dilap

7 ms·

1.
by dilap 5d ago
Yeah that's fair, though my understanding (perhaps wrong!) is Tesla unlike Waymo doesn't need super-detailed maps; it could in theory go across the state right now (but it doesn't have legal authorization). (In Austin it co
2.
by dilap 5d ago
Yeah that's a good point -- I'm not sure, but it definitely felt like there was no remote control going on, just based on smoothness and speed. But for sure the ability exists, so good pts re true autonmy and scaling.
3.
by dilap 7d ago
Isn't Tesla already there? I took a fully-autonomyous Cybercab to a coffee shop yesterday; it doesn't even have a steering wheel. It was just a short drive, but it handled various complications (a garbage truck, double-parked cars
4.
by dilap 28d ago
GLM 4.7 is gone (at least for us), with no suitable replacement from Cerebras. I think all they care about now is hardware and OpenAI hosting.
5.
by dilap 1mo ago
This is neat! Would love the ability to make "true noon" be at the top of the clock, instead of 12.
6.
by dilap 1mo ago
Years and years ago I saw this story on k5 when it came out and never read it -- but from commentary I always thought was some sort of hard sci-fi exploration of what happens when we get AGI/ASI. So all these years later AI is real, I
7.
by dilap 1mo ago
It's exactly because everyone's tone is different that I'm confident it's AI. Granted, a bit of AI-influencing phrasing in human writing is common, but the idea that someone sat down, worked hard, and came back with an e
8.
by dilap 1mo ago
You're too kind. It's obviously AI copy, and a huge negative signal (to me, anyway).
9.
by dilap 2mo ago
Don't worry, we'll get there soon enough.
10.
by dilap 3mo ago
Have you tried iterating on style feedback in AGENTS.md? I've been reasonably successful using this to get it to output code in a terse, non-defensive style that matches my hand-written code.
11.
by dilap 3mo ago
Reload should fix (did for me).
12.
by dilap 4mo ago
Doesn't look minified, just very dense, almost like progcomp code. First time I've seen an LLM spit out that style of code, I'm impressed!
13.
by dilap 4mo ago
It's just Claudeslop. It's everywhere. An epidemic. If you're familiar it stands out instantly. (Would you let someone else talk for you? In real life? Like open up your mouth and let a TTS system spit out the sounds and pick
14.
by dilap 4mo ago
Why Codex GPT-5.5 High instead of Extra High, I wonder?
15.
by dilap 4mo ago
Yeah, it has a tendency to default to "smallest local hack that will work" and code as defensively as possible. BUT I have had great success using AGENTS.md and becoming better at prompting to get it to not be like this. Basic app
16.
by dilap 4mo ago
Personally I don't find this to be true anymore! It's not always great and does still will often tend towards unneeded complexity (especially if not pushed a bit), but I often find GPT 5.5 writing code I would have written myself.
17.
by dilap 4mo ago
way easier to work on a focussed c codebase you own than a mature unwieldy c++ codebase you don't. but it's fine, people will take that work and port to llamacpp and everyone wins. (the ux of ds4 is fantastic too -- it's dead
18.
by dilap 5mo ago
FWIW, I've found Codex with GPT-5.4 to be better than Opus-4.6; I would say it's at least worth checking out for your use case.
19.
by dilap 5mo ago
Deepseek R1 was a publically-available, MoE model that was getting a ton of attention before llama4. Llama4 didn't get much attention because it wasn't good.
20.
by dilap 6mo ago
i use a an auto-layout tool, so having windows stacked on top of each other is super-common for me, and the fact that they all peak thru each other (like the screenshot in the blog) looking absolutely terrible drives me crazy
21.
by dilap 6mo ago
I wish I could remember the details -- I know I got annoyed with things being slow and when I got a new computer decided to go the no-homebrew route. I'm using nix, and it seems fine so far, but I also really don't understand it a
22.
by dilap 6mo ago
Horses for courses, but I've stopped using brew 'cuz it's too slow, so this might bring me back! Edit: no, it won't...
23.
by dilap 6mo ago
I thought this was parody until the last sentence.
24.
by dilap 6mo ago
I still remember the very first time I saw an emoji -- just an old-school dumb phone, and my friend sent me a message with an emoji, which the phone, amazingly, was able to display. I had no idea such a capability even existed, and wondered
25.
by dilap 7mo ago
Welllll I'm not so sure that phrase is well-suited for your intended meaning, then. (Also, tangentially, I think could argue thinking models w/ the elided thought prelude satisfy "having hidden state where it first forms an o
26.
by dilap 7mo ago
Ha, thanks!
27.
by dilap 7mo ago
I don't know for a fact, but I'd bet a few digits of cold hard cash it's a SwiftUI rewrite that is to blame. (Any1 in the know want to chime in?) And yeah, it's terrible. Apple doesn't make good apps anymore. (This
28.
by dilap 7mo ago
The fundamental nature of the model is that it consumes tokens as input and produces token probabilities as output, but there's nothing inherently "predictive" about it -- that's just perspective hangover from the histor
29.
by dilap 7mo ago
That description is really only fair for base models†. Something like Opus 4.6 has all kinds of other training on top of that which teach it behaviors beyond "predict most probable token," like problem-solving and being a good cha
30.
by dilap 7mo ago
MIT ( https://gist.github.com/karpathy/8627fe009c40f57531cb1836010... )

More ›