ouraf

Isn't that more or less how GPT-4 works? multiple "expert" LLMs giving input depending on the context?[0]

[0]https://the-decoder.com/gpt-4-architecture-datasets-costs-an...

the biggest issue is if you have too many specialists and spin a lot of them to reply to the same query and after that discard the less optimal answers.

Your answer quality might improve, but the computing costs could skyrocket without some smart filtering and distribution before you reach any LLM

I've been seeing a bit of buzz about "grounded" models. Other than setting temperature to Zero and adding better context matching (with a vector database and embeddings, for example), what other changes in methodology or implementation makes it better?

It is as bad in medicine, if not worse. Recently a very influential paper about alzheimer was pulled[1] because of almost everything you said, but for biology instead: data fabrication, overestimating numbers, etc.

There's also a case related with cloning (which includes stem cell research) and massive fraud in South Korea 20 or so years ago.[2]

It's a complicated matter: research is a beacon of light on top of a mountain of failed experiments. And not even the government will fund a mountain of failures if your shining discovery at the end can't bring returns to society (or shareholders) in greater value than what you took.

And the pharmaceutical industry never settled for small results. And, unlike ML, it's much harder to test things as an outsider (you can hate the hype, but openAI, hugging face and all these "open" language models makes it much easier to learn, test, tweak and improve things without a master's degree in data engineering.).

It's a grimy area to research even as a pastime. There's a lot of recorded practices both in research and production that makes you question how low can a person go for money

1: https://www.science.org/content/article/potential-fabricatio...

2: https://youtu.be/ett_8wLJ87U

Isn't discord attachment security really lax? In a "anyone with an image or file link can download it regardless of connection to the server or role" level of lax?

You'd need to be really bold or poorly advised to consider that role

Linux Mint on a m2 drive will probably boot and open anything really fast and keep a comfortable experience similar to windows desktop

OpenAI + PineconeDB -> Langchain -> Quivr to get a decent UI

right now i'm trying to prove to the company i work that you can turn the endless rules and daily information stream into something people can find easily using sematic search (vector DB) and an AI to summarize categorize, generate embeddings and auto update the DB, while also interpreting the content of the search result and giving a more digestible answer and a source link if the end user needs more information.

Keyword based search is a pain when different products have similar names or the internal search doesn't filter words like "of" and "and" from the keyword search.

After that, if the tests work and i can push the idea forward, it'll probably be OpenAI + a local vector DB (Chroma?) + custom made search page for internal usage

there's a modicum of irony in posting this on an Ycombinator-enabled site. But at least they give enough leeway to keep this post here, so that's a good thing

Can anyone do a "explain lime I'm five" summary of the problem and how the solution would work when typescript is converted down to JavaScript again so the interpreter takes it? Looking from the top it seems like it would add more complexity (and potentially new bugs or exceptions) than it could make things faster and safer.

IF typescript was 100% its own thing and not built on top of JS, it would make complete sense to do it, but with a js engine below it...

Nothing gets close to GPT 3.5 and 4 if you need to send and receive prompts in languages other than English.

Sadly, that niche is still better served by openAI.

Ooh, i think i remember this one: It had the cosmic coincidence of one of its researchers being called Timothy MOSCA.

https://www.smithsonianmag.com/smart-news/see-the-first-comp...

good to see it got some more attention.

adult

It can't get any worse than twitter, can it?

will not work. Companies just can't allow dissenting voices they cant moderate. It's too dangerous

Some years ago there was a browser extension called "dissenter" that was exactly what you suggested and a bit more. It was shot down. Hard.

An old professor that specialized in helping students prepare to university or civil servant tests used to say that if you want to understand a subject, you need to do two things about it:

- Make some kind of logical scheme to help with disambiguation (things like which legal process comes first, of the Socrates-Plato-Aristotle order of philosophers and ideas)

- Write the long form of the subject everything by hand. Pen on paper. If it's a law or bill, write it fully, if it's a scientific concept, write every part of it that is related to the subjects that can appear on the test.

It's slow, tiresome and a complete nightmare for students with anxiety problems (trust me, i know), but as you write, you very slowly burns the knowledge that fills your logical scheme into your brain.

That's what the people that succeed in hard tests do to dominate nebulous topics. I don't know if it translates to actual mastery in different areas, but it's a very strong studying technique

Not really. Mainstream media publishes something about AI at least once a week. It might not have adoption in the everyday life, just like most tools are known, but only fully explored by their niche.

There's a lot of overhype, A LOT. And many startups are reinventing the wheel in a more expensive way just to ride the wave, but unlike crypto, LLMs are tools that can solve more "dev/user" level problems, unlike crypto, that was supposed to solve a "global financial system" problem, and since that's too hard to grasp, people used it as investments, because it's closer to their real lives and needs.

Well, the man did say twitter would win the fight against the bots or die trying.

Both results are positive IMO

i misread it as "soon-to-be-dead" and almost had a heart attack.

Bad eyesight aside, not as a learning material for you, but purchase some fairytale books or illustrated stories. It's good to make you and your child closer over a bedtime story or admiring pictures until they can read by themselves.

If anything major comes out of this, is probably EVEN MORE prompts and popups asking for permission to use your data. even with GDPR, data collection and sales never stopped, it just made things more annoying by transforming every webpage into a granular term of service to continue doing the same.

It isn't even turned off by default. Many sites just give you an "i accept" button or even if you want to manage the preferences, the "accept all choices" button is where the "confirm my choices" should be.

Bigger companies will just append this to their TOS and push it down the customer's throat. That if MS doesn't settle out of court and the case gets thrown together with any major oppositon to the data mining

My idea is that the end user is an employee, not a customer of the company. Since they have a general understanding and the questions won't be as broad, i hope one of the open models with a less strict license is enough for the task. Stability AI released a model with creative commons CC BY license (allows commercial use with attribuition, although Creative Commons is a weird license for a gigantic database) and there are some "LLaMA" clones (open llama, redpanda, vicuna, etc.) that i need to look deeper into to understand how it was done and how it didn't inherit the original's license

I think Reddit's executives are implementing these changes to create amazing revenue forecasts out of thin air and sell the company for the next sucker. Maybe Meta or Microsoft will buy Reddit out and ruin it further.

So let's say that just hypothetically, I train an image generating model with every available work drawn or designed by Akira Toriyama, then I use this model to make a Journey to the West comedy manga.

The government won't arrest me for that, but can private corporations still sue me for using their work to train the model?

Ok, i wasn't specific enough, that's on me.

My previous phone was a Xiaomi. Part of the reason they're cheaper than competitors with similar hardware is because that thing is filled to the brim with useless ads, extra tracking way beyond just crash analytics for the manufacturer and the regular google stuff and even some useless "phone booster" app embedded in the settings.

After waiting the two weeks so Xiaomi would allow me to unlock the bootloader (which is bullshit, by the way), i installed LineageOS.

The problem is that the regular updates with security patches sometimes would cause a boot loop. Some times is something as simple as a system file becoming read only, sometimes no thread on XDA developers could guide me to the right solution, and then i just do a factory reset and start over.

But aside from flashing lineage and an ad blocker, my usage is the same as anyone else: answer calls, use messaging apps, browse the web, play games and edit some documents when i'm far away from my PC.

My current phone is a motorola, though, and while i think the official ROM is pretty disappointing (too many pop ups nagging me to sign up to motorola or check motorola apps), as long as i can flash magisk and an ad blocker, i'm a happy person.

Well, guess it won't work for me. Rooted phone and prone to boot loops and hard resets whenever i get curious with some stuff. Every authenticator stored in the phone can't be recovered when you install a new ROM (google tries to send the authorization pop up to the previous version of the phone and, of course, it'll never get there)

Guess that's the price i pay for using the hardware in a slightly different way than what google and the manufacturer wanted me to.

The company will proabaly scrap the brand and rebuild it around the Huffington Post. not just "select roles", like change the facade and keep the guts. They bought it in 2020 and the HuffPo brand itself is way less maligned than buzzfeed.

It's also a chance of re-hiring the best staff for cheaper wages.

They took that decision after windows 8 and all the push for windows phones failed to compete with android and iOS.

Now they just focus hard on offering software for people who pays and ads for people who doesn't. They can't just dictate the technology trends and styles like before (Metro lost space to Material, so they can't even influence the look & feel of computers like before).

One of the few parts where Microsoft is still focusing on being cutting edge and force their proprietary ecosystem is XBox. They're still trying to be unique and stand out against Sony and Nintendo

I wonder how much tinkering you need to make an enterprise version that removes all the time sink and telemetry.

not everyone that deals with sensible data has IT education beyond windows and office, and you can't demand every bank clerk and hospital receptionist in the world use arch Linux for safety reasons.

On the other hand, these click bait articles in the main UI are timesinks that could reduce productivity in a company

Honestly, there's too much garbage in the code dump they made.

Maybe an UML graph or even a presentation or written guide on how they measure and apply each weigh or group policy would make it easier to have some solid take on how it works

Going a little more "end user needs vs new tech offers", the intent is on the right place, but the output isn't helping as much as normal programmatic tools

At least for me, what would make this a killer app would be the ability of reading a document or pdf or big text dump and 1: identify "possible fields" (first name, date of birth), "probable fields" (middle name or other fields that are part of data set but doesn't appear in every line) and "probable junk data" (page numbers, page headers, useless pdf padding"

2: allow selection or tuning of these fields to generate regex to catch or remove only the data related to the parsed fields.

I THINK there's something done with pandas (pandoc?)that can help tearing a document apart and getting fields or basic doc structure, but AI would need to take it from here and present it in a clear, concise and optionally explained way so a busy office worker could just copy the regex filter in a spreadsheet formula or program function

I hate to be devil's advocate, but archive.org has a problem with curation of their free library. Some stuff there breaks some terms of service of the original owners, like the leaked PS2 SDK https://archive.org/details/ps2_sdks, old and irrelevant as it may be today. Or cracked games that are still availiable online https://archive.org/details/gtasa210apk

I wish them to stay providing service, but maybe a quick checkup about "why is this stored here" or "isn't any other site more capable or well equipped to store and defend the right to access this content" would help giving it more of a defense against copyright holders

maybe they can use the expenses maintaining docker images of projects that are beneficial to a broader FOSS user community to pay less taxes or something, or brag about their phylantropy on accounting reports.

I don't believe the revenue from these avenues would save or kill the company anyway, so better make a lemonade of these lemons, right?