Paperspace Introduces Toolkit For AI Developers Looking To Provision GPUs In The Cloud

ARTICLE TITLE HERE

Paperspace, a startup looking to fast track AI development, introduced a new product Wednesday to make it easier for developers to access GPU-powered cloud infrastructure.

The company out of Brooklyn, N.Y. has focused on facilitating the use of GPUs for machine learning and deep learning projects that need their computational power. Gradient is a toolkit with functionality that streamlines deployment of those GPU-accelerated instances, Dillon Erb, a Paperspace co-founder, told CRN.

Paperspace was founded three years ago on the premise that GPUs "were going to really enter into the cloud space in a big way," Erb said. With that vision, Paperspace built a software layer to automate provisioning and managing that infrastructure.

[Related: Here's Who Made Gartner's 2018 Magic Quadrant For Data Science And Machine Learning]

id
unit-1659132512259
type
Sponsored post

Early use cases were visual effects, data-intensive computing, CAD and virtual desktops. As artificial intelligence broke into the mainstream in recent years, however, more customers turned to the startup for training machine learning and deep learning algorithms.

Gradiant makes accessing the infrastructure required by those intelligent apps as easy as it is for any web app, driving industry adoption of AI, Erb said.

"Now they have to cobble together a bunch of open source projects," he said. "Paperspace gives you an out-of-the-box solution."

Graphics processors, historically used for gaming, in recent years have become a common tool for training machine learning models because of their computational advantages over CPUs.

While the infrastructure has become more ubiquitous, "if you're a company and want to start building machine learning or integrating it into existing business processes, there aren't a lot of tools to do that," Erb said.

"Paperspace can really automate the infrastructure side of that, so you're working on the things you're trying to solve," he said.

Paperspace operates infrastructure for most customers, serving as their IaaS provider, but its platform can also run in the public cloud.

Gradient rapidly integrates with Jupyter Notebooks, which have become the de facto standard for interacting with machine learning code. Developers can add one line of Python code to run their apps on a GPU-powered cloud, with orchestration taking place behind the scenes. There's also a new GPU job runner.

"You give us a container, some code and the command you want to run, and we run that as performant and efficiently as possible," Erb said.

Paperspace is on a mission accelerate development of AI products and services across industry verticals, he said.

The company has already launched a partner program to take its solution to the enterprise and hopes Gradient will pour fuel on an emerging channel ecosystem.

"We [have] kicked off a channel partner program at Paperspace," Erb said. "That's something we're excited to grow, and Gradient really opens that up significantly for us."

Early partners were smaller solution providers specializing in graphics and analytic niches. But more "traditional big guys" are coming into the fold as machine learning becomes democratized and prevalent across industry applications, he said.