Solving Math Word Problems via Cooperative Reasoning induced Language Models (ACL 2023)

Visualization

MCTS Log

Data preparation

put the dataset under data/

Fine tune generator

Set the hyperparameters in train.slurm and execute bash train.slurm

Fine tune verifiers

Set the hyperparameters in train_verifier.slurm and execute bash train_verifier.slurm

MCTS

After fine-tuning, specify the model path in mcts.slurm, execute bash mcts.slurm. Note that the provided script will not produce reasonable outputs unless the generator and verifiers are properly fine-tuned.

Requirements

pytorch-lightning==1.6.4
torch==1.10.0
python==3.8
cuda==11.1

Citation

Please consider citing our paper and starring this repo if you find them helpful. Thank you!

@article{zhu2022core,
         title={Solving Math Word Problem via Cooperative Reasoning induced Language Models},
         author={Zhu, Xinyu and Wang, Junjie and Zhang, Lin and Zhang, Yuxiang and Gan, Ruyi and Zhang, Jiaxing and Yang, Yujiu},
         journal={arXiv preprint arXiv:2210.16257},
         year={2022}
}

Feel free to open an issue if you have any questions.