Hello!
As a handsome local AI enjoyer™ you’ve probably noticed one of the big flaws with LLMs:
It lies. Confidently. ALL THE TIME.
(Technically, it “bullshits” - https://link.springer.com/article/10.1007/s10676-024-09775-5
I’m autistic and extremely allergic to vibes-based tooling, so … I built a thing. Maybe it’s useful to you too.
The thing: llama-conductor
llama-conductor is a router that sits between your frontend (OWUI / SillyTavern / LibreChat / etc) and your backend (llama.cpp + llama-swap, or any OpenAI-compatible endpoint). Local-first (because fuck big AI), but it should talk to anything OpenAI-compatible if you point it there (note: experimental so YMMV).
Not a model, not a UI, not magic voodoo.
A glass-box that makes the stack behave like a deterministic system, instead of a drunk telling a story about the fish that got away.
TL;DR: “In God we trust. All others must bring data.”
Three examples:
1) KB mechanics that don’t suck (1990s engineering: markdown, JSON, checksums)
You keep “knowledge” as dumb folders on disk. Drop docs (.txt, .md, .pdf) in them. Then:
>>attach <kb>— attaches a KB folder>>summ new— generatesSUMM_*.mdfiles with SHA-256 provenance baked in- `>> moves the original to a sub-folder
Now, when you ask something like:
“yo, what did the Commodore C64 retail for in 1982?”
…it answers from the attached KBs only. If the fact isn’t there, it tells you - explicitly - instead of winging it. Eg:
The provided facts state the Commodore 64 launched at $595 and was reduced to $250, but do not specify a 1982 retail price. The Amiga’s pricing and timeline are also not detailed in the given facts.
Missing information includes the exact 1982 retail price for Commodore’s product line and which specific model(s) were sold then. The answer assumes the C64 is the intended product but cannot confirm this from the facts.
Confidence: medium | Source: Mixed
No vibes. No “well probably…”. Just: here’s what’s in your docs, here’s what’s missing, don’t GIGO yourself into stupid.
And when you’re happy with your summaries, you can:
>>move to vault— promote those SUMMs into Qdrant for the heavy mode.
2) Mentats: proof-or-refusal mode (Vault-only)
Mentats is the “deep think” pipeline against your curated sources. It’s enforced isolation:
- no chat history
- no filesystem KBs
- no Vodka
- Vault-only grounding (Qdrant)
It runs triple-pass (thinker → critic → thinker). It’s slow on purpose. You can audit it. And if the Vault has nothing relevant? It refuses and tells you to go pound sand:
FINAL_ANSWER:
The provided facts do not contain information about the Acorn computer or its 1995 sale price.
Sources: Vault
FACTS_USED: NONE
[ZARDOZ HATH SPOKEN]
Also yes, it writes a mentats_debug.log, because of course it does. Go look at it any time you want.
The flow is basically: Attach KBs → SUMM → Move to Vault → Mentats. No mystery meat. No “trust me bro, embeddings.”
3) Vodka: deterministic memory on a potato budget
Local LLMs have two classic problems: goldfish memory + context bloat that murders your VRAM.
Vodka fixes both without extra model compute. (Yes, I used the power of JSON files to hack the planet instead of buying more VRAM from NVIDIA).
!!stores facts verbatim (JSON on disk)??recalls them verbatim (TTL + touch limits so memory doesn’t become landfill)- CTC (Cut The Crap) hard-caps context (last N messages + char cap) so you don’t get VRAM spikes after 400 messages
So instead of:
“Remember my server is 203.0.113.42” → “Got it!” → [100 msgs later] → “127.0.0.1 🥰”
you get:
!! my server is 203.0.113.42?? server ip→ 203.0.113.42 (with TTL/touch metadata)
And because context stays bounded: stable KV cache, stable speed, your potato PC stops crying.
There’s more (a lot more) in the README, but I’ve already over-autism’ed this post.
TL;DR:
If you want your local LLM to shut up when it doesn’t know and show receipts when it does, come poke it:
- Primary (Codeberg): https://codeberg.org/BobbyLLM/llama-conductor
- Mirror (GitHub): https://github.com/BobbyLLM/llama-conductor
PS: Sorry about the AI slop image. I can’t draw for shit.
PPS: A human with ASD wrote this using Notepad++. If it the formatting is weird, now you know why.
Hallucination is mathematically proven to be unsolvable with LLMs. I don’t deny this may have drastically reduced it, or not, I have no idea.
But hallucinations will just always be there as long as we use LLMs.
Spite based inference?
You dirty pirate hooker.
I don’t believe you.
This is aaesome. Ive been working on something similar. Youre not likely to get much useful from here though. Anything AI is by default bad here
Well, to butcher Sinatra: if it can make it on Lemmy and HN, it can make it anywhere :)
This + Local Wikipedia + My own writings would be sick
I’m not claiming I “fixed” bullshitting. I said I was TIRED of bullshit.
So, the claim I’m making is: I made bullshit visible and bounded.
The problem I’m solving isn’t “LLMs sometimes get things wrong.” That’s unsolvable AFAIK. What I’m solving for is “LLMs get things wrong in ways that are opaque and untraceable”.
That’s solvable. That’s what hashes get you. Attribution, clear fail states and auditability. YOU still have to check sources if you care about correctness.
The difference is - YOU are no longer checking a moving target or a black box. You’re checking a frozen, reproducible input.
That’s… not how any of this works…
Please don’t teach me to suck lemons. I have very strict parameters for fail states. When I say three strikes and your out, I do mean three strikes and you’re out. Quants ain’t quants, and models ain’t models. I am very particular in what I run, how I run it and what I tolerate.
I think you missed the guy this is targeted at.
Worry not though. I get it. There isn’t a lot of nuance in the AI discussion anymore and the anti-AI people are quite rude these days about anything AI at all.
You did good work homie!
Thank you! I appreciate you.
PS: Where’s the guy this should be targeted at?
I want to believe you, but that would mean you solved hallucination.
Either:
A) you’re lying
B) you’re wrong
C) KB is very small
D) None of the above.
I didn’t “solve hallucination”. I changed the failure mode. The model can still hallucinate internally. The difference is it’s not allowed to surface claims unless they’re grounded in attached sources.
If retrieval returns nothing relevant, the router forces a refusal instead of letting the model free-associate. So the guarantee isn’t “the model is always right.”
The guarantee is “the system won’t pretend it knows when the sources don’t support it.” That’s it. That’s the whole trick.
KB size doesn’t matter much here. Small or large, the constraint is the same: no source, no claim. GTFO.
That’s a control-layer property, not a model property. If it helps: think of it as moving from “LLM answers questions” to “LLM summarizes evidence I give it, or says ‘insufficient evidence.’”
Again, that’s the whole trick.
You don’t need to believe me. In fact, please don’t. Test it.
I could be wrong…but if I’m right (and if you attach this to a non-retarded LLM), then maybe, just maybe, this doesn’t suck balls as much as you think it might.
Maybe it’s even useful to you.
I dunno. Try it?
So… Rag with extra steps and rag summarization? What about facts that are not rag retrieval?
Parts of this are RAG, sure
RAG parts:
- Vault / Mentats is classic retrieval + generation.
- Vector store = Qdrant
- Embedding and reranker
So yes, that layer is RAG with extra steps.
What’s not RAG -
KB mode (filesystem SUMM path)
This isn’t vector search. It’s deterministic, file-backed grounding. You attach folders as needed. The system summarizes and hashes docs. The model can only answer from those summaries in that mode. There’s no semantic retrieval step. It can style and jazz around the answer a little, but the answer is the answer is the answer.
If the fact isn’t in the attached KB, the router forces a refusal. Put up or shut up.
Vodka (facts memory)
That’s not retrieval at all, in the LLM sense. It’s verbatim key-value recall.
- JSON on disk
- Exact store (!!)
- Exact recall (??)
Again, no embeddings, no similarity search, no model interpretation.
“Facts that aren’t RAG”
In my set up, they land in one of two buckets.
-
Short-term / user facts → Vodka. That for things like numbers, appointments, lists, one-off notes etc. Deterministic recall, no synthesis.
-
Curated knowledge → KB / Vault. Things you want grounded, auditable, and reusable.
In response to the implicit “why not just RAG then”
Classic RAG failure mode is: retrieval is fuzzy → model fills gaps → user can’t tell which part came from where.
The extra “steps” are there to separate memory from knowledge, separate retrieval from synthesis and make refusal a legal output, not a model choice.
So yeah; some of it is RAG. RAG is good. The point is this system is designed so not everything of value is forced through a semantic search + generate loop. I don’t trust LLMs. I am actively hostile to them. This is me telling my LLM to STFU and prove it, or GTFO. I know that’s a weird way to operate maybe (advesarial, assume worst, engineer around issue) but that’s how ASD brains work.
The system summarizes and hashes docs. The model can only answer from those summaries in that mode
Oh boy. So hallucination will occur here, and all further retrievals will be deterministically poisoned?
Oh boy. So hallucination will occur here, and all further retrievals will be deterministically poisoned?
Huh? That is the literal opposite of what I said. Like, diametrically opposite.
Let me try this a different way.
Hallucination in SUMM doesn’t “poison” the KB, because SUMMs are not authoritative facts, they’re derived artifacts with provenance. They’re explicitly marked as model output tied to a specific source hash. Two key mechanics that stop the cascade you’re describing:
- SUMM is not a “source of truth”
The source of truth is still the original document, not the summary. The summary is just a compressed view of it. That’s why it carries a SHA of the original file. If a SUMM looks wrong, you can:
a) trace it back to the exact document version b) regenerate it c) discard it d) read the original doc yourself and manually curate it.
Nothing is “silently accepted” as ground truth.
- Promotion is manual, not automatic
The dangerous step would be: model output -> auto-ingest into long-term knowledge.
That’s explicitly not how this works.
The Flow is: Attach KB -> SUMM -> human reviews -> Ok, move to Vault -> Mentats runs against that
Don’t like a SUMM? Don’t push it into the vault. There’s a gate between “model said a thing” and “system treats this as curated knowledge.” That’s you - the human. Don’t GI and it won’t GO.
Determinism works for you here. The hash doesn’t freeze the hallucination; it freezes the input snapshot. That makes bad summaries:
- reproducible
- inspectable
- fixable
Which is the opposite of silent drift.
If SUMM is wrong and you miss it, the system will be consistently wrong in a traceable way, not creatively wrong in a new way every time.
That’s a much easier class of bug to detect and correct. Again: the proposition is not “the model will never hallucinate.”. It’s “it can’t silently propagate hallucinations without a human explicitly allowing it to, and when it does, you trace it back to source version”.
And that, is ultimately what keeps the pipeline from becoming “poisoned”.
Huh? That is the literal opposite of what I said. Like, diametrically opposite.
The system summarizes and hashes docs. The model can only answer from those summaries in that mode. There’s no semantic retrieval step.
No, that’s exactly what you wrote.
Now, with this change
SUMM -> human reviews
That would be fixed, but will work only for small KBs, as otherwise the summary would be exhaustive.
Case in point: assume a Person model with 3-7 facts per Person. Assume small 3000 size set of Persons. How would the SUMM of work? Do you expect a human to verify that SUMM? How are you going to converse with your system to get the data from that KB Person set? Because to me that sounds like case C, only works for small KBs.
Again: the proposition is not “the model will never hallucinate.”. It’s “it can’t silently propagate hallucinations without a human explicitly allowing it to, and when it does, you trace it back to source version”.
Fair. Except that you are still left with the original problem of you don’t know WHEN the information is incorrect if you missed it at SUMM time.
Woof, after reading your “contributions” here, are you this fucking insufferable IRL or do you keep it behind a keyboard?
Goddamn. I’m assuming you work in tech in some capacity? Shout-out to anyone unlucky enough to white-knuckle through a workday with you, avoiding an HR incident would be a legitimate challenge, holy fuck.
Hallucination isn’t nearly as big a problem as it used to be. Newer models aren’t perfect but they’re better.
The problem addressed by this isn’t hallucination, its the training to avoid failure states. Instead of guessing (different from hallucination), the system forces a Negative response. That’s easy and any big and small company could do it, big companies just like the bullshit
^ Yes! That. Exactly that. Thank you!
I don’t like the bullshit…and I’m not paid to optimize for bullshit-leading-to-engagment-chatty-chat.
“LLM - tell me the answer and then go away. If you can’t, say so and go away. Optionally, roast me like you’ve watched too many episodes of Futurama while doing it”
A very tailored to llms strengths benchmark calls you a liar.
https://artificialanalysis.ai/articles/gemini-3-flash-everything-you-need-to-know (A month ago the hallucination rate was ~50-70%)
Buuuuullshit. Asked different models about the ten highest summer transfer scorers and got wildly different answers. They then tried to explain why amd got more wrong numbers.
This is amazing! I will either abandon all my other commitments and install this tomorrow or I will maybe hopefully get it done in the next 5 years.
Likely accurate jokes aside this will be a perfect match with my obsidian volt as well as researching things much more quickly.
I hope it does what it I claim it does for you. Choose a good LLM model. Not one of the sex-chat ones. Or maybe, exactly one of those. For uh…research.
I have a Strix Halo machine with 128GB VRAM so I’m definitely going to give this a try with gpt-oss-120b this weekend.
Show off :)
You’re self hosting that, right? I will not be held responsible for some dogey OpenRouter quant hosted by ToTaLlY NoT a ScAM LLC :)
Of course, self hosted with llama-swap and llama.cpp. :)
This is the way. Good luck with OSS-120B. Those OSS models, they
- really
- like
- bullet
- points
gpt-oss is pretty much unusable without custom system prompt.
Sycophancy turned to 11, bullet points everywhere and you get a summary for the summary of the summary.
re: the KB tool, why not just skip the llm and do two chained fuzzy finds? (what knowledge base & question keywords)
re: the KB tool, why not just skip the llm and do two chained fuzzy finds? (what knowledge base & question keywords)
Yep, good question. You can do that, it’s not wrong. If your KB is small + your question is basically “find me the paragraph that contains X,” then yeah: two-pass fuzzy find will dunk on any LLM for speed and correctness.
But the reason I put an LLM in the loop is: retrieval isn’t the hard part. Synthesis + constraint are. What a LLM is doing in KB mode (basically) is this -
-
Turns question into extraction task. Instead of “search keywords,” it’s: “given these snippets, answer only what is directly supported, and list what’s missing.”
-
Then, rather that giving 6 fragments across multiple files, the LLM assembles the whole thing into a single answer, while staying source locked (and refusing fragments that don’t contain the needed fact).
-
Finally: it has “structured refusal” baked in. IOW, the whole point is that the LLM is forced to say “here are the facts I saw, and this is what I can’t answer from those facts”.
TL;DR: fuzzy search gets you where the info lives. This gets you what you can safely claim from it, plus an explicit “missing list”.
For pure retreval: yeah - search. In fact, maybe I should bake in a >>grep or >>find commands. That would be the right trick for “show me the passage” not “answer the question”.
I hope that makes sense?
It does. Great work, btw.
Thank you. I appreciate you saying so!
-
Responding to my own top post like a FB boomer: May I make one request?
If you found this little curio interesting at all, please share in the places you go.
And especially, if you’re on Reddit, where normies go.
I use to post heavily on there, but then Reddit did a reddit and I’m done with it.
https://lemmy.world/post/41398418/21528414
Much as I love Lemmy and HN, they’re not exactly normcore, and I’d like to put this into the hands of people :)
PS: I am think of taking some of the questions you all asked me here (de-identified) and writing a “Q&A_with_drBobbyLLM.md” and sticking it on the repo. It might explain some common concerns.
And, If nothing else, it might be mildly amusing.
Okay pardon the double comment, but I now have no choice but to set this up after reading your explainations. Doing what TRILLIONS of dollars hasn’t cooked up yet… I hope you’re ready by whatever means you deam, when someone else “invents” this
It’s copyLEFT (AGPL-3.0 license). That means, free to share, copy, modify…but you can’t roll a closed source version of it and sell it for profit.
In any case, I didn’t build this to get rich (fuck! I knew I forgot something).
I built this to try to unfuck the situation / help people like me.
I don’t want anything for it. Just maybe a fist bump and an occasional “thanks dude. This shit works amazing”
I wouldn’t know how to get this going, but I very much enjoyed reading it and your comments and think that it looks like a great project. 👍
(I mean, as a fellow autist I might be able to hyperfocus on it for a while, but I’m sure that the ADHD would keep me from finishing to go work on something else. 🙃)
Ah - ASD, ADHD and Lemmy. You’re a triple threat, Harry! :)
Glad if it was entertaining, if even a little!
I really need this. Each time I try messing with GPT4All’s “reasoning” model, it pisses me off. I’m selective on my inputs, low temperature, local docs, and it’ll tell me things like tension matters for a coil’s magnetic field. Oh and it spits out what I assume is unformatted LATEX so if anyone has an interface/stack recommendation please let me know
I feel your pain. Literally.
I once lost … 24? 26? hrs over a period of days with GPT once…it each time confidently asserting “no, for realz, this is the fix”.
This thing I built? Purely spite driven engineering + caffeine + ASD to overcome “Bro, trust me bro”.
I hope it helps.
I’m probably going to give this a try, but I think you should make it clearer for those who aren’t going to dig through the code that it’s still LLMs all the way down and can still have issues - it’s just there are LLMs double-checking other LLMs work to try to find those issues. There are still no guarantees since it’s still all LLMs.
I haven’t tried this tool specifically, but I do on occasion ask both Gemini and ChatGPT’s search-connected models to cite sources when claiming stuff and it doesn’t seem to even slightly stop them bullshitting and claiming a source says something that it doesn’t.
Yeah, this is different. Try it. It gives you cryptogenic key to the source (which you must provide yourself: please be aware. GIGO).
How does having a key solve anything? Its not that the source doesn’t exist, it’s that the source says something different to the LLM’s interpretation of it.
Yeah.
The SHA isn’t there to make the model smarter. It’s there to make the source immutable and auditable.
Having been burnt by LLMs (far too many times), I now start from a position of “fuck you, prove it”.
The hash proves which bytes the answer was grounded in, should I ever want to check it. If the model misreads or misinterprets, you can point to the source and say “the mistake is here, not in my memory of what the source was.”.
If it does that more than twice, straight in the bin. I have zero chill any more.
Secondly, drift detection. If someone edits or swaps a file later, the hash changes. That means yesterday’s answer can’t silently pretend it came from today’s document. I doubt my kids are going to sneak in and change the historical prices of 8 bit computers (well, the big one might…she’s dead keen on being a hacker) but I wanted to be sure no one and no-thing was fucking with me.
Finally, you (or someone else) can re-run the same question against the same hashed inputs and see if the system behaves the same way.
So: the hashes don’t fix hallucinations (I don’t even think that’s possible, even with magic). The hashes make it possible to audit the answer and spot why hallucinations might have happened.
PS: You’re right that interpretation errors still exist. That’s why Mentats does the triple-pass and why the system clearly flags “missing / unsupported” instead of filling gaps. The SHA is there to make the pipeline inspectable, instead of “trust me, bro.”.
Guess what? I don’t trust you. Prove it or GTFO.
The hash proves which bytes the answer was grounded in, should I ever want to check it. If the model misreads or misinterprets, you can point to the source and say “the mistake is here, not in my memory of what the source was.”.
Eh. This reads very much like your headline is massively over-promising clickbait. If your fix for an LLM bullshitting is that you have to check all its sources then you haven’t fixed LLM bullshitting
If it does that more than twice, straight in the bin. I have zero chill any more.
That’s… not how any of this works…
Fair point on setting expectations, but this isn’t just LLMs checking LLMs. The important parts are non-LLM constraints.
The model never gets to “decide what’s true.” In KB mode it can only answer from attached files. Don’t feed it shit and it won’t say shit.
In Mentats mode it can only answer from the Vault. If retrieval returns nothing, the system forces a refusal. That’s enforced by the router, not by another model.
The triple-pass (thinker → critic → thinker) is just for internal consistency and formatting. The grounding, provenance, and refusal logic live outside the LLM.
So yeah, no absolute guarantees (nothing in this space has those), but the failure mode is “I don’t know / not in my sources, get fucked” not “confidently invented gibberish.”
I have no remarks, just really amused with your writing in your repo.
Going to build a Docker and self host this shit you made and enjoy your hard work.
Thank you for this!
Same sentiment. Tonight it run on my systems XD.
There are literally dozens of us. DOZENS!
I’m on a potato, so I can’t attach it to something super sexy, like a 405B or a MoE.
If you do, please report back.
PS: You may see (in the docs) occasional references that slipped passed me to MoA. That doesn’t stand for Mixture of Agents. That stood for “Mixture of Assholes”. That’s always been my mental model for this.
Or, in the language of my people, this was my basic design philosophy:
YOU (question)-> ROUTER+DOCS (Ah shit, here we go again. I hate my life)
|
ROUTER+DOCS -> Asshole 1: Qwen (“I’m right”)
|
ROUTER+DOCS -> Asshole 2: Phi (“No, I’m right”)
|
ROUTER+DOCS -> Asshole 3: Nanbeige (“Idiots, I’m right!”)
|
ROUTER+DOCS (Jesus, WTF. I need booze now) <- (all assholes)
|
–> YOU (answer)
(this could have been funnier in the ASCII actually worked but man…Lemmy borks that)
EDIT: If you want to be boring about it, it’s more like this
PS: If you like it, let other people in other places know about it.
Thank you <3
Please let me know how it works…and enjoy the >>FR settings. If you’ve ever wanted to trolled by Bender (or a host of other 1990s / 2000s era memes), you’ll love it.
THIS IS AWESOME!!! I’ve been working on using an obsidian vault and a podman ollama container to do something similar, with VSCodium + continue as middleware. But this! This looks to me like it is far superior to what I have cobbled together.
I will study your codeberg repo, and see if I can use your conductor with my ollama instance and vault program. I just registered at codeberg, if I make any progress I will contact you there, and you can do with it what you like.
On an unrelated note, you can download wikipedia. Might work well in conjunction with your conductor.
Please enjoy :) Hope it’s of use to you!
EDIT: Please don’t yeet wikipedia into it. It will die. And you will be sad.
