A lightweight chatbot application that detects user emotions from text using a fine-tuned DistilBERT model and generates empathetic responses. Built using FastAPI, deployed via Docker and AWS.
🚀 Features
🎭 Emotion detection using fine-tuned BERT (DistilBERT) model
🤖 Context-aware chatbot responses with emojis
💬 Lightweight frontend interface (HTML + JS)
🐳 Dockerized for easy deployment
☁️ EC2-hosted with public access
🧱 Tech Stack
Frontend: HTML, JavaScript (fetch API)
Backend: FastAPI, Python
Model: Hugging Face Transformers (DistilBERT)
Deployment: Docker, AWS EC2
📁 Project Structure
🔧 Setup Instructions
- Clone the Repository
git clone https://github.com/Rishabh000/MindEase.git
cd MindEase
- Build and Run with Docker
docker build -t chatbot-app .
docker run -d -p 80:8000 --name chatbot chatbot-app
Access at: http:///
💬 API Endpoints
Method Endpoint Description
GET / Loads the chatbot UI
POST /chat Accepts a message and returns emotion-based response
Example request:
POST /chat { "message": "I'm feeling low today..." }
Example response:
{ "input": "I'm feeling low today...",
"emotion": "sadness",
"response": "😢 I'm here for you. Would you like to talk about what’s making you feel this way?" }
Acknowledgments:
Transformers by Hugging Face, FastAPI, ClearML for training tracking