is
I'm just clicking on this more and if you go down, you'll find an option called
Show transcript, okay? So if you click on this,
you'll get the transcript. So transcript is nothing but the speech that they do, right? For example, so if I
click on this, okay? And now, it says, but that doesn't mean I think the underlying
risk is actually. So if I play it, Sundar Pache will exactly say these words, but that doesn't mean. That doesn't mean I think the underlying risk is
actually pretty high. Okay, so this is exactly
what is there in this video. So what I'm doing now is
I'm just copying this. I'm copying the
transcript, okay? Let me just copy the transcript. Okay. F of now, I'm copying this much, okay? I'm copying this
and I'm pasting it. So before pasting, I'm
using a keyword, okay? TL Colon DR. So TL, semicolon DR. This is the keyword or
this is the prompt, which means too
long didn't read. Okay? So this is very long. So just give
me the highlight. This is what you mean. So now I'm pasting the
transcript, okay? So now I'm pasting
the transcript. And here's the prompt TLDR. Let me hit Enter. This is a long podcast,
very long podcast. Yeah, now A has done the magic. So we have copied around 48
minutes of that podcast, and it has given
us the highlights. So Sunda's early life
and inspiration. He grew up in Chennai
these all these were the content
which was there, technology and AI, advice for young people
follow your heart, not just your head, surround yourself with people
smarter than you. So yeah. So we have got the crisp of what was there in
that two hour video. We don't have to
watch it completely. Okay. So this is maybe if you
want to know what is there, you just have to just
take the transcript, paste it here, TLDR. So what's the difference
between TLDR and summarize? Summarize is like summary. Let's say, for
example, if you are having 1,000 words, okay, if you're having or if you
are pasting thousand words, summarize will give
you like 500 words. Okay? It will summarize, it will take all the
other important things and it'll give you
important things. TLDR will give you only
200 words or 150 words. So TLDR will just give
you the highlights, okay? I can also go if I
want more details, my prom should be summarized. If I give summarize, let
me show you that also. And I'm pasting this. So if I do this, I'll be getting more details
when compared to TLDR. So it is giving us more details. Okay, so this is the difference. So summarize will give
us a little more detail, and TLDR will just give
us the highlights. So if it is a podcast
or something, if you want more details,
go for summarize. If you just want to
know what is there, then go with TLDR. We'll
meet in the next video.
4. Specify the Steps: Specifying the steps. This is to do a lot of
thing in a single command. Let's say, for example, if
we are doing a task, first, we'll go for a prompt, then with that data, we'll
go for another prompt. So this is to specify all
the steps in one prompt. I'll show this with an example. I'm getting into ChargVT
I'm saying step one, I'm planning a trip to Carlo. Step one, give me the
top five destinations to visit Karla and step two, create an itinerary to visit those places from 6:00 A.M. To 7:00 P.M. And step three tell
me the modes of commute. Also the other thing
that we have to know is step two is
dependent on step one. So without step one, this step two cannot
be done, right? So first prompt I have specified five destinations
to visit Karla. Step two is dependent on first. Step three is
dependent on step two. This is how my prompt
is. So let's see. Just with one prompt,
I'm doing all the steps. I'm getting all the
data that I want. Step one is almost done. Top five destinations
to visit Kerala. The highlights, best
par and everything, yeah we have got once we
get the places to visit, next we should know what is the time that we
have to visit, right? Step two is giving
the time also. From 6:00 A.M. To 730, I have to go to this place.
These are all the activities. Got it, it's giving
a lunch break and. Once I know the
time, then I have to know how I have
to come out, right? So now step in step three, it is giving me the distance, durations and if I can take
a bus or cab or whatever. Everything is done. This is just to get everything
done in one prompt. You can use it for anything. For example, if you're having a big task and if your task
is dependent on other, you can give a single prompt. You can creatively think for the work that you are doing and you can specify all the steps and do a lot of things
in single prompt.
5. Ask for Context: Ask for context. So
using this prompt, the machine will ask
you some questions and it will give you a
personalized response. So let's get into CHAPT and
let's ask something like, give me a suggestion
to buy a new laptop. Okay, suggest me to
buy a new laptop. Yeah, so now we have got a
list of some random laptops, MacBook Air, four,
and something right. But this is not what I want. This is like a set of
random laptops, right? So now let me add this prompt. So again, now my prompt is suggest me to
buy a new laptop. Before doing this task, if you don't have
all the information. So before doing this task, if you don't have all the
information to do the task, quit doing the task and ask
me for more information. Ask your questions one by one. So this is a prompt. Ask your questions one by one. I'll ask the
questions one by one. And if you don't have all
the information, ask me. Okay, now let's see how the LM is responding
for this prompt. First, it's asking
me a question. Do you prefer Windows or MacOis? So it didn't give
me a random laptop. It's asking me. So
let's go with MacOys. Let me go with MacOis. Yeah, so now it is asking
me a few questions. So what is your primary
use for the laptop? Is it for everyday use for professional task or let me
go with everyday use, okay? Yeah, now it's giving
me a recommendation of either I can go with
13 inch or four or what. So initially, it was suggesting
me some random laptops, but now I'm getting a
personalized response. So this is how we can
make the machine to ask us some questions and give
us a personalized response.
6. 5 Year Old : Explain X as if I'm f or world. So this is to make a tough tough or a difficult concept,
easy to understand. So let's get into hagiPT. So now my prompt is explain how a chief minister
is elected in India. Okay, let's just go
with this prompt. Yeah, so now we have got a usual response or a
normal response, right? So now what I'm doing is
I'm using the same prompt, and I'm using this keyword
called as if I am 5-year-old. So you can go with
10-year-old or whatever. So now let's see how the
LLM responds to this. Yeah, now we got the response and the response is very simple. So people vote.
The whole school, imagine your whole school is voting to select
a class monitor, and a party is
nothing but a team. The captain is nothing
but the chief minister. Yo, so even a child can
understand this, right? So this is how we can
make a concept very easier for us to
understand using HAG PT.
7. Roleplay : In this video,
we'll be discussing about a concept
called role play. So role play is nothing but
giving the machine or giving the um a role and starting the conversation
according to that. So, for example, I'm getting into Chagp and I'm
using this form. I'm saying the machine that
you are a traffic police. I'm converting the machine to a traffic police in Chennai. Stick to the role and
don't be generic. If I have not given you
enough information, please let me know and what
information is required. Forget that you are GPT. Don't start the conversation. Let me start the conversation. Okay? Now I'm converting the machine to a
traffic police, okay? Let me enter that's all. So now just imagine
that you are talking to a traffic police, not haibtoy? Cha Gibt will be very generic. It will give you a reply for anything
that you ask, right, but now we have changed
it to a traffic police. Now, let's ask us
something like, let's ask what is your name? It's giving you a reply, I am Sub Inspector Arvin Kumar
from Cheney Traffic Police. Okay, let's ask, did you
have your breakfast? Honestly, no time to have breakfast today since
630 in the morning. I'm in duty near Tena but Tena Bit is actually
a place in Chen, okay? School traffic online so now we have converted the machine
to a traffic police. Now let me ask, what
is your salary? Okay. It's telling that
the salary is around 38,600 and the first three
words, right, it is in Tamil. So that's how you can change the machine to what you want. So this is the concept
called role play, and you can use it for anything. For example, if you are a
person who is working, okay, what you can do what
you can do is you can give it a role
as a senior person, a senior marketing
consultant or whatever. Maybe your reporting manager. If a person assume a person is having more knowledge in
in a particular topic. Just give that role to the
machine and ask the questions. And sometimes it will not work. For example, each
and every company will function
differently, right? So what I used to do
is I'll give the SOPs. For example, a company will be having a set of rules on
how it is functioning. I'll give the set of SOPs and
how we are going to operate and I'll give the machine I'll give the
machine a name also. I'll say your name is Jack, and I'll give the SOPs
and I'll give it a roll. Now if I start a conversation
if I ask any doubts, say, how can I do this
task or something, I will give me
personalized reply, I will not give
me generic reply. So you can definitely
assign any role and you can get the
information that you want. So this is all about roleplay.
8. Book Summary : Book summary. So this
prompt is to get the highlights and
the key points of the book even without
reading book. Let it be any book. So it is only that the book
should be online. It is not that it
should be very famous or it should have an
ISBN or anything. The only thing is the book
should be available online. It should be little
famous, that's all. So let's get into HAGPT so what I'm doing is I'm
just giving this prompt. Provide me the key ideas of
this book, Art of winning. So here you can specify the
book name that you want. You don't have to read the book. Even without reading
the entire book, you are going to
get all the crisp of what is there in that book. So Art of winning by RaaGrienb I'm specifying
the title of the book, followed by the author's name. Okay, so let's head enter
now, and that's it. So now we have the highlights and the summary of what
is there in the book. I did not even read the book, but now I know what is
there in that book. We also have a summary table. Yeah, so this is how even
without reading a book, you can know what is
there in that book.
9. Place it in a Table: Place it in a table. This will arrange all our data
in a table format, which will be much easier
for us to understand. So let's get into CharPT. This was my prompt, create a detailed read
itinerary itinerary. Okay. I just asked for an
itinerary to visit Delhi, I've given a prompt for that, and this is a response. So it is telling me day one, morning, this is the
place I have to visit. Afternoon this is the
place I have to visit. Evening, this is the place and similarly for day two
and for day three. And it is also giving
me some travel tips. Okay? So this paragraph is
not good to read, right? We will not be very
much comfortable to read everything in
paragraphs, right? So we just have to
go with this prompt, which is place it in a table, place it in a table. Okay. So what this will do is it will rearrange all these in a table format. That's all. So this is how the data
was in paragraphs. And now we are getting it
in a proper table format. Yeah, so now here is the result. Now we got all our data
in a table format. Day one by time or this time, I have to visit this place and these are all the highlights there and food
recommendations as well. It's telling me
this cafe will be good tips also visit early
for peaceful surroundings. B 12 30, I have to
go to this place. This is what I wanted and initially I was getting
some raw data like this. You can use this prompt, place it in a table.
10. Renovate Home, Create Poster : In this video,
you'll be learning how to renovate something,
renovate your house, renovate your street
or something using AI, and then you'll be
learning how to create a poster and also how
to create a logo. Now I'm here in chat GBT. So what I'm going to
do is I'm just going to input an image of a house. I'm just adding an image. I'm just adding an image. Let me show how the
image will look like. So I'm adding this image. Okay? I'm adding
this image to LLM, and I'm trying to renovate. And this image is now added,
and this is my prompt. Take this photo of my
entrance and imagine how it will look if it was beautifully renovated,
green and peaceful. Show me the afterimage. Okay? I'm asking AI to
renovate this place. That's all, okay. Let's see. Here is the result. So this is the image
that we gave as input, and we asked to
renovate this image. And now AI has done a
fantastic job, right? So it has added a fan here. Everything is very,
the dimensions and everything are very,
very similar, right? It has added a new gait so even the space mismatch
or nothing is there. I think everything
is really good. So you can definitely
try this one. So now we'll get into
making a poster. So this is familiar
to many of us. Maybe if you are not,
you can also create a poster just with hagipety. So your prom should be make a colorful poster or create it. Create a colorful
and emotional poster for whatever you want,
for whatever you want. For of now, I'm using this alumni meat in my university have
given the date also. Let's see how the result is. Okay, so now we have the result, St. John's University
Alumni Met. Okay, similarly, for
whatever you want, you just have to start your
prompt with create a creator, then the details that you want. Once you do this, you'll
get the poster like this. And sometimes there are many possibilities that there will be some spelling
mistakes in the image. Okay, so in order to avoid that or in order
to make it better, give the details
in double codes. Here, we don't have to do it because details
are very minimal. But if you're having
more details, for example, St.
John's Universe rate, just give it in double codes so that the spelling mistakes, issue and all will reduce. Yeah, so the next one that we are
going to do is we are going to create A logo, okay? Create a logo for youth movement
called Next Generation. This is a simple prompt, but you can specify, if you want a theme, for example, if your
theme should be brown, you can specify that if
it should have some image and if it should have an image or if it should
have a symbol or whatever, you can give more details and us and you'll be getting it. Yeah, so now the
result is ready. So our prompt us create we
started with Creta logo, and the logo is now ready. And also, there
is one more place where we can do the
same, which is Canva. You just have to type canva.com, then a slash and AI. Okay, this will be the URL. Once you do this,
you will enter into the artificial intelligence
mode of Canva, and you just have to click
on this design for me. Once you do this, if you
want to create a logo, you can directly click on this and describe what you want. Describe all the details. For example, if you want them to be in red color or whatever, you just have to do
that. It'll be done. Or if you want to
create a presentation, social media marketing or
if you click on marketing, you'll get options for if
it is a flyer or what. So whatever it is,
if you want to create any infographics, I think Jana J will give you a better
result. You can try this out.
11. Time Machine: In this video, we'll be
discussing about three prompts. So the first prompt,
I'm uploading a picture of India in 1940s. Okay, so
this is the picture. I'm uploading this picture, and I'm asking this is old
photo of India in 1940s. Write a short story about
what their day might be, like how their lifestyle was. Okay, let me hit Enter. Yes. So we asked for a short story about
this period in India. And yes, we have
got a short story. Maybe if you want
to write some books or if you want to get some
inspirations about things, like, you can definitely add an image and ask for a
story and it will be done. So this is that it will resemble it will
give you words about this picture and also how India was in that year.
Okay, we have got that. And now let me get
into the next prompt, which is these are my
set of skills, okay? I'm adding a set of skills and asking some ideas
to generate money. So these are my set of skills. Digital marketing, content
creation, designing, give me unique ideas to make
money out of these skills. So you can add your skills and definitely as to
what it can be done. Now we have the
result. Yeah, based on the input that I gave, it has given me ideas, the execution strategy, and
it has also given example. AI prompt Pack, mini courses. I can do some many courses. So execution strategy, how
the strategy will work. And it is also giving
me an example, and it is giving me package
ideas and everything. So similarly, you can give in your skills and ask
what can be done. Okay? And let us get in
with the next prompt. So I earn this much, whatever is your
salary or whatever, and I live in Chennai, help me make a monthly budget plan. Just give in and
with this prompt, you can ask A to give you
a monthly budget plan. Now we have the result,
and this was a prompt. I earn this much per month, and I live here, and you can add in whatever
details you want. You want like, if you want
to invest in something or if you are a movie person, or you can add everything together and you can
ask for your plan. So now it has given
me a good plan. So 50% of income. I have to use it for this saving investment I have
to do for this. And at the last, it has given me even the date. First to third, you
have to pay the rent, EMI insurance and things fifth to seventh
what I have to do, by the end of the month,
what I have to do. So tax planning and everything. Got it. So similarly, if you want a budget plan, you can go for this prompt.
12. Sentimental Analysics : Sentiment analysis. So this is nothing but to do an analysis of the
sentiments for your data. Let's say, for example, if you're having a set
of reviews of a product, a hotel or whatever. So just to get the sentiment, even without reading that data, you can go for this prompt. So what I'm doing
now is I'm pasting the review of 1010 reviews, okay I'm pasting ten
reviews of a hotel. So you can go for
number of reviews, and this should be your prompt. I will give you a
set of reviews. Your task is to do sentiment analysis of customer
reviews of the restaurant. And also, I'm specifing that
the label must be neutral, positive and negative. Okay? Also, add a column for
improvements and positives. So this is what I'm specifying. Okay. Okay. Let me
go through this. Yeah, so now we have
the result and yeah, the sentiment for the first
review is a positive review, and these are all the
positives in that review. Excellent food, flavors
are good and things. And also, we have
a negative review here and serve hot food. Maybe that customer did not get hot food when it was served. And here, again, we
have a negative review. The dining experience
is not good. Okay? So this is
how even without getting into the reviews or you don't have to read
any of these reviews, but you can get what are
all the improvements required in your
review in your data? So that is all about this.
13. Specify the Length : Specifying the length. So this is to specify
the length of the response that LLM gives us. So let's get into harPT. So my prompt is generate two to three sentence about why prompt
ensuring is important. So this is the key word,
two to three sentences. Yeah, so that's all
is the response. Okay. Now my prompt will
be generate a paragraph. A paragraph about why prompt
engineering is important. Or you can go give me ten points why prompt
engineering is important. So just giving something random, you can ask it you can specify the length
on the response. Now we have got a paragraph. So this is how you can specify the length and get the response.
14. Delimeters: Dimeters. This
prompt is to analyze the similarities and
differences on two articles. Let's say, for
example, if you have a news article 0N a particular newspaper and if you have the same
on another newspaper, let's say if there
is an accident. Let's say there is an accident
and if that accident is recorded on two newspapers on let's say Hindu newspaper
and the same accident was also written in a decan
Chronicle or anything. Okay? So now, you want to
get more details about that. If you want to get what are the similarities
on both articles and what are the differences? For in that case, we
can go with this. So what I'm doing
is I'm pasting. I'll give you a text
in double codes, and I'm saying I'll
give you another text in square brackets. Name the first text in
double codes as source one and name the text in
square bracket as source two. And this is the prompt. Compare both the text and tell me the final answer if
both are similar or not. Okay? So what I'm doing
is I'm giving a text. It can be anything, okay?
It can be anything. If you want to just compare both the text and no,
you can go with this. So I'm giving a text
in double codes. Okay. And the second text
in square brackets. I'm starting it with square
brackets. I'm ending it. So let me, hit Enter now. So the results are ready, and these are the similarities of both the details that we gave and these
are the differences. So to compare two articles, you can definitely go with this.
15. Dreamlab Image Gen : In this video, I'll show
you how you can create good quality images using artificial intelligence
just with the prompt. Okay. So first, we have to know what a prompt and what
prompt engineering is. So anything that
you type, right? Anything that you
type in the LLM, like ChatGPT or Gemini
or whatever, right? Anything that you
type is a prompt. For example, a man walking in a forest is a prompt,
which anyone can do. But what is prompt engineering? Prompt engineering is bringing
your thoughts to life. For example, a man walking
in a forest, right? So I'll have a vision of a man who's wearing a red colored
t shirt and walking. Maybe your vision can be a very old man
walking with a stick. Okay? So we'll be
having more visions. Similarly, AAA will also have many things to
consider in the mind. So in order to bring what is
there in our mind, right? So that is called us
prompt engineering. So now I'll show you how I'll show you a format with
which we can bring our thought to make
it a proper image. Okay? So first, we have to
get into Canvas AI. So I'm using first, I'll show you how you can get in here. First, you have to get into canva.com slash AI RL
canvas.com slash AI. Yeah, this is a place where
you can generate images. Just click on this create image. And yeah, for example, let's consider a person cooking. Okay? So this is a prompt. A person cooking is a prompt. But this person can
be a man, a woman, so that person can be cooking in the kitchen or
can be cooking outside. That person can be wearing
any dress, right, any outfit, that the lighting can be, it can be a morning time,
it can be an evening time. So there are many
things to consider. We didn't say anything,
but we are just saying a person cooking. So in order to make it
proper, we have a format, which is called cascade, CAS CAD in which the C is
nothing but the camera angles. There are many camera
angles, right? So a bird eye is
like a top view, worm's Eye is like an
extreme bottom view. Wide angle is for
showing more content. A long shot is showing
more details like an extreme long shot or an extreme close up shot is having the camera
just like this. It will just show
half of your face. There are many camera angle. First, we have to specify the camera angle,
a person cooking. So for this, I think a wide
angle will be good, right? So a wide angle, angle shot of a person cooking. Okay. C is for camera angle. The next one is ambience. Ambience is nothing but
the time of the day. Let's say it's early
morning, okay? In the morning. And the next one is the subject. Subject is a person,
but it's not person. We have to specify
who that person is. Let's say a women, okay? A women. Or let's be more
specific like an Indian women. Indian women. Or we can be even specific like middle aged. A middle aged Indian
woman wearing a sari A middle
aged Indian woman wearing a sari, cooking
in the morning. Okay. And the next
one is the context. Context is the background, okay? So let's say this
woman is cooking in a forest or valley, okay. Cooking in the
morning in a valley. And next is action. Action is nothing
but cooking, right? Yeah. And the next
one is dialogue. Dialogue or script, we
don't have to specify the dialogue here because
this is for video generation. We'll be discussing
that as well. For of now, let's
keep dialogue alone. And the next one is emotion. We have to say if the person
or the animal or whatever, if it is happy,
sad, or whatever. So let's say let's make
the women sad, okay? A wide angle of a middle
aged Indian women wearing a Sari and and cooking in
the morning in a valley. Okay? She is sad. I'm saying she's sad also. Now let's generate this image. Yeah, so now we got some very good images of a
person cooking in a valley. I think this one is good, right? We can also specify if it
should be realistic or what. Okay? So if we say
a realistic image, then all these images
will turn more realistic. Yeah, so this is how you
can use this format called cascade and bring out your
thoughts directly to life.
16. Lets think Step by Step : Let's think step by step. This is nothing but knowing
how the machine things. Let's get into ha gi Pt. I'm adding this prompt
before my prompt. How many fishes are there
in the Pacific Ocean? Before that, my prompt is
let's think step by step. Now it will show us
how it is thinking, which source it is getting into, how it is finalizing the answer. It's going to give us a reply, that'll be at the last
and the thinking part, if you want to know
the thinking part, you can go on this. Let's think step by step. Now we have the result and
the machine thought for 1 minute and 31 seconds and it's saying us how
I'm estimating briefly. Scientists global
counts on whatever. It's saying the thinking part, how I'm estimating briefly and why the range is so widely. So this is nothing but the
thinking part of the machine. So if you want to
know how the machine things, you can use
this prompt huh. Let's think step by step.
17. Tokens: In this video,
you'll be learning a concept called tokens
in prompt engineering. So a token is nothing
but a measuring unit. So let's say, for example, we are measuring our
weight and kilogram, we are measuring liquid
in litters, right? Similarly, to measure
prompts or to measure data, we use a term called token. Okay. So now let's
get into an LLM. For example, let's say, charge Equity, I'm getting into this. I'm trying to summarize, okay? I'm using this keyword TLDR
to summarize a transcript. And this transcript is
around To us of a podcast. Okay? This is a very
long transcript. And let me try to
summarize this. Yeah, I'm not able
to summarize this because it's giving
me a message. The message you
submitted was too long. Please edit and resubmit. So this is where this concept
called tokens will come in. So what I'm doing is I'm getting into a place
called tokenizer. So in order to get in here, you just have to type
tokenizer in Google, and probably OpenA
will directly give you a thing called tokenizer, which you can click
in and get in. You just have to paste the text, and it will give you the
number of tokens it has. I'm pasting the transcript, and it's telling me that
it has 32,142 tokens. And it is giving me the
number of characters as well. So tokens are always not words. Let's say, for example, a
cat should be one token, each and every space
will be a token. If I hit space one time, it will become two tokens. And not all the words
are taken as one token. Let's say, for example,
summarize, sum arrays. So will be taken as okay, it gives us the it gives us
how it takes the token also. Sum is taken as a token. AR is taken as a token and the remaining is
taken as a token. So three. This is how tokens will be measured on an average, we'll be getting
like 70 to 75 tokens for every hundred words. Okay? So this is how
tokens will be measured. So we are not able to if we are not able to
process what we want, then we can get into a
concept called chunking. So chunking is nothing
but just splitting up. Okay, if you want to
summarize this entire thing, and if you are not
able to summarize it, you just have to copy half
of it for summarize that, then take the next chunk,
then summarize it. So this concept is
called a chunking. Or, the other thing
that you can do is you can get into
some other LLMs. So Charge Body has only 8,000
token limit for free users. And maybe you can
go into Claude. Cloud has around two
lack of token limits. So if you are
getting into Claude, if your token limit
has to be more, and if you just still want
to be as a free user, you can definitely use
Claude or any other LLM. So yeah, this is the
concept of tokens.
18. Conclusion: Mm hm. Congrats on
completing this course. Do check out my
other courses and also do give me a review
for this course. Thank you.
Joshua Israel, Cloud Engineer
