You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I downloaded a dataset and noticed that there are many extra spaces in the labels. For example, it appears as \alpha _ { k } instead of \alpha_{k}. Why is this happening, and what impact could it have on model training?
The text was updated successfully, but these errors were encountered:
The extra spaces in the labels arises from the dataset’s design to maintain consistent formatting.
Specifically, the dataset is constructed following the format and conventions used in the pix2tex and im2latex-100k datasets. The extra spaces are added during the normalization process to standardize the formatting of the formula. This ensures a fairer evaluation across different models by providing a consistent representation of LaTeX expressions, regardless of small formatting differences. While the additional spaces slightly increase the sequence length, they do not have a negative impact on model performance.
I downloaded a dataset and noticed that there are many extra spaces in the labels. For example, it appears as \alpha _ { k } instead of \alpha_{k}. Why is this happening, and what impact could it have on model training?
The text was updated successfully, but these errors were encountered: