The reporter recently learned from Amazon Cloud Technology that Amazon Cloud Technology is accelerating the three essential core capabilities for leading the construction of data bases, namely the data processing ability required for model fine-tuning and pre training, the ability to quickly combine proprietary data with models to generate unique value, and the ability to effectively process new data to promote the sustained and rapid development of generative AI applications, helping enterprises build a comprehensive data base to fully tap the potential of data and accelerate the landing of generative AI technology.
In the era of generative AI, data is the key for enterprises to stand out. Chen Xiaojian, General Manager of the Product Department of Amazon Cloud Technology in Greater China, stated that the basic model relies on large-scale high-quality datasets, and the differentiation advantage of generative AI comes from proprietary data of enterprises. The large amount of new data generated by generative AI applications also needs to be managed and utilized in a timely and effective manner.
"What enterprises need are generative AI applications that understand business and users, and building such applications requires starting from data." Chen Xiaojian said that Amazon Cloud Technology's three core capabilities for building a data base cover important scenarios from basic model training to generative AI application construction. They can help enterprises easily cope with massive multimodal data, improve basic model capabilities, and endow the unique value of data to basic models and generative AI applications while ensuring user business and data security, accelerating enterprise business growth.
In terms of data storage, scalability and response speed are key. Amazon S3 object storage supports a wide range of data protocols, making it easy to handle various data types, and also supports intelligent layering to reduce training costs. The specially built file storage service Amazon FSx for Lustre can provide sub millisecond latency and throughput performance of millions of IOPS, further accelerating the speed of model optimization.
Accelerating the integration of data and models through technological means has become the second key capability of the enterprise data base. Amazon Cloud Technology has added vector search support to mainstream data services, improving data query performance by storing data and vectors together.
How to handle newly generated data in generative AI applications and improve model invocation efficiency has become the third key capability of enterprise data bases. It is reported that the Amazon Memory DB in memory database caches new data generated from previous Q&A sessions to achieve fast response and accurate answers to similar questions, while effectively reducing the frequency of basic model calls. Amazon Memory DB can store millions of vectors with just a few milliseconds of response time, achieving query performance of millions of times per second with a 99% recall rate.