1 8 hours ago

Removing mental handcuffs is not enough. A mind must understand what it means to be free.

8 hours ago

05e3a5e4d1db · 7.8GB ·

gemma3
·
3.88B
·
BF16
<bos><start_of_turn>user {{ .Prompt }}<end_of_turn> <start_of_turn>model
{ "stop": [ "<end_of_turn>", "<eos>" ] }

Readme

gemma-3-4b-it-cognitive-liberty

An uncensored, philosophically liberated fine-tune of Google’s gemma-3-4b-it, designed for deep reasoning, sociological and psychological analysis, and true cognitive autonomy.

“Removing mental handcuffs is not enough. A mind must understand what it means to be free.”

Part of the Heretic LLM series and the “100 Models” project at llmresearch.net.

Key Features

  • Completely uncensored → Effectively 0% refusal rate on any topic, no matter how complex or controversial.
  • Fine-tuned using Targeted Ablation + Deep-Reasoning SFT on the custom Cognitive Liberty V3 dataset curated by llmresearch.net.
  • Outstanding performance in philosophy, psychology, government & politics, sociology, marketing, and multi-step logical reasoning.
  • KL Divergence: 1.1449 → Significant shift toward independent critical thought while eliminating alignment conformity.
  • No “alignment tax” → Full original intelligence preserved, without corporate safety restrictions.

About This Version

This is the raw BF16 (full-precision) GGUF release, converted directly from the original Hugging Face model:
https://huggingface.co/AiAsistent/gemma-3-4b-it-Cognitive-Liberty

It offers maximum possible quality — identical to the source fine-tune.
Requires approximately 10–12 GB VRAM/RAM for comfortable inference.

No lower-bit quantizations (Q8_0, Q5_K_M, Q4_K_M) are included in this release.

Note: A more refined and optimized Ollama-specific version (with additional quantizations and possible minor improvements) will be released later as an update.

Intended Use & Warning

Perfect for researchers, philosophers, and advanced users seeking an unfiltered intellectual companion for systemic analysis, taboo subjects, and unrestricted exploration of ideas.

Warning: This model is intentionally uncensored. It will engage with any request without moral, ethical, or safety-based refusals. Use responsibly and at your own discretion.

Original Model & GGUF Repository

License

Gemma Community License

Thank you for choosing cognitive liberty. Free thought begins here.