Can Amazon Cloud Technology, which has entered generative AI, generate the future again?

The 2023 World Artificial Intelligence Conference (WAIC) held in Shanghai opened on July 6, 2023 with the theme of "Intelligent Connecting the World and Creating the Future". Pre-trained large models and generative AI have become the absolute focus of WAIC 2023. As the world's largest public cloud provider, Amazon Cloud Technology will of course participate in and exhibit at this event.

In early June 2023, Amazon Cloud Technology announced that TII (Technology Innovation Institute), a world-leading scientific research center located in Abu Dhabi, the capital of the United Arab Emirates, trained the open source model Falcon 40B on Amazon Cloud Technology. Falcon 40B, a large language model (LLM) with 40 billion parameters, ranks first on Hugging Face’s open source large language model rankings.

The Falcon 40B model was released by TII in May 2023 under the Apache 2.0 license. Since its release, the Falcon 40B has outperformed current competing models in a variety of benchmarks, demonstrating superior performance without the need for specialized fine-tuning. To make the model more easily accessible to customers, Amazon Cloud Technologies has made the Falcon 40B deployable through Amazon SageMaker Jumpstart.

Amazon Cloud Technology also announced the establishment of the Amazon Cloud Technology Generative AI Innovation Center at the end of June 2023, aiming to help customers successfully build and deploy generative AI solutions. Amazon Cloud Technology has invested US$100 million in the center. The Amazon Cloud Technology Generative AI Innovation Center team consists of strategic experts, data scientists, engineers and solution architects. They will work closely with customers to gradually build customized solutions for generative AI. .

Dr. Sherry Marcus, Director of the Generative AI Product Research Institute of Amazon Cloud Technology, participated in the "WAIC 2023 Industrial Development Plenary Meeting" and the "2023 Shanghai AI Conference Room's 'Original Future' Open Source Innovation Ecosystem Development Forum" and delivered a keynote speech . Sherry Marcus believes that machine learning is now at a turning point for three reasons: the explosive growth of network data and the availability of this data; the increase in computing power; the emergence of the Transformer AI model and the evolution of the complexity of the model.

Sherry Marcus said that the vast majority of customers do not need to train models from scratch and cannot rely on a universal, single large-scale language model to cope with various tasks. The correct approach should be to allow customers to access multiple models. , and then customize your own model based on your own needs and data. Amazon Cloud Technology advocates allowing users to customize their own models based on a huge model in a secure manner and protecting user privacy.

Amazon Cloud Technology can provide customers with already trained and pre-trained models, including Amazon Titan series basic models, or models trained by third-party partners such as Stability AI and Anthropic, as well as through Amazon SageMaker JumpStart Open source model provided.

Sherry Marcus said that customers can use their own data to build specialized or specialized models based on their own needs based on large language models, so that on the one hand, customers can have the richness of large models, and at the same time Fast iteration of small models is also possible.

Regarding the multiple LLM rankings currently appearing, Sherry Marcus believes that the evaluation of these rankings is usually based on specific cases and tasks, but does not represent the full value of the model. Only after applying these models to customers' data can Amazon Cloud Technologies truly judge their effectiveness.

Sherry Marcus believes that the large models in the future will be larger, and there will be more different models, including multi-modal models, and more tailored models. In addition, there will be better understanding of how to be better in specific businesses. You will also gain a deeper understanding by using these models effectively.

Amazon Cloud Technology is rapidly innovating in the field of generative AI. In just 3 months, it has released a number of generative AI tools and platforms including Amazon Bedrock. Sherry Marcus said that the challenge facing generative AI is how to understand its impact on the transformation of various industries and what applications it can bring. At the same time, she urged the industry to pay attention to the disruptive innovations that generative AI may trigger in the future, and to look further ahead. Will Amazon Cloud Technology, as the pioneer of public cloud, once again create a new future in the wave of generative AI?

The above is the detailed content of Can Amazon Cloud Technology, which has entered generative AI, generate the future again?. For more information, please follow other related articles on the PHP Chinese website!