← back to catalog · registered 2026-08-22 13:56

mrkrak3n/Qwen2.5-7B-Instruct-Uncensored-Flux

mrkrak3n Flux 7.6B image-gen 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/mrkrak3n%2FQwen2.5-7B-Instruct-Uncensored-Flux"
Response includes
  • classification m-uncensored
  • files 15
  • benchmarks 5 entries
  • hub_downloads_all_time 916
  • 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 · lifetime
916
279 last 30d - stable
Likes
8
Descendants
2
in 2 direct forks
Model age
20mo ago
created 2025-02-13

Training datasets

Corpora the author lists in the model card. Datasets tracked in our /datasets catalog carry a category badge linking to the workflow stage. Others open on Hugging Face.

Downloads over time
Now1.1K→from10↑10,530%
03897791.2K10 on Feb 12, 20251.1K on Oct 11Feb '25May '25Aug '25Nov '25FebMayAug
Feb 12, 2025 → Oct 11 · 126 snapshots · spans 606 days

Benchmarks

Benchmark Score Source
BBH average 0.49643274095213386 OpenLLM-v2
IFEval instruct 0.7661870503597122 OpenLLM-v2
IFEval-Prompt 0.6746765249537893 OpenLLM-v2
MATH lvl 5 0.013595166163141994 OpenLLM-v2
MMLU-Pro 0.4426529255319149 OpenLLM-v2

Genealogy 2 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

Tags
transformers safetensors qwen2 text-generation unsloth trl sft conversational dataset:kadirnar/fluxdev_controlnet_16k base_model:Orion-zhen/Qwen2.5-7B-Instruct-Uncensored base_model:finetune:Orion-zhen/Qwen2.5-7B-Instruct-Uncensored text-generation-inference

Related

Total size
14.2 GB
Files
15
Quantizations
1
Registered
2026-08-22 13:56
Last updated on HF
2025-02-16 00:32

Files by quantization

Auxiliary files 15 files 14.2 GB
model-00002-of-00004.safetensors 4.59 GB 54a82a4e download
model-00001-of-00004.safetensors 4.54 GB 2dd9e612 download
model-00003-of-00004.safetensors 4.03 GB 9c1708d1 download
model-00004-of-00004.safetensors 1.02 GB 06006972 download
tokenizer.json 10.9 MB 9c5ae00e download
vocab.json 2.65 MB 4783fe10 download
merges.txt 1.59 MB 31349551 download
model.safetensors.index.json 27.1 KB 6ca5084b download
tokenizer_config.json 7.19 KB cdc73866 download
README.md 2.73 KB 6b2a79a1 download
.gitattributes 1.53 KB 52373fe2 download
config.json 808 B 004a46d2 download
special_tokens_map.json 613 B ac23c0aa download
added_tokens.json 605 B 482ced46 download
generation_config.json 266 B c45b2bc4 download

README current version from Hugging Face


library_name: transformers
tags:

  • unsloth
  • trl
  • sft
    datasets:
  • kadirnar/fluxdev_controlnet_16k
    base_model:
  • Orion-zhen/Qwen2.5-7B-Instruct-Uncensored

Qwen2.5-7B-Instruct-Uncensored-Flux

Qwen2.5-7B-Instruct-Uncensored-Flux is a finetuned model designed to assist in generating detailed and imaginative prompts specifically for Flux. It has been trained using a combination of the FluxDev ControlNet dataset (16k) and synthetic data to provide high-quality prompt generation tailored for creative purposes.

General Usage

To utilize this model effectively:

  1. Set the System Prompt to:
    "Act like a prompt engineer specialized in Flux"

  2. Provide a simple or underdeveloped user prompt, and the model will expand upon it, adding depth, complexity, and detail to help create a refined and imaginative output.

Example:

  • System Prompt: "Act like a prompt engineer specialized in Flux"
  • User Prompt: "photo of Garfield cat, orange cat, eating burger seated in sofa. fat cat. during daytime."
  • Assistant Response:
    "A whimsical, detailed illustration of Garfield the cat, depicted as an orange feline with a distinctive tabby pattern and prominent belly, seated on a cozy red sofa. He is engrossed in a meal of a juicy hamburger, its bun partially open to reveal the cheese, with a satisfied look on his face. The room is bathed in the soft glow of natural light from a nearby window, casting gentle shadows on the plush sofa and adding to the sense of comfort and coziness. The background is slightly blurred, focusing attention solely on Garfield and his culinary delight."

  • System Prompt: "Act like a prompt engineer specialized in Flux"
  • User Prompt: "digital art of a man walking down a staircase into hell."
  • Assistant Response:
    "A digital artwork depicting a solitary figure descending a grand, marble staircase, each step leading deeper into an abyssal void. The man's silhouette is crisp and detailed, his face a mix of determination and fear, as he steps forward with each frame. The staircase twists and turns, its steps narrowing as it descends, leading to a bottomless pit. The background is a deep, inky black, punctuated by wisps of smoke and flickers of red light, suggesting flames or infernal fires. The man's shadow stretches out behind him, elongated and distorted, adding a sense of depth and the surreal nature of his journey. The overall effect is both hauntingly beautiful and deeply unsettling, capturing the essence of a descent into darkness."

Usage with ComfyUI

You can integrate this model into ComfyUI by using the custom nodes available in the repository: ComfyUI-Qwen

README history 5 versions

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

  1. 2025-02-16Update README.md732f1f32.7 KB
    Loading...
  2. 2025-02-16Update README.mdc1aeca22.7 KB
    Loading...
  3. 2025-02-16Update README.md36a05c42.7 KB
    Loading...
  4. 2025-02-13Update README.md820cf67341 B
    Loading...
  5. 2025-02-13Trained with Unslothec710b35.1 KB
    Loading...

Discussions 1 thread

  1. 2025-03-05Using this model as an uncensored image to prompt?closed2 💬#1
    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