NonToxicCivilBert or 'CivilBert' is a further finetuned model of bert-based-uncased using only the non-toxic data of the Jigsaw-unintended-bias dataset to make the predicted tokens less toxic. We are working further to make this model better and less toxic.
You can use it directly using the transformers:
- from transformers import AutoTokenizer, AutoModelForMaskedLM
- tokenizer = AutoTokenizer.from_pretrained("Ashokajou51/NonToxicCivilBert")
- model = AutoModelForMaskedLM.from_pretrained("Ashokajou51/NonToxicCivilBert")
- Downloads last month
This model can be loaded on the Inference API on-demand.