En To Dzo Nllb Mul Mt Nlp M4
This model is a fine-tuned English to Dzongkha translation model based on facebook/nllb-200-distilled-600M
Downloads 22
Release Time : 1/7/2025
Model Overview
An English to Dzongkha machine translation model based on the NLLB-200 distilled model, focusing on high-quality text translation tasks
Model Features
High-Quality Translation
Achieves a BLEU score of 19.2 on the evaluation set, indicating good translation quality
NLLB-Based Architecture
Uses Facebook's NLLB-200 distilled model as the foundation, with strong multilingual capabilities
Optimized Training
Fine-tuned over 9 epochs, reducing training loss from 2.2141 to 0.8253
Model Capabilities
English to Dzongkha text translation
Semantic-preserving translation
Handling medium-length texts (average generation length of 15-16 words)
Use Cases
Language Services
Document Translation
Translate English documents into Dzongkha
Achieves fluent translation while preserving original semantics
Cultural Exchange
Facilitates cultural content exchange between English and Dzongkha
Featured Recommended AI Models
ยฉ 2025AIbase