This is the offcial repo for the ACL-2022 paper "Learning to Reason Deductively: Math Word Problem Solving as Complex Relation Extraction".
- transformers
pip3 install transformers
- Pytorch > 1.7.1
- accelerate package
pip3 install accelerate
(for distributed training)
Reproduce the results, simply run our scripts under scripts
folder.
For example, reproduce the results for Math23k
dataset,
bash scripts/run_math23k.sh
We also provide the Roberta-base-DeductiveReasoner checkpoints that we have trained on the Math23k, MathQA and SVAMP datasets. We do not provide the 5-fold model checkpoints due to space limitation.
Dataset | Link |
---|---|
Math23k (train/dev/test setting) | Link |
MathQA | Link |
SVAMP | Link |
- We plan to have a easier or more user-friendly way to load our pretrained model, by using the
from_pretrained
function from Hugginface.
If you find this work useful, please cite our paper:
@InProceedings{jie2022math,
author = "Jie, Zhanming and Li, Jierui and Lu, Wei",
title = "Learning to Reason Deductively: Math Word Problem Solving as Complex Relation Extraction",
booktitle = "Proceedings of ACL",
year = "2022"
}