Apple has made a significant stride in on-device artificial intelligence with OpenELM (Open-source Efficient Language Models). This suite of open-source, large language models is designed to run directly on your iPhone, iPad, or Mac, bringing powerful AI capabilities closer to you. This approach prioritizes user privacy by keeping your data on your device, rather than relying on remote servers.
OpenELM is a family of eight, compact language models specifically optimized for efficient operation on your Apple devices. Ranging from 270 million to 3 billion parameters in size, these models are trained on publicly available datasets. You can access them through the Hugging Face Hub, a popular platform for AI developers to share and collaborate on code.
What sets OpenELM apart is its use of a layer-wise scaling strategy. This translates to more efficient allocation of resources within the model, resulting in higher accuracy. Apple has released both pre-trained models and models further tuned with specific instructions, demonstrating their commitment to open research and collaboration in the field of AI. As outlined by Apple:
OpenELM, a state-of-the-art open language model. OpenELM uses a layer-wise scaling strategy to efficiently allocate parameters within each layer of the transformer model, leading to enhanced accuracy. For example, with a parameter budget of approximately one billion parameters, OpenELM exhibits a 2.36% improvement in accuracy compared to OLMo while requiring 2x fewer pre-training tokens.
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.
By making OpenELM open-source, Apple invites the broader research community to explore and improve upon these models. This fosters transparency, allowing researchers to identify and address potential biases in the data and model design. In addition to this, developers and businesses can adapt OpenELM for specific applications, accelerating innovation.
While OpenELM isn’t integrated into Apple devices yet, rumors suggest iOS 18 will unveil exciting new AI features. The expectation is that Apple will leverage OpenELM to enhance user experiences on your devices while keeping your privacy a top priority.
Overall, Apple’s introduction of OpenELM marks a significant step forward for on-device AI processing. This approach prioritizes user privacy and fosters collaboration within the AI research community. With OpenELM’s potential integration into iOS 18, we can expect more powerful and secure AI experiences on our Apple devices in the near future.
