CQCL/lambeq

Feature Request : Dataset visualisation

Closed this issue · 1 comments

Given the dynamic nature of NLP data points, it would be really useful if we could visualize a dataset based on how noisy and/or rugged the dataset is to provide a manner of countering caveats. Another approach would be to enable a regularization term to allow for a better fit to the dataset.

I am not sure what you mean by noisy or rugged datasets. Also sounds like there are two separate ideas going on here.