CoMA-7B

CoLLaMA: A Multi-task Instruction Dataset and Large Language Model for Code

<p align="center" width="100%"> <img src="https://i.postimg.cc/J7Ds1tw6/CoLLaMA.jpg" width="40%" height="20%"> </p>

Model details

Trained in June 2023.

CoMA comprises a fine-tuned coding LLM and a multi-task instruction tuning dataset featuring 77K data samples encompassing 8 diverse tasks.

Please refer to the README of the GitHub repository for detailed information.

Training dataset

The model was trained on a 77k rows instruction following dataset, which is released in the GitHub repository.

Citation

<div> <div align="center"> <a target='_blank'>Gang Hu¹</span>&emsp; <a target='_blank'>Xi Wen¹</span>&emsp; <a target='_blank'>Xin Liu¹</a>&emsp; <a href='https://jimin.chancefocus.com/' target='_blank'>Jimin Huang²</a>&emsp; <a target='_blank'>Qianqian Xie*³</a>&emsp; </div> <div> <div align="center"> ¹School of Information Science & Engineering, Yunnan University&emsp; ²ChanceFocus AMC&emsp; ³School of Computer Science, Wuhan University&emsp; </div>

@misc{Hu2023CoLLaMA,
      title={CoLLaMA: A Multi-task Instruction Dataset and Large Language Model for Code}, 
      author={Gang Hu and Xi Wen and Xin Liu and Jimin Huang and Qianqian Xie},
      year={2023},
}