Skip to content
Projects

Strifee/arabic2english

View on GitHub

Strifee/arabic2english has 29 GitHub stars and 6 forks. GitHot measured 0 stars in 24 hours. Last checked 2026-11-15 15:58 UTC. Last push 2026-04-01.

How these counts are measured

Measured star growth

in 24h
0
in 7 days
Update delayed
in 15 days
Update delayed
in 1 month
Update delayed

Net star changes over overlapping periods, not daily totals. Missing measurements are left blank. Stars measure attention, not software quality.

About

Arabic to English machine translation with Transformers and Pytorch

Machine Learning

README.md

Original

Arabic2English - Arabic to English Translator

This is a PyTorch implementation of an Arabic to English Neural Machine Translation built using Transformers architecture (Attention Is All You Need)

Why Arabic→English is Hard

Arabic is morphologically rich — one word can express what takes a full sentence in English. It's also right-to-left with complex root-based derivation. Standard seq2seq models struggle with this — Transformers handle it better due to attention mechanisms capturing long-range dependencies.

Model Architecture

  • Custom Transformer (Attention Is All You Need)
  • 3 encoder + 3 decoder layers
  • 8 attention heads
  • Embedding size: 256
  • Trained from scratch — no pretrained weights

Setup and Requirements

1. CUDA:
install CUDA before installing the required packages or check if it is already installed

2. Clone the Translate repo:

$ git clone https://github.com/Strifee/arabic2english.git

3. install requirements:

pip install -r requirements.txt

if you have problem with CUDA package try this:

conda install -q pytorch torchvision cudatoolkit=11 -c pytorch-nightly

Data

Arabic to English Translation Sentences :

Arabic to English Translation Sentences is a dataset for machine translation between English and Arabic.

Training

1. Clone the Translate repo:

$ git clone clone https://github.com/Strifee/arabic2english.git

2. Training

$ python translate.py

2. Regularization

Hyperparameters :

BATCH_SIZE = 16
learning_rate = 0.0001
num_epochs = 30

num_heads = 8
num_encoder_layers = 3
num_decoder_layers = 3

max_len= 230
dropout = 0.4
embedding_size= 256

Before regularization :

image

After regularization :

image

Results

"I'm ready" -> 'أنا مستعد'
"i'm lucky"  -> 'انا محظوظ'
"I'm sad" -> 'أنا حزين'