M2M-100 is a multilingual, many-to-many machine translation model developed by Facebook AI Research. It enables direct translation between 100 languages without relying on English as an intermediate. Built using FAIRSEQ, this model supports training, evaluation, and inference for diverse multilingual datasets.
Key Features
Supports direct translation between any of 100 languages
Does not require pivoting through English
Built on FAIRSEQ, optimized for research and scalability
Pre-trained models and configuration scripts included
Capable of zero-shot and supervised translation
High translation quality with strong BLEU performance on benchmark datasets
Android DevHub
.png)
.png)
.png)
.png)
