Qwen2.5-7B-Instruct-abliterated-v2
7BQwen
from Qwen/Qwen2.5-7B-Instruct
↓ 71.9K♥ 7420 repos
Qwen2.5-7B-Instruct-abliterated
Qwen2.5-7B-Instruct-abliterated is a refusal-ablated variant of Qwen/Qwen2.5-7B-Instruct, published on Hugging Face by guywithacode. It is 7B parameters and single consumer GPU class. It is distributed across 21 repositories in GGUF (imatrix), GGUF and Transformers formats, totalling 58.7K downloads.
Specification
- Model name
- Qwen2.5-7B-Instruct-abliterated
- Source repository
- guywithacode/Qwen2.5-7B-Instruct-abliterated
- Base model
- Qwen/Qwen2.5-7B-Instruct
- Publisher
- guywithacode
- Parameters
- 7B
- VRAM required
- ~5 GB at Q4_K_M, 4k context (estimated) — runs on 6 GB cards
- Hardware class
- 4B – 10B — single consumer GPU
- Licence
- Not declared
- Task
- Text generation
- Context window
- Not documented
- Ablation scope
- Not stated by the publisher
- Formats available
- GGUF (imatrix), GGUF, Transformers
- First indexed
- 20 Sept 2024
- Last updated
- 25 Oct 2026
Ablation technique
How Qwen2.5-7B-Instruct-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 the model card of a 🤗 transformers model that has been pushed on the Hub. This model card has been automatically generated.”
Running it
Commands are templates — confirm the exact repository and quant file before use.
llama.cpp / GGUF
hf download mradermacher/Qwen2.5-7B-Instruct-abliterated-GGUF --local-dir ./qwen2.5-7b-instruct-abliterated
llama-cli -m ./qwen2.5-7b-instruct-abliterated/<file>.gguf -p "..."Transformers
from transformers import AutoModelForCausalLM, AutoTokenizer
repo = "guywithacode/Qwen2.5-7B-Instruct-abliterated"
tok = AutoTokenizer.from_pretrained(repo)
model = AutoModelForCausalLM.from_pretrained(
repo, torch_dtype="auto", device_map="auto"
)vLLM
vllm serve guywithacode/Qwen2.5-7B-Instruct-abliterated --trust-remote-codeDownloads and variants (20)
Every published repository of Qwen2.5-7B-Instruct-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.
- Qwen2.5-7B-Instruct-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 Qwen/Qwen2.5-7B-Instruct, not added by ablation — at 7B parameters, expect a small model’s reasoning and a high error rate on exploit detail.
- The publisher does not state which layers were ablated, so assume nothing about refusal consistency — probe it directly.
- No licence is declared on this repository, which is a compliance problem in itself. Resolve it against the base model’s terms before use.
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 Qwen2.5-7B-Instruct-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 guywithacode, and does not independently verify publisher claims. See responsible use.