/lora_cammi

Primary LanguagePython

Stable Diffusion LoRA Training

Objective

Train a Stable Diffusion LoRA model on a provided set of images and evaluate its performance.

Table of Contents

  1. Environment Setup
  2. Training Steps
  3. Evaluate the Model
  4. Challenges

Environment Setup

  1. Clone the Repository git clone https://github.com/yourusername/lora_cammi.git cd lora_cammi

  2. Install Dependencies Ensure you have Python 3.x installed. Install the required Python packages using pip: torch>=1.9.0 torchvision>=0.10.0 transformers>=4.11.3 diffusion>=0.4.0 numpy>=1.21.2 matplotlib>=3.4.3 Pillow>=8.3.2

  3. Run the main script

    python main.py

Prepare Data: Place your low-resolution images in a directory and specify the input directory path in the code.

Training Steps

Initialize Model: The LoRA model is initialized with specified parameters such as input channels, hidden channels, number of layers, and kernel size.

Prepare Dataset: The dataset is prepared using the CustomImageDataset class, which loads the images and applies transformations.

Train the Model: The model is trained using Mean Squared Error (MSE) loss and Adam optimizer. Training progresses over multiple epochs.

Save the Model: After training, the model state is saved to a file (lora_model.pth) for future use.

Evaluate the Model

Evaluation Metrics=>

The model's performance is evaluated using two metrics:

Peak Signal-to-Noise Ratio (PSNR)
Structural Similarity Index (SSIM)

Evaluation Process:

PSNR and SSIM are calculated between the generated images and the original high-resolution images.

Challenges