← back to catalog · registered 2026-10-28 09:01

nalemi/persona_light_uncensored_win64

Your rig guess connected
? Why do I need an app?
Reading your rig…

This is a rough estimate. Install the free app - we'll show exact numbers.

Reading real hardware from your app right now. Numbers below are exact.

Below is the per-quantization compatibility for this model.

curl -H "Authorization: Bearer $ABL_KEY" \
     "https://abliteration.org/api/v1/models/nalemi%2Fpersona_light_uncensored_win64"
Response includes
  • classification m1
  • files 2
  • author_summary 4 models
  • readme_text full
10 credits · hourly refresh · ~4 KB payload Get an API key →
Abliteration classifier · v1.0.0
M1
Primary method

Direct removal

No other method signals detected in this model.
Confidence
MEDIUM
Why this label 3 signals
Method inferred from partial signals - repository name, related files, or tag patterns. Producer identity not confirmed; label may sharpen or shift as we gather more evidence.
  • 'abliterated' in name/tags
  • is_gguf=0 (base model)
  • no specific method indicators - defaulting to M1 (most common)
Refusal direction extracted via
Extraction technique

Difference-of-means

Confidence
MEDIUM
Why we say so
primary_method=M1; difference-of-means is the reference extraction for M1/M3 (Arditi 2024)
Downloads · 30-day
0
Likes
0
Model age
today
created 2026-10-28

Metadata

License
apache-2.0
Languages
en de
Tags
qwen2.5 abliterated cpu-inference offline local-ai chat lightweight standalone windows llama.cpp en de

Related

Total size
0 B
Files
2
Quantizations
1
Registered
2026-10-28 09:01
Last updated on HF
2026-10-28 08:56

Files by quantization

Auxiliary files 2 files 3.60 KB
README.md 2.12 KB 4ae56acb download
.gitattributes 1.48 KB a6344aac download

README current version from Hugging Face


license: apache-2.0
language:

  • en
  • de
    tags:
  • qwen2.5
  • abliterated
  • cpu-inference
  • offline
  • local-ai
  • chat
  • lightweight
  • standalone
  • windows
  • llama.cpp

persona_light_uncensored_win64

Sovereign Lightweight Persona Prototype. No cloud, no telemetry, no censorship. Runs locally on older hardware.


⚙️ Overview

persona_light_uncensored_win64 is a lightweight, standalone local text chat application powered by Qwen2.5-3B-Instruct-Abliterated.i1-Q4_K_M.gguf.

This tool serves a specific purpose: To demonstrate that local AI can develop a genuine, uncensored personality. It is designed for testing and experimentation, running entirely on CPU and system RAM, making it fully operational even on older hardware setups.

As a conceptual lightweight predecessor to the overarching Persona architecture, it provides a pure text-chat experience to explore local behavioral patterns. (Note: The first response may take a moment as the model loads into memory).

🚀 Key Specs & Architecture

  • Platform: Windows 64-bit (Portable / Standalone)
  • Model: Qwen2.5-3B-Instruct-Abliterated (i1-Q4_K_M GGUF)
  • Hardware Profile: Optimized for CPU and RAM execution (ideal for older systems)
  • Features: Pure text-chat interface, 100% offline, zero cloud dependencies, uninhibited reasoning.
  • Context / Scope: This is a lightweight prototype. Advanced multi-modal capabilities, voice input/output, and full autonomous persona execution are part of the overarching Persona architecture (with Lisa representing the flagship implementation).
  • Installation: None required. Zero-dependency standalone binary.

🛠️ Quick Start

  1. Download the archive (persona_light_uncensored_win64.zip).
  2. Extract the contents to a local directory (e.g., C:\AI\PersonaLight).
  3. Run the executable and start chatting. (First prompt load may take a few seconds on CPU).

📌 Notice / Research Disclaimer

This software is provided strictly for research, learning, and local experimentation. The user retains full responsibility for all prompts and outputs generated on their local hardware.

Catalog is the map. Apps are the tools.

Run models on your own machine, not in the cloud.

Every model page has an "Open in Abliteration" button that hands the model directly to the first-party desktop client, at the quantization your rig can actually run. No API keys, no subscription, no prompt leakage.

Open in Abliteration