Fyafulla AI: Tamang Language Translation and Language Model

Machine Translation and Language Modeling Project

The Machine Translation Project at Fyafulla.com is designed to bridge language barriers between English, Nepali, and Tamang. It aims to provide accurate, real-time translations to promote accessibility, language learning, and cultural preservation.

While major tech companies have created powerful translation tools for widely spoken languages, low-resource languages like Tamang are often left behind. This project is an independent effort to bring modern AI technology to a language that is deeply important to millions of people.

Tamang Language Modeling Project (Tamang GPT)

Alongside translation, our Language Modeling Project is designed to generate natural Tamang sentences based on an input word or phrase, as well as compose complete Tamang sentences from scratch.

Trained specifically on Tamang linguistic patterns, our Tamang GPT prototype produces coherent, grammatically structured Tamang text to assist learners, educators, and writers.

Project Goals

The primary goals of the project are to:

Methodology: How We Train the AI

The project uses state-of-the-art Natural Language Processing (NLP) architectures, specifically Neural Machine Translation (NMT) and GPT-based models. To train our translation systems, we use parallel text corpora—datasets where English and Nepali sentences are paired with their Tamang equivalents. For monolingual language modeling, we train on large collections of verified Tamang texts.

A Critical Challenge: Unlike English, there is very little digital text available in Tamang, as much of the language has historically been oral. This means our project involves collecting, transcribing, and verifying sentences alongside native speakers so the AI can learn accurate patterns.

Data and Training

Our parallel dataset spans tens of thousands of aligned sentences and continues to expand for stronger translation accuracy, alongside larger monolingual corpora for language modeling. The data preparation workflow involves several steps:

Expected Impact

Our AI translation system and language model empower Tamang speakers to communicate more easily online, support new language learners, and contribute to the preservation and growth of the Tamang language in the digital age. They also serve as a foundation for future AI-based applications, including speech tools and interactive learning platforms.

By making Tamang accessible on the web, we hope to encourage younger generations to connect with their heritage and ensure the language continues to thrive in a digital world.

Explore the Project

Want to see the technology in action? Try our English-Nepali-Tamang Translator, test Tamang GPT, or read How the AI Works.