Skip to content

Latest commit

 

History

History
166 lines (120 loc) · 7.34 KB

README.md

File metadata and controls

166 lines (120 loc) · 7.34 KB

SmartGPT

SmartGPT: Intelligent Conversational Assistant

SmartGPT is an advanced conversational AI application designed for engaging and meaningful interactions between users and a sophisticated chatbot. Built using Streamlit and FastAPI, it offers an intuitive interface that makes communication seamless.

Key features include dynamic tone selection, allowing users to customize the chatbot's responses to suit various contexts, from formal to casual. SmartGPT also enables real-time data retrieval, ensuring users receive accurate information on topics such as flights and weather updates.

A standout feature is the PDF contextual querying, where users can upload documents, allowing the chatbot to answer questions based on their content. This is particularly beneficial in academic and professional settings.

SmartGPT supports session management, enabling users to create and view multiple chat sessions, ensuring continuity in their interactions. With secure user authentication, users can easily register and log in, fostering a safe environment for discussions.

Overall, SmartGPT combines powerful natural language processing with user-centric features, making it an ideal tool for customer support, education, research, and personal assistance.

SmartGPT Frontend

Description

The frontend for SmartGPT is built using Streamlit, providing a simple, interactive user interface to interact with the SmartGPT chatbot. It allows users to upload PDFs for context, select a tone for conversation, and manage sessions. Users can register, log in,craete new sessions and view their chat history.

Features

  • User Authentication: Users can register, log in, and create new sessions.
  • Tone Selection: Users can choose the tone of conversation (e.g., formal, casual).
  • File Upload: Upload PDF files for contextual question-answering.
  • Session Management: Create new sessions, view chat history, and manage ongoing sessions.
  • Chat Interface: Interactive chat with the assistant that adapts based on user preferences.

Technologies Used

  • Streamlit: For the frontend interface.
  • FastAPI: Backend API interaction (not covered in this README).
  • Python: For handling logic and requests to the backend.

Authentication

  • Register: Enter your name, email, and (optional) password to create a new user.
  • Login: Enter your registered name, email, and password to log in.

Session Management

  • Create New Session: Start a new conversation session with the chatbot.
  • View History: View previous chat history (available after logging in).

Tone Selection

  • Choose the tone of the chatbot's response from the dropdown menu (e.g., formal, casual).
  • For real-time data, it is imperative to select the real-time option. This choice ensures that the information provided is up-to-date and accurate. For instance, if one inquires about the available flights from Destination A to Destination B within the last 24 hours, the system will deliver the most current and precise data.

Modify Tone Options

If you wish to add more tones, modify the dropdown options in the Streamlit app's sidebar.

tone_options = ["formal", "casual", "friendly", "sarcastic"]
st.selectbox("Choose Tone", tone_options)

API Integration

Ensure that your backend API (FastAPI) is running and accessible at the defined endpoint.

Screenshots

SmartGPT Frontend

SmartGPT Backend

Description

The SmartGPT backend is a robust and scalable system developed using FastAPI, leveraging modern AI technologies to deliver intelligent and contextually relevant responses through an interactive chatbot interface. The architecture incorporates various powerful libraries, including LangChain for conversational AI, Azure OpenAI for advanced language models,Tavily which is a swearch engine optimized for LLMs and RAG, aimed at efficient, quick, and persistent search results and PostgreSQL for efficient data storage and retrieval.

Features

  • User Management: Users can register, log in, and manage their profiles.
  • Session Management: Create and manage chat sessions, view session history.
  • Customizable Responses: Allows users to select different tones for the chatbot's responses (e.g., formal, casual).
  • Real-Time Data Retrieval: Integrates with external APIs for real-time information.
  • PDF Querying: Supports uploading PDFs to enhance the chatbot's context and knowledge base.

Technologies Used

  • FastAPI: For building the backend APIs.
  • SQLAlchemy: For database interactions with PostgreSQL.
  • PostgreSQL: For storing user data, session details and Chat History
  • PGVector: For similarity search with embeddings.
  • LangChain: For handling language model interactions and retrieval-augmented generation.

Installation

  1. Clone the repository:

    git clone https://github.com/Suwarrna-2001/SmartGPT.git

  2. Install the dependencies:

    poetry install
    • poetry install: This command will install the dependencies specified in your pyproject.toml file. If you haven't locked your dependencies yet, you can do so with poetry lock, but typically, running poetry install will automatically handle that for you.

  3. Navigate to the project directory:

    cd SmartGPT-Backend

  4. Create a virtual environment and activate it:

    poetry shell

  5. Set up the .env file with the following keys:

    AZURE_OPENAI_API_KEY=<Your Azure OpenAI Key>
AZURE_OPENAI_ENDPOINT=<Your Azure OpenAI Endpoint>
TAVILY_API_KEY=<Your Tavily API Key>
CONNECTION_STRING=<Your PostgreSQL Connection String>

  6. Start the FastAPI application(Backend):

    poetry shell
uvicorn app:app --host 127.0.0.1 --port 3003

  7. Run the Streamlit application:

    poetry shell
cd services
streamlit run ui_chat_service.py --server.port 8502

Running the Application

  1. Run the FastAPI application:

    uvicorn app:app --reload

  2. Access the API documentation in your browser at http://localhost:3003/docs.

API Endpoints

User Management

  • POST /register-user: Register a new user with name and email.
  • POST /login: Authenticate a user and create a session.

Session Management

  • POST /session: Create a new chat session.
  • GET /user/{user_id}/sessions: Retrieve all sessions for a specific user.
  • GET /messages/{session_id}: View chat history for a specific session.

Normal Quering:

  • POST /talk-to-bot: API endpoint processes user queries by invoking a conversational chain based on the provided session ID and tone, returning intelligent responses from the chatbot.

PDF Querying

  • POST /up: Upload a PDF for contextual question-answering.
  • POST /query-pdf: Query the uploaded PDF for specific information.

All Sessions:

  • GET /all-sessions: Returns all the sessions
  • GET /all-users-info: Returns all the info of the registered users.

Database Structure

The application uses PostgreSQL with the following tables:

  • User Table: Stores user information.
  • Session Table: Manages user sessions linked to their chat history.

Refer: