AI-powered
podcast player
Listen to all your favourite podcasts with AI-powered features
The Challenges of Streaming Data
One of the things that I've noticed about when people are talking about these applications is that it's usually working on streaming data. So you really shrink the solution space down for the small models from what the large model is and you can do it with a lot less resources than I guess. Yes, that latency thing that I think Andy alluded to. A lot of industrial industry problems are around streaming data. It's classic sensor data. You've got some machine that's vibrating and you want to know does that vibrating mean that you're about to fail in the next day in which case I should tighten your bolts now before you fail. No one wants to wait 50 milliseconds or something.