@wlxd's banner p

wlxd


				

				

				
3 followers   follows 4 users  
joined 2022 September 08 21:10:17 UTC

				

User ID: 1039

wlxd


				
				
				

				
3 followers   follows 4 users   joined 2022 September 08 21:10:17 UTC

					

No bio...


					

User ID: 1039

Looking at this, I wonder if it’s not some kind of reverse rug pulling, where insiders buy on the dip, knowing reversal is coming.

More seriously, I wonder how aware Trump is that this constant flip flopping is destroying his ability to make credible threats in future.

What this means in practice is that if talk to the perp, and you indicate or imply in any way whatsoever that he is expected to answer your questions, the conversation is now custodial interrogation, and if you don't mirandize the guy first, your case is fucked.

Oh, we can continue applying the doctrine to illegal wiretaps just fine, that's not my problem with it. My problem is things like, if you fail to recite a specific magic incantation before your suspect confesses to the crime, you must disregard that confession.

Can you explain how it is a problem? It's not immediately clear to me, and it's apparently not immediately clear to most of the legal systems around the world, given that they do not subscribe to the extensive application of this doctrine.

The government has no responsibility to right wrongs in general. There are some specific laws that apply to some specific contexts, for example Federal Tort Claims Act (which would not apply here), but as a general matter, government enjoys sovereign immunity, which means that unless some specific law applies, it is under no obligation to compensate you if it wrongs you.

Your analogy misses some key important points (e.g. the neighbor house party should be a strict invite only event, and you only got into it because you literally snuck through the window), but more importantly, the issue is not compensation for plane ticket. I'd be happy to give that guy $1000 for him to fuck off. He can buy plane ticket to US with it, but he will not be admitted into the country, because the US government is under no obligation whatsover to admit any noncitizens into the country.

The fruit from the poisonous tree doctrine as applied in the US is pretty stupid. It is beyond retarded that good faith procedural errors can allow obviously guilty men go free. Most of the rest of the world does not have it, or does not have it to the same extent as US does.

We can nitpick on what we mean by “visible”, but at the end of the day, that’s really not a high bar to meet. The only visible form of political expression I ever engaged in was anonymous posting on SSC/TheMotte. Most of my friends don’t do even that.

They can, however, move out themselves and their stuff out of the blast radius, which actually does make a difference.

It has only partly happened, which is even worse. We are full on papieren bitte, except where it would help deporting the illegals.

Falsely claiming that you’re a US citizen to an ICE agent conducting a raid is a federal felony (18 U.S.C. § 1001). I suspect that ICE will have enough circumstantial evidence to support a reasonable belief that you’re lying, which will allow them to arrest you, and unless you help them in their investigation, you’ll be sitting in jail until they do in fact determine you lied. This is limited by detention capactity, of course.

He is also claiming the right to do it in a way which would probably prevent a US citizen caught up accidentally getting relief through the courts.

That depends on what happens with them after they get deported. If they are set free in some foreign country, they can literally buy a plane ticket and come back here. Sure, they will not get a relief that will prevent them from getting detained and removed from the country in the first place, but I don’t see how this is relevant. It is perfectly normal and common to be able to get any relief only after illegal action of government has already been inflicted upon you. It really is not substantially different than getting illegally arrested and jailed: the government needs to take utmost care to avoid doing it, but when you are getting arrested, you cannot get court relief right there and then.

This is true, but it’s also true that calling your recruiters to check up on status often does make a difference, and usually expedites the entire process. As anyone who has ever worked in any large company, pinging people regularly to update their status and nudge them towards doing or accelerating the work they owe you, is in fact a significant part of your work, and does make a difference. Recruiters/HR is no different, and pushing them does work.

Only if all your friends also live within walking distance, which is unlikely in any case.

Putting parking lot in the back makes little difference. It only matters if you’re already on the street with the front entrance. If you’re a block away, you need to cross the parking lot anyway.

More generally, without a car you are effectively disabled, in the literal sense: not fully able to participate in society in the same way the majority is. Most societies make some accommodations to the disabled, but there are real limits and trade offs involved. For example, people who have trouble walking (which in US is I suspect a bigger group than people without access to a car) typically appreciate a lot being able to drive right to their destination, and park right at the entry.

Point is, it is not clear to me why we should cater to your disability, to the detriment of the majority.

It's easy being an AI advocate, I just have to wait a few weeks or months for the people doubting them to be proven wrong haha.

Often you don’t even need to wait, doubters often say things that are wrong already when they say it. Remember Gary Marcus? He was big a couple years back, but everyone learned to ignore him after basically everything he said was wrong.

The LLM is still talking to the image generator. It just does so using native tokens and vectors instead of going through a text encoder layer in-between.

This is very confusingly stated. The second sentence is correct, but in the first one, it’s confusing to say that LLM is talking to image generator, because the LLM and the image generator are literally the same thing.

They marry legal immigrants present in the US or marry foreigners outside the US.

You missed one option, that is, they marry illegal immigrants, but otherwise this my point: all family based immigration is downstream of previous immigration events.

Yes, legal immigration.

This is only true in the most literal sense, as illegal immigrants are not eligible for petitioning for family based immigration, but overall it misses my point: US citizens can and do petition for family based immigration for their illegally present parents, spouses, and siblings. There is a whole legal industry for that, just search Google for “green card for undocumented parents”, you’ll find many immigration law companies adveritising their services. If your spouse, parent or sibling did not enter unlawfully, but eg. overstayed their visa, there is hardly any legal issue preventing adjusting their status. Even if you entered unlawfully, all you need to do is sneak out of US, and then pretend you’ve never been here illegally; perfectly viable for many illegals who haven’t generated federal record of their presence.

You've provided no evidence of this.

Sorry, do you actually believe that people who enter or stay illegally are a complete dead end from family based immigration purposes, or are you just asking for evidence this to be obnoxious? The existence of the legal industry dedicated to legalizing parents and spouses is evidence. You can find many businesses that help with that in Google.

They are being sponsored by US citizens or permanent residents.

OK, and how exactly do you think US citizens come about to have foreign family? Again, think it through. These US citizens are almost universally downstream of some relatively recent immigration event. They either became naturalized citizens, or were born to immigrants. People who were born to two American-born parents are highly unlikely to have any foreign family that they could even consider getting in here.

So, basically all family based immigration is downstream of recent immigration. If you ignore the family based legal immigration, then legal immigration has been absolutely overwhelmed by illegal immigration for many years now.

What I am arguing here is that large chunk, if not a majority of legal family based immigration is downstream of some illegal immigration event that happened in recent history. For example, US citizen born to illegal parents, or formerly illegal aliens who legalized their presence in some way (and there are ways to do it for quite a lot of people).

Therefore, when you say that

that the type of immigration that will be paused under this administration - asylum claims from people entering through the southern border (…) - is a fraction of the overall immigration growth since the 60s.

You are missing the forest for trees: yes, family based chain migration have brought more people here than fake asylum claimants, but the point is that fake asylum claimants will cause more family based migration in future, so reducing the former also reduces the latter.

Only US citizens or permanent residents can sponsor their own family members. The "illegal" spouse cannot sponsor anyone

Yes, that’s why I said “get married to legal residents”.

unless they acquire permanent residency, which is very difficult if they entered the US without a visa.

Anything involving legal immigration system is “very difficult” if you talk to pro immigration advocates, but as it happens, I personally know multiple people who went exactly this route, and it worked out for them.

I would think it's a small portion of the overall family based visas.

Who do you think are the millions of people coming on family based visa? Are you suggesting that 200k of former employment based green card recipients are bringing 5 family members from abroad each?

I think I know what you're getting it, but just so I understand, can you explain how the parents and siblings are "illegal" in this situation?

You sneak through the border with a husband and a kid. You give birth to an another child. Your child is a citizen. Once he becomes old enough, he sponsors your and his brother’s family based green card.

Showers are annoying but what’s way worse is water saving dishwashers. They run for way longer, and do bad job cleaning.

Roughly 85% of new immigrants every year are immediate relatives + other sponsored family members + employees.

Of legal immigrants, yes, but 1) recently, we have had more of illegals immigrants than legal, and 2) legal family migration is to a large degree downstream of illegal immigration. Illegals come here, and get married to legal residents, which enables them, and their families to start chain migration. They give birth to children who have been treated as US citizens, which again allows them to bring their illegal parents and older siblings through family process, and also to get married to illegals who then are legalized.

As it happens, Trump actually tries to do something about that, with his EO stopping birthright citizenship to illegals. We will see what SCOTUS says about this, but the implications of it could be enormous. For example, it will discourage illegals from starting families in here, because their children will be in bad legal situation. It will make it easier to deport parents of small children if they are no longer US citizen children.

You seem to fundamentally misunderstand what multi modality is.

The LLMs are not trained on text corpora. They are trained on tokens. These tokens are just numbers. The training process is good at finding patterns in the numbers they are fed, and predicting subsequent numbers.

Where do the numbers come from? Well, in text only LLMs, there is a tokenizer stage that takes text input and translates it into string of numbers. These numbers don’t directly correspond to individual letters or words, and so LLMs never really see those. This is why, for example, they had trouble telling how many rs are in strawberry, because what you’re asking them is how many rs are in 2645, 675, 15717 (this is the tokenization of Strawberry). It’s actually pretty magical that they manage to figure out that there are any rs in these numbers at all, that they are able to learn detokenization.

So how about different modalities? Well, these are represented as just different tokens, different numbers. For us humans some numbers represent image fragments, some represent sound snippets, and some represent pieces of text. For a multimodal LLM though, it’s all the same, it’s all just numbers. When you train them, you feed text, images, and sounds directly into the model (after tokenizing), you don’t do anything like “first transcribe sound to text through some different model and then feed it as text”. The sounds are never mapped 1-1 to English tokens, they are mapped 1-1 to sound tokens, which is just a different set of numbers than those representing Unicode text.

Because of this, for an LLM, not only sounds and words are intercompatible, but in fact there is literally no difference between the two on a technical level. There is of course a difference on a semantic level: for example, it learns quickly that text and sound tokens never mix, because it’s never fed training data that’s intermingled sound and text. Instead, after sound comes another sound (eg response to a dialogue) or a bunch of text (a transcription). But, to a model, there is no fundamental difference between text and sound and image.

Thanks!

I don’t understand what you mean, eg

But audio tokens are not intercompatible with text tokens, for obvious reasons

What exactly do you mean by “not intercompatible”?