Nafidia is betting on artificial data

NVIDIA has acquired a GRETEL industrial company for nine numbers, according to two people with direct knowledge of the deal.

The sources say that the acquisition price exceeds the latest evaluation of Greell at a value of $ 320 million, although the accurate purchase conditions are still unknown. Gretel and its team of approximately 80 employees will be folded in NVIDIA, where its technique will be published as part of the growing AI group of Cliant Giant of artificial intelligence services based on the core group of developers.

The acquisition comes at a time when NVIDIA introduces artificial data generation tools, so that developers can train their artificial intelligence models and set them for specific applications. In theory, artificial data can create a semi-VINIT supply from artificial intelligence training data and helps solve the problem of scarcity of data waving on the horizon in the artificial intelligence industry because ChatGPT has gone in 2022-although experts say that the use of artificial data in artificial intelligence reaches its own risks.

A NVIDIA spokesman rejected the comment.

Greesl was founded in 2019 by Alex Watson, John Mairez, and Ali Gulshan, who is CEO. The startup offers an artificial data platform and a set of application programming facades for developers who want to create models of obstetric artificial intelligence, but they cannot access adequate training data or have privacy concerns about using real people’s data. Gretel does not build and license its FRONTIERE AI models, but the current open source models are adjusted to add privacy and differential safety features, then collect those together to sell them. The company raised more than 67 million dollars of investment capital financing before acquisition, according to Cook.

A GRETEL spokesman also refused to comment.

Unlike data created by man or in the real world, artificial data is generated and designed as data simulation in the real world. Supporters say this makes data generation required to build artificial intelligence models more developing, less intense of employment, and more accessible to smaller artificial intelligence developers or less resources. Privacy protection is another major selling point for artificial data, making it an attractive choice for health care providers, banks and government agencies.

NVIDIA already offers artificial data tools for developers for years. In 2022, omniver Replicator, which gives developers the ability to generate three -dimensional and accurately accurate data to train nervous networks. Last June, NVIDIA started offering a family of AI open models that generate artificial training data for developers for use in building or refining LLMS. It is called NEMOTRON-4 340B developers, and these mini models can be used by developers to pay their synthetic data through “health care, financing, manufacturing, retail and every other industry.”

During its main presentation at the annual developer conference in NVIDIA this Tuesday, NVIDIA COFENTERNAL and CEO JenSen Huang spoke about the challenges facing the industry in quickly limiting artificial intelligence.

“There are three problems that we focus on,” he said. “One, how can you solve the data problem? How and where do you create the data needed to train artificial intelligence? What is the structure of the model? Then three, what are the scaling laws?” Huang continues to describe how the company is now using the generation of artificial data in its robot platforms.