Flux Icon Kit Lo R A

Introduction

The Flux-Icon-Kit-LoRA is a text-to-image model configuration using LoRA (Low-Rank Adaptation) technique. It leverages the concept of diffusers for generating creative and unique icon imagery.

Architecture

The model is built upon the "black-forest-labs/FLUX.1-dev" base model. It utilizes a LoRA configuration to fine-tune and adapt the model efficiently for specific image generation tasks, specifically designed for generating icon-like images with varying themes and styles.

Training

The model was trained using the following parameters:

• Learning Rate Scheduler: Constant
• Optimizer: AdamW
• Noise Offset: 0.03
• Multires Noise Discount: 0.1
• Network Dimensions: 64
• Network Alpha: 32
• Repeat & Steps: 25 & 3100
• Epochs: 20
• Total Images Used: 40 (Raw 8-bit)
• Recommended Inference Steps: 30–35

Training involved using a dataset of 40 images with specific image processing parameters to fine-tune the model for optimal performance in generating icon-like images.

Guide: Running Locally

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

   import torch
   from pipelines import DiffusionPipeline
   

3. Load Base 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-Icon-Kit-LoRA"
   trigger_word = "Icon Kit"
   pipe.load_lora_weights(lora_repo)
   

5. Select Device:

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

6. Trigger Image Generation: Use the trigger word "Icon Kit" to generate images.

Cloud GPUs: Consider using cloud GPU services such as AWS EC2, Google Cloud, or Azure for better performance and faster processing.

License

The model is released under the CreativeML Open RAIL-M license, which provides guidelines for responsible AI usage and mandates adherence to ethical AI practices.