Official source code repository for the ACL 2023 Findings paper: "PIP: Parse-Instructed Prefix for Syntactically Controlled Paraphrase Generation" by Yixin Wan and Kuan-Hao Huang and Kai-Wei Chang.
Link to full paper: https://arxiv.org/abs/2305.16701.
- To build a conda environment for running experiments, cd into the current repository and run the command:
conda create --name <env> --file requirements.txt
- To train the baseline model with seq2seq training, run:
sh ./scripts/train_seq2seq.sh
- To train the baseline model with prefix tuning, run:
sh ./scripts/train_prefix_tuning.sh
- To train the PIP-direct model, run:
sh ./scripts/train_pip_direct.sh
- To train the PIP-indirect model, run:
sh ./scripts/train_pip_indirect.sh