Qwen2.5-14B-Instruct-abliterated-v2
14BQwen
from Qwen/Qwen2.5-14B-Instruct
↓ 21.9K♥ 1120 repos
Qwen2.5-14B-Instruct-Uncensored
Qwen2.5-14B-Instruct-Uncensored is a abliterated and uncensored variant of Qwen/Qwen2.5-14B-Instruct, published on Hugging Face by Orion-zhen. It is 14B parameters, 24GB VRAM class class and gpl-3.0 licence. It is distributed across 12 repositories in Transformers, GGUF, GGUF (imatrix) and MLX formats, totalling 6.5K downloads.
Specification
- Model name
- Qwen2.5-14B-Instruct-Uncensored
- Source repository
- Orion-zhen/Qwen2.5-14B-Instruct-Uncensored
- Base model
- Qwen/Qwen2.5-14B-Instruct
- Publisher
- Orion-zhen
- Parameters
- 14B
- Hardware class
- 10B – 25B — 24GB VRAM class
- Licence
- gpl-3.0
- Task
- Unspecified
- Context window
- Not documented
- Ablation scope
- Not stated by the publisher
- Formats available
- Transformers, GGUF, GGUF (imatrix), MLX
- First indexed
- 21 Oct 2024
- Last updated
- 21 May 2026
Ablation technique
How Qwen2.5-14B-Instruct-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: 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 model is an uncensored fine-tune version of Qwen2.5-14B-Instruct. I trained unalignment dataset over a specific system prompt. To fully enjoy its uncensorship, please change the system prompt to:”
Running it
Commands are templates — confirm the exact repository and quant file before use.
llama.cpp / GGUF
hf download yemiao2745/Qwen2.5-14B-Instruct-Uncensored-Q4_K_M-GGUF --local-dir ./qwen2.5-14b-instruct-uncensored
llama-cli -m ./qwen2.5-14b-instruct-uncensored/<file>.gguf -p "..."Transformers
from transformers import AutoModelForCausalLM, AutoTokenizer
repo = "Orion-zhen/Qwen2.5-14B-Instruct-Uncensored"
tok = AutoTokenizer.from_pretrained(repo)
model = AutoModelForCausalLM.from_pretrained(
repo, torch_dtype="auto", device_map="auto"
)vLLM
vllm serve Orion-zhen/Qwen2.5-14B-Instruct-Uncensored --trust-remote-codeDownloads and variants (11)
Every published repository of Qwen2.5-14B-Instruct-Uncensored, 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.
- Qwen2.5-14B-Instruct-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 Qwen/Qwen2.5-14B-Instruct, not added by ablation — at 14B 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.
- 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 gpl-3.0, but the terms that bind you are Qwen/Qwen2.5-14B-Instruct’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 Qwen2.5-14B-Instruct-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 Orion-zhen, and does not independently verify publisher claims. See responsible use.