Facebook made known a few days ago that you're betting on PyTorch as your default AI framework, since its current artificial intelligence models perform trillions of operations every day alreadyl bet on Pytorch, it seeks to satisfy this growing demand for workload As the company said that by migrating all their systems, they will be able to innovate much more quickly while ensuring a more optimal experience for all their users.
For those unaware of PyTorch, they should know that is an open source machine learning library which is based on the Torch library. It was created by Facebook's artificial intelligence research unit and it is already used to power a wide range of artificial intelligence applications, such as computer vision and natural language processing models.
Examples of PyTorch AI models include customizing user feeds and stories on Instagram, and identifying and removing hate speech on Facebook.
Adopting PyTorch as Facebook's default AI framework helps ensure that all experiences on our technologies will run optimally at Facebook scale and for everyone, regardless of device, operating system, or internet connection quality. that they may have
Facebook mentions that this migration also means that you can work together with a community more closely never:
PyTorch not only makes our research and engineering work more effective, collaborative and efficient, it also allows us to share our work as open source PyTorch libraries and learn from the advancements made by the thousands of PyTorch developers around the world. .
One of the reasons to go to PyTorch is that the process from research to production of AI has traditionally been tedious and complex, and another of the main problems to be addressed is that researchers were forced to choose between AI frameworks optimized for research or production, but not for both.
Today, more than a year into the migration process, there are more than 1.700 PyTorch-based inference models in full production on Facebook, and 93 percent of our new training models, those responsible for identifying and analyzing content. on Facebook, they're on PyTorch.
"This new iteration merged Python-based PyTorch with production-ready Caffe2 and merged graphical and immediate run modes, providing flexibility for research and performance optimization for production," Facebook wrote on its blog. "PyTorch engineers at Facebook introduced a family of tools, libraries, pre-trained models and data sets for each stage of development, enabling the developer community to rapidly create and implement new AI innovations at scale."
In other words, Facebook is choosing PyTorch because it is a unique framework for research and production AI models which provides flexibility to experiment and also the ability to launch AI on a large scale when it's ready for prime time. That makes it possible to deploy new models in minutes rather than weeks, Facebook said, while reducing the infrastructure and engineering burden that comes with maintaining two different artificial intelligence systems.
The goal of our PyTorch migration is to create a smoother end-to-end developer experience for our engineers and developers. We want to accelerate our process from research to production by using a single platform that allows us the flexibility to experiment along with the ability to launch AI models at production scale.
PyTorch it also has an advantage when it comes to running AI models directly on devices like smartphones. This is because Facebook has created the PyTorch Mobile framework that reduces binary sizes at runtime to ensure that PyTorch AI models can run on devices with minimal processing power.
Source: https://ai.facebook.com