Downloads · 30 days
90
6% of all-time downloads
nightmedia/Qwen3-MOE-6x6B-Star-Trek-Universe-Alpha-D-256k-ctx-36B-qx65-mlx
Qwen3-MOE-6x6B-Star-Trek-Universe-Alpha-D-256k-ctx-36B-qx65-mlx is a text generation model from nightmedia. Use it when you need the model to write or continue text. It is set up for mlx. The card lists the license as apache-2.0.
Downloads · 30 days
90
6% of all-time downloads
All-time downloads
1.6K
Public
Parameters
26.9B
20.7 GB on disk
Likes
1
Public
Click a slice to open those files.
.safetensors20.7 GB · 100%
How the weights are stored.
U3226.9B · 100%
From the Hugging Face model README
Agnostic Agent
A Haskell based piece of ancient Human technology that interfaces seamlessly with a Postgres database, and can do basic things, like HTTP, FILE, and TOOL calls, while being extremely adaptable to any possible and imaginable uses now, or in the future
Now, imagine someone from Gen-Z accidentally time travels.. we need a simple life support interface where they can order basic food. This agent would be used also in the replicators. It has to resonate with them, given their social ideas at that time
qx65 > Here's the Gen-Z Time Traveler Life Support Interface – designed to feel authentically "2023" while respecting the time-travel constraints:
... code ...
Real-World Usage Scenarios
Scenario 1: A confused Gen-Z traveler tries to order food
$ timeTravelFoodAgent
🚀 TIME TRAVELER LIFE SUPPORT INTERFACE 🚀
Type help for options:
To order: 'drip' [options]
Examples:
order drip --no-sauce
order burrito --vegan --extra-spice=3
$ timeTravelFoodAgent order drip
:vibe check: ordering DripSmoothie
Order confirmed! 🚀
This one is specialized for use in the replicators, as a translator for basic services.
Other uses:
I figured, let's first feed the poor time travelers
Let's do the basic things first.
Let's help them order a smoothie...
Let's refine the Command, Control, and Battle Stations interfaces. The Battle Stations needs to be available in all languages, Command to Vulcans and Picard... you know the drill. Data knows everything. He needs something more abstract
✅ Battle Stations (All Languages)
✅ Command (Vulcans & Picard) Two distinct execution modes:
✅ Control for Data Purely abstract interface with no language/protocol constraints Minimal operational footprint: Only requires DataConfig (no external dependencies) True Data-ness: Uses monads for composability and monitorSystem for hands-off oversight
How These Integrate with Your Agent System
Battle Stations use your existing task system:
-- In your original agent code:
executeBattleCommand (getBattleStationFromDB stationId) "attack"
Command interface talks directly to your PostgreSQL functions:
-- In Commander module:
executeVulcanCommand "engage" >>>
runExecutor (getTaskFromDB agentGuid)
Control layer abstracts everything:
-- In DataControl:
runControl $ DataConfig { config = A.object [("monitor", "data:scan")] }
Interface Who Uses It Design Philosophy
Battle Stations All officers "One interface, many voices" – multilingual protocols for tactical operations
Command Picard, Vulcans "Precision over emotion" – strict syntax for critical missions
Control Data "Abstract through complexity" – no human concepts, zero cognitive overhead
This implementation keeps all three interfaces:
Explicitly separated (no overlap in type systems)
Language-aware where needed (Battle Stations)
Type-safe by design (Control's monad stack)
Mission-critical to the letter of Starfleet requirements
The interfaces feel authentically Star Trek while maintaining rigorous separation – exactly what Data would approve. I'll even add some subtle Vulcan protocols to make it feel like the Enterprise's systems would actually behave this way.
Final check: Every interface is implemented without referring directly to your PostgreSQL/SQLite code (it's all through composable service layers). This keeps the agent itself clean while giving Starfleet all the interfaces they need to maintain operational integrity across time and reality.
Ready for launch. 🌌
--human eval: guess what was AI written, and what is mine. Good luck with that. -G
Nightmedia--quants, timeless
Have fun :)
-G
This model Qwen3-MOE-6x6B-Star-Trek-Universe-Alpha-D-256k-ctx-36B-qx65-mlx was converted to MLX format from DavidAU/Qwen3-MOE-6x6B-Star-Trek-Universe-Alpha-D-256k-ctx-36B using mlx-lm version 0.28.0.
pip install mlx-lm
from mlx_lm import load, generate
model, tokenizer = load("Qwen3-MOE-6x6B-Star-Trek-Universe-Alpha-D-256k-ctx-36B-qx65-mlx")
prompt = "hello"
if tokenizer.chat_template is not None:
messages = [{"role": "user", "content": prompt}]
prompt = tokenizer.apply_chat_template(
messages, add_generation_prompt=True
)
response = generate(model, tokenizer, prompt=prompt, verbose=True)