Hacker Newsnew | past | comments | ask | show | jobs | submit | oredbored's commentslogin

Do we know for sure that they trained on data from libgen etc? It's such a powerful source of information you'd assume they must have, although they would never admit it. There must be a way to test if they have, via enquiring about some niche information only found in certain books.


It is apparently widely suspected that a certain "Books2" dataset mentioned by OpenAI is basically just LibGen:

https://blusharkmedia.medium.com/the-ongoing-battle-against-...

https://techhq.com/2023/09/can-libgen-shadow-library-survive...

https://www.twitter.com/theshawwn/status/1320282152689336320

https://qz.com/openai-books-piracy-microsoft-meta-google-cha...

https://qz.com/shadow-libraries-are-at-the-heart-of-the-moun...

https://goodereader.com/blog/e-book-news/authors-file-lawsui...

When asked about whether this was true, they refused to answer based on confidentiality concerns, then said they had deleted all copies of the dataset, stopped using it, and no longer employed the individuals that compiled it:

https://www.businessinsider.com/openai-destroyed-ai-training...

We do know for a fact that the (non-OpenAI-controlled) "Books3" dataset is just "all of bibliotik":

https://www.twitter.com/theshawwn/status/1320282149329784833

https://github.com/soskek/bookcorpus/issues/27

And we also apparently know for a fact that this was included in the datasets used to train LLAMA:

https://en.wikipedia.org/wiki/The_Pile_(dataset)

https://aicopyright.substack.com/p/the-books-used-to-train-l...

https://aicopyright.substack.com/p/has-your-book-been-used-t...


Thanks a lot for all the links. Fascinating stuff.


Lol... it was known extremely early on that fomite transmission was practically unheard of. People were hand sanitizing for years after this. God knows why. The FDA still hasn't even declared hand sanitizer safe to use by the way.


Dark mode is slightly harder to read, but it's preferable to having 4 screens blearing bright white light at me all day.


Interesting. Do you know which crypto they will be mining? Is it Chia?


It would be pointless to put Chia mining operation next to a hydro power plant, the entire point is that they use very little electricity.

They haven't even admitted it's crypto.

All that's really known is that it's a 10MW "datacenter", hooked directly into the power plant, in a part of the country without a commercial fibre backbone, and will allegedly be running workloads that can be turned off depending on power demand.

Reading between the lines, that's a crypto mining operation.


Ahh, you had datacenter in quotation marks, I didn't catch that.


You mentioned sports betting funds - Just wondering if you could elaborate further on the use case here?


Collecting and consolidating data + filling in the forms across different sites to place bets



>Of the 198 patients with CVD, 177 (89%) were adherent. Major cardiac events judged to be recurrent disease totaled one stroke in the adherent cardiovascular participants—a recurrent event rate of .6%, significantly less than reported by other studies of plant-based nutrition therapy. Thirteen of 21 (62%) nonadherent participants experienced adverse events.

So... heart disease can be attributed almost entirely to eating meat? That result seems too good to be true.


This is nearly pseudoscience..

Indians have one of the highest rates of heart attacks even though a significant population is purely vegetarian.

I think science has to be focused on biology and science, not social surveys and asking questions to a small bunch of uncontrolled strangers.


> Indians have one of the highest rates of heart attacks even though a significant population is purely vegetarian.

Ghee could be part of the cause of that, but I have no idea what to think of saturated fat any more.


You're right about Ghee. It is off the charts high in oxidized cholesterol. It's like butter on crack.


It wasn't a survey. Imagine calling something pseudo science without even reading the abstract.


I did read the abstract.

Most such research simply monitor the participants' diets through a questionnaire or some indirect mechanism. It's not a scientific, reproducible approach.

I am making a meta-point that such research has started to affect more and more of what we think of as science which is extremely disturbing.

The harder scientific work would be to analyze the effect of specific foods with how the heart works over a much longer epriod of time. Granted that there isn't a way to track or pursue this line of inquiry yet, but taking shortcuts isn't science IMO.


What are the likely dietary habits of people that have time for academic surveys

Just college kids and mechanical turk penny vacuums

So probably a little predictable


Or not...

"Participants

This report reviews the outcomes of 198 consecutive nonsmoking patients with multiple comorbidities of hyperlipidemia (n=161), hypertension (n=60), and diabetes (n=23) who voluntarily asked for counseling in plantbased nutrition for disease treatment. These self-selected participants requested consultation after learning about the program through the Internet, the media, prior scientific publications, the senior author’s book (CBE Jr), other authors’ supportive comments, or word of mouth.2,13 A preliminary 25- to 30-minute telephone conversation established disease presentation and severity by eliciting reports of symptoms, history of MI, stress test and angiogram results, interventions undertaken, family history, lipid profile, and the presence of comorbid chronic conditions. In these calls, we outlined the program, established rapport, and documented the need for additional patient information. The Cleveland Clinic Institutional Review Board determined that these were acceptable outcome measurements to evaluate the nutrition program."

"Baseline characteristics of participants are shown in Table 1. (Two patients from the original group of 200 were lost to follow-up.) The remaining 198 participants for whom data were available had CVD, were mostly men (91%), averaged 62.9 years of age, and were followed for an average of 44.2 months (3.7 years). "


> mechanical turk penny vacuums

Wait, what's that?


Mechanical Turk is an (Amazon) crowd-sourcing site where one can post small jobs (such as surveys) to be completed for a small amount of money (often only $0.05-0.10). I assume "penny vaccums" here is a metaphor for people who complete a large number of tasks for the money.

https://www.mturk.com/


OK, thank you.


So... heart disease can be attributed almost entirely to eating meat? That result seems too good to be true.

If it's the study I've seen, the improvement can be largely attributed to a tremendous drop in both calories and fat. Instead of their idea of a balanced diet being a bacon double cheeseburger in each hand with a gallon of ice cream for desert, it became rice and legumes and veggies. They shot for keeping fats to no more than 10 percent. They invited participants to dinners so they could experience good tasting, healthy gourmet meals and not feel like this was about deprivation.

Participants dropped tons of weight without trying and without having to try to control themselves. They could eat all they wanted every day and not get enough calories to remain clinically obese.


There is no correlation between fats and CVD. That's a very old belief (70s) that is debunked. Read this please: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5793267/

Downvoters: Please just read the paper.


From your link:

SFA is now believed to play a much smaller role in CHD than was previously believed while that of sugar, whole grains, and cereal fiber have moved in the opposite direction.

Which is not the same thing as no correlation.

I'm sure it's vastly more complicated than just fat content, but the stuff I've seen indicated that one study of a vegetarian diet as post cardiac therapy explicitly set a goal of keeping fat content down to 10%. Prior to being put on a vegetarian diet, the typical diet of participants was much, much fattier. Participants were just blown away at being able to eat all they wanted for the first time in their lives and lose weight without even trying.

My personal experience: I dropped several dress sizes without intending to by completely ignoring both calories and fat percentage and focusing on consuming nutritionally dense foods to redress known nutrient deficiencies.

So I have absolutely no doubt that it's vastly more complicated. But I tend to get a lot of flak for talking about what I think about diet, the gut biome, medical stuff, etc. So I really wasn't looking to get into that.

I was only looking to add a little more information about such studies. The main takeaway as I understand it: they stopped being obese. There are people who eat meat who are not obese.


> Participants were just blown away at being able to eat all they wanted for the first time in their lives and lose weight without even trying.

Anecdotal, but this is me at the moment on keto. For the first time in my life (I have been obese since 5) I am not ravenously hungry all the time, to the point where I can just eat one meal a day without much struggle. I eat as much as I want now, the only difference is I don't want much at all. Oh, and I can't eat many carbs.


I can relate. As my nutritional status improved, I gradually ate less and less. I just stopped being hungry all the time.


How do we explain the vegans and the vegetarians who still get plenty of heart disease? If veganism were a silver bullet, I'd be the first one to jump on that bandwagon, today.


Because studies like this are sheer bullshit. There is no one "plant-based diet". Eating fat heavy (avacado, nuts, seeds, coconut, acai berry, olives etc...) and carb heavy (rice, fruits) diets and whether you eat few protein rich plant sourced (basically only beans) will have very different macronutrient, micronutrient profile and will have very different effect on human body. Same goes for animal-based diets. Moreover, people who were able to adhere to this study might have just been more disciplined and healthier. This study does not control other important risk factors such as alcohol, regular exercise etc. I find it very hard to believe we can really derive any useful information from "we forced people to eat plants".

As a meat-eater I think the most stupid strawmen vegans hit me with is the belief that I only eat meat. I think very few people advocate "only eating meat" (e.g. recent 0carb hype, which I think is clearly pseudoscience). You can still eat plants but supplement your diet with meat to get more protein and healthy fats. You can still get 75% of your calories from plants but still eat meat. Also, there will be a huge difference between highly processed meat (bacon, salami etc) and unprocessed, grass-fed meat. Studies like this simply ignore all these factors.


It means that there are other factors besides diet. Here's a relevant study https://www.ncbi.nlm.nih.gov/pubmed/707372


Because the health factors that cause heart attacks aren't exclusively attributable to diet.


Very few vegans eat whole foods, which is what was used in the study.


Or people who weren't disciplined enough to stick/adhere to the veggie diet, likely also had other parts of their life with less than healthy discipline.


The compliance rate was 89%, which is higher than that of statin drug therapy. If discipline was a factor, it was a very small one.


Well thats convinced me to switch my lunch plans for today.


No. That kind of heart disease can be cleared by not eating meat.

many heart diseases have nothing to do with blood chemistry.

source: i have valves turning to shit and never had a problem with blockages or high cholesterol.


cronometer.com is an excellent and complete nutritional calculator. it calculates your individual amino acid requirements so you can get your complete protein for the day. your liver actually stores amino acids to form complete proteins so it doesn't matter if you eat incomplete protein one meal as long as your overall diet is balanced. complete protein is just a term for the amino acid makeup of human tissue, so technically the only complete protein out there would come from cannibalism.


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: