README.md

---
library_name: transformers
tags: [bangla, sentiment-analysis, NLP, transformers, onubhuti]
---

# 🌟 Onubhuti: Bangla Sentiment Analysis Model

**Onubhuti** is a sentiment analysis model specifically trained to classify Bangla text into **Positive**, **Neutral**, or **Negative** sentiments. Fine-tuned from **FB XLM-Roberta**, it leverages state-of-the-art NLP techniques for robust performance. This model can be applied to various natural language processing (NLP) tasks such as customer reviews, social media analysis, and more.

---

## 📖 Model Details  

### 📋 Model Description  

**Onubhuti** is a transformer-based model fine-tuned on Bangla text for sentiment classification tasks. It has been trained to classify Bangla sentences into three sentiment categories:

- **Positive**: Sentiments that express approval, happiness, or general positivity.
- **Neutral**: Sentiments that are neutral or have no clear positive or negative sentiment.
- **Negative**: Sentiments that express disapproval, dissatisfaction, or general negativity.

The model uses **FB XLM-Roberta**, a transformer model pre-trained on multilingual data and further fine-tuned specifically for Bangla. This enables the model to understand nuances in sentiment for Bangla text with high accuracy.

- **Developed by:** Nazmus Sakib Apurba
- **Funded by:** Not Funded
- **Shared by:** Tamim
- **Model type:** Transformer-based classification model
- **Language(s):** Bangla (বাংলা)
- **License:** Apache 2.0 (or any relevant license)
- **Finetuned from model:** FB XLM-Roberta

---

### 🔗 Model Sources  

- **Repository:** [Your Hugging Face model URL](#)  
- **Paper:** [Optional, link to relevant paper if available]  
- **Demo:** [Optional, link to any demo or related page]  

---

## 🚀 Uses

**Onubhuti** is designed for sentiment analysis tasks and can be used in various contexts:

- **Social Media Monitoring:** Analyzing tweets, Facebook posts, or other social media content for sentiment.
- **Customer Feedback Analysis:** Classifying customer reviews, survey responses, or feedback.
- **Content Moderation:** Identifying positive, negative, or neutral comments in a user-generated content environment.

### 🛠️ Installation and Usage

1. **Installation:**  
   To use this model, you need to install the necessary libraries:
   ```bash
   pip install transformers torch
   ```

   ```bash
   # Load model directly
   from transformers import AutoTokenizer, AutoModelForSequenceClassification

   tokenizer = AutoTokenizer.from_pretrained("Apurba3036/onubhuti-bangla-sentiment")
   model = AutoModelForSequenceClassification.from_pretrained("Apurba3036/onubhuti-bangla-sentiment")
   ```
   ```bash
   💡 Example Output:
   
   Bangla text: এই জায়গাটা এত সুন্দর যে ভাষায় বর্ণনা করা কঠিন। পাহাড় আর সবুজ গাছপালার মিশেলে প্রকৃতি যেন তার সেরা রূপ এখানে মেলে ধরেছে। আমি এই জায়গায় বারবার আসতে চাই।
   Predicted sentiment: Positive
   ```