Finally The Most Awaited AI Tool Launched By Apple, Open ELM Apple has unveiled a new streamlined language model called OpenELM, representing the Open Efficient Language Model. During the introduction of this freshly developed language model, Apple stated, “We are proud to present Open ELM, a collection of open-source efficient language models.” They further explained that it employs a layer-wise invocation strategy to optimally distribute parameters within each layer of the transformer model, resulting in heightened security and improved efficiency.
As per Apple’s announcement, OpenELM represents a cutting-edge language model that optimally allocates parameters across the layers of the transformer model, thereby enhancing precision. It comprises eight models featuring four distinct parameter sizes: 270 million, 450 million, 1.1 billion, and 3 billion. Notably, all these models have undergone training using publicly accessible datasets.
According to sources, Apple is actively engaged in various AI-related projects, collaborating with researchers in the field to enhance OpenELM and develop additional tools of a similar nature. Apple recently released this model, coinciding with the impending launch of their newly developed software, iOS 18. This suggests that the forthcoming software update is likely to integrate advanced generative AI capabilities.
The release of OpenELM sheds light on Apple’s strategic efforts to integrate AI capabilities across its product lineup, encompassing iMacs, MacBooks, iPhones, and iPads. These newly introduced language models are now accessible via the Hugging Face Hub, a platform facilitating the sharing of AI codes within the community. According to a white paper issued by Apple, OpenELM consists of eight distinct models, with four being pre-trained using the CoreNet library and the remaining four being fine-tuned variations. Notably, this language model demonstrates a remarkable improvement in accuracy, boasting a 2.36% enhancement compared to existing models such as OLMo.
What sets it apart from other language models is its capability for on-device processing, eliminating the need for reliance on third-party services. Typically, AI processing involves transferring data to cloud servers for computation, followed by the transmission of responses back to the user. This approach significantly enhances user privacy and security. Additionally, it offers cost savings to companies by eliminating the need to maintain large servers for processing AI commands and tasks.
