finetuned the bert model again to classify things right

new data new training

added finetuned categorical classification model + more evaluated dataset modifications

created the dataset for categorical classification

removed commas

added dataset for content moderation