saithound

9 ms·

91.
by saithound 1y ago
No, the post's article is not about the sort of shape identification task discussed by GP. Or indeed any image recognition task: it's a paper about removed context in language. Fwiw, I did test GP's task on ChatGPT 4o directl
92.
by saithound 1y ago
Cool. Since ChatGPT 4o is actually really good at this particular shape identification task, what, if anything do you conclude about its intelligence?
93.
by saithound 1y ago
That's _exactly_ what the LLM did: the article's authors decided to count that as a failure.
94.
by saithound 1y ago
It's very regional. Buses meant for right-hand drive markets like the UK, Australia and Japan are (with very few exceptions) shaky, low-entry configuration two door junk, or double-decker one door junk. If that's what you're
95.
by saithound 1y ago
Needs (2024) in the title.
96.
by saithound 1y ago
FWIW I fully agree.
97.
by saithound 1y ago
Since developers working a tplaces like Parkes Observatory use LLMs regularly, it seems like experience ("13-year-olds" versus "senior engineers" at the two extremes) doesn't explain this gap as well as you imply. T
98.
by saithound 1y ago
I do not disagree with your broader point, but its worth noting that The Economist article deliberately framed its analysis around datasets that also wouldn't capture the economic slowdown! That choice itself is telling.
99.
by saithound 1y ago
I'm not sure either. I'm pointing out that The Economist is presenting misleading metrics deliberately: based on the "reliable American data" that they chose, you wouldn't see evidence of an ongoing recession either
100.
by saithound 1y ago
I don't even lean toward the worst-case AI narratives, but it sure feels like Economist journos will keep pushing our "here's why AI wont take your job" articles, even as their own writers get quietly pushed out by ChatG
101.
by saithound 1y ago
This would be a wrong argument even if your premise about Switzetland was factually true (it's not). It's like praising Danish architecture for its earthquake-resistance since no Danish building ever collapsed in an earthquake. It
102.
by saithound 1y ago
CMU professors can't build AI agents, and decide to brag about it. That's the article. "We tried something, and we couldn't make it work. Therefore it must be impossible to do." I agree with the article's main
103.
by saithound 1y ago
Levy war, conclude peace, contract alliances, establish commerce. Also, mandate schooling, mandate vaccinations, forbid polluting the environment, set clocks forward, and various other acts. Governments exist so that we can coordinate doing
104.
by saithound 1y ago
OpenAI o3 and o4-mini are massive disappointments for me so far. I have a private benchmark of 10 proof-based geometric group theory questions that I throw at new models upon release. Both new models gave inconsistent answers, always with w
105.
by saithound 2y ago
Has anyone added another formally proven layer or component? Yes, they're being added all the time. Recently added features: - Support for a bunch of new architectures, including RISC-V. - Mixed criticality scheduling, which provides c
106.
by saithound 2y ago
> If you leave the detector in place, but don't record the data that was measured, the interference pattern is back. This is not remotely true. It looks like you read an explanation of the quantum eraser experiment that was either f
107.
by saithound 2y ago
> My most senior engineer on the team never met a deadline in his life. I went over this sentence a few times before it started making sense. Noting this for others who run into it. What's meant is that the most senior engineer fai
108.
by saithound 2y ago
> But seriously, according to that link, it seems to me like the zero or empty type is more suitable. That's a bit different. The empty type is only suitable for functions that never return (e.g. loop infinitely, crash the program).
109.
by saithound 2y ago
Shadowing always has been a feature, doubly so in languages which lack linear types. It is a promise to the reader (and compiler) that I will have no need of the old value again. Notice that applying the naming convention you suggest does n
110.
by saithound 2y ago
> There's lots of factual information here you can know and use to make informed "guesses" (if you will). The gambler who learned the entire observable history of a tumbling RNG will not be in a better position to take the
111.
by saithound 2y ago
There are thousands of gamblers whose job is to literally predict the tumbling random number generators in the slot machines they play, and will be rewarded with thousands of dollars if they do a good job. They are not experts. As said abov
112.
by saithound 2y ago
> Because the models have continually matched the quality they claim. That's very far from true. "Yes, I know that the HuggingFace arena and coding assistant leaderboards both say that OpenAI's new model is really good, bu
113.
by saithound 2y ago
I imagine that the leader of the Singularity Institute (now Center for Applied Rationality) using the workshops as a recruitment ground for his personal psychedelic drug cult, and his followers then killing a bunch of innocent people, did n
114.
by saithound 2y ago
Yes, it looks all but certain that OpenAI gamed this particular benchmark. Otherwise, they would not have had a contract that prohibited revealing that OpenAI was involved with the project until after the o3 announcements were made and the
115.
by saithound 2y ago
Fwiw (not color-blind) I can see red, green and black pixels. The graph doesn't look weird to the naked eye. Find the interactive visualiser by scrolling down, and switch it to "Files in Anna's Archive [md5]". This will
116.
by saithound 2y ago
Girard is one of the top logicians of his time. He can write standard mathematical prose when he wants to. In turn, when he chooses to use the continental philosophy style, it's a deliberate choice: people who can't read it are no
117.
by saithound 2y ago
Of course. Pretty hard to deny it when transformers learmed to produce useful human language not by figuring out how it relates to some Tarskian "true reality" but by looking at a lot of text and figuring out its internal use, exa
118.
by saithound 2y ago
Since I have the exact same thought (that you are coming at this from a software engineering direction, not a computer architecture/engineering one), this is unlikely to be further productive :( You say you want students to be able t
119.
by saithound 2y ago
Very good question. My working assumption throughout was that the people in a computer architecture class already had 1 or 2 semesters of other programming courses where they worked in a high-level language, and are looking to learn how com
120.
by saithound 2y ago
I answered a similar question by anyfoo downthread, the point about implementing a PIT should explain what I mean, and why I think that emulating a simple real architecture might be more useful while requiring similar effort.

More ›