1 paper · 1 filter
Avyay Sadhu, Alvaro Velasquez, Lekai Chen
Running a language model on edge hardware provides private and low-latency reasoning without a network connection, and yet the small models that fit on such devices are unreliable…