gemma-3-27b-instruct-uncensored
vision
1,568 Pulls Updated 2 weeks ago
Readme
🧠 nidum-gemma-3-27B-Instruct-Uncensored (GGUF)
This repository provides the GGUF-quantized versions of nidum-gemma-3-27B-Instruct-Uncensored for use with Ollama and other GGUF-compatible backends.
📌 Available Quantized Models
The following quantized versions are available:
| Model Variant | Size |
|---|---|
q8_0 |
~28GB |
q6_k |
~21GB |
q5_k_m |
~18GB |
q3_k_m |
~12GB |
🚀 Installation
Step 1: Install Ollama
If you haven’t installed Ollama, do so using:
curl -fsSL https://ollama.com/install.sh | sh
⚡ Usage
Run a Quantized Model
After installing Ollama, run the model directly:
ollama run nidumai/nidum-gemma-3-27b-instruct-uncensored:q8_0
To use another quantization, replace q8_0 with q6_k, q5_k_m, or q3_k_m:
ollama run nidumai/nidum-gemma-3-27b-instruct-uncensored:q6_k
📥 Downloading the Model Locally
If you prefer to download and store the models locally:
ollama pull nidumai/nidum-gemma-3-27b-instruct-uncensored:q6_k
This will store the model on your system for offline use.
📜 Model Details
- Base Model:
Gemma-3-27B - Training Data: Multi-turn conversational AI
- Uncensored: Designed for unrestricted responses
- Inference Engine: Optimized for GGUF-compatible backends
⚙️ Advanced Configuration
You can adjust temperature and stopping conditions when running:
ollama run nidumai/nidum-gemma-3-27b-instruct-uncensored:q8_0 --temperature 0.7 --stop "<end_of_turn>"