Dark Thing Flux Lo R A

Introduction

Dark-Thing-Flux-LoRA is a text-to-image model under development by prithivMLmods. It uses LoRA (Low-Rank Adaptation) for generating images with specified prompts, focusing on themes like "Dark Creature."

Architecture

The model operates as an extension of the black-forest-labs/FLUX.1-dev base model, integrating LoRA technology to enhance image generation. It employs a diffusion pipeline with parameters such as AdamW optimizer and constant LR scheduler to refine image outputs.

Training

• Parameters:

  • LR Scheduler: Constant
  • Optimizer: AdamW
  • Network Dimensions: 64
  • Network Alpha: 32
  • Epochs: 10
  • Image Dimensions: 1024 x 1024 (default)

• Noise Parameters:

  • Noise Offset: 0.03
  • Multires Noise Discount: 0.1
  • Multires Noise Iterations: 10

• Training Data:

  • Total Images: 15

The model remains in training and not finalized, potentially containing artifacts.

Guide: Running Locally

1. Set Up Environment:

   • Install necessary libraries, including torch and pipelines.

2. Load the Model:

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

3. Load LoRA Weights:

   lora_repo = "prithivMLmods/Dark-Thing-Flux-LoRA"
   pipe.load_lora_weights(lora_repo)
   

4. Set Device:

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

5. Use Trigger Words:

   • Trigger image generation using the term "Dark Creature."

Recommendation: Utilize cloud GPUs for optimal performance, such as AWS EC2 or Google Cloud's AI Platform.

License

The model is licensed under the CreativeML OpenRAIL-M license.