PrismML Develops Tiny Language Model for Qualcomm's Snapdragon-Powered Smart Glasses

PrismML, an AI Lab founded by Caltech researchers and advised by UC Berkeley’s Ion Stoica, has developed a compact version of its language models designed for smart glasses running on Qualcomm’s Snapdragon chips. On Wednesday, at Qualcomm’s Snapdragon Summit, the company introduced PrismML’s 1-bit Bonsai LLM, capable of being operated locally on AI smart glasses using the Snapdragon AR1 Gen 1 Platform.

• As reported by TechCrunch, PrismML’s technology is notable for compressing larger language models significantly (by a factor of 4) without losing much accuracy in performance benchmarks.
• The specific model for smart glasses comes with a 2-billion-parameter design focused on vision and language, enabling users to interpret what they're seeing in real time.
• PrismML’s broader objective is to establish open-weight AI that operates on local devices, optimizing existing computational capabilities and reducing reliance on private AI ventures.
• Releasing this model for Qualcomm is an incremental step toward their mission, though no commercial product announcements for smart glasses utilizing PrismML have been made as yet.