Browsers with AI
- Inicie sesión ou rexístrese para enviar comentarios
I have been playing with small amount of money in the stock market for a few weeks. I watch Bloomberg and read YahooFinance.com. The impression is that a vast amount of money is being invested and huge $ values are being grown right now all centered on AI. Software companies, companies with big data centers, energy companies to power data centers, CPU and other hardware makers have been growing in value incredibly fast. Lots of cash flowing that way.
Recently I read a bit about a browser called Perplexity Comet with AI built in. I looked at a few sites like https://usefulai.com/tools/ai-browsers and see a trend happening like https://www.operaneon.com/ which I read is an Opera browser with AI built in that will cost $20 a month to use.
Will the free software community be excluded from use and development of AI. The financial cost to develop, power and maintain the data center required to perform complicated tasks seems like it might be hurdle FSF et al won't be able to get over. So it would be a long time before Abrowser would have AI built in. I can't imagine wanting AI in my browser anyway, but that's because I am old and like things the way they used to be. Rotary dial phones on party lines, black and white TVs with only 6 or 7 channels available and newfangled transistor radios blasting the hits on AM radio, yeah, those were the days.
In addition to browsers with AI, the Ubuntu based Archon OS is out there. I think it likely talented developers that love Trisquel might be forced to use AI powered development environments at work to keep up now days.
Well, just trolling. Would be interested in anybody's thoughts about this AI boom/paradigm shift happening right now.
Right now, AI is scary as heck. Apps like Sora 2 can make incredibly real looking fake news in seconds. Perhaps cyber criminals, many of them state sponsored can make zero day exploits just as fast. And of course really great things can be done using it as well. I read some people are creating their own AI doctor to chat with and getting very accurate diagnosis. And on and on.
>"I read some people are creating their own AI doctor to chat with and getting very accurate diagnosis. And on and on."
I heard on TheLinuxCast podcast that some people are making their own AI girlfriends or boyfriends and falling in love with them. It's a strange new world we are living in.
I wonder how soon we will see people starting to marry their AI creations?
This AI crap you mentioned andy is more creepy then my current avatar of Gin Ichimaru of Bleach.
>"more creepy then my current avatar of Gin Ichimaru of Bleach"
Well, to be fair, nothing is more creepy than an avatar of Gin Ichimaru of Bleach. Except maybe an avatar of Hamlet holding a skull.
Or that trash anime Death Note... that anime is hella creepy.
If you wonder why I hate it, half of it is the fanbase and their hatred towards light and the fact they ignore that Ryuk is the real evil. the other half is that it is depressing how it ends.
The fanbase though I swear is as toxic as fascists. They all are glad he dies and ignore the actual monster that I mentioned above.
If your curious why I say he is the real evil, I can tell you in another thread, etc...
Oh and on topic, the other thing that its as creepy as, is macbeth. I have similar feelings of Macbeth to Death Note too btw.
Of course, we all hate Macbeth, it's only natural.
>Macbeth does murder sleep: the innocent sleep,
Sleep that knits up the ravelled sleeve of care,
The death of each day's life, sore labor's bath,
Balm of hurt minds, great nature's second course,
Chief nourisher in life's feast.
Murdering sleep is very rude. Fascist, I might say.
The three witchesin Macbeth remind me of Ryuk from death note... they find entertainment and delight in a good soul being tainted by evil and the process of their destruction.
Those who tempt people into darkness and delight in their suffering afterwards are pure evil.
Thus, why I hate Ryuk and why I hate the death note fanbase. They ignore the real pure evil and focus on one of the victims.
Light was corrupted by the power, yes he was evil, but he was also a victim.
The demons are always the worst:
Aka, Ryuk and the three witches
The three witches tempted Macbeth and that's how he started to fall in macbeth.
Similar situations that irritate me. One just has the worst fanbase alive...
>"One just has the worst fanbase alive..."
Worse than the systemd and Rust fan bases? Hard to believe my friend, hard to believe.
I've probably accumulated thousands of systemd related downvotes in various forums over the years, and I'm a person who frequently uses systemd (with Trisquel) and only points out some of its flaws and shortcomings. systemd fans are relentless - you either accept systemd entirely with no criticism whatsoever, or you are their avowed enemy.
The Rust fans seems to follow a similar playbook. No criticism, even legitimate criticism, will be tolerated.
The ruthless anti-systemd mob was far worse. It is the one that caused the resignation of developers, typically within the Debian project, as https://en.wikipedia.org/wiki/Systemd#History explains:
Between 2013 and 2014, the Debian Technical Committee engaged in a widely-publicized debate on the mailing list about which init system to use as the default in Debian 8 before settling on systemd. Soon after, Debian developer Joey Hess, Technical Committee members Russ Allbery and Ian Jackson, and systemd package maintainer Tollef Fog Heen resigned from their positions, citing the extraordinary levels of stress caused by disputes on systemd integration within the Debian and FOSS community that rendered regular maintenance virtually impossible.
And most of that mob did not rely on pointing out actual systemd's flaws and shortcomings. It was some kind of religious fanaticism, where every blow is allowed. They were setting up sites with huge lists of lies regarding systemd and conspiracy theories. There were articles literally using the religious vocab using terms such as the "apocalypse", the "shism", the "divide" or the "exodus" in three different articles by Paul Venezia I linked to there: https://trisquel.info/forum/systemd-free-trisquel-variant#comment-149916
Ian Jackson resigned due to the inappropriate manipulations and drama caused by the pro-systemd side in that debate, and I think Russ Allbery did too if I recall correctly.
And I don't recall there being an "anti-systemd" side - there was one side that wanted Debian to embrace systemd only, and one side that wanted to continue Debian's policy of officially supporting multiple init systems, which would have included systemd.
Regardless, that Debian debate has nothing to do with me, and I've been a frequent systemd user myself for many years. I was just noting that when I do critique systemd for some of its more obvious issues that I've drawn an enormous amount of animosity from systemd's rabid fanbase. Which, frankly, I find extremely weird - why does an init system have a rabid fanbase spread out across the web and on so many different forums? Very strange. If it made any sense to accuse someone, such as RedHat, of astroturfing the issue, I would. But that doesn't make any sense - what would be the financial interest for RedHat in doing such a thing?
Ian Jackson resigned due to the inappropriate manipulations and drama caused by the pro-systemd side in that debate, and I think Russ Allbery did too if I recall correctly.
Ian Jackson was indeed against systemd. Russ Allbery on the other hand voted for it: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=727708#6729
I don't recall there being an "anti-systemd" side
- https://ihatesystemd.com
- https://nosystemd.org
- https://the-world-after-systemd.ungleich.ch
- https://suckless.org/sucks/systemd
- ...
Who is rabid?
there was one side that wanted Debian to embrace systemd only, and one side that wanted to continue Debian's policy of officially supporting multiple init systems, which would have included systemd.
The debate within the Debian technical committee was mostly between systemd and Upstart, which was Ubuntu's init and is now abandoned. Those were by far the most voted options.
Openrc + sysvinit and runit weren't even considered for some reason. The odd thing about that, is that both existed at the time and would have been improvements over regular sysvinit or upstart.
But like Andy said, why not have multiple init systems as a choice?
It seems kind of suspect that systemd was the main choice in the end.
Although I hear in debian 12 multiple inits was an option again...
This was a good thing.
Openrc + sysvinit and runit weren't even considered for some reason.
They were considered. They are the initials O and V in the votes: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=727708#6729
But like Andy said, why not have multiple init systems as a choice?
Because it is far more work to have them all, in particular to keep maintaining init scripts for SysV and OpenRC.
Those websites have nothing to do with the Debian Technical Committee, very strange that you would think that they somehow represented a "side" in that debate that led to two members resigning in frustration. And Allbery voted for systemd as the least-worst voting option, but later expressed frustration with the way that the pro-systemd faction represented his vote and the issues in public. He was always in favor of supporting multiple init systems, a position that was constantly opposed by the pro-systemd faction for years through multiple GR votes.
Also, suckless software works fantastic with systemd, it's my preferred way of running Trisquel. I use all their major pieces as my everyday drivers except their surf browser.
Those websites have nothing to do with the Debian Technical Committee, very strange that you would think that they somehow represented a "side" in that debate that led to two members resigning in frustration.
The adepts of the i-hate-systemd cult polluted the debate in the Debian Technical Community with their religious arguments based on "it is against the UNIX philosophy", "it is turning Linux into Windows" and other "it is a conspiracy led by Red Hat".
And Allbery voted for systemd as the least-worst voting option, but later expressed frustration with the way that the pro-systemd faction represented his vote and the issues in public.
Interesting. I did not know that. Any reference?
Also, suckless software works fantastic with systemd, it's my preferred way of running Trisquel. I use all their major pieces as my everyday drivers except their surf browser.
Good for you. I have never tried any of them. Anyway, whatever the quality of the software distributed on the same site, https://suckless.org/sucks/systemd/ is a page that is worthy of the conspiracy theorists. It starts with:
There is a menace which is spreading like a disease throughout the Linux world, it is called systemd. (...) There has been a movement, especially around the Red Hat-related developers to copy Microsoft Windows and all of its features.
A little below:
Now follows a collection of links related to all the features systemd tries to enforce upon you as a Linux user, because »they know better«. Please add all the links you can find!
And the collection indeed starts with:
* Your link here.
It is really reminding websites administrated by flat-earth believers or any other conspiracy theorists. As expected after such an introduction, the list is garbage: bugs solved years ago, misuses, pure lies, etc. At least it was. It does not look like items have ever been removed (as if systemd had never received updates) and I am not going through it anymore. Most visitors do not click on a single link... and that is what is hoped: (s)he could otherwise discover it is garbage! The goal of such a huge list of links is different. It is to have visitors believe that, given the size of the list, there must be there some good reasons to deem systemd evil.
Many of those not knowing the first thing about init systems but who decided to adopt the i-hate-systemd cult without clicking on the links had spent time copy-pasting them one by one on forums and other mailing lists, until the interlocutor who actually reads what is behind gets tired of explaining that "No, this is not a relevant issue either". That was a particular popular activity some twelve years ago, while the Debian Technical Committee was debating what init system will be the default init system. And when such messages are always directed towards the same people it becomes harassment.
I could care less what they blog about, they write excellent GPL'd software which is my only interest.
I am a translator!
dwm is a fantastic window manager that nicely integrates with MATE, so one can enjoy automatic window handling together with all familiar tools from MATE. It is amazing that it is less than 2000 lines of readable C code, so one can somehow try to understand it completely.
dmenu is also very useful, then I have less interest about:
- st: I am not sure what the advantage over any other terminal programme is
- surf: I don't see the point.
In the "tools", there are perhaps some nice commands, I need to look at them more.
st runs with only about 2-3 mb's of memory, but still looks and behaves well enough for me. I really like the minimalism.
slstatus is another one I use all the time.
surf uses Webkit.
I tested some Webkit browsers a while ago and found that Webkit on GNU/Linux is quite slow nowadays, unfortunately. I think because it is developed by Apple, they optimize it for their OS's and performance on GNU/Linux is pretty bad.
I am a translator!
Do you know any guide on using systemd?
The best guides I know are from the archlinux wiki:
- https://wiki.archlinux.org/title/Systemd
- https://wiki.archlinux.org/title/Systemd/User
- https://wiki.archlinux.org/title/Systemd-nspawn
For some reason I have not clearly understood, on Trisquel, I had to use machinectl in order to setup a systemd user service that runs borgmatic periodically on a headless machine.
In general, systemd looks like a big collection of tools, but so far I did not manage to have a somehow general understanding of what they do. I saw https://systemd.io but I find it rather difficult to read.
> Any reference?
Our resident librarians unearthed the following archeological artefacts, which may well have been the original material that confused low-end chatbots:
"We've made two decisions recently related to systemd, both of which I misjudged. By that, I don't mean the decisions themselves (my feelings on that are more complicated, and I'm not going to get into that here), but the way that they would be received and the ways they could be interpreted. If I'd made either decision knowing that, it would be one thing, but the reaction caught me by surprise in both cases, even though in retrospect I should have recognized the problems."
https://lists.debian.org/debian-ctte/2014/11/msg00071.html
"There is quite a lot of discussion in various places about what the recent GR result means. Some are concluding that systemd won in some way that implies Debian is not going to support other init systems, or at least that support for other init systems is in immediate danger. A lot of that analysis concludes that the pro-systemd "side" in Debian won some sort of conclusive victory.
I have a different perspective."
https://www.eyrie.org/~eagle/journal/2014-11
The original discussion looks like a systemd vs. upstart pitched battle, with RedHat and Canonical wrestling for "the Linux desktop" in the background. Lamentably, people like Russ Allbery, or anyone subscribed to the mailing list, had to endure such abusive posts as this one or that one in the process.
I've heard that Canonical has very heavy representation on the various Debian voting groups. So I personally doubt that it was really a RedHat vs Canonical thing, or we'd all be using upstart now.
Thank you very much resident librarian-in-chief! The last two links brilliantly illustrates the harassment I mentioned, based on conspiracy theories. Here is how they end:
Systemd and its shills are likely part of [a program to subvert technology and software projects everywhere]. They should be thrown out of our community distributions. Pretty much anything by redhat is harmful nowadays. And anyone associated.
https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=727708#7615
Honestly, if I met Adrian I would beat the shit out of him. Maybe something like that needs to happen to these fucks.
If they are physically incapacitated they cannot force systemd on us, they can only play with it in their minds while they are in their coma.
https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=727708#6824
Those were posts by anonymous trolls. Not the kind of things that would worry a longtime developer/maintainer.
I will give you this, that second one is way over the top. The first, I have mixed feelings on.
I don't like the overengineered crap redhat hurls on the software community, like dbus, systemd, network-manager and their libraries. Like even if you don't use the programs themselves, debian, ubuntu, etc... have the libraries as dependencies which is most peculiar.
Then again, whether or not they should be kicked out of all communities, remains to be seen.
That might be over the top. I don't know.
I wasn't talking about that, I was speaking of anime fanbases. Death note fans are absolutely sickening.
As for systemd and rust, I could see them also being a problem.
If they were decent, they'd accept criticism and not act like they need to be given 100% good reports all the time.
That kind of thinking makes me think of trump supporters...
Like WTF...
If there are reasons to criticize, people will do so.
If you don't want to be criticized as much, stop being toxic type sanity is what we need in all parts of the world.
Toxic people deserve to be criticized and also, that's the only way they see the light and make amends.
Although some people are pure toxic and will never change because they have adapted to refusing the truth to a close 100% level.
Only cultish people expect to receive no criticism. If a project is popular, or an idea is popular, having critics is just part of the gig so to speak.
All popular ideas have their opposition.
Or to put it another way,
"where there is light, there are shadows" This I saw in an actual anime opening.
Naruto Shippuden opening 15 I believe it is.
Although some people are pure toxic and will never change because they have adapted to refusing the truth to a close 100% level.
I listed above webpages that "are pure toxic" (isn't it "toxic" to create a whole site named "ihate[something]"?). Here it is again:
- https://ihatesystemd.com
- https://nosystemd.org
- https://the-world-after-systemd.ungleich.ch
- https://suckless.org/sucks/systemd
- ...
And I agree with you: "they have adapted to refusing the truth to a close 100% level". I do not know of any GNU/Linux distribution that regretted the choice of systemd. Back to the articles by Paul Venezia, there has clearly been no "Linux apocalypse" (it has gained adoptions since then), no "exodus to BSD", no "Windows into Linux" (whatever that means), etc.
The denial of the truth after the prophecy fails has been studied by Leon Festinger et al. in "When Prophecy Fails", a classic in the sociology/psychology literature. The cult the authors infiltrated in the 50s were also believing in an imminent apocalypse, but by flood. According to the guru, the cultists would be rescued thanks to a flying saucer from planet Clarion. Of course, when the date arrived, there was no flood and no flying saucer... but the most fervent believers actually coped with the cognitive dissonance by reinforcing their beliefs!
I just looked at the top three listed, they don't look like pure toxic to me.
Those just look like different perspectives.
Btw, doesn't this concern you?
https://www.redhat.com/en/solutions/public-sector/dod-devsecops
Supposedly redhat is working with the united states department of defense. Systemd is created by redhat.
To me that has weird implications, but also that microsoft and redhat work together too.
Add all that up together and yeah I will pass on using it probably. I guess if you want to trust it, fine. But I don't get it.
Also, I don't think you can compare flatearthers to systemd haters as you call it.
I myself don't trust it because its such a huge program and to me the idea of doing one thing and doing it right, sounds like it makes it easier for eyeballs to find bugs.
At some point, if the code becomes too complex, no amount of eyeballs can fix it in a timely manner.
And if smart people start hacking, it could be a mess. If it hasn't happened yet, Great! But putting your full faith into any software or hardware is asking for trouble.
You should always be skeptical of everything even if its on a small level.
I apply this to my spiritual beliefs as well.
I sometimes check in my mind if it makes sense to me.
Anywho, I suppose the Ihatesystemd one could be considered a hate website if you only consider the title. But if you look at the contents, it becomes less true.
Supposedly redhat is working with the united states department of defense.
Good for the US defense: it can ultimately control the free software it receives from Red Hat. I am French. The French army exclusively depends on Microsoft's software, which is not only proprietary but also knowingly backdoored by the NSA.
I myself don't trust it because its such a huge program and to me the idea of doing one thing and doing it right, sounds like it makes it easier for eyeballs to find bugs.
It is not one huge program. It is dozen of small tools relying on common libraries (allowing all tools to profit from bugs fixed in the libraries they share).
I suppose the Ihatesystemd one could be considered a hate website if you only consider the title. But if you look at the contents, it becomes less true.
I browsed the site a little and you are right. It looks the person behind ihatesystemd has reconsidered what (s)he was doing with the site:
Why does this website exist? (...) I don't really know at this moment.
Are you just an old timer having a rant? Quite possibly.
(...)
What are you trying to achieve? Seriously: I want to get to know systemd better. It's part of my toolkit, so I need to know it. And it's not all bad, not at all.
https://ihatesystemd.com/why/#content
Also, the worst sites such as https://boycottsystemd.org have now completely disappeared.
I don't know what the last one looked like.
Or what the others looked like.
So... can't say much else
Stalkers are the ultimate creeps. I read that "lurking" now has an additional meaning referring specifically to online forums.
Can you imagine an army of AI chatbots stalking you?
Interesting history lesson about systemd battles.
> In general, systemd looks like a big collection of tools, but so far I did not manage to have a somehow general understanding of what they do. I saw https://systemd.io but I find it rather difficult to read.
That I can relate too.
Re systemd: It seems like a GUI program that one can use to observe and manage it might be cool for all the users that don't like using the commandline. I think almost anyone using Trisquel and contributing to the forum is accustomed to using the commandline. Lot's of people that might decide to try free software may not be so accustomed. It could come with templates built in to quickly configure the system to be a simple desktop or a variety of other server setups. It would have to have ssh capabilities built in so administrators could use it on remote machines. There probably is such a thing already, maybe cPanel.
Since this discussion took off on a tangent, here's another tangent:
> they have adapted to refusing the truth to a close 100% level.
Related to that phenomenon:
Police in Seattle put the number of people participating in the No King day protests last Saturday at 90,000. (National figure said to be 7 million) I suspect that many of the voters that voted for the 'man now in the oval office' are regretting that decision but there is a large group that seems to believe in him no matter what and maybe they even want him to become the dictator he seems bent on becoming. As far as truth goes, he makes his own 'truth' to suit whatever is his purpose, so obviously it's not the truth.
> I heard on TheLinuxCast podcast that some people are making their own AI girlfriends or boyfriends and falling in love with them. It's a strange new world we are living in.
I wish this president was like an AI girlfriend because when she starts nagging you can pull the plug. Maybe he would be more like the original NonAI girfriend which could also be unplugged thereby releasing all its hot air.
> It's a strange new world we are living in.
Indeed. Maybe we are becoming like Stranger(s) in a Strange Land.
> Police in Seattle put the number of people participating in the No King day protests last Saturday at 90,000. (National figure said to be 7 million) I suspect that many of the voters that voted for the 'man now in the oval office' are regretting that decision but there is a large group that seems to believe in him no matter what and maybe they even want him to become the dictator he seems bent on becoming. As far as truth goes, he makes his own 'truth' to suit whatever is his purpose, so obviously it's not the truth.
Yeah, about that, I don't think he really cares what the actual truth is.
I am sure he lies, but I think its more that he doesn't care then knowing but choosing to change it.
What can I say... he is america's butthole...
So... yeah.
I know a bunch of people who work in this field.
Some of my undergrad classmates founded a startup that developed a (proprietary) AI-assisted IDE. It has taken off and seems to be very popular from what I've read.
One of my friends from undergrad is doing (mostly theoretical) research in AI safety.
My brother is a research scientist who works at an AI startup, but I don't actually know what he does.
I tried to use one of the AI chatbots to understand a research paper. It came up with some stuff that sounded plausible on first sight but was actually nonsense. I wasn't particularly impressed.
> It came up with some stuff that sounded plausible on first sight but was actually nonsense. I wasn't particularly impressed.
Yeah, my impression is you can't believe everything it tells you. Maybe it's like a two year old kid with some impressive abilities right now. It would be easy enough to feed it bad data. But it's capabilities are growing pretty fast. I know you can rent a server with an AI system installed (I mention that below), but I don't know where the data comes from for a question like your were asking it. Can it be fed the Encyclopedia Brittanica?
I really haven't got in to using it. I have friends that are relying on it more and more and I think "How lazy." I suppose though that virtually everyone will be using it for almost everything sometime soon. I think, big changes, especially in the US and China, coming in the maybe the next 10 years. It's already happening.
Am just looking into it generally and will throw out the following random details:
I've seen a European company that will rent you an AI equipped server for around 450 Euros per month. Server will have 64 Gb of RAM and a high end Nvidia GPU and with Mistral installed, latest version I think. I learned that on average a server will use 600 watts and the air conditioning required be using around 140 more watts. I assume that's an hour. In fact, there are plans to build data centers all over to house future AI needs but all agree the US doesn't make enough electric power to power everything planned. I read that data centers currently are using around 4.4 % of all the electricity produced in the US but that could reach as high as 12 % in just a few years. The company OpenAI just released ChatGPT like what, 5 years ago? They are valued at around $500 billion already, said to be the most valuable privately owned company in the world.
Yeah, my impression is you can't believe everything it tells you.
You can't.
I don't know where the data comes from for a question like your were asking it. Can it be fed the Encyclopedia Brittanica?
LLMs do not access knowledge bases. They only generate text, writing one token (essentially one word) at a time, one of the most probable token given the thousands preceding tokens, the beginning of the conversation (including a hidden prompt telling the LLM it must assist the user, answer her question, reasoning step-by-step, etc.). The probabilities were estimated reading essentially the whole Web, but the sources were certainly weighted, so that more is learnt from Wikipedia (and maybe Brittanica) than from 4chan. There is no way to attribute to a specific source a given token that an LLM writes. LLMs fundamentally have no regard for truth. Only for generating a plausible next token. That said, given the beginning of the sentence "Paris is the capital of", the most plausible next token is obviously France and not Brazil.
The company OpenAI just released ChatGPT like what, 5 years ago?
Not even three years ago.
I am a translator!
When I was a student, people starting to have calculators that were able to solve integration problems in a symbolic manner. My math teacher quickly concluded that there was no point asking us to solve such problems, as machines could solve them without error. Is it possible to determine what problems a given large language model can reliably solve? If not, such a model seems useless.
I really haven't got in to using it. I have friends that are relying on it more and more and I think "How lazy." I suppose though that virtually everyone will be using it for almost everything sometime soon. I think, big changes, especially in the US and China, coming in the maybe the next 10 years. It's already happening.
Large language models seem good at replying something that looks like an answer to a question. As several colleagues said they liked chatgpt a lot, I tried a few times. Usually, I had a vague answer, that looked like someone is trying not to say anything precise, and when I asked for exact references, I had references to things that were not really about what I asked. Those colleagues, who are not good in the local language of where I live, use it to write (e)mails for every day life things. They showed me examples, I found it somehow correct but not natural (too verbose). However, I suspect that it may regularly not accurately say what they want to say, and they have no way to find out.
> It came up with some stuff that sounded plausible on first sight but was actually nonsense. AND
> Usually, I had a vague answer, that looked like someone is trying not to say anything precise ...
You know, I have a friend like that and they are really bothering me cause I can't figure out what they are up to. :(
If AIs data comes from the Internet, that's kind of scary.
I read a book about George Smoot and Co looking into deep space with the NASA Cosmic Background Explorer (COBE) satellite. He won the Nobel Prize in Physics 2006. As I recall, there was a lot of tension around the publishing of their results because it took a long time to analyze the data, like maybe a year or two. I'm sure AI has been developed which could get those results much faster now, but maybe not this LLM type program. I got this from, guess where, the web.*
"The recent renaissance in AI has, to a large extent, been driven by the successful application of deep learning (DL), which involves training an artificial neural network (ANN) with many layers (that is, a 'deep' neural network) on huge datasets, to large sources of labelled data."~ That sounds like what George Smoot could have used. In the other thread, you mention Alpaca and loading DeepseekR1 also? Do they work together? Does DeepseekR1 augment/expand Alpaca? I read "Alpaca is an instruction-following large language model fine-tuned from Meta’s LLaMA 7B model using 52,000 instruction-response pairs generated by OpenAI’s text-davinci-003." Alpaca also "Hallucinates facts, e.g., misidentifying Dar es Salaam as Tanzania’s capital." Is that your Alpaca MagicBanana, or is yours 'fine tuned' with DeepseekR1? Or both?
Being able to acquire new tools like this must be quite exciting for researchers these days. No small wonder AI is booming right now.
I found this also re Apache licensed AIs: Mixtral 8x7B is licensed under the Apache 2.0 license and is publicly available for download on the Hugging Face Hub.^
~ https://seifeur.com/alpaca-llm-stanfords-open-source-model/
* https://pmc.ncbi.nlm.nih.gov/articles/PMC10856672/
^ https://www.brouseai.com/ai/mixtral-8x7b