rf15

22 ms·

481.
by rf15 2y ago
Do you have a source on the phonemes situation? First time I hear of it, and I'm pretty sure I've seen people learn them past that point. I know they struggle with it, but, as an example, even japanese people can learn the effecti
482.
by rf15 2y ago
Imagine how much research is held back/misguided because of this... I'm happy I'm out tbh, a good r&d team seems so much easier to work in.
483.
by rf15 2y ago
> Nvidia managed to get it right I don't think they did. If you work in the space and watched it develop over the years, you can see that there's been (and still are) plenty of jank and pain points to be found. Their spread mos
484.
by rf15 2y ago
Their plugins are a (very) mixed bag, but saying to not use their products is a bit too alarmist if you ask me - the baseline IDE is doing fairly well, and teamcity and doing your GitHub-specific PR-stuff from within intelliJ is kind of nic
485.
by rf15 2y ago
> briefly grounding the reset pin of a discrete TPM device with a pair of tweezers Ah, the tweezers strike again, just not for Nintendo this time. Truly the most universal hardware hacking tool.
486.
Top scientist alleges he was barred from studying mystery brain illness(theguardian.com)
62 points by rf15 2y ago|9 comments
487.
by rf15 3y ago
Whenever I travel in a group through Shiphol, at least one piece of luggage gets missing every time. Yes, they manage to get it back to you eventually (although I did lose an expensive pair of running shoes there), but the amount of hoops y
488.
by rf15 3y ago
"We're sorry we got caught" in all its beauty. But I wish other companies, especially telcos, would be a little more bothered by this.
489.
by rf15 3y ago
A lot of comments here assume that two countries are either "at war" or not, even though it's much more of a gradient in which military escalation is one of the last steps. In fact, you can financially support an endeavour of
490.
by rf15 3y ago
We might not be in a direct, conventional, people-shooting-each-other war, but the trade war with them is absolutely raging and we are pretty evenly matched. It's a nightmare. One of the CCP's greatest strengths is to exploit&#x2F
491.
by rf15 3y ago
This, together with the endless philosophising around dark energy, dark matter and whatnot paints a pretty strong arrow towards our models having some flaws when it comes to their large-scale application. I hope to live long enough to see w
492.
by rf15 3y ago
Also note that this is likely not a good sign like "Oh, they're not extinct after all!". You may well have found their last survivor and that's it.
493.
by rf15 3y ago
It kind of is!
494.
by rf15 3y ago
> I am generally skeptical whether such a highly formal approach can help us answer the practical question of whether we can get LLMs to answer 'I don't know' more often I feel like writing an entire paper about the practi
495.
by rf15 3y ago
The actual anomaly is not the temperatures, but how absurdly rapid they change year over year.
496.
by rf15 3y ago
I'm almost at the point where my hiring decisions actually turn the other way - rattle down the SOLID principles and I feel like you're more occupied rote learning random SE bits than internalising what they were born out of. Ever
497.
by rf15 3y ago
I think the "We are in control" thing is a dead giveaway that this is fake. Communicating with the victim might be essential to get access, but afterwards it's just about extracting whatever you need as fast as possible in my
498.
by rf15 3y ago
> Plants don't kill each other They absolutely do. From simple concepts like resource competition in which trees deplete the ground or outcompete each other for sunlight (e.g. any forest floor is usually not covered in grass) to par
499.
by rf15 3y ago
Disappointing to see that this article is just an AI-based transcription and nothing more. Why not link the original video here?
500.
by rf15 3y ago
"Build for our future platform now. Prepare to be a customer."
501.
by rf15 3y ago
I highly doubt this would work considering you train and sample over how common patterns and associations are. A unique pattern that is not dominating the dataset will just be forgotten. Sampling especially will look to reduce these unique
502.
by rf15 3y ago
This is also the argument I'd have against this entire idea of a sleeper agent LLM: if it is just a tiny point in the dataset, it'll probably just get washed out, if not in training directly then the second you apply quantization.
503.
by rf15 3y ago
This is a good reminder that readers of HN should take to heart: Just because a popular github issue gets posted to HN doesn't mean that you should head over and do your posting there. Keep it here, usually the associated experts are a
504.
by rf15 3y ago
The implication was that they added that text later, I thought.
505.
by rf15 3y ago
Have you seen stock photo sites lately? Many of them happily claim they're real photos, too.
506.
by rf15 3y ago
This strikes me as a shortsighted, risky, and frankly unsustainable attitude for a company. It's no surprise they're struggling.
507.
by rf15 3y ago
It is important to note here that SWA is not supported by all models, e.g. Mixtral is not working correctly.
508.
by rf15 3y ago
Ed is the standard text editor, after all.
509.
by rf15 3y ago
People will get fired due to hallucinated bad reports, won't they.
510.
by rf15 3y ago
I feel like comparing it to biological systems glances over like half the points the parent makes. Also of course it's a sophisticated autocorrect, there's no system for agency, which is what we also desire from a proper AGI.

More ›