McCreary County Record
In the world of artificial intelligence, small language models are making a big impact. These tiny AI models are gaining popularity for their ability to run on local devices without the need for cloud computing. One major player in this trend is Apple, who recently introduced OpenELM, a set of these small language models that can run directly on smartphones.
Available on the platform Hugging Face under an Apple Sample Code License, OpenELM features models with a range of 270 million to 3 billion parameters. Despite their smaller size, these models offer significant performance improvements compared to larger models on the market. Apple’s approach with OpenELM includes a layer-wise scaling strategy to enhance model performance while using fewer tokens.
In addition to OpenELM, Apple released CoreNet, a library used to train the models, along with reproducible training recipes for replicating model weights. This move aligns with Apple’s commitment to transparency, as the company shares source code, model weights, and training materials to empower the open research community.
While not yet integrated into consumer devices, Apple’s upcoming iOS 18 update may include new AI features that leverage on-device processing for user privacy. In the future, Apple may even collaborate with Google or OpenAI for more complex, off-device AI processing to enhance Siri’s capabilities.
Overall, Apple’s foray into small language models with OpenELM signifies a new era in AI technology, where on-device processing and user privacy take center stage. Stay tuned for updates on how Apple’s innovative approach will shape the future of AI on your smartphone.
“Social media scholar. Reader. Zombieaholic. Hardcore music maven. Web fanatic. Coffee practitioner. Explorer.”