Get started with AI for free using Phi-3 [Pt 6]

Key Takeaways

hey folks in this session we'll experiment with AI models using the oama tool and the five free model directly from a browser for that we've built an online playground that you can use for free on any machine that can run a web browser so you don't even need a fast GPU or CPU and this is meant to be interactive so go ahead and open that link that you'll also find in the video description so you can play along so if you open this link it will open this GitHub repository from there select the Cod space button in there and it will ask to create a Cod space a c space is an online development environment that you can use directly from your browser you need a GitHub account for that so if you don't have one you can create one for free next select create Cod space and then it will take a few minutes uh the first time to create the environment so you can just go grab a cup of coffee and wait till it's ready now we have something familiar it's the very same vs code editor that you know but right into your browser so first let's use the terminal to check that everything works so type in oyama Run f free so by running this command I can chat with the model it's like a minimal chat GPT right in my terminal I'm hearing my cat mirring in the background so for inspiration let's say create an IU about a Hungry Cat and there I get the response from the model so everything looks good in there so can play a bit with the model and once you're done you can use contrl D to close the chat next let's open this oama notebook in there this is a Jupiter notebook and it allows to mix markdown text along with code to create interactive experiences first we Define the model in there if you want to play with different model you can do it by changing the constant in there just make sure that you don't load the model with oama first now for every code block that you see you can modify the code if you want and you can get this button on the left that you can use to run the code make sure to run it at least once every time you see one to make sure you set the context right for the notebook next we set up the open AI client so that we can make calls to the model o provides a compatible API so we just have to set the bu URL in there and put an API key and F don't need it let's run it and now we can play with the SDK just notice uh this npm prefix here in the import it's because we're using Doo as our runtime to run typescript in the notebook here it wouldn't be needed uh if you're using njs so now that we have set up the open client we can use it to gener a chat completion so again uh we use the SDK we specify the model name and send it our prompt let's run it and here we get the result now it's your turn to play with the notebook we have more example here with prompt engering few shot learning and even retrival augmented generation so once you've completed the notebook we also have a few samples in this folder in there you can open any of them like for example this classic if you one and then you can run directly the example using the Run button at the top so here what I'm trying to do is extract the sentiment of a text using Json format and here we get the result so this is mostly positive not too much aggressive and in French now it's your turn to play with J Ai and the best thing is that you don't even have to set up anything stay tuned for the next video we have plenty more content coming for you to learn more about how you can use AI more effectively and deploy your apps when it's time to shine