streamlit_application

AI Chatbot Documentation

Overview

This project is a Streamlit-based web application that implements a personal gym chatbot. It uses the Ollama API for local AI processing and provides a user-friendly interface for interacting with different language models.

Features

• Dark-themed user interface with custom CSS styling
• Configurable model selection (llama2, mistral, codellama)
• Adjustable temperature setting for AI responses
• Chat history persistence using Streamlit session state
• Real-time chat interface with user and assistant messages
• Clear chat functionality
• Error handling for API requests

Prerequisites

• Python 3.8+
• Streamlit
• Requests library
• Ollama server running locally on port 11434

Installation

1. Install required Python packages:

pip install streamlit requests

2. Ensure Ollama is installed and running locally:

• Follow Ollama's official documentation for installation
• Start the Ollama server: ollama serve

Project Structure

project_directory/
│
├── app.py           # Main application code
├── README.md        # This documentation file

Usage

1. Run the Streamlit app:

streamlit run app.py

2. Access the application through your web browser (typically at http://localhost:8501)

3. Configure settings in the sidebar:

• Select desired model
• Adjust temperature slider
• View application information

4. Interact with the chatbot:

• Enter messages in the text input field
• Click "Send" to get AI responses
• Use "Clear Chat" to reset the conversation

Code Explanation

Imports

• streamlit: For creating the web interface
• requests: For making API calls to Ollama
• json: For handling JSON data

Page Configuration

st.set_page_config(
    page_title="AI Chatbot",
    page_icon="🤖",
    layout="wide"
)

Sets up the Streamlit page with a title, icon, and wide layout.

CSS Styling

Custom CSS is applied using st.markdown with unsafe_allow_html=True to create a dark theme and style various UI components.

Session State

if 'messages' not in st.session_state:
    st.session_state.messages = []

Initializes chat history storage using Streamlit's session state.

Sidebar Configuration

with st.sidebar:
    st.title("⚙️ Configuration")
    model = st.selectbox(...)
    temperature = st.slider(...)

Creates a sidebar for model selection and temperature adjustment.

Main Interface

• Displays the chat history in a container
• Shows user and assistant messages with proper formatting
• Uses markdown for styling

Ollama Integration

def query_ollama(prompt, model_name, temp):
    response = requests.post("http://localhost:11434/api/generate", ...)

Handles API calls to the local Ollama server for generating responses.

Chat Functionality

• Text input for user messages
• Send button to trigger AI response
• Clear chat button to reset conversation
• Spinner during API processing

Error Handling

• Checks for valid user input
• Handles API request errors
• Displays warning messages when appropriate

Limitations

• Requires local Ollama server
• Limited to supported models (llama2, mistral, codellama)
• No persistent storage for chat history
• Basic error handling for API failures

Future Improvements

• Add persistent storage for chat history
• Implement message streaming
• Add support for more models
• Enhance error handling
• Add conversation export functionality

Troubleshooting

• Ensure Ollama server is running on port 11434
• Verify model availability in Ollama
• Check internet connection for package installation
• Monitor Streamlit logs for errors