Ollama with ComfyUI and Open WebUI | LLAMA3 & Phi3 | PDF Search Setup for Enhanced Performance

[Music] welcome to day JS this is a big day for me because today I want to show you how you can use llama 3 and stable diffusion together from a same system so that uh it'll make your life much easier here and give you a complete uh option to use the text generation and image generation together so for that I will use uh confi for the stable diffusion image generation and for Lama 3 I will use AMA and I will download the Lama 3 Model okay so in our previous video I showed you how you can install confi uh with zeloda on Windows machine for the AMD GPU users if you are Nvidia users you can directly install on confi and for the text generation I use AMA so let's start to find AMA search with AMA and if you go to their side you can see you have a download option click download and and download the windows version if you are using Linux you can download Linux or Mac so let's download this okay so let's open it and run the AMA setup it will start installing so let's install it so you can see it is actually um already getting the rockem so it should run perfectly okay and AMA is running background I get a uh you know notification you cannot see this because it's out of the range uh but it is actually running so that's a good news so let's uh start AMA and to do that I'm going to the terminal and here I want to download the model so let's go to the model here and the first one is Lama 3 so I want to use Lama 3 just copy this and it will download the 8 billion parameter version I also ran 70 billion version it runs well but it's low for this demonstration I want to use the 8 billion parameter so that it can run very fast okay so let's run the command AMA run Lama 3 and it will download the Lama 3 okay so Lama 3 is downloaded and it's ready so let's ask it a question uh tell me a jokes so you can see it's super fast okay so I'm going to close it now and uh for testing purpose I will also download the fe3 let's download it just copy this command and if you run this it will download the F3 okay so it's downloaded the fe3 so let's check K tell me sh and it is also super fast okay so let's close it and in previous version previous videos I show you the confi installation so let's run confi to run that I need to go to the directory so I installed it on the C directory and inside SD folder so let's go to that folder and here I need to run this command start.bat and I want listen so that uh I can use it locally the API will be enabled and auto launch so that it will launch automatically it started but it didn't Auto launch so let's launch it okay so here I have one model that is sdxl base model so let's start it let's generate a simple image here okay so that's that's a great picture I really like these kind of pictures so the next thing is we have Ama and inside AMA we have Lama 3 model and F3 model and we have confy y for image generation and inside that we have the sdx model so we have two different softwares now we need to uh find a way so that we can Bond them together for that we will use open we y so if you go to the open web UI you can go there and you can find the docks and everything so for that first I will go to the uh kab repository and I will open the W again from here so here is the open uh starting page so let's start with getting started here you can run this on a Docker machine but I will not use Docker so let's go and see if there there is any other option here you can see build and install Okay so uh you prefer not to run with uh Docker I want to run it locally so uh this is the way we can do this so let's open another uh command PRT and let's change our directory I in install every inside this SD directory so I will also install this on this directory open Wy okay so let's first copy this so it will clone the uh open w y That's fine so let's go open you inside this directory the second thing I need to create a cond environment for this okay I will use cond for that so let's use cond create name open YB this name and I want to install the python is equal to 3.10 so let's create this so so this command will create a virtual environment for python with the python version 3.10 and okay so we'll start the cond activate y you can see the W environment is started the next thing you can see we need nodejs so I will install nodejs inside this cond okay I made a mistake we need python 3.11 I install python 3.10 so let's deactivate this and create it again okay with the same command just 3.11 and I want to replace the existing one it's pretty simp simple so let's activate the cond environment from here now we need to install the nodejs so I will install nodejs inside the cond environment so let's search no js on and you can find the anakonda one just click here and you will get the command here so just copy this and paste it it will install the nodejs let's press yes here we need 20.10 at least so this is 20.12 so that's fine the this is done so let's go to the directory okay because this command is not uh working on Windows so for Windows I find it easier to go to the folder and just find the en. example and paste it copied it and just rename this delete everything after EnV and it will ask you to change it yes that's end so this is this part I did it manually okay the next thing is uh we need to run this command npn minus I so let's do this sorry so n PM minus sorry I okay the next command is npm run build so let's put it here oh no no I can't copy this I'm sorry somehow my copy and paste is not working okay so it's done now let's go to this directory and we need to install the requirements with this command okay so it's installed so let's run it so the Run command is bash. start. and it will not work with Windows so we need the windows version so let's check the directory here we have start windows. bat so let's start with this and I will put l some let's check it and I will also put use is in load let's let's check whether it bulls or not okay so it started now the problem is if I click here it will show me this error so just change it to the 00003 Local Host and it should work the first time you need to sign up okay so this looks almost almost like the chat GB and I will close these windows I don't need them from here you can see it should automatically detect AMA so if you go to the model it is giving you the options remember we downloaded the Lama 3 and fre3 here so they're here and we're running confi we need a way to uh you know connect confi with this so for that let's go to the settings and here you can see images from here you can also use automatic uh 11 1111 and also sdex with this but if you click here you can use open AI also but I'm using confi okay and I need to copy and paste this link and if I paste it here and just click here it is actually server is connected verified so let's on this and when I on this image generation it will give me the option to select the model I have only one model is DXL base and I want the uh Dimension 1,24 and 576 is the 16 by 9 ratio and for the set steps I want 20 50 is way higher okay so let's save it saving successfully close it and now we need to select a model we can select Lama 3 llama will uh create this so when you first select the model it will take a little bit time but after that it will not take any any extra time okay let's take it it's a nice one right so just let's uh create a create a prompt for stable diff sh you create an image for thisas C let's check it what it will say okay now it's uh actually describe the scene so here you can see the generate image option click here and it should generate an image and if I go to the St diffusion you can see it's generating you can see based on the description it generated a image so it's really amazing I really like this so now here I can I will show you another option uh if you click here in the plus sign it'll add another model you can add another model here so I will down load fe3 okay uh for the fe3 like uh let's create another scene for this let's see what it will says so you can see it is actually using the F3 uh latest model and again it will take a little bit time because uh it should it need to load the F3 model okay so it's actually generated huge one a big one and if you click here you can see the Lama 3 Lama 3 is also uh generate the response as well as fe3 so we get both at the same time so let's create one for the Lama 3 Let's see what it will generate okay I think it's good now if I go to the F3 response and here if I click it again let let's see what it will generate wow so both generated image and I think this one is much better than the Lama 3es one but you can generate as much as you want so that's the that's great really great so let's download the discover Canada PDF let's download a version and sorry uh we need to save it I and save it to the download folder and let's open it so from here I need to find a so about the citizenship test okay just remember this I will go there and here click on the plus file and if I go to the downloads and discover large it will now upload the Discover large file and here about the citizenship test tell me about the citizen she test and it will generate both response for f three and Lama 3 that's great right it will take a little time because it is actually now searching on the PDF and you can see you can see it's it's really working perfectly and here is the file it searches this is the Lama 3 response and the this is usually written test but it could be interview let's see see here is all the information and this is same so it is working perfectly right so that's um all for today so with this open UI open w we can connect both uh L language model and stab diffusion model together and we can generate images so let's check with the image generation let's see what type of image it generates for this wow so it is generating an image for this response and it's let's check for this one so it works actually great seamlessly together so that's all for today if you like this video please like And subscribe it will help me to generate more videos like this and have a great day thank you [Music]

Channel: Tech Jotters

Views: 1,677

Keywords: Ollama, ComfyUI, Open WebUI, LLAMA3, Phi3, LLM Models, PDF Search, AI Integration, Tech Tutorial, AI Setup, Language Models, Performance Optimization, AI Configuration, Tech Guide, AI Development, Unified Platform, PDF Analysis, AI PDF Search, Machine Learning, Advanced AI, Software Installation, Step-by-Step Guide, AI Performance, Ollama Integration, ComfyUI Setup, Open WebUI Configuration.

Id: g8C5ICIkiXs

Length: 19min 38sec (1178 seconds)

Published: Sat May 18 2024