Berenices Alpha D W Lo R A

Introduction

The Berenices-Alpha-DW-LoRA model, developed by prithivMLmods, is a text-to-image generator leveraging the StableDiffusionXLPipeline. It excels in creating detailed and dynamic images based on specific prompts, using techniques like LoRA (Low-Rank Adaptation) to refine outputs.

Architecture

• Base Model: stabilityai/stable-diffusion-xl-base-1.0
• Scheduler: EulerAncestralDiscreteScheduler
• Optimizer: AdamW
• Network Dim: 64
• Network Alpha: 32

The model architecture includes parameters for noise offset, multiresolution noise discount, and iterations to fine-tune image outputs.

Training

The model utilizes a constant learning rate scheduler, with training parameters set to optimize image generation quality, including:

• Epochs: 15
• Repeat Factor: 20
• Noise Offset: 0.03
• Multires Noise Discount: 0.1

Guide: Running Locally

To run the Berenices-Alpha-DW-LoRA model locally, follow these basic steps:

1. Install Dependencies:

   pip install torch diffusers
   

2. Set Up Pipeline:

   from diffusers import StableDiffusionXLPipeline
   pipe = StableDiffusionXLPipeline.from_pretrained(
       "stabilityai/stable-diffusion-xl-base-1.0",
       torch_dtype=torch.float16,
       use_safetensors=True,
   )
   

3. Load Model Weights:

   pipe.load_lora_weights("prithivMLmods/Berenices-Alpha-DW-LoRA", weight_name="Berenices-Alpha-DW-LoRA.safetensors", adapter_name="DW")
   pipe.set_adapters("DW")
   pipe.to("cuda")
   

4. Run Inference: Use trigger prompts like "Deadpool" or "Wolverine" to generate images.

For optimal performance, consider using cloud GPUs from providers like AWS, GCP, or Azure.

License

The model is licensed under the CreativeML OpenRAIL-M license, allowing for creative and research purposes with proper attribution.