Amit Sangani, Director of Partner Engineering at Meta, is the instructor of this course. It's great to have you here, Amit. Thank you, Andrew. It's great to be here. I'm excited to introduce you to all the exciting capabilities and use cases of LAMA collection of models. LAMA has been a game changer for AI developers because Meta has published the model ways online so that anyone can download, modify, play with, and work on applications using them. This is in contrast to the closed source models, which can also be very powerful and very useful, but that you can access only via API calls. Many teams in big and small companies, as well as just individuals building cool applications, have been using Lama as a result. Because LAMA weights are freely available, many people, including me, That's right, Andrew. The models are free to download so that everyone in the AI community can use them for building generative AI applications, modify them, and carry out additional training and therefore drive research and innovation. We are seeing millions of downloads and I'm grateful to all the developers using LAMA to build amazing things for other people. First, there's a set of base models. These are LMs that have been trained to repeatedly predict the next word based on say internet text but haven't received any additional training to modify their behavior. These base models are useful to developers who want to continue training models to perform well on specific tasks. These chat models are ideal for powering chatbots and for following your instructions to get questions answered or to get tasks done. The last set of models have received additional training specifically to make them good at understanding and writing computer code. While these models may seem like they're most useful for software engineers, they're also making it easier for many people that are newer to coding to write, debug, and learn code on their own. And you'll get to try out all of these different You'll start off by prompting a LAMA model to help you write a birthday card. In the process, you'll learn the details of how LLM inputs are actually formatted. For example, with start and end tokens for different parts of the input. You'll also prompt LAMA to help you classify the sentiment You'll also learn about prompt engineering, including an important technique called few-shot learning via in-context learning. Specifically, by giving the model one or two examples of how you would like it to respond to a prompt, you can get a model to give its response in a similar way. You'll also learn about chain of thought prompting and also how to use code-specific LAMA models. Many developers are using Code LAMA to help with their coding Lastly, you will also learn about LAMA guard, This is a key step for many businesses wanting to deploy LLNs. And with that, let's move on to the next video and get started.