Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
dofm
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
13 ms
·
241.
▲
by
dofm
1mo ago
Well isn't that going to be an interesting test of corporate ideology ;-)
242.
▲
by
dofm
1mo ago
> If you can take any book and turn it into a model, because it's "transformative enough", and "AI learns just like a person does", then surely a model distilling another model is transformative and fair use. I m
243.
▲
by
dofm
2mo ago
The Axios article? Probably not slop, at least not quite in the way you might think. This is how Axios writes — it’s their house style for these forms of articles — and have done for several years before LLMs. This style (Axios call it “Sma
244.
▲
by
dofm
2mo ago
The Open Community Licence is a horrible mistake, even with the recent modifications.
245.
▲
by
dofm
2mo ago
I mean this more generally, beyond that team. Oracle have used their ownership of Java as a thumping great cudgel before, and my own feeling is that the Google lawsuit basically ended Java’s appeal as a language on which others would risk a
246.
▲
by
dofm
2mo ago
Java could be in better hands. Maybe will be, one day, when Oracle finally ends up fighting the organisational fire that has started in their debt pile.
247.
▲
by
dofm
2mo ago
Interesting. I come at this from the other direction. I was your stereotypical guitar-playing, photographer, software nerd; I would never have tried making physical things until I realised I wanted to make photography accessories. (Also som
248.
▲
by
dofm
2mo ago
But it's a machine tool and you should be able to modify and improve it . Bambu don't want you to have that flexibility. And when they do break in unusual ways (and they still do), their customer service is not good enough to m
249.
▲
by
dofm
2mo ago
Prusa make machine tools , IMO. (I don't understand people who build print farms from Bambu Lab machines at all; I do get using Prusas.)
250.
▲
by
dofm
2mo ago
Depends on the niche and application you are interested in. Though I would caveat my assertion by saying that if the specific form factor of the A1 Mini (which is a nice thing) matters to you, and it well might, there’s probably no altern
251.
▲
by
dofm
2mo ago
Bambu and Prusa are both engaged in harm. But Prusa is only engaged in self -harm.
252.
▲
by
dofm
2mo ago
At this point there are so many alternatives that we should fully be into the “friends don’t let friends buy Bambu Labs products” but alas we are not. It’s absolutely insane how little they would have to do to be good community members and
253.
▲
by
dofm
2mo ago
Yes. xhigh can not just overdo the answer, it can also trip itself up and end up writing worse code. Even in the lower reasoning levels I find I want to like Qwen 3.8 27B and mostly don’t; it’s OK in the low reasoning effort, though. Muse
254.
▲
by
dofm
2mo ago
This actually is meaningful news, I think. Pretty wide audience appeal in the local LLM space too.
255.
▲
by
dofm
2mo ago
This is exhausting and would have been a lot easier to read in the more, how to put it, natural writing style the author had until early 2025. What changed, I wonder? ;-/
256.
▲
by
dofm
2mo ago
The issue is not RAM size. It's memory bandwidth! The 12B and 31B models will be useless on a Pi 5; maybe the 12B can be persuaded to run, but it may not manage more than one token per second. It only manages 17GB/s memory bandwid
257.
▲
by
dofm
2mo ago
Cool — but is that model really the right choice for the task? I guess it is only 3B active which helps a lot but is Gemma 4 E4B not more practical?
258.
▲
by
dofm
2mo ago
I’m using the Unsloth 4-bit quant. To change the reasoning strength you just put text in the system prompt. From memory it is: Reasoning strength: low
259.
▲
by
dofm
2mo ago
Have you tested Muse Glimmer in low reasoning strength? Token generation is slow (and prefill is) but you will likely find it solves actual problems faster than Qwen 3.6 35B-A3B.
260.
▲
by
dofm
2mo ago
Right but the tenor of it is different. In the 90s if you went more than a fortnight of job search after you graduated and didn’t get to being offered an interview, either you only had a third class degree or there was something very wrong
261.
▲
by
dofm
2mo ago
You would in the UK but we are a condiments culture. Quite normal to find two kinds of tomato relish, two kinds of mustard, and if the café is satanic, salad cream or piccalilli. But I don’t think I would dream of asking for salt or pepper
262.
▲
by
dofm
2mo ago
> In Europe you'll wait 20 minutes to get the attention of a waiter for salt or ketchup until your food is cold. You could ask for the salt or ketchup when the meal arrives. Also Europeans are not, I think, accustomed to asking for
263.
▲
by
dofm
2mo ago
Right but what percentage of people doing any given class end up specialising in it? It's not their job to "inject" it but universities are going to have to grapple with the notion that if they want to keep admitting students
264.
▲
by
dofm
2mo ago
There's also the Open University in the UK — an extraordinary institution. AFAIK it's also accredited in the USA. (I have been thinking about an OU degree for a second act)
265.
▲
by
dofm
2mo ago
> Agreed. A university is a place for learning how to ask questions, and it runs on curiosity. This is a rather oversimplified, old-fashioned take on the purpose of university. University in the 21st century is surely a straightforward h
266.
▲
by
dofm
2mo ago
This is shockingly vapid and wishy-washy even for PG. Leaving this weird, rather Valley-specific, even YC-specific, navel-gazing definition of "founder" aside, universities should do what they promise to do for their students. And
267.
▲
by
dofm
2mo ago
That figure likely doesn't include crypto mining, either. That's another 0.5% or so. 0.5% is about the entire energy consumption of two Swedens or half a Thailand.
268.
▲
by
dofm
2mo ago
Redditor is too easy. Conveying the errors typical in a Raise The Colours group on Facebook, that's the hard stuff.
269.
▲
by
dofm
2mo ago
I have found it intriguingly difficult in my own tests to make LLMs deliberately fuck up grammar in any way similar to how humans do. I suppose this is actually an architectural limitation.
270.
▲
by
dofm
2mo ago
I am observing that it is a bad analogy. China is not a threat to the daily life of individual Americans in the way Russia has always presented a threat to Ukrainians. China isn't even a threat to the the daily life of individual Ame
More ›