Prompt Engineering 101: Zero-shot, One-shot, and Few-shot prompting

hey guys what's up welcome back to another video of pronged engineering course where in this video we'll be discussing a type of prompt engineering and that is known as short prompting which are zero shot prompting one shot prompting and few short prompting so before starting this video let me just demonstrate this slab to you guys so we'll be using prompts to get output from chat DBT in an effective way to enhance focus and precision of our output and also we'll be applying prompt engineering techniques to improve our output as always so moving forward uh let's

just see what is short prompted so short promptings basically it has a three variations which entail giving chat gbt complete autonomy to generate any response with zero shot prompting and you know you can place parameters and you know you can give a reference data as well which happens in one and few shot prompting so if it's still unclear uh don't worry the next example will make you guys more clear so in zero short prompting you don't give any previous data or guidelines you know before completing requests so let's just say you are giving any prompt

to chat gbt right so for zero short prompting you won't be giving any previous background details or any other information so you will right away get into your front without any extra information or any heads or detail but in one shot prompting you are only allowed to give only one piece of data or guideline complete the request and similarly for few short obviously by the name is suggested you can add multiple pieces of data or guidelines to complete your request so yeah I hope it gets clear to you guys and let's just now move forward

and see what's the from formula for zero one and few shot prompting so for zero shot prompting you straight away a right prong with no reference that's the formula for that and for one shot prompting you put one example or one reference and then straight away after that you give your problem related to that example or reference and in few shot obviously uh you use examples One Two Three or as many as you want as your reference and then you write that prompt related to that reference alright so these will be a few of the

example that we'll be doing in today's less without further Ado let's just jump right into our chat GPT alright so here I am with my charger video open and first of all I will see how the zero shot prompting works so I'm just gonna give it an example to write me a YouTube script for my tech review channel so in this example as you guys can see like I didn't give any information or further anything like you know what exactly what is my topic for my tech review Channel or what Gadget am I reviewing in

so yeah I didn't give any information so yeah let's just see how Chad gbd will get output for this all right yeah so as you guys can see with my first example of zero shot prompting didn't give me like a YouTube strip for any specific topic or any specific Gadget it just provided me a template which I have to edit and you know fill it up according to my need which is quite a bit of a hustle if you ask me and yeah so this is not something clear maybe topics or you can say the

things you provided in this IBM not in need of that so yeah this is a very you can say undirectional way of using chat gbt with just zero shot prompting so if you move forward let's just do uh our second example which was one shot prompting so for that I'm gonna add Sajid now to write me his YouTube script for my tech review Channel related to the latest iPhone so yeah in this I just provided one kind of information or reference which is for the latest iPhone so let's just see whether it will improve uh

the output for us this time or not so let's just hit enter and see that all right yeah so now you can see like as soon as I provided uh with a little bit of information it was just one reference which is latest iPhone it just provided me a complete dinner first of all you start up with the unboxing then you can do like a footage of iPhone packaging and you know now it is a little bit more precise but you know you cannot say like that it's a perfect way of using charge video that's

a perfect way to you know master your prompt engineering techniques so still like you know uh as I told you guys earlier as well like the more you put effort into chat GPT the more you're gonna prompt your activity the more best and best kind of outputs you're gonna gain from AI so yeah let's just now move forward and do our last example uh which was few short prompting where we'll be providing more than one reference to chat GPT and you know we'll try to get the best output possible alright so this time I said

DPT to write a five minute YouTube script on the latest iPhone camera specification for my tech review Channel start with a 10 second hook and rotate a phone for each main point so yeah as you guys can see I provide multiple kind of information and reference which igbt now I said you need you need to make a five minute YouTube strip that is a one reference then I have the second one as latest iPhone camera specification and a third one to make a 10 second hook and rotate a photo for each main point so yeah

we have like three to four references in this example so let's just hit enter and see like whether we'll get a perfect output for that or not all right yeah so now you can see like you know I got the perfect output from Chad jibril just the way I want it so this is all uh for the iPhone camera in this example because I wanted to make a script on the latest iPhone camera specifications first of all it says you need to uh go for the high resolution capture then you have a stunning cityscape photo

to display on screen then we have a little bit of information about night mode then we have for portrait mode as well and then we have for Ultra wide angle which is also very important and then we have for advanced video capabilities and you know with that now I can say like you know I did use charge DVD in a very perfect way providing as much information as required and you know to get the best output possible from chat gbt you know that's that's the way to use AI in my view you know you should

prove to be more knowledgeable to AI I know you need to give like as good information as possible to get that perfect and the best outcome awesome so yeah that was pretty much it for today's video I hope you guys liked it and don't forget to watch more videos on this course and I'll catch you guys in the next one take care bye foreign