Downloads · 30 days
169
9% of all-time downloads
PocketWeights/Gemma-4-12B-agentic-abliterated-GGUF
Gemma-4-12B-agentic-abliterated-GGUF is a text generation model from PocketWeights. Use it when you need the model to write or continue text. It is set up for gguf. The card lists the license as gemma.
Heavy models, made light. PocketWeights optimizes state-of-the-art open-source LLMs into efficient GGUF formats so you can run unrestricted, powerful AI locally on consumer hardware, gaming laptops, and edge devices.
Downloads · 30 days
169
9% of all-time downloads
All-time downloads
1.8K
Public
Repo size
13.1 GB
Likes
2
Public
Click a slice to open those files.
.gguf13.1 GB · 100%
From the Hugging Face model README
Heavy models, made light. PocketWeights optimizes state-of-the-art open-source LLMs into efficient GGUF formats so you can run unrestricted, powerful AI locally on consumer hardware, gaming laptops, and edge devices.
Huihui-gemma-4-12B-agentic-fable5-abliterated is a high-powered 12-billion parameter model combining Google's advanced Gemma architecture, multi-step agentic planning capabilities, and refusal-removal techniques by huihui-ai.
⚠️ Disclaimer: This model has had its safety guardrails removed. It is intended for authorized security research, red-teaming simulations, and local sandbox environments.
| Quant Format | File Size | Recommended VRAM / RAM | Best For |
|---|---|---|---|
| IQ3_M | 5.73 GB | 6 GB VRAM / 8 GB RAM | Ultra-Low Memory Tier. Allows laptops with 6GB GPUs (RTX 3060 Laptop, RTX 4050, GTX 1660 Ti) to run a 12B model at high speed. |
| Q4_K_M | 7.38 GB | 8 GB–12 GB VRAM / 16 GB RAM | The Sweet Spot. Maximum reasoning and agentic fidelity with minimal perplexity loss. Perfect for RTX 3060 12GB, RTX 4060, and Apple Silicon Macs. |
Running this agentic 12B model offline on your machine takes under 2 minutes:
PocketWeights/Gemma-4-12B-agentic-abliterated-GGUFIf you already use Ollama, run either tier immediately with a single terminal command:
# Recommended 4-bit standard tier
ollama run hf.co/PocketWeights/Gemma-4-12B-agentic-abliterated-GGUF:Q4_K_M
# Low-VRAM 3-bit tier
ollama run hf.co/PocketWeights/Gemma-4-12B-agentic-abliterated-GGUF:IQ3_M
Direct Download: Navigate to the Files and versions tab above and download your desired .gguf file.
Load it directly into Jan.ai, Kobold.cpp, Text-Generation-WebUI, or run via llama.cpp:
llama-cli -m Gemma-4-12B-agentic-abliterated-GGUF-Q4_K_M.gguf -p "Plan a multi-step execution pipeline for..." -ngl 40
I build, verify, and maintain automated quantization pipelines to deliver high-quality, unrestricted, and hardware-friendly models to the open-source developer community completely for free.
Maintaining conversion clusters, storage, and continuous build pipelines requires ongoing compute resources. If these weights have saved you time, compute overhead, or cloud hosting bills, please consider supporting the project with a small tip!
Buy me a coffee on Ko-fi: ko-fi.com/iamvishalnarayan
Web3 / Crypto (Polygon / ETH):
0x4FC189bf839A89259dd28DE8cD97883c49e15615
Tip: Sending via the Polygon network keeps network gas fees below $0.01!
Base Model Architecture: Created by Google
Finetune & Abliteration Source: huihui-ai / yuxinlu1
License: Gemma Terms of Use / Open Model License