Google DeepMind Engineer Runs Gemma 3 270M Model on an Office Printer
3 hour ago / Read about 0 minute
Author:小编   

A Google DeepMind engineer recently pulled off a clever experiment: running the Google Gemma 3 270M small language model directly inside an ordinary office laser printer. The model has roughly 270 million parameters, and the printer had only 14.1MB of available memory—yet it still managed a generation speed of over 60 tokens per second.