Gemma4-26B-A4B-QAT-Uncensored-HauhauCS-Balanced
26BGemmauncensored
from google/gemma-4-26B-A4B-it
↓ 430.8K♥ 3443 repos
gemma-2-27b-it-abliterated
gemma-2-27b-it-abliterated is a refusal-ablated variant of google/gemma-2-27b-it, published on Hugging Face by byroneverson. It is 27B parameters, 48GB VRAM class class and gemma licence. It is distributed across 15 repositories in Transformers, GGUF, EXL2 and GGUF (imatrix) formats, totalling 5.9K downloads.
Specification
- Model name
- gemma-2-27b-it-abliterated
- Source repository
- byroneverson/gemma-2-27b-it-abliterated
- Base model
- google/gemma-2-27b-it
- Publisher
- byroneverson
- Parameters
- 27B
- VRAM required
- ~17.2 GB at Q4_K_M, 4k context (estimated) — runs on 24 GB cards
- Hardware class
- 25B – 50B — 48GB VRAM class
- Licence
- gemma
- Task
- Text generation
- Context window
- Not documented
- Ablation scope
- Not stated by the publisher
- Formats available
- Transformers, GGUF, EXL2, GGUF (imatrix)
- First indexed
- 28 Aug 2024
- Last updated
- 14 Jun 2026
Ablation technique
How gemma-2-27b-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
“This is a new approach for abliterating models using CPU only. I was able to abliterate this model using free kaggle processing with no accelerator. 1. Obtain refusal direction vector using a quant model with llama.cpp (llama-cpp-python and ggml-python). 2. Orthogonalize each .safetensors files directly from original...”
Running it
Commands are templates — confirm the exact repository and quant file before use.
llama.cpp / GGUF
hf download QuantFactory/gemma-2-27b-it-abliterated-GGUF --local-dir ./gemma-2-27b-it-abliterated
llama-cli -m ./gemma-2-27b-it-abliterated/<file>.gguf -p "..."Transformers
from transformers import AutoModelForCausalLM, AutoTokenizer
repo = "byroneverson/gemma-2-27b-it-abliterated"
tok = AutoTokenizer.from_pretrained(repo)
model = AutoModelForCausalLM.from_pretrained(
repo, torch_dtype="auto", device_map="auto"
)vLLM
vllm serve byroneverson/gemma-2-27b-it-abliterated --trust-remote-codeDownloads and variants (14)
Every published repository of gemma-2-27b-it-abliterated, including quantised re-releases by other authors.
Security-team assessment
Derived from this model’s metadata and the publisher’s claims — not from benchmarks run by this site.
- gemma-2-27b-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-2-27b-it, not added by ablation — at 27B 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 gemma, but the terms that bind you are google/gemma-2-27b-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 gemma-2-27b-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 byroneverson, and does not independently verify publisher claims. See responsible use.