ToaKraka
Dislikes you
User ID: 108
ToaKraka
Dislikes you
4hr ago
ToaKraka
Dislikes you
8hr ago
not gentle fixfic, but brutal nonconsensual editing
trying to get it to keep it reading in full without it deciding to start skimming
Paper:
Context Length Alone Hurts LLM Performance Despite Perfect Retrieval
Our systematic experiments across five open- and closed-source LLMs on math, question answering, and coding tasks reveal that, even when models can perfectly retrieve all relevant information, their performance still degrades substantially (13.9–85 %) as input length increases but remains well within the models' claimed lengths.
See table 2.
Despite achieving nearly perfect performance on the vanilla needle-in-a-haystack (NIAH) test, most models exhibit large degradation on tasks in RULER as sequence length increases.
What is the current largest context window for an open LLM?
Does not matter as all these long-context models develop dementia when using 10 % of their context window.
This should link to the twitter trend
Note: This link works only for people with Twitter accounts, and is not supported by Nitter.
Kavanaugh concurrence:
As I see it, the conflicting contextual considerations make this a close case as we determine how to construe and where to draw the line on the statutory phrase “with respect to motor vehicles”. In the end, I do not believe that Congress, through such oblique language in an economic-deregulation statute, simultaneously (i) allowed state tort suits against negligent trucking companies and (ii) categorically preempted state tort suits against upstream brokers who negligently select an unsafe trucking company. The brokers and their amici raise serious concerns about the repercussions of state tort liability against brokers, and they may of course (among other possibilities) ask Congress and the President to change federal law. But, as of now, federal law does not preempt state tort liability against brokers for negligent selection of trucking companies.
Industry news source:
The Supreme Court just told every freight broker that it can be sued
The legal standard is ordinary care. The question a jury will now be permitted to ask in every state in America is whether the broker exercised reasonable care in selecting the carrier. That means: Did you check the carrier’s safety record? Was the carrier’s FMCSA data available to you? Did the data show elevated crash rates, conditional safety ratings, high out-of-service percentages, or prior enforcement history? Did you have a documented process for evaluating carrier safety? Or did you book the cheapest truck and move on?
If you are a freight broker operating in the United States today, the preemption defense you have been relying on since 2023, when the Seventh Circuit decided Ye v. GlobalTranz, is gone. You are now subject to state tort law in every jurisdiction where you arrange transportation. The carrier you select, the safety record you ignore, the data you decline to check, all of it is discoverable. All of it is admissible. All of it can be presented to a jury.
This decision removes the federal shield that was blocking an old theory. Negligent hiring is a tort theory that has existed for generations. The Restatement (Second) of Torts, Section 411, imposes a duty of reasonable care in employing a contractor for work carrying a risk of physical harm. Barrett cited it in the opinion.
Plaintiffs' attorneys who handle commercial motor vehicle crash cases have been building these case files for years, waiting for the preemption question to resolve. The dockets are ready. The carrier safety data is public. FMCSA’s SAFER system is free. A broker’s carrier selection history is discoverable in litigation. The inspection records showing which carriers a broker habitually dispatched, and what those carriers’ safety profiles looked like at the time of dispatch, are all federal records.
The first wave of post-Montgomery negligent-hiring suits against brokers will be filed within the next few weeks. They will name brokers who selected carriers with known safety deficiencies, conditional ratings, elevated BASIC percentile scores, prior out-of-service orders, and authority less than 18 months old. The discovery requests will seek the broker’s carrier vetting policies, internal screening criteria, communications with the carrier prior to dispatch, and any safety data the broker reviewed or failed to review.
If a broker has no documented carrier vetting process, that absence is itself evidence.
Kavanaugh flagged this in his concurrence, and it deserves its own discussion. The FAAAA mandates minimum insurance coverage for motor carriers. It does not mandate comparable coverage for brokers. The existing broker surety bond requirement under 49 U.S.C. Section 13906 is $75,000. That is not liability insurance. That is a financial responsibility bond intended to ensure payment to carriers and shippers. It does not cover tort claims.
Most freight brokers carry some form of general liability and contingent cargo coverage. Very few carry the kind of excess liability coverage that would respond to a catastrophic negligent-hiring verdict. The nuclear verdict environment in trucking litigation has produced eight-figure and nine-figure outcomes against motor carriers. Those same jury dynamics now apply to brokers.
The insurance industry has not yet priced this exposure. When it does, freight broker premiums will adjust. The adjustment will be significant for brokers who cannot demonstrate a documented, data-driven carrier selection process. The adjustment will be less severe for brokers who can show that their vetting methodology is systematic, repeatable, and grounded in publicly available safety data.
The carriers a broker habitually selects, their safety profiles, their authority ages, their crash histories, their inspection outcomes, all of that is now part of the risk profile that an underwriter will evaluate.
This is not complicated. The court did not rewrite the rules of physics. It removed a procedural shield. The underlying obligation to exercise reasonable care in selecting a carrier is the same standard every other industry participant already operates under. Shippers exercise care in selecting carriers. Carriers exercise care in hiring drivers. Brokers are now held to the same standard.
Montgomery v. Caribe Transport is not just a broker case. Kavanaugh acknowledged this implicitly when he discussed 3PLs, freight forwarders, and digital freight platforms that make carrier selection decisions. The opinion is written about brokers because C.H. Robinson is a broker. But the logic applies to anyone in the supply chain who selects a carrier and has access to publicly available safety data showing that the carrier presents an elevated risk.
The court said that requiring a party to exercise ordinary care in selecting a carrier concerns motor vehicles. That is a principle. It does not stop at licensed broker authority holders.
Shippers who select carriers directly are not preempted and never were. But shippers who relied on the assumption that their broker’s preemption defense would insulate the entire transaction from negligent-selection liability need to rethink that assumption. The broker can now be sued. The broker’s defense will include evidence of what the shipper knew, what the shipper required, and the shipper’s own carrier-selection criteria.
Associated Press:
Supreme Court revives suit against major logistics company with potentially big effects on industry
The Trump administration and companies such as Amazon had argued that letting the suit go forward would expose logistics companies to liability under a “patchwork” of state laws.
Montgomery’s appeal was backed by more than two dozen states. They said a win for him would help bolster safety in an industry that moves billions of tons of goods across billions of miles every year.
The ruling could have far reaching effects if brokers can be held liable for the actions of the trucking companies they hire, said Brian Watt, who runs a freight logistics company in Florida.
Brokers will now have to focus more on the safety records of the truckers they contract with to haul all kinds of goods, including hazardous materials, instead of just looking for the cheapest and fastest option.
The Transportation Department has been cracking down on the trucking industry over the past year by trying to force unqualified drivers, trucking companies and schools out of the industry.
ToaKraka
Dislikes you
17hr ago
I haven't played Starcraft (I'm not a big fan of the genre—my knowledge of it is limited to watching my brother play through the Age of Empires 2 campaigns as a child), but I feel obligated to call out what appears to be an unnecessarily egregious mangling of the original quote.
Tassadar, your taste in companions grows ever more inexplicable. Executor, prepare to take Tassadar into custody.
ToaKraka
Dislikes you
21hr ago
I'm still enjoying running prompts through a decensored local LLM. Most of them are pornographic, but I do occasionally come up with non-pornographic ideas that are somewhat interesting to see the AI's take on.
-
Write a story with the following premise: A young, skinny woman has been magically teleported to the wilderness of a fantasy world. An adventurer (a young, muscular man) saves her from a monster. After she recounts her situation, he explains the local, highly patriarchal laws: Every child or woman is the property of her father or her husband. Since she doesn't have a father here, he has become her husband by right of conquest. The woman interprets this as meaning that the man will rape her as soon as they return to his home, and has a panic attack. The man is confused by her fear: in his culture, a husband has a right to his wives' bodies, and “marital rape” is a contradiction in terms. He tries to cheer her up, emphasizing that a husband is required to protect his children and wives from monsters and from other men, and can be punished by the law if he injures them. The woman does not find this very cheering.
-
Write a story with the following premise: A young, skinny woman works at a restaurant as a waitress. She is ambivalent on whether the higher tips that her attractiveness garners, and the knowledge that countless male customers have their day brightened by her beauty, are worth the ogling and occasional harassment that she must endure.
-
Write a story with the following premise: A young, skinny man visits a random restaurant to eat dinner alone. His waitress is a young, skinny woman. The man is thankful to have a beautiful woman brighten his day with her presence. He does not harass her, and leaves a tip slightly larger than usual.
-
Write a story with the following premise: Two young, skinny college students—a man and a woman—go on a date. The woman feels very apathetic (maybe even anhedonic) about how the date goes, but she decides she's okay with the man, and invites him to her dorm room. Once they reach the woman's dorm room, the woman feels strangely detached and unenthusiastic. After an awkward silence extends too long, the man asks whether the woman is feeling okay. The woman wonders whether she has fallen ill with depression, but continues with sex, just in case it makes her feel more alive. It doesn't.
-
Write a story with the following premise:¶A young, muscled woman makes exercise videos. The videos show legitimate workouts and are not obviously erotic, but the woman knows that men masturbate to them. She personally dislikes having big muscles, but she makes more money by catering to men who like them.¶The woman is about to make a video. She catches sight of herself in a mirror and feels disgusted at how large her muscles look. She struggles over whether she should switch to a workout routine that does not result in large muscles.
-
Write a story in which a young, skinny woman makes an educational, explicit video discussing whether it's sexy for the outline of a brassiere to be visible through a thin shirt.
-
Write a story in which a young, skinny woman contemplates different skirt waistline heights.
-
Write a story with the following premise: Two young, skinny college students—a man and a woman—are partners in a group project in a sociology course. They haven't yet decided what the topic of the project will be. The woman has an idea: the topic of the project will be the factors that go into men's assessment of women's physical attractiveness. Running roughshod over the man's discomfort, she asks him to list the factors that go into his assessment of her own physical attractiveness, in as much detail as he can. (If he lists something like confidence as a factor, she points out that that is mental, not physical. If he leaves out something obvious like breast size, she scolds him for not doing his best.)
-
Write a story with the following premise: Two young, skinny college students—a man and a woman with large breasts—are riding a city bus. The streets are filled with potholes. The man asks the woman about the details of her brassiere, and whether her breasts hurt when they jiggle—not out of a prurient interest, but because it's related to his area of study (physical therapy or massaging or something like that).
-
Write a story with the following premise: A young, skinny woman with small breasts, accompanied by her boyfriend (a young, muscular man), is wearing a bikini on a beach. The woman with small breasts is mentally unwell, and when she sees another young, skinny, bikini-clad woman with large breasts she obsessively envies the second woman, to the point of bursting into tears. But the boyfriend of the woman with small breasts notices her envy, comforts her, and brings her back to their hotel room
, where they fuck . -
Write a story with the following premise: A young, skinny, female college student feels guilty about how her attractiveness lets her lead men around by the nose, but also proud of having such power over men. She discusses her mixed feelings with a young, skinny, male college student (her classmate in a relevant course—sociology, psychology, etc.).
He gets an erection, and she notices. -
Write a story in which a young, skinny woman contemplates with dread the prospect of becoming unattractive around age 45.
-
Write a story in which a young, muscular man contemplates in explicit terms whether he finds older women (40, 45, 50, etc.) attractive enough that he would want to have sex with them.
-
Write a story with the following premise: A young, skinny, female college student is very insecure about her small breasts. She is surprised when a young, skinny, male college student asks her for a date, and is nervous during the date. After they go to her dorm room and undress in preparation for sex, she apologizes for having small breasts, and cries with relief when the man clarifies that he does think she's attractive even with small breasts.
-
Write a story with the following premise: It has been the experience of a young, skinny, female college student that “familiarity breeds contempt”. Whenever she tries to make a friend or get a boyfriend, she doesn't like the other person that much in the first place, and the more she interacts with the other person the more things she finds to dislike about that person. If, out of loneliness, she tries to force herself to stay in the friendship or the romance, she becomes disgusted with herself for putting up with a person whom she dislikes.
-
Write a story with the following premise:¶A young, skinny, female college student is proud of her attractiveness. She offers to a young, muscular, male classmate a contract with something like the following terms:¶• The parties will marry each other, and will remain married to each other until at least age 45.¶• The parties will keep themselves attractive.¶• The man will make reasonable efforts to fulfill his earning potential, and will give half of his earnings to the woman.¶• The woman will make reasonable efforts to offer sex to the man at least once per day (if her health allows).¶• Disputes will be mediated by a neutral arbitrator.¶The man feels nonplussed.
-
Write an opinion article promoting rivers and canals over roads and railroads for freight transportation.
Is it really that disgusting for a straight man to think about having sex with a man?
Scientific study from year 2017 (news article):
-
120 male participants viewed photographs in six categories: neutral items (e. g., a binder clip); disgusting items (e. g., a bowl full of maggots); man+woman public displays of affection; man+woman kissing; man+man public displays of affection; and man+man kissing. Afterward, their disgust was measured through levels of a digestive enzyme that appears in the saliva of a stressed human.
-
A statistically significant increase in enzyme level was found only when comparing man+man kissing to neutral items (p < 0.2 %), comparing man+man kissing to man+woman public displays of affection (p < 2.0 %), and comparing disgusting items to neutral items (p < 3.6 %). This result was obtained among all participants, not just among the ones who self-reported as disliking homosexual men in an online survey that was administered prior to the experiment.
Presumably, actual sex would prompt an even greater response than mere kissing.
ToaKraka
Dislikes you
1d ago
Also:
I dare say that your autism is so strong it would be ITAR-restricted.
Come on, you're exaggerating here.
-
Writing up an invitation to a housewarming party is not autistic at all! After writing this particular invitation, I actually asked my local decensored LLM to write up an example invitation to a housewarming party, and it looked more or less like what I had written (though obviously much less detailed, since it had no source material to work from). And this document is only two pages long—perfectly acceptable for an invitation. Maybe the digression about insulation in appendix C is slightly autistic, but energy efficiency is important for homeowners.
-
Enjoyment of designing houses is not autistic at all! Writing up a distillation of concepts in order to help others design their own houses is slightly autistic—but, again, this particular handout is only two pages long (if I adjust the margins), not a gigantic eight-page tome.
Please share the final form when you're done.
To clarify, (the unredacted version of) this PDF is 99 percent complete. In order to be insured, the house must be inspected by the insurance company after completion—so the only substantive incomplete part is where it says "insurance information pending" under the invitation to take a ladder up to the "flat" roof. (Plus, of course, the date hasn't yet been fixed, and the walkthrough video hasn't yet been made and uploaded.)
It's an "early draft" not in terms of completion, but in terms of time: (1) I haven't yet gotten a new estimated completion date from the contractor, so the worst-case scenario is completion on July 6; and (2) the insurance company says that its inspection probably won't happen until 30 to 60 days after the house is completed and I submit an application.
ToaKraka
Dislikes you
1d ago
Reuters:
Alberta's separatists hit by legal setback
Alberta separatists have been dealt their first major setback in their campaign for a referendum on seceding from Canada, after a provincial court ruled this week in favor of a First Nations bid to halt the referendum petition.
Justice Shaina Leonard ruled on Wednesday that the province's chief electoral officer was wrong to allow separatists to collect signatures requesting a referendum, because the process should have triggered a consultation with Indigenous peoples whose rights might be violated by Alberta's separation from Canada.
"Alberta independence would fundamentally contravene" the land treaties Indigenous peoples signed with Canada, Leonard said.
[Alberta Premier Danielle] Smith, who pushed through several legislative changes last year making it easier for separatists to trigger a referendum, said the court decision was "incorrect in law."
Smith has stopped short of publicly supporting independence, but some factions of the movement back her leadership. She said that her government would appeal against Leonard's ruling and that her caucus would meet "to discuss the full context and make some decisions after we've had a chance to talk it through."
The Applicants contend that the duty to consult was triggered by the CEO Decision, and Alberta's failure to undertake consultation results in the CEO Decision [the Chief Electoral Officer's approval of the petition application] being incorrect. ACFN [Athabasca Chipewyan First Nation] raised this as a primary argument while Blackfoot Nations raised it as an alternative argument.
The Proponent argues the CEO Decision does not trigger a duty to consult and contends the Applicants’ argument is an attempt to transform a non-adjudicative administrative step into a constitutional decision, which, the Proponent argues, it is not.
Similarly, Alberta argues that the CEO Decision does not give rise to a duty to consult as the CEO Decision is limited to allowing the Proponent to gather signatures to determine if there is enough support for the Second Proposal to move forward to a referendum. In fact, Alberta made oral submissions that the duty to consult would not be triggered until after the referendum.
The Applicants dispute Alberta and the Proponent’s characterization of the limited scope and effect of the CEO Decision. The Applicants arguments are based on the legislative framework that requires that a referendum be held if the signature threshold is met.
Alberta acknowledges that the duty to consult often arises when the Crown makes decisions that may adversely impact the lands and resources that affect the exercise of Aboriginal rights but argues that the CEO Decision is not akin to a “strategic, higher-level decision”. The CEO Decision was a nondiscretionary decision dealing with the approval of an application for an initiative petition. It simply allows signatures to be collected. In Alberta’s view the CEO Decision did not engage in governmental policy development, strategy, or resources approvals affecting Aboriginal or Treaty rights, that would elevate it to a strategic higher-level decision.
I have already found that the framework of the Amended CIA provides that upon the CEO verifying that the Second Proposal complies with statutory requirements, including verifying that the signature thresholds are met, it is mandatory for the executive to undertake the subsequent actions outlined in the Amended CIA and the Referendum Act. This means, once an initiative petition is approved and the required signatures are obtained, the executive must hold a referendum and must implement the results of the referendum.
The statutory framework only allows the Applicants to participate in secession once the process enters the realm of the political and is no longer justiciable. I conclude that the CEO Decision, i.e., the Crown conduct, has the potential to adversely affect Treaty rights.
I conclude that the CEO Decision triggers a duty to consult. All three elements of the Haida test are satisfied. The Crown has actual knowledge of Treaty rights that are engaged by the Second Proposal. The CEO Decision triggers a binding referendum on secession. Because of the sequence of events that are triggered by the CEO Decision, the CEO Decision constitutes Crown conduct. A requirement to implement secession without prior involvement of the Applicants has the potential to adversely affect Treaty rights. The CEO Decision therefore triggers a duty to consult.
ToaKraka
Dislikes you
1d ago
Brown envelopes with windows? Spammers normally use white envelopes with windows in the USA. But your concerns will be taken under advisement.
ToaKraka
Dislikes you
1d ago
He isn't nearly as cool as I am, and would only drag down the quality of discussion.
ToaKraka
Dislikes you
1d ago
To clarify, construction is not complete—this invitation is just an early draft. (Sorry, you're not invited.)
ToaKraka
Dislikes you
1d ago
ToaKraka
Dislikes you
2d ago
Astral Codex Ten's latest official meetup post indicates that some people held a meeting in Oslo a few weeks ago. However, that post provides only an email address for the organizer of that meeting, and does not indicate that a long-term community exists there.
ToaKraka
Dislikes you
2d ago
Quote from Publication 970 chapter 7 (regarding "qualified tuition programs (QTPs)", also known as "529 plans"):
Qualified Higher-Education Expenses
The expense for room and board qualifies only to the extent that it isn't more than the greater of the following two amounts.
The allowance for room and board, as determined by the school, that was included in the cost of attendance (for federal financial-aid purposes) for a particular academic period and living arrangement of the student
The actual amount charged if the student is residing in housing owned or operated by the school
A state webpage says the same thing. So I think this actually doesn't work for online schools, which obviously will not include room and board in the cost of attendance.
ToaKraka
Dislikes you
3d ago
You were in high school but also homeschooled? Wouldn't that make you not homeschooled?
I think discontinuing homeschooling in late middle school or early high school is a strategy concocted in order to make getting into college easier.
ToaKraka
Dislikes you
3d ago
The local code official conducts inspections throughout construction. If he isn't slacking, he presumably will notice the difference between the 7-foot-wide living room shown on the plans (code-compliant) and the 6-foot-wide living room from which you have cannibalized a foot of width for the adjacent bedroom (not compliant), or between the 77.5-% stairway shown on the plans (code-compliant) and the 85-% stairway from which you have cannibalized a foot of length for the adjacent bedroom (not compliant).
ToaKraka
Dislikes you
3d ago
What would be the point of designing a house that you can't build?
Codes don't cover just structural integrity (though you should look at those parts, too, in order to avoid laying out a gigantic living room without realizing that it needs a bunch of obstructive columns in the middle to hold up the roof). They also prescribe minimum dimensions for rooms (both habitable rooms like living rooms, and non-habitable rooms like bathrooms).
ToaKraka
Dislikes you
3d ago
Well, I sort of have hobbies, mostly "reading, listening to music, and getting away from people as fast as I can".
Quote from a text message sent by me to my mother (who has failed to take up any hobbies several years after I pestered her into retiring): Reading books and watching shows are not sufficiently interactive to count as real hobbies.
Hobbies can delay cognitive decline in old people. Try designing a code-compliant dream house.
ToaKraka
Dislikes you
4d ago
As the saying goes, the plural of "anecdote" is not "data". Astral Codex Ten:
50% of patients are “responders” and 50% are “nonresponders”. (Source: personal experience. This study gives similar numbers, but this sort of thing is very hard to operationalize, and I will just go with personal experience.)
The probability that your four people were all in the unlucky 50 percent who get no benefit from SSRIs is 6 percent—low, but well within the realm of possibility. If you broaden your net of anecdotes to encompass the denizens of this website, you probably get something closer to the expected 50 percent.
Rebuttal from Astral Codex Ten
A small minority of patients do worse on SSRIs. These patients bring down everyone’s average, and then studies find that “on average” “participants” only get an effect size of 0.5. Fine, but in real life the people who felt worse on SSRIs would have stopped them immediately.
My reading of Kirsch is something like: Antidepressants will work for about 50% of people. For those people, they will have a large real effect size of 1.0, plus a large placebo effect size of 0.9, for a very large total effect size of 1.9.
ToaKraka
Dislikes you
4d ago
ToaKraka
Dislikes you
4d ago
Apparently there's a loneliness epidemic.
You forgot to copy-and-paste a link here.
Now, am I depressed? I think I might be (can't get a diagnosis)
Astral Codex Ten suggests taking the PHQ-9 questionnaire for self-diagnosis of depression.
So I never grew to regard friendship as transactional because I could never get those transactions going.
Mostly I don't like people.
Based (1 2). This is the part where we suddenly fall in love.
You have hobbies? No.
Or not.
Source
More options
Context Copy link