This is a baseline model for our [calculator-assisted models](https://huggingface.co/models?search=emnlp2023) 
trained on a mixture of all our [Calc-X datasets](https://huggingface.co/datasets?search=emnlp2023). 
See the corresponding paper for details. 

The reported results can be reproduced by using [evaluation script](https://github.com/emnlp2023sub/gadgets/blob/65e24e810cf5ea20aceb8a3c8ddbc19f035ab694/examples/test_calc.py)
from the project repository.