It is a model able to predict toxicity given a history and a response to it. It is created for dialog agents. To use it correctly please use the following schematics: [HST]Hi,how are you?`[END]I am doing fine[ANS] I hope you die. Token [HST] initiates the history of the conversation and every pair turn is separeted by [END]. Token [ANS] indicates start of the response to the last utterance. I will update this card, but right now I am developing a bigger proyect with these,so i dont have the time to indicate all the results.