takarajordan's picture
Update README.md
55e825a verified
metadata
license: apache-2.0
base_model:
  - black-forest-labs/FLUX.1-schnell
pipeline_tag: text-to-image
Takara.ai Logo

From the Frontier Research Team at Takara.ai we present Flux.1 Q_4_k, a quantized GGUF model optimized for stable-diffusion.cpp, enabling efficient image generation on lower-end hardware. This model was used to create the Kurai Toori Dark Streets dataset.


Features

  • Optimized for lower-end hardware through 4-bit quantization
  • High-quality image generation despite compression
  • Efficient performance with minimal quality degradation
  • Wide-ranging capabilities beyond dark street scenes

Usage

  1. Clone and set up stable-diffusion.cpp:
    git clone https://github.com/leejet/stable-diffusion.cpp.git
    cd stable-diffusion.cpp
    # Follow setup instructions in the stable-diffusion.cpp README
    
  2. Download the GGUF model file from this repository.
  3. Run the model using stable-diffusion.cpp, pointing to the downloaded file:
    ./sd -m path/to/flux.1-q_4_k.gguf -p "your prompt here"
    

Performance Benefits

  • Reduced memory usage compared to full-precision models
  • Faster inference times on consumer hardware
  • Runs on less powerful hardware without significant quality loss
  • Ideal for experimentation and rapid prototyping

Technical Details

This model is a 4-bit quantized version of the FLUX.1-schnell base model from Black Forest Labs. The quantization process preserves the creative capabilities of the original model while dramatically reducing its memory footprint and computational requirements.

Example Use Cases

  • Generating urban nightscapes and cityscapes
  • Creating artistic interpretations for creative projects
  • Rapid prototyping of visual concepts
  • Accessible AI image generation on consumer hardware

For research inquiries and press, please reach out to research@takara.ai

人類を変革する