English to Hindi Translator

---

Overview

The English to Hindi Translator is a machine translation system that leverages the MarianMTModel from Hugging Face's Transformers library. It translates English sentences into Hindi using a neural machine translation approach. The system is designed to process English text, translate it into Hindi, and output the translated text.

Currently, the translation system operates with a dataset of English-Hindi sentence pairs. Future enhancements will include better preprocessing techniques and more sophisticated translation features.

---

Project Goals

• Sentence Translation: Translate English sentences into Hindi using the MarianMTModel.
• Training and Fine-Tuning: Train the model on a custom dataset of English-Hindi sentence pairs.
• User-Friendly Translation: Provide a function to easily translate sentences and obtain results in Hindi.

---

Features

• Custom Dataset: Utilizes a dataset of English-Hindi sentence pairs for training the model.
• Translation Model: Employs MarianMTModel for translating English sentences to Hindi.
• Training Loop: Implements a training process with manual tokenization and padding.
• Translation Function: Allows translation of individual English sentences into Hindi.

---

Dataset

The dataset used for training consists of English-Hindi sentence pairs stored in modified_hin.txt. Examples from the dataset include:

• Wow! - वाह!
• Help! - बचाओ!
• Jump. - उछलो.
• Hello! - नमस्ते।
• Cheers! - वाह-वाह!
• Got it? - समझे कि नहीं?
• I'm OK. - मैं ठीक हूँ।
• Awesome! - बहुत बढ़िया!
• Come in. - अंदर आ जाओ।
• Get out! - बाहर निकल जाओ!
• Go away! - चले जाओ!
• Goodbye! - ख़ुदा हाफ़िज़।
• Perfect! - उत्तम!
• Welcome. - आपका स्वागत है।

---

Model Architecture

• MarianMTModel: Utilizes Hugging Face's MarianMTModel for machine translation.
• Feature Extraction: Extracts features from English sentences for translation to Hindi.
• Translation Algorithm: Maps English sentence features to Hindi using the pre-trained MarianMTModel.

---

Training

The model is trained using PyTorch with the following process:

1. Data Loading: Loads and preprocesses the dataset with a custom Dataset class.
2. Tokenization & Padding: Performs manual tokenization and padding of sentences.
3. Training Loop: Executes forward and backward passes, computes loss, and updates model parameters.

---

Usage

To translate a sentence from English to Hindi, use the translate_sentence function. For example:

english_sentence = "Hello!"
translated_sentence = translate_sentence(english_sentence)
print(f"English: {english_sentence}")
print(f"Hindi: {translated_sentence}")