DeepMind AI group moves from Torch framework to Google’s own TensorFlow

Google’s DeepMind artificial intelligence (AI) research group today announced that for all future research it will use TensorFlow, a machine learning library that Google open-sourced last year, instead of Torch, an older framework.

The move suggests that some of Google’s brightest AI minds are convinced of the promise of Google’s own open source software; TensorFlow is now good enough for DeepMind.

“We believe that TensorFlow will enable us to execute our ambitious research goals at much larger scale and an even faster pace, providing us with a unique opportunity to further accelerate our research programme,” Koray Kavukcuoglu, a research scientist at Google DeepMind and one of Torch’s core contributors, wrote in a blog post.

This is important because of DeepMind’s considerable capabilities — earlier this year its AlphaGo AI player of the ancient Chinese board game Go beat top-ranked Go player Lee Sedol.

To be sure, DeepMind is not Google’s only AI research unit. Google also has the larger Google Brain team.

In a letter to Alphabet shareholders yesterday, Google chief executive Sundar Pichai played up the importance of AI. “We’ve been building the best AI team and tools for years, and recent breakthroughs will allow us
to do even more,” Pichai wrote.

There are several other open-source deep learning frameworks to choose from, but it would only be right for Google’s internal groups to gradually align with its own open source tooling.

“Our transition to TensorFlow represents a new chapter, and I feel very excited about the prospect of DeepMind contributing heavily to another great open source machine learning platform that everyone can use to advance the state-of-the-art,” Kavukcuoglu wrote.

Google said earlier this month that TensorFlow could now support training across multiple machines, not just one.