🏛️ RAMT-LaBSE English → Malay Legal Translation Model

RAMT-LaBSE is a Retrieval-Augmented Machine Translation (RAMT) model based on
Facebook M2M100, fine-tuned specifically for English → Malay legal text translation.

This model improves translation quality in:

• statutes
• contracts
• court documents
• legal terms
• policy & regulatory content

It uses LaBSE embeddings during training for retrieval-augmented sentence matching.

---

📂 Model Architecture

• Base model: facebook/m2m100_418M
• Tokenizer: M2M100 SentencePiece
• Fine-tuned on: English–Malay legal parallel corpus
• Technique: Retrieval-Augmented MT + Contrastive Alignment
• Optimized for: Precision, legal terminology, formal structure

---

🚀 Quick Use (Transformers)

from transformers import M2M100ForConditionalGeneration, M2M100Tokenizer

model_name = "mytranslatenisa/ramt-labse"

tokenizer = M2M100Tokenizer.from_pretrained(model_name)
model = M2M100ForConditionalGeneration.from_pretrained(model_name)

text = "The agreement shall be legally binding upon both parties."

tokenizer.src_lang = "en"
encoded = tokenizer(text, return_tensors="pt")

generated = model.generate(
    **encoded,
    forced_bos_token_id=tokenizer.get_lang_id("ms")
)

print(tokenizer.decode(generated[0], skip_special_tokens=True))