Andrew Ng introduces a new course on building AI applications with Hugging Face open source models. Learn to select models from the Hub, use the Transformers library and Pipeline API, combine models for multimodal tasks like image and speech recognition, and deploy apps via Hugging Face Spaces.
I'm excited to introduce open source models with Hugging Face built in partnership with Hugging Face Hugging Face tools have been transformative for AI builders and they help you quickly build many possible AI applications by letting you grab and deploy any of a huge variety of already trained open source models that you can then assemble into a new application. This course teaches you some of the best practices as well as what it feels like to quickly grab an image recognition model here, maybe grab an LLM there, maybe grab a speech recognition model there, and so on, and to use that to put together some creative application maybe in just tens of minutes. In this course you use the Hugging Face Transformers library and a number of open source models to process text, audio and images. For example, you see how to combine models to build something to help a person with a visual impairment by describing an image aloud. You could do this by applying a trained object detection model to identify the objects in the image and then you can apply a text to speech model to narrate a summary of what's in the image. All of the models you use in this course are open source, meaning that the models and their weights are openly available for for anyone to download. Hugging Face's work to make open source models more accessible has been a huge boost to the AI community. I'm delighted to introduce the instructors. Eunice Belkodar is a machine learning engineer at Hugging Face, Mark sun, also a machine learning engineer at Hugging Face and Maria Harlusova, a member of technical staff at Hugging Face.
Thanks Andrew. In this course you'll learn some tips on searching and selecting models from among the thousands of open source models on a Hugging Face hub. You'll interact with the models using Pipeline object from the HuggingFace Transformers library which handles much of the preprocessing of inputs and post processing of outputs so that you can get an AI enabled application working with a few lines of code.
You'll also wrap your AI applications such as an image narration assistant for the visually impaired inside a user friendly interface using the Gradle library. You'll also deploy an AI enabled image captioning service as an API using Hugging Face Spaces so that anyone can make an API call over the Internet and use your application.
You will build the components that can be combined into a voice assistant by bringing together automatic speech recognition and text to speech models.
You will also use open source model to perform many of the natural language tasks that the LLM can do, summarizing text translating to another language and chatting with users like a chatbot.
I hope this course will open you up to many sources of opportunities to build AI powered applications.
