Sentiment + Sarcasm Analyzer (Gradio + MCP)

This project is a lightweight Gradio application that performs sentiment analysis and sarcasm detection using Hugging Face Transformers. It is designed to run on CPU and was developed as part of the Hugging Face MCP Course. The app is fully compatible with the Hugging Face MCP server architecture.

Live Demo

👉 Launch the app on Hugging Face Spaces

Related MCP server: Grumpy Senior Developer MCP

Architecture Overview

Models (CPU-only):
- distilbert-base-uncased-finetuned-sst-2-english: Sentiment analysis
- helinivan/english-sarcasm-detector: Sarcasm detection
Frontend: Gradio UI
Backend: Python with Hugging Face Transformers
MCP Integration: Hugging Face MCP-compatible (gradio[mcp])

Features

Sentiment classification: "positive" or "negative"
Sarcasm detection with a probability score
CPU-compatible (no GPU required)
Simple and clean Gradio interface

Output Format

The app returns a structured JSON response with four fields:

{ "assessment": "positive", "confidence": 1.0, "sarcasm_detected": true, "sarcasm_confidence": 0.97 }

Gradio Interface

The interface provides the following controls:

Element	Description
Textbox	Enter text to be analyzed
Submit	Run the sentiment and sarcasm analysis
Clear	Reset the input/output

Setup Instructions

1. Clone the repository

git clone https://github.com/YOUR_USERNAME/mcp-sentiment cd mcp-sentiment

2. Create a virtual environment

python -m venv .venv # Then activate: .venv\Scripts\activate # Windows source .venv/bin/activate # macOS/Linux

3. Install dependencies

pip install -r requirements.txt

Make sure gradio[mcp] is included for MCP compatibility.

4. Add Hugging Face token

Create a .env file:

HF_TOKEN=your_token_here

5. Run the app locally

python app.py

Deploy to Hugging Face Spaces

git init git remote add origin https://huggingface.co/spaces/YOUR_USERNAME/mcp-sentiment git add . git commit -m "Deploy MCP app" git push -u origin main

Once pushed, the MCP server endpoint will be live at:

https://YOUR_USERNAME-mcp-sentiment.hf.space/gradio_api/mcp/sse