Apple releases OpenELM language model, hints at the arrival of AI on iPhones

Apple is a technology company that many organizations are keeping an eye on in the direction of AI development. Although Apple still hasn’t revealed its development plans.Gen AI for commercial useSame as competitors But the release of a new set of AI models, like OpenELM, is showing that. The company’s goal is to focus on Running AI on devices

On April 24, 2024, researchers from Apple published OpenELM (Open-source Efficient Language Models) Large-scale AI language model (LLM) suiteopen source that can run on the device without connecting to a cloud server

OpenELM comes in four sizes: 270 million parameters, 450 million parameters, 1.1 billion parameters, and 3 billion parameters (parameters are the variables that the model learns and uses to make decisions from the trained dataset).

For example, model Phi-3 atMicrosoftThe recently released model has 3.8 billion parameters, compared to Google Gemma’s 2 billion parameter model. These models use less power. and is ideal for working on small mobile devices such as smartphones and laptops.

The research also says that OpenELM uses a method of allocating parameters to each model layer to increase accuracy. The results are The model performed 2.36% more accurate than OLMo, but used twice as many untrained tokens.

In summary, the strengths of OpenELM are that it is small, energy efficient, uses little CPU and RAM resources, and is suitable for use on mobile devices. Being open source makes it free for developers to use and customize. Run models on devices without connecting to a cloud server.

This model will help increase the potential of work such as writing emails, summarizing messages, writing articles, helping write emails, translating languages, analyzing data. or can be used for entertainment purposes You can also create creative works.

Reference: theverge venturebeat