oza75/nllb-600M-mt-french-bambara

Overview

The demo of the model can be found here: Bambara Translator

Project

This model is part of a larger initiative dedicated to enabling Bambara speakers to access global knowledge without language barriers. Our goal is to eliminate the need for Bambara speakers to learn a secondary language before they can acquire new information or skills. By providing this model, we aim to support the creation of tools for bambara language, thus democratizing access to knowledge.

Bambara Language

Bambara, also known as Bamanankan, is a Mande language spoken primarily in Mali by millions of people as a mother tongue and second language. It serves as a lingua franca in Mali and is also spoken in neighboring countries (Burkina Faso, Ivory Coast etc...). Bambara is written in both the Latin script and N'Ko script, and it has a rich oral tradition that is integral to Malian culture.

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 2e-05
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 1
• mixed_precision_training: Native AMP

Framework versions

• Transformers 4.44.2
• Pytorch 2.4.0+cu121
• Datasets 3.0.0
• Tokenizers 0.19.1