Google makes partnership with Facebook on AI technology

This past Tuesday, Google and Facebook announced a partnership to enable the open-sourced machine learning framework PyTorch to work with Tensor-Processing Units (TPUs). This partnership could signal a new age of collaboration towards AI research.

“Today, we’re pleased to announce that engineers on Google’s TPU team are actively collaborating with core PyTorch developers to connect PyTorch to Cloud TPUs. The long-term goal is to enable everyone to enjoy the simplicity and flexibility of PyTorch while benefiting from the performance, scalability, and cost-efficiency of Cloud TPUs.” – Director of Product, Rajen Sheth

PyTorch is Facebook’s open-source framework which enables development of mathematical programs like those used in Artificial Intelligence research. Such frameworks allow researchers to develop arbitrarily complicated mathematical computational graphs and automatically calculate derivatives.

TPUs are computer chips designed by Google specifically for AI systems. According to Google, TPUs are 15x to 30x faster than conventional Graphical Processing Units (GPUs).

Why TPUs on Pytorch Matter

TPUs on Pytorch
The combination of large amounts of data and training neural networks on GPUs was the catalyst for the current success in deep learning systems. Neural networks that could take months to train, can train in just a few hours when using GPUs. As deep learning has matured, neural networks and datasets have become much larger. These networks can now take months to train on GPUs. Google’s proprietary TPUs offer a way for these huge systems to train much faster. Faster training means researchers can run experiments much faster, thereby increasing the speed of AI research.

Why this partnership is good for AI research

Traditionally, Google and Facebook have ran their independent AI research through Google DeepmindGoogle Brain and Facebook AI Research. As a result, the AI tooling echo system has become split on a Tensorflow (Google’s AI framework) vs Pytorch debate. While competition has allowed both frameworks to move at breakneck speed, it’s also made research reproducibility more difficult.

Google makes partnership with Facebook on  AI technology

If this announcement signals a more collaborative approach to AI research, we could potentially see improved interoperability between these two frameworks. The result could make AI deployment on smartphones easier, unify the tooling ecosystem around these frameworks, and improve reproducibility of research results.

“The long-term goal is to enable everyone to enjoy the simplicity and flexibility of PyTorch while benefiting from the performance, scalability, and cost-efficiency of Cloud TPUs,” he added.
PyTorch 1.0 accelerates the workflow involved in taking breakthrough research in AI to production deployment, Facebook said.

(Source)

Sharing is caring
  •  
  •  
  •  
  •  
  •  
  •  
Manish Kumar Barnwal

Manish Kumar Barnwal

Hye guys, I am Manish a professional Digital Marketer, Blogger & Youtuber. And I am just 18 years old. Tech Affair 24 is my dream blog, here on my blog I share articles related to Technology, Digital Marketing, Tech News and Online Earning.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.