Model Card: hassan-IA/amalaz-en2ha-micro

amalaz-en2ha-micro is a small, fine-tuned translation model that converts English → Hassaniya Arabic. It is based on Helsinki-NLP/opus-mt-en-ar and trained on dah parallel dataset (~3 k examples).

This model aims to improve translation coverage for Hassaniya which is linguistically distinct from Modern Standard Arabic.

Example Usage

from transformers import AutoTokenizer, AutoModelForSeq2SeqLM

model_name = "hassan-IA/amalaz-en2ha-micro"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForSeq2SeqLM.from_pretrained(model_name)

text = "Dah drinks tea 3 times everyday"
inputs = tokenizer(text, return_tensors="pt")
outputs = model.generate(**inputs)

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

Output:

الداه يشرب اتاي اثلت مرات كل يوم