Instruction-Aware Contextual Compressor

Welcome to the Instruction-Aware Contextual Compressor repository! This project is designed to compress context for RAG in LLM.

Key Features

• Contextual Compression: Intelligently compresses data based on the context it is used in.
• Instruction-Aware: Adapts compression techniques based on the instructions provided.

Getting Started

To get started with the Instruction-Aware Contextual Compressor, follow these simple steps:

1. Clone the Repository

   git clone https://github.com/howard-hou/instruction-aware-contextual-compressor.git

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

   pip install -r requirement.txt

Reproducing Experimental Results

• 1. Ensure to download the proper checkpoint: howard-hou/IACC-compressor-small
• 2. Acquire the appropriate dataset via the provided link
• 3 Acquire the appropriate document collection via the provided line
• 4. Run the following command to reproduce the results: bash run_compress_dataset.sh

Checkpoints

weights and checkpoints are stored in the huggingface model hub.

• ranker
• compressor

Citation

If you find this project useful, please consider citing it:

@article{howard-hou2022instruction-aware-compressor,
  title={Instruction-Aware Contextual Compressor},
  author={Howard Hou},
  journal={GitHub},
  volume={1},
  number={1},
  pages={1-10},
  year={2022},
  publisher={GitHub}
}

License

This project is licensed under the Apache-2.0 License.