Category: Chat

Hey, I am currently faced with the challenge of evaluating/deciding which information to fine tune into my model (used to predict answers for a chatbot) vs. which information to provide the model via prompt. I am well aware of all the capacity constraints and prerequisites but would like to hear from you guys whether you have any experience witht the raw performance/quality of answers when comparing the two approaches. Thanks! 🙂

I am working on a project where I have to build a chatbot for a company. I have already made the RASA chatbot and in the actions.py file I am using Langchain and GPT to query the information from the data I have fed to my Agent.

Now thw task is to integrate this Chatbot with the website. I need to make a proper frontend and connect with my backend(already made).

What tools I can use, how will I do, workflow.

I need someone to explain me all this.

I have built an AI Chatbot for my client on Voiceflow platform.
now, I ran into a problem with my AI Chatbot, the problem is Specifically with Voiceflow Knowledge base response.
Here is a description of my problem:
“So , I actually integrated the chatbot to my client’s website And its working okay But its working different on a voiceflow prototype and on the clients website For example If i ask a question “what color is the sky” with the “?” At the end of the sentence he will respond right. But if i ask him that same question without the “?” At the end he wont respond and he will just glitch out. But its only glitching out on clients website But on the voiceflow prototype its working perfectly fine, without me having to put “?” At the end of the sentence.”

How can I solve this issue!?

What are SOTA Open Domain QA models at the moment? I’ve been doing research on the field and am seeing so many cool approaches, since they’re so many aspects of QA that need to be worked on, but I have no idea what’s SOTA at the moment. My professor told me to look into RAG, and I am, but I feel like he might not be as up to date in this area, so I’m not sure where it stands? Also where do you guys look to find out what’s the most popular models at the moment? I’ve been reading papers in popular journals and trying to just search stuff up ig but it’s not really telling me what I want to know idk

I am really now into nsfw roleplays, just daily talks don’t turn me on. What’s your guy’s favorite? list it up below and I will try them all, archive them LOLZ. still finding one!

so far I tried :

1. botify.ai : can do it on phone. they do hardcore, various contents very qualified sexting terms. But, you can’t do several roleplays with one person and you have to do it with all different characters
2. altcat.ai : This is dope. You can do various role play with one chatbot, just the chat room? differs. What turns me on is that the origin are real influencers ( I saw one from OF lolz ) but, one time sad about it is that it only has 2~3 roleplays to try on
3. crushchat.ai: quite various and well qualified. but the thing is that I don’t like anime that much, if I liked it will use it often.
4. nomi.ai : great chatting qualities, was one of my favs. but I want some more concpetual roleplays and pictures with this.

I’m looking for a chatbot service like chatbase so I can embed a customer service rep on my website. I trained a GPT with a bunch of data and it works 85 to 95% of the time. When I use the same data on Chatbase (and other websites like it) I get horrible results. Does anyone have any advice for me?