Assistance with AI Model Integration and Customization Settings

  • Resolved scotlandtravel

    (@scotlandtravel)


    1 month ago

    Hi, (In short I’m trying to make it a ai assist bot which tells use prices based on custom pricing structure )

    I hope you’re doing well! I’m currently working on developing an AI-powered chatbot that assists users in calculating taxi fares based on distance and vehicle type using a custom pricing formula. My goal is to fine-tune the bot using OpenAI’s gpt-3.5-turbo model, ensuring that it can dynamically calculate prices based on user input (e.g., pickup and destination locations, distance, vehicle type, etc.).What I’m Trying to Achieve:

    1. Fine-Tuned Pricing Model:
      • The model will calculate taxi prices using a formula:
        Total Price = Initial Fee + (Distance * Price per Mile).
      • I’ve created a training dataset that teaches the model this pricing structure.
    2. Controlled API Usage:
      • I would like the AI bot to only invoke OpenAI (ChatGPT) when absolutely necessary, specifically for price-related queries.
      • For other types of queries, I’d prefer that the bot handle them locally, without connecting to OpenAI, to reduce unnecessary API usage.
    3. Website Crawling:
      • I would like to give the AI bot access to crawl my company’s website so it can extract data (e.g., FAQs, location-specific details, pricing info, etc.) to improve its responses and provide more accurate, up-to-date information to users.

    Questions:

    1. Settings for Controlled API Usage:
      What settings should I enable or customize so that my AI bot only connects to OpenAI for price-related queries? I’d like to avoid unnecessary calls for non-pricing-related queries. Is there a way to implement keyword filtering or intent detection to control when the bot connects to the API?
    2. Enabling Website Crawling for the Bot:
      How can I allow the AI bot to crawl specific sections of my website to retrieve information? What permissions or technical steps are required to set up this capability within the AI engine?

    I appreciate any guidance or specific settings you can recommend to optimize the bot’s performance and achieve these goals.

    Looking forward to your advice!

Viewing 3 replies - 1 through 3 (of 3 total)
  • Thread Starter scotlandtravel

    (@scotlandtravel)

    Note: I have fine-tuned using a dataset on chatgpt 3.5 yet when testing in playground its still giving generalized answers

    Thread Starter scotlandtravel

    (@scotlandtravel)

    Any help on this?

    Plugin Support Val Meow

    (@valwa)

    Hey @scotlandtravel! ??

    I recommend starting with our documentation, which will answer most of your questions. You can learn more here: https://docs.meowapps.com.

    Fine-tuning a model is a specific and complex process. Refer to the “Let It Know About Your Business” section in the documentation to learn more about this. Please note that GPT-3.5 is now deprecated.

    There is no “local chatbot” within AI Engine. The plugins call different services to retrieve the AI model response and display it. If you want a locally running AI model, you’ll need to host it yourself (using Ollama, for example), set up the environment, and use AI Engine filters to dynamically switch between the remote and self-hosted model.

    Regarding API usage, please refer to the documentation here.

    For web search information, please read the documentation here.

Viewing 3 replies - 1 through 3 (of 3 total)
  • You must be logged in to reply to this topic.