← back to catalog · registered 2026-11-21 09:04

HirCoir/Qwen3.8-27B-Uncensored

HirCoir 27B GGUF multimodal second-order
Your rig guess connected
? Why do I need an app?
Reading your rig…

This is a rough estimate. Install the free app - we'll show exact numbers.

Reading real hardware from your app right now. Numbers below are exact.

Below is the per-quantization compatibility for this model.

curl -H "Authorization: Bearer $ABL_KEY" \
     "https://abliteration.org/api/v1/models/HirCoir%2FQwen3.8-27B-Uncensored"
Response includes
  • classification m-uncensored
  • files 7
  • author_summary 1 models
  • readme_text full
10 credits · hourly refresh · ~4 KB payload Get an API key →
Abliteration classifier · v1.0.0
M-U
Primary method

Uncensored (method unknown)

No other method signals detected in this model.
Confidence
LOW
Why this label 3 signals
Weak or ambiguous signals. Best guess based on catalog patterns; treat as tentative and check the evidence below.
  • 'uncensored' in name/tags but no 'abliterated' marker
  • method not identifiable from author declaration alone
  • may be DPO fine-tune, prompt engineering, or unknown technique
Refusal direction extraction

No specific extraction method could be identified for this model. The producer either did not document it or used a proprietary pipeline.

What is a refusal direction? →
Downloads · 30-day
0
Likes
0
Model age
1d ago
created 2026-11-21
Downloads over time
Now 0 → from 0 ↑ 0%
0011 0 on Nov 21 0 on Nov 22 Nov
Nov 21 → Nov 22 · 2 snapshots · spans 1 day

Genealogy 0 direct forks

Full fork graph →

This model's place in the market. Above: what it was derived from. Below: the tree of everything derived from it.

Metadata

License
apache-2.0
Quantizations
IQ4
Tags
gguf llama.cpp vision qwen3 base_model:orcarouter/Qwen3.8-27B-Uncensored-GGUF base_model:quantized:orcarouter/Qwen3.8-27B-Uncensored-GGUF license:apache-2.0 endpoints_compatible region:us imatrix conversational

Related

Total size
14.3 GB
Files
7
Quantizations
3
Registered
2026-11-21 09:04
Last updated on HF
2026-11-21 07:40

Files by quantization

IQ4 1 file 14.3 GB
Qwen3.8-27B-Uncensored-IQ4_XS.gguf 14.3 GB c52df3e4 download
F16 1 file 888 MB
mmproj-Qwen3.8-27B-Uncensored-f16.gguf 888 MB add205b7 download
Auxiliary files 5 files 17.2 KB
Qwen3.8-27B-Uncensored.ipynb 4.41 KB f9901af6 download
.gitattributes 4.27 KB 5cd24b1a download
tunnel.py 3.59 KB 44385071 download
README.md 2.74 KB dab5d21e download
start-qwen.sh 2.21 KB 96eabf97 download

README current version from Hugging Face


license: apache-2.0
base_model: orcarouter/Qwen3.8-27B-Uncensored-GGUF
tags:

  • llama.cpp
  • gguf
  • vision
  • qwen3

Qwen3.8-27B Uncensored — llama.cpp + visión

Paquete listo para correr Qwen3.8-27B Uncensored (IQ4_XS) con llama.cpp en 2x Tesla T4, con entrada de imagen (mmproj) y chat web.

Contenido

Archivo Descripción Tamaño
Qwen3.8-27B-Uncensored-IQ4_XS.gguf Modelo Qwen3.8-27B abliterado (sin refusals), cuantizado IQ4_XS 15.3 GB
mmproj-Qwen3.8-27B-Uncensored-f16.gguf Proyector de visión, permite entrada de imagen 931 MB
llama.cpp/ Build b12600 con CUDA 12.8 + runtime incluido (libcudart, libcublas) ~1.1 GB
start-qwen.sh Script que arranca el servidor con la config probada
tunnel.py Crea túneles públicos *.gradio.live con el binario frpc de Gradio, sin depender de servicios externos
Qwen3.8-27B-Uncensored.ipynb Cuaderno que hace todo: descarga, arranca el servidor, el chat y los túneles
qwen-chat/app.py Chat web con Gradio, historial persistente

Uso rápido

bash start-qwen.sh

El servidor queda en http://127.0.0.1:8080 (API compatible con OpenAI en /v1) y el chat web en http://127.0.0.1:7860.

Túneles públicos

start-qwen.sh levanta automáticamente un túnel *.gradio.live para cada puerto con el binario frpc de Gradio, así el servidor queda público sin depender de servicios externos. Las URLs quedan en tunnel-8080.log y tunnel-7860.log (línea TUNNEL_URL=...). Si un túnel se cae, el script genera otro solo y muestra la nueva URL. No imprime logs, solo la URL.

También se pueden lanzar a mano:

python3 tunnel.py 8080   # API de llama.cpp
python3 tunnel.py 7860   # chat web

Configuración del servidor

  • Contexto: 131072 tokens
  • KV cache: q8_0
  • MTP (decodificación especulativa) activado, --spec-draft-n-max 2
  • Reasoning: off
  • Visión: mmproj cargado
  • GPUs: 2x Tesla T4 (15 GB c/u), -ngl 99

API

curl http://127.0.0.1:8080/v1/chat/completions \
  -H 'Content-Type: application/json' \
  -d '{
    "model": "qwen3.8-27b-uncensored-iq4_xs",
    "messages": [{"role": "user", "content": "Hola"}]
  }'

Para imágenes, envía el contenido como image_url dentro del mensaje. La API lo acepta gracias al mmproj.

Origen

Aviso

Este modelo tiene la alineación de seguridad removida y responde a peticiones que el modelo base rechazaría. Úsalo solo para fines legítimos, como auditar tus propios proyectos o investigación.

README history 4 versions

The author's README evolved over time. Click a version to see its content at that point.

  1. 2026-11-21 Upload README.md with huggingface_hub ffe03be 2.7 KB
    Loading...
  2. 2026-11-21 Upload README.md with huggingface_hub 8e42521 2.6 KB
    Loading...
  3. 2026-11-21 Upload README.md with huggingface_hub f83f6b7 2.5 KB
    Loading...
  4. 2026-11-21 Upload README.md with huggingface_hub 98655e0 2 KB
    Loading...
Catalog is the map. Apps are the tools.

Run models on your own machine, not in the cloud.

Every model page has an "Open in Abliteration" button that hands the model directly to the first-party desktop client, at the quantization your rig can actually run. No API keys, no subscription, no prompt leakage.

Open in Abliteration