gujlish-sentiment-analysis

Gujlish & Hinglish Sentiment Analysis Model (MuRIL Fine-Tuned)

This model is a fine-tuned version of google/muril-base-cased trained on a custom synthetic dataset of 150,000 code-mixed samples. It is designed to detect sentiment (Positive, Negative, Neutral) in Indian languages, specifically handling Code-Switching (mixing Gujarati/Hindi with English).

Project Overview

• Problem: Standard AI models struggle with "Gujlish" (Gujarati written in English script) or "Hinglish".
• Solution: This model uses MuRIL (Multilingual Representations for Indian Languages) and was fine-tuned on a massive synthetic dataset covering diverse topics like Tech, Movies, Food, and Daily Life.
• Performance: Achieved 100% Accuracy on the validation set.

Intended Use

This model is ideal for:

• Analyzing social media comments (YouTube, Instagram) in India.
• Processing audio transcriptions from mixed-language conversations.
• Customer feedback analysis for Indian apps.

How to Use

from transformers import pipeline

# Load the model
classifier = pipeline("text-classification", model="marshal-yash/gujlish-sentiment-analysis")

# Test on Code-Mixed Text
text = "Aa movie bahu saras che, maja aavi!"
result = classifier(text)

print(result)
# Output: [{'label': 'Positive', 'score': 0.99}]