# 🧠 MarianMT-Text-Translation-AI-Model-"en-fr"

A **sequence-to-sequence translation model** fine-tuned on English–French sentence pairs. This model translates English text into French and is built using the Hugging Face `MarianMTModel`. It’s ideal for general-purpose translation, educational use, and light regulatory or formal communication tasks between English and French.

---

## ✨ Model Highlights

- 📌 Based on [`Helsinki-NLP/opus-mt-en-fr`](https://huggingface.co/Helsinki-NLP/opus-mt-en-fr)
- 🔍 Fine-tuned on a cleaned parallel corpus of English-French sentence pairs
- ⚡ Translates from **English → French**
- 🧠 Built using **Hugging Face Transformers** and **PyTorch**

---

## 🧠 Intended Uses

- ✅ Translating English feedback, emails, or documents into French
- ✅ Cross-lingual support for customer service or regulatory communication
- ✅ Educational platforms and language learning

---

## 🚫 Limitations

- ❌ Not suitable for informal slang or code-mixed inputs
- 📏 Inputs longer than 128 tokens will be truncated
- 🤔 May produce less accurate translations for highly specialized or domain-specific language
- ⚠️ Not intended for legal, medical, or safety-critical translations without expert review

---

## 🏋️‍♂️ Training Details

| Attribute          | Value                              |
|--------------------|----------------------------------|
| Base Model         | `Helsinki-NLP/opus-mt-en-fr`     |
| Dataset            | Parallel English-French corpus   |
| Task Type          | Translation                      |
| Max Token Length   | 128                              |
| Epochs             | 3                                |
| Batch Size         | 16                               |
| Optimizer          | AdamW                            |
| Loss Function      | CrossEntropyLoss                 |
| Framework          | PyTorch + Transformers           |
| Hardware           | CUDA-enabled GPU                 |

---

## 📊 Evaluation Metrics

| Metric     | Score   |
|------------|---------|
| BLEU Score | 27.82   |

---

## 🔎 Output Details

- Input: English text string  
- Output: Translated French text string

---

## 🚀 Usage

```python
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
import torch

model_name = "AventIQ-AI/MarianMT-Text-Translation-AI-Model-en-fr"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForSeq2SeqLM.from_pretrained(model_name)
model.eval()

def translate(text):
    device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
    finetuned_model.to(device)
    inputs = tokenizer(text, return_tensors="pt", padding=True, truncation=True).to(device)
    outputs = finetuned_model.generate(**inputs)
    return tokenizer.decode(outputs[0], skip_special_tokens=True)

# Example
print(translate("Hello, how are you?"))
```
---
## 📁 Repository Structure
```
finetuned-model/
├── config.json               ✅ Model architecture & config
├── pytorch_model.bin         ✅ Model weights
├── tokenizer_config.json     ✅ Tokenizer settings
├── tokenizer.json            ✅ Tokenizer vocabulary (JSON format)
├── source.spm                ✅ SentencePiece model for source language
├── target.spm                ✅ SentencePiece model for target language
├── special_tokens_map.json   ✅ Special tokens mapping
├── generation_config.json    ✅ (Optional) Generation defaults
├── README.md                 ✅ Model card

```
## 🤝 Contributing
Contributions are welcome! Feel free to open an issue or pull request to improve the model, training scripts, or documentation.