Apple releases new open-source language model to power on-device AI capabilities

apple ai

Apple has released OpenELM (Open-source Efficient Language Models), a family of open-source large language model (LLM). The Cupertino-based tech giant intends to run this directly on its devices instead of via cloud services.

The OpenELM AI models are available on Hugging Face page, a community for sharing AI codes.

Introducing the SLMs, Apple stated, “We introduce OpenELM, a family of Open-source Efficient Language Models. OpenELM uses a layer-wise scaling strategy to efficiently allocate parameters within each layer of the transformer model, leading to enhanced accuracy.”

OpenELM: All the details

According to Apple, OpenELM is a state-of-the-art language model that effectively distributes parameters around the transformer model's layers, improving accuracy. Eight models with four distinct parameter sizes 270M, 450M, 1.1B, and 3B are reportedly included in OpenELM. All of the models have been trained using publicly available datasets.

Apple also stands to gain strategically from OpenELM. With the information freely shared, Apple will be able to work with other researchers in the field, allowing others to improve and add to OpenELM. Furthermore, this open atmosphere will draw top talent for the organisation, such as scientists, engineers, and AI experts.

The release of OpenELM comes a few weeks before WWDC in June when Apple is expected to reveal iOS 18. Several new generative AI-based features are anticipated to be launched with iOS 18. OpenELM's release provides insight on the groundwork Apple is laying for AI integration in its devices, including iMacs, MacBooks, iPhones, and iPads.