Once when I was 14, my little brother and I were lying in beds next to each other, and I tried to walk him through a thought experiment about how social context influences aesthetics.
In another world where Jesus was nailed to a giant plastic My Little Pony figurine, would you feel the same weight and reverence when walking into an old Catholic church and seeing their equestrian "crucifix"? Would everything we've all been immersed in since birth to create that feeling ab…
Having little kids: *so tired, all the time*
Having big kids: “It’s ok mom, you can go back to bed! (Since you’re not feeling well this morning)”
I love having big kids 🥰
Twenty-five years and my life is still
Trying to get up that great big hill of hope
For a destination
And I realized quickly when I knew I should
That the world was made up of this brotherhood of man
For whatever that means
And so I cry sometimes
When I'm lying in bed
Just to get it all out
What's in my head
And I am feeling a little peculiar
And so I wake in the morning
And I step outside
And I take a de…
Time course of visual plasticity following adult-onset #deafness (in cats) https://www.nature.com/articles/s41598-026-39490-8
User A wrote a nifty little tool in Python. On Windows. Several weeks of work.
User B wants to use it as well.
They copy it, try to run it. Doesn't work.
Call me for help.
A: "It works on my machine. B must configure his python exactly like I do!"
Me: "Can't we run it in a virtual environment (venv). Or with a docker image?"
A: "Hm, I would have to learn how to do that. Is this really necessary?"
Me:
I have been feeling in need of a little magic these days. Have you? There are wondrous pavilions everywhere, but we do not always see them.
https://salrandolph.substack.com/p/from-the-yellow-crane-pavilion
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.
The little owl, Athena noctua, is well-known as the companion of the goddess Athena, though it is uncertain how she became associated with the nocturnal birds. In Europe, the owl therefore represents wisdom and is the symbol of many a university.
The ancient Greeks showed #Athena with her favourite bird in their art, like for instance this bronze statuette of Athena flying her owl, dated ca.…
The entire machinery of online discourse around building and creating has been so thoroughly captured by entrepreneurial "logic"
that we've lost the language to describe what it feels like to simply make a thing that helps someone,
give it away, and move on with your life.
I've been feeling this for a while now, and I suspect a lot of folks who have the itch to build feel it too, even if they haven't articulated it.
I'm saddened but not surprised. Why are our politicians - from #Newsom to mayors to supervisors - so vocal about mandatory treatment and the need for conservancy or jail, and so unwilling to ensure the necessary services are available?
California’s new ‘war on drugs’: thousands arrested, few get treatment, data shows