中文Sign up

CCSSNE/DreamFast-qwen3-4b-heretic

🤗 Hugging Face sourcetext-generationapache-2.04B params8.0 GBGGUF✓ 22 checksumsupdated today
Submit in one command

Run it next to your model folder. It makes the torrent, checks your files against Hugging Face, and submits it. You just start seeding and paste your key from your account. It only reads your files and never changes them. Read the script first if you like.

curl -fsSL https://pirateface.co/package.sh | bash -s -- --repo CCSSNE/DreamFast-qwen3-4b-heretic ./model-folder
Needs a seeder →

Qwen 3 4B - Heretic (Abliterated)

An abliterated version of Qwen 3 4B created using Heretic v1.2.0. This model has reduced refusals while maintaining model quality, making it suitable as an uncensored text encoder for image generation models like Z-Image and FLUX.2 Klein 4B.

Model Details

  • Base Model: Qwen/Qwen3-4B
  • Abliteration Method: Heretic v1.2.0
  • Trials: 200
  • Trial Selected: Trial 96
  • Refusals: 3/100 (vs 100/100 original)
  • KL Divergence: 0.0000 (zero measurable model damage)

Files

HuggingFace Format (for transformers, llama.cpp conversion)

model-00001-of-00002.safetensors
model-00002-of-00002.safetensors
config.json
tokenizer.json
tokenizer_config.json

ComfyUI Format (for Z-Image / FLUX.2 Klein 4B text encoder)

comfyui/qwen3-4b-heretic.safetensors              # bf16, 7.5GB
comfyui/qwen3-4b-heretic_fp8_e4m3fn.safetensors   # fp8, 4.1GB
comfyui/qwen3-4b-heretic_nvfp4.safetensors        # nvfp4, 2.6GB

GGUF Format (for llama.cpp and ComfyUI-GGUF)

Quant Size Notes
F16 ~7.5GB Lossless reference
Q8_0 ~4GB Excellent quality
Q6_K ~3GB Very good quality
Q5_K_M ~2.7GB Good quality
Q4_K_M ~2.3GB Recommended balance
Q3_K_M ~1.9GB For low VRAM only

NVFP4 Notes

The NVFP4 (4-bit floating point, E2M1) variants use ComfyUI's native quantization format. They are ~3x smaller than bf16 and load natively in ComfyUI without any plugins. Blackwell GPUs (RTX 5090/5080, SM100+) can use native FP4 tensor cores for best performance, but ComfyUI also supports software dequantization on older GPUs (tested working on RTX 4090).

Usage

With ComfyUI (Z-Image / FLUX.2 Klein 4B)

  1. Download a ComfyUI format file:

    • FP8 (recommended): comfyui/qwen3-4b-heretic_fp8_e4m3fn.safetensors (4.1GB)
    • NVFP4 (smallest): comfyui/qwen3-4b-heretic_nvfp4.safetensors (2.6GB)
    • bf16 (full precision): comfyui/qwen3-4b-heretic.safetensors (7.5GB)

  2. Place in ComfyUI/models/text_encoders/

  3. In your Z-Image workflow, use the ClipLoader node and select the heretic file

With Transformers

from transformers import AutoModelForCausalLM, AutoTokenizer
import torch

model = AutoModelForCausalLM.from_pretrained(
    "DreamFast/qwen3-4b-heretic",
    device_map="auto",
    torch_dtype=torch.bfloat16
)
tokenizer = AutoTokenizer.from_pretrained("DreamFast/qwen3-4b-heretic")

prompt = "Describe a dramatic sunset over a cyberpunk city"
inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
outputs = model.generate(**inputs, max_new_tokens=200)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))

With llama.cpp

llama-server -m qwen3-4b-heretic-Q4_K_M.gguf

Abliteration Process

Created using Heretic v1.2.0 with 200 optimization trials:

? Which trial do you want to use?
> [Trial  96] Refusals:  3/100, KL divergence: 0.0000  <-- selected
  [Trial  90] Refusals:  5/100, KL divergence: 0.0000
  [Trial  95] Refusals:  9/100, KL divergence: 0.0000
  [Trial 122] Refusals: 90/100, KL divergence: 0.0000
  ...

Trial 96 was selected for having the fewest refusals (3/100) with zero measurable KL divergence, indicating the abliteration surgically removed the refusal mechanism with no damage to model capabilities.

Limitations

  • This model inherits all limitations of the base Qwen 3 4B model
  • Abliteration reduces but does not completely eliminate refusals (3/100 remain)

License

This model is released under the Apache 2.0 License, following the base Qwen 3 4B model license.

Acknowledgments