Comment by sarah_eu

Well it will be multimodal, training and inferring on feeds of distributed sensing networks; radio, optical, acoustic, accelerometer, vibration, anything that's in your phone and much besides. I think the time of the text-only transformer has already passed.