Conversational AI + question answering model

I would like to train a model that does both conversations and questions answering based on articles we feed to it, and ending up with an informed chatbot. Is GPT-j-6B the way to go ?
There may be trade-offs to make in terms of performance, model size, training time, hardware resources. Could you suggest the best open source models that fits my goals? with different sizes and If I can train them with hugging face APIs. Thank you.

1 Like

Question: did you ever get what you were asking for, regarding conversational AI ?
The reason I’m asking is, i-too am looking for a similar model application.
Very interested in any info you feel up for sharing!

I would recommend checking out LangChain: GitHub - hwchase17/langchain: ⚡ Building applications with LLMs through composability ⚡. Allows for building Conversational Agents with Huggingface.

So, I made gpt-j and chatbot separately, I also made question-answer bot separately, but to do both at the same time, you must have a different dataset from other datasets, then with this dataset, the gpt-j model can be fine-tuned, I recently started such a project, I will try it myself.

Hello there!
Just came across your post , as s REFERENCE in one of my emails. from this person:(Profile - conceptofmind - Hugging Face Forums)
Anyway; i really only have 1 question: if you wanted to create a ChatGPT-3 Android app, while USING an android phone, which software solution would you choose?
This is were i find myself at the moment.
Any advice/help you could offer a real Noob, would be GREATLY APPRECIATED :+1:.
Thank you.

You can make an android application using java or kotlin, but if you say it is difficult, you can also use python kivy.But I recommend java because it has more community than python kivy so if there is a place where you can’t do it while creating an application, you can find a quick answer to your question.In addition, I know that there is no chatgpt api, I think you meant gpt-3, but gpt-3 is also paid after a while.