Wildmimic

For the moment i can live with the “not completely open source compliant” situation - they still respect the privacy of their users, and it would also mean to start looking for something that works nearly as well as grayjay.

I actually tried heliboard, but with swiping depending on the library i decided against them back then, and before that changes i do not see any real difference in the current situation (one has a closed source blob, the other is not true open source).

its not that the license of futos keyboard is awful for an end-user, its that in case It gets abandoned you cant make a fork and finance it by adapting the links.

I am not an FOSS purist - i was using Windows until about 2 years ago, and have been using MS operating systems dating back to MS-DOS 5. I acknowledge the importance of FOSS software and development for both the ecosystem as a whole and to make sure that there is available software for end users that doesn’t come with a thousand strings attached and that respects privacy (that actually was the reason for leaving windows behind). I can live with Futos license, as long as they keep working on it - the moment they drop development tho and dont update their license to reflect the new reality, i’ll drop it like a hot potato.

Plague my ass. Something like plague would not make high officials travel to the area and it also wouldn’t cause the censor apparatus to scrub everything. Plague is treated with standard antibiotics and supporting measures when needed. There’s no need for anyone coming from Moscow for something like this.

These technologies are pretty much unable to completely genocide mosquitos unless deployed worldwide in an massive effort, which noone wants to actually do - they are really only problematic in areas where mosquitos and blood-borne diseases which get transmitted by them happen to coexist.

There’s a large difference between 15 years of science to make sure something doesn’t do more harm than it prevents, and taking a decade before someone even declares themselves as responsible for evaluating anything.

That only means that whatever they found at his residences is fucked up enough to land him in prison for the rest of his life.

I was not aware of that. Damn, this is why we can’t have nice things. Back to looking for a useable swipe-keyboard on my phone that i don’t feel icky using. Currently it’s still better than using Gboard from a privacy perspective, but i’d rather try to get something less problematic.

Edit: I read the addendum fully and in peace now, and it seems they have at least rectified the situation surrounding donations/grants. They fall into the same trap as so many coders do, as in “I don’t do politics” to make sure they are as palatable as possible to as large an audience as possible. They do not realize that what they do by default is politics, and taking an official stance is an important part of that. I am somewhat confused by that too, because i am very sure by Louis Rossmans behavior in general that he is a very political person.

Edit2: Ok, after looking a bit more into the issue (mainly https://www.youtube.com/watch?v=zaoXHT8BfV4 and https://danb.me/blog/futo-open-source-definition/ ) it seems that in general they are mainly guilty of Open-Source-Washing their projects. Their Licences are more restrictive than what is commonly known as FOSS, which they have every right to do so, but they keep using the term “Open Source” where it does not meet the commonly accepted definition of that term. Their licences are still pretty permissive in regards to openness what exactly you are running on your devices, but especially the redistributing situation (not allowing to change payment links, which also means any forks of their software would not have any financing opportunities; not allowing to place “ads and malware”, which, while i fully agree with it, is still not in line with the open source definition) does not meet the requirements of “Open Source”.

I feel a bit less icky now, phew.

This is completely irrelevant - by default local models do not have access to the internet, because you have to provide/activate the relevant tools. They can’t access the web by themselves regardless of where you are, the environment has to be set up for it. Even if you provide a web search tool, they can’t do anything outside of providing search parameters - what happens with them is defined by the tool, not the LLM.

These things work by providing the LLM with a system prompt that explain them what they have to do to get access to specific tools. They get provided the specific syntax they have to use it, and the environment they are running inside is then parsing the correctly formatted plain text the LLM outputs.

Specific example for the system prompt in Newelle, a all-in-one suite on Linux to run models, to provide tools in general and the web search tool in detail:

# Tools  

## Overview  
You have access to tools that extend your capabilities. Use them when they are relevant to the user's request.  

## Invocation Format  
When using a tool, output **only** a single valid JSON object:  

\`\`\`json  
{  
  "tool": "tool_name",  
  "arguments": {  
    "arg_name": "arg_value"  
  }  
}  
\`\`\`  

## Rules  
1. Output only the JSON object — no explanations, markdown, or extra text before or after.  
2. Ensure valid JSON: no comments, trailing commas, or extra text.  
3. Use only the tools listed below and only their defined arguments.  
4. **After invoking a tool, stop generating immediately.** Wait for the result before continuing.  
5. Some tools are shown in **compact form** (only name and description, no `parameters`), marked "(compact: ...)". Calling a compact tool directly with guessed or missing arguments is an error and will be rejected.  
## Available Tools  

\`\`\`  
{TOOLS}  
\`\`\`  

The Search tool in detail:

{  
  "name": "search",  
  "description": "Perform a search query on the internet, you can specify the number of results to return and if you want to only return the links and titles.",  
  "parameters": {  
    "type": "object",  
    "properties": {  
      "query": {  
        "type": "string"  
      },  
      "tool_uuid": {  
        "type": "string"  
      },  
      "only_links": {  
        "type": "boolean"  
      },  
      "max_results": {  
        "type": "integer"  
      }  
    },  
    "required": [  
      "query"  
    ]  
  }  
}  

Edit: I can’t for the love of god find out how to escape the 3 backticks for the JSON-Object correctly without either breaking the code blocks or the backslashes showing up. In the original prompt, the backslashes do not exist.

Insane. They have gone completely insane. I am one of the people who say Machine Learning and LLMs are a good technology, and that the amount of ressources used by the common ChatGPT-User is not worth writing home about (as long as they aren’t generating slop videos, which are using remarkable amounts of ressources for the arguably crappy results), but what these people are doing has nothing to do with anything even remotely rational. All of this is a financial desaster waiting to happen, just for militaries and spy agencies to scoop up what has been built to turn it against all of us.

Local Models do not betray you.

Also, if you wanna take notes on your PC, use something like this: https://notes.futo.tech/ (sadly the server side is a bit unstable on my machine at the moment, but the tool itself works fine on my devices, and it’s in active development. Since FUTO’s keyboard replaced gboard on my phone a few months ago after years of simply being not good enough yet, i am pretty sure they will keep improving it.

Vehicle safety in the us for pedestrians is worse than ever. The amount of huge and heavy SUVs who hit people litterally like a truck with their high hoods, which also have huge dead spots in driver vision have counteracted the technical improvements made.

I postulate that corporate social media has only one goal outside of making money: influencing public opinion. Any algorithms that optimize for engagement with ideas that are hostile to democracy should be forbidden, just as influencing foreign elections in general.

As we can see with matchmakers in the US having a fee to find a tradwife for neonazis of around 25k$, they already do not get pussy and are losers.

That’s why i postulate that they will be bailed out, for the simple reason that the US military and 3-letter-agencies do not care if you can make money with this tech. They care about the amount of control it gives them.

If they get bailed out it means that the Military-Industrial complex has it’s fingers so deep in OpenAI and Anthropic that they wear them like a glove.

Edit: the news actually are developing faster than i thought. Jay Clayton, a Three-Letter-Agency guy being called upon to be responsible for AI by Trump makes my prediction a certainty.

Because there is no way he gets 1 Trillion in an IPO the moment their finances become public, and that is the bare minimum he needs so he doesn’t get ripped to shreds.

damn, a dremel and some resin to make it appear unmodified again are enough? the dremel is probably optional - my soldering iron would probably suffice, and some transparent tape well placed could also do the job in a pinch. If Meta wanted these glasses to not be so easily modable, they would have implemented at lease SOME check - make the camera look for the light of the led or smth - make it flash a OTP code to be recognized by the cam, which is fucking easy to do

The political compass is fine, it’s just the US which has sawed off the outer third of the left side x-axis, while eliminating the y-axis altogether, bunching authoritarian and libertarian groups into a constantly infighting sack of cats.

There is a difference between preventing misinformation and outright preventing installation of your software based on domain names, and NOT MAKING THIS PUBLIC and grabbing the list of “banned” domains from an external server (which already shows that they knew this is fucked up). Not even the Russia-Fans Nutomic and Descartes stoop so low.

The first one i can understand. But dbzer0 is not a place of misinformation, (neither is anarchist.nexus for that matter - we are not a homogenous group and i dare you to find me posting misinformation outside of honest mistakes, and it is obvious when something i write is an personal opinion) and there is NOT ONE reason outside of personal animosity to do this.

The Fediverse is not a place for hidden mechanisms like that. They were upfront with their implementation and that they have an opinionated software, which i can understand completely. The swift change towards hidden triggers to prevent usage of their open-source software by whoever they do not like at the moment (and that’s exactly what you are getting into with the mechanism they implemented), and their reaction after being found out, doubling down and inciting defederations instead of doing the right thing, disqualifies this person as a whole in this space.

Edit: The important thing here is: they made it about themselves, not for a larger goal like combating disinformation or making the Fediverse a better place. Their personal animosity vs. specific worldviews tainted their pretty good product. As I said, i have enough criticism regarding the lemmy devs (especially that the finances of lemmy.ml and donations towards lemmy development are intertwined), but even with their left-auth worldview, they did not prevent anyone from using their software because they are “are thinking the wrong way”.

could you please explain? what else outside of hiding censor mechanisms and “incompatibility” because of “i don’t like you” on external URLs could have a similar impact?

Go for PyLova instead please (the fork of PieFed without a diva dev that implements hidden blocklists and gets pissy if he gets found out).

Well, it would be smarter than if you trained it on 4chan, but not by much.

It would also be a bit less racist, but again not by much.

But it would be well versed in memeology and where to find e-thots on onlyfans.