Multi-Model Chatbot

This project is a ChatGPT-like web application built with Streamlit.
It lets you chat with different language models through a unified interface, choose which model to use, and tweak generation parameters (temperature, top_p, max tokens) while seeing responses stream in real time.

---

Features

• Multi-model support via a configurable model_cfg.json file
• Model selection in the UI (choose the active model from a dropdown)
• Live parameters control: temperature, top_p, max_tokens
• Chat interface with conversation history and recent chats sidebar
• Custom system prompt loaded from utils/system_prompt.txt
• Streaming responses from the backend API client

---

Project Structure

• app.py – Main Streamlit app (UI, sidebar, chat logic)
• utils/api_client.py – API client used to load models and stream responses
• utils/model_cfg.json – Configuration file describing available models (name, context length, purpose, etc.)
• utils/system_prompt.txt – System prompt injected at the beginning of each conversation

Note: You can edit model_cfg.json and system_prompt.txt to adapt the chatbot to your own models and use cases.

---

Installation

It is recommended to use a virtual environment (conda or venv).

# 1. Clone the repository
git clone https://github.com/yaghmo/chatbot.git
cd chatbot

# 2. Create a virtual environment (example with conda)
conda create -n chatbot_env python=3.10
conda activate chatbot_env

# 3. Install dependencies
pip install -r requirements.txt

Available Models

All models run on GPU:

Model Name	Device	Speed
TheBloke/Mistral-7B-Instruct-v0.2-GGUF	GPU but needs CPU	Fast
TheBloke/Mistral-7B-Instruct-v0.2-GGUF	CPU	Slow

Usage

After installing dependencies, Start the api server then launch the interface with:

python api_server.py
# wait 2-3 secs

streamlit run interface.py

or simply start the launcher, it will handle both files.

python launch.py

A web application will open where you can interact with models you select from settings, the cahtbot will then answer the user back.

Examples