![](https://lemmy.dudeami.win/pictrs/image/d81210bb-1088-48f7-9496-7df8ed9470c0.png)
![](https://lemmy.world/pictrs/image/c47230a8-134c-4dc9-89e8-75c6ea875d36.png)
I think the point would to be make them like cigarette warning labels. At the moment the text can be hidden on a bottle or can in tiny text. It needs to be a big ugly white box with a black border and large text that gets people’s attention.
I like to code, garden and tinker
I think the point would to be make them like cigarette warning labels. At the moment the text can be hidden on a bottle or can in tiny text. It needs to be a big ugly white box with a black border and large text that gets people’s attention.
Semi-cold? That’s extra, you’ll be lucky to afford it. The affordable water been sitting out on the pavement for a few weeks.
Sadly it wasn’t a bid to open source the AI, rather than a bid for payment.
My question would be, why do you need a more powerful server? Are you monitoring your load and seeing it’s overloaded often? Are you just looking to be able to hook more drives to it? Do you need to re-encode video on the fly for other devices? Giving some more details would help someone to give a more insightful answer. I personally am using a Raspberry Pi 4, Chromebox w/ an i7, an old HP rack server, and an old desktop PC for my self hosting needs, as this is cheaper than buying all new hardware (though the electricity bill isn’t the greatest haha, but oh well). If you are just looking for more storage, using the USB 3.0 slots on the Raspberry Pi 4b you can add a couple extra SSDs using a NVMe to USB 3.0 enclosure. For most purposes the speeds will be fine for most applications.
As for SSD vs HDD, SSD hands down. The only reason you’d pick an HDD is if your trying to get more storage cheaper and don’t mind a higher rate of failure. If your data is at all valuable, and it almost always is, redundancy should be added as well.
And as for running Linux, if it can’t run Linux I wouldn’t want to own it.
Edit: Fixed typo
This might help, sorry if it doesn’t, but here is a link to CloudFlares 5xx error code page on error 521. If you’ve done everything in the resolution list your ISP might be actively blocking you from hosting websites, as it is generally against the ISPs ToS to do such on residential service lines. This is why I personally rent a VPS and have a wireguard VPN setup to host from the VPN, which is basically just a roll your own version of Tailscale using any VPS provider. This way you don’t need to expose anything via your ISPs router/WAN and they can’t see what you are sending or which ports you are sending on (other than the encrypted VPN traffic to your VPS of course).
I use my own router with DD-WRT in-between the ISPs router/modem and my LAN, and use a different subnet. I haven’t had any issues with this myself, and my router just sees the ISP router/modem as the WAN.
I’ve never ran this program, but skimmed the documentation. You should be able to use the SHIORI_DIR
(or a custom database table following those instructions) along with the -p
argument for launching the web interface. A simple bash script that should work:
export SHIORI_DIR=/path/to/shiori-data-dir
shiori serve -p 8081
To run multiple versions, I’d suggest setting up each instance as a service on your machine in case of reboots and/or crashes.
Now for serving them, you have two options. The first is just let the users connect to the port directly, but this is generally not done for outward facing services (not that you can’t). The second is to setup a reverse proxy and route the traffic through subdomains or subpaths. Nginx is my go-to solution for this. I’ve also heard good things about Caddy. You’ll most likely have to use subdomains for this, as lots of apps assume they are the root path without some tinkering.
Edit: Corrected incorrect cli arguments and a typo.
TL;DR: The bot is configured to condense certain instances and communities. At the moment, only beehaw.org is marked to be condensed.
Quickly looking at the source code, it seems ReplyToPostsCommand
uses a SummaryTextWrapper
, which contains an iterable for both CondensedSummaryTextWrapperProvider
and DefaultSummaryTextWrapperProvider
. The DefaultSummaryTextWrapperProvider
has a priority of -1_000
(so it’s always checked last) and is set to always return true
on the supports(Community $community): bool
. CondensedSummaryTextWrapperProvider
references the config/services.yaml for it’s supports(Community $community): bool
call which lists 0 condensed communities and 1 condensed instance, being beehaw.org.
If you are expecting a more windows-like experience, I would suggest using Ubuntu or Kubuntu (or any other distro using Gnome/KDE), as these are much closer to a modern Windows GUI. With Ubuntu, I can use the default file manager (nautilus
) and do Ctrl+F
and filter files via *.ext
, then select these files then cut and paste to a new folder (drag and drop does not seem to work from the search results). In Kubuntu, the search doesn’t recognize *
as a wildcard in KDE’s file manager (dolphin
) but does support drag/drop between windows.
In my humble opinion, we too are simply prediction machines. The main difference is how efficient our brains are at the large number of tasks given for it to accomplish for it’s size and energy requirements. No matter how complex the network is it is still a mapped outcome, just the number of factors weighed is extremely large and therefore gives a more intelligent response. You can see this with each increment in GPT models that use larger and larger parameter sets giving more and more intelligent answers. The fact we call these “hallucinations” shows how effective the predictive math is, and mimics humans abilities to just make things up on the fly when we don’t have a solid knowledge base to back it up.
I do like this quote from the linked paper:
As we will discuss, we find interesting evidence that simple sequence prediction can lead to the formation of a world model.
That is to say, you don’t need complex solutions to map complex problems, you just need to have learned how you got there. It’s never purely random attempts at the problem, it’s always predictive attempts that try to map the expected outcomes and learn by getting it right and wrong.
At this point, it seems fair to conclude the crow is relying on more than surface statistics. It evidently has formed a model of the game it has been hearing about, one that humans can understand and even use to steer the crow’s behavior.
Which is to say that it has a predictive model based on previous games. This does not mean it must rigidly follow previous games, but that by playing many games it can see how each move affects the next. This is a simpler example because most board games are simpler than language with less possible outcomes. This isn’t to say that the crow is now a grand master at the game, but it has the reasoning to understand possible next moves, knows illegal moves, and knows to take the most advantageous move based on it’s current model. This is all predictive in nature, with “illegal” moves being assigned very low probability based on the learned behavior the moves never happen. This also allows possible unknown moves that a different model wouldn’t consider, but overall provides what is statistically the best move based on it’s model. This allows the crow to be placed into unknown situations, and give an intelligent response instead of just going “I don’t know this state, I’ll do something random”. This does not always mean this prediction is correct, but it will most likely be a valid and more than not statistically valid move.
Overall, we aren’t totally sure what “intelligence” is, we are just an organism that has developed more and more capabilities to process information based on a need to survive. But getting down to it, we know neurons take inputs and give outputs based on what it perceives is the best response for the given input, and when enough of these are added together we get “intelligence”. In my opinion it’s still all predictive, its how the networks are trained and gain meaning from the data that isn’t always obvious. It’s only when you blindly accept any answer as correct that you run into these issues we’ve seen with ChatGPT.
Thank you for sharing the article, it was an interesting article and helped clarify my understanding of the topic.
Disclaimer: I am not an AI researcher and just have an interest in AI. Everything I say is probably jibberish, and just my amateur understanding of the AI models used today.
It seems these LLM’s use a clever trick in probability to give words meaning via statistic probabilities on their usage. So any result is just a statistical chance that those words will work well with each other. The number of indexes used to index “tokens” (in this case words), along with the number of layers in the AI model used to correlate usage of these tokens, seems to drastically increase the “intelligence” of these responses. This doesn’t seem able to overcome unknown circumstances, but does what AI does and relies on probability to answer the question. So in those cases, the next closest thing from the training data is substituted and considered “good enough”. I would think some confidence variable is what is truly needed for the current LLMs, as they seem capable of giving meaningful responses but give a “hallucinated” response when not enough data is available to answer the question.
Overall, I would guess this is a limitation in the LLMs ability to map words to meaning. Imagine reading everything ever written, you’d probably be able to make intelligent responses to most questions. Now imagine you were asked something that you never read, but were expected to respond with an answer. This is what I personally feel these “hallucinations” are, or imo best approximations of the LLMs are. You can only answer what you know reliably, otherwise you are just guessing.
Most oil is not economically salvaged due to the low cost of extraction from wells. At best they’ll try to burn it off, at worst they just won’t give a shit.