You are building an ML feature that runs locally on a mobile device, such as ranking, personalization, or language understanding. The model must fit device limits while keeping quality high across a wide range of hardware, OS versions, and usage patterns.
What considerations do you take into account for model optimization on-device?
You are building an ML feature that runs locally on a mobile device, such as ranking, personalization, or language understanding. The model must fit device limits while keeping quality high across a wide range of hardware, OS versions, and usage patterns.
What considerations do you take into account for model optimization on-device?