Apple Releases Open-Source LLMs To Run On-Device

(RTTNews) - Apple (AAPL) has recently launched several open source large language models called OpenELM or Open-source Efficient Language Models to "empower and enrich the open research community, facilitating future research efforts".

OpenELM, released on the AI code sharing community Hugging Face Hub, can run on-device instead of cloud servers.

"Apple has been developing a large language model — the algorithm that underpins generative AI features — and all indications suggest that it will be entirely on-device. That means the technology is powered by the processor inside the iPhone, rather than in the cloud," Bloomberg reports.

The tech giant stated that there are four OpenELM models in total - four per-trained and four instruction-tuned, in sizes - 270 million, 450 million, 1.1 billion and 3 billion parameters.

Notably, all models use a layer-wise scaling strategy to efficiently allocate parameters within each layer of the transformer model, aiming to improve efficiency and accuracy.

"Diverging from prior practices that only provide model weights and inference code, and pre-train on private datasets, our release includes the complete framework for training and evaluation of the language model on publicly available datasets, including training logs, multiple checkpoints, and pre-training configurations", Apple said.

Moreover, Apple is expected to launch iOS 18 with AI capabilities during the Worldwide Developers Conference.