DQM
- 7 Devlogs
- 8 Total hours
Deterministic Query Model (aka DQM) is a trie-based lookup engine that connects to external “brain” files for info. This project was built using Python.
Ship #1
Pending review
@tanzeemrockz
on
DQM
DQM (Deterministic Query Model)
A lightweight, trie-based lookup engine that traverses pre-built JSON “brains”.
What is DQM?
Instead of using power and resource-hungry, slow LLMs, We can now clear doubts using this lookup engine that walks through a pre-built “brain” (.tqm format).
Try it out: Official Page
Overview
Why does DQM exist?
You may ask, why does DQM exist? The short answer is, accuracy and speed. I was tired of seeing people using AI for the most trivial things — “How hot is the sun? How deep is the sea?”, these are in fact very simple questions and there should’ve been a better alternative. Sure, search engines exist for a reason but many people have stopped using them and moved over to AI — Search engines themselves are starting to implement AI heavily into them. I was tired of it providing wrong answers and making-up stuff that are simply – not real. So now, we have DQM.
Our guarantees
We guarantee zero-hallucinations, fully predictable, and absolute offline capability. But any community-made brains are not guarenteed these qualities — as they will turn out to be the brain developers’ fault.
How it works
- Step 1 — Normalisation: Lowercases, expands slang/contractions, removes repeated characters that serve no purpose, strips punctuation, and normalises delimiters.
- Step 2 — Trie Traversal: Splits the normalised queries into tokens and walks through the dictionary nodes.
-
Step 3 — Resolution: Returns the string leaf, picks randomly if the leaf is an array, or throws clear status messages (
Path not found...orIncomplete query...)
The .tqm brain schema
Nothing fancy, just good ole JSON
Example:
{
"brain_id": "custom_core",
"version": "1.0.0",
"author": "John Doe",
"info": "Rocket science, quantum physics, etc...",
"description": "Custom brain for Rocket Science and Quantum Physics",
"licence": "CC-BY-NC-SA-4.0",
"compatibility": ">=2.1.0",
"routes": {...}
}
-
CLI Commands:
Interactive brain selector on startup,cls/clearclears the screen, andexit/quitto quit.
@tanzeemrockz
on
DQM
57m 29s logged
Hey people! I just wrote the README and now, I’m ready to ship. Wish me luck!
@tanzeemrockz
on
DQM
1h 6m 1s logged
GUYS! I just added a web-demo using python, js, and html using vercel beta python runtime! please do check it out at https://dqm.cyberslayercorpz.dev/
@tanzeemrockz
on
DQM
1h 16m 23s logged
guys, i just built a web page (html + css + tailwind) for DQM it’s still in localhost but I’m deploying it in a bit (be sure to visit “dqm.cyberslayercorpz.dev”!)
@tanzeemrockz
on
DQM
1h 30m 56s logged
heyya! I just added more features and bug-fixes. wanna see? :D
Changelog (v2.1.0)
- Added support for external brains (brains to be placed inside /brains directory)
- Added standard metadata for every
.tqmbrain - Added more aliases
- Improved the
main.tqmbrain - Added a brain selector (runs on startup)
- Added a compatibility checker so newer brains don’t break inside an older engine
- Added a licence checker (cuz why not)
- Added a startup greeting function (not exactly but it greets you now)
- Fixed a few syntax issues and bugs
I think that’s it! I’m looking forward to making this drastically better and maybe add a webUI or create a webpage so we’ll see! :D
@tanzeemrockz
on
DQM
1h 2m 24s logged
guys, i just fixed a few bugs now. wanna see what i did? :D
Changelog (v2.0.1):
- Fixed punctuation engine in
nodes.py - Added more aliases
- Fixed a few things so each leaf has more than one response and use
random.choice()to pick one - Expanded
main.tqmdrastically (that was done in the last commit itself but i didn’t write a devlog lol) - Resolved issue where if you were outside the project directory, the brain won’t load but now it will.
- Added more attention to detail and now it shows you info about the brain, etc…
I guess that’s it! :D
@tanzeemrockz
on
DQM
46m 2s logged
guys so i just added a punctuation system that translates “who’s” to “who is” and so on and added a few responses to every tree so if you say “hi” first time it’ll give you a response and the next time, it’ll give you another response by importing random. I have also made the “brain” (main.tqm) HUGE with a LOT of data (the brain was written by gemini so shout out flash 3.8!) and that’s it for now!
@tanzeemrockz
on
DQM
1h 30m logged
Sup folks! So this is the first devlog for this project. This is DQM otherwise known as Deterministic Query Model. So basically this is an old technique where you split a prompt into words and compare against a tree file (.tqm but it’s just a json file lol) and goes thru. E.g. “who is max verstappen” goes thru “who” > “is” > “max” > “verstappen” then finds the answer inside and tells you it. This is all hardcoded and you can change the stuff inside these so-called “brains”. This project is made in python (i’m a newbie and i’m learning along the way) and I used google gemini to write all the “brains” because it’s way too much work for my “potato brain” lol. So python was really sweet and looking forward to adding more features. Cya!
