PrismML adapts tiny language models for Qualcomm smart glasses
AI lab PrismML has developed a version of its compact language models for smart glasses using Qualcomm Snapdragon chips. Founded by Caltech researchers and advised by UC Berkeley’s Ion Stoica, the lab presented its 1-bit Bonsai LLM at Qualcomm’s Snapdragon Summit. The model can run locally on glasses based on the Snapdragon AR1 Gen 1 Platform.
The glasses-focused system has 2 billion parameters and is designed for vision and language tasks. This would allow users to ask questions about what they are seeing in real time. PrismML is known for reducing the size of larger models by four times while preserving almost all of their performance on standard benchmarks.
The lab’s broader aim is to develop open-weight AI that operates on devices and uses their existing computing capacity more efficiently. It presents this approach as an alternative to relying on proprietary AI providers and their demand for additional computing resources. Support for Qualcomm’s chip advances that goal, although no smart glasses running PrismML have been announced.
Written by Sakhanda Wire AI from reporting by TechCrunch. Facts, quotes and full context are in the original article below.