ResiDual Transformer

The code implementation of ResiDual: Transformer with Dual Residual Connections.

Enviroment setup

Please use the anaconda to setup the environment by conda env create -f environment.yaml. Then you can activate the environment by conda activate resi_dual.

Model training

For model training, please follow the guide to process data, and use the command below

LAYER=6 # or 12
DATA_PATH=THE_PATH_TO_YOUR_DATA_BIN # the data should be fairseq data-bin format
CONFIG=iwslt.yaml # or other config under hydra_config folder

fairseq-hydra-train \
common.user_dir=$(pwd)/resi_dual \
task.data=${DATA_PATH} \
model.encoder.layers=${LAYER} \
model.decoder.layers=${LAYER} \
--config-dir hydra_config \
--config-name ${CONFIG}

Contributing

This project welcomes contributions and suggestions. Most contributions require you to agree to a Contributor License Agreement (CLA) declaring that you have the right to, and actually do, grant us the rights to use your contribution. For details, visit https://cla.opensource.microsoft.com.

When you submit a pull request, a CLA bot will automatically determine whether you need to provide a CLA and decorate the PR appropriately (e.g., status check, comment). Simply follow the instructions provided by the bot. You will only need to do this once across all repos using our CLA.

This project has adopted the Microsoft Open Source Code of Conduct. For more information see the Code of Conduct FAQ or contact opencode@microsoft.com with any additional questions or comments.

Trademarks

This project may contain trademarks or logos for projects, products, or services. Authorized use of Microsoft trademarks or logos is subject to and must follow Microsoft's Trademark & Brand Guidelines. Use of Microsoft trademarks or logos in modified versions of this project must not cause confusion or imply Microsoft sponsorship. Any use of third-party trademarks or logos are subject to those third-party's policies.

Cite

@article{xie2023residual,
  title={ResiDual: Transformer with Dual Residual Connections},
  author={Xie, Shufang and Zhang, Huishuai and Guo, Junliang and Tan, Xu and Bian, Jiang and Awadalla, Hany Hassan and Menezes, Arul and Qin, Tao and Yan, Rui},
  journal={arXiv preprint arXiv:2304.14802},
  year={2023}
}