In this informal live‑demo tutorial, the presenter walks through downloading and installing Meta's Llama 3.1 (8‑billion‑parameter) model using the Ollama platform. After directing viewers to the Ollama download page and selecting the 8B model, he copies the generated terminal command, runs it, and shows the model responding to simple chat prompts. He then asks the model to generate a complete Python Pygame implementation of the classic Pong game, copies the code into an IDE, creates a virtual environment, installs Pygame, and executes the program to demonstrate that the generated game runs (albeit with some control quirks). Throughout, he comments on the model’s fast response speed and higher perceived accuracy compared to his previous local model, Mistral, and invites viewers to share their own experiences. The video ends with a light‑hearted note about his gaming skills and a sign‑off.
Chapters
Okay, I probably just suck at video games. And I'm doing this live on the internet where it's going to stay. You know, my kids are going to watch this when they grow up. Meta recently released Llama 3.1. And in this video, I'm just going to show you how to run it using Ollama. And Ollama allows you to run open source models locally on your own system. And if you don't know, or if you don't have Ollama already, just come here to ollama.com/download. And you can download for Mac, Linux, or Windows, depending on what system you're using. And after you've downloaded it and make sure that it's set up correctly, just come to models and you'll see Llama 3.1 here as the model. I'm choosing to use 8 billion model parameter, which is 4.7 gigabytes there because I don't have a lot of memory. And once you click on it, you will see here a kind of pre-made prompt for you that you use to download that model on your system. So once you copy that, go ahead to your terminal and just paste that command there and you will see it running. So let's just delete some portion of this that is not required. So it's ollama run 3.1, 8 billion parameters. So for me, I've already run this, which will pull the model. But if you haven't, you'll see it pulling the manifest and doing all the extra stuff to make sure the system is set to your system. So for me, as you can see, I've already set. For you, once it does pull, it will show you this message, you know, send message. And maybe we can ask it a very simple question. Uh, let's just say what's up, you know. Let's see what— okay, that's a unique greeting. How can I assist you today? Let's just say what's up again. You're sticking with the same theme, huh? What's up? It's like a British slang phrase that means what's up. Indeed. How's your day going so far? Okay, cool. Uh, now let's ask it to generate maybe a game, right? Because that's always a good thing. So, you know, generate a robust Pong game, because I see everybody doing Snake games. I mean, let me do a generator for Pong game in Python. And actually, I'm going to give it a shot and run this live for you. So once you generate it, I want to give it a try and see how it— if it's going to play or if it's going to work in the first run, right? And so I think that you can see it from what it's doing, it's just importing Pygame and then We have the system of course there, and that is setting the variables there. Okay, so seems like it's doing pretty good. One thing also while this is going, uh, just a note, for me it seems like it's doing this really fast. What you can see on the screen right now, it's real time. I'm not speeding up anything, it's just its actual speed. So it seems like it's doing very well and it, it outputs very fast. So I've been using Mistral a lot because I feel like it's been, it's been really fast on my system. But I think I'm going to replace it with this because this one seems much faster and more accurate actually. So I don't know what your experience has been with using, uh, this model on your local system, but yeah, feel free to let me know. So it's finished. It's just giving here the features of the game. I don't need all that right now, so I just want to see if it works. So let's just copy this code right here. I'm just going to move it all the way up to the beginning. I'm not going to go through what the code actually does, but I'm just going to check to see if this code works. So spin up our— we have our IDE set up here. I've already imported, uh, I've created a virtual environment and I've imported Pygame. So pip install pygame in that case. So let's go up here and make sure that there's no errors. So this is just the code itself. I haven't modified anything. I just copied and pasted in here. It doesn't seem to be throwing any errors. At least it's not showing any errors in the screen right there. So that's good. But, you know, game time now. Let's see if it's gonna play the game. So run the file, give it a moment. Oh, look at that! Uh-oh, it actually plays! So it has 2 paddles, as you can see. All right, all right. Oh, please— oh, it didn't. Come on, I can't, I can't play with the other side. You can tell I haven't been playing enough video games in my life to be able to control this. Ah, I see, I see. Now I see. I can control using The WSD, I believe. Oh, I'm too slow, but we're just gonna give this more try and see. All right. Oh well, the other paddle disappeared. Okay, I probably just suck at video games, and I'm doing this live on the internet where it's gonna stay, you know, my kids are gonna watch this when they grow up. But, uh, okay. So as you can see, it did pretty well. And I'm probably just gonna keep playing this after this. Um, but give it a shot. Let me know what you think, uh, about this model. Uh, and, uh, just wanted to show you how to do that real quick. So see you in the next video. Goodbye.
