But training LLM's is also a task one can do whenever you have a spare GPU-minutes.
replyI wonder why they don't have some kind of scheduler which makes sure there are never any idle minutes. One would imagine they at least would have autoscaling on their production serving workload and use the freed compute capacity for model training for example.
I doubt they're inferencing on their training hardware
replyIs the electricity cost far greater than the marketing value?
replyThe marginal electricity cost is zero.
replyOr specifically, electricity was already paid for with the pre-paid capacity.
replyNot using it would not save them any money, they already paid for it.
The first is a physical quantity that can be written down.
replyThe second is approximately no better than astrology.
The second point is, sadly, true of quite a lot of aspects of software, including "design" and "quality"
reply[dead]
reply