In Context Lo R A

Introduction

In-Context LoRA fine-tunes text-to-image models to generate image sets with customizable intrinsic relationships. It can be adapted for various tasks and optionally conditioned on another set using SDEdit. This repository includes multiple models across diverse tasks, detailed in the MODEL ZOO section.

Architecture

The In-Context LoRA models are built upon the FLUX.1-dev base model by black-forest-labs. They utilize a combination of techniques to enhance text-to-image generation by fine-tuning intrinsic relationships within image sets.

Training

The models are extensively documented in the accompanying paper, which details their training methodologies. The paper can be accessed on arXiv.

Guide: Running Locally

1. Clone the Repository:

   git clone https://github.com/ali-vilab/In-Context-LoRA.git
   cd In-Context-LoRA
   

2. Install Dependencies:
   Ensure you have Python and pip installed, then run:

   pip install -r requirements.txt
   

3. Download Model Weights:
   Access the Files & versions tab to download the desired model weights in Safetensors format.

4. Run the Model:
   Load the model in your desired framework, ensuring compatibility with your local environment.

5. Suggested Environments:
   For optimal performance, especially for large-scale tasks, consider using cloud GPUs via platforms like AWS, Google Cloud, or Azure.

License

The In-Context LoRA models use the FLUX base model, and users must comply with FLUX's license. For more details, see the FLUX License.