PyTorch, an open source machine learning software based on the Torch library, is used for applications such as computer vision and natural language processing. With the software that started to be developed by Facebook (Meta) in 2018, important projects have been signed so far. Now, a new era begins with PyTorch.
PyTorch also got Intel support behind it
Founded in September 2022, the PyTorch Foundation set out to get more open management, more collaboration and contribution. With the release of the beta version of PyTorch 2.0 in December 2022, the organization received its first fruits. PyTorch 2.0 has so far managed to reach 428 different contributors who have provided new code and capabilities to the open source project.
The primary focus for PyTorch 2.0 is ‘performance’. The Accelerated Transformer, which is at the center of Large Language Models (LLMs) and artificial intelligence, can be shown as the main objective of the project, enabling the connection between different concepts.
Transformers stand out as the most fundamental technology that enables productive artificial intelligence to reach this level, including OpenAI’s models such as GPT-4. Transformers prepared with PyTorch 2.0 provide an important support for artificial intelligence training and inference.
While maintaining the PyTorch 2.0 claim, one of the biggest companies supporting this project is Intel. Although the importance of graphics cards is often underlined in Artificial Intelligence and Machine Learning, Intel states that processors also play an important role in this process.
That’s why Intel announced that they are preparing important optimizations for their processors for PyTorch 2.0. With the new optimization, Intel users are expected to get top performance even when using the largest models.
With ChatGPT, companies are now taking more serious steps in artificial intelligence. The PyTorch 2.0 project is one of the biggest proofs of this. So what do you think about PyTorch 2.0 and Intel optimization? You can share your views with us in the comments section.