README.md

metadata

license: apache-2.0
language:
  - ar
  - he
pipeline_tag: translation
widget:
  - text: رحت أشتري كتاب
base_model:
  - Helsinki-NLP/opus-mt-ar-he
tags:
  - text2text-generation

Arabic to Hebrew Translator (Israeli Arabic Dialect)

This model is a fine-tuned version of Helsinki-NLP/opus-mt-ar-he on an Israeli Arabic dialect corpus, trained for 15 epochs. It aims to translate informal conversational phrases in Israeli Arabic into Hebrew, making it suitable for applications involving colloquial or everyday expressions.

Example Usage

Before using the fine-tuned model, load the tokenizer from the base model:

from transformers import pipeline, AutoTokenizer

# Load tokenizer from the base model
tokenizer = AutoTokenizer.from_pretrained("Helsinki-NLP/opus-mt-ar-he")

# Initialize the translation pipeline with the fine-tuned model and base tokenizer
trans = pipeline("translation", model="HebArabNlpProject/mt-ar-he", tokenizer=tokenizer)
print(trans("رحت أشتري كتاب"))