Thank you for sharing the great work. Are there any hyper-parameters that should be changed when using the cased dataset? I only changed the dataset and the pre-trained model, but got even poorer results compared to the uncased model. By the way, I'm wondering why the article part in the cased data is still uncased like the photo below.
Thank you for sharing the great work. Are there any hyper-parameters that should be changed when using the cased dataset? I only changed the dataset and the pre-trained model, but got even poorer results compared to the uncased model. By the way, I'm wondering why the article part in the cased data is still uncased like the photo below.