Tootfinder

Opt-in global Mastodon full text search. Join the index!

No exact results. Similar results found.
@annsev@troet.cafe
2026-03-13 13:16:43

RE: #Söder
Er hat sich seither selbst in einem ungeheueren Ausmaß korrumpiert.

@aredridel@kolektiva.social
2026-04-14 14:22:42

So to follow up on this, I've caught it in action. Models, when quantized a bit, just do a bit more poorly with short contexts. Even going from f32 (as trained) to bf16 (as usually run) to q8 tends to do okay for "normal" context windows. And q4 you start feeling like "this model is a little stupid and gets stuck sometimes” (it is! It's just that it's still mostly careening about in the space of "plausible" most of the time. Not good guesswork, but still in the zone). With long contexts, the probability of parameters collapsing to zero are higher, so the more context the more likelihood you are to see brokenness.
And then at Q2 (2 bits per parameter) or Q1, the model falls apart completely. Parameters collapse to zero easily. You start seeing "all work and no play makes jack a dull boy” sorts of behavior, with intense and unscrutinized repetition, followed by a hard stop when it just stops working.
And quantization is a parameter that a model vendor can turn relatively easily. (they have to regenerate the model from the base with more quantization, but it's a data transformation on the order of running a terabyte through a straightforward and fast process, not like training).
If you have 1000 customers and enough equipment to handle the requests of 700, going from bf16 to q8 is a no-brainer. Suddenly you can handle the load and have a little spare capacity. They get worse results, probably pay the same per token (or they're on a subscription that hides the cost anyway so you are even freer to make trade-offs. There's a reason that subscription products are kinda poorly described.)
It's also possible for them to vary this across a day: use models during quieter periods? Maybe you get an instance running a bf16 quantization. If you use it during a high use period? You get a Q4 model.
Or intelligent routing is possible. No idea if anyone is doing this, but if they monitor what you send a bit, and you generally shoot for an expensive model for simple requests? They could totally substitute a highly quantized version of the model to answer the question.
There are •so many tricks• that can be pulled here. Some of them very reasonable to make, some of them treading into outright misleading or fraudulent, and it's weirdly hard to draw the line between them.

@NFL@darktundra.xyz
2026-05-07 18:20:45

Colts release CB Kenny Moore II after failing to find trade nfl.com/news/colts-release-cb-

@grahams@mastodon.social
2026-05-04 23:08:27

Matrix Music. Tonight. 8pm EDT. UncertainFM uncertain.fm/

Falling Green Matrix Text.

Green text overlaid promoting a radio show, details in the main post.
@metacurity@infosec.exchange
2026-03-06 16:57:35

Mullin’s appointment to lead DHS raises questions about future of CISA
nextgov.com/people/2026/03/mul

@cowboys@darktundra.xyz
2026-02-28 16:54:36

Cowboys receive troubling warning following George Pickens' decision sportingnews.com/us/nfl/dallas

@Techmeme@techhub.social
2026-03-28 01:20:52

Worth, which aims to help financial services onboard and underwrite SMBs, raised a $30M Series A led by Fulcrum Equity Partners, following a $25M seed round (Brian Contreras/Inc)
inc.com/brian-contreras/fintec

@karlauerbach@sfba.social
2026-03-16 22:30:01

Now here is something designed to shake confidence:
Flying, unmanned warships controlled by an on-board, but distributed, AI...
"Airbus is Preparing 2 Uncrewed Combat Aircraft from Kratos for 1st Flight With a European Mission System'"

The price of silver smashed through the $90 mark Wednesday morning
for the first time in weeks
as the metal recoups gains it lost after a late January crash,
though the precious metal remains far from the record highs it previously hit above $120.
KEY FACTS
The price of silver is $90.68 as of 9:25 a.m. EST, up nearly 4% but down slightly from a high of $91.18 reached earlier in the morning.

@ascendor@social.tchncs.de
2026-04-26 11:58:50

RE: troet.cafe/@Fellingshaeuser/11
Höchststrafe für die Grünen-Hater. So kacke es ist, dass die C*U unsere Demokratie und unsere Welt zerstört für ihren Profit - ich kann (muss) über diese Schlagzeile schmunzeln.