Cj1m / MemorAI

Experimenting with giving LLMs memory

MemorAI

MemorAI is a conversational AI chatbot that uses its memory to provide more context-aware and personalized responses. By leveraging OpenAI's GPT-4, MemorAI can store and recall information from previous interactions to enhance the user experience.

Features

• Context-aware conversation using stored memory
• Utilizes GPT-4 for natural language understanding and generation
• Extensible memory storage system with support for Pinecone and Weaviate databases
• Easy to use and interact with through a simple command-line interface

Getting Started

Prerequisites

• Python 3.8 or higher
• An API key for OpenAI's GPT-4
• Either a local Weaviate DB or a Pinecone API key

Installation

1. Clone the repository:

git clone https://github.com/your-username/MemorAI.git

2. Change to the MemorAI directory:

cd MemorAI

3. Install the required dependencies:

pip install -r requirements.txt

4. Rename Config.example.py to Config.py and configure API keys:

OPENAI_API_KEY = "your_openai_api_key"
PINECONE_API_KEY = "your_pinecone_api_key" # Only required for BRAIN_TYPE="Pinecone"

# Choose between 'Pinecone' and 'Weaviate'
BRAIN_TYPE = "Pinecone"

# Pinecone settings
PINECONE_ENVIRONMENT = "your_pinecone_environment" # Only required for BRAIN_TYPE="Pinecone"

# Weaviate settings
WEAVIATE_URL = "your_weaviate_url"

Usage

1. Run the main script:

python Main.py

2. Interact with MemorAI using the command-line interface:

User: Hi, my name is Christopher. Nice to meet you!
MemorAI: Hello Christopher, it's nice to meet you too! How may I assist you today?

3. MemorAI will learn from conversation and will remember details between sessions.

License

This project is licensed under the MIT License. See the LICENSE file for details.

Acknowledgments

OpenAI for providing the GPT API Weaviate and Pinecone for their vector database software and services.