angoragoats

42 ms·

61.
by angoragoats 2mo ago
.name is in no way a "nonstandard TLD." It has been in existence for a quarter of a century and was part of the first batch of new gTLDs approved after the dot-com boom had begun, in 2000. For the first couple of years of .name&#x
62.
by angoragoats 2mo ago
Yes, I am aware you can do this, and it doesn’t refute anything I said.
63.
by angoragoats 2mo ago
That isn't responsive to anything I said. I'll summarize my position in case you're confused: * You said that humans are just as hallucinatory as LLMs when speaking about anything other than their flat or kids, and I think th
64.
by angoragoats 2mo ago
Twitter is dead. You are at the funeral. Given that Nitter et al are now mostly gone, please stop posting Twitter links to HN, as it appears to be against the site rules.
65.
by angoragoats 2mo ago
> Gees... it never occurred to me that such an obvious thing needs a proof. I didn’t ask for a proof, I asked for a source, and it is not obvious to me. Your claim is that “humans are just as hallucinatory [as LLMs] when speaking about p
66.
by angoragoats 2mo ago
> Yeah, but humans are just as hallucinatory when speaking about pretty much anything other than their flat or kids. Do you have a source for this? I vehemently disagree. > 99% of people learn by memorising facts, not by testing reali
67.
by angoragoats 2mo ago
> HN submissions are usually long-form content, so it's rare to see X links anyway. There are, at minimum, tens of Twitter links that I see daily which make it to the first couple of pages of the site. A lot of them are AI-related c
68.
by angoragoats 2mo ago
Sorry, forgot to add the link! Edited my post to add it like I intended.
69.
by angoragoats 2mo ago
No, it’s not anymore, since the HN FAQ[0] specifically states: “It's ok to post stories from sites with paywalls that have workarounds.“ There don’t appear to be workarounds for Twitter anymore. [0] https://news.ycombinator.
70.
by angoragoats 2mo ago
No, it's not in the video. The video was not focused on performance per watt at all, and the best attempt that the video makes at measuring performance per watt actually shows the opposite: that the 5090 system is more efficient than t
71.
by angoragoats 2mo ago
You are both correct. He originally wanted to buy it because his feelings were hurt by lots of Twitter users (like the Elon Jet guy), and I can only speculate that he thought it’d fill the emotional void in his soul to own it and do what he
72.
by angoragoats 2mo ago
And people who downvote comments because the facts therein aren’t stated in a way they like are snowflakes.
73.
by angoragoats 2mo ago
The name of the party is the Democratic Party. I am aware of Mamdani using that phrase approximately 5 years before he ran for NYC mayor. Who else has said that?
74.
by angoragoats 2mo ago
Yes, but LLMs are hallucination machines by design. What I mean by this is that everything the LLM generates is a hallucination, untethered from truth or reality, since the data is generated purely based on statistics around what word is
75.
by angoragoats 2mo ago
> Twitter is good if you similarly curate your experience and don’t let it push you into garbage. I want to follow the guy who tracked Elon's jet and documented its fuel usage and CO2 emissions. Oops, can't make that part of my
76.
by angoragoats 2mo ago
> only saying Musk is bad because he's using drugs I never said any such thing. Please don’t misrepresent my words.
77.
by angoragoats 2mo ago
> Are you anti-drugs? No. > Do you think drug users are bad people No, no one is a bad person only because they use drugs. > or using drugs is condemnable As a general statement, no. But my answer is yes, your actions are condemnab
78.
by angoragoats 2mo ago
Agreed for sure. I think part of the problem is that at least for a while, he was good at selling the lie that he was an intelligent engineer, which caused a lot of people (especially tech-savvy ones) to trust him more than the average rand
79.
by angoragoats 2mo ago
Option A: You occasionally encounter power tripping mods on a site where you are free to create your own community and become the mod yourself. Option B: a power tripping drug-addled Nazi billionaire owns the entire site and can change the
80.
by angoragoats 2mo ago
Sir, you are voluntarily using a website owned by an unstable Nazi, who purchased it because his feelings were hurt and people were making fun of him there. Now his brain droppings make it into everyone’s feed by default, from what I unders
81.
by angoragoats 2mo ago
Where do you live? Because I’m guessing you’re wrong.
82.
by angoragoats 2mo ago
Sure, I assume you mean with userscripts or custom CSS. I can’t be bothered; just going to continue to visit the site less.
83.
by angoragoats 2mo ago
> Now, on the flipside, the yoke also exists in the Tesla Model S Plaid, which has a traditional manual rack. In here it is the worst god-damned thing to ever curse a vehicle. Yes, as I said, bad product decisions.
84.
by angoragoats 2mo ago
> Musk is aggressively optimistic on what can be achieved and by when That’s putting it quite charitably, IMO. > but rational on product decisions Hard disagree. You only need to look to his entire tenure at DOGE for literally hundred
85.
by angoragoats 2mo ago
I learned from experience at least a decade ago that you should not take anything he says seriously or as an assertion of fact. Countless examples[0] prove that this is the case. [0] https://elonmusk.today
86.
by angoragoats 2mo ago
You can parallelize inference with much cheaper GPUs as well! I’ve got a 4060 ti 16gb, and I’m thinking about getting another. I previously specced out a cluster using multiple 3090s. At the time, the 3090s were going for $700 on eBay. They
87.
by angoragoats 2mo ago
You seem to be acting deliberately obtuse. > What about [...] performance per watt which I’ve now mentioned ~~four~~ five times and you’ve ignored the previous ~~three~~ four times?
88.
by angoragoats 2mo ago
Nah, I don’t think I will.
89.
by angoragoats 2mo ago
Have you ever actually set up a multi-GPU system for inference? Based on my experience you are drastically overstating the problem. Both tensor and pipeline parallelism (without NVLink) produce a machine which is faster than any Mac on the
90.
by angoragoats 2mo ago
I flag and hide every single Twitter post that I see. HN’s code seems to be stuck in the 2000s, so there’s no easy way to automatically block or filter beyond that. To be honest, I just end up coming to this site less than I used to.

More ›