4 ms·
So not just versioned (which cloud software often is) but running locally? That's a pretty poor fit for a large language model, since users won't have sufficie
by jefftk 4y ago
So not just versioned (which cloud software often is) but running locally? That's a pretty poor fit for a large language model, since users won't have sufficient hardware and use is very bursty.
- mejutoco 4y agoThey might be able to copy the model weights, if running locally.
- kridsdale1 4y agoNot yet. In a few years, consumer GPUs or dedicated ML units (which Apple SOCs have) and local storage and ram will be large enough to run “a few years old” LLMs. The cutting edge will always be in the cloud but a market for personal AI does exist.