[rspen]

Great idea, but the trained model is still a little bit lacking. Many times the choices didn't make much sense coming from the question. That being said this is a super ambitious chatbot and I love seeing this type of thing implemented

[nickwalton00]

Thanks! I just used the vanilla GPT-2 model, but I'm planning on fine tuning it on some fantasy data. If OpenAI releases their large model I would love to try it out and see how much it improves it.