It is fun for you to play with Stable Diffusion or Open AI's GPT4 but the revolution is in making applications on the Web for anyone to use.

This session is a techy overview of the landscape of all the tools you can use to get from something on a web page into a machine learning model and back. This is hard because your ordinary web hosting service can't really run these models and often you need a GPU on the server.

First we will look at connecting from say p5 to services with an API like OpenAI or Replicate (and look at doing it with a proxy server to hide your API key for the API). If you need a little more control than the API offers, we will look at prototyping in a notebook like Google Colab and connecting that to the web with Ngrok or using Huggingface spaces with quick and dirty web interfaces and APIs. Finally we will look at "deploying" your ML work with a dedicated (expensive) GPU with something like Huggingface Endpoints or TensorDock.