Strong Compute raises $7.8M to accelerate ML training pipelines – TechCrunch
strong calculation, a Sydney, Australia-based startup that helps developers remove bottlenecks in their machine learning training pipelines, announced today that it has raised a $7.8 million funding round. dollars. The round includes a total of 30 funds and angels, including Sequoia Capital India, Blackbird, Folklore and Skip Capital, as well as Y Combinator, Starburst Ventures and founders and engineers from companies like Cruise, Waymo, Open AI, SpaceX and Galactic Virgin.
The company, which was part of Y Combinator Winter Bundle 22, promises that its optimizations can speed up the training process by 10 to 1000 times, depending on the model, pipeline and framework. As the founder of Strong Compute, Ben Sands, who also co-founded the company AR Metatold me, the team recently made breakthroughs where they were able to take Nvidia’s reference implementation, which their client LayerJot used, to run 20 times faster.
“It was a big win,” Sands said. “It really gave us the feeling that there’s nothing that can’t be improved.” He didn’t really want to reveal all the details of how the team’s optimizations work, but he noted that the company is now hiring mathematicians and building tools that give it a more detailed view of how their user’s code interacts with CPUs and GPUs at a much deeper level than was previously possible.
As Sands pointed out, the company’s current goal is to start automating much of the ongoing work to optimize the training process – and that’s something the company can now address, thanks to this round of funding. “Our goal now is to have serious development partners in self-driving, medical and aviation, to see what will actually go mainstream very well,” he explained. “We now have the resources to have an R&D team that doesn’t have to deliver something in a two-week sprint, but can actually look at what real core technology is that could take a year to get a win. but it can really help with this automated analysis of the problem.
The company currently has six full-time engineers, but Sands plans to double that number over the next few months. In part, it’s also because the company is now attracting interest from large enterprises that often spend $50 million or more on their compute resources (and Sands noted that the market is essentially bimodal, with customers spending less from $1 million or $10 to $100 million, with only a few players in the middle).
However, all companies trying to build ML models suffer from the same problem: training models and running experiments to improve them still take a long time. This means that well-paid data scientists working on these problems spend a lot of time in a waiting pattern, waiting for results to arrive.Strong Calculate solves the basketball court problem,” said Nikhil Abraham, CFO of SteadyMD. “The long training periods meant that our best developers had to shoot hoops all day, waiting on machines.”
And while some of that inbound interest is coming from the financial industry and companies looking to optimize their natural language processing models, Strong Compute remains focused on computer vision for now.
“We’ve only scratched the surface of what machine learning and AI can do.” said Folklore partner Tanisha Banaszcyk. “We love working with founders who have long-term ambition and visions that will endure through generations. Having invested in autonomous driving, we know how important speed to market is – and see the impact Strong Compute can have in this market with its purpose-built platform, deep understanding of the $500 billion market, and world-class team.