Text Classification on Civil Comments

Metric: AUROC (higher is better)

LeaderboardDataset

Loading chart...

Results

Submit a result

#	Model↕	AUROC▼	Extra Data	Paper	Date↕	Code
1	RoBERTa Focal Loss	0.9818	No	A benchmark for toxic comment classification on ...	2023-01-26	Code
2	RoBERTa BCE	0.9813	No	A benchmark for toxic comment classification on ...	2023-01-26	Code
3	DistilBERT	0.9804	No	A benchmark for toxic comment classification on ...	2023-01-26	Code
4	HateBERT	0.9791	No	A benchmark for toxic comment classification on ...	2023-01-26	Code
5	AlBERT	0.979	No	A benchmark for toxic comment classification on ...	2023-01-26	Code
6	BERTweet	0.979	No	A benchmark for toxic comment classification on ...	2023-01-26	Code
7	ResNet + RoBERTa finetune	0.97	No	PyTorch Frame: A Modular Framework for Multi-Mod...	2024-03-31	Code
8	Unfreeze Glove ResNet 44	0.966	No	A benchmark for toxic comment classification on ...	2023-01-26	Code
9	Unfreeze Glove ResNet 56	0.9639	No	A benchmark for toxic comment classification on ...	2023-01-26	Code
10	Compact Convolutional Transformer (CCT)	0.9526	No	A benchmark for toxic comment classification on ...	2023-01-26	Code
11	Trompt + OpenAI embedding	0.947	No	PyTorch Frame: A Modular Framework for Multi-Mod...	2024-03-31	Code
12	ResNet + OpenAI embedding	0.945	No	PyTorch Frame: A Modular Framework for Multi-Mod...	2024-03-31	Code
13	Trompt + RoBERTa embedding	0.885	No	PyTorch Frame: A Modular Framework for Multi-Mod...	2024-03-31	Code
14	ResNet + RoBERTa embedding	0.882	No	PyTorch Frame: A Modular Framework for Multi-Mod...	2024-03-31	Code
15	LightGBM + RoBERTa embedding	0.865	No	PyTorch Frame: A Modular Framework for Multi-Mod...	2024-03-31	Code
16	PaLM 2 (few-shot, k=10)	0.8535	No	PaLM 2 Technical Report	2023-05-17	Code
17	PaLM 2 (zero-shot)	0.7596	No	PaLM 2 Technical Report	2023-05-17	Code