wombyz / mbagpt Goto Github PK

Chatbot with dynamic access to Hormozi, Buffett and Branson knowledge bases + ChatGPT API fallback for unrelated Qs

Python 100.00%

mbagpt's Introduction

MBAGPT: Chatting with Multiple Data Sources

MBAGPT is a chat application that leverages the power of GPT-3.5-turbo to provide conversational responses with access to multiple data sources. It allows users to ask questions and receive answers from different knowledge bases (vectorDBs), such as Alex Hormozi, Warren Buffett and Richard Branson as well as answer general queries with the ChatGPT API as a fallback.

Features

Chat interface for interacting with the chatbot powered by GPT-3.5-turbo.
Integration with Hormozi, Buffett and Branson databases for retrieving relevant documents.
Semantic search functionality to provide informative snippets from the databases.
Intent classification to route user queries to the appropriate database.
HTML templates for displaying chat history and messages.
Persistence of embeddings using the Chroma vector store.
OpenAI API key integration for authentication.

Installation

Clone the repository:

git clone https://github.com/wombyz/MBAGPT.git

Install the required dependencies:

pip install -r requirements.txt

Set up your credentials:

Sign up on the OpenAI website and obtain an API key.
Create a new file called "secrets.toml" in the .streamlit folder.
Set your OpenAI API key (required) and pinecone creds (optional) in the secrets.toml file or as an environment variable.
Update the code in the app to use the correct method for accessing the API key.

Run the indexing script to create the vector databases:

python indexing.py

This script will create the Buffet and Branson vector databases by indexing the documents. Make sure to have the necessary PDF documents in the appropriate directories (./docs/buffett/ and ./docs/branson/) before running the script.

Run the application:

streamlit run app.py

Usage

Access the application by navigating to http://localhost:8501 in your web browser.
Enter your prompt in the input box and press Enter.
The chatbot will process your prompt and provide a response based on the available data sources.
The chat history will be displayed on the screen, showing both user and assistant messages.

Contributing

Contributions are welcome! If you would like to contribute to this project, please follow these steps:

Fork the repository.
Create a new branch for your feature or bug fix.
Implement your changes and ensure that the code passes all tests.
Submit a pull request with a detailed description of your changes.

License

This project is licensed under the MIT License.

mbagpt's People

Stargazers

Watchers

mbagpt's Issues

NameError: name 'buffet_handler' is not defined

Whenever I use something related to buffet I get error, the others work, i clone the repo and followed the instructions, what's wrong?

Failed to build tiktoken ERROR: Could not build wheels ...

Log :

Building wheels for collected packages: tiktoken
  Building wheel for tiktoken (pyproject.toml) ... error
  error: subprocess-exited-with-error

  × Building wheel for tiktoken (pyproject.toml) did not run successfully.
  │ exit code: 1
  ╰─> [38 lines of output]
      running bdist_wheel
      running build
      running build_py
      creating build
      creating build\lib.win-amd64-cpython-312
      creating build\lib.win-amd64-cpython-312\tiktoken
      copying tiktoken\core.py -> build\lib.win-amd64-cpython-312\tiktoken
      copying tiktoken\load.py -> build\lib.win-amd64-cpython-312\tiktoken
      copying tiktoken\model.py -> build\lib.win-amd64-cpython-312\tiktoken
      copying tiktoken\registry.py -> build\lib.win-amd64-cpython-312\tiktoken
      copying tiktoken\_educational.py -> build\lib.win-amd64-cpython-312\tiktoken
      copying tiktoken\__init__.py -> build\lib.win-amd64-cpython-312\tiktoken
      creating build\lib.win-amd64-cpython-312\tiktoken_ext
      copying tiktoken_ext\openai_public.py -> build\lib.win-amd64-cpython-312\tiktoken_ext
      running egg_info
      writing tiktoken.egg-info\PKG-INFO
chain.
      [end of output]

  note: This error originates from a subprocess, and is likely not a problem with pip.
  ERROR: Failed building wheel for tiktoken
Failed to build tiktoken
ERROR: Could not build wheels for tiktoken, which is required to install pyproject.toml-based projects

Solution :

Install a Rust Compiler: You can install Rust on your system. The recommended way to do this is through rustup, which is a toolchain installer for Rust. Once Rust is installed, you should be able to build the tiktoken package. Make sure the Rust compiler (rustc) is in your system's PATH.

Problems with Python 3.11 it didn't support Chromadb, so this one requires numpy <1.24 and this is no compatible with Python 3.11

Problems with Python 3.11 it didn't support Chromadb, so this one requires numpy <1.24 and this is no compatible with Python 3.11

Assuming I have no Python installed and, of course, any library. Is there any way to recreate a clean install of the repository and all the required libraries needed

When I try to run python indexing.py, get a lot of errors, like:

Need to install (to chromadb/hnswlib, see below): Microsoft C++ Build Tools, see how in https://github.com/bycloudai/InstallVSBuildToolsWindows
Need, also:

tabulate
unstructured
pdf2image
hnswlib
chromadb

The last 2 ones cannot install on my computer due to the commented incompatibility, so not sure if I need any another lib.

So finally I cannot test this great app :-(

Thanks in advance

Julio Falcón

Error with setting up secrets.toml file

Hi, I am having trouble when running the following instructions under step 3

Create a new file called "secrets.toml" in the .streamlit folder.
Set your OpenAI API key (required) and pinecone creds (optional) in the secrets.toml file or as an environment variable.
Update the code in the app to use the correct method for accessing the API key.

Could you please help with

Sharing an example of how the "secrets.toml" file should be set up or how it should look like?
Sharing an example of how does the modified code looks (to use the correct method for accessing the API key) for both (a) index.py and (b) app.py?

Thank you very much!