Apple presents eight elegant language paradigms for AI.

Known as OpenELM (Open-source Efficient Language Models), Apple has published a collection of eight compact AI language models designed for on-device usage. These versions fit immediately onto a smartphone due to their compact size.

The Features and Specifications of OpenELM
There are two varieties of OpenELM models: four that are "trained" and four that are instruction-tuned. The models have a maximum context window
of 2048 tokens and were trained on publically accessible datasets. The models are far smaller than previous models like OpenAI's GPT-3 and Meta's Llama 3 series, with parameter counts ranging from 270 million to 3 billion.

Models for OpenELM
These are the eight OpenELM models:

OpenELM-270M

OpenELM-450M

OpenELM-1_1B

OpenELM-3B

OpenELM-270M-Guide

OpenELM-450M-Guide

OpeningELM-1_1B-Detail

OpenELM-3B-Guide

The Potential Uses of OpenELM
Even while these models are currently mostly proof-of-concept research models, Apple may use them as the foundation for future on-device AI products.


A Comparison of OpenELM with Other Models
This release is in line with Microsoft's ambitions to create helpful, small-scale, locally executable AI language models. Microsoft Phi-3 models, for example, are designed to reach a comparable degree of language comprehension and processing capability.