Technological change is about to break out. Who will unleash the potential of enterprise-level generative AI?

242f4691aeb706ae1309dca529a49c35.gif

Editor: A Mao

Design: Muyou

As the founder and leader of global cloud computing, Amazon Cloud Technology releases a large number of new services, new functions and new applications at the re:Invent global conference at the end of November every year, and this year is certainly no exception.

There are so many flowers that the eyes can’t see them all. Because of this, the Amazon Cloud Technology re:Invent China city tour held after the annual global conference always becomes one of the most popular and most watched series of technology events at the end of the year.

Even if there was snow and traffic jams that day, it could not slow down the technical crowd at all. On December 12, Amazon Cloud Technology re:Invent China Tour Beijing Station opened grandly, and even before the conference started, the venue was already packed.

According to Chen Xiaojian, general manager of the product department of Amazon Cloud Technology Greater China , more than 80% of startup unicorn companies with a global valuation of more than US$1 billion choose to operate their workloads on Amazon Cloud Technology.

f814fdec86d421c147e2515073ccc4d1.jpeg

For start-ups, their technological innovations must be built on the safest and most reliable platform. This is why companies widely value and choose Amazon Cloud Technology - Amazon Cloud Technology that has reshaped IT infrastructure. Fundamentally, we have achieved universal access to technology, allowing companies even on their first day of business to start on the top platform.

Chen Xiaojian said that the emergence of generative AI has allowed more and more companies to see more and greater possibilities for realizing business value. To this end, Amazon Cloud Technology has created a new paradigm of enterprise-level generative AI to help customers create their products.

"From responsible AI to data security capabilities, we have made multi-dimensional updates. Amazon Cloud Technology always puts enterprise-level stringent requirements at the top of the product design concept, which also ensures that we are an enterprise enterprise construction and application generation The best choice for modern AI," he said.

Amazon Cloud Technology hopes to use its own efforts to lower the threshold for the application of generative AI in thousands of industries, while taking into account scale and cost, and protecting data security and privacy, from three different levels, to help enterprises use generative AI for their most important applications. create optimal business scenarios, thereby reshaping industries and changing lives.

Under the hood, Amazon Cloud Technologies provides the infrastructure for basic model training and inference. At this conference, Amazon Cloud Technology released the self-developed chip Amazon Trainium 2 - as can be seen from the name, this is a dedicated chip for generative AI and machine learning training.

Currently, a number of customers, including Adobe and Deutsche Telekom, are already using Amazon Trainium 2 to design, deploy and support new generative AI applications. In fact, Amazon Cloud Technology’s search team is also using Amazon Trainium 2 to build large-scale deep learning models.

As the world's first supplier to bring NVIDIA's latest chip H100 GPU to the cloud, Amazon Cloud Technology's Amazon EC2 P5 instances bring amazing performance, reducing costs by 40% and being four times faster than their predecessors. So many.

According to Chen Xiaojian, Amazon Cloud Technology will also provide the first cloud AI supercomputer equipped with NVIDIA Grace Hopper super chip and Amazon Cloud Technology UltraClusters technology; the first NVIDIA DGX cloud using NVIDIA’s latest chip GH200 NVL32 will soon be launched on Amazon Cloud Technology; in addition, The two companies will also launch a series of in-depth cooperation in AI training, research and development, customized development, etc.

Through the Amazon SageMaker HyperPod tool, Amazon Cloud Technology can help customers significantly shorten training time by 40%. It can not only automatically detect and locate hardware faults, and automatically replace faults, but also "continue" training at breakpoints when a fault occurs. Nor will all previous efforts be wasted.

In the middle tool layer, Amazon Cloud Technology provides the Amazon Bedrock platform for generative AI applications . As a builder's tool, Amazon Bedrock provides some industry-leading open source models, including the Amazon Titan model developed by Amazon Cloud Technology.

ccd249d6722d66800852c1f2a4944f98.jpeg

Chen Xiaojian said that Amazon Bedrock supports Anthropic's latest Claude 2.1 model, which not only provides an industry-leading 200K mark context window, but also greatly improves accuracy, making it very suitable for large-scale tasks such as language modeling, text generation and dialogue systems. .

Around Amazon Bedrock, Amazon Cloud Technology has carried out a series of enhancements in pre-training, fine-tuning, agents and strategies to ensure the intensity and depth of training while also ensuring that any customer data will not be used to train the underlying model.

At the same time, Amazon Cloud Technology has also newly released Amazon Titan Text Express , Amazon Titan Multimodal Embeddings and Amazon Titan Image Generator services to provide customers with generative AI applications in terms of text generation, multi-modal search and image generation preview. support.

At the top application layer, Amazon Cloud Technology has launched a new generative AI assistant Amazon Q - which is also one of the biggest highlights of this re:Invent global conference. As a chatbot, Amazon Q does not face consumers, but various employees in work scenarios, helping them complete their daily work efficiently and accurately.

a8b70e4e2623afe341fa6b6a4d3d8925.jpeg

Currently, Amazon Q can already leverage business data from many well-known applications, including Microsoft 365, Slack, Salesforce, Dropbox, and Amazon Cloud Technology’s own Amazon S3.

Chen Xiaojian pointed out that Amazon Q is not only an expert in enterprise business, it is also an expert in business intelligence. By introducing Amazon Q into Amazon QuickSight 's BI application, it can respond to customer requests within seconds, create accurate, beautiful and clear descriptions of monthly business changes, and provide decision-making basis for enterprise customers.

Of course, Amazon Q is also a contact center specialist—after all, it still has “chat” in its DNA. By bringing it to Amazon Connect , Amazon Q can detect customer issues based on real-time conversations and automatically respond, make recommendations, and provide relevant information.

"As a leader in technology, Amazon Cloud Technology has always been committed to bringing value to our customers. Through the re:Invent global conference, we use technology to help customers reshape their businesses and help them do better - this period The journey has just begun." Chen Xiaojian said at the end.

( Personal point of view, only for reference )

81aad935a4a6cf8a8ef9d6b5f93571f5.png

Guess you like

Origin blog.csdn.net/pangtout/article/details/135301588