Google Expands TensorFlow Open Source Tools to Accelerate Machine Learning Development

Join top executives in San Francisco on July 11-12 to hear how leaders are integrating and optimizing AI investments for success.. Learn more

The big artificial intelligence (AI) news at Google I/O today is the release of the company’s PaLM 2 large language model, but that’s not the only AI news at the event.

The company is also rolling out a number of open source machine learning (ML) technology upgrades and enhancements for the growing TensorFlow ecosystem. TensorFlow is an open source technology effort, led by Google, that provides machine learning tools to help developers build and train models.

Google is launching its new DTensor technology at Google I/O. This technology brings new parallelism techniques to ML training, helping to improve model training and scaling efficiency.

Image Credit: Google

There is also a preview version of the TF Quantification API, which is intended to help make models more resource-efficient overall and thus reduce development cost.


transform 2023

Join us in San Francisco on July 11-12, where top executives will share how they’ve integrated and optimized AI investments to achieve success and avoid common pitfalls.

Register now

A key part of the TensorFlow ecosystem is the Keras API set, which provides a set of deep learning capabilities based on the Python language in addition to the core TensorFlow technology. Google announces a couple of new Keras tools: KerasCV for computer vision (CV) applications and KerasNLP for natural language processing (NLP).

“A big part of what we’re seeing in terms of tools and the open source space is really driving new capabilities, new efficiency and new performance,” Alex Spinelli, Google’s vice president of product management for machine learning, told VentureBeat. . “Absolutely, Google is going to embed amazing, amazing AI and ML into its products, but we also want to create kind of a rising tide that lifts all boats, so we’re really committed to our open source strategies and enabling developers in general.” .

TensorFlow is still the ‘workhouse’ of machine learning at Google

In an era where extensive language models (LLMs) are all the rage, Spinelli stressed that it is now even more critical than ever to have the right ML training tools.

“TensorFlow is still the workhorse of machine learning today,” he said. “It is still… the fundamental underlying infrastructure [in Google] that drives a lot of our own machine learning developments.”

To that end, DTensor updates will provide more “horsepower” as ML training requirements continue to grow. DTensor introduces more parallelization capabilities to help streamline training workflows.

Spinelli said that ML in general is increasingly hungry for data and computing resources. As such, it is extremely important to find ways to improve performance to process more data to meet the needs of ever larger models. New updates to Keras will provide even more power, with modular components that truly allow developers to create their own natural language processing and computer vision capabilities.

Even more power will come to TensorFlow thanks to the new JAX2TF technology. JAX is a research framework for AI, widely used at Google as a computational library, to build technologies like the Bard AI chatbot. With JAX2TF, models written in JAX will now be easier to use with the TensorFlow ecosystem.

“One of the things that we’re really excited about is how these things will turn into products and seeing the developer community flourish,” he said.

PyTorch vs. TensorFlow

While TensorFlow is the workhorse of Google’s ML efforts, it’s not the only open source ML training library.

In recent years, the open source PyTorch framework, originally created by Facebook (now Meta), has become increasingly popular. In 2022, Meta contributed PyTorch to the Linux Foundation, creating the new PyTorch Foundation, a multi-stakeholder effort with an open government model.

Spinelli said that what Google is trying to do is support developer choice when it comes to ML tools. He also noted that TensorFlow isn’t just an ML framework, it’s a complete ecosystem of tools for ML that can help support training and development for a wide range of use cases and deployment scenarios.

“This is the same set of technologies, essentially, that Google uses to build machine learning,” Spinelli said. “I think we have a really competitive offering if you really want to build high-performance systems at scale and want to know that they will work across all infrastructures of the future.”

One thing Google apparently won’t do is follow Meta’s lead and create an organization separate from the TensorFlor Foundation.

“We’re pretty comfortable with the way it’s playing out today and the way it’s being managed,” Spinelli said. “We’re pretty comfortable with some of these great updates that we’re rolling out now.”

VentureBeat’s mission is to be a digital public square for technical decision makers to gain insights into transformative business technology and transact. Discover our informative sessions.


Scroll to Top