jigsaw_unintended_bias100K

Predict whether online social media comments are toxic based on their text and additional tabular features providing information about the post (e.g. likes, rating, date created, etc.). This dataset originates from a 2019 Kaggle competition (https://www.kaggle.com/c/jigsaw-unintended-bias-in-toxicity-classification) in which the 1st place solution utilized dataset-specific tricks such as a Bucket Sequencing Collator, auxiliary domain-specific prediction tasks for models, and a custom mimic loss function for training. Dataset found from the paper: Benchmarking multimodal automl for tabular data with text fields. arXiv preprint arXiv:2111.02705.