Z-code multilingual model representation diagram

Project Z-Code

新闻与深度文章

Z-Code multilingual model

新闻报道 | Microsoft Translator Blog

Microsoft is on a quest for AI at Scale …

Photo of a young child writing on a chalkboard. The word “hello” is written on the chalkboard in multiple languages.
微软研究院博客

| Krishna Doss Mohan 和 Jann Skotdal

Today, we’re excited to announce that Mi…

新闻报道 | VentureBeat

Today, Microsoft announced that Microsof…

新闻报道 | Microsoft AI Blog

Microsoft announced today that 12 new la…

DeepSpeed MoE powers eight times bigger models using expert-parallelism + ZeRO-Offload compared with expert-parallelism only. A graph shows supported model sizes on NVIDIA A100 GPUs. DeepSpeed MoE scales near-linearly with respect to the number of GPUs. Z-code MoE (10B) consistently outperforms other systems on BLEU scores for an in-house 50 language test dataset. Read more in the blog post. 
微软研究院博客

| DeepSpeed Team 和 Z-code Team

Today, we are proud to announce DeepSpee…

新闻报道 | Microsoft AI – Cognitive Services Blog

The extractive summarization feature in …

新闻报道 | Microsoft AI – Cognitive Services Blog

It has been a year since we released (in…

新闻报道 | customers.microsoft.com

微软研究院博客

| Xuedong Huang

At Microsoft, we have been on a quest to…