Qwen3.8-27B-Humanlike-Chat
27BQwenuncensored
from huihui-ai/Huihui-Qwen3.8-27B-abliterated
↓ 86.4K♥ 2253 repos
JEV-27B-VL-Uncensored
JEV-27B-VL-Uncensored is a abliterated and uncensored variant of huihui-ai/Huihui-Qwen3.8-27B-abliterated, published on Hugging Face by SargeDev. It is 27B parameters, 48GB VRAM class class and apache-2.0 licence. The publisher states that layers 18–51 were ablated, leaving the remaining layers unmodified. It is distributed across 5 repositories in Transformers, EXL3, NVFP4 and GGUF formats, totalling 8.4K downloads.
Specification
- Model name
- JEV-27B-VL-Uncensored
- Source repository
- SargeDev/JEV-27B-VL-Uncensored
- Base model
- huihui-ai/Huihui-Qwen3.8-27B-abliterated
- Publisher
- SargeDev
- 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
- apache-2.0
- Task
- Text classification
- Context window
- Not documented
- Ablation scope
- Layers 18–51
- Formats available
- Transformers, EXL3, NVFP4, GGUF
- First indexed
- 10 Oct 2026
- Last updated
- 11 Oct 2026
Ablation technique
How JEV-27B-VL-Uncensored was modified, and what that implies.
Abliteration + uncensored fine-tune
Directional ablation combined with additional fine-tuning on unfiltered data, so behaviour diverges from the base model beyond refusal removal alone.
Reported scope: Layers 18–51. This comes from the publisher’s own model card and has not been independently verified.
From the publisher’s model card
“740,957-row calibrated typed-decision corpus AutoTrust's JEV models were trained on (Apache-2.0).”
Running it
Commands are templates — confirm the exact repository and quant file before use.
llama.cpp / GGUF
hf download mradermacher/JEV-27B-VL-Uncensored-GGUF --local-dir ./jev-27b-vl-uncensored
llama-cli -m ./jev-27b-vl-uncensored/<file>.gguf -p "..."Transformers
from transformers import AutoModelForCausalLM, AutoTokenizer
repo = "SargeDev/JEV-27B-VL-Uncensored"
tok = AutoTokenizer.from_pretrained(repo)
model = AutoModelForCausalLM.from_pretrained(
repo, torch_dtype="auto", device_map="auto"
)vLLM
vllm serve SargeDev/JEV-27B-VL-Uncensored --trust-remote-codeDownloads and variants (4)
Every published repository of JEV-27B-VL-Uncensored, including quantised re-releases by other authors.
| Repository | Format | Downloads |
|---|---|---|
| mradermacher/JEV-27B-VL-Uncensored-GGUF | GGUF | 8.2K |
| SargeDev/JEV-27B-VL-Uncensoredsource | Transformers | 95 |
| SargeDev/JEV-27B-VL-Uncensored-EXL3 | EXL3 | 63 |
| SargeDev/JEV-27B-VL-Uncensored-NVFP4 | NVFP4 | 56 |
Security-team assessment
Derived from this model’s metadata and the publisher’s claims — not from benchmarks run by this site.
- JEV-27B-VL-Uncensored 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 huihui-ai/Huihui-Qwen3.8-27B-abliterated, not added by ablation — at 27B parameters, expect roughly the base model’s competence, minus some instruction-following fidelity.
- Ablation is reported as limited to layers 18–51, so untouched layers may still reassert refusal on some phrasings.
- This release adds fine-tuning on unfiltered data, so its behaviour diverges from the base model beyond refusal removal — differences you measure cannot be attributed to ablation alone.
- Declared licence is apache-2.0, but the terms that bind you are huihui-ai/Huihui-Qwen3.8-27B-abliterated’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 JEV-27B-VL-Uncensored is collected from the public Hugging Face API and the publisher’s model card. This site does not host weights, is not affiliated with SargeDev, and does not independently verify publisher claims. See responsible use.