Meta has transferred the management and control of PyTorch to the recently established PyTorch Foundation. The foundation will continue to develop the AI platform and its ecosystem.
The tech giant stated it’s time to transfer responsibility for the AI platform to the PyTorch Foundation. A “new home” should allow development to accelerate.
The PyTorch Foundation was recently established by the Linux Foundation. The administrator of the Linux ecosystem maintains oversight. According to the organization, the AI platform is housed in a neutral environment. This should be beneficial as the importance of and popularity of AI grows.
Development of PyTorch
PyTorch was developed in 2016 by Facebook, the precursor to Meta. The deep learning framework has since supported hundreds of AI projects. One of the platform’s major advantages is that developers and data scientists can use Python as the main programming language for their AI projects and models. Like Python, PyTorch is very popular.
The AI framework was designed for GPU hardware. PyTorch also supports so-called ‘tensors’, allowing data to be created and manipulated through methods unavailable in other frameworks. As a result, PyTorch can be used for various tasks. The framework has several modules that extend its functionality to diverse deep learning applications.
The PyTorch Foundation will work closely with the user community and developers who maintain the framework. The foundation will also focus on further expanding the ecosystem. To this end, several partners sit on the board, including AMD, AWS, Google Cloud, Meta Platforms, Microsoft and Nvidia. The number of partners should increase in the coming years.
In addition, a strong focus on the PyTorch community is expected. According to the foundation, developers are the engine behind the framework’s success. The organization promises to ensure that the development and decision-making process becomes transparent and accessible to all stakeholders.