Skip Navigation

User banner
Posts
16
Comments
44
Joined
3 yr. ago

users lookin at my profile counter:

(its mostly myself)

heyhi ~ i make posts and hope that peeps comment on them so i can comment on them and kinda have a conversation sorta.

come and have a tea with me 🍵 <3

  • yes but is a very different thing. reasoning tokens dont always provide insight and they very commonly contain mistakesyespecially among qwen models.

    they are not code-verifiable sources.

  • im not really looking for something like this, im just disappointed to see that todays chatbots with web search still just make u

    • either trust the LM with its sources
    • or read thru ever source completely to actually trust it

    even tho... It feels so obvious to just show that preview of exactly the text describing the answer to be most immediately trustworthy.

  • yea no ur right, grep works great fir doin codebase search. dunno what this other peeps is going on about, for the case of file search, grep rules.

  • yyyyes exactly.

  • i know that generally, quoting stuff from files works well. the point here is less about being useful but more about being 100% verifiable.

    indexing things with line numbers absolutely works, but u gotta actually put line numbers in the tool output. Meaning, the read_source_code should return the line number at the start if each line, e.g.

     
        
    1:extends Control
    2:
    3:# comment here
    4:and so on
    
    
      

    it eats up tokens, but does make line index hit rate almost always correct.

  • Free Open-Source Artificial Intelligence @lemmy.world

    a mockup of what "trushworthy LM search" could look like

  • queer person sues employer for forcing them to see an american flag on the way to work

  • its largely sitting in rp thingies and such.

    currently, training a video gen model on corn is simply too expensive, and tech companies like openai and alibaba would rather not train on it, as it costs them extra thousands to train a type of video which is restricted to be generated anyway.

    data curation has become one of the pillows improving models performance lately.

    • makes it easier to wear all the clothing u could want
    • doesnt get weirdly stuck on clothing
    • doesnt make u look real dumb when getting erection
    • just.... way prettier
    • much comfier to lick (very opinionated, I know)
    • doesnt.... feel as delicate. a blood-enduced meat sausage feels like its prime for breaking off or- having problems when hitting walls (which is a genuine problem for some apparently)

    i think thats it....

  • its so weird that that is now mistrals "medium" size...

    weird to see em growing up like this...

  • Free Open-Source Artificial Intelligence @lemmy.world

    SKILL.md: The Feature to end them all

  • what volume of markdown files are we talking?

    also, just so i understand this right:you are looking fir a markdown editor which has a chat window on the side which can look at other files to assist in writing.is that correct?

    which editor do you use right now for editing the files? does it need to support vim-movements? (if u dont know what that is, it doesnt matter)

    what exactly would the LM be assisting in? should it be to just read files and respond, or edit them itself aswell, or suggest edits?

    depending on the amount of files, a simple index and read-tool functionality might be enough. Here is how you would create such an index:

    1. LM looks at each md file
    2. Then generates a one-liner about the content, like:

      Static functions and variables in Godot, using autoload scripts and scenes (e.g. loading screens, overlays, save systems, anything permanently loaded)

    3. store that into a file alongside each files path, perhaps like this:
     plaintext
        
    ~/Documents/
      file.md (one-liner here)
      another_file.md (one-liner here)
      topic/
        file_about_topic.md (one-liner here)
    [...]
    
    
      

    These three steps can be done using any coding agent you got lying around using this prompt

     plaintext
        
    Goal: FILE_INDEX.txt which contains all files in <folder with the documents> in tree form (reduce redundant dir paths) with a concise and keyword-heavy one-liner about the files content.
    
    FILE_INDEX.md format:
    
    ~/SomeDir/SomeDirBelowIt/
      a_file_in_that_dir.md (description)
      another_file.md (description)
      dir_inside_it/
        file_in_that_dir.md (description)
    
    How: First use `tree` to see the directories contents, then use subagents to delegate the generation of the descriptions. They should out them into files themselves. An agent can summarize 5 files at mist in one go. Tell each agent exactly what name the output file should have. Finally, after having received all descriptions with their file names, combine the files all into one final FILE_INDEX.txt
    
    
      

    This index can then be did into any agent to let it find files quicker, without having to hope for good chunking settings in a RAG pipeline.


    all this was written by a human, even if it might not seem as such.

  • (wait why is this on here? just looks like some shilling thing)

  • i wish it were an art >o< thatd be so cool---

  • moltbook is an open source thing, no?...

    imean- hes not gonna grow the social media, that doesnt make sense yet.

    hes (zuck 🍬) sees more in moltbook than there is.

  • mygosh habeck ~

    i kinda liked him in a person kind of way.... he also seemed much more honest than most politicians...

    (dunno much about him otherwise tho- besides the infamous Bäcker Insolvenz thingy and his party 🎂🎈 >v< )

  • I mean, they aren't deploying language models for thos, as far as I know.

    they are using..... u know?.... more classic deep learning models which are hard-trained on rule-based feedback, so like, stuff which is explicitly made for this one task.

    which... feels reasonable?

    today's transformer architecture makes this kinda stuff valid, so as long as someone with actual experience and knowledge is using the model as a draft kinda thing, it feels like a good starting point.... (but eh, what I say doesn't matter, I don't work in the chip industry)

  • sounds.... reasonable? >~<

  • I believe its still called Microsoft 365. the Microsoft 365 copilot is its own subscription they r advertising here. mayb im off tho-

  • oh umm.... well.... ur gona be disappointed,,.. that community - weeellll...... its nothing special, i made it for funzies mostly and moated some dum stuff on there.... have a look at that stuff first.,... i....,... wouldnt recommend it-

  • 196 @lemmy.blahaj.zone

    rules in schnitzel!?!?

  • Free Open-Source Artificial Intelligence @lemmy.world

    tool-result insertion for LLM agents: personal findings

  • 196 @lemmy.blahaj.zone

    my scarf rules me

  • 196 @lemmy.blahaj.zone

    i wana rule like her ~

  • 196 @lemmy.blahaj.zone

    my head does not rule

  • 196 @lemmy.blahaj.zone

    imma jus post this n u look at it n dun judge, is this rule oki?

  • Free Open-Source Artificial Intelligence @lemmy.world

    how does one interact with MCP servers without mcp-libraries?

  • Selfhosted @lemmy.world

    IPv4 address reaches website, but domain doesn't... (wrong community?)

  • Free Open-Source Artificial Intelligence @lemmy.world

    For tool-result data: how do u let the llm know?

  • Free Open-Source Artificial Intelligence @lemmy.world

    Before you buy and courses: Read this free prompting guide! (no login required)

    www.promptingguide.ai
  • Free Open-Source Artificial Intelligence @lemmy.world

    How should one treat fill-in-the-middle code completions?

  • Free Open-Source Artificial Intelligence @lemmy.world

    code-completion model (Qwen2.5-coder) rewrites already written code instead of just completing it

  • Free Open-Source Artificial Intelligence @lemmy.world

    code-completion model (Qwen2.5-coder) rewrites already written code instead of just completing it

  • 3DPrinting @lemmy.world

    Does one need special drivers for running 3D printers?