hungryhobbit

6 ms·

91.
by hungryhobbit 3mo ago
There are no "attention heads" or fixed number of things a model can pay attention to ... or at least not exactly. After every prompt the model decides "I have a weight of 1 to distribute between every token in my context&quo
92.
by hungryhobbit 3mo ago
The real question here isn't "why are we sacrificing the education of the next generation to buy tools that make us feel safe but (logically, statistically) will only prevent one in a million child deaths?" ... the answer to
93.
by hungryhobbit 3mo ago
And (judging by the number of Medium articles I've been turned away from) 95% of all authors leave it on.
94.
by hungryhobbit 3mo ago
Pick Medium, it's great ... if you want a paywall to turn away half your visitors!
95.
by hungryhobbit 3mo ago
Languages change. That word "literally", which you just used, meant one thing for decades (centuries?) ... and now it also (literally) means the exact opposite. You can be the crabby old guy complaining about how the youth ruined
96.
by hungryhobbit 3mo ago
I think he was being a little hyperbolic, but the core concern (that programmers look competent right up until you have them write a FizzBuzz) is very real. (Source: I've helped hire lots of programmers.)
97.
by hungryhobbit 3mo ago
I honestly couldn't tell if the author was writing satire (about AI writing detection?) or is just an idiot who doesn't understand why we have language rules to facilitate communication. I want to think it's the latter, but i
98.
by hungryhobbit 3mo ago
I don't understand: why would anyone strap their phone to their face? What problem does this solve?
99.
by hungryhobbit 3mo ago
I suspect it's because Stepenson stopped editing his work a long time ago. He used to write a ton and edit it down to a normal-sized book, but once he got popular he just decided to (seemingly) publish everything he wrote, without edit
100.
by hungryhobbit 3mo ago
I swear, early on he wrote well , which means he wrote 1000 pages and edited them down to a normal book. Once he got popular I imagine he was like "I'm popular, screw that editing stuff, I'm just publishing every word I writ
101.
by hungryhobbit 3mo ago
His earlier stuff was much better. If you tried Snow Crash or Diamond Age and didn't like them, he's not the author for you, but if you tried his more recent crap like Anathem I would strongly advise you not to judge him based
102.
by hungryhobbit 3mo ago
Something is seriously wrong with an author when they think they need to invent an entire new language (and force the reader to learn half of it) just for a book. ENGLISH IS A PERFECTLY FINE LANGUAGE, AND EVERY GREAT WORK OF ENGLISH WAS WR
103.
by hungryhobbit 3mo ago
I <3 Stephenson's earlier works: Snow Crash, Diamond Age, and even the lesser stuff (the university one, In the Beginning There Was the Command Line, etc.) were all incredibly well written, and I'd strongly recommend them. Bu
104.
by hungryhobbit 3mo ago
So having tons of AIs quoting various literary works and reproducing knock-offs of them has a positive effect on those books' sales? I think you're wrong: there is absolutely damage to the authors and publishers from what the AI
105.
by hungryhobbit 3mo ago
ROFLMAO ... you think this is a liberal problem?
106.
by hungryhobbit 3mo ago
Pretty short/dumb take.
107.
by hungryhobbit 3mo ago
It should be required that every pay-walled article posted here includes a paywall bypass link. NO ONE IS PAYING THE WSJ FOR A SUBSCRIPTION TO READ THIS NOISE!
108.
by hungryhobbit 3mo ago
The "real" cost is what people pay, not what the AI companies claim.
109.
by hungryhobbit 3mo ago
Who is paying $200/day? My company has me on a $30/month plan (I assume they pay annually?). I use Claude constantly and only hit usage limits when I try to do 4+ projects at once. I can't imagine why anyone would need almos
110.
by hungryhobbit 3mo ago
Piracy created AI.
111.
by hungryhobbit 3mo ago
Its not a "study" at all, it's a marketing piece.
112.
by hungryhobbit 3mo ago
The founders left, and it went downhill as a result.
113.
by hungryhobbit 3mo ago
Wikipedia is supposed to be a reference. SO was supposed to help programmers program. The former continues to succeed, with hiccups, because they remember their goal. The latter is dead, and it failed because it tried to be the former
114.
by hungryhobbit 3mo ago
I disagree. Stack Overflow was made to succeed, not fail ... originally. Then the founders left, and meta basically chronicles the echo chamber of mistakes made. It's not the intent: that was to help coders code. Meta was dominated b
115.
by hungryhobbit 3mo ago
I feel like "do X untill tests are green" is sort of the prototypical /goal case.
116.
by hungryhobbit 3mo ago
The key thing is that normally Claude grades Claude's work, so you do get a"pinky promise, as all of Claude's training bias tells it to plunge ahead (even if that means "fixing" a test by commenting it out). With &#
117.
by hungryhobbit 3mo ago
? I'm pretending to part of a group I'm not? Is "people who read web articles on monitors" a group (to pretend to be a part of)?
118.
by hungryhobbit 3mo ago
Why not address the objective evidence the OP provided? To an impartial observer, it seemed quite overwhelming.
119.
by hungryhobbit 3mo ago
I think "If you’re using OpenRouter, you’re already looking to bypass frontier models." is false. Our company uses both Claude subscriptions and OpenRouter ... and a lot of what we use OpenRouter for is more Claude. We do a litt
120.
by hungryhobbit 3mo ago
From the article it doesn't even sound like he cares about pelicans at all, and doesn't think they are a good way to compare models anymore ... but people are used to seeing the test now, and it does serve as a common "hello

More ›