Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
selcuka
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
7 ms
·
31.
▲
by
selcuka
1mo ago
> maybe that's not Google's use case in mind That would be interesting when they also own Google Meet.
32.
▲
by
selcuka
1mo ago
I'm not sure how the examples given in this thread, such as brain scans, are "unscientific" while simply saying "nah, I believe everybody else is exactly like me" is "rational voice"?
33.
▲
by
selcuka
1mo ago
> why do we default to believing that there are suddenly two types of people Because we know that there are already more than two types of people, even when we ignore aphantasia? So there is nothing new here. Also if you read the FA, it
34.
▲
by
selcuka
1mo ago
In fact, the term typical mind fallacy was born from aphantasia [1]: > Galton gave people some very detailed surveys, and found that some people did have mental imagery and others didn't. [...] Though another psychologist, Dr. Berma
35.
▲
by
selcuka
1mo ago
> that's what people mean when they talk about seeing something in their mind. That's an oversimplification. I have a friend who says that they can visualise things mentally exactly like in real life, with their eyes closed or
36.
▲
by
selcuka
1mo ago
> Why do people ask about dreams? This has no connection with visualization. What you can "see" when your eyes are closed has no connection with visualisation either. There are no photons hitting your retina. So aphantasia and
37.
▲
by
selcuka
1mo ago
> I want to know if aphantasia is associated with any other mental models. For example, driving, map reading, spatial awareness. Aphantasia is usually associated with spatial memory difficulties [1]. Anecdotal, but I am aphantasic (almos
38.
▲
by
selcuka
1mo ago
> most people don't even remember what a bicycle looks like Not only people. LLMs too.
39.
▲
by
selcuka
1mo ago
"it"
40.
▲
by
selcuka
1mo ago
> LLMs inserting code comments makes zero sense by definition. Not completely, because anyone who reviews the code in the future does not have access to your original prompt, so theoretically leaving a comment that explains the "why
41.
▲
by
selcuka
1mo ago
There is a difference between "not providing visa sponsorship" and "using X as a proxy for not hiring non-citizens".
42.
▲
by
selcuka
1mo ago
> Companies can though use proxies So it's a decision between violating the law and acting unethically?
43.
▲
by
selcuka
1mo ago
> Foreign workers are here for the benefit of the United States But that's the point, isn't it? Some foreign workers who "worked the US for the benefit of the US" are now unable to come back to the US. Isn't that
44.
▲
by
selcuka
1mo ago
They accept more tokens these days, but they are still more accurate with a shorter context [1]. [1] https://arxiv.org/abs/2307.03172
45.
▲
by
selcuka
2mo ago
Not necessarily. I've been using a couple of RackNerd VPSs for non-critical stuff for a few years now and they have been rock solid. Hetzner and Contabo are also reputable providers with competitive pricing.
46.
▲
by
selcuka
2mo ago
Kevin Bacon and Kyra Sedgwig married in 1988, and they are both celebrities.
47.
▲
by
selcuka
2mo ago
> Do we have actual data on whether or not celebrities have more volatile relationships? Yes, they are legitimately more volatile [1] > I find that the overall divorce rate for celebrities during their first 16 years of marriage is 52
48.
▲
by
selcuka
2mo ago
> isn't this obviously the case? It is, but people often act like it isn't and everybody has an equal chance. Therefore it is necessary to frequently restate the obvious.
49.
▲
by
selcuka
2mo ago
> When told to a statistically illiterate person? Which is a rounding error from 100% according to the GP.
50.
▲
by
selcuka
2mo ago
I agree that the examples on that site are not very good. What about [1, 2, 3] + [4, 5, 6] == "1, 2, 34, 5, 6" or parseInt(0.000001) == 0 parseInt(0.0000001) == 1 or "" + 5 == "5"
51.
▲
by
selcuka
2mo ago
> Your link is talking about transistor count. The article is talking about single-threaded performance. But Moore's Law has always been about transistor count, not performance.
52.
▲
by
selcuka
2mo ago
There is a more recent Qwen with 4B parameters, but there is no 3.8 (or 3.6): https://huggingface.co/Qwen/Qwen3.5-4B
53.
▲
by
selcuka
2mo ago
The article refers to it as AGENT.md, but the standard name is AGENTS.md (plural).
54.
▲
by
selcuka
2mo ago
Pi even installs ripgrep and fd if it can't find them in the path.
55.
▲
by
selcuka
2mo ago
Agreed. It's a great logo and I would be very impressed if an AI made it.
56.
▲
by
selcuka
2mo ago
> There's no official Qwen 3.8 4B (only 27B and 2.4T.. at least for now), so if not a typo you've downloaded some third party model/finetune. Probably this: https://huggingface.co/empero-ai/Qwen3.8-4B-
57.
▲
by
selcuka
2mo ago
Those people exist. What is worse is that they don't even bother to provide the correct context to the model, or even use the same model to answer your questions. So, for example, they submit a PR using an LLM, and when you ask them to
58.
▲
by
selcuka
2mo ago
Exactly. You can simply close all open tickets with <WONTFIX> and claim that you've cleared 5 years of engineering work in 5 minutes. It doesn't mean anything.
59.
▲
by
selcuka
2mo ago
> I'm too lazy to figure out how to switch it off. Press MODE, then press ALARM ON/OFF.
60.
▲
by
selcuka
2mo ago
> in the old days if you didn't provide a domain name, the finger command would default to hitting `localhost`. It still does (at least on my MacBook): $ finger myusername Login: myusername ... ... No Mail. No
More ›