Skip to content

Language to Language Transformer model from scartch using pure Pytorch where I used my transformer model for translation task. from the paper "Attention is all you Need" 2017 using pytorch.

License

Notifications You must be signed in to change notification settings

Esmail-ibraheem/LlTRA-Model

Repository files navigation

LlTRA-Model.

Transformer_Diagram_2 drawio

LlTRA stands for: Language to Language Transformer model from the paper "Attention is all you Need", building transformer model:Transformer model from scratch and using it for translation using pytorch.

Problem Statement:

In the rapidly evolving landscape of natural language processing (NLP) and machine translation, there exists a persistent challenge in achieving accurate and contextually rich language-to-language transformations. Existing models often struggle with capturing nuanced semantic meanings, context preservation, and maintaining grammatical coherence across different languages. Additionally, the demand for efficient cross-lingual communication and content generation has underscored the need for a versatile language transformer model that can seamlessly navigate the intricacies of diverse linguistic structures.


Goal:

Develop a specialized language-to-language transformer model that accurately translates from the Arabic language to the English language, ensuring semantic fidelity, contextual awareness, cross-lingual adaptability, and the retention of grammar and style. The model should provide efficient training and inference processes to make it practical and accessible for a wide range of applications, ultimately contributing to the advancement of Arabic-to-English language translation capabilities.


Dataset used:

from hugging Face huggingface/opus_infopankki


Configuration:

this is the settings of the model, You can customize the source and target languages, sequence lengths for each, the number of epochs, batch size, and more.

def Get_configuration():
    return {
        "batch_size": 8,
        "num_epochs": 30,
        "lr": 10**-4,
        "sequence_length": 100,
        "d_model": 512,
        "datasource": 'opus_infopankki',
        "source_language": "ar",
        "target_language": "en",
        "model_folder": "weights",
        "model_basename": "tmodel_",
        "preload": "latest",
        "tokenizer_file": "tokenizer_{0}.json",
        "experiment_name": "runs/tmodel"
    }

Search algorithm used:

Greedy Algorithm for finding which token has the maximum probability.


Training:

I used my drive to upload the project and then connected it to the Google Collab to train it:

  • hours of training: 4 hours.
  • epochs: 20.
  • number of dataset rows: 2,934,399.
  • size of the dataset: 95MB.
  • size of the auto-converted parquet files: 153MB.
  • Arabic tokens: 29999.
  • English tokens: 15697.
  • pre-trained model in collab.
  • BLEU score from Arabic to English: 19.7

Some Results:

SOURCE: العائلات الناطقة بلغة أجنبية لديها الحق في خدمات الترجمة عند اللزوم.
TARGET: A foreign-language family is entitled to interpreting services as necessary.
PREDICTED: in a native language, it is provided by the services of the services for the elderly.
--------------------------------------------------------------------------------
SOURCE: قمت بارتكاب جرائم وتُعتبر بأنك خطير على النظام أو الأمن العام.
TARGET: you have committed crimes and are considered a danger to public order or safety
PREDICTED: you have committed crimes and are considered a danger to public order or safety
--------------------------------------------------------------------------------
SOURCE: عندما تلتحق بالدراسة، فستحصل على الحق في إنجاز كلتا الدرجتين العلميتين.
TARGET: When you are accepted into an institute of higher education, you receive the right to complete both degrees.
PREDICTED: When you have a of residence, you will receive a higher education degree.
--------------------------------------------------------------------------------
SOURCE: اللجنة لا تتداول حالات التهميش والتمييز المتعلقة بالعمل.
TARGET: The Tribunal does not handle cases of employment-related discrimination.
PREDICTED: The does not have to pay and the work.
--------------------------------------------------------------------------------
SOURCE: يجب عليك أيضاً أن تستطيع إثبات على سبيل المثال بالوصفة الطبية أو بالتقرير الطبي بأن الغرض من الدواء هو استخدامك أنت الشخصي.
TARGET: In addition, you must be able to prove with a prescription or medical certificate, for example, that the medicine is intended for your personal use.
PREDICTED: You must also have to prove your identity with a friend or friend, for example, that the medicine is intended for your personal use.
--------------------------------------------------------------------------------
SOURCE: إذا كان لديك ترخيص إقامة في فنلندا، ولكن لم تُمنح ترخيص إقامة استمراري، فسوف تصدر دائرة شؤون الهجرة قراراً بالترحيل.
TARGET: If you already have a residence permit in Finland but are not granted a residence permit extension, the Finnish Immigration Service makes a deportation decision.
PREDICTED: If you have a residence permit in but are not granted a residence permit, the Service makes a decision.

check the theoretical part: Theoretical part

developing process:

  1. https://youtu.be/oPaQGmBDxh4
  2. https://youtu.be/Nm5CUo7ol18?si=5U3m4IM7crJ23Zzu
  3. https://youtu.be/O-jfimyP6Tw?si=ucPweGo2b7gh2rrI

About

Language to Language Transformer model from scartch using pure Pytorch where I used my transformer model for translation task. from the paper "Attention is all you Need" 2017 using pytorch.

Topics

Resources

License

Stars

Watchers

Forks

Packages

No packages published