zui-c / rlfn Goto Github PK

Check the config file config.yaml

python train.py

python inference.py --dataset 2014

Not save MathBERTa's params

GPT-4V test on CROHME 2014 datasets done by JiaQi Han

If you aim to use this model for real-world applications, it's recommended to train on larger datasets. Finding data that matches your specific application scenario can greatly enhance the model's performance. Also, make sure to update vocabulary-related files, including 'words_dict.txt' and 'token.json'.

We would like to acknowledge the work done on SAN and its modified version CAN. Additionally, we have utilized MathBERTa in our work. These work have been valuable references for our project.