Translation Training

Translation training focuses on developing robust and accurate machine translation (MT) models, primarily by improving the quality and quantity of training data and refining model architectures. Current research emphasizes leveraging diverse data sources, including automatically generated and post-edited parallel corpora, and exploring techniques like context-aware multi-modal transformers for sign language translation and streaming data processing for efficiency. These advancements aim to enhance MT performance across various language pairs, particularly for low-resource languages, and improve the reliability of automatic translation evaluation metrics.