The release of OpenELM is intended to empower the open research community and allow for the investigation of risks and data and model biases. Developers and companies can use the models as they are or modify them. Apple has not yet brought these AI capabilities to its devices, but it is expected that iOS 18 will include new AI features and rumors suggest that Apple plans to run its large language models on-device for privacy purposes. The open sharing of information is also a recruitment tool for Apple to attract top engineers, scientists, and experts.
Key takeaways:
- Apple has released several open-source large language models (LLMs) called OpenELM, designed to run on-device rather than through cloud servers. These models are available on the Hugging Face Hub.
- The OpenELM models use a layer-wise scaling strategy to improve accuracy and efficiency. They have shown a 2.36% improvement in accuracy compared to OLMo while requiring 2x fewer pre-training tokens.
- Apple is releasing these models to empower the open research community and allow for investigation into risks, data, and model biases. Developers and companies can use these models as they are or modify them.
- Apple is expected to bring these AI capabilities to its devices with iOS 18, planning to run its large language models on-device for privacy purposes.