I've been exploring AI for quite a few months now and you have seen that with you know my content about Lang chain or the free code cam Cod I did uh for Lang chain with open eyes apis and today I wanted to show you how you can run large language models locally and that's where AMA comes in they just recently released the windows version it was available on Mac and Linux but now they have released Windows version which is in preview by the way so we'll be check checking out AMA and how you can integrate that with Lang chain to run your models locally but also build those llm powered apps so if you go to ama. com you can see that it's available for Mac OS Linux and windows now which is awesome so if click on download and click on download for Windows or any of the other operating systems which you might be on it'll download the exe file and now let's go through the installation steps here so it looks like it wants me to close obs yes so I'll install amaama and then continue the recording so the installation was successful and now if I go to my terminal type in ama and you can see that we have few different options and available commands within AMA so you have serve create Show run pull push list copy remove and help and I'll go over the important ones that'll that you'll need to get started with running a model locally on your machine but before that let me do a quick explanation on how AMA works so the single binary file that we downloaded and installed on our machine comes with the server and the client and the client in our case is the CLI so later when I'll run AMA run fi so fi or fee is a model by Microsoft and you can look it up I kind of posted about it on my Twitter you know I wanted to check this model out so when you run this this command this basically does an API call to the server so our CLI acts as a client and does an API request to the server or you could build your own app that utilizes this API so you'll be doing rest API calls to this AMA server that's a little bit about how AMA works and now if you go into models on ama. com you can look at different models that are available for you to pull so just like you to get pull to pull a repository you'll do AMA pull F to run the F or fee model locally so let's go ahead and do that so I'll pull the Manifest and then you'll see that it's 1.
6 GB the model size and I'll start downloading it okay so now we have successfully pulled a model from Ama so now you need to run that model so if I do run fee and it should start the model right and now we are basically making the API call to the AMA server that's running on our machine so let's say I don't know why our skies blue and there you go it generates a response saying the color of the sky is determined by SC ing of sunlight in the Earth's atmosphere and then you can kind of read through this so now we have a large language model well in Fe case it's called small language model because of the amount of the parameters that it has I believe it's 1. 2 or 1. 7 uh billion or 2.
