Microsoft says it is making upgrades to Translator and other Azure AI services powered by a new family of AI models that researchers are calling Z-code, offering a more efficient language model than other large-scale providers.
According to Microsoft, Z-code is a new family of artificial intelligence models developed by Microsoft that takes advantage of the shared linguistic elements across multiple languages via transfer learning, which Microsoft says applies knowledge from one task to another related task. This is designed to improve quality for machine translation and other language-understanding tasks, in addition to extending those capabilities beyond the world’s most common language to underrepresented languages.
The company says the models use a sparse “Mixture of Experts” approach that is more efficient since it only needs to engage a portion of the model to complete a task, while other architectures have to activate an entire AI model to run every request. This allows scaling while keeping the amount of compute constant.
Microsoft says it is using NVIDIA GPUs and Triton Inference Server to deploy and scale them efficiently for high-performance inference.
According to the company, Z-code models have been deployed to improve common language understanding tasks, including name entity recognition, text summarization, custom text classification and key phrase extraction across Azure AI services. However, this is the first time a company has publicly demonstrated that it can use this new class of models to power machine translation products.
The Z-code-based translation model is available now, first by invitation to customers using document translation in Translator, a Microsoft Azure Cognitive Service and part of Azure AI.
Microsoft says Z-code is part of the company’s larger XYZ-code initiative designed to combine models for text, vision, audio and multiple languages to create more powerful and integrative AI systems that can speak, hear, see and understand people better.
Xuedong Huang, Microsoft’s technical fellow and Azure AI chief technology officer, says the company’s goal is to help everyone and every organization communicate better via improved translations.
“With Z-code we are really making amazing progress because we are leveraging both transfer learning and multitask learning from monolingual and multilingual data to create a state-of-the-art language model that we believe has the best combination of quality, performance and efficiency that we can provide to our customers,” Huang said in a blog.
If you enjoyed this article and want to receive more valuable industry content like this, click here to sign up for our digital newsletters!
Leave a Reply