Defining open-source models

0:09 hi I'm Corey cigared pace and I'm part of the AI Cloud aacy team here at Microsoft and I have the pleasure of presenting to you lesson number 16 of our generative AI for beginners course covering working with open source models in this lesson we're going to cover what exactly are open source models some of the benefits of why would want to use open source models and also understanding where we can find the different open source models that are available to build our applications we also have some learning goals here we want to understand the difference between these open source models when helping us determine which ones to use how to actually use these in an application and where to find more detail about these open source models that are available so what are exactly open source models standard definition that we sort of taking from open source soft software would put this category of that all this information when we were talking about large language models whether it's the training data being publicly available the full model weights that were you for training the evaluation code that was also used to evaluate the training as well as fine-tuning the model the training metrics that we determine during that

1:21 process and then last but most importantly is also an open source license in terms of being able to use this uh as a developer uh freely without any sort of restrictions this is the sort of criteria that uh an open source software would have if you translate that over to open- Source models if we look at the available models that are out there now there's actually only a few models that sort of fit all of these criterias perfectly one series of models that are it's worth looking at is uh the olmo models from Allen AI Olo or is short for actually open language models H and these models are like I mentioned a series of models the most