A Performance Evaluation of a Quantized Large Language Model on Various Smartphones
It addresses the problem of enabling private and efficient on-device AI for smartphone users, but is incremental as it applies existing quantization methods to new hardware.
This paper evaluated the performance of a quantized large language model on various Apple iPhone models, measuring thermal effects and interaction speeds to assess on-device inference feasibility, with results showing specific performance metrics across different smartphone generations.
This paper explores the feasibility and performance of on-device large language model (LLM) inference on various Apple iPhone models. Amidst the rapid evolution of generative AI, on-device LLMs offer solutions to privacy, security, and connectivity challenges inherent in cloud-based models. Leveraging existing literature on running multi-billion parameter LLMs on resource-limited devices, our study examines the thermal effects and interaction speeds of a high-performing LLM across different smartphone generations. We present real-world performance results, providing insights into on-device inference capabilities.