Arabic-English Neural Machine Translation Using Transformer

**Project Overview**

This project implements an Arabic-to-English neural machine translation system using Transformer-based techniques. It includes data preprocessing, separate SentencePiece BPE tokenizers for Arabic and English, model training, and translation generation.

**Key Features**

- Preprocessing and splitting the bilingual dataset into training, validation, and test sets.

- Training SentencePiece tokenizers with BPE to process Arabic and English text.

- Training and saving a neural machine translation model.

- Generating translations using greedy decoding and beam search.

- Evaluating translation quality using BLEU score.

- Building an interactive web interface with Gradio that supports Arabic-English language selection, text input, translated output, and translation history.

**Technologies Used**

Python, PyTorch, Transformer-based Neural Machine Translation, SentencePiece, Byte Pair Encoding (BPE), Gradio, and BLEU evaluation.

The project demonstrates a complete workflow for developing and testing a bilingual machine translation application.