adw

10 ms·

151.
by adw 3y ago
It’s breaking into Cotton Eye Joe by Rednex (which is another piece of contemporary Eurodance in the same vein).
152.
by adw 3y ago
Not all courts have the power to set precedent. Small claims courts in England don’t.
153.
by adw 3y ago
Slashdot was way way earlier. Also massively overcomplex (metamoderation) and full of trolls and much narrower. Reddit was basically Slashdot 4chan, in a meaningful sense.
154.
by adw 3y ago
ICLR, ICML, KDD, ICCV, ECCV and NeurIPS are all major "AI" conferences and none of them has AI in the name. NeurIPS comes closest. "AI" has historically not been a useful distinction for anyone in the field. I actually t
155.
by adw 3y ago
This debate is a classic. AI has always been an overloaded term and more of a marketing signifier than anything else. The rule of thumb is, historically, "something is AI while it doesn't work". Originally, techniques like A*
156.
by adw 3y ago
HN has always been a tense standoff between a few cliques, the first two being the ostensibly intended audience; * competent and curious engineers * entrepreneurs, who live on a continuum where one end is... * ...hucksters and snake-oil pur
157.
by adw 3y ago
Equal temperament was, originally, what allowed us to build physical instruments which can be played in all keys; but as a _result_ of that, it's what allows us to write music which fluently changes key (beyond the relative minor or, a
158.
by adw 3y ago
Python has https://docs.python.org/3/library/collections.html#collectio... (which I'm sure would raise objections over the volume of code – https://github.com/python/cpython/blob&#x2
159.
by adw 3y ago
You're neglecting the extensive theory which has developed around the "Heathrow Terminals 1, 2, and 3" opener. Go west, young competitor!
160.
by adw 3y ago
This is very cool and very educational. Don't deploy it, though. :-) I needed something like this once, but at a little bit larger scale (few tens of thousands of documents). The answer, as always, was [sqlite]( https://www.s
161.
by adw 3y ago
It's, in practice, applied gradient-based optimization on a geometrically-defined objective, which means if you squint hard enough it's machine learning. (I am a crystallographer turned machine learning engineer.)
162.
by adw 3y ago
Yep. It's not noise on the MIDI signal, obviously, it's crosstalk induced by USB circuitry. It gets worse when bus power is involved; USB-powered instruments are notorious for it.
163.
by adw 3y ago
5-pin MIDI out is still the best MIDI out, at least for serious applications.
164.
by adw 3y ago
2k lines is (probably) not a lot of data.
165.
by adw 3y ago
It is a unit of text the LLM processes. :-) Everyone uses (byte pair encoding)[ https://en.wikipedia.org/wiki/Byte_pair_encoding ] to generate their tokens; the tokens are whatever emerge from this. They will typically c
166.
by adw 3y ago
America’s superpower is immigration. The core teams at the biggest and best players are recruited globally; the high end of software engineering is, economically, the NFL or the Premier League.
167.
by adw 3y ago
It’s shown up in one of the bigger Magic-podcast Discords too. You nailed the writing on this. And “always validate your inputs”!
168.
by adw 3y ago
Loaded language is _entirely_ appropriate when you're talking about a frozen war (which, let's be clear, is what we are talking about here).
169.
by adw 3y ago
This is sort of a distinction without a difference. It's an autoregressive sequence model; the distinction is how you're encoding data into (and out of) a sequence of tokens. LLMs are autoregressive sequence models where the &quot
170.
by adw 3y ago
Received wisdom has been “not competitive”. I wonder how the MSR folk feel about all of this, too.
171.
by adw 3y ago
Ikea has the wildest legal structure, but yes, a lot of IKEA is technically owned by a couple of "nonprofits" which happen to pay out a lot of money to the Kamprad family.
172.
by adw 3y ago
We're agreeing. The issue is that what I've said is true _as defined_ (the value of work is what you can get paid for it; if hiring someone to do something costs X, and you can earn Y>X doing something else, then it's a de
173.
by adw 3y ago
This is just, in an economic sense, another application of comparative advantage. If you can get paid more than your childcare costs, it's an economic loss not to do exactly that. You get what you measure.
174.
by adw 3y ago
The whole backstory here is, to say the least, complex, extends over about a decade, and is important subtext for the last few SY records. Starting around 2003, Jim O'Rourke (go listen to Bad Timing and Insignificance, which are both a
175.
by adw 3y ago
Most elite universities have been forced (by market forces) into doing what Harvard is doing. The proportion of firsts being awarded by Cambridge, for example, had already inflated substantially 20 years ago. The solution involves having, r
176.
by adw 3y ago
It's an experimental discipline implemented _using_ computers. Programming is necessary but definitely not sufficient. I'm biased, but in my experience, stats, the math-heavy end of economics, and computational physics/chemis
177.
by adw 3y ago
... I somehow shouldn't be surprised that industrial snakebite-and-black exists. The favourite drink of UK goths everywhere. I can feel Andrew Eldritch sneaking up behind me. He has a drum machine. ( https://en.wikipedia.org&
178.
by adw 3y ago
I feel like Flexport should lose their right to post wanted ads here permanently over this. It’d be a largely symbolic punishment, let’s be real, but unethical behavior like this shouldn’t be allowed to pass.
179.
by adw 3y ago
Transformers have been industrially ubiquitous for a lot longer than GPT-3, initially in machine translation.
180.
by adw 3y ago
Oh, yeah, absolutely; I’m post-Eternal September too!

More ›