Flux Cute 3 D Kawaii Lo R A

Introduction

The Flux-Cute-3D-Kawaii-LoRA model is a text-to-image diffusion model designed to generate 3D kawaii-style images based on text prompts. It utilizes the Diffusers framework and is developed as a LoRA (Low-Rank Adaptation) model.

Architecture

• Base Model: black-forest-labs/FLUX.1-dev
• Training Parameters:
  • LR Scheduler: Constant
  • Optimizer: AdamW
  • Network Dimensions: 64
  • Network Alpha: 32
  • Epochs: 18
• Image Processing:
  • Noise Offset: 0.03
  • Multires Noise: Discount 0.1, Iterations 10
  • Repeat & Steps: 25 & 3270
• Image Dimensions: Best at 768 x 1024, default 1024 x 1024

Training

The model was trained with:

• Total Images: 24
• Labeling: florence2-en (natural language & English)
• The model is still under training and may contain artifacts or perform poorly in some scenarios.

Guide: Running Locally

1. Setup Environment: Ensure you have Python and PyTorch installed.
2. Import Required Libraries:

   import torch
   from pipelines import DiffusionPipeline
   

3. Initialize Model:

   base_model = "black-forest-labs/FLUX.1-dev"
   pipe = DiffusionPipeline.from_pretrained(base_model, torch_dtype=torch.bfloat16)
   

4. Load LoRA Weights:

   lora_repo = "strangerzonehf/Flux-Cute-3D-Kawaii-LoRA"
   pipe.load_lora_weights(lora_repo)
   

5. Setup Device:

   device = torch.device("cuda")
   pipe.to(device)
   

6. Generate Images: Use the trigger word "Cute 3d Kawaii" to generate images.

For optimal performance, running on cloud GPUs such as those provided by AWS, GCP, or Azure is recommended.

License

The model is licensed under the creativeml-openrail-m license, which allows for certain uses, adaptations, and redistributions under specified conditions.