OrcaText

This repository provide an implementation of Orca, which is described in this paper:

Orca: Enhancing Role-Playing Abilities of Large Language Models by Integrating Personality Traits

Yuxuan Huang

Motivating Example:

Orca Architecture:

Usage

1. Clone this project, install dependency packages.

git clone https://github.com/Airoura/OrcaText.git
cd OrcaText
pip3 install -r requirements.txt

2. Pipeline:

bash scripts/pipeline.sh

3. Train

   The processed dataset is in Alpaca format, and you can train LLaMA using LLaMA-Factory.

4. Test

   We use OrcaBench to comprehensively evaluate the role-playing ability of the model, which is also our job.

Datasets and Checkpoints

data

• OneDrive
• BaiduNetdisk

We are preparing to release all checkpoints.

Reference

To cite this work please use:

@misc{huang2024orcaenhancingroleplayingabilities,
      title={Orca: Enhancing Role-Playing Abilities of Large Language Models by Integrating Personality Traits}, 
      author={Yuxuan Huang},
      year={2024},
      eprint={2411.10006},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2411.10006}, 
}

License

Orca is released under Apache-2.0 license, see LICENSE for details.