Apple has released 20 machine learning models and four datasets on Hugging Face. The move allows developers to experiment with AI applications that run on a user device rather than in the cloud.
The added models on Hugging Face are of tremendous value because Apple is also releasing the Core machine learning (ML) framework. Core ML allows developers to integrate machine learning models into the applications they are developing. In turn, the app will combine the integrated models with user data to make accurate predictions.
AI on the device
Unique to Apple is that these predictions are made on the user’s device. Traditionally, AI tools send user data to the AI developer’s cloud and repositories. Apple, on the other hand, promised to put security first when it comes to AI with its own AI story: Apple Intelligence.
Clement Delangue, CEO of Hugging Face, praised this secure approach in a comment to VentureBeat: “Core ML models run solely on the user’s device and eliminate the need for a network connection. This keeps your app blazing fast and user data private.”
Commitment to Hugging Face
Hugging Face is an online platform where AI models and code can be shared with developers. Apple has already collaborated with the platform in recent months. For example, in April 2024, eight open-source LLMs were shared so that developers could discover and address risks in the models, among other things.
Tip! Apple and OpenAI make good money from AI, but not from each other