eicker 74 points a month ago

We are repeating an old pattern in computing: throw more hardware at the problem until efficiency becomes impossible to ignore. Bigger models have delivered remarkable gains, but they’re increasingly expensive. The next breakthroughs may come less from adding parameters and more from smarter architectures, better algorithms and more efficient inference.

path: 0 24785681, hotness: undefined, score: 74, children: 8

Vlyn 47 points a month ago

DeepSeek has really led the way here, especially as they are a bit more hardware constrained. Plus they openly publish their findings and release open source models, so high hopes there.

It's probably China's play to pop the AI bubble, but I'm all for it (:

save

path: 0 24785681 24786608, hotness: undefined, score: 47, children: 5

ScoffingLizard -4 points a month ago

I wonder what all is in the deepseek code that is malicious. I'd like to try it but don't want a million Mb/s of tracker shit across my network and can't run it myself.

save

path: 0 24785681 24786608 24797417, hotness: undefined, score: -4, children: 4

Rothe 20 points a month ago

Except there likely won't be a lot of further breakthroughs if we burn down our planet faster than we already do.

This is all an expenditure of vast amounts of energy for literally no gain for anybody except a handful of billionaires and their corporations.

save

path: 0 24785681 24792133, hotness: undefined, score: 20, children: 0

JustDorky 3 points a month ago

That's literally exactly what Chinese researchers are doing at DeepSeek and they've built frontier models with that philosophy

save

path: 0 24785681 24799490, hotness: undefined, score: 3, children: 0

Our Rules