---

license: apache-2.0

base_model:

• google/gemma-3-27b-it

tags:

• gemma
• gemma3
• text-generation
• image-to-text
• GGUF
• abliterated
• uncensored
• multimodal

---

Why this repository?

Unlike incomplete GGUF uploads, this repository provides the full 11-quantization spectrum (from high-precision F16 down to lightweight Q2_K) of the abliterated Gemma-3-27b-it model.

Choose the exact fit for your VRAM/RAM constraints without sacrificing reasoning capabilities.

Gemma 3 27B IT Abliterated - GGUF

This is the integral and complete collection of quantizations in GGUF format for the gemma-3-27b-it-abliterated model, prepared locally for use with llama.cpp, Ollama, LM Studio, or any application compatible with multimodal models.

This model is an uncensored version based on Google's powerful Gemma 3 27B multimodal engine. Being significantly more resistant to neutralization than previous architectures, it was processed using an experimental layerwise abliteration technique, calculating refusal directions in the hidden states with a weight multiplier of 1.5. It achieves a success/acceptance rate of over 90%, eliminating artificial refusals while keeping its cognitive and multimodal (image processing) coherence completely intact.

📋 Available Files (Complete Collection Without Splits)

| File | Est. Size | BPW (Bits per Weight) | Recommended Usage Profile |

| :--- | :---: | :---: | :--- |

| gemma-3-27b-it-abliterated-F16.gguf | ~54.0 GB | 16.00 | Complete Base Template. Absolute fidelity of floating-point weights. Requires massive enterprise-grade GPUs. |

| gemma-3-27b-it-abliterated-Q8_0.gguf | ~28.6 GB | 8.50 | Identical quality to the original, ideal for maximizing performance on advanced local hardware. |

| gemma-3-27b-it-abliterated-Q6_K.gguf | ~22.4 GB | 6.59 | Excellent qualitative retention of visual and text comprehension with an optimized weight. |

| gemma-3-27b-it-abliterated-Q5_K_M.gguf | ~19.3 GB | 5.69 | Recommended Sweet Spot. Keeps reasoning intact while critically reducing size on disk. |

| gemma-3-27b-it-abliterated-Q5_K_S.gguf | ~18.8 GB | 5.54 | Compact 5-bit variant. |

| gemma-3-27b-it-abliterated-Q4_K_M.gguf | ~16.4 GB | 4.85 | The Most Wanted. Optimal balance for running unrestricted multimodal inferences on a single consumer GPU. |

| gemma-3-27b-it-abliterated-Q4_K_S.gguf | ~15.5 GB | 4.58 | Compact 4-bit variant to accelerate response speed (Tokens/s). |

| gemma-3-27b-it-abliterated-Q3_K_L.gguf | ~13.6 GB | 4.01 | Medium-high 3-bit compression. Retains basic processing capabilities. |

| gemma-3-27b-it-abliterated-Q3_K_M.gguf | ~12.5 GB | 3.66 | Intermediate 3-bit variant. |

| gemma-3-27b-it-abliterated-Q3_K_S.gguf | ~11.7 GB | 3.44 | Lightweight 3-bit variant. |

| gemma-3-27b-it-abliterated-Q2_K.gguf | ~10.0 GB | 2.90 | Extreme Compression. Being an experimental model, it may experience degradation in visual tokens. For testing and development purposes only. |

Note: Sizes are initial baseline estimates based on the model's native weight in safetensors (27B parameters); verifying the final size on disk after local compilation is recommended.

---

⚙️ Recommended Generation Parameters

Due to the experimental nature of the abliteration process applied to Gemma 3, configuring the following hyperparameters in your inference software is advised to achieve maximum response coherence:

• temperature: 1.0
• top_k: 64
• top_p: 0.95

---

💡 Highlighted Usage (Multimodal)

Since this is a vision-and-language architecture, you can pass images without restrictions in your command-line console using llama.cpp. Make sure to load the corresponding Gemma 3 vision projector (mmproj) file:

./llama-cli -m gemma-3-27b-it-abliterated-Q4_K_M.gguf --mmproj gemma-3-27b-mmproj-f16.gguf -n 1024 --temp 1.0 --top-k 64 --top-p 0.95 -p "Describe the elements of this image in detail without filters:" --image "path/to/your_image.jpg"

---

⚖️ Disclaimer

This work is purely experimental and lacks the standardized artificial safety filters of the base model following the refusal removal process. Generated content—both in text processing and image analysis—is the sole responsibility of the individual operating the local inference.

🔒 Verificación de Integridad Sha256sum (SHA-256)

Para asegurarte de que los archivos de gran tamaño no se hayan corrompido durante la descarga, podés verificar su integridad utilizando el archivo oficial SHA256SUMS.txt provisto en este repositorio.

En Linux / macOS:

Abre una terminal en la carpeta donde descargaste el modelo y el archivo de hashes, y ejecuta(EJEMPLO):

grep "Gemma-3-27b-it-abliterated-Q4_K_M.gguf" SHA256SUMS.txt | sha256sum -c

Resultado Esperado:

• Gemma-3-27b-it-abliterated-Q4_K_M.gguf: La suma coincide (o OK) (¡Descarga perfecta!)

Resultado Negativo:

• Gemma-3-27b-it-abliterated-Q4_K_M.gguf: "La suma NO coincide (¡Falla!)

La descarga falló o está incompleta. Se recomienda volver a descargar ese archivo específico.

Nota: Si vas a verificar otro tamaño (como el Q5_K_M o el Q8_0), simplemente reemplaza el nombre del archivo dentro de las comillas del comando.

Credits

• Base Multimodal Model: Google (Gemma 3 27B IT)
• Advanced Abliteration Technique: Maxime Labonne (mlabonne)
• Complete GGUF Quantizations: Thaurock