Facebook open sources tools for bigger, faster deep learning models

Facebook on Friday open sourced a handful of software libraries that it claims will help users build bigger, faster deep learning models than existing tools allow.

The libraries, which [company]Facebook[/company] is calling modules, are alternatives for the default ones in a popular machine learning development environment called Torch, and are optimized to run on [company]Nvidia[/company] graphics processing units. Among the modules are those designed to rapidly speed up training for large computer vision systems (nearly 24 times, in some cases), to train systems on potentially millions of different classes (e.g., predicting whether a word will appear across a large number of documents, or whether a picture was taken in any city anywhere), and an optimized method for building language models and word embeddings (e.g., knowing how different words are related to each other).

“‘[T]here is no way you can use anything existing” to achieve some of these results, said Soumith Chintala, an engineer with Facebook Artificial Intelligence Research.

That team was formed in December 2013 when Facebook hired prominent New York University researcher Yann LeCun to run it. Rob Fergus, one of LeCun’s NYU colleagues who also joined Facebook at the same time, will be speaking on March 19 at our Structure Data conference in New York.

A heatmap showing performance of Facebook's modules to standard ones on datasets of various sizes. The darker the green, the faster Facebook was.

A heatmap showing performance of Facebook’s modules to standard ones on datasets of various sizes. The darker the green, the faster Facebook was.

Despite the sometimes significant improvements in speed and scale, however, the new Facebook modules probably are “not going to be super impactful in terms of today’s use cases,” Chintala said. While they might produce noticeable improvements within most companies’ or research teams’ deep learning environments, he explained, they’ll really make a difference (and justify making the switch) when more folks are working on stuff at a scale like Facebook is now — “using models that people [previously] thought were not possible.”

Perhaps the bigger and more important picture now, then, is that Friday’s open source releases represent the start of a broader Facebook effort to open up its deep learning research the way it has opened up its work on webscale software and data centers. “We are actually going to start building things in the open,” Chintala said, releasing a steady stream of code instead of just the occasional big breakthrough.

Facebook is also working fairly closely with Nvidia to rework some of its deep learning programming libraries to work at web scale, he added. Although it’s working at a scale beyond many mainstream deep learning efforts and its researchers change directions faster than would be feasible for a commercial vendor, Facebook’s advances could find their way into future releases of Nvidia’s libraries.

Given the excitement around deep learning right now — for everything from photo albums to self-driving cars — it’s a big deal that more and better open source code is becoming available. Facebook joins projects such as Torch (which it uses), Caffe and the Deeplearning4j framework being pushed by startup Skymind. Google has also been active in releasing certain tooling and datasets ideal for training models.

It was open source software that helped make general big data platforms, using software such as Hadoop and Kafka, a reality outside of cutting-edge web companies. Open source might help the same thing happen with deep learning, too — scaling it beyond the advances of leading labs at Facebook, Google, Baidu and Microsoft.

Project Ara will offer at least three chip choices, including a Tegra K1

Ahead of the second Project Ara developer’s conference in January, Paul Ermenko, project head, has shared a few more details about what to expect from Google’s ambitious modular phone on his Google Plus page.

One tidbit Eremenko revealed is that the Project Ara team has been working on a module that uses an [company]Nvidia[/company] Tegra K1 processor, which comes from same line of chips that are used in Google’s Nexus 9 tablet. [company]Google[/company] calls it an “application processor” or an “AP,” and it’s a module which houses the CPU, the GPU, RAM, cellular modem, and other core system components. There will also be an AP made with Marvell’s silicon, a company that makes decidedly lower-powered chipsets, including those that power Google’s Chromecast. The [company]Marvell[/company] AP will use the PXA1928, which is a 64-bit quad-core chip based on ARM Cortex A53 cores.

These two new chip module reference designs are in addition to a previously-announced Rockchip-based AP expected to be demoed in early 2015. These three chips will likely cost varying amounts, which fits in with the Project Ara philosophy of offering modular choice for devices starting as inexpensive as $50. But the Tegra K1-based AP indicates that there will be Project Ara options that emphasize performance, as well.

Based on job listings and previous statements made by team members, it looks like Project Ara is gearing up for a “market pilot” next year, although that doesn’t necessarily mean modular phones will be ready for mainstream consumers — not tinkerers — by next holiday season.

There are still several significant challenges that need to be addressed: The devices are still in prototype form, and only in the 2nd half of 2014 did the team demonstrate a Project Ara device successfully booting up in public. In addition, the key to Project Ara is the modules it will work with — whether they house improved processors, cameras, or even unusual sensors like blood glucose meters. The module store is in the works, but it’s still not a concept that’s been tried or tested before. Next year is shaping up to be a key period for Google’s modular phone experiment.

Samsung counter-sues Nvidia in major fight over chip patents

Nvidia, a pioneer in the field of graphic chips for smartphones and video games, sued Samsung in September for allegedly using patent-infringing Qualcomm chips in many of its devices. Now, Samsung has returned the favor with a counter-suit that claims Nvidia is violating the Korean companies’ own patents related to buffering techniques for semi-conductors. Samsung also claims that Nvidia engaged in false advertising by claiming that its Shield tablet has the world’s fastest mobile processor. Such counter-suits are common in patent cases, and this is likely just the beginning of a multi-year throw-down in the chip industry. You can get background on the case here, see more details about Samsung claims here, and read Nvidia’s response to the counter-suit here.