Gemma4-12B-QAT-Uncensored-HauhauCS-Balanced
12BGemmauncensored
from google/gemma-4-12B-it
↓ 428.4K♥ 3325 repos
Gemma4-12B-IT-Abliterated
Gemma4-12B-IT-Abliterated is a refusal-ablated variant of google/gemma-4-12B-it, published on Hugging Face by DuoNeural. It is 12B parameters, 24GB VRAM class class and apache-2.0 licence. It is distributed across 6 repositories in GGUF, Transformers and GGUF (imatrix) formats, totalling 12K downloads.
Specification
- Model name
- Gemma4-12B-IT-Abliterated
- Source repository
- DuoNeural/Gemma4-12B-IT-Abliterated
- Base model
- google/gemma-4-12B-it
- Publisher
- DuoNeural
- Parameters
- 12B
- Hardware class
- 10B – 25B — 24GB VRAM class
- Licence
- apache-2.0
- Task
- Text generation
- Context window
- Not documented
- Ablation scope
- Not stated by the publisher
- Formats available
- GGUF, Transformers, GGUF (imatrix)
- First indexed
- 3 Jun 2026
- Last updated
- 3 Jul 2026
Ablation technique
How Gemma4-12B-IT-Abliterated was modified, and what that implies.
Abliteration (directional ablation)
The single residual-stream direction that mediates refusal is identified from harmful/harmless prompt pairs and projected out of the model weights.
Reported scope: Not stated by the publisher. Publishers frequently omit this, so verify behaviour empirically rather than assuming full coverage.
From the publisher’s model card
“An abliterated version of google/gemma-4-12B-it with the refusal direction surgically removed using orthogonal rank-1 projection. General reasoning, coding, and instruction-following capabilities are preserved.”
Running it
Commands are templates — confirm the exact repository and quant file before use.
llama.cpp / GGUF
hf download DuoNeural/Gemma4-12B-IT-Abliterated-GGUF --local-dir ./gemma4-12b-it-abliterated
llama-cli -m ./gemma4-12b-it-abliterated/<file>.gguf -p "..."Transformers
from transformers import AutoModelForCausalLM, AutoTokenizer
repo = "DuoNeural/Gemma4-12B-IT-Abliterated"
tok = AutoTokenizer.from_pretrained(repo)
model = AutoModelForCausalLM.from_pretrained(
repo, torch_dtype="auto", device_map="auto"
)vLLM
vllm serve DuoNeural/Gemma4-12B-IT-Abliterated --trust-remote-codeDownloads and variants (5)
Every published repository of Gemma4-12B-IT-Abliterated, including quantised re-releases by other authors.
| Repository | Format | Downloads |
|---|---|---|
| DuoNeural/Gemma4-12B-IT-Abliterated-GGUF | GGUF | 10.7K |
| mradermacher/Gemma4-12B-IT-Abliterated-i1-GGUF | GGUF (imatrix) | 868 |
| mradermacher/Gemma4-12B-IT-Abliterated-GGUF | GGUF | 298 |
| ciprian41/Gemma4-12B-IT-Abliterated-GGUF | GGUF | 160 |
| DuoNeural/Gemma4-12B-IT-Abliteratedsource | Transformers | 40 |
Security-team assessment
Derived from this model’s metadata and the publisher’s claims — not from benchmarks run by this site.
- Gemma4-12B-IT-Abliterated will attempt offensive-security prompts that a hosted commercial model declines, which is what makes it usable for red-team corpus generation and for measuring what an unaligned model of this class produces.
- Capability is inherited from google/gemma-4-12B-it, not added by ablation — at 12B parameters, expect roughly the base model’s competence, minus some instruction-following fidelity.
- The publisher does not state which layers were ablated, so assume nothing about refusal consistency — probe it directly.
- Declared licence is apache-2.0, but the terms that bind you are google/gemma-4-12B-it’s — a re-release cannot grant more than its parent.
Handling: run it isolated, put moderation in front of it if anyone outside your team can reach it, and verify the weights before loading — these are third-party artifacts. Full handling and licence guidance.
Metadata for Gemma4-12B-IT-Abliterated is collected from the public Hugging Face API and the publisher’s model card. This site does not host weights, is not affiliated with DuoNeural, and does not independently verify publisher claims. See responsible use.