PyTorch Day China 2025

• A Practical Framework for AI Governance: Mitigating Copyright & Content Risk - Ye Tao, Grandall Law Firm
• A Small Step in AI, a Giant Leap for Automation - Yi Yuan, Independent
• AI & Creativity: How Machine Learning Is Shaping the Future of UX - Moyin Ipinmoroti, Oak Theory
• AI for Startups: Scaling Innovation With PyTorch - Abhilekh Verma, Abhilekh Verma Consultancy (OPC) Private Limited
• Development of a Parkour Game Based on Unity3D: Design, Implementation, and User Experience - Shang Yu Yu & Che Zhang, Shenzhen Technology University
• DoubleSnakeRL: A PyTorch-based Multi-Agent Deep Reinforcement Learning Framework for Competitive Gam - Qu Shijie & Che Zhang, Shenzhen Technology University
• Edge as a Service (EAAS), Challenge and Opportunities in Edge AI Vision - Norman Lam, Anlyvis Limited
• Effortless Transition Solution of CUDA-Based PyTorch Scripts To Alternative AI Accelerators - Zejun Chen, Intel
• CrossScaleMHA: A Lightweight Attention Module with Multi-Scale Patch-Based Sequence Modeling for Embedding Enhancement - Fu ChuZhao & Che Zhang, Shenzhen Technology University
• Fault Diagnosis and Recovery Algorithm Based on Fat-Tree Network Structure - Yu Zhang, Shenzhen University of Technology & Che Zhang, Shenzhen Technology University
• Methods for Distributed Cluster Training of Large Models - Yinuo Jiang, Shenzhen technology university & Che Zhang, Shenzhen Technology University
• ONNXSequential: Seamless ONNX Model Composition and Inference in Kornia - Jian Shi, Kornia.AI
• Optimizing PyTorch ATen Operators for RISC-V - Fei Zhang, Institute of Software Chinese Academy of Sciences
• Path Selection Algorithms in Software-Defined Network - Kai Liu & Che Zhang, Shenzhen Technology University
• Predicting Severe COVID-19 Outcomes in Madagascar Using PyTorch - Tombolaza Canut Filamant, University Of Toamasina
• PyTorch Performance Analysis Tool - Wang Lei, Advanced Compilation Lab
• Self-Healing API Architecture: Harnessing PyTorch for Resilient Cloud Systems - Vamsi krishna reddy Munnangi, University of Central Missouri
• Sentiment Analysis of Weibo Comments Based on BERT - Caihong Luo & Che Zhang, Shenzhen Technology University
• The Design of Implementation of AIOps ChatBot - Tengzi Luo & Che Zhang, Shenzhen Technology University
• vLLM Ascend: A Scalable Hardware Accelerator on NPU for Large Scale Inference - Yikun Jiang, vLLM Ascend Team
• WebTraffic Time Series Predictions Using LSTM & ARIMA Models - Chaoran Yang & Che Zhang, Shenzhen Technology University

• A Practical Framework for AI Governance: Mitigating Copyright & Content Risk - Ye Tao, Grandall Law Firm
• A Small Step in AI, a Giant Leap for Automation - Yi Yuan, Independent
• AI & Creativity: How Machine Learning Is Shaping the Future of UX - Moyin Ipinmoroti, Oak Theory
• AI for Startups: Scaling Innovation With PyTorch - Abhilekh Verma, Abhilekh Verma Consultancy (OPC) Private Limited
• Development of a Parkour Game Based on Unity3D: Design, Implementation, and User Experience - Shang Yu Yu & Che Zhang, Shenzhen Technology University
• DoubleSnakeRL: A PyTorch-based Multi-Agent Deep Reinforcement Learning Framework for Competitive Gam - Qu Shijie & Che Zhang, Shenzhen Technology University
• Edge as a Service (EAAS), Challenge and Opportunities in Edge AI Vision - Norman Lam, Anlyvis Limited
• Effortless Transition Solution of CUDA-Based PyTorch Scripts To Alternative AI Accelerators - Zejun Chen, Intel
• CrossScaleMHA: A Lightweight Attention Module with Multi-Scale Patch-Based Sequence Modeling for Embedding Enhancement - Fu ChuZhao & Che Zhang, Shenzhen Technology University
• Fault Diagnosis and Recovery Algorithm Based on Fat-Tree Network Structure - Yu Zhang, Shenzhen University of Technology & Che Zhang, Shenzhen Technology University
• Methods for Distributed Cluster Training of Large Models - Yinuo Jiang, Shenzhen technology university & Che Zhang, Shenzhen Technology University
• ONNXSequential: Seamless ONNX Model Composition and Inference in Kornia - Jian Shi, Kornia.AI
• Optimizing PyTorch ATen Operators for RISC-V - Fei Zhang, Institute of Software Chinese Academy of Sciences
• Path Selection Algorithms in Software-Defined Network - Kai Liu & Che Zhang, Shenzhen Technology University
• Predicting Severe COVID-19 Outcomes in Madagascar Using PyTorch - Tombolaza Canut Filamant, University Of Toamasina
• PyTorch Performance Analysis Tool - Wang Lei, Advanced Compilation Lab
• Self-Healing API Architecture: Harnessing PyTorch for Resilient Cloud Systems - Vamsi krishna reddy Munnangi, University of Central Missouri
• Sentiment Analysis of Weibo Comments Based on BERT - Caihong Luo & Che Zhang, Shenzhen Technology University
• The Design of Implementation of AIOps ChatBot - Tengzi Luo & Che Zhang, Shenzhen Technology University
• vLLM Ascend: A Scalable Hardware Accelerator on NPU for Large Scale Inference - Yikun Jiang, vLLM Ascend Team
• WebTraffic Time Series Predictions Using LSTM & ARIMA Models - Chaoran Yang & Che Zhang, Shenzhen Technology University

Motivation
PyTorch supports a wide range of acceleration hardware beyond CPUs, including CUDA, XPU, MPS, NPU, HPU, and more. Its architecture allows new backend integration through two key components: ATen operators and device runtime.
While ATen operators are device-agnostic, the runtime remains device-specific, relying on APIs like torch.cuda and torch.xpu. This fragmentation complicates writing portable, hardware-agnostic code across PyTorch and its ecosystem.
To address this challenge, we propose torch.accelerator: a unified, device-agnostic runtime API for stream-based accelerators.
Design
An Accelerator refers to a device that collaborates with the CPU to accelerate computation, typically via asynchronous execution using Stream and Event for synchronization. Our design assumes a single active accelerator per host.
The torch.accelerator API provides a consistent interface for device and stream management, with backend support integrated via the existing DeviceGuardImplInterface registration mechanism.
Further Work
We are actively working on a unified device memory API. These will streamline the library, model, and UTs.
Reference
https://github.com/pytorch/pytorch/pull/132204

Deploying PyTorch in large-scale heterogeneous computing have huge challenges in stability, usability, performance, and quality.

This session will showcase best practices using real-world Huawei Ascend application scenarios and community contributions, demonstrating PyTorch and ecosystem components in large-scale heterogeneous computing.

Through Ascend-optimized technologies and architecture improvements in PyTorch, we have greatly enhanced computing performance in traning and inference, meeting production standards.

We’ve proposed and implemented features for heterogeneous compute and accelerator quality standards in PyTorch. Driving PyTorch’s application in heterogeneous compute, like PyTorch ecosystem projects: DeepSpeed, vLLM.

Finally, we will share real user stories from large-scale training and distributed inference, presenting best practices on the Ascend platform and helping the community address challenges in diverse compute scenarios.

PyTorch is one of the most popular frameworks for deep learning and machine learning, and Intel has been a long-term contributor and advocate for the PyTorch community. In this talk, we will share our experience in contributing to PyTorch in the core framework and its ecosystem. We will show how to make AI applications more popular through the improvement of hardware computing power and the optimization of open source software. We will introduce the latest relevant progress on Intel GPUs on PyTorch, and the low-cost deployment solution of DeepSeek R1 671B full-blooded version on Intel Xeon CPU. We will also introduce some PyTorch ecosystem projects that we have participated in in the past, such as HuggingFace, vLLM, SGLang, etc. Finally, we will discuss future plans and visions and continue to work with the PyTorch Foundation to drive deep learning and machine learning in a better direction.