Build your own LLM AI on a Raspberry Pi

hi welcome to wiscat okay today so what we're going to do is uh this is a followup on uh poster presentation that uh my friend Luke and I did at a conference recently where what we did was we demonstrated creating your own uh AMA AI server so we built an AI server running an llm and what we were mainly trying to do is um demonstrate that you don't actually need to have some really hardcore Hardware or something to do it um you can actually run it on a Raspberry Pi a Raspberry Pi just pictur it over my shoulder here is actually this tiny little thing this is this is the whole unit right and oh the green screen is coming through yay but anyway um this is how small it is right it's tiny it's a tiny little thing and this can run your llm right there are newer versions um editions Raspberry Pi 5 I've actually only got a Raspberry Pi 4 here so I don't even have the latest edition but it was enough to actually build a a unit a self-contained unit that with no internet connection was able to actually serve its own uh Wi-Fi and um uh basically people could connect to it and use the AI inside of it without even so much as an internet connection so it's completely Standalone and so we thought you know that that'd be pretty fun to show off um why did we do it uh well because we can um to demonstrate that it's possible uh so it's it's got no real practical use I don't think but um if you actually do want to do it for practical reasons uh you might want to have some bigger Hardware of course if you want to have multiple people using it but for example for just a handful of people who want to try it out and have a little bit of fun these things are awesome I raspberry pies are awesome in general so um they're just a lot of fun so what we did was we took uh AMA which is an open- Source platform for running uh llms and um well basically AIS and we shoved it into a Raspberry Pi and then we got the Raspberry Pi to have its own ad hoc uh Wi-Fi network so that people can just connect to the pi using their own smartphones or tablets or something like that and then uh that'll all just work so uh what do you need to actually do these sorts of things well um if I just I'm going to bring up that that's a nice background so yeah let's good this one so here are the ingredients for making one of these devices so you're going to need a Raspberry Pi yay got one and um of course you can get a Raspberry Pi 5 and the new AI hat and stuff like that that might even work better but at the moment I'm just running a simple old Raspberry Pi uh you will also need one micro SD card these are what uh the Raspberry Pi is actually um put the OS into and that's where the software is installed and so it doesn't have to be all that big even and just you know say uh a 32 GB one should be ample for this purpose uh you're also going to need a a Micro SD card reader I've got a little USB one here um so this is just so that we can plug this into the computer and put the OS on there now when it comes to um software what are we going to need well we're going to need the Raspberry Pi imager software which is this one in the background here and I've already installed it on my machine uh you're going to need anable um if if you want to do it exactly the way that I did it you don't have to uh have anible you can type in the commands and everything by hand but uh anible makes that easier and uh generally you're going to need git as well but git also uh is so that you can actually download the GitHub repository which I'm about to um introduce you to and links in the description below for that and also because we're going to be dealing with an AI you want to hammer all right you want a hammer now the hammer is mainly for the purpose of you know if the AI actually becomes sensient and decides that it's going to take over humanity and destroy us all then you know this thing is fortunately small enough that you can you know take care of it so Safety First anyway uh let's get into it shall we uh so the first thing that we're going to want to do is we're going to want to do our imager thing so I'll shrink myself down and I'll get rid of this uh you don't need the terminal just yet so what I am going to want to do next is to grab my Raspberry Pi images so I've downloaded this um by the way actually I'll I'll just quick mention of uh where to download that from oh this is the Alama web page and I'll show you these things in a bit um just because I want to uh I'll get start the Imaging the that takes a bit of time so once it's done I'll do it but the imager itself is from the raspberry piie uh.com page uh they have this lovely Tool uh it works on uh Linux Windows and Mac and so for me it was really easy is just installed it like that in one command and bang I had the Raspberry Pi imager that's actually all you need for installing the image and I'll show you just how easy it is and there are actually other things that are other images and everything that are really cool in there like Moodle box and things like that which are make it really easy to make one of these that actually runs Moodle but anyway um that's what I'm using to do the installation and here's the raspberry pie imager itself so uh what I need to do is actually grab my SD card uh where' I put it right and I've got my little card reader thing here and I'm going to insert my card reader in the right way I think that's the right way and I'm going to plug this into my machine here usba a did I get it the right way no turn it around did I get that the right way no uh turn it around again I mean it's 50/50 why does it always take three times that's not going in at all jeez jeez Louise what's going on down here all right there we go it's in usba this time I had to turn around four times I mean that's ridiculous okay anyway uh it's plugged in so that means I can actually choose my storage and I can say yeah I want uh this uh um I'm I'm not overriding my hard disk in my computer that's the main thing you want to check there uh it's going into the sdmmc blah blah blah blah blah all right next one we want to do is we want to choose the operating system now choose the operating system for what we're going to install the pie and there are tons of options in here and they're um really quite good uh so I think this is like the ones I've used recently um oh no no these are just the first three that pop up at the top there are ones that you can use for a desktop um Ros I mean this this thing can become a desktop computer that's pretty impressive for something in this you know this size um so you can actually use it as a full-on desktop if you want um it's not going to be breaking any speed limits or anything but anyway this time what we're going to do is do raspberry pios other oh and by the way while I'm here I might as well just show you in here uh there are sorry uh other general purposes no no uh there's a special T version uh specific purpose OS and in here you can actually see Moodle boxes right there so if you want to try out Moodle box and I'll probably do another video following up on um the changes that have happened in mood box lately but they're awesome and they're in there as well there are also a bunch of other um Freedom box cool uh anyway uh there's there's a whole bunch there are a whole bunch of options in there but what we're going to do today is we're going to just use general purpose Raspberry Pi OS one and what we want is this one here Raspberry Pi OS light now light basically means that it's command line only we don't want anything additional we don't need a desktop environment on the p and it's going to take resources if we use the Raspberry Pi desktop environment so we don't want that we want it to have no desktop environment it's just going to be just you know the Bare Bones of of what you need to actually switch it on so we're going to go there and now what we're going to do is going to come down here and we're going to click on this um gear icon or Cog icon as the case may be and we're going to do a few things we're going to change a few settings here now the settings I'm going to pull over to the side here for a second cuz I do do want to go to my GitHub page here that has the py Lama project right why did I call it py Lama because it's Raspberry Pi and AMA put them together and if I called it llama P that sounds really weirdly carnivorous so um anyway I'm going to go with uh py for this one okay a Raspberry Pi we've got that Debian based OS like um uh Raspberry Pi OS light 64bit that's the one I just chose and now I'm going to set the installation parameters here so you can see in the instructions of clearly laid out here host name should be set to py Lama right and one of the cool things about this is that Pi raspberry posos uh will'll actually make py. local this is actually going to become the the the URL that we're going to need to access it because the multicast uh address is going to be py. local and that's automatic Ally going to be um shouted out across our network uh later on so it'll make connecting to it and everything a lot easier um also uh I mean I I think actually on Mac OS uh this is multicast is implemented in a a thing that they call Bonjour um anyway that's you know that that's what we want we want the host name to be set there now we're also going to want to be able to connect to this thing so we're going to enable SSH and I'm going to use uh pass public key authentication only because I want to use an anable Playbook with this I want to be able to use the public key authentication to actually connect to it and do everything automatically um so I'm going to set up my uh public key authentication with my public key uh here this is uh the public key for my SSH for this computer that I'm using right now so I've got I've already pasted that in um before so that's that's all I need right I'm going to set a username and password as well and I'm going to change it it's saying here uh set the username to Pi or change the anible unit I'm I'm just going to set it to PI right so Pi so username should be pi and the password should be I don't know password yeah um you should probably use a more secure one by the way this is not secure in the least so um if you're looking for a tutorial on how to make a secure py Lama server this ain't it um as I said this is not for a practical purpose so much as this is just for fun all right now we're not going to need the uh Wireless L because we're actually going to use anible to um uh connect uh to it over the wired Lan we're going to actually plug this into my router uh here so I can control it from this computer uh and when anible does set it up it's going to set it up as um it's going to set it up as a uh its own Wi-Fi access point so going to do that I might well local settings are not really that necessary but anyway um because I'm not going to be directing connecting directly to it I'm going to be sshing into it so but anyway uh if I go I don't know JP for the keyboard layout why not anyway uh that's that done because I got a Japanese keyboard wow anyway uh eject media when finish play sound when finished enable Telemetry yeah no thanks uh but anyway that's basically all you want to do is make sure importantly that the py Lama host name is set the username is set to Pi and the authorized Keys uh that you've added the authorized key from the machine that you're using to do the install okay we're going to hit save on that and now we can do the writing and click on that all existing data will be erased oh no well it's brand new thing so who cares uh you do want to continue yeah I do boom and it's going to ask me for a password because it's writing to a dis uh generic SD yep it is the SD card good because I don't want to destroy my computer okay anyway that's that's running and it's writing to the SD card and as I said that's going to take a little bit of time so while that's happening uh let's talk a little bit more about AMA itself so AMA itself here you can see get up and running with large language models now of course we're going to I mean there are plenty of options here llama 3 5 3 mistl Gemma Etc um the different models you can choose which models you want to actually run uh once it's up and running so if I go across to uh the models section here we can actually see which model we're going to be running so today we've got llama 353 I yeah that's a multilingual one right 23 languages in stuff like that but these ones here like IR that's an 8 billion at the minimum and 35 billion parameter uh model uh the 8 billion one might run sort of okayish on a Raspberry piie mistl also be a bit slow but it would probably handle it I mean the Raspberry p that I've got also has 8 GB of RAM it's kind of an important point there because um these things chew up a lot of Ram uh when they're running but things like mixol you know and like 22 billion parameters or you know 70 billion parameters for llama 2 or something you're not going to be running anything that big in fact we want a nice small one so that we can run it on a Raspberry Pi and so the one that I'm going to be running today is surprisingly named for a Lar large language model is this one here tiny llama now tiny llama is a little 1.1 billion llama module on three trillion tokens and stuff like that so you can get more information about this one here um of course the actual information cool logo uh but the actual information um inside of it uh because it's only 1.1 billion parameters yeah it's not going to be super you know um it's it's not going to be able to answer all your questions and stuff like that um and I it's really really unlikely that it's going to become sentient and want to take over the world uh but I've got my hammer just in case anyway um this is what we're going to run and because it's small enough we can run it on a Raspberry Pi uh if you're running on say your laptop or something like that you may want to go with a different model uh and choose a different model that has you know more capabilities or four more specific ones so there are ones for example like Commander code Gemma um this is for programming coding tasks and things like that uh lava is actually for describing pictures and stuff like that and um there are a whole bunch of you know different ones there there's a Chinese llama um I've seen like medical ones there's Medi Lama or something like open hermies is not a bad one as well well that one's I've seen an implementation of that one with um sort of like a game thing so you can actually have it to make one of those you know uh you've entered the room there's a chair on the right hand side and a chest on the left what do you do a open the chest B sit in the chair and stuff like that so that's pretty pretty fun too anyway there are tons of them I mean look at all these and these are all freely available so you can set them up on your own uh device no worries oh there's there's the medical one yeah Med Lama 2 uh fine tune to answer medical questions based on an open source medical data set woohoo so um there's a whole bunch of uh specific purposes and if you choose uh an AI for the specific purpose you're interested in um you can try it and see how well you fair um you know there so much diversity out there that uh I really can't tell you which one's the best one because I don't know anyhow um the other the next thing is that uh yeah we're going to need use anible to run it and by the way anible if you want to install that um there are installation instructions on this again I'll leave the code uh the uh links to this page in the uh description below um and by the way while I'm at it I'm just going to stop right here and say hey like And subscribe if you like this kind of content because um you know that really helps out the channel apparently um I don't know all the other YouTube say that so I'm I'm checking in all right so uh here the other thing in here is that we're using anible to do what is the question so if we look at the um setup. yaml file here this is um wait what is that I don't even remember oh this is okay so this is just um I don't I didn't know that that file was in there I thought I just called it Docker composed. yaml that probably oh okay okay so this is basically the docker compose file and what this will do is actually this will call um a few of these different images here this is the docker compos oh sorry yeah yeah yeah sorry I I've completely forgotten how I did it the setup. yaml file what that does is that actually sets up the pi with its basic install and all it does is actually first of all installs the server packages so it'll install htop because I like being able to see what's going on it'll install git to the um to the server probably not NE needed actually oh look at that I can now remove the SD card from the reader yay uh it's going to install snapd um I don't remember why I put snapd in there it's going to install dock compost and Docker compost absolutely is necessary because later on we're going to use that to set up the actual uh AMA thing itself it's going to copy the docker compos file across to the pi um put it in the Pi's home directory just so that you know it's going to be in slome slpy but it's going to be there so that we can uh we can just run it straight away you know without having to mess around copying files across and then the next thing is it's just going to run a couple of commands and you could run these just by copy and pasting the commands in here nmcl and stuff like that and it's just going to grab these built-in commands and uh the these commands and run these to set up the Wi-Fi interface to set up the WPA on it it's like the only secure thing in there it's going to set up a password which is py WiFi um so there you go uh and it's going to set up a Wi-Fi as an access point um so that it becomes a hot spot that we can just connect to directly and I will uh try to demonstrate connecting using my tablet here anyway that's uh what that's going to do then after that we're going to run the docker compose uh up file and that should actually start up install um a copy of or Lama and uh the web user interface so that we can actually access it from browser um they're actually two separate parts but the docker compos file installs them both so if you want to have a look in the docker compos file see what that does uh here's actually allas the the background stuff is up here at the top um amaama Lama is going to just grab it and make a little directory for it a little volume for it to store some data like the model and it's just going to give it a command uh container name then it's going to install open web UI which is a a web user interface for uh and other I it doesn't have to be you can actually use this open web UI to make a a website that connects to chat GPT or or open AI or something like that but in this case we're connecting to the AMA inside the machine that's going to um connect to the orama base URL it's going to specify that we want to connect to theama um system we just installed and it's going to make it available on Port 80 so that we can just you know HTTP straight directly to it and that's pretty much all this does um there's not actually all that much to it um but it's in a nice little package here that makes it easy so anyway um Let's uh let's get the the dis is done so it's easier to pull out the uh USB than it is to uh plug it in and I'm popping out my my little SD card there it is micro SD card and I'm going to plug this in to the back part where the USB micrb slot is on the p all right and now I'm going to actually plug this in to the oh I can't get that on camera um but okay basically I'm going to plug it in to the um to the the wired internet connection here right so that's going to connect it to the same Lan as my computer that I'm using right now to do the install and it's also going to give it an internet connection so I can actually install that other stuff like the AMA um stuff and get how and stuff like that so I'm going to do that and oh I forgot to uh one more thing that these things do need is electricity and I forgot to plug this in so I'm going to pause the recording while I go and plug this thing in okay okay so now I've got that plugged in the back there so I'm going to plug the actual power into the pie right now and I'm going to fire it up okay and it is starting to boot up now in the next in the next uh few minutes what it's going to do is it's actually going to boot up the operating system it'll you know it's a first boot for the operating system itself so it's then going to um connect to uh well it's going to set up the SD card make sure that the uh the actual image takes up the full SD card C disc and do you know basic boot up things so basically once you plugged it in you want to wait a few minutes and um it'll actually do a reboot once and then it'll be ready to use so it takes a few seconds to do it anyway um yeah so we're actually getting pretty close to uh done I mean we've done the the part that takes the most time is actually Imaging the the Pama sort of a thing really uh except we do need to I am going to need to um SSH into it so let's see uh SSH uh Pi at and remembering that we go Pi Lama whoops Lama p p pil mama no no no P Lama do local and here's where the bonjour stuff um really comes in the py. local is going to allow me to connect to it directly like this and looks like it's not cooked yet not ready yet okay could not resolve okay let's give it a few more minutes I can see that the the power LED on it is actually flashing to say that it's doing stuff it's working it's busy um you know come back later okay the red light went out is that that's not a good thing is it okay uh oh there we go all right now actually I've done this once before with a different pi and so here we go remote host identification has changed it's possible something someone is doing something nasty yeah uh okay um if you've seen my other video on SSH um you probably know what this means but what this means is that last time I tried to connect to py. local it had a different signature on it of course because this one is a brand new installation so uh this computer has never seen it before and this is saying that the remote host identification the py when we asked you py who are you have I seen you before uh it's not a known host so it's got a um anyway this friendly message tells me how to disregard it now don't just if if you do if you are sshing into service and you see this be careful because it could well mean that um uh someone is doing something very nasty um but in this case it's not it's just that I rebuilt this this is a fresh install of py local and things like that so I'm just going to run that command which does uh take care of that it's going to you know save my old known host file and it's going to um remove these uh known hosting so then I can ask this say into it and it'll ask me do You Know This Server um and I will say yeah yes please type yes or no I thought I did all right anyway there we go so now I'm logged in sshed into the py Lama and if I try htop um oh Hop's installed how about that fancy that Raspberry Pi OS has htop by default so it's not really going to be necessary it's not going to be a problem either all right so anyway the the important thing is that I can log in uh that means that I can check that the pie Lama the Raspberry Pi itself is ready for um you know action so in here what I'm going to do is I'm going to go to wait let me grab that all right I'm going to C put you over there all right so now behind me is the GitHub repository and what I do want to do is I just want to GitHub this I'm going to clone this locally on my my computer machine I'm no longer logged into um py Lama the Raspberry Pi so I'm just going to copy this code here and I'm going to go um just in the throwaway directory get clone blah right and if you don't have um I mean that that was fast right if you don't have it you can actually just grab the code and download as zip unload if you don't want to install git on your home computer by all means feel free uh knock yourself out with that so anyway now that I've got that in here if I look in here I've got my uh py Lama is there oh that's the poster that I was showing you before uh so I can CD into py this is the directory of that git um GitHub repo I just found and we can see that we've got the dock compos yaml is in there uh we've got the inventory. yaml the readme ID and the setup. yaml so actually before I do I'm going to have a quick look at the inventory file here okay so so hosts R pie the username is going to be Pi the host is py. local and that's using the same bonjour thing again and it's going to you know say strict host G chicken no this actually what this does is prevents that um rude message from before from coming up and um cuz anible will also stop if that's not there so um I must have got I don't remember actually putting that in but I must have got tired of it um telling me no access toight but anyway if we go back here then we've got the next command to run is just this one here anable Playbook d i inventory. yaml uh setup yaml so this is just going to do the setup stuff here so I'm just in here where I've got my setup yaml file and I can just paste this in here and run it and that's going in and installing all the server packages that we need um it's going to be going through those uh tasks that we were talking about before setup yo so it's going to install the server packages of htop uh get iyn snapd and Docker compose I could probably cut out snapd ryn I definitely need because for this sync operation I'm pretty sure I could actually just have that download it directly from the GitHub repo yeah anyway uh well actually this is copy so it probably doesn't even use rsync okay hey stop it um anyway this is going to do the installation stuff that we need it's also going to um update the cache which might take a few little while yeah okay GitHub chill out taking a while and after it does that it's going to copy across the dock compos file set up the Wi-Fi interfaces and then that's starting setup will be done and we can just do our Docker compose stuff right by the way if I come over here and go py. local you'll see that I I've got nothing here and it's even trying to go to https that's cute yeah you think I've got https here all right it's Firefox it's going to keep trying to do that but anyway we should this is where we'll be able to connect to the open web UI uh thing after it's installed okay um I'm going to just pause the video until this uh Playbook starts moving on again o bang bang bang okay oh as soon as it started moving it really moved fast so anyway um it finished installing the um server packages on the Raspberry Pi it set up the docker compos file it set up the Wi-Fi interface set up Wi-Fi um psk set up a password and set up the access point and so now I should actually be able to well theoretically connect to it and because it's plugged into my local internet connection it's going to act like a uh a Wi-Fi access point just like my normal Wi-Fi router and stuff like that so if you know and it's got the username password also in there as from um set up as here so the password is actually going to be Pama Wi-Fi and the name is going to be py Wi-Fi with capitals and a a hyphen in there so that's pretty cool all right so let's go to the next one where're we're actually going to log into the um Pi again so sshp py. loal right now we're in here and if we LS if we look in this directory we can see that we've got this just single file this Docker compos an file that's the docker compos that was copied across when we did this here and so now if I go back to oh no no no no no not that this one here go back to the read me sort of thing we've got this last one here this is the last one this is actually going to install all the docker use Docker to install the images so if I just paste that in there and now I hit it and it should do the magic and so here it is it's created a uh a Docker Network there it's created the volumes and now it's pulling orama from um from the web so it's basically downloading the AMA packages uh for well theama [Music] um theama uh Docker container the the the image it's pulling the image from the the internet and it's going to set it up as a container now the thing is that remembering um if we go back to the library here this is going to install AMA the server on this pi and it's not going to install any of the models though so the models will have to get a little bit later okay um it's moving on to open web UI now and so open webui is coming in um a little bit of thing I'm just going to mention though um recently did another video uh where Jose Cruz hi um came in helped me out in installing all on a Mac and the way we did it we we actually did it using Docker but I've since seen um Docker and uh Native installs on an M1 Mac and the native install was Blazing fast and the um the docker was slow as a wet week so uh this is if you're doing you know an install on your your machine uh locally to run it locally on your laptop or something like this the docker install might not be the best for you um depending on what sort of Hardware you're running so I just thought I'd uh mention that there um so anyway at the moment we're just installing the or and empty shell and then what we're going to do next is we're going to log into the open web UI once it's built and we're going to download our first uh llm which will be tiny llama so come on guys you can do it's got a fair bit it's got like a gigabyte of um stuff to download so um I'm going to let that uh go and and I'll restart the video recording once that's done okay we're coming up on the end now um so yeah all the pulls and the extractions are completed now it's creating the ol uh container and creating the open web uui container and this especially running for the first time can take a little bit of time but uh and it is and maybe I started this oh wait a minute there we go all's done and the web UI come on come and there it is it's done and that's it it's basically built so now if we do uh pseudo doer PS for example we can see that I've got we've got our two um two uh images the open web UI and orama are both running on this this Raspberry Pi so now if we go to py. local get rid of the yes in HTTPS because it's just HTTP and there we can see it has started up here and so this is because I'm on the same Lan um of course I can connect to it just using py. local um so I'm going to enter an email address Adam at something but I mean it can't send email so it doesn't really matter there uh in password I'm just going to put in something literally I just put in s o m e t h iing g again this is going over HTTP um don't reuse a password on this one because the connection is not secure and as soon as you use any password here um it's gone over the the internet um separately where oh oh wait a minute sorry I'm trying to hit the sign in button and I can't because I need to sign up first right and the first user is um the first user is the um the uh going to be the admin user so if I go just to sign sign up instead of sign in and then bang I'm connected here and this is uh pretty much it running so H yeah we get the what's new in open web UI blah blah blah okay let's go yay and here it is how can I help you today and it looks like we're ready to go but we're not quite ready yet because as I said we have the empty shell we do not actually have a model yet so I'm going to full screen this one here and down the bottom leftand corner and open web UI and you can do these things um you can go AMA pull something something by you know uh executing ex executing a command to the the docker container directly if you want but um we've got ourselves a little uh web UI here so why don't we use the web UI so I'm going to go here to settings and I'm going to first of all I'm going to double check that the connection is okay orama 11434 right is what it's saying if we verify the connection o we get Server Connection verified that's nice and the next one that we want to do is actually go to models now models here this is where we're going to go um to our library here and remember we want to install Tiny llama so here's how simple it is you you just basically um you grab tiny llama and you go to your open web UI and it says here pull a model from Al llama.com and so you just got to put that in there tiny llama and that's it let's download that baby and pulling manifest and it's just going to start installing my model look at that see it's downloading the model directly to the Raspberry Pi now and setting it up so that I can run it and that's all you really need to do to get it running so uh it's coming it's coming and because it's tiny llama it's not going to take very long it's as I said Tiny it's a 1 billion parameter model it's probably about what 3 GB or something um it's not a very big file at all there are a few steps that it has to take to actually get it it up and running uh while it's doing this let's go back to that uh that thing here and HTP and just look at how um actually you know what I will but I'm going to put this over this side and I'll put the web browser over this side yay all right and so now did I just interrupt that I hope not anyway uh let's go all right download canceled a that's not what I meant to do oops whoops okay if you interrupt it Midway um it you get some success and some cancellation all right right so let me see come on if I refresh it ah tiny L it did make it in there okay so I've actually got Ty llama is now available here so let's ask it a question um because that's it we're we're ready to go and oh I can set this as default so I will set this as default here we go so let's uh let's ask it um let's ask it would you like to be a Moodle box or are you happy as a p py Lama right let's let's ask it it's I sorry I I made a note of this while I was waiting um so here we go would you like to be a Moodle box or are you happy as a p let's ask it a good existential question see what happens and uh by the way I'm leaving the uh HTP in the right hand side so that you can see the the thing light up when this starts so let's give it a try and boom look at that the the processing cores on the the P they're working they're working it's doing something what's it doing is it going to start to me come on wow look at that CPU there what do we got in there yeah just saying yeah it's the all server is doing it as a machine learning expert and experienced data analysist I am not in any position to Give opinions or judgments however it is safe to say that you're referring to a hybrid robotic system called Moodle box which combines The Best of Both Worlds human intelligence and automation then no I do not see myself as is one of them at pyas we strive to create systems that are intelligent reliable efficient and sustainable our products are designed to work alongside humans and make our world a better place so while Moodle box is an interesting concept I believe it falls outside the scope of our mission as pyas a that's so wow okay let's let's ask it somehow who is wise cat yeah who is wise cat let's see I me you ask it questions like this that it's obviously not going to know the answer to and expect hallucinations um CU that's all it's got would you like to be a Moodle box are you I mean I like it ask ask the AI for its its opinion on its own existence why not all right come on who is wisecat you should know this come on I am your [Laughter] father okay I mean arguably right okay let me see I get back over there I'm growing bigger just can you see me I'm over here come on Pama wow even to the p it's taking a fair while at the moment I might open the hood on this wi is cat formerly known as wise cat with a Canadian startup founded in 2017 by David lur all right I don't know there might actually exist a might exist I'm going to it's not me anyway um okay so that's that's cruising along nicely I did say though that you can connect to it using uh this uh like a a a uh a tablet or something right so let's do that let's do that next um what do you mean fingerprint doesn't match bite me show a different finger yeah gave it the middle finger that worked all right yeah yeah all right okay so let's go over so the first thing that we need to do to actually use this and this is where the poster came in um for the demonstration was in order to use it we need to connect to the uh the py Wi-Fi here and so I'm going to do that using this QR code here and by the way if you build it using the exact same um uh system that I did sorry I'm going to make myself disappear you can actually use these QR codes if you follow the exact same thing because this is like py. local is the the website so what we're doing is we're scanning this QR code to connect to the um the Wi-Fi network and then we'll scan this one on the right hand side to connect to the website it just has like HTTP Pama local that's that's what this one does okay you know what I I will bring myself back here and I am going to scan this um QR code so I need to grab my QR code scanner and okay it's found that QR code and it's saying connections connect to Pama Wi-Fi it is connecting right checking the quality of your internet connection probably pretty rubbish actually because it's going through a Raspberry Pi that's running py Lama right now okay uh internet may not not be available okay uh connect only this time sure okay so that's that part done I've now connected to the py. WiFi and next one I'm going to scan the one on the right hand side um which is behind me but from my perspective it's right there so I'm going to scan that QR code now and there it is and uh open link and what comes up is the py Lama right there whoa you can see my green screen okay so now if I enter my my email Adam [Music] tomu wise car no no no I need to be able to spill my own stuff Uhn net and password I think I just put in as what was it was just password right P SS w r d is that what I did or did I I think I went with something uh password's incorrect okay it's probably something s o m e t h i n g sign in and whoa save password no I don't want to save password all right um but yeah that has got me in I am now logged in look at that so now um oh and this one's not in dark mode but I've got the same thing and um up the top you yeah I've got py. local and I can ask it a question um uh what is Moodle let's ask it what is Moodle ask it what is Moodle right and I can hit enter on that okay and reduce this keypad and it's answering right and it's saying Moodle is a web-based learning management system designed specifically for educational institutions perhap particularly those that offer online courses or distance education programs blah blah blah blah blah yep so that's pretty good all right um so uh that's pretty much how it how it runs now actually if you want to try it out by all means uh I'm going to get out of the way again and let you have at that there's the uh there are the QR codes that you need um you can just jump on in and set it up okay the other thing that I would uh just like to mention before we go um is that this the Wi-Fi version right if if I actually unplug the network connection of course this one that's you know connecting via the wired connection won't work anymore but if I actually unplug it or or just you know um switch it off and then plug it in somewhere else I could take this to the the top of Mount Fuji or something like that and with a battery pack and run it and without even an internet connection use the um the AI to actually just run it'll work um can work anywhere without an internet connection um don't believe me seriously try it yourself um the it won't require an internet connection to run because everything is self-contained inside the pie so um by the way I I actually have two um the first one I took the cover off the first one this one is actually called plan a let's see I don't know can't see that but look it well trust me it says on the front plan a and this one's Plan B so always have a plan B uh but the thing is that it doesn't need an internet connection now once the model has been downloaded the model file is stored inside the pi so it doesn't need to con it's not sending that data to op open AI or to you know Google or anyone else this is self-contained inside this unit um so and and it's running on a Raspberry Pi it's running on this I mean seriously well this is Plan B but plan a is working this time yeah so and that that's pretty much uh it to actually introduce you can choose your own models and everything like that um but that's how you take AMA and stick it in a Raspberry Pi and get it work well that's one way of doing it um take that knowledge go forth give it a try um find out a better way to do it or something make your own YouTube video about it um seriously I and comment with the YouTube video you make in the comments on this video because I'd love to see it um okay so anyway that's all it is uh all for today um thank you very much for watching uh like And subscribe and do all the good stuff and uh until next time uh I'll uh you know enjoy your AI see you

Channel: WiseCat

Views: 2,105

Keywords:

Id: 6NCjTof5snk

Length: 51min 57sec (3117 seconds)

Published: Tue Jun 11 2024