I want to create a chatbot that do not require training on nlp and as per my research dialog gpt is the one that could be useful. I tried it but didn’t get much answers. It is not working very good. Anybody facing same issue or am I doing something wrong. Please guide.
I’m having the same issue. After configuring DialogGPT, I see a warning in the chat with this message:
It looks like this is a recently introduced feature, and it may not be fully operational yet.
Hello Chehak,
Welcome to Kore.ai Community, we are glad to have your here.
We have analysed your concern and below are the suggestions that you need to check:
-
Dialog gpt works only with GPT-4.0 models for conversation management as of now, can you check if you are using GPT-4.0.
-
Enable the conversation type you would want to work (dialogs, faqs, intents, knowledge from Search AI)
-
Make sure the changes you’re doing are synchronized; it is just like how we train utterances while using ML or KG, unless you sync it, Dialog GPT might not work.
-
Finally, fine tune the name of the task or dialog by adding a brief explanation in description.
I am still not getting required answers from my bot, it’s like it is not using LLM and rag to its full capacity. I have few areas of doubt:
- Is there anything we can use to get results completely RAG based without creating any dialogs and all?
- Do we have to create a complete dialog flow in dialog gpt too for it to work like we do in Automation AI.
PS: Our use case is to make a chat bot just like a RAG based agent that answers user’s queries using context from uploaded documents and files without any training.
@chehak it should work for your use case you have described. If you uploaded docs to SearchAI did you initiate sync or initial training?
It is just a button click once you have done that you should get the answers from docs using RAG etc