---
title: 📝 BERT + ELMo Sentence Embeddings Generator
emoji: 🌍
colorFrom: gray
colorTo: red
---
# 📝 BERT + ELMo Sentence Embeddings Generator

This is a Streamlit application that generates sentence embeddings using BERT and ELMo. It supports single sentence input as well as bulk analysis via CSV uploads.

---

## 🌟 Features

- **Model Options**: Choose from multiple BERT variants or ELMo.
- **Input Options**: 
  - Single sentence
  - CSV file with a `sentence` column
- **Embedding Generation**:
  - BERT (Base and Large, Cased and Uncased)
  - ELMo
- **Output**: Download embeddings in a CSV file containing:
  - Original sentence
  - Source (Single or CSV)
  - Embedding vector
- **Preview**: View a preview of embeddings for the first 5 sentences.

---

## 🖥️ How It Works

1. Select the desired model from the dropdown menu.
2. Enter a sentence or upload a CSV file with sentences.
3. Click **Generate Embeddings** to compute embeddings.
4. View the summary including:
   - Source type
   - Model used
   - Embeddings shape
5. Download the embeddings CSV for further use.

---

## 🧩 Use Cases

- **Semantic Search**: Find similar sentences or phrases.
- **Clustering**: Group sentences with similar meanings.
- **Machine Learning**: Use embeddings as features for classification or regression.
- **Visualization**: Explore embeddings in 2D/3D space for analysis.

---

## ⚠️ Notes

- Ensure CSV files have a column named `sentence`.
- BERT and ELMo produce embeddings of different dimensions.
- Single sentence input returns embeddings of shape `(1, embedding_dim)`.
- CSV input returns embeddings of shape `(num_sentences, embedding_dim)`.