Apple’s AI push has been slow to say the least, especially when compared to the rapid developments underway at its competitors, namely Microsoft and Google. While Samsung, Google, and even Nothing have a plethora of AI features on their respective devices, iPhones have been left behind as Apple catches up in the AI race. However, it is actively trying to make progress and has recently been in talks with Google and OpenAI about a possible deal that would allow their AI models to be used on iPhones, but this is still in development.
Today, Apple researchers released a family of four lightweight AI models in the Hugging Face Model Library that can run on the device, hinting at their future use on devices such as the iPhone, the iPad and the Mac.
Apple launches four open source AI models
According to the article on Hugging Face, the family of AI models is known as “Open Source Efficient Language Models” or OpenELM. These templates were designed to efficiently complete small tasks, like writing emails. Apple says OpenELM is trained on publicly available datasets using the CoreNet library which includes RefinedWeb, Deduplicated PILE, a subset of RedPajama, and a subset of Dolma v1.6, totaling approximately 1 .8 trillion tokens. It was released with four settings: 70 million, 450 million, 1.1 billion, and 3 billion settings.
For those who don’t know, metrics measure the number of variables the AI model can learn from while making decisions. These are based on the dataset the AI model was trained on.
According to Apple, the OpenELM family of AI models was launched to “empower and enrich the open research community by providing access to cutting-edge language models.”
Apple’s AI push
The iPhone maker has been experimenting with AI for some time now. Last year, it released a machine learning framework called MLX that allows AI models to perform better on its Apple Silicon-powered devices. Additionally, it also launched an image tool called MLLM-Guided Image Editing or MGIE.
Last month, it was revealed that Apple researchers had achieved a breakthrough in training AI models on text and images. A research paper on the same topic was published on March 14. Titled “MM1: Methods, Analysis, and Insights from Multimodal LLM Pre-Training,” it demonstrates how using multiple architectures for training data and models can help achieve the state of the art. results on several benchmarks.
It is also said to be working on its own Large Language Model (LLM), at the heart of which is a new framework known as Ajax which could provide a ChatGPT-like application, nicknamed “Apple GPT”. Collaboration between different Apple departments. , such as software engineering, machine learning and cloud engineering, would be underway to make this LLM project a reality.
The release of the OpenELM family of AI models certainly paints an intriguing picture of AI development at Apple. However, since no base model has been released yet, it will still be a while before Apple devices such as the iPhone and Mac can finally take advantage of it.
