[andrewgazelka]

My understanding was training on ChatGPT output was against OpenAI ToS. Is this incorrect for this use case (training BERT)?