Apple: efficient device/server foundation-model adaptation

What changed

Integrated device and server language models with mixed-bit compression, task adapters and post-training procedures including teacher-committee rejection sampling and mirror-descent-based preference optimization.

Scientific scope

Integrated device and server language models with mixed-bit compression, task adapters and post-training procedures including teacher-committee rejection sampling and mirror-descent-based preference optimization. One original integrated training/inference system, not a new invention of LoRA, GQA or low-bit quantization. Device latency is hardware/prompt dependent; provider privacy principles and preference tests are not independent privacy or safety guarantees.

Evidence

The linked original sources describe the stated contribution and its evidence limits. WRS has not independently rerun the experiments or certified a proof. One original integrated training/inference system, not a new invention of LoRA, GQA or low-bit quantization. Device latency is hardware/prompt dependent; provider privacy principles and preference tests are not independent privacy or safety guarantees.

Contributors

• Source-listed authors and technical contributors · SCIENCE: Source-listed authors or explicitly credited technical contributors for the cited versions. Family entries retain source-specific attribution; a combined list is not one original-paper byline or equal individual work. Names alone do not establish resolved identities.

• Apple Intelligence Foundation Language Models
• Apple: efficient device/server foundation-model adaptation — author/company record

Provisional WRS assessment, edition 2026-10-16.1. Independent expert calibration is pending. This grade does not certify correctness or assign ownership.

Full scientific record · Browse contribution grades · Request a correction