@FeepingCreature's banner p

FeepingCreature


				

				

				
0 followers   follows 0 users  
joined 2022 September 05 00:42:25 UTC
Verified Email

				

User ID: 311

FeepingCreature


				
				
				

				
0 followers   follows 0 users   joined 2022 September 05 00:42:25 UTC

					

No bio...


					

User ID: 311

Verified Email

Always the question, categories for what purpose? Whether adoptive mothers and mothers are the same or different categories will depend on whether you're a woman trying to commiserate about pregnancy (obviously different) or a child classifying their experience of motherhood (usually indistinguishable). Without clarifying what task we are trying to differentiate, the question is unanswerable; when clarifying, the categories almost disappear.

Let me just google these:

  • Policy Press seems to be attached to Bristol University.

  • Stylus Publishing doesn't offhand have an association with any university, though they certainly depict themselves as a university publisher.

  • Multilingual Matters "is an independent academic book publisher based in Bristol, UK"

  • And St. Martins just seems to be a freeform publisher.

So depending on qualification, this is either 0/4, 1/4 or 3/4. What do you consider an "academic press"?

To be fair, abortionists are maybe just smart enough to know that it'd be bad play. And at the end of the day, the way to deceive someone is doublethink- just avoid noticing the contradiction. Humans are very good at this. The unusual behavior from a human would be requiring consistency based on the actual cognitive features of the embryo or baby. "I don't see how that helps me win this conflict."

From the fact that you are allowing @foreverlurker to walk

Was the parent comment edited...?

First, @foreverlurker [...] you are banned

Or it is literally just a parameter (PDF).

At a high level, we first identify a sparse set of attention heads with high linear probing accuracy for truthfulness. Then, during inference, we shift activations along these truth-correlated directions. [...] ITI results in a significant performance increase on the TruthfulQA benchmark.

As an aside, when I mentioned I was excited about this disclosure to a deeply Blue family member, they suggested I've shifted right.

This is revolting to me. I could see myself telling a family member "Hey, it's cool you're interested in that, but note that people may perceive that as a Party X aligned topic." I couldn't see myself going "hey, being interested in that makes you party X aligned." To a family member? Do I trust them that little?

The world is full of contradictions and ambiguities. Sometimes one catches your interest and you go off to investigate it. It makes me sick that this good, healthy, praiseworthy act is apparently inseparable from political affiliation.

I don't think the instigators should be tortured to death. I think your quotes on the everyone are very wide.

It's getting more "categories are made for man", IMO. At some point you'd expect these changes to start dissolving the gender binary altogether, or at least make it inapplicable.

**Transhumanists

You are correct that it's a wedge issue, and we're impatiently waiting on the other side of the wedge. Morphological freedom soon!

**Likewise, gender abolitionists

Yeah many trans people are surprisingly gender essentialist, but once the body is cosmetic their opinion won't matter, the ensuing social drift will be unstoppable.

I think you're getting downvoted because you're putting something in that comment that wasn't there to begin with.

Touching a soldering iron and jumping in a housefire both result in burns, and yet they have almost nothing to do with one another. Just because two things may be comparable in one very specific way doesn't mean they're comparable in any other way. This goes especially for the spectrum of human perversions, which famously spans (almost?) every human activity, and double-especially for porn, which is known for not tracking reality particularly closely in the first place.

Seems like a motte/bailey.

Motte: As a group, homos must replenish from non-homosexuals' offspring, necessarily; if they didn't do this, they'd already not exist.

Bailey: Homosexuality is a choice; what's more, it is actually a movement agentically interested in swelling its numbers. To do so, they must make our kids gay.

The problem with the bailey is the assumption that ... either, kids should rather be gay but stay their lives firmly in the closet; or the way that this looks from the outside, kids should just "choose" to "be" straight. Which, as far as I can tell, is and remains largely impossible.

I mean, obviously it's possible for gays to procreate with women, since it's what happened historically. You can stick your dick anywhere, what you cannot change easily or at all is what you need to see in front of you, in reality or your mind's eye, to get hard in the first place.

I don't think the interns were put there as a last-ditch motivation attempt for otherwise-NEETs.

I don't know if this is true, and I don't think it's very likely, but it would be hilarious to me if "jews" turned into a dogwhistle for "the cultural elite".

This is funny to me because Christians have been and still are guilty of doing all of those things: cut off parts of genitals, "sterilization", and IMO teaching eternal punishment in hell is at least as bad as convincing them their parents are trying to commit genocide.

And of course the child grooming.

A secular humanist could maybe make this argument. A christian should attend to the beam in their own eye.

  • -17

As long as the rockets continue to fly, and land, and the competitors continue to largely not do that, or do that plainly worse, at some point you're asking me to disbelieve the evidence of my own eyes.

I don't believe in ideas, I believe in dollars per kg to orbit.

You can take Elizabeth Holmes as another example, though I don't think the comparison is fair, since Musk delivers something tangible, but this would be my guess to what's happening with investor money. Generate hype, get money, use it to make something to generate more hype, repeat.

I mean, if the "something" is "deliver the product you are paid for", I think you have defined every public company ever as a scam.

I'm saying that if he's getting the governments to pay him 3x the price tag that he's advertising to everyone else, then that cheaper price tag is arguably fake, and governments are subsidizing it.

Everybody does that though. It's only fake if the cheap price is dumping. If you're not selling the cheap segment to dumping prices, it's just called "not leaving free money on the table." A company that didn't try to segment the market like that would be incompetent as a business.

I think one more aspect is reflectivity: the degree to which a system integrates knowledge of its own operation into its schema. For instance, a search engine that can show a "Google is down" page, or that lists the number of results, or that finds Google help pages on search as search results, has (basic) reflectivity. It seems plausible to me that a lack of reflectivity is a big part of what's holding LLMs back and causing hallucinations and the like: they may be confident or uncertain, but they cannot condition on their confidence.

It's nonsensical, but it's nonsense that screens off the other nonsense. You're basically double-counting criticism. If you grant that the kids return to being regular kids at the age they left, that already gives you all the nonsensicality required for the hazy memories as well.

since there's no vetting system, you can't even claim that people attracted this way are going to do a worse job.

Sure you can, if you think the populations selected by either will be different. You don't need vetting; effort acts as a filter as well.

Mod decision prediction market when (only half-joking)

I mean, I guess if you're putting a person who has aged a year back into a younger body, you're already applying god-tier/superintelligence-tier modification. At that point, "how" you go about it, if you do it parsimoniously by applying some general effect to the brain, or if you literally just rewrite every atom, you're not so much putting your thumb on the scale as grabbing the scale and tilting it whichever way you like. Of course Aslan could have made them remember clear as day and without any doubts if he wanted to do that. So yeah, I guess I agree that "well, they're remembering hazily because they were de-aged" doesn't do anything, because it's not like de-aging is a primitive process that would have to be further enhanced for clear memories. - Then again, Aslan has never seemed as all-powerful or unconstrained as capital-G God to me. Maybe it is a primitive process.

It's a joke.

The thing I don't understand is how you can possibly train for uncertainty.

The model needs to "learn the feeling of not being sure". But whether it's sure or not always depends on its state of knowledge at the time, and that state of knowledge will never be represented in its training set. Additionally and relatedly, you cannot train a LLM to "notice when it's saying something wrong" without indirectly training it to say something wrong, then say it notices.

You would have to inspect the network and somehow determine when it is objectively uncertain, and to what degree, and then synthesize a training task based on that actual uncertainty. That level of interpretability is pretty beyond us at the moment.

Right, I guess I'm saying if you wanted to train a specific response to a level of uncertainty, it would be difficult to construct the training samples.

Evidently, the model has figured out that something should be hooked up to its uncertainty. But I have no clue how you'd make that happen intentionally.