Code Monkey home page Code Monkey logo

gpt-oracle-trainer's Introduction

gpt-oracle-trainer

Twitter Follow Open Main Version In Colab

Overview

Creating a chatbot that can accurately answer questions about a product or service's documentation is a complex task. You can fine-tune on the documentation itself (but results won't be conversational), use embeddings (risks losing relevant context), etc. gpt-oracle-trainer is an experimental tool that aims to simplify this process and potentially produce better results.

Simply provide your product or service's documentation, describe your product or service, select a temperature for data generation, and choose the number of training examples to generate per document. The system will then generate the dataset in the correct format, train the model, and allow you to test it.

Features

  • Data Generation: The system generates a question-and-answer dataset based on your documentation, service description, temperature, and number of examples. The data is formatted correctly for model training.

  • Model Training: The system trains the model using the generated dataset.

  • Model Testing: Test the trained model with a custom prompt.

How to Use

  1. Open the notebook in Google Colab or in a local Jupyter notebook.

  2. Add your OpenAI API key to the line openai.api_key = "OPENAI API KEY HERE".

  3. Paste your documentation into the docs list, each document as a separate string.

  4. Define your service_name_and_description, temperature, and number_of_examples_per_doc. For example:

service_name_and_description = "MosaicML, a platform that makes it easier to train and fine-tune large AI models"
temperature = .7
number_of_examples_per_doc = 10
  1. Run the cells to generate the dataset, train the model, and test it.

  2. The final trained model will be saved under the name you specify in new_model.

Contributions are welcome! Some ideas:

  • improve the data generation process
  • add more customization options for the model training

License

This project is MIT licensed.

Contact

Matt Shumer - @mattshumer_

Project Link: https://github.com/mshumer/gpt-oracle-trainer

Lastly, if you want to try something even cooler than this, sign up for Personal Assistant (most of my time is spent on this). It's basically an AI that can operate your web browser to complete tasks for you.

gpt-oracle-trainer's People

Contributors

mshumer avatar

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    ๐Ÿ–– Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. ๐Ÿ“Š๐Ÿ“ˆ๐ŸŽ‰

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google โค๏ธ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.