Post

'A team of Stanford University computer scientists find that SLMs can find the same — or better — answer to the kind of queries that dominate current chat requests almost every time and cut compute and energy costs by between 60 and 80 per cent.' www.ft.com/content/4858...

11/9/26, 7:20 pm80 reposts314 likes

Susan Taylor

@sataylormd.bsky.social

11 Sept

Thank you!

Macungie Boro

@macungie.bsky.social

12 Sept

Yes, sure a local model can do a lot of stuff if given an unlimited amount of time. I've run similar local models on a local machine with a top of line CPU, with a 32GB modern GPU, and it doesn't even come close the speed of output as those Frontier models.

Ntrn Chck

@ntrnchck.bsky.social

11 Sept

Because most of what it can do with less errors is giving general information, suggestions, and general coding ... So I guess that's already a "solved" problem ... Let's leave it at that

paul-warner.bsky.social

@paul-warner.bsky.social

13 Sept

I strongly suspect Google has figured this out long ago. I expect caching really helps at their scale too

Shash of the Grove ⚧️ Hippie Chick

@bearofthegrove.bsky.social

12 Sept

OK, but the AI, for the people pushing it, is about raping the working classes, then disposing of the bodies. I mean, they've said it, just not so bluntly.

Rage Against the Maniac

@phillipse16.bsky.social

11 Sept

Context? Meaning?

Roaring 20s Redux

@mishoiliev.bsky.social

11 Sept

Context is that the AI data centre build up may be wasteful as SLMs can be run on a Raspberry Pi

Rage Against the Maniac

@phillipse16.bsky.social

11 Sept

Thank you!

gselvara.bsky.social

@gselvara.bsky.social

11 Sept

SLMs (Small Language Models) are language models with a much smaller number of parameters than large language models (LLMs) — typically ranging from a few million up to a few billion parameters, compared to LLMs which can have hundreds of billions.