search

unitaryai / detoxify

Trained models & code to predict toxic comments on all 3 Jigsaw Toxic Comment Challenges. Built using ⚡ Pytorch Lightning and 🤗 Transformers. For access to our API, please email us at contact@unitary.ai.
https://www.unitary.ai/
Apache License 2.0
893 stars 115 forks source link

Question regards use case #80

Open TheMrguiller opened 1 year ago

TheMrguiller commented 1 year ago

Hi,

I am new in the development of NLP models and I have in mind using your model as an assistant to fine-tune a chit-chat bot. I have seen in several issues that you don't recommend using it. Is it true? Also what kind of model as experts do you suggest using, even between your current models?

Thank you in advance,

laurahanu commented 1 year ago

You can use it if you want to check the toxicity of your bot replies. Either the unbiased or the multilingual version should work well. However, since it hasn't been trained on this kind of data, it might not work as well as on user comments online.